article
stringlengths 0
745k
| abstract
stringlengths 0
92.7k
|
|---|---|
a torus @xmath0 with fundamental group marking is uniformized by the complex plane @xmath1 with variable @xmath2 and a lattice generated by @xmath3 and @xmath4 , @xmath4 in the upper half plane @xmath5 .
the change of marking equivalence relation for tori is given by the action of the modular group @xmath6 on @xmath5 .
the grtzsch and rauch variational formulas provide that the differential of the moduli parameter @xmath4 is represented by the quadratic differential @xmath7 . with the analogy to higher genus moduli in mind , consider @xmath5 as the teichmller space , @xmath6 as the mapping class group and the quotient as the moduli space .
the compactification of the quotient @xmath8 is given by introducing the coordinate @xmath9 for a neighborhood of infinity .
the differential of the nonzero moduli parameter @xmath10 is @xmath11 , which formally vanishes at infinity or equivalently @xmath12 is represented by the quadratic differential @xmath13 . in particular , quadratic differentials model the logarithmic derivative of the moduli parameter at infinity .
equivalently , the moduli cotangent @xmath14 at infinity includes a @xmath10 factor . the infinitesimal deformation space of a pair @xmath15 , a compact riemann surface and a distinguished point , is the cohomology group @xmath16 for the canonical bundle @xmath17 and the inverse of the point bundle . by kodaira - serre duality ,
the dual infinitesimal deformation space is @xmath18 , the space of holomorphic quadratic differentials with possible simple poles at @xmath19 .
quadratic differentials with simple poles give the moduli cotangent space .
our goal is to generalize this result for nodal stable curves and describe the moduli cotangent space as a coherent analytic sheaf on the family .
our discussion begins with the local geometry of the model case , the family of hyperbolas @xmath20 in @xmath21 .
consider for @xmath22 positive , the singular fibration of @xmath23 over @xmath24 with the projection @xmath25 .
the family @xmath26 is the local model for the formation and deformation of a node .
the general fiber is an annulus and the special fiber is the union of the germ of the coordinate axes with the origin the node .
the projection differential is @xmath27 .
the vertical line bundle @xmath28 over @xmath29 has non vanishing section @xmath30 and the relative dualizing sheaf @xmath31 has non vanishing section @xmath32 .
we consider curve families with noded stable fibers .
a proper surjective map @xmath33 of analytic spaces is a family of nodal curves provided at each point @xmath34 , either @xmath35 is smooth with one dimensional fibers , or the family is locally analytically equivalent to a locus @xmath36 in @xmath37 over @xmath38 with @xmath39 vanishing at @xmath40 .
the locus @xmath41 is the loci of nodes .
we consider families with first order vanishing of @xmath39 ; for a neighborhood of a node the family is analytically equivalent to a cartesian product of @xmath26 and a complex manifold .
stability of fibers is the condition that each component of the nodal complement in a fiber has negative euler characteristic .
negative euler characteristic ensures that the automorphism group of a component is finite .
the general fiber of a family @xmath33 of nodal curves is a smooth riemann surface .
the locus of noded curves within the family is a divisor with normal crossings . for a family @xmath33 ,
the relative dualizing sheaf is isomorphic to the product of canonical bundles @xmath42 .
degenerating family @xmath43.,width=230,height=230 ] the relative dualizing sheaf @xmath44 provides a generalization of the family of canonical bundles for a family of riemann surfaces . for @xmath45 positive
, sections of @xmath46 over open sets of @xmath47 , generalize families of holomorphic @xmath45-differentials for riemann surfaces .
a section @xmath48 of @xmath46 on a neighborhood of a node @xmath49 is given as @xmath50 with @xmath51 holomorphic in @xmath52 .
for an annulus fiber and the mapping @xmath53 into a fiber , the section is given as @xmath54 sections of @xmath55 generalize families of holomorphic quadratic differentials and in the language of bers are families of regular @xmath56-differentials @xcite .
infinitesimal opening of a node is described by the variational formula for the parameter @xmath57 . for @xmath57 nonzero
, the pairing of the infinitesimal variation of @xmath57 with the section @xmath48 is given in lemma [ plumderiv ] as the @xmath58 multiple of the @xmath59-constant coefficient in the laurent expansion of @xmath60 .
the appearance of the scaling @xmath57-factor is intrinsic to the variation of a node .
the present appearance of the @xmath57-factor is dual to the appearance in the torus example . the noded fibers sub family of @xmath49 is @xmath61 over @xmath62 with the restriction of the section @xmath48 to the fibers given as @xmath63 on @xmath64 and as @xmath65 on @xmath66 .
the residue of @xmath48 at the node is @xmath67 .
the residue is coordinate independent and well defined modulo a sign ; for @xmath45 even the residue is well defined .
the kernel of the residue map to @xmath1 defines a subsheaf of @xmath55 as follows . for a family @xmath33 ,
let @xmath68 be the @xmath69 component of the loci of nodes in @xmath70 and let @xmath71 be the skyscraper sheaf supported on @xmath72 .
the vanishing residue subsheaf is defined by the exact sequence @xmath73 in the preparatory lemma [ frameeval ] , we describe local bases of sections of @xmath55 over suitable opens sets of @xmath47 .
the bases have a direct relationship to coordinate cotangent frames for the open sub family of riemann surfaces .
lemma [ frameeval ] provides the basic tool for understanding the vanishing residue subsheaf and the moduli cotangent sheaf . in section 4
, we show that the direct image @xmath74 , with presheaf of sections of @xmath75 over open sets of @xmath47 , is locally free and that the corresponding sequence of direct images is also exact .
we then combine the coordinate cotangent frame for sub families property , the definition of @xmath75 and a formal argument to establish the main result .
the vanishing residue exact sequence is natural for admissible families . the direct image of the vanishing residue subsheaf is naturally identified with the cotangent sheaf for admissible families .
we discuss applications .
a section of the moduli cotangent sheaf over an open set of the base is a section of @xmath55 over the open set with residues vanishing . in the torus example
the section vanishes along the fiber and necessarily the residue vanishes . a description of the moduli cotangent fiber at a nodal curve is provided by the isomorphism between a locally free sheaf and an analytic vector bundle . for deformations of a riemann surface ,
moduli cotangents are represented by quadratic differentials on the surface .
for deformations of a nodal stable curve , moduli cotangents are represented by equivalence classes of sections of the vanishing residue subsheaf - it is not sufficient to specify the restriction of a section to a fiber of a family . in lemma [ cotfib ] , we find that the moduli tangent - cotangent pairing extends to noded fibers .
we also consider the basic geometry of the divisor of noded curves within the family .
we find that the conormal sheaf of the divisor is the subsheaf of @xmath55 with sections over base open sets vanishing along all noded fibers .
the log - cotangent sheaf of the divisor is the direct image of @xmath55 , with sections of @xmath55 over base open sets .
the works of masur @xcite and hubbard - koch @xcite provide a foundation for the present considerations . in section 3 , we give the construction of a standard nodal stable family of curves following masur .
the construction begins with an open set in teichmller space parameterizing riemann surfaces with distinguished points and combines plumbing copies of the family @xmath26 .
the constructed family provides the setting for our considerations .
hubbard and koch introduce the notion of a @xmath76-marking for families of nodal stable curves , a marking modulo dehn twists about the elements of a multi curve .
they construct a universal analytic family for a @xmath76-marking and show that the deligne - mumford compactification of the moduli space of riemann surfaces is analytically described by quotients of @xmath76-marked universal families . from the results of hubbard and koch , our considerations apply to the deligne - mumford compactification . in section 5 , we generalize the laurent coefficient map of lemma 4 for expansion ( [ laurcoeff ] ) .
the generalization is a period about a collar latitude of the product of a section of the vertical line bundle and a section of @xmath55 .
the generalization enables direct comparison of the infinitesimal plumbings @xmath77 and @xmath20 of a node .
let @xmath33 be a proper family of stable curves with a nodal fiber @xmath78 with a neighborhood of a node analytically equivalent to the cartesian product of @xmath26 and a parameter space @xmath62 .
consider the local coordinates @xmath79 and @xmath80 , @xmath81 , for neighborhoods of the inverse images of the node on the normalization of @xmath78 .
for the plumbing @xmath77 of @xmath78 contained in @xmath33 and @xmath48 a section of @xmath75 on a neighborhood of @xmath78 , the initial plumbing tangent evaluates as @xmath82 the formula generalizes the standard fact that for the deligne - mumford compactification the normal sheaf to the divisor of noded curves is the direct sum of products of tangent lines at distinguished points on the normalized curves .
in section 6 we present three applications for the vanishing residue sheaf .
the first involves the action of the automorphism group of a stable curve on infinitesimal deformations .
the quotient of the action on the moduli tangent space provides a local model for the deligne - mumford compactification .
we apply the main theorem and give a detailed description of the automorphism action on the vanishing residue sheaf . in the second application
, we give a detailed discussion of plumbing an elliptic curve family to a @xmath83-pointed @xmath84 .
we describe the initial moduli cotangent space and the action of the elliptic curves and pointed-@xmath84 involutions on moduli parameters and moduli cotangents . in the third application
, we apply a standard construction for plumbing an abelian differential and compare two approaches for calculating the variation of period .
i would like to especially thank john hubbard and sarah koch for valuable conversations .
for a complex manifold , we write @xmath85 for the sheaf of holomorphic functions , @xmath0 for the holomorphic tangent sheaf , @xmath86 for the holomorphic cotangent sheaf and @xmath17 for the canonical bundle , the determinant line bundle of @xmath86 . equivalently @xmath0 is the sheaf of holomorphic vector fields and @xmath86 the sheaf of holomorphic @xmath3-forms . we introduce a singular fibration of a neighborhood @xmath29 of the origin in @xmath21 over a neighborhood @xmath87 of the origin in @xmath1 . for positive constants @xmath22 , with @xmath88 and @xmath24 ,
the family @xmath26 is the singular fibration with projection @xmath89 .
the differential of the projection @xmath90 vanishes only at the origin ; the @xmath91 fiber crosses itself at the origin . for @xmath92 , solving for @xmath2 the fiber of @xmath93 is @xmath94 with @xmath95 and for @xmath91 , the fiber is the union of discs @xmath96 with @xmath97 and @xmath98 with @xmath99 in @xmath21 .
the family @xmath29 over @xmath87 is a family of annuli degenerating to a one point union of a @xmath2 disc and a @xmath100 disc in the axes of @xmath21 .
alternatively , @xmath29 over @xmath87 is a representative of the germ at the origin of the family of hyperbolas limiting to the union of coordinate axes .
.,width=240,height=240 ] the vector field @xmath101 is vertical on @xmath102 , since @xmath103 vanishes .
let @xmath104 be another vertical vector field , non vanishing on @xmath102 .
since on @xmath102 , @xmath105 is rank one , it follows that @xmath106 , for @xmath39 a section of @xmath107 . by hartog
s theorem @xcite , @xmath39 is analytic on @xmath29 and since @xmath108 are non vanishing on @xmath102 , it follows that @xmath39 is non vanishing on @xmath29 .
the observations provide that the condition _ vertical vector field _ defines a line bundle @xmath109 over @xmath29 and @xmath110 represents a non vanishing section ( the vector field @xmath110 , a section of @xmath111 , vanishes at the origin ; the corresponding line bundle section is non vanishing ; @xmath109 is not a sub bundle of @xmath111 ) . in general , vertical vector fields that vanish at most in codimension @xmath56
correspond to non vanishing vertical line bundle sections .
the @xmath112-fiber of @xmath29 over @xmath87 is an example of an open noded riemann surface @xcite , alternatively an open nodal curve .
the @xmath112-fiber is normalized by removing the origin of @xmath21 to obtain a @xmath2 disc , punctured at the origin and a @xmath100 disc , punctured at the origin .
the origins are filled in to obtain disjoint discs .
analytic quantities on the @xmath112-fiber , lift to analytic quantities on the normalization . the sheaf of regular @xmath3-differentials , @xcite , equivalently the dualizing sheaf @xcite , ( * ? ? ? * dualizing sheaves , pg .
82 ) associates to the @xmath112-fiber : abelian differentials @xmath113 with at most simple poles respectively at the origin for the @xmath114 discs and the important residue matching condition @xmath115 .
bardelli provides an exposition on the basics of families of stable curves @xcite .
he discusses the topology of fibers , the arithmetic genus , the dualizing and relative dualizing sheaves , line bundles and divisors , riemann roch , as well as khler differentials and first order deformation theory .
the fiber tangent spaces of @xmath102 over @xmath87 are subspaces of @xmath116 .
the meromorphic differential @xmath117 is a functional on the fiber tangent spaces of @xmath102 . for the @xmath112-fiber , it is immediate that @xmath118 is a section of the dualizing sheaf .
the differential @xmath118 satisfies the relations @xmath119 the differential is uniquely determined modulo the submodule @xmath120 by each relation . to motivate the definition of the _ relative dualizing sheaf _ for @xmath29 over @xmath87
, we consider the coset of @xmath118 in @xmath121 .
the finite pairing @xmath122 provides that the coset has holomorphic sections on @xmath102 .
let @xmath123 be another meromorphic differential on @xmath29 with @xmath124 holomorphic and non vanishing on @xmath29 .
on @xmath102 , the quotient sheaf @xmath121 is pointwise rank one and thus @xmath125 for @xmath39 holomorphic on the domain . again by hartog s theorem @xcite , the function @xmath39 is analytic on @xmath29 .
the relation @xmath126 and non vanishing of the first quantity imply that @xmath39 is non vanishing on @xmath29 . in particular ,
the differentials @xmath123 with @xmath124 non vanishing , considered as elements of @xmath121 , define a line bundle over @xmath29 ; the line bundle is not a sub bundle of @xmath111 .
the differentials @xmath123 correspond to non vanishing holomorphic sections of the line bundle .
the constant relation @xmath127 , shows that the line bundle is the dual of the vertical line bundle @xmath109 .
the relative dualizing sheaf @xmath128 is defined to have sections of the relative cotangent bundle @xmath129 , given by differentials satisfying the polar divisor and residue conditions ( * ? ? ?
iii ) , ( * ? ? ?
* dualizing sheaves , pg .
in particular , we have the sheaf equality @xmath130 and the quotient @xmath121 is the intended cokernel .
we have described the relative dualizing sheaf and shown that it is dual to the vertical line bundle @xmath109 .
we will use that for the families considered , the restriction of the relative dualizing sheaf to a fiber of the family is the dualizing sheaf of the fiber ( * ? ? ?
since the total space @xmath29 is smooth , there is a description of @xmath128 in terms of the canonical bundle @xmath131 and pullback @xmath132 of the canonical bundle of @xmath87 ( * ? ? ?
* definition 3.5 ) , ( * ? ? ? * dualizing sheaves , pg .
84 ) ; in particular @xmath133 where @xmath134 denotes the dual . as above , a differential @xmath123 , satisfying the polar divisor and residue conditions
, determines a coset in @xmath121 . a differential @xmath123 and a non vanishing section @xmath135 of @xmath136 together determine the element @xmath137 in @xmath138 .
the association @xmath139 is independent of the particular choice of @xmath135 and realizes the sheaf isomorphism ( [ omegak ] ) .
we are interested in general curve families @xmath33 including nodal curves or equivalently analytic families of noded riemann surfaces ( possibly open ) with smooth total space @xmath70 and base @xmath47 .
we require that the projection @xmath35 is a submersion on the complement of a codimension @xmath56 subset .
families of open or compact riemann surfaces and the above nodal family are included in the considerations .
for each node of a fiber , we assume that for a neighborhood the family @xmath33 is analytically equivalent to a cartesian product of the standard nodal family @xmath26 and a complex manifold parameter space @xmath62 .
the vertical line bundle @xmath28 and relative dualizing sheaf @xmath44 are defined for the family @xmath33 .
the vertical line bundle is defined from @xmath140 and the relative dualizing sheaf is @xmath141 .
the positive powers @xmath46 of the relative dualizing sheaf , alternatively the sheaf of regular @xmath45-differentials @xcite , is defined to have sections @xmath48 of @xmath142 with at most order @xmath45 poles at the image of the nodes on the normalizations of nodal fibers and the residue matching @xmath143 for the forms @xmath144 on the normalization .
the power @xmath46 is dual to the power @xmath145 of the vertical line bundle .
the cartesian product of @xmath26 and @xmath62 provides a local model for the family geometry .
we present an explicit isomorphism for the sheaf isomorphism @xmath146 restricting domains as necessary , let @xmath135 be a non vanishing section of the relative dualizing sheaf @xmath44 and @xmath4 a non vanishing section of @xmath147 .
consider the association between sections @xmath148 of @xmath46 and @xmath45-canonical forms for @xmath70 , sections @xmath149 of @xmath150 - the association is given by the formula @xmath151 observations are in order . on the submersion set for @xmath35 ,
a non vanishing section of @xmath152 and the pullback of a local frame for @xmath153 together form a frame for @xmath154 .
it follows that the product @xmath155 is a non vanishing section of the canonical bundle @xmath156 on the submersion set . with the codimension @xmath56 condition , the product is non vanishing in general - consequently the ratio @xmath157 is an analytic function on @xmath70 . the relation can be inverted to give a formula for @xmath149 in terms of @xmath148 . in particular , the association provides a local isomorphism of sheaves .
the right hand side depends on the choice of @xmath4 , but is homogeneous of degree zero in @xmath135 , and so is independent of the particular choice of @xmath135 .
the association establishes a canonical isomorphism between @xmath46 and @xmath150 twisted by @xmath158 .
the isomorphism @xmath159 is a general form of the isomorphism ( [ omegak ] ) .
most important for our considerations , @xmath45-canonical forms present a local model for sections of powers of the relative dualizing sheaf , alternatively a model for families of regular @xmath45-differentials .
we use the local cartesian product description of the family to define annuli in the fibers and annular laurent expansions of sections of @xmath46 .
the annuli and expansion coefficients depend on the local analytic equivalence of @xmath33 to a cartesian product .
first , the fibers of @xmath26 , the annuli @xmath160 or the one point union of @xmath2 and @xmath100 discs , define annuli or a nodal region in the fibers of @xmath33 .
the differential @xmath32 is a non vanishing section of @xmath31 and gives a non vanishing section of the relative dualizing sheaf for the cartesian product .
the annulus @xmath161 maps into the @xmath57-fiber of @xmath29 by @xmath162 or by @xmath163 with the differential @xmath32 pulling back to @xmath164 .
the @xmath56 factor enters in comparing expansions in @xmath94 to expansions in @xmath59 .
in particular the differential @xmath32 has residues @xmath165 when considered on the normalization , while the pulled back differential @xmath166 has residue @xmath167 . for all our considerations we define the residue on the normalization and the annular laurent coefficients in terms of the pull back to the annulus @xmath168 .
a section @xmath48 of @xmath169 is given as @xmath170 since @xmath52 is a coordinate triple for @xmath171 .
in particular for the coefficient function given by a power series @xmath172 the pullback of the @xmath45-differential to @xmath168 for a @xmath57 nonzero fiber has the annular laurent series @xmath173 with @xmath59-constant term @xmath174 the @xmath112-fiber of @xmath29 is the union of @xmath64 with coordinate @xmath2 and @xmath66 with coordinate @xmath100 .
the @xmath64 and @xmath66 restrictions of @xmath51 are the functions @xmath175 the @xmath59-constant term is long recognized as important in the analysis and variational theory of embedded annuli @xcite .
the residue of the differential @xmath48 at the node of the @xmath91 fiber is simply the zeroth coefficient @xmath176 .
the laurent @xmath59-constant term is important in the considerations of the following sections .
we concentrate on the square of the relative dualizing sheaf .
[ deflaur ] for a section @xmath48 of @xmath55 and a local analytic equivalence of @xmath33 to a cartesian product @xmath26 and @xmath62 , define the @xmath59-constant coefficient @xmath177 [ laurdef ] we consider @xmath178 as a sheaf map @xmath179 defined on open sets given as suitable cartesian products .
the sheaf map depends on the cartesian product equivalence .
the value of @xmath180 at @xmath91 is the fourfold multiple of the residue of @xmath48 on the normalization - the residue is intrinsically defined .
we describe the local deformation space , as well as , the associated family for a nodal stable curve , equivalently for a compact noded riemann surface .
the description includes the family tangent and cotangent coordinate frames .
the discussion closely follows ( * ? ? ?
* sections 2 , 5 and 7 ) , ( * ? ? ?
* section 3 ) and especially ( * ? ? ?
* section 2 ) .
similar treatments are found in @xcite and @xcite . a riemann surface with nodes @xmath182 is a connected complex space , such that every point has a neighborhood analytically isomorphic to either the unit disc in @xmath1 or the germ at the origin of the intersection of the coordinate axes in @xmath21 .
the special points of @xmath182 are the nodes .
we write @xmath183 for the nodal complement @xmath184 .
the normalization of @xmath182 is @xmath183 with the removed nodes considered as distinguished points .
functions , line bundles and sections of line bundles on @xmath182 lift to corresponding quantities on @xmath183 with a matching condition .
the components of @xmath183 are the parts of @xmath182 .
provided @xmath182 is compact , each component of @xmath183 is described as a compact surface minus a finite number of points .
we now assume that each part has negative euler characteristic .
in particular a noded riemann surface is equivalent to a stable curve @xcite . in the kodaira - spencer setup
the infinitesimal deformation space of a compact complex manifold @xmath185 is the ech cohomology @xmath186 @xcite .
the infinitesimal deformation space of a pair @xmath15 , a riemann surface and a distinguished point , is @xmath16 for @xmath17 the canonical bundle and the inverse of the point line bundle @xmath19 ( @xmath187 is the line bundle with a section with divisor @xmath19 ) . by the dolbeault isomorphism @xmath188 @xmath189 the sheaf of smooth vector fields vanishing at @xmath19 @xcite .
the elements of the dolbeault group are equivalence classes of @xmath190 forms with values in smooth vector fields vanishing at @xmath19 - smooth beltrami differentials .
a beltrami differential @xmath191 with support disjoint from @xmath19 represents a trivial infinitesimal deformation exactly when there is a smooth vector field @xmath192 , vanishing at @xmath19 , with @xmath193 . by kodaira - serre duality
the dual of the infinitesimal deformations is @xmath18 , the space of holomorphic quadratic differentials with a possible simple pole at @xmath19 . in terms of a local coordinate @xmath2 , the integral pairing for a beltrami differential @xmath194 and a holomorphic quadratic differential @xmath195 is @xmath196 for @xmath197 the euclidean area element in @xmath2 .
since holomorphic sections vanishing on an open set vanish identically , it follows that given an open set in @xmath182 there exists a basis for @xmath198 of elements supported in the open set .
we describe local coordinates for the teichmller space of @xmath182 @xcite .
associated to a beltrami differential of absolute value less than unity is a deformation of @xmath182 .
specifically given an atlas @xmath199 for the surface and a beltrami differential , define new charts as follows : for @xmath2 the coordinate on @xmath200 and @xmath201 the local expression for @xmath191 , let @xmath202 be a homeomorphism solution of the beltrami equation @xmath203 on @xmath204 @xcite .
the new atlas @xmath205 defines the new surface @xmath206 .
local holomorphic coordinates for the teichmller space of @xmath207 are given as follows : choose @xmath208 compactly supported , spanning the dolbeault group @xmath209 for @xmath210 small , @xmath211 satisfies @xmath212 and @xmath213 is a riemann surface .
the assignment @xmath214 for @xmath215 small , is a local coordinate for teichmller space .
we will use that a chart @xmath216 for @xmath182 disjoint from the supports @xmath217 is a chart for @xmath218 ; a fixed set of charts can be used for a neighborhood of the distinguished points .
the initial variation of a family @xmath218 can be found as follows .
a one - parameter family of solutions of the beltrami equation @xmath219 has an initial variation @xmath220 $ ] satisfying the @xmath221-potential equation @xmath222_{\bar z}=\nu$ ] .
the initial variation @xmath222 $ ] is a vector field and for deformations of the pair @xmath15 , vector fields vanish at @xmath19 .
the data @xmath223\circ z_{\alpha})\}$ ] is a @xmath112-cocycle with values in @xmath189 .
moreover solutions of the @xmath221-potential equation are unique modulo holomorphic functions .
the ech coboundary of @xmath223\circ z_{\alpha})\}$ ] is the deformation class in @xmath16 .
we are interested in the spaces of regular quadratic differentials for a proper family @xmath33 of stable curves .
as above , the total spaces @xmath70 and @xmath47 are smooth ; @xmath35 is a submersion on the complement of a codimension @xmath56 subset ; each part of a fiber has negative euler characteristic . for a sheaf @xmath38 on @xmath70 , the direct image presheaf @xmath224 assigns to open sets @xmath225 in @xmath47 sections of @xmath38 on @xmath226 .
we review that the direct image @xmath227 of the relative dualizing sheaf square is a locally free sheaf of the expected dimension - the direct image is the sheaf of holomorphic sections of a vector bundle .
first we observe that the dimension of the space of regular quadratic differentials on the fibers of @xmath33 is constant .
the fibers of the family are compact noded riemann surfaces ; the distinguished points on the normalization of a fiber are inverse images of nodes .
a regular quadratic differential , equivalently a section of the dualizing sheaf square , is a meromorphic quadratic differential on each part with possible double poles at the inverse images of nodes and residues equal for each pair of nodal inverse images . by riemann roch and the negative euler characteristic hypothesis
, for a part of genus @xmath228 with @xmath229 distinguished points the dimension of the space of quadratic differentials with possible double poles is @xmath230 .
a node corresponds to a pair of distinguished points on the normalization and a pair of distinguished points contributes @xmath231 to the dimension of quadratic differentials before imposing the equal residue condition
. a node net contributes @xmath83 to the dimension ; or equivalently each distinguished point on the normalization contributes a net @xmath232 to the dimension . with this counting convention ,
the contribution to the dimension of regular quadratic differentials from a part of the normalization is @xmath233 , the @xmath234 multiple of the euler characteristic of the part .
the euler characteristic is additive - the sum of euler characteristics of parts of a fiber coincides with the euler characteristic of the general fiber .
it follows that the dimension of the space of regular quadratic differentials on a fiber is a constant , equal to the @xmath234 multiple of the euler characteristic of the general fiber . in the following , we write @xmath235 for the genus of the general fiber .
the following result on families of differentials is proposition 5.1 of @xcite and proposition 4.1 of @xcite .
masur proves the result by considering the sheaf of canonical forms on @xmath70 and considering the short exact sequences given by forming poincar residues onto successive hyperplane intersections with the family to define the fibers in @xmath70 .
direct image sheaves are formed .
grauert s theorem @xcite is applied to show that the direct image sheaves are coherent .
kodaira - serre duality and a sequence chase are used to show that the poincar residue maps are surjective , his first result .
he applies the result to construct local sections of @xmath236 with prescribed polar divisors .
a general argument about products of @xmath3-forms is applied to find the desired families of regular quadratic differentials .
we give a sketch of the much simpler hubbard - koch argument @xcite .
[ locfree ] the direct image sheaf @xmath227 is locally free rank @xmath237 .
a regular quadratic differential on a fiber of @xmath33 is the evaluation of a section @xmath227 .
begin with a proper analytic map @xmath33 of analytic spaces and a coherent sheaf @xmath238 on @xmath70 .
for a point @xmath239 , consider the fiber @xmath240 and @xmath241 the coherent sheaf on @xmath78 whose stalk at @xmath242 is the tensor product @xmath243 of the stalk of @xmath238 with the structure sheaves of @xmath70 and @xmath78 . by a form of the cartan - serre theorem with parameters the direct image of @xmath238 on @xmath47 is a locally free sheaf provided the dimensions of the cohomology groups @xmath244 are constant . for the sheaf @xmath55 the higher cohomology groups vanish and from the above discussion the zeroth cohomology has constant rank @xmath237 .
the direct image sheaf is locally free of rank @xmath237 .
the second statement of the lemma follows since the dimension of regular quadratic differentials equals the rank of the direct image sheaf .
the general fiber of @xmath33 is a smooth riemann surface with restrictions of sections of @xmath227 to a fiber giving @xmath245 , the space of holomorphic quadratic differentials on the fiber . by hypothesis a neighborhood of each node in @xmath33 is analytically equivalent to a cartesian product with @xmath26 .
the pairing of the infinitesimal variation of the parameter @xmath57 , @xmath92 , with @xmath245 is fundamental to our considerations . for the pairing ( [ serrepair ] )
, the following lemma is given as formula 7.1 in @xcite , formula ( 4 ) in @xcite and in proposition 9.7 in @xcite .
the proof is to give a quasiconformal map from a @xmath57-fiber of @xmath26 to a @xmath246-fiber , compute the initial @xmath246-derivative of the beltrami differential and compute the pairing ( [ serrepair ] ) with a quadratic differential on the @xmath57-fiber .
the quasiconformal map can be given to preserve concentric circles and as a rotation near the boundary .
[ plumderiv ] the pairing of the infinitesimal @xmath57-variation , @xmath57 nonzero , with a section @xmath247 of the direct image sheaf @xmath227 on the @xmath57-fiber is @xmath248 we examine the geometric role of the @xmath57-scaling in the following sections .
we now construct a general family for varying a nodal stable curve / a riemann surface with nodes @xmath249 . in the next section ,
we describe the hubbard - koch theorem for the universal property of the family .
varying @xmath249 is given as a combination of quasiconformal deformations of a relatively compact set in the normalization of @xmath249 and gluings to copies of the family @xmath26 .
the construction only uses that beltrami differentials define finite deformations .
ech style sliding deformations could equally be used in the construction in place of beltrami differentials .
we start with the normalization of @xmath249 : a riemann surface @xmath182 with formally paired distinguished points @xmath250 with corresponding local coordinates @xmath251 near @xmath252 , @xmath253 , @xmath254 near @xmath255 , @xmath256 and with beltrami differentials @xmath257 a basis for the cohomology group @xmath258 .
we require that the domains of the local coordinates @xmath259 are mutually disjoint and that the union of local coordinate domains is disjoint from the union of beltrami differential supports .
let @xmath260 with @xmath215 small and let @xmath213 be the deformation described above by quasiconformal maps .
the parameter @xmath261 is a local coordinate for the teichmller space of @xmath182 .
the coordinates @xmath259 are holomorphic for @xmath218 , given the disjoint support hypothesis .
we next plumb the family @xmath218 to @xmath262 copies of the family @xmath26 . choose a positive constant @xmath263 , such that the local coordinate domains contain the discs @xmath264 . for @xmath265
, remove from @xmath218 the discs @xmath266 and @xmath267 to obtain an open surface @xmath268 .
we now write @xmath62 for the connected domain of the parameter @xmath261 and @xmath87 for the disc @xmath269 .
boundary neighborhoods of individual fibers of @xmath270 are glued to boundary neighborhoods of individual fibers of @xmath26 .
given @xmath271 , identify @xmath272 in the boundary neighborhood @xmath273 to the point @xmath274 in the fiber of a @xmath69 factor of @xmath26 and identify @xmath19 in the boundary neighborhood @xmath275 to the point @xmath276 in the fiber of a @xmath69 factor of @xmath26 .
the result is a family of compact possibly noded riemann surfaces @xmath277 with projection - in brief the family given by a boundary neighborhood gluing of @xmath270 and @xmath262 copies of the plumbing .
the plumbings determine labeled annuli in the fibers of the family .
the constructed family @xmath277 is an @xmath181 family .
the total space and base of an @xmath181 family are complex manifolds and the projection is proper and a submersion except on the combined nodal set of codimension @xmath56 . by construction ,
a neighborhood of a node in an @xmath181 family is analytically equivalent to a cartesian product of @xmath26 and a complex manifold .
the dimension count for regular quadratic differentials provides for @xmath235 the genus of a smooth fiber that the dimensions satisfy @xmath278 .
[ frameeval ] for an @xmath181 family @xmath277 , there is a neighborhood of the origin in @xmath279 with sections @xmath280 of the direct image sheaf @xmath281 with the following pairings on @xmath282 @xmath283 for @xmath284 , @xmath285 with @xmath286 the kronecker delta and @xmath287 the coefficient map for the @xmath69 annulus in a fiber .
on @xmath288 the restriction of the sections @xmath289 to the fibers are the cotangents for the coordinates @xmath181 .
we begin by showing that the pairings of sections of the direct image sheaf with the tangents @xmath290 and the coefficient maps @xmath287 are holomorphic on a neighborhood of the origin in @xmath279 . the initial riemann surface @xmath182 has atlas @xmath199 and the surface @xmath218 has atlas @xmath205 for @xmath291 solutions of @xmath292 on @xmath200 .
the composition rule for quasiconformal maps provides that on @xmath293 the beltrami differential for the infinitesimal variation @xmath290 is @xmath294 for @xmath215 small , the beltrami differentials @xmath295 are bounded with compact support on the surface @xmath182 .
a section @xmath247 of the direct image sheaf @xmath281 has uniformly bounded @xmath296 integral on the support of the beltrami differentials in the fibers .
it follows that the pairings @xmath297 are bounded . by the basics of teichmller theory the pairings
are holomorphic on @xmath298 . by the riemann extension theorem ,
the pairings are consequently holomorphic on the domain of @xmath247 in @xmath279 .
the discussion of sections of @xmath299 provides that the coefficient maps @xmath300 are holomorphic on the domain of @xmath247 . by lemma [ locfree ] , there are sections @xmath301 with the given pairings and laurent coefficients at the origin @xmath302 . from the above ,
the matrix of evaluating pairings and laurent coefficients is holomorphic and the identity at the origin .
the matrix inverse applied to the sections @xmath301 gives the desired sections with the desired pairings and laurent coefficients .
the statement about cotangents follows immediately from lemma [ plumderiv ] . for all applications of lemma [ frameeval ] , we restrict the base of the family to provide that the sections @xmath280 and relations are defined for the family . for an @xmath181 family @xmath277 and the direct image sheaf @xmath227 , referring to lemma [ frameeval ] , the sections @xmath280 are the first @xmath181 frame and the sections @xmath289 are the second @xmath181 frame .
we consider proper families of stable curves @xmath33 including nodal curves with @xmath70 and @xmath47 smooth manifolds .
we assume a neighborhood of each node is analytically equivalent to a cartesian product of @xmath26 and a complex manifold parameter space .
the cartesian product structure provides that the locus of nodes is a disjoint union of codimension @xmath56 smooth subvarieties .
a section of @xmath55 has a canonical residue at a node ; the residue is computed on the normalization .
we consider the vanishing residue subsheaf @xmath75 of @xmath55 and the associated exact sequence .
the direct image sheaf @xmath74 is shown to be locally free and the sequence of direct images is exact .
the hubbard and koch characterization and description of a universal family are applied to show that the direct image @xmath74 is the cotangent sheaf for an admissible family .
applications include that first @xmath181 frames are frames for the log - cotangent sheaf of the divisor of noded curves and second @xmath181 frames are coordinate cotangent frames . for a family @xmath33 as above .
on @xmath70 , the locus @xmath303 is the @xmath69 component of the loci of nodes and @xmath304 is the skyscraper sheaf with fiber @xmath1 on @xmath305 and @xmath112 fiber on the complement .
the vanishing residue subsheaf @xmath75 of @xmath55 is defined by the exact sequence @xmath306 where @xmath307 is the residue on @xmath305 and the @xmath112 map on the complement . although we do not provide the general argument , the relative dualizing sheaf , vanishing residue subsheaf and exact sequence are natural for analytic families @xmath33 . on the complement of nodes in a family @xmath33 , the sheaves @xmath75 and @xmath55 coincide ; @xmath75 generalizes the notion of families of holomorphic quadratic differentials .
the expansion ( [ fzws ] ) for sections of @xmath169 shows that at a node the stalk of @xmath75 has rank @xmath56 over the structure sheaf with generators @xmath2 and @xmath100 in terms of the local coordinate @xmath52 ; the sheaf @xmath75 is not locally free . for an @xmath181 family @xmath277 ,
the elements of the second @xmath181 frame of lemma [ frameeval ] are sections of @xmath75 .
the direct image @xmath74 is a locally free sheaf with basis the second @xmath181 frame .
the sequence of direct images is exact @xmath308 the first statement follows from lemma [ frameeval ] , noting that the fourfold multiple of the residue at a node is the laurent coefficient evaluated at @xmath91 . by lemma [ locfree ] , @xmath281 is locally free rank @xmath237 and from lemma [ frameeval ] , the first @xmath181 frame elements are linearly independent on fibers , and thus are independent over @xmath309 . to consider direct images
, we work with presheaves on @xmath279 .
given the inclusion @xmath310 , an element @xmath247 of @xmath74 is a linear combination @xmath311 for @xmath312 and @xmath313 the first @xmath181 frame . on the locus of the @xmath69 node @xmath303 the residues @xmath314 vanish and @xmath315 .
it follows that @xmath316 vanishes on the divisor @xmath317 .
the coordinate @xmath318 vanishes to order one on the divisor ; it follows that @xmath319 for @xmath320 .
the observation applies for each node and consequently @xmath247 is a combination over @xmath309 of the second @xmath181 frame .
the direct image of @xmath75 is consequently locally free with basis the second @xmath181 frame .
the first and second @xmath181 frames are respectively frames over @xmath309 for sections of @xmath321 and @xmath75 on @xmath322 . the constant section @xmath3 is a frame for @xmath304 over @xmath309 on @xmath305 . by lemma [ frameeval ] , the @xmath323 element of the first frame maps by the combined residue to the constant section @xmath324 of @xmath304 and @xmath112 for the remaining skyscrapers .
the combined residue is surjective . by the above argument
, an element of the kernel of the combined residue map is a combination of second frame elements and the sequence is exact at @xmath281 .
we use the work of hubbard and koch to relate @xmath181 families @xcite .
hubbard and koch study proper flat families of nodal curves ( * ? ? ?
* definition 3.1 ) .
a proper surjective analytic map @xmath33 of analytic spaces is a family of nodal curves provided at each point @xmath34 , either @xmath35 is smooth with one dimensional fibers or the family is locally isomorphic to the locus of @xmath36 in @xmath325 over @xmath62 with @xmath39 vanishing at the image of @xmath40 .
we work with the case of first order vanishing of @xmath39 , where families are locally analytically equivalent to the cartesian product of @xmath26 and a complex manifold ( * ? ? ?
* chapter 10 , proposition 2.1 ) .
families are further assumed to be @xmath76-marked ,
@xmath76 a multi curve on a reference surface ( * ? ? ?
* definitions 5.1 , 5.2 and 5.5 ) ; a @xmath76-marking is a continuous section into a family of cosets of mapping classes from a reference surface to the fibers of the family .
they show in their proposition 5.7 that a proper flat family @xmath33 of stable curves has a @xmath76-marking if and only if there is a hausdorff closed set @xmath326 such that : i ) each component of @xmath327 is either an annulus or a one point union of discs ( a degenerate annulus ) and ii ) @xmath328 is a trivial bundle of surfaces with boundary .
the multi curve @xmath76 represents the annuli that can degenerate in the family . by construction @xmath181
families are proper flat with fibers stable curves with hausdorff closet set the complement of the plumbing annuli .
an @xmath181 family is an example of their @xmath76-marked plumbed analytic family @xmath329 ( * ? ? ?
* sections 8.2 , 8.3 ) .
they also construct a standard model family @xmath330 ( * ? ? ?
* definition 7.1 , section 7.1 ) . by their theorem 9.11 ,
the induced map from a neighborhood of the origin in the base of an @xmath181 family into @xmath331 is a @xmath76-marking preserving topological equivalence of families .
their theorem 10.1 provides that for an @xmath181 family @xmath277 , @xmath76-marked by plumbing annuli , there is a unique analytic map of @xmath279 into @xmath331 such that the @xmath76-marked family @xmath330 pulls back to a @xmath76-marked family analytically equivalent to @xmath277 .
in particular , @xmath181 families provide local analytic descriptions for the standard model family @xmath330 .
furthermore , for a pair of @xmath181 families with a common fiber , there are neighborhoods of the base points and @xmath76-markings such that the families over the neighborhoods are uniquely @xmath76-marked equivalent under a biholomorphism .
basic considerations provide that standard model families @xmath330 and @xmath181 families are locally _ kuranishi _ ( * ? ? ?
* chapter 11 , section 4 ) .
a deformation of a curve @xmath249 is a family of noded curves @xmath33 and basepoint @xmath332 with a prescribed analytic equivalence of @xmath249 to the fiber @xmath333 .
a deformation @xmath33 of a curve @xmath249 is kuranishi provided for any deformation of the curve @xmath334 and basepoint @xmath335 , that the family @xmath334 is locally at @xmath336 the pullback of the family @xmath33 locally at @xmath337 by a unique morphism respecting the equivalences to @xmath249 .
existence of a kuranishi family for a curve is equivalent to stability of the curve ( * ? ? ?
* chapter 11 , theorem 4.3 ) .
in particular , a standard model family @xmath330 with basepoint @xmath338 is locally the pullback of a kuranishi family . at the same time
, the kuranishi family is locally @xmath76-marked by plumbing annuli .
the hubbard and koch theorem 10.1 provides that the kuranishi family is locally the pullback of the standard model family .
in particular , the families are locally equivalent for neighborhoods of the designated fibers . the local uniqueness property for the kuranishi family transfers to the standard model family .
kuranishi families have many general properties ( * ? ? ?
* chapter 11 , section 4 ) .
an admissible family is a proper @xmath76-marked family @xmath33 of stable curves with a neighborhood of each node analytically equivalent to a cartesian product of @xmath26 and a complex manifold .
hubbard and koch use their standard model family to define an analytic structure for the quotient of the augmented teichmller space by local actions of mapping class groups ( * ? ? ?
* section 12 ) . in theorem 13.1 , they show that the constructed space is analytically equivalent to the analytic form of the deligne - mumford compactification of the moduli space of riemann surfaces . in particular @xmath181 families
provide local analytic descriptions for the local manifold covers of the deligne - mumford compactification .
[ main ] the vanishing residue sequence ( [ vanseq ] ) is natural for admissible families . the direct image of the vanishing residue sheaf is naturally identified with the cotangent bundle for admissible families .
the proof is formal given the present considerations ; we use simplified notation .
we use local descriptions in terms of @xmath181 families . for @xmath76-marked families
@xmath277 and @xmath339 with a common fiber , there is a biholomorphic equivalence of the families over neighborhoods of the common base point ; restrict the families to the neighborhoods .
if the number of nodes on the common fiber is not maximal for the families then @xmath262 and @xmath340 can be distinct .
the relative dualizing sheaf is characterized either by its dualizing property or as the dual of the vertical line bundle .
the nodal loci are characterized by self crossing of the fiber of the projection and the residue on the normalization is intrinsic .
a biholomorphism @xmath192 pulls back the second family to the first family .
the vanishing residue exact sequence for the second family pulls back to the vanishing residue exact sequence for the first family . on the domain of @xmath192
the coordinate cotangents are related by @xmath341 for an analytic jacobian . by lemma [ frameeval ] , on the open
set @xmath342 the cotangents are represented by the second @xmath181 frames @xmath343 for @xmath344 and @xmath345 for @xmath346 . on the open set the second frames satisfy the relation @xmath347 the left and right hand sides independently give @xmath237 analytic sections of @xmath348 and @xmath74 .
the relation is consequently satisfied on the domain of @xmath192 .
since @xmath345 and @xmath343 are local bases for the locally free sheaves @xmath349 and @xmath74 and @xmath350 is the cocycle for the cotangent bundle , it follows that the direct image of @xmath75 is analytically equivalent to the cotangent bundle .
the equivalence does not depend on the particular representation as an @xmath181 family .
for an admissible family the equivalence is given by any local representation .
for an admissible family the form for a local frame of @xmath74 is determined by general considerations .
begin with the restriction map from the relative dualizing sheaf to the dualizing sheaf of a fiber . by the dimension
count leading to lemma [ locfree ] , for a noded fiber the space of sections of the dualizing sheaf square vanishing at nodes has codimension the number of nodes . by the same count
, for each node there is a local section of the dualizing sheaf square with unit residue at the node and vanishing at other nodes . now
the product of a function defining the associated locus of nodal curves in the base and a section of the relative dualizing sheaf square with constant unit residue is a section of @xmath74 corresponding to the node . as sections of @xmath74 , the constructed products are locally independent and independent from sections with non trivial restrictions to the fiber .
the collection of sections provides a local frame for @xmath74 .
we will see below that a section of @xmath74 vanishing on noded fibers is a conormal to the locus of noded curves .
[ cotanframe ] for an @xmath181 family the second @xmath181 frame is the coordinate cotangent frame . to understand the description of the cotangent fibers , we review the isomorphism between locally free sheaves and analytic vector bundles . associated to an analytic vector bundle @xmath351 over a connected complex manifold @xmath185 is the sheaf @xmath352 of analytic sections .
the sheaf is locally free . in particular
if @xmath351 is trivial over an open set @xmath225 , the space of sections over @xmath225 is isomorphic to @xmath353 for @xmath354 the rank of @xmath351 .
conversely , if for an open cover @xmath355 of @xmath185 , the space of sections of a sheaf @xmath356 is given as locally free by isomorphisms @xmath357 , then the cocycles @xmath358 define invertible maps of @xmath359 with the compatibility condition @xmath360 on @xmath361 .
the cocycle @xmath362 defines a rank @xmath354 vector bundle on @xmath185 ( * ? ? ?
ii ) . given a locally free sheaf @xmath356 , the fibers of the corresponding vector bundle are described by an elementary sheaf construction . for the fiber at @xmath272 in @xmath185 , introduce the sky - scraper sheaf @xmath363 supported at @xmath272 ; a complex number defines a section of @xmath363 and @xmath363 is an @xmath364 module by evaluating functions at @xmath272 .
the vector bundle associated to @xmath356 has fiber at @xmath272 the cohomology group @xmath365 .
a germ at @xmath272 of a section of @xmath356 defines an element of the fiber by extending by zero for the complement of @xmath272 .
germs @xmath336 and @xmath366 at @xmath272 of sections of @xmath356 define the same element of the fiber provided for representatives on a neighborhood of @xmath272 , that @xmath367 for @xmath39 in @xmath368 with @xmath369 or equivalently @xmath370 for @xmath371 a representative of a germ of a section and @xmath372 in @xmath368 with @xmath373 .
an example of the fiber construction is for the rank one sheaf @xmath374 for a point on a riemann surface . for germs of sections @xmath375 and @xmath376 at @xmath272 , each with divisor @xmath272 , then the germs satisfy @xmath377 for @xmath39 a germ of @xmath378 at @xmath272 . in the @xmath272-fiber @xmath379 ,
the class of @xmath375 is the @xmath380 multiple of the class of @xmath376 .
we apply the elementary sheaf construction to describe the fibers of the cotangent bundle of an @xmath181 family .
[ cotfib ] let @xmath277 be an @xmath181 family with second @xmath181 frame @xmath238 and @xmath381 , the ideal sheaf of a point in the base .
the cotangent space of @xmath279 at @xmath272 is identified with @xmath382 .
there is a well defined pairing between beltrami differentials supported away from the nodes on the fiber @xmath383 , the functionals @xmath384 and the cotangent vectors at @xmath272 given by forming limits with representatives .
a frame for the cotangent sheaf of the @xmath181 family is given by the coordinate frame @xmath238 .
the description of the cotangent fiber at @xmath272 comes from the above sheaf description of a vector bundle fiber . from the proof of lemma [ frameeval ] , the pairings of the tangent functionals with @xmath238 are holomorphic on a neighborhood of the origin in @xmath279 .
it follows that elements in @xmath385 have pairings vanishing at @xmath272 .
the conclusion follows .
we can describe the cotangent fiber at a point for the curve @xmath249 by giving sections of the dualizing sheaf square and local coordinates at the nodes .
the normalization of @xmath249 is a riemann surface @xmath182 with formally paired distinguished points @xmath250 .
assume that local coordinates @xmath251 near @xmath252 , @xmath253 and @xmath254 near @xmath255 , @xmath256 are given .
consider @xmath386 the space of meromorphic quadratic differentials on @xmath182 with at most simple poles at the distinguished points @xmath250 ; @xmath386 is the cotangent space of the teichmller space of @xmath182 .
given an @xmath181 family @xmath277 containing a fiber isomorphic to @xmath249 , we describe sections of @xmath74 representing the cotangents at @xmath249 . by lemma [ locfree ] , each section @xmath135 of the dualizing sheaf square is the evaluation of a section @xmath387 of the direct image @xmath281 .
the @xmath181 family is equivalent to a family with local coordinates @xmath259 for plumbing data .
the coordinates define sheaf maps @xmath388 . by lemma [ frameeval ] , there are also sections @xmath389 , @xmath390 , of the direct image with @xmath59-coefficients @xmath391 , @xmath392 , for @xmath286 the kronecker delta . for @xmath393 ,
the system of equations @xmath394 , @xmath392 , for functions @xmath395 , has a local solution . solving the system defines a linear map depending on the data of local coordinates : for @xmath393 , associate @xmath396 , a section of @xmath74 .
furthermore for @xmath397 , the products @xmath398 , @xmath390 , define sections of @xmath74 vanishing on the fiber isomorphic to @xmath249 . by lemma [ cotfib ] ,
the pairings at @xmath249 of the elements @xmath399 , @xmath393 , @xmath390 , with beltrami differentials supported away from the nodes and with the infinitesimal @xmath318-variations are determined completely by the elements of @xmath386 and the local coordinates @xmath259 .
we will see in theorem [ compplum ] that the cotangent vectors depend on the second order expansions of the local coordinates .
the local cartesian product description at a node provides that for an admissible family @xmath33 , the locus @xmath400 of noded curves is a divisor with normal crossings . in general for a normal crossing divisor in a complex manifold @xmath185 , there are local analytic coordinates @xmath401 with @xmath402 for @xmath403 .
we are interested in three cotangent constructions associated to the divisor @xmath404 of noded curves .
the first is the dual of the normal sheaf of @xmath404 , the conormal @xmath405 , defined by the exact sequence @xmath406 for @xmath407 the sheaf of @xmath3-forms on @xmath185 and @xmath408 the sheaf of khler differentials for @xmath404 @xcite .
the conormal is the set of cotangent vectors to @xmath185 supported on @xmath404 and is the annihilator of the tangents to the intersection of @xmath404 components .
for example , at a point of @xmath409 , the conormal is the annihilator of the intersection @xmath410 of tangent spaces . at the same point ,
the conormal sections are generated by the differentials @xmath411 and the khler differentials are generated by @xmath412 .
the second cotangent construction is the log - cotangent bundle @xmath413 for a normal crossing divisor .
the log - cotangent is a subsheaf of the sheaf of meromorphic @xmath3-differentials with @xmath414 the sheaf of analytic @xmath3-differentials . at the above point of @xmath404 ,
the sheaf is locally generated by the differentials @xmath415 .
the log - cotangent bundle is locally free of rank @xmath416 .
the third cotangent construction , supported on @xmath404 , is the space of cotangents represented on the normalizations of the stable curves by holomorphic quadratic differentials ; the differentials are holomorphic at the nodal inverse images . in theorem
[ compplum ] , we show that the space of such cotangents is the annihilator of the infinitesimal plumbing tangents . for an @xmath181 family @xmath277 with parameters
@xmath417 the divisor @xmath404 of noded curves is @xmath418 with @xmath419 .
the locus @xmath404 is the codimension @xmath262 locus with parameters @xmath420 .
[ logpolframe ] for an @xmath181 family the log - cotangent bundle is generated by the first @xmath181 frame .
the conormal of @xmath404 is generated by the elements @xmath421 of the second @xmath181 frame .
the statements are direct consequences of corollary [ cotanframe ] and lemma [ frameeval ] .
we generalize the laurent coefficient map of definition [ laurdef ] for elements of @xmath422 . evaluating the limit of
the generalization will not require a special choice of coordinates .
the generalization enables comparison of infinitesimal plumbings and extends period expansions of quadratic differentials @xcite to the limiting case of a nodal fiber . for the family @xmath26 ,
the vertical line bundle @xmath28 is dual to the relative dualizing sheaf @xmath31 . for a section @xmath423 of @xmath28 , nonzero at the node ,
the adjoint residue value , defined modulo a sign , is the residue of the dual section @xmath424 at the node .
the @xmath425 fibers of the family @xmath26 are annuli . for @xmath48 a section of @xmath31 and @xmath426 a loop in a fiber winding once around the annulus , then the integral @xmath427 is the @xmath428 multiple of the zeroth laurent coefficient of the restriction of @xmath48 to the fiber with the sign of the integral uniquely determined by the orientation of @xmath426 . .
, width=240,height=240 ] given @xmath423 a section of @xmath28 with adjoint residue value @xmath165 , a collection of loops @xmath429 in the fibers @xmath430 , is admissible provided @xmath431 . for @xmath247 a section of @xmath422 and @xmath423 with admissible loops @xmath429 , the laurent pairing is @xmath432 we consider properties of the pairing .
the product @xmath433 is a section of @xmath31 and the integrals only depend on the homology classes of the loops .
the integrals do not depend on a choice of coordinates .
the pairing is analytic in @xmath57 and for @xmath247 a section of the vanishing residue sheaf @xmath75 , then @xmath434 vanishes at @xmath91 . in particular for @xmath247 a section of @xmath75 the limit @xmath435 exists . to find additional properties of the limit
, we introduce expansions for @xmath423 and @xmath247 and calculate . for the local cartesian product description @xmath171
, the considerations will only involve continuity in @xmath62 and so we omit @xmath62 dependence in the expansions .
we begin with a section of @xmath31 and its dual section of @xmath28 . from section [ locgeom ] , for an annulus @xmath168 with variable @xmath59 mapped into a fiber of @xmath29 by @xmath436 , then a @xmath3-differential @xmath437 pulls back to @xmath438 on @xmath168 . for a loop @xmath426 in @xmath168 with @xmath439 ,
the integral @xmath440 equals the @xmath441 multiple of the @xmath59-constant of @xmath51 ; see formula ( [ zconst ] ) .
provided @xmath442 , then in a neighborhood of the origin @xmath443 is an analytic section of @xmath28 with dual section @xmath48 satisfying the hypothesis @xmath444 .
we use the coordinate @xmath94 for @xmath29 to give expansions for @xmath423 a section of @xmath28 and @xmath247 a section of @xmath422 : @xmath445 and @xmath446 where @xmath447 represents degree @xmath354 terms and higher in the variables .
the product has the expansion @xmath448 and the pullback to the annulus @xmath168 has the expansion @xmath449 and the laurent pairing is @xmath450 in particular the coefficient of @xmath57 is determined by the first order expansion of @xmath423 at the node and the second order expansion of @xmath247 .
the first order expansion of the section @xmath423 is determined by the restriction of @xmath423 to the @xmath91 fiber . in the particular case
@xmath451 the laurent pairing coincides with the laurent coefficient , specifically @xmath452 ; compare the expansion ( [ lpform ] ) to formula ( [ zconst ] ) .
in @xcite it is shown that the coefficient functionals @xmath453 , @xmath454 are the evaluations of the _ infinitesimal origin slidings _ for the coordinates @xmath114 of the inverse images of the node on the normalization ; the functionals are the action of the infinitesimal displacement of the location of possible poles for quadratic differentials on the normalization . in the definition of sliding ( * ? ? ? * pgs . 405 , 406 ) , the integral is for a negatively oriented loop about the distinguished point and as noted on ( * ? ? ?
409 ) the factor @xmath455 is omitted from the beltrami - quadratic differential pairing ; compare to the present formula ( [ serrepair ] ) .
the infinitesimal plumbing formula below agrees with the earlier formula ( * ? ? ?
* corollary ) obtained by constructing a two - parameter family interpolating between a pair of plumbings .
the @xmath181 coordinates depend on choosing plumbings .
the following formula gives the initial effect of a change of plumbing data .
to apply the formula for the second @xmath181 frame of lemma [ frameeval ] , note that in a neighborhood of the @xmath69 node , @xmath456 for the second frame element @xmath457 .
[ compplum ] let @xmath33 be a proper family of stable curves with a nodal fiber @xmath78 with a neighborhood of a node analytically equivalent to the cartesian product of @xmath26 and a parameter space @xmath62 .
consider the local coordinates @xmath79 and @xmath80 , @xmath81 , for neighborhoods of the inverse images of the node on the normalization of @xmath78 .
for the plumbing @xmath77 of @xmath78 contained in @xmath33 and @xmath247 a section of @xmath75 on a neighborhood of @xmath78 , the initial plumbing tangent evaluates as @xmath458 for @xmath459 and the given partial derivatives .
the essential matter is a change of coordinates for a neighborhood of a node represented in two ways as a product with @xmath29 .
if a neighborhood of the node is given as the cartesian product of the plumbing @xmath77 and a parameter space , then the initial plumbing tangent is evaluated by lemma [ plumderiv ] and a limit . as noted above
, for @xmath460 coordinates for @xmath29 , the evaluation is equivalently given by the laurent pairing with the section @xmath461 of @xmath28 and a limit .
the section has dual section @xmath462 of @xmath44 .
we now calculate the expansion of @xmath48 for a second description in terms of @xmath29 with coordinates @xmath94 . in general for a change of variable @xmath463 ,
@xmath464 then @xmath465 .
as noted above , the first order expansion of @xmath48 is determined by its restrictions to the coordinate axes in @xmath29 . on the @xmath2-coordinate axis ,
respectively @xmath100-coordinate axis , in @xmath29 the variable @xmath100 , respectively @xmath2 , and its differential vanish .
apply the general change of variable for the coordinate axes .
we have the following relation on the coordinate axes in @xmath29 @xmath466 the dual section of @xmath28 satisfies @xmath467 we apply lemma [ plumderiv ] to find that the initial plumbing tangent is given as the limit @xmath468 .
we apply formula ( [ lpform ] ) and the observation @xmath469 to obtain the final formula .
an easy application is the standard formula for the normal sheaf to the divisor @xmath404 of noded curves ( * ? ? ?
* props . 3.31 , 3.32 ) .
the normal sheaf is the quotient of the tangent sheaf by the intersection of the tangent sheaves of components of @xmath404 .
for an @xmath181 family @xmath277 with parameters @xmath470 , the intersection of components of @xmath404 has parameters @xmath471 .
the normal sheaf is the quotient of the @xmath57 parameter tangents modulo the @xmath261 parameter tangents .
the infinitesimal sliding deformations are represented by compactly supported beltrami differentials on the normalization of a curve .
in particular , the infinitesimal slidings are tangents to the intersection of components of @xmath404 .
theorem [ compplum ] provides the relation @xmath472 between the plumbings @xmath77 and @xmath20 of a node .
the factor @xmath473 is the cocycle for the tensor product of tangent lines to the normalized curve at the inverse images of the node . in particular , the normal line is the product of tangent lines .
hyperbolic metrics provide a norm for the limit pairing @xmath474 as follows . for an @xmath181 family @xmath277 with all @xmath57 parameters nonzero , the fibers are compact riemann surfaces with uniformization hyperbolic metrics depending smoothly on the parameters . for negative curvature metrics
, there is a unique geodesic in each free homotopy class of closed curves .
the geodesic in the class of the @xmath69 annulus has length expansion @xmath475 locally uniformly in @xmath261 with @xmath476 for a single plumbing and @xmath477 in general ( * ? ?
* example 4.3 ) . for a single plumbing ,
the length expansion provides that @xmath478 .
we can define a norm for a single plumbing pairing by considering for @xmath423 a section of @xmath28 and @xmath247 a section of @xmath74 the limit @xmath479 for @xmath423 with adjoint residue value @xmath165 and @xmath247 vanishing on the noded fiber , the definition provides a norm for the conormal to the divisor of noded curves .
we consider that @xmath181 families provide local manifold covers for the deligne - mumford compactification and describe the local actions of the relative mapping class groups .
we begin with the hubbard - koch discussion of markings ( * ? ? ?
* sections 2 , 5 ) . for @xmath76 a labeled multi curve on a compact reference surface @xmath62 ,
let @xmath480 be the group of mapping classes fixing the free homotopy classes of @xmath76 and the components of @xmath481 .
let @xmath482 be the extension of @xmath480 of mapping classes permuting the @xmath76 free homotopy classes and components of @xmath481 .
for @xmath483 the identification space given by collapsing @xmath76 , let @xmath484 be the group of isotopy classes of homeomorphisms of @xmath483 that fix each individual collapsed curve and fix the components of @xmath485 .
let @xmath486 be the extension of isotopy classes of homeomorphisms of @xmath483 . for every multi curve subset @xmath487 ,
define @xmath488 to be the homeomorphisms of @xmath489 , stabilizing each component of @xmath489 and homotopic to a composition of dehn twists about elements of @xmath490 .
associated to an admissible family @xmath33 are local families over @xmath47 of equivalence classes of maps from @xmath62 to the fibers of @xmath70 . in particular for @xmath239 , consider the maps of @xmath62 to @xmath491 taking elements of a multi curve @xmath492 bijectively to the nodes on the fiber .
a pair of such maps @xmath493 are _ equivalent _ provided for @xmath494 that @xmath495 is homotopic to @xmath496 on @xmath489 by a homotopy of homeomorphisms from @xmath489 to @xmath497 .
the set of equivalence classes of maps is @xmath498 .
the union @xmath499 inherits a topology by considering the compact - open topology for maps of @xmath62 into @xmath70 .
the space @xmath500 has a projection @xmath501 to @xmath47 . a @xmath76-marking for a family @xmath33
is defined as a continuous section of the projection .
the mapping class group @xmath482 acts on @xmath76-markings by a right action of precomposition of maps . in analogy to the local topological triviality of families of compact surfaces , hubbard - koch show that the projection @xmath501 has discrete fibers and a local section through each point ( * ? ? ?
* theorem 5.4 ) .
a direct characterization of @xmath76-markable families is provided ( * ? ? ?
* proposition 5.7 ) .
let @xmath502 with @xmath76-marking @xmath503 be the hubbard - koch universal @xmath76-marked family .
the natural map from a connected open set in the base of an @xmath181 family into the base @xmath504 provides a @xmath76-marking preserving analytic equivalence of the @xmath181 family to the pullback family .
the universal property provides that a mapping class @xmath505 which acts by a right action by precomposition on @xmath76-markings , acts by a left action on the base @xmath504 by an analytic equivalence @xmath506 and on the total space by an analytic equivalence @xmath507 to give the family @xmath508 with marking @xmath509 ( * ? ? ?
* theorem 10.1 ) .
if each neighborhood of a basepoint @xmath510 of the family intersects its @xmath506 translate , then @xmath506 fixes @xmath510 and the mapping class @xmath511 is realized by an automorphism acting on the @xmath510 fiber ( * ? ?
* proposition 2.6 ) .
conversely , automorphisms of a fiber act by a left action on maps of @xmath62 to the fiber . by discreteness and existence of @xmath512 marking sections , automorphisms of a fiber of @xmath502 locally act through the mapping class action . for a stable curve @xmath249 , consider a @xmath76-marked family @xmath502 containing a fiber isomorphic to @xmath249 . as described above
, the automorphism group @xmath513 acts on a neighborhood of the point in @xmath504 representing @xmath249 .
let the normalization of the curve @xmath249 be a riemann surface @xmath182 with paired distinguished points @xmath250 with local coordinates @xmath251 near @xmath252 , @xmath253 and @xmath254 near @xmath255 , @xmath256 .
the local coordinates define sheaf maps @xmath515 . by lemma [ frameeval ] , there are sections @xmath389 , @xmath390 , of the direct image @xmath516 with coefficients @xmath391 , @xmath392 , for @xmath286 the kronecker delta . in the discussion following lemma [ cotfib ] ,
the cotangent space at @xmath249 is described in terms of : @xmath386 , the meromorphic quadratic differentials with at most simple poles at distinguished points , the local coordinates @xmath259 and the linear map @xmath517 , extending sections of the dualizing sheaf square .
[ aut ] with the above conventions , @xmath513 acts by a right action on the cotangent fiber at @xmath249 by : the map @xmath507 pulls back @xmath386 , pulls back the local coordinates @xmath259 and pulls back the linear map @xmath28 . for @xmath513
acting on local coordinates by permutations and multiplications by unimodular numbers , then @xmath513 acts by : pulling back @xmath386 , stabilizing the operator @xmath28 and on the cotangents @xmath518 , @xmath519 , @xmath520 by permutations and multiplications by unimodular numbers .
we showed in theorem [ main ] that an analytic equivalence of families acts by pulling back the vanishing residue sheaf .
in particular maps @xmath507 act by pulling back @xmath521 the direct image of the vanishing residue sheaf .
in particular for @xmath522 and associated map @xmath507 , we have by definition of the extending sheaf sections map that @xmath523 for @xmath393 . in general the coordinates @xmath524 and @xmath525 are distinct and extensions of sections are not preserved by pulling back . the group @xmath513 acting on the normalization @xmath182 has cyclic stabilizer at each distinguished point .
local coordinates can be chosen with each stabilizer acting by rotations . for such coordinates , @xmath513 acts by permutations and rotations of coordinates .
for such coordinates , @xmath513 acts preserving extensions of sections of @xmath386 to sections of @xmath74 .
further for such coordinates , @xmath513 acts by a permutation of the sections @xmath389 , @xmath526 and by the same permutation on the coordinate indices and by multiplication by unimodular numbers on the coordinate products .
we describe plumbing elliptic curves ( flat tori ) to a projective line @xmath84 pointed at @xmath112 , @xmath527 and @xmath3 .
the general plumbing gives a genus @xmath56 surface with hyperelliptic involution .
we consider the family cotangent frame and the action of the automorphism group of the initial fiber .
the construction begins with an elliptic curve @xmath528 with variable @xmath2 , universal cover @xmath1 and lattice deck transformation group generated by @xmath3 and @xmath529 .
an elliptic curve has quadratic differentials @xmath530 and @xmath531 for the weierstrass @xmath532-function .
the @xmath532-function is doubly periodic with singular part @xmath533 at the origin ; @xmath534 has unit residue at the origin .
an elliptic curve has the involution @xmath535 with the differentials @xmath530 and @xmath534 involution invariant . the projective line @xmath84 with variable @xmath100 has abelian differentials @xmath536 with residues @xmath3 at @xmath537 and @xmath527 at @xmath510 .
the projective line @xmath84 pointed at @xmath112 , @xmath527 and @xmath3 has the involution @xmath538 .
define a nodal stable curve @xmath249 by pairing distinguished points as follows .
pair the origins on @xmath539 and @xmath84 to form a first node @xmath540 , and pair the points @xmath527 and @xmath3 on @xmath84 to form a second node @xmath541 .
the involutions @xmath542 and @xmath543 of components define involutions on the stable curve by extending by the identity on the remaining component .
define plumbings of the stable curve @xmath249 as follows . at the node @xmath540 , use the local coordinates @xmath544 , @xmath545 and write @xmath546 for plumbing . at the node @xmath541 , use the local coordinates @xmath547 , @xmath548 and write @xmath549 for plumbing .
the elliptic curves @xmath539 and plumbings define a family @xmath550 with general fiber a genus @xmath56 surface over a @xmath83-dimensional base @xmath551 with base parameters @xmath552 .
following corollary [ aut ] , the action of the @xmath249 involutions on the local coordinates at distinguished points is as follows .
the involution @xmath542 maps @xmath553 to @xmath554 and fixes the remaining coordinates .
the involution @xmath543 fixes @xmath553 , maps @xmath555 to @xmath556 , @xmath557 to @xmath558 and @xmath559 to @xmath560 .
it follows immediately that the involutions each act on the @xmath551 parameters by the map @xmath561 .
we describe sections of the @xmath249 dualizing sheaf square and following the discussion after lemma [ cotfib ] , describe the extension of the sections by the linear map @xmath562 to sections of the relative dualizing sheaf of @xmath550 over @xmath551 . given the action of the involutions on the @xmath563 , @xmath110 local coordinates , by corollary [ aut ] the involutions @xmath542 and @xmath543 stabilize the linear map @xmath562 . to define sections of the @xmath249 dualizing sheaf square ,
begin with the @xmath356 node and consider @xmath564 . by construction , the section has residue @xmath324 at @xmath540 and residue @xmath112 at @xmath541 with simple poles at the points @xmath565 .
the section is invariant by the involutions @xmath542 and @xmath543 .
we write @xmath566 for the extension @xmath567 to a section of the relative dualizing sheaf @xmath568 with @xmath569 and @xmath570 .
the section @xmath566 is invariant by the involutions acting on @xmath550 , since the dualizing sheaf section is invariant and the involutions stabilize the map @xmath562 .
next consider the @xmath84 node and the section @xmath571 extended by zero on @xmath356 . by construction , the section has residue @xmath112 at @xmath540 and @xmath324 at @xmath541 .
the section is invariant by the involutions .
we write @xmath572 for the extension @xmath573 to a section of the relative dualizing sheaf @xmath568 with @xmath574 and @xmath575 .
similar to @xmath566 , the section @xmath572 is invariant by the involutions acting on @xmath550 . finally consider the section @xmath576 on @xmath356 , extended by zero on @xmath84 .
the section is invariant by the involutions .
we write @xmath577 for the extension @xmath578 to a section of the relative dualizing sheaf @xmath568 with vanishing @xmath178 coefficients .
similar to @xmath566 and @xmath572 , the section @xmath577 is invariant by the involutions acting on @xmath550 .
we consider the expansions of the sections @xmath579 at nodes as examples of our overall considerations . for a node @xmath580 and a section @xmath48 of @xmath581 given as @xmath582 @xmath261 representing the remaining variables ,
the interpretation of initial derivatives of @xmath583 is as follows .
the vanishing residue condition is @xmath584 .
the derivatives @xmath585 and @xmath586 are the coefficients of the terms @xmath587 and @xmath588 in the expansion of @xmath48 on the branches of the normalization of the node .
finally the derivative @xmath589 is the @xmath57-linear term of the function @xmath180 of @xmath57 . at the node @xmath540 ,
the local coordinates for @xmath550 are @xmath590 and @xmath4 . at the node @xmath541 ,
the local coordinates for @xmath550 are @xmath591 and @xmath4 . for the sections @xmath579 , the expansions @xmath592 modulo @xmath593 remainders are as follows . the section @xmath594 has expansion @xmath595 at @xmath540 and @xmath112 at @xmath541 .
the section @xmath596 has expansion @xmath597 at @xmath541 and @xmath112 at @xmath540 .
the section @xmath577 has @xmath112 expansion at each node .
it now follows from lemma [ plumderiv ] that on the domain of sections , the pairing of plumbing tangents with the sections @xmath579 is given as follows : @xmath598 pairs to the values @xmath599 and @xmath600 pairs to the values @xmath601 .
furthermore by the grtzsch and rauch variational formula , for @xmath602 the section @xmath577 represents the cotangent @xmath603 and thus @xmath604 pairs with the sections @xmath579 to the values @xmath605 . for @xmath602 ,
the sections @xmath579 are the coordinate cotangent frame for the base @xmath551 .
combining the @xmath542 and @xmath543 invariance of the sections @xmath606 and the involution action on coordinates @xmath552 gives the expected action @xmath607 on cotangent representatives . by theorem [ dm ] , the @xmath181 family @xmath550 over @xmath551 with @xmath249 automorphism action provides a local analytic description for the deligne - mumford compactification .
the action of the involutions follows general expectations , ( * ? ? ?
* chapter 11 , proposition 4.11 ) .
the product @xmath608 acts on the fiber @xmath249 as the limit of hyperelliptic involutions and for genus @xmath56 the action is everywhere trivial .
the action of a single involution is the standard _
half dehn twist for a @xmath3-handle _ , or equivalently the _ @xmath56-torsion associated with an elliptic tail .
_ we apply a standard construction for plumbing an abelian differential and then calculate the variation of its period .
the plumbing provides an example of a section of the direct image of @xmath75 as a cotangent sheaf section and a demonstration for rauch s period variation formula @xcite .
we calculate the period variation by two approaches and then compare .
begin with a compact riemann surface @xmath182 with a canonical homology basis @xmath609 , given by representative cycles and with points @xmath610 disjoint from the cycles . by riemann - roch
there is a meromorphic abelian differential with residue @xmath527 at @xmath510 and @xmath3 at @xmath537 @xcite .
the meromorphic differential is unique modulo the analytic differentials .
a unique differential @xmath611 of the third kind is determined by the condition of vanishing periods on the given homology basis .
introduce local coordinates @xmath563 at @xmath537 and @xmath110 at @xmath510 , such that @xmath612 , and such that the charts include the discs @xmath613 .
coordinates @xmath614 are analytically specified by the condition that on the discs , @xmath611 is respectively given as @xmath615 and as @xmath616 .
we can use the coordinates to describe points near @xmath537 or @xmath510 . introduce an arc @xmath617 from @xmath618 to @xmath619 , disjoint from the basis of cycles and contained in @xmath620 .
the plumbing of @xmath182 is as follows .
given @xmath57 nonzero , remove the closed discs @xmath621 and @xmath622 . overlap the annuli @xmath623 and @xmath624 by @xmath580 to obtain the plumbed surface @xmath625 . since the identification is @xmath626 , the differential @xmath611 plumbs to a differential @xmath627 on @xmath625 .
introduce an oriented arc @xmath429 from @xmath3 to @xmath57 in the annulus @xmath628 and disjoint from @xmath629 .
the arc concatenation @xmath630 determines an oriented cycle on @xmath625 , disjoint from the original basis of cycles .
the image of the positively oriented circle @xmath631 also defines a cycle @xmath632 . by construction
the integral of @xmath627 over @xmath632 is unity and the genus of @xmath625 is one greater than the genus of @xmath182 .
the combination of images of cycles @xmath633 is a canonical homology basis for @xmath625 .
the abelian differential @xmath627 is the dual to @xmath632 relative to the @xmath168 cycles of the basis .
the period @xmath634 is an entry in the riemann period matrix of @xmath625 .
we set @xmath635 and from the above construction @xmath636 .
the exponential of the period @xmath637 extends to an analytic function on the disc @xmath638 .
the derivative of the function is @xmath639 .
we can also calculate the derivative by combining rauch s variational formula ( * ? ? ?
* theorem 2 ) and lemma [ plumderiv ] .
preliminary observations are necessary .
first , by construction in a neighborhood of the plumbing , the @xmath57 family @xmath625 coincides with the family @xmath26 .
second , @xmath625 locally embeds into a fiber of @xmath26 by the map @xmath640 and consequently @xmath641 is the pullback of @xmath642 on @xmath29 .
third , rauch s formula is given in terms of the @xmath56-form @xmath643 for the beltrami - quadratic differential pairing , whereas the present formulas are given in terms of the @xmath56-form @xmath197 ; see ( [ serrepair ] ) . applying the period variation formula , lemma [ plumderiv ] and the observations gives @xmath644 in a neighborhood of the plumbing , @xmath627 is given as @xmath645 and @xmath646 combining contributions , the derivative of the exponential of the period is @xmath647 , matching the direct calculation . the derivative formula can be expressed as the differential of a function by @xmath648 .
the family @xmath625 is one dimensional with coordinate @xmath649 with @xmath650 a constant .
the derivative formula can also be given as @xmath651 , with the right hand side a second frame element , in particular a section of the direct image sheaf @xmath74 .
enrico arbarello , maurizio cornalba , and phillip a. griffiths . ,
volume 268 of _ grundlehren der mathematischen wissenschaften [ fundamental principles of mathematical sciences]_. springer , heidelberg , 2011 . with a contribution by joseph daniel harris .
e. arbarello , m. cornalba , p. a. griffiths , and j. harris . ,
volume 267 of _ grundlehren der mathematischen wissenschaften [ fundamental principles of mathematical sciences]_. springer - verlag , new york , 1985 .
lipman bers .
spaces of degenerating riemann surfaces . in _ discontinuous groups and riemann surfaces ( proc .
maryland , college park , md . , 1973 )
_ , pages 4355 .
ann . of math .
studies , no .
princeton univ . press ,
princeton , n.j .
, 1974 .
scott a. wolpert .
eometry of the weil - petersson completion of teichmller space . in _
surveys in differential geometry viii : papers in honor of calabi , lawson , siu and uhlenbeck _ , pages 357393 .
press , cambridge , ma , 2003 . | an analytic approach and description are presented for the moduli cotangent sheaf for suitable stable curve families including noded fibers . for sections of the square of the relative dualizing sheaf ,
the residue map at a node gives rise to an exact sequence .
the residue kernel defines the vanishing residue subsheaf . for suitable stable curve families ,
the direct image sheaf on the base is locally free and the sequence of direct images is exact .
recent work of hubbard - koch and a formal argument provide that the direct image sheaf is naturally identified with the moduli cotangent sheaf .
the result generalizes the role of holomorphic quadratic differentials as cotangents for smooth curve families .
formulas are developed for the pairing of an infinitesimal opening of a node and a section of the direct image sheaf .
applications include an analytic description of the conormal sheaf for the locus of noded stable curves and a formula comparing infinitesimal openings of a node .
the moduli action of the automorphism group of a stable curve is described .
an example of plumbing an abelian differential and the corresponding period variation is presented . |
the sheer size of the many - body hilbert space makes treating strongly correlated systems adequately extremely difficult or even impossible .
examples of considerable interest are the superconducting doped fullerenes . even for a single c@xmath1 molecule
a full many - body calculation is still a challenge and calculations for solids made of fullerenes are simply out of question . in this situation
we are forced to restrict our attention to only the most relevant degrees of freedom .
for the doped fullerenes these are the electrons in the @xmath0-band .
starting from _ ab initio _ density functional calculations we set up a tight - binding hamiltonian that describes the electrons in the @xmath0-band only . including correlation effects we arrive at a generalized hubbard hamiltonian that can be treated by quantum monte carlo ( qmc ) . using the fixed - node approximation we study the screening of an point charge by the electrons in the @xmath0-band .
we find that the screening is surprisingly efficient even for strong correlations , almost up to the mott transition .
this has important implications for superconductivity in the doped fullerenes .
given that molecular vibration energies are of the same order as electronic energies , retardation effects are inefficient at reducing the electron - electron repulsion .
it is therefore not clear how the weak electron - phonon attraction can lead to superconductivity .
efficient metallic screening , as found in our calculations , can , however , reduce the electron - electron repulsion enough to allow for an electron - phonon driven superconductivity .
but the screening does , of course , also affect the coupling to the phonons .
it turns out that , due to screening , the alkali and @xmath2 modes couple only weakly , while the coupling to the @xmath3 modes is not affected .
therefore , although being electron - phonon driven , superconductivity in the doped fullerenes differs in important ways from the conventional picture of superconductivity .
treating correlations in the fullerenes is quite difficult . even for a single c@xmath1 molecule
a full many - body calculation is still a challenge and simulations of fullerenes , which are solids made of c@xmath1 molecules are well beyond current computational capabilities .
solid c@xmath1 is characterized by weak inter - molecular interactions .
hence the molecular levels merely broaden into narrow , well separated bands @xcite . doping the solid with alkali metals has the effect of filling the band originating from the molecular @xmath0-level with the weakly bound valence electrons of the alkali atoms .
since the @xmath0-level is three - fold degenerate the corresponding band can hold up to six electrons per molecule ; for a@xmath4c@xmath1 , e.g. , the band is half - filled .
when we are interested in the low - energy properties of the @xmath0 electrons , it is a good approximation to focus only on the region around the fermi level ; i.e. on the @xmath0-band , projecting out all the other bands @xcite . that way we arrive at a tight - binding hamiltonian comprising only the @xmath0 orbitals , that reproduces the _ ab initio _ band structure remarkably well @xcite . to obtain a realistic description we have to include the correlation arising from the coulomb repulsion among the electrons on the same molecule .
the resulting hamiltonian describes the interplay between the hopping of electrons between different molecules and the coulomb repulsion among the electrons on the same molecule @xmath5 here @xmath6 creates an electron of spin @xmath7 in the orbital with index @xmath8 on molecule @xmath9 , the @xmath10 are hopping matrix elements between orbitals @xmath11 and @xmath12 on neighboring molecules , and @xmath13 are occupation operators .
varying the value of the interaction term @xmath14 , we can study the effect of correlations .
the physical value for the doped fullerenes is @xmath15ev @xcite , which has to be compared to the width of the @xmath0-band , @xmath16ev .
we now give a very brief outline of the quantum monte carlo method for determining ground states .
the basic idea is to use the hamiltonian to project out the ground - state from some trial function @xmath17 , that we have guessed : @xmath18 to see how this works let us assume we knew the expansion of the trial function in eigenfunctions of @xmath19 @xmath20 thus the component with the lowest eigenenergy is damped least during the projection , i.e. , if @xmath21 , in the limit of large @xmath22 the ground - state component will dominate . in practice
we use for hamiltonians @xmath19 with a spectrum that is bounded , both , from below _ and _ above , an iterative projection of the form @xcite @xmath23|\psi^{(n-1)}\rangle , \ ] ] which , for small but finite @xmath22 , also exactly projects to the ground - state .
we see that the basic operation in ( [ qmciter ] ) is a matrix - vector product . since we are working in a many - body hilbert space ,
the dimension of the vectors is , however , in general enormous ; see table [ dimh ] for an illustration . to understand the monte carlo method for doing the iteration we first rewrite ( [ qmciter ] ) in configuration space . here
@xmath24 denotes a configuration of the electrons in real space @xmath25 we see that the matrix @xmath26 maps configuration @xmath24 into configurations @xmath27 .
we clearly can not follow every possible new configuration @xmath27 since that would lead to an exponential growth in the number of configurations as we iterate .
the idea of monte carlo is then to sample _ only one _ of the configurations @xmath27 with a probability @xmath28 . to do that we want to interpret the matrix elements of @xmath26 as probabilities .
they are , however , in general not normalized and can even be negative @xmath29 normalization introduces the need for population control , while negative matrix elements introduce the sign problem @xcite .
rrll @xmath30 & & + + 4 & @xmath31 & @xmath32 & @xmath33 + 8 & @xmath34 & @xmath35 & @xmath36 + 16 & & @xmath37 & @xmath38 + 32 & & @xmath39 & @xmath40 + 64 & & @xmath41 & @xmath42 + 108 & & @xmath43 & @xmath44 +
in conventional superconductors the electron - phonon interaction leads to an effective electron - electron attraction .
this attraction is , of course , counteracted by the coulomb repulsion between the electrons . in the conventional picture this repulsion is , however , strongly reduced by retardation effects @xcite .
the resulting effective coulomb repulsion is described by the dimensionless coulomb pseudopotential @xmath45 , which is believed to be of the order of @xmath46 . for the doped fullerenes the situation is different : retardation effects are inefficient @xcite .
therefore the screening of the coulomb interaction becomes important for reducing the electron - electron repulsion .
assuming that the random phase approximation ( rpa ) is valid for the electrons within the @xmath0-band it was found that efficient metallic screening significantly reduces the coulomb pseudopotential @xcite . in this scenario
the coulomb pseudopotential @xmath47 is substantially larger than that for conventional superconductors , but it is not too large to prevent superconductivity from being driven by the electron - phonon interaction . for strongly correlated systems like the doped fullerenes
the use of the rpa seems , however , highly questionable . to address this question ,
we have investigated how well the rpa describes the screening within the @xmath0-band .
it is clear that the rpa properly describes the screening when the kinetic energy is much larger than the interaction energy , i.e. rpa works well in the weakly correlated limit . for strong correlations ,
where the coulomb energy dominates , the rpa gives qualitatively wrong results : introducing a test charge @xmath48 on a molecule , rpa predicts that almost the same amount of electronic charge moves away form that molecule , since for a coulomb integral @xmath14 much larger than the band width @xmath49 the gain in potential energy dominates the cost in kinetic energy .
the rpa neglects , however , that in this limit , when an electron leaves a molecule it has to find another molecule with a missing electron , or there will be a large increase in coulomb energy .
it is not clear for what value of the coulomb interaction @xmath14 this qualitative breakdown of the rpa starts , and up to which values of @xmath14 rpa gives still accurate results . to find out we have performed quantum monte carlo calculations . to study the screening of a point charge @xmath48 on molecule with index @xmath50 we consider the hamiltonian @xmath51 which differs from ( [ hamil ] ) only by the last term describing the interaction with the test charge . as a trial function
we use a generalized gutzwiller function of the form @xmath52 where @xmath53 is a slater determinant , @xmath54 is the number of double occupancies in the system , and @xmath55 is the number of electrons on the molecule with the test charge . calculating the expectation value @xmath56 by variational monte carlo and the mixed estimator @xmath57 by fixed - node diffusion monte carlo , we obtain the ground - state expectation value @xmath58 from the extrapolated estimator @xmath59 . to estimate the accuracy of our approach we have compared the results of the quantum monte carlo ( qmc ) calculations with the results from exact diagonalization for a cluster of 4 molecules .
we find that the qmc calculations are accurate up to very large values of the coulomb interaction @xmath14 .
performing qmc calculations for clusters of sizes @xmath60 32 , 48 , 64 , 72 , and 108 molecules , where exact diagonalization is not possible ( cf .
table [ dimh ] ) , we have extrapolated the screening charge @xmath61 to infinite cluster - size , assuming a finite - size scaling of the form @xmath62 @xcite .
the finite - size extrapolation gives only a small correction to the screening charge found for the larger clusters .
results for @xmath63 are shown in figure [ scr ] . for small values of @xmath14
the rpa somewhat underestimates the screening , a behavior also found in the electron gas @xcite . for intermediate values of @xmath14 ( @xmath64 )
the rpa still gives surprisingly accurate results , while for larger @xmath14 it rapidly becomes qualitatively wrong , as discussed above .
we thus find efficient , rpa - like screening even for quite strong correlations close to the mott transition .
as pointed out in the preceeding discussion , efficient metallic screening helps to reduce the effective electron - electron repulsion , i.e. the coulomb pseudopotential .
but the screening also affects the electron - phonon coupling .
at first it might appear that efficient screening is not really helpful for superconductivity .
phonons couple to the electrons by perturbing the potential seen by the electrons .
an example is the longitudinal mode of a jellium .
efficient screening tends to weaken the coupling to such phonons , since it reduces the perturbation considerably : the bare coupling constant @xmath65 is reduced to @xmath66 , where @xmath67 is the dielectric constant @xcite . to some extent , such a reduction is also at work in the fullerenes .
an example are the alkali phonons .
each c@xmath1 molecule is surrounded by 14 alkali ions that are bound with quite weak force constants .
they should therefore respond strongly when an electron arrives on a c@xmath1 molecule .
this was , however , not confirmed experimentally ; an alkali isotope effect could , e.g. , not be observed @xcite . given the efficient metallic screening this finding can be naturally understood .
when an electron arrives on a c@xmath1 molecule , other electrons leave the molecule , which thus stays almost neutral .
the alkali ions then see only a small change in the net charge and therefore couple weakly .
to analyze the situation more closely , we consider the change in the energy of the molecular orbital @xmath68 under a deformation of the molecule with amplitude @xmath69 @xmath70 where @xmath71 is the electron - phonon coupling .
an illustration is given in figure [ elphc ] .
the change of the on - site energy @xmath72 will induce a response of the @xmath0 electrons . since the effect of a point charge @xmath48 is just to shift the on - site energy by @xmath73 ( cf .
( [ hamilq ] ) ) , the screening charge induced by the change @xmath74 is given by @xmath75 where @xmath76 measures the efficiency of the screening : @xmath77 .
the total screening charge induced by the molecular deformation @xmath69 is then given by @xmath78 including screening , the shift in the molecular levels is then given by @xmath79 for molecular solids like the doped fullerenes the electron - phonon coupling is given by the dimensionless constant @xcite @xmath80 given efficient metallic screening @xmath81 , the coupling to phonons that cause a net shift of the molecular levels @xmath82 will be reduced , while modes that leave the center of gravity of the molecular levels unchanged @xmath83 will not be affected .
such modes are the @xmath3 modes in c@xmath1 . for these modes efficient
screening serves to reduce the electron - electron repulsion without affecting the electron - phonon coupling .
we have described how to construct a model for the @xmath0 electrons in the doped fullerenes , that can be analyzed by many body techniques . using quantum monte carlo
we have calculated the static screening of a point charge .
we find that the rpa works surprisingly well , almost up to the mott transition .
the metallic screening helps to reduce the electron - electron repulsion in the doped fullerenes , where retardation effects are inefficient .
but the screening in general also reduces the electron - phonon coupling . in a molecular solid there can be , however , intra - molecular modes that are not screened , examples being the @xmath3 modes in the fullerenes .
we thus find that , although superconductivity in the fullerenes is driven by the electron - phonon coupling , it differs in important points from the textbook picture of superconductivity .
erwin , electronic structure of the alkali - intercalated fullerides , endohedral fullerenes , and metal - adsorbed fullerenes , in w.e .
billups and m.a .
ciufolini ( eds . ) , _ buckminsterfullerenes _ ( vch , new york , 1993 ) , pp . 217 - 253 .
lwdin , _ j. chem .
_ 19 ( 1951 ) 1396 .
e. koch , o. gunnarsson , and r.m .
martin , fixed - node dmc for fermions on a lattice : application to the doped fullerides , in m.p .
nightingale and c.j .
umrigar ( eds . ) , _ quantum monte carlo methods in physics and chemistry _ ( kluwer , 1999 ) , pp
. 447 - 462 .
brhwiler , a.j .
maxwell , a. nilsson , n. mrtensson , and o. gunnarsson , _ phys .
b _ 48 ( 1993 ) 18296 ; r.w .
lof , m.a .
van veenendaal , b. koopmans , h.t .
jonkman , and g.a .
sawatzky , _ phys .
_ 68 ( 1992 ) 3924 .
n. trivedi and d.m .
ceperley , _ phys .
b _ 40 ( 1989 ) 2737 .
ten haaf , h.j.m .
van bemmel , j.m.j .
van leeuwen , and w. van saarloos , _ phys .
_ 72 ( 1994 ) 2442 ; d.f.b .
ten haaf , h.j.m .
van bemmel , j.m.j .
van leeuwen , w. van saarloos , and d.m .
ceperley , _ phys .
b _ 51 ( 1995 ) 13039 ; e. koch , o. gunnarsson , and r.m .
martin , _ phys .
b _ 59 ( 1999 ) 15632 . for
recent applications of fixed - node monte carlo see , e.g. : m. boninsegni , _ phys .
rev . b _ 52 ( 1995 ) 15304 ; e. koch , o. gunnarsson , and r.m .
martin , _ phys .
b _ 54 ( 1996 ) r11026 ; f. aryasetiawan , o. gunnarsson , e. koch , and r.m .
martin , _ phys .
rev . b _ 55 ( 1997 ) r10165 ; m. calandra and s. sorella , _ phys .
b _ 57 ( 1998 ) 11446 ; a.c .
cosentini , m. capone , l. guidoni , and g.b .
bachelet , _ phys .
b _ 58 ( 1998 ) r14685 ; e. koch , o. gunnarsson , and r.m .
martin , _ phys .
b _ 60 ( 1999 ) .
schrieffer , _ theory of superconductivity _
( benjamin , new york , 1964 ) ; d.j .
scalapino , the electron - phonon interaction and strong - coupling superconductors , in r.d . park ( ed . ) , _ superconductivity _ ( dekker , new york , 1969 ) ; v.l .
ginzburg and d.a .
kirzhnits , _ high - temperature superconductivity _ ( consultants bureau , new york , 1982 ) .
o. gunnarsson , _ rev .
_ 69 ( 1997 ) 575 .
o. gunnarsson and g. zwicknagl , _ phys .
_ 69 ( 1992 ) 957 ; o. gunnarsson , d. rainer , and g. zwicknagl , _ int . j. of mod .
b _ 6 ( 1992 ) 3993 .
e. koch , o. gunnarsson , and r.m .
martin , _ phys . rev
. lett .
_ 83 ( 1999 ) 620 .
l. hedin and s. lundqvist , effects of electron - electron and electron - phonon interactions on the one - electron states of solids , in h. ehrenreich , d. turnbull , and f. seitz ( eds . ) , _ solid state physics _ , vol .
23 ( academic press , new york , 1969 ) .
ebbesen , j.s .
tsai , k. tanigaki , h. hiura , y. shimakawa , y. kubo , i. hirosawa , and j. mizuki , _ physica c _ 203 ( 1992 ) 163 ; b. burk , v.h .
crespi , a. zettl , and m.l .
cohen , _ phys .
_ 72 ( 1994 ) 3706 .
m. lanoo , g.a .
baraff , m. schluter , and d. tomanek , _ phys .
. b _ 44 ( 1991 ) 12106 ; d. rainer , _ j. low temp . phys .
_ 10 ( 1986 ) 371 . | we use the doped fullerenes as an example of how realistic systems can be described by simple models . starting from the band structure we set up a tight - binding model that describes the @xmath0 conduction band . adding correlation terms we arrive at a generalized hubbard hamiltonian that we treat using quantum monte carlo . to address the problem of superconductivity in the doped fullerenes , we study the screening of a point charge .
we find surprisingly efficient metallic screening even for strong correlations , almost up to the mott transition , and discuss the implications on superconductivity , in particular the effect of the efficient screening on the coulomb pseudopotential and the electron - phonon coupling . and superconductivity , coulomb pseudopotential , electron - phonon coupling , hubbard model , orbital degeneracy , fullerenes , quantum monte carlo pacs : 74.70.wz 71.10.fd |
in a number of physical problems one needs to consider the model in the background of manifolds with boundaries on which the dynamical variables obey some prescribed boundary conditions . in quantum field theory , the imposition of boundary conditions on the field operator gives rise to a number of physical consequences .
the casimir effect is among the most interesting phenomena of this kind ( for reviews see @xcite ) .
it arises due to the modification of the quantum fluctuations of a field by boundary conditions and plays an important role in different fields of physics , from microworld to cosmology .
the boundary conditions in the casimir effect may have different physical natures and can be divided into two main classes . in the first one ,
the constraints are induced by the presence of boundaries , like macroscopic bodies in qed , interfaces separating different phases of a physical system , extended topological defects , horizons in gravitational physics , branes in high - energy theories with extra dimensions and in string theories . in the corresponding models
the field operator obeys the boundary condition on some spacelike surfaces ( static or dynamical ) .
the original problem with two conducting plates , discussed by casimir in 1948 @xcite , belongs to this class . since the original research by casimir
, many theoretical and experimental works have been done on this problem for various types of bulk and boundary geometries .
different methods have been developed including direct mode - summation and the zeta function techniques , semiclassical methods , the optical approach , worldline numerics , the path integral approach , methods based on scattering theory , and numerical methods based on evaluation of the stress tensor via the fluctuation - dissipation theorem .
the recent high precision measurements of the casimir force allow for an accurate comparison between the experimental results and theoretical predictions . in the second class , the boundary conditions on the field operator
are induced by the nontrivial topology of the space .
the changes in the properties of the vacuum state generated by this type of conditions are referred to as the topological casimir effect .
the importance of this effect is motivated by that the presence of compact dimensions is an inherent feature in many high - energy theories of fundamental physics , in cosmology and in condensed matter physics .
in particular , supergravity and superstring theories are formulated in spacetimes having extra compact dimensions .
the compactified higher - dimensional models provide a possibility for the unification of known interactions . models of a compact universe with nontrivial topology may also play an important role by providing proper initial conditions for inflation in the early stages of the universe expansion @xcite . in condensed matter physics , a number of planar systems in the low - energy sector are described by an effective field theory .
the compactification of these systems leads to the change in the ground state energy which is the analog of the topological casimir effect .
a well - known example of this type of systems is a graphene sheet . in the long wavelength limit
, the dynamics of the quasiparticles for the electronic subsystem is described in terms of the dirac - like theory in two - dimensional space ( see ref .
@xcite ) . the corresponding effective 3-dimensional relativistic field theory , in addition to dirac fermions , involves scalar and gauge fields ( see @xcite and references therein ) .
the single - walled carbon nanotubes are generated by rolling up a graphene sheet to form a cylinder and for the corresponding dirac model one has the spatial topology @xmath0 .
for another class of graphene - made structures , called toroidal carbon nanotubes , the background topology is a 2-dimensional torus , @xmath1 .
many authors have investigated the casimir energies and stresses associated with the presence of compact dimensions ( for reviews see refs .
most97,duff86,khan14 ) . in higher - dimensional models
the casimir energy of bulk fields induces an effective potential for the compactification radius .
this has been used as a stabilization mechanism for the corresponding moduli fields and as a source for dynamical compactification of the extra dimensions during the cosmological evolution .
the casimir effect has also been considered as a possible origin for the dark energy in both kaluza - klein - type and braneworld models @xcite .
extra - dimensional theories with low - energy compactification scale predict yukawa - type corrections to newton s gravitational law and the measurements of the casimir forces between macroscopic bodies provide a sensitive test for constraining the parameters of the corresponding long - range interactions @xcite .
the influence of extra compactified dimensions on the casimir effect in the classical configuration of two parallel plates has been recently discussed for scalar @xcite , electromagnetic @xcite and fermionic @xcite fields .
the vast majority of the works on the influence of the copmactification on the properties of the quantum vacuum in the casimir effect has been concerned with global quantities such as the force or the total energy .
more detailed information on the vacuum fluctuations is contained in the local characteristics . among the most important local quantities , because of their close connection with the structure of spacetime , are the vacuum expectation values ( vevs ) of the vacuum energy density and stresses .
for charged fields , another important characteristic is the vev of the current density . due to the global nature of the vacuum
, this vev carries information on both global and local properties of the vacuum state . besides , the vev of the current density appears as a source of the electromagnetic field in semiclassical maxwell equations , and ,
hence , it is needed in modeling a self - consistent dynamics involving the electromagnetic field . in models with nontrivial topology
, the nonzero current densities in the vacuum state may appear as a consequence of quasiperiodicity conditions along compact dimensions or by the presence of gauge field fluxes enclosed by these dimensions .
note that the gauge field fluxes in higher - dimensional models will also generate a potential for moduli fields and this provides another mechanism for moduli stabilization ( for a review see @xcite ) .
the vev of the fermionic current density in spaces with toroidally compactified dimensions has been considered in @xcite . in the special case of a 2-dimensional space ,
application are given to the electrons in cylindrical and toroidal carbon nanotubes , described within the framework of the effective field theory in terms of dirac fermions .
the vacuum currents for charged fields in de sitter and anti - de sitter spacetimes with toroidally compact spatial dimensions are investigated in bell13,beze15 .
finite temperature effects on the charge density and on the current densities along compact dimensions have been discussed in beze13 and @xcite for scalar and fermionic fields , respectively .
the changes in the fermionic vacuum currents induced by the presence of parallel plane boundaries , with the bag boundary conditions on them , are investigated in @xcite . in the present paper
we consider the effect of two parallel plane boundaries on the vacuum expectation value of the current density for a charged scalar field in background spacetime with spatial topology @xmath2 , where @xmath3 stands for a @xmath4-dimensional torus .
the organization of the paper is as follows . in the next section the geometry of the problem
is described and the hadamard function is evaluated in the region between the plates for general robin boundary conditions . by using the expression for the hadamard function , in section [ sec:1plate ] , we evaluate the current density in the geometry of a single plate . the corresponding asymptotics are discussed in various limiting cases and numerical results are presented . in section [ sec:2plates ] the current density is investigated in the region between two plates .
the main results of the paper are summarized in section [ sec : conc ] .
an alternative representation of the hadamard function is given in appendix .
we consider @xmath5-dimensional flat spacetime with spatial topology @xmath6 , @xmath7 ( for a review of quantum field - theoretical effects in toroidal topology see ref .
the set of cartesian coordinates in the subspace @xmath8 will be denoted by @xmath9 and the corresponding coordinates on the torus by @xmath10 . if @xmath11 is the length of the @xmath12th compact dimension then one has @xmath13 for @xmath14 , and @xmath15 for @xmath16 .
our main interest in this paper is the vev of the current density for a quantum scalar field @xmath17 with the mass @xmath18 and charge @xmath19 .
the equation for the field operator reads@xmath20where @xmath21 , @xmath22 and @xmath23 is the vector potential for a classical gauge field .
we assume the presence of two parallel flat boundaries placed at @xmath24 and @xmath25 , on which the field obeys robin boundary conditions @xmath26with constant coefficients @xmath27 , @xmath28 , and with @xmath29 being the inward pointing normal to the boundary at @xmath30 . here , for the further convenience we have introduced a special notation @xmath31 for the @xmath32th spatial dimension .
note that robin boundary conditions in the form ( [ rob ] ) are gauge invariant ( for the discussion of various types of gauge invariant boundary conditions see @xcite ) . in what follows
we will consider the region between the plates , @xmath33 .
for this region one has @xmath34 .
the expressions for the vevs in the regions @xmath35 and @xmath36 are obtained by the limiting transitions .
the results for dirichlet and neumann boundary conditions are obtained from those for the condition ( [ rob ] ) in the limits @xmath37 and @xmath38 , @xmath39 , respectively .
robin type conditions appear in a variety of situations , including the considerations of vacuum effects for a confined charged scalar field in external fields @xcite , gauge field theories , quantum gravity and supergravity espo97,luck91 , braneworld models @xcite and in a class of models with boundaries separating the spatial regions with different gravitational backgrounds @xcite . in some geometries
, these conditions may be useful for depicting the finite penetration of the field into the boundary with the `` skin - depth '' parameter related to the coefficient @xmath27 .
it is interesting to note that the quantum scalar field constrained by robin condition on the boundary of cavity violates the bekenstein s entropy - to - energy bound near certain points in the space of the parameter @xmath40 @xcite .
in addition to the boundary conditions on the plates , for the theory to be completely defined , we should also specify the periodicity conditions along the compact dimensions .
different conditions correspond to topologically inequivalent field configurations @xcite . here
, we consider generic quasiperiodicity conditions,@xmath41with constant phases @xmath42 , @xmath43 .
the special cases of the condition ( [ percond ] ) with @xmath44 and @xmath45 correspond to the most frequently discussed cases of untwisted and twisted scalar fields , respectively . as it will be seen below , one of the effects of nontrivial phases in ( [ percond ] ) is the appearance of nonzero vacuum currents along compact dimensions ( for a discussion of physical effects of phases in periodicity conditions along compact dimensions see @xcite and references therein ) . for a scalar field ,
the operator of the current density is given by the expression @xmath46 , \label{jl}\]]@xmath47 .
its vev is obtained from the hadamard function @xmath48with @xmath49 being the vacuum state , by using the formula @xmath50 in the discussion below we will assume a constant gauge field @xmath23 .
though the corresponding field strength vanishes , the nontrivial topology of the background spacetime leads to the aharonov - bohm - like effects on physical observables . in the case of a constant gauge field @xmath23 , the latter can be excluded from the field equation and from the expression for the vev of the current density by the gauge transformation @xmath51 , @xmath52 , with the function @xmath53 . in the new gauge one has @xmath54
however , unlike to the case of trivial topology , here the constant vector potential does not completely disappear from the problem .
it appears in the periodicity conditions for the new field operator:@xmath55where now the phases are given by the expression@xmath56 in the discussion below we shall consider the problem in the gauge @xmath57 omitting the prime . for this gauge , in ( [ feq ] ) , ( [ rob ] ) , ( [ jl ] ) one has @xmath58 and in the expressions ( [ jl1 ] ) the term with the vector potential is absent . from the discussion above it follows that in the problem at hand the presence of a constant gauge field is equivalent to the shift in the phases of the periodicity conditions along compact dimensions .
the shift in the phase is expressed in terms of the magnetic flux @xmath59 enclosed by the @xmath12th compact dimension as@xmath60where @xmath61 is the flux quantum and @xmath62 is the @xmath12th component of the spatial vector @xmath63 . in the discussion below the physical effects of a constant gauge field will appear through the phases @xmath64 . in particular ,
the vevs of physical observables are periodic functions of these phases with the period @xmath65 . in terms of the magnetic flux
, this corresponds to the periodicity of the vevs , as functions of the magnetic flux , with the period equal to the flux quantum .
for the evaluation of the hadamard function in ( [ jl1 ] ) we shall use the mode - sum formula@xmath66where @xmath67 form a complete set of normalised positive- and negative - energy solutions to the classical field equation obeying the boundary conditions of the model . in the region between the plates , introducing the wave vectors @xmath68 and @xmath69 ,
these mode functions can be written in the form@xmath70 e^{i\mathbf{k}_{\parallel } \cdot \mathbf{x}_{\parallel } \mp i\omega _ { \mathbf{k}}t } , \label{phik}\]]where @xmath71 , @xmath72 , @xmath73 , and @xmath74 stands for the coordinates parallel to the plates . for the momentum components along the dimensions @xmath75 , @xmath76 ,
one has @xmath77 , whereas the components along the compact dimensions are quantized by the periodicity conditions ( [ percond2]):@xmath78with @xmath16 .
we will denote by @xmath79 the smallest value for the energy in the compact subspace , @xmath80 .
assuming that @xmath81 , we have@xmath82this quantity can be considered as the effective mass for the field quanta .
now we should impose on the modes ( [ phik ] ) the boundary conditions ( rob ) with @xmath58 . from the boundary condition on the plate at @xmath83 , for the function @xmath84 in ( [ phik ] ) one
gets @xmath85from the boundary condition on the second plate it follows that the eigenvalues for @xmath86 are solutions of the equation @xmath87where @xmath88and @xmath89 is the separation between the plates .
formula ( eigeq ) can also be written in the form@xmath90unlike to the cases of dirichlet and neumann conditions , for robin boundary condition the eigenvalues of @xmath86 are given implicitly , as solutions of the transcendental equation ( [ eigeq2 ] ) .
this equation has an infinite number of positive roots which will be denoted by @xmath91 , @xmath92 , and for the corresponding eigenvalues of @xmath86 one has @xmath93 . for @xmath94 or @xmath95
there are no other roots in the right - half plane of a complex variable @xmath96 , @xmath97 ( see @xcite ) . in the remaining region of the plane @xmath98 ,
the equation ( eigeq2 ) has purely imaginary roots @xmath99 , @xmath100 .
depending on the values of @xmath101 , the number of @xmath102 can be one or two .
in the presence of purely imaginary roots , under the condition @xmath103 , there are modes of the field for which the energy @xmath104 becomes imaginary .
this would lead to the instability of the vacuum state . in the discussion below
we will assume that @xmath105 .
note that in the corresponding problem on background of spacetime with trivial topology the stability condition is written as @xmath106 .
now , by taking into account that @xmath107 , we conclude that the compactification , in general , enlarges the stability range in the space of parameters of robin boundary conditions .
the coefficient @xmath108 in ( [ phik ] ) is found from the orthonormalization condition @xmath109where the integration over @xmath110 goes in the region between the plates . substituting the functions ( [ phik ] ) , one gets@xmath111\sin ( y)/y\right\ } ^{-1}}{(2\pi ) ^{p}av_{q}\omega _ { \mathbf{k } } } , \label{ck}\]]where @xmath96 is a root of the equation ( [ eigeq2 ] ) and @xmath112 is the volume of the compact subspace .
the function @xmath113 is defined by the relation@xmath114first we shall consider the case when all the roots of ( [ eigeq2 ] ) are real and @xmath91 . having the complete set of normalized mode functions , the mode - sum ( [ g1n ] ) for the hadamard function is written in the form@xmath115 \sin \lambda _ { n } } , \label{g11}\end{aligned}\]]where @xmath116 , @xmath117 , @xmath118 , and @xmath119 , @xmath120 . in ( [ g11 ] ) , the energy for the mode with a given @xmath121 is written as@xmath122and@xmath123here and in what follows we use the notation@xmath124note that @xmath125 and @xmath126 . in ( [ g11 ] ) , the eigenvalues @xmath127 are given implicitly and this expression is not convenient for the evaluation of the vevs . in order to obtain an expression in which the explicit knowledge of @xmath127 is not required
, we apply to the series over @xmath128 the abel - plana - type summation formula @xcite@xmath129\sin \lambda _ { n } } & = & -\frac{% \pi f(0)/2}{1-b_{2}-b_{1}}+\int_{0}^{\infty } duf(u ) \notag \\ & & + i\int_{0}^{\infty } du\frac{f(iu)-f(-iu)}{c_{1}(u)c_{2}(u)e^{2u}-1 } , \label{sumfor}\end{aligned}\]]where , for the further convenience , the notation@xmath130is introduced . in ( [ sumfor ] )
we have assumed that @xmath94 .
the changes in the evaluation procedure in the case @xmath131 will be discussed below .
for the series in ( [ g11 ] ) , we take in the summation formula@xmath132note that @xmath133 and the first term in the right - hand side of ( [ sumfor ] ) is absent .
the use of the summation formula ( [ sumfor ] ) with ( [ flamb ] ) allows us to write the hadamard function in the decomposed form@xmath134where @xmath135 . here
, the part@xmath136comes from the first integral in the right - hand side of ( [ sumfor ] ) and corresponds to the hadamard function in the geometry of a single plate at @xmath137 when the second plate is absent .
this function is further decomposed by taking into account that the part in ( [ gj ] ) coming from the first term in the right - hand side of ( [ gj ] ) , @xmath138is the hadamard function for the boundary - free geometry .
after the integration over the components of the momentum along uncompactified dimensions , this function can be presented in the form@xmath139with the notations @xmath140where @xmath141 is the macdonald function . consequently , the hadamard function in the geometry of a single plate is written as@xmath142where the second term in the right - hand side is induced by the presence of the plate at @xmath30 . for the further transformation of the boundary - induced part in ( [ g1j ] ) we rotate the integration contour over @xmath143 by the angle @xmath144 . in the summation over @xmath145 the integrals over the intervals @xmath146 cancel each other and we get@xmath147this expression is well suited for the investigation of the current density . with the representation ( [ g1jb ] ) , the hadamard function in the region between the plates , given by ( [ g(1 ) ] ) ,
is decomposed into the boundary - free , single plate - induced and second plate - induced contributions .
an alternative expression for the hadamard function is obtained in appendix . in deriving ( [ g(1 ) ] ) and
( [ g1jb ] ) we have assumed that @xmath148 . in the case @xmath149 , the quantum scalar field in the geometry of a single plate at @xmath83 has modes with @xmath150 for which the dependence on the coordinate @xmath110 has the form @xmath151 . in the case
@xmath152 , for a part of these modes the energy is imaginary and the vacuum is unstable . in order to have a stable vacuum , in what follows , for non - dirichlet boundary conditions , we shall assume that @xmath153 and the mode with @xmath154 corresponds to a bound state .
for @xmath149 and in the absence of purely imaginary roots of ( [ eigeq2 ] ) , in the right - hand side of the summation formula ( [ sumfor ] ) the residue terms at @xmath155 should be added ( see @xcite ) .
now the integrand in ( [ g1j ] ) has a simple pole at @xmath156 and after the rotation the contribution of the residue at that pole should be added .
this contribution cancels the additional residue term in the right - hand side of ( [ sumfor ] ) . in the case when the equation ( [ eigeq2 ] ) has purely imaginary roots the corresponding contributions have to be added to the mode - sum ( [ g11 ] ) for the hadamard function . but the corresponding contributions should also be added in the left - hand side of ( [ sumfor ] ) and the further evaluation procedure remains the same .
hence , the expressions ( [ g(1 ) ] ) and ( [ g1jb ] ) are valid for all values of the coefficients in the robin boundary conditions .
the only restrictions come from the stability of the vacuum state : @xmath153 and @xmath157 . in the presence of compact dimensions with @xmath158
one has @xmath107 and these conditions are less restrictive than those in the case of trivial topology .
the current density in the boundary - free geometry is obtained by using the hadamard function ( [ g0b ] ) and has been investigated in @xcite .
the corresponding charge density and the current densities along uncompact dimensions vanish . as it can be seen from ( [ g(1 ) ] ) and
( [ g1jb ] ) , the same holds in the case of the boundary - induced contributions in the vevs .
hence , the only nonzero components correspond to the current density along compact dimensions .
in this section we investigate the vev of the vacuum current density in the geometry of a single plate at @xmath30 .
this vev is obtained with the help of the formula ( [ jl1 ] ) by using the hadamard function from ( g1jb ) .
the component of the vev of the current density along the @xmath12th compact dimension is presented in the decomposed form@xmath159where @xmath160 is the current density in the boundary - free geometry and @xmath161 is the contribution induced by the presence of the plate .
the current density in the boundary - free geometry has been investigated in @xcite and for the completeness we will recall the main results .
the current density is given by the formula@xmath162where @xmath163 , @xmath164 , and @xmath165 .
the current density @xmath160 is an odd periodic function of @xmath64 with the period @xmath166 and an even periodic function of @xmath167 , @xmath168 , with the same period .
this corresponds to the periodicity in the magnetic flux with the period of flux quantum .
an alternative expression for the current density in the boundary - free geometry is given by the formula@xmath169where we have defined the function @xmath170and@xmath171 in the model with a single compact dimension ( @xmath172 ) the representations ( [ jl0a ] ) and ( [ jl0b ] ) are identical . when the length of the @xmath12th compact dimension , @xmath11 , is much larger than the other length scales , the behavior of the current density crucially depends whether the parameter@xmath173is zero or not . for @xmath174 , which is realised for a massless field with @xmath175 , @xmath176 , to the leading order we have @xmath177 in this case , the leading term in the expansion of @xmath178 coincides with the current density in @xmath179-dimensional space with a single compact dimension of the length @xmath11 . for @xmath180 and for large values of @xmath11 one has@xmath181and the current density
is exponentially suppressed . in the opposite limit of small values for @xmath11 , to the leading order
we get@xmath182the leading term does not depend on the mass and on the lengths of the other compact dimensions and coincides with the current density for a massless scalar field in the space with topology @xmath183 .
now we turn to the investigation of the plate - induced contribution in the current density . by using the expression for the corresponding part in the hadamard function from ( [ g1jb ] ) , we get the following expression@xmath184with the notations @xmath185 for the distance from the plate and@xmath186recall that , in order to have a stable vacuum state with @xmath187 , we have assumed that @xmath153 . under this condition ,
the integrand in ( [ jlj(1 ) ] ) is regular everywhere in the integration range .
the integral in ( [ jlj(1 ) ] ) is evaluated in the special cases of dirichlet and neumann boundary conditions with the result @xmath188where the upper and lower signs correspond to dirichlet and neumann boundary conditions , respectively .
note that , in the problem with a fermionic field , obeying the bag boundary condition on the plate , the boundary - induced contribution vanishes for a massless field @xcite .
let us consider the behavior of the plate - induced contribution in asymptotic regions of the parameters . at large distances from the plate , @xmath189 ,
one has @xmath190 . assuming that @xmath191 , the dominant contribution in ( [ jlj(1 ) ] ) comes from the region near the lower limit of the integration and from the term with @xmath192 , @xmath193 .
to the leading order we find@xmath194and the current density is exponentially small .
note that the suppression is exponential for both massive and massless field . for points close to the plate , @xmath195 , in ( [ jlj(1 ) ] ) the contribution of the terms with large values of @xmath196 dominates and this formula is not convenient for the asymptotic analysis and for numerical evaluations . in the case
@xmath197 , an alternative expression is obtained by using the representation ( [ g1alt3 ] ) for the hadamard function .
the first term in the right - hand side of this representation corresponds to the geometry with uncompactified @xmath12th dimension and does not contribute to the current density along that direction . in the geometry of a single plate at @xmath30 the part in the hadamard function induced by the compactification
is given by the first term in the figure braces of ( g1alt3 ) . from this part , by making use of ( [ jl1 ] ) , for the vev of the @xmath198th component of the current density we get@xmath199where we have defined the function@xmath200the part with the first term in the right - side of ( [ gzj ] ) corresponds to the current density in the boundary - free geometry . in this part the integration over @xmath96 is done with the help of the formula@xmath201and one gets the expression ( [ jl0b ] ) .
extracing the boundary - free part , for the plate - induced contribution from ( [ jlalt ] ) we find@xmath202 in the case of single compact dimension one has @xmath172 , @xmath203 , and the corresponding formula for the plate - induced contribution in the current density is obtained from ( [ jl1alt ] ) omitting the summation over @xmath204 and putting @xmath205 .
an important issue in quantum field theory with boundaries is the appearance of surface divergences in the vevs of local physical observables .
examples of the latter are the vevs of the field squared and of the energy density .
these divergences are a consequence of the oversimplification of a model where the physical interactions are replaced by the imposition of boundary conditions for all modes of a fluctuating quantum field .
of course , this is an idealization , as real physical systems can not constrain all the modes ( for a discussion of surface divergences and their physical interpretation see @xcite and references therein ) .
the appearance of divergences in the vevs of physical quantities indicates that a more realistic physical model should be employed for their evaluation on the boundaries .
an important feature , which directly follows from the representation ( [ jl1alt ] ) , is that the vev of the current density is finite on the plate .
this is in sharp contrast with the behavior of the vevs for the field squared and energy - momentum tensor .
the finiteness of the current density on the boundary may be understood from general arguments .
the divergences in local physical observables are determined by the local bulk and boundary geometries .
if we consider the model with the topology @xmath206 with the @xmath12th dimension having the topology @xmath207 , then in this model the @xmath12th component of the current density vanishes by the symmetry .
the compactification of the @xmath12th dimension to @xmath208 does not change both the bulk end boundary local geometries and , hence , does not add new divergences to the vevs compared with the model on @xmath209 .
in deriving ( [ jl1alt ] ) we have assumed that @xmath197 . in the case
@xmath149 the contribution of the bound state should be added to ( [ jl1alt ] ) . for @xmath210 , this contribution
is obtained from the corresponding part in the hadamard function , given by ( gjbound ) , and has the form@xmath211 in what follows for simplicity we shall consider the case @xmath148 .
recall that , the representation ( [ jlj(1 ) ] ) is valid for all values of @xmath27 from the range of the vacuum stability . for dirichlet and neumann boundary conditions ,
after the evaluation of the integral in ( [ jl1alt ] ) by using the formula@xmath212one gets@xmath213where the upper and lower signs correspond to dirichlet and neumann conditions , respectively . for a single compact dimension with the length @xmath214 and with the phase @xmath215 in the periodicity condition for a massless field this gives@xmath216now , combining the expressions ( [ jl0b ] ) and ( [ jl1altdn ] ) , we see that in the case of dirichlet boundary condition the boundary - free and plate - induced parts of the current density cancel each other for @xmath217 and , hence , the total current vanishes on the plate . for neumann condition
the current density on the plate is given by @xmath218note that the normal derivative of the current density on the plate vanishes for both dirichlet and neumann boundary conditions : @xmath219 .
this is not the case for general robin condition .
let us consider the behavior of the plate - induced contribution in the current density in the limit @xmath220 . in this investigation
it is more convenient to use the representation ( [ jl1alt ] ) . for @xmath221 ,
the dominant contribution in the integral of ( [ jl1alt ] ) comes from the region near the lower limit of the integration and from the term @xmath222 , @xmath192 , @xmath223 , in the summation .
the argument of the function @xmath224 in the integrand is large and we can use the asymptotic expression @xmath225 . after some intermediate calculations , for the leading term we get@xmath226here ,
we have additionally assumed that @xmath227 for @xmath228 . for @xmath175 , @xmath193 , @xmath176 ,
the dominant contribution in ( [ jl1alt ] ) comes from the term @xmath192 , @xmath223 , with the leading term @xmath229here , @xmath230 is the plate - induced contribution in the current density for @xmath179-dimensional space with topology @xmath231 ( see ( [ jl1alt ] ) for the case @xmath232 and , hence , @xmath205 ) .
if the length of the @xmath233th compact dimension is large , @xmath176 , the dominant contribution to the sum over @xmath234 comes from large values of latexmath:[$% @xmath234 by the integration in accordance with@xmath236the integral over @xmath237 is evaluated by using the formula ( [ intrel ] ) . as a result , from ( [ jl1alt ] ) , to the leading order , we obtain the current density along the @xmath12th compact dimension for the spatial topology @xmath206 with the lengths of the compact dimensions @xmath238 .
now let us consider the limiting case when @xmath11 is large compared with the other length scales in the problem , @xmath239 , @xmath176 .
the dominant contribution in ( [ jl1alt ] ) comes from the term @xmath192 , @xmath240 .
for @xmath180 we find@xmath241where , for non - neumann boundary conditions ( @xmath242 ) , we have assumed that @xmath243 . for @xmath174
the leading term is given by the expression@xmath244comparing with the corresponding asymptotics ( [ jl0llarge ] ) and ( jl0llargeb ) , we see that for non - neumann boundary conditions , in the both cases @xmath180 and @xmath174 , the leading terms in the boundary - induced and boundary - free parts of the current density cancel each other
. an equivalent representation for the plate - induced current density is obtained from ( [ jl1alt ] ) rotating the integration contour in the complex plane @xmath96 by the angle @xmath245 for the term with @xmath246 and by the angle @xmath247 for the term with @xmath248 .
the integrals over the intervals @xmath249 are cancelled and we find@xmath250where@xmath251and @xmath252 is the bessel function .
the equivalence of the representations ( [ jlj(1 ) ] ) and ( [ jl1alt3 ] ) can also be directly seen by applying to the series over @xmath253 in ( [ jlj(1 ) ] ) the relation@xmath254the latter is a direct consequence of the poisson s resummation formula . after using ( [ sumform3 ] ) in ( [ jlj(1 ) ] ) , we introduce a new integration variable @xmath255 and then pass to polar coordinates in the @xmath256-plane .
the integration over the polar angle is expressed in terms of the bessel function and the representation ( [ jl1alt3 ] ) is obtained .
another expression is obtained by applying to the series over @xmath253 in ( [ jlj(1 ) ] ) the summation formula ( [ abelplan1 ] ) . for the series in ( [ jlj(1 ) ] )
one has @xmath257 and the first integral vanishes . as a result
, the plate - induced part in the vev of the current density is presented as@xmath258{x^{2}+\omega _ { \mathbf{n}_{q-1}}^{2}})-\cos \tilde{\alpha}_{l}% } \notag \\ & & \times \int_{0}^{x}dy\frac{(1-y^{2}\beta _ { j}^{2})\cos \left ( 2yz_{j}\right ) + 2y\beta _ { j}\sin \left ( 2yz_{j}\right ) } { ( 1+y^{2}\beta _ { j}^{2})\left ( x^{2}-y^{2}\right ) ^{(1-p)/2}}. \label{jlj(1)alt}\end{aligned}\]]for dirichlet and neumann boundary conditions we obtain@xmath259{x^{2}+\omega _ { % \mathbf{n}_{q-1}}^{2}})-\cos \tilde{\alpha}_{l}}. \label{jlj(1)altdn}\ ] ] in figure [ fig1 ] , for the simplest kaluza - klein model with a single compact dimension of the length @xmath214 and with the phase @xmath215 ( @xmath260 ) , we have plotted the total current density , @xmath261 , for a massless scalar field in the geometry of a single plate as a function of the distance from the plate and of the phase @xmath215 .
the left / right panel correspond to dirichlet / neumann boundary conditions . as has been already noticed before , in the dirichlet case the total current density vanishes on the plate . [ cols="^,^ " , ]
in the present paper we have investigated the influence of parallel flat boundaries on the vev of the current density for a charged scalar field in a flat spacetime with toroidally compactified spatial dimensions , assuming the presence of a constant gauge field .
the effect of the latter on the current is similar to the aharonov - bohm effect and is caused by the nontrivial topology of the background space . along compact dimensions
we have considered quasiperiodicity conditions with general phases .
the special cases of twisted and untwisted fields are the configurations most frequently discussed in the literature . by a gauge transformation , the problem with a constant gauge field is mapped to the one with zero field , shifting the phases in the periodicity conditions by an amount proportional to the magnetic flux enclosed by a compact dimension in the initial representation of the model . on the plates we employed robin boundary conditions , in general , with different coefficients on the left and right plates .
the robin boundary conditions for bulk fields naturally arise in braneworld scenario and the boundaries considered here may serve as a simple model for the branes .
we considered a free field theory and all the information on the properties of the vacuum state is encoded in two - point functions .
here we chose the hadamard function .
the vev of the current density is obtained from this function in the coincidence limit by using ( [ jl1 ] ) . for the evaluation of the hadamard function
we have employed a direct summation over the complete set of modes . in the region between the plates
the eigenvalues of the momentum component perpendicular to the plates are quantized by the boundary conditions on the plates and are given implicitly , in terms of solutions of the transcendental equation ( [ eigeq2 ] ) .
depending on the values of the robin coefficients , this equation may have purely imaginary solutions @xmath262 . in order to have a stable vacuum with @xmath263 ,
we assume that @xmath105 .
compared to the case of the bulk with trivial topology , this constraint in models with compact dimensions is less restrictive .
the eigenvalues of the momentum components along compact dimensions are quantized by the periodicity conditions and are determined by ( [ kltild ] ) .
the application of the generalized abel - plana formula for the summation over the roots of ( eigeq2 ) allowed us to extract from the hadamard function the part corresponding to the geometry with a single plate and to present the second - plate - induced contribution in the form which does not require the explicit knowledge of the eigenmodes for @xmath86 ( see ( [ g(1 ) ] ) ) .
in addition , the corresponding integrand decays exponentially in the upper limit .
a similar representation , ( [ g1jb ] ) , is obtained for the hadamard function in the geometry of a single plate .
the second term in the right - hand side of this representation is the boundary - induced contribution .
an alternative representation for the hadamard function , ( [ g1alt3 ] ) , is obtained in appendix , by making use of the summation formula ( [ abelplan1 ] ) .
the second term in the right - hand side of this representation is the contribution induced by the compactification of the @xmath12th dimension .
the vevs of the charge density and the components of the current density along uncompact dimensions vanish .
the current density along compact dimensions is a periodic function of the magnetic flux with the period equal to the flux quantum .
the component along the @xmath12th compact dimension is an odd function of the phase @xmath64 and an even function of the remaining phases @xmath264 , @xmath176 .
first we have considered the geometry with a single plate .
the vev of the current density is decomposed into the boundary - free and plate - induced parts .
the boundary - free contribution was investigated in @xcite and we have been mainly concerned with the plate - induced part , given by ( [ jlj(1 ) ] ) . for special cases of dirichlet and neumann boundary conditions
the corresponding expression is simplified to ( [ jlj(1)dn ] ) .
the plate - induced part has opposite signs for dirichlet and neumann conditions . at distances from the plate larger than the lengths of compact dimensions
the asymptotic is described by ( [ jlj(1)far ] ) and the plate - induced contribution is exponentially small . for the investigation of the near - plate asymptotic of the current density
it is more convenient to use the representation ( jl1alt ) for the general robin case and ( [ jl1altdn ] ) for dirichlet and neumann conditions . from these representations
it follows that the current density is finite on the plate .
this property is in sharp contrast with the behavior of the vevs of the field squared and of the energy - momentum tensor which diverge on the plate .
for dirichlet boundary condition the current density vanishes on the plate and for neumann condition its value on the plate is two times larger than the current density in the boundary - free geometry .
the normal derivative of the current density vanishes on the plate for both dirichlet and neumann conditions .
this is not the case for general robin condition .
the behavior of the plate - induced part of the current density along @xmath12th dimension , in the limit when the lengths of the other compact dimensions are much smaller than @xmath11 , crucially depend wether the phases @xmath264 , @xmath176 , are zero or not . for @xmath265 one has @xmath266 and the corresponding asymptotic expression is given by ( jl1smallr ) . in this case the plate - induced contribution is exponentially suppressed .
for @xmath175 , @xmath176 , the leading term in the asymptotic expansion , multiplied by @xmath267 , coincides with the corresponding current density for @xmath179-dimensional space with topology @xmath268 . in the limit when the length of the @xmath12th dimension is much larger than the other length scales of the model ,
the behavior of the plate - induced contribution to the current density is essentially different for the cases @xmath180 and @xmath174 . in the former case
the leading term is given by ( [ jlllarge ] ) and the current density is suppressed by the factor @xmath269 . in the second case , for the leading term one has the expression ( [ jlllargeb ] ) and its behavior , as a function of @xmath11 , is power law . in both cases and for non - neumann
boundary conditions , the leading terms in the boundary - induced and boundary - free parts of the current density cancel each other . for the current density in the region between the plates we have provided various decompositions ( ( [ jl2 ] ) , ( [ jldec ] ) , ( [ jltot ] ) for general robin boundary conditions and ( [ jl2dn ] ) , ( [ jl2dn2 ] ) , ( jldn ) for special cases of dirichlet and neumann conditions ) . in the case of dirichlet
boundary condition the total current vanishes on the plates .
the normal derivative vanishes on the plates for both dirichlet and neumann cases . in the limit when the separation between the plates is smaller than all the length scales in the problem ,
the behavior of the current density is essentially different for non - neumann and neumann boundary conditions . in the former case
, the total current density in the region between the plates tends to zero .
for neumann boundary condition on both plates , for small separations the total current density is dominated by the interference part and it diverges inversely proportional to the separation ( see ( jlsmalla4 ) ) .
the results of the present paper may be applied to kaluza - klein - type models in the presence of branes ( for @xmath270 ) and to planar condensed matter systems ( for @xmath271 ) , described within the framework of an effective field theory .
in particular , in the former case , the vacuum currents along compact dimensions generate magnetic fields in the uncompactified subspace .
the boundaries discussed above can serve as a simple model for the edges of planar systems .
n. a. s. was supported by the state committee of science ministry of education and science ra , within the frame of research project no .
15 rf-009 .
in this section we derive an alternative representation for the hadamard function which is well suited for the investigation of the near - plate asymptotic of the current density .
the starting point is the representation ( [ g11 ] ) .
we apply to the corresponding series over @xmath253 the summation formula @xcite@xmath272f(u ) \notag \\ & & \qquad + i\int_{0}^{\infty } du\,[f(iu)-f(-iu)]\sum_{\lambda = \pm 1}\frac{% g(i\lambda u)}{e^{ul_{l}+i\lambda \tilde{\alpha}_{l}}-1 } , \label{abelplan1}\end{aligned}\]]where @xmath273 is given by ( [ kltild ] ) .
the part in the hadamard function coming from the first term in the right - hand side of ( [ abelplan1 ] ) coincides with the hadamard function for the geometry of two plates in @xmath274-dimensional space with topology @xmath275 and with the lengths of the compact dimensions @xmath276 ( the @xmath12th dimension is uncompactified ) .
we will denote this function by @xmath277 . as a result , under the assumption @xmath197 ,
the hadamard function is decomposed as@xmath278 \sin \lambda _ { n } } \notag \\ & & \times \int_{\omega _ { \mathbf{k}}^{(l)}}^{\infty } du\,\frac{\cosh ( \delta t% \sqrt{u^{2}-\omega _ { \mathbf{k}}^{(l)2}})}{\sqrt{u^{2}-\omega _ { \mathbf{k}% } ^{(l)2}}}\sum_{\lambda = \pm 1}\frac{e^{-\lambda u\delta x^{l}}}{% e^{ul_{l}+i\lambda \tilde{\alpha}_{l}}-1 } , \label{g1alt}\end{aligned}\]]where @xmath279 , @xmath280 , and @xmath281 . here
, the second term in the right - hand side vanishes in the limit @xmath282 and is induced by the compactification of the @xmath12th dimension from @xmath207 to @xmath208 with the length @xmath11 . by making use of the relation@xmath283with@xmath284we
rewrite the formula ( [ g1alt ] ) in the form@xmath285 \sin \lambda _ { n } } , \label{g1alt2}\end{aligned}\]]with @xmath286 .
now , by using the summation formula ( [ sumfor ] ) for the series over @xmath128 we get the final representation@xmath287 in this expression , the part with the first term in the figure braces is the contribution to the hadamard function induced by the compactification of the @xmath12th dimension for the geometry of a single plate at @xmath30 and the part with the second term in the figure braces is induced by the second plate .
note that the contribution of the first term in the right - hand side of ( [ g1alt3 ] ) to current density along the @xmath12th dimension vanishes . in deriving the representation ( [ g1alt3 ] ) we have assumed that @xmath148 . for this case , in the region between the plates , all the eigenvalues for the momentum
@xmath86 are real and in the geometry of a single plate there are no bound states . for @xmath149 , in the application of the summation formula ( [ sumfor ] ) to the series over @xmath128 in ( [ g1alt2 ] ) the contribution from the poles @xmath288 should be added to the right - hand side of ( [ sumfor ] ) .
this contribution comes from the bound state in the geometry of a single plate at @xmath30 . for this bound state
the mode function has the form @xmath289 with @xmath290 . assuming that @xmath291 , the contribution from the bound state to the hadamard function in the geometry of a single plate is given by the expression@xmath292 where @xmath293 is the heaviside unit step function . in the case
@xmath294 the corresponding expression is more complicated .
mostepanenko , n.n .
trunov , _ the casimir effect and its applications _ ( clarendon , oxford , 1997 ) ; g. plunien , b. mller , w. greiner , phys . rep . *
134 * , 87 ( 1986 ) ; e. elizalde , s.d .
odintsov , a. romeo , a.a .
bytsenko , s. zerbini , _ zeta regularization techniques with applications _ ( world scientific , singapore , 1994 ) ; k.a .
milton , _ the casimir effect : physical manifestation of zero - point energy _ ( world scientific , singapore , 2002 ) ; m. bordag , g.l .
klimchitskaya , u. mohideen , v.m .
mostepanenko , _ advances in the casimir effect _ ( oxford university press , oxford , 2009 ) ; _ casimir physics _ , edited by d. dalvit , p. milonni , d. roberts , f. da rosa , lecture notes in physics vol .
834 ( springer - verlag , berlin , 2011 ) . m.j .
duff , b.e.w .
nilsson , c.n .
pope , phys . rep . * 130 * , 1 ( 1986 ) ; r. camporesi , phys .
196 * , 1 ( 1990 ) ; a.a .
bytsenko , g. cognola , l. vanzo , s. zerbini , phys . rep . * 266 * , 1 ( 1996 ) ; a.a .
bytsenko , g. cognola , e. elizalde , v. moretti , s. zerbini , _ analytic aspects of quantum fields _ ( world scientific , singapore , 2003 ) ; e. elizalde , _ ten physical applications of spectral zeta functions _
( springer verlag , 2012 ) .
e. elizalde , phys .
b * 516 * , 143 ( 2001 ) ; c.l .
gardner , phys .
b * 524 * , 21 ( 2002 ) ; k.a .
milton , grav . cosmol . * 9 * , 66 ( 2003 ) ; a.a .
saharian , phys .
d * 70 * , 064026 ( 2004 ) ; e. elizalde , j. phys .
a * 39 * , 6299 ( 2006 ) ; a.a .
saharian , phys .
d * 74 * , 124009 ( 2006 ) ; b. green , j. levin , j. high energy phys .
11 ( 2007 ) 096 ; p. burikham , a. chatrabhuti , p. patcharamaneepakorn , k. pimsamarn , j. high energy phys . 07 ( 2008 ) 013 ; p. chen , nucl .
* 173 * , s8 ( 2009 ) .
mostepanenko , i.yu .
sokolov , phys .
a * 125 * , 405 ( 1987 ) ; j.c .
long , h.w .
chan , j.c .
price , nucl .
b * 539 * , 23 ( 1999 ) ; r.s .
decca , d. lpez , e. fischbach , g.l .
klimchitskaya , d. e. krause , v. m. mostepanenko , ann .
( n.y . ) * 318 * , 37 ( 2005 ) ; phys .
d * 75 * , 077101 ( 2007 ) ; g.l .
klimchitskaya , v.m .
mostepanenko , eur .
j. c * 75 * , 164 ( 2015 ) .
cheng , phys .
b * 643 * , 311 ( 2006 ) ; h.b .
cheng , phys .
b * 668 * , 72 ( 2008 ) ; s.a .
fulling , k. kirsten , phys .
b * 671 * , 179 ( 2009 ) ; k. kirsten , s.a .
fulling , phys .
d * 79 * , 065019 ( 2009 ) ; e. elizalde , s.d .
odintsov , a.a .
saharian , phys .
d * 79 * , 065023 ( 2009 ) ; l.p .
teo , phys .
b * 672 * , 190 ( 2009 ) ; l.p .
teo , nucl .
b * 819 * , 431 ( 2009 ) ; l.p .
teo , j. high energy phys .
11 ( 2009 ) 095 .
k. poppenhaeger , s. hossenfelder , s. hofmann , m. bleicher , phys .
b * 582 * , 1 ( 2004 ) ; a. edery , v.n .
marachevsky , j. high energy phys .
12 ( 2008 ) 035 ; f. pascoal , l.f.a .
oliveira , f.s.s .
rosa , c. farina , braz .
* 38 * , 581 ( 2008 ) ; l. perivolaropoulos , phys . rev .
d * 77 * , 107301 ( 2008 ) ; l.p .
teo , phys .
d * 83 * , 105020 ( 2011 ) .
s. bellucci , a.a .
saharian , a.h .
yeranyan , phys .
d * 89 * , 105006 ( 2014 ) ; s. bellucci , a.a .
saharian , n.a .
saharyan , eur .
j. c * 74 * , 3047 ( 2014 ) ; e.r . bezerra de mello , a.a .
saharian , arxiv:1408.6404 , to appear in int . j. theor
. phys .
j. scherk , j.h .
schwartz , phys .
b * 82 * , 60 ( 1979 ) ; y. hosotani , phys .
b * 126 * , 309 ( 1983 ) ; a. higuchi , l. parker , phys .
d * 37 * , 2853 ( 1988 ) ; y. hosotani , ann . phys .
* 190 * , 233 ( 1989 ) ; a. actor , class .
quantum grav .
* 7 * , 663 ( 1990 ) ; k. kirsten , j. phys . a : math
* 26 * , 2421 ( 1993 ) ; c. ccapa ttira , c.d .
fosco , a.p.c .
malbouisson , i. roditi , phys .
a * 81 * , 032116 ( 2010)*. * a.a .
saharian , _ the generalized abel - plana formula with applications to bessel functions and casimir effect _
( yerevan state university publishing house , yerevan , 2008 ) ; report no . ictp/2007/082 ; arxiv:0708.1187 .
d. deutsch , p. candelas , phys .
d * 20 * , 3063 ( 1979 ) ; p. candelas , ann .
( n.y . ) * 143 * , 241 ( 1982 ) ; v. sopova , l.h .
ford , phys .
d * 66 * , 045026 ( 2002 ) ; n. graham , r.l .
jaffe , v. khemani , m. quandt , m. scandurra , h. weigel , nucl .
b * 645 * , 49 ( 2002 ) ; n. graham , r.l .
jaffe , h.weigel , int .
a * 17 * , 846 ( 2002 ) ; n. graham , r.l .
jaffe , v. khemani , m. quandt , m. scandurra , h. weigel , phys .
b * 572 * , 196 ( 2003 ) ; k.a .
milton , phys .
d * 68 * , 065020 ( 2003 ) ; k.a .
milton , j. phys .
gen . * 37 * , r209 ( 2004 ) ; n. graham , r.l .
jaffe , v. khemani , m. quandt , o. schrder , h. weigel , nucl .
b * 677 * , 379 ( 2004 ) ; k.a .
milton , i. cavero - pelaez , and j. wagner , j. phys .
gen . * 39 * , 6543 ( 2006 ) ; a.a .
sahariana , eur .
phys . j. c * 52 * , 721 ( 2007 ) ; k.a .
milton , phys .
d * 84 * , 065028 ( 2011 ) ; f.d .
mazzitelli , j.p .
nery , a. satz , phys .
d * 84 * , 125008 ( 2011 ) ; r. estrada , s.a .
fulling , l. kaplan , k. kirsten , z. liu , k.a .
milton , j. phys .
theor . * 41 * , 164055 ( 2008 ) ; r. estrada , s.a .
fulling , f.d .
mera , j. phys . a : math
* 45 * , 455402 ( 2012 ) ; n. bartolo , r. passante , phys .
a * 86 * , 012122 ( 2012 ) ; r. passante , l. rizzuto , s. spagnolo , eur .
j. c * 73 * , 2419 ( 2013 ) . | we evaluate the hadamard function and the vacuum expectation value ( vev ) of the current density for a charged scalar field , induced by flat boundaries in spacetimes with an arbitrary number of toroidally compactified spatial dimensions .
the field operator obeys the robin conditions on the boundaries and quasiperiodicity conditions with general phases along compact dimensions .
in addition , the presence of a constant gauge field is assumed .
the latter induces aharonov - bohm - type effect on the vevs .
there is a region in the space of the parameters in robin boundary conditions where the vacuum state becomes unstable .
the stability condition depends on the lengths of compact dimensions and is less restrictive than that for background with trivial topology .
the vacuum current density is a periodic function of the magnetic flux , enclosed by compact dimensions , with the period equal to the flux quantum .
it is explicitly decomposed into the boundary - free and boundary - induced contributions . in sharp contrast to
the vevs of the field squared and the energy - momentum tensor , the current density does not contain surface divergences .
moreover , for dirichlet condition it vanishes on the boundaries .
the normal derivative of the current density on the boundaries vanish for both dirichlet and neumann conditions and is nonzero for general robin conditions . when the separation between the plates is smaller than other length scales , the behavior of the current density is essentially different for non - neumann and neumann boundary conditions . in the former case
, the total current density in the region between the plates tends to zero .
for neumann boundary condition on both plates , the current density is dominated by the interference part and is inversely proportional to the separation .
pacs numbers : 03.70.+k , 11.10.kk , 04.20.gz |
the non - equilibrium nature of the spin glass phase has been extensively investigated since it was discovered in low frequency ac - susceptibility measurements @xcite and dc - relaxation experiments in the early 1980 s @xcite .
the interpretation of the experimental results and the design of new experimental procedures emanate essentially from two different theoretical approaches : hierarchical phase space models @xcite and real space droplet / domain models @xcite .
a high level of phenomenlogical and theoretical insight into the phenomena has by now been acquired .
there remain , however , unresolved problems as to the interpretation and reproducibility of non - equilibrium results .
these shortcomings do in part also prevent a useful judgement in - between the applicability of the different models to real 3d spin glasses . the recently reported memory effect@xcite , observable in low frequency ac - susceptibility experiments , not only elucidates some paradoxal features of the spin glass phase , but an in detail study of related phenomena @xcite also emphasizes the importance of : cooling / heating rates , wait times , thermalisation times etc . , i.e the detailed thermal history of the sample on the results from low frequency ac - susceptibility and dc - magnetic relaxation experiments .
these complicated non - equilibrium phenomena should be considered in the perspective that an ac - experiment at only a couple of decades higher frequency , @xmath1 100 hz ( observation time 1/@xmath2 2 ms ) , in spite of a strong frequency dependence , simply shows an equilibrium character when measured in ordinary ac - susceptometers .
the non - equilibrium processes of the system are only active on observation time scales governed by the cooling / heating rate and during a halt at constant temperature these evolve with the time the sample has been kept at constant temperature .
in this paper we report results from low field zero field cooled ( zfc ) relaxation experiments , i.e. measurements of the response function , at one specific temperature in the spin glass phase .
the parameter we vary in a controlled way is the thermal history in the spin glass phase .
the results imply that the reponse function is governed by : the cooling procedure in a rather narrow region just above the measurement temperature , the wait time before the response function is measured , and the cooling / heating procedure in a rather narrow region just below the measurement temperature if an additional undercooling has been carried out .
it is also clearly shown that the thermal history , in the spin glass phase , at temperatures well enough separated from the measurement temperature is irrelevant to the reponse function at @xmath3 .
the results are put in relation to a phenomenological real space fractal domain picture that is introduced and discussed in more detail elsewhere @xcite .
the investigation is also motivated by a lack of agreement in the detailed behaviour of the non - equilibrium dynamics , when measured on one and the same spin glass material in different magnetometers .
the sample is a bulk piece of ag(mn 11at@xmath0 ) with a spin glass temperature @xmath4 35 k made by drop - synthesis technique .
the experiments were performed in a non - commercial squid magnetometer optimised for dynamic studies in low magnetic fields @xcite . the sample is cooled in zero magnetic field from a temperature above @xmath5 , using a controlled thermal sequence , to the measurement temperature , @xmath3= 27 k , where a weak magnetic field ( @xmath6= 1 oe ) is applied after the system been kept at @xmath3 a certain wait time , @xmath7 .
the relaxation of the magnetisation is then recorded as a function of time elapsed after the field application . in our figures of the relaxation data we show the relaxation rate , i.e. the logarithmic derivative of the response function @xmath8=1/@xmath6 d@xmath9/dlog@xmath10 , which is the quantity that most clearly exposes changes of the response function after the different thermal procedures .
the relaxation rate is related to the out - of - phase component of the ac - susceptibility via @xmath8(@xmath10)@xmath11 - 2/@xmath12 @xmath13(@xmath14 ) at @xmath10=1/@xmath14 @xcite , and @xmath15 is also the quantity that most instructively has been used to visualise the memory phenomenon in spin glasses mentioned above @xcite . to acquaint with the sample ,
the field cooled and zero field cooled magnetisation is plotted vs. temperature in fig .
the curves are measured in a field of 40 oe .
the cusp in the zfc susceptibility around 35 k closely reflects the spin glass temperature @xmath5 of the sample .
the classic aging experiment is performed by cooling the sample directly to a measurement temperature below @xmath5 , wait a controlled amount of time and then switch the magnetic field on or off .
not much attention has been given to the influence of the cooling rate . in fig .
2 , the relaxation rate , @xmath16 , is plotted vs. log@xmath10 for three different wait times .
the results for two substantially different , but still in a logarithmic time perspective rather similar , cooling rates are plotted , 3 k / min open symbols and 0.5 k / min solid symbols .
the wait time dependence of the data displays the characteristics of the ageing phenomenon in spin glasses ( a similar ageing phenomenon is also an inherent property of other disordered and frustrated magnetic systems @xcite ) .
the response is visually sensitive to the cooling rate , @xmath17 , but the influence of @xmath17 decreases as the wait time increases . for the @xmath7=10 s curves ,
rapid cooling yields a maximum in @xmath8(@xmath10 ) at a shorter observation time than slow cooling .
the two curves are clearly different and do not coalesce anywhere in the measured time interval . for the longer wait times , the position of the maxima do not differ appreciably , but the magnitude is somewhat higher for the rapidly cooled curves . at short observation times , for the @xmath7=@xmath18 and @xmath19 s curves , there are no or weak differences in the relaxation rates , but after some time they start to deviate and there remains a cooling rate dependence all through the long time part of our observation time window . concentrating on the position of the maximum in the relaxation rate and identifying this with an effective age of the system , @xmath20 , this parameter is apparently governed by the cooling rate and the wait time . when the wait time is short , @xmath20 is governed by the cooling rate ,
whereas for longer wait times , it is dominated by the wait time , @xmath7 .
the position of the maximum in the relaxation rate is closely equal to the wait time for @xmath7 = @xmath18 and @xmath19 s while for the @xmath7 = 10 s the maximum is delayed about a decade in time .
similar tendencies as to the evolution of the effective age with increasing wait time have earlier been reported in connection with a specific method , ` field quenching ' , to achieve a well defined initial state for ageing experiments @xcite . to further investigate how the thermal sequence on approaching the measurement temperature influences the response function , temperature shift experiments under controlled cooling and heating rates were performed . in the negative temperature shift experiment ,
the system is kept a certain wait time at a shift temperature , @xmath21 , above @xmath3 .
therafter the system is cooled to the measurement temperature where the field is applied immediatly after the sample has become thermally stabilised .
the positive temperature shift experiment follows a similar procedure , but the temperature @xmath21 is below @xmath3 .
3 shows the relaxation rate when the system has been subjected to ( a ) negative and ( b ) positive temperature shifts .
the wait time at @xmath21 is 1000 s and @xmath3 = 27 k. the cooling and heating rates are 0.5 k / min .
two reference curves measured after the sample has been cooled at 0.5 k / min directly to @xmath3 are included in the figure .
one corresponds to a wait time @xmath7 = 1000 s and the other is measured without any wait time prior to the field application ( @xmath22 0 ) . from fig .
3(a ) it is seen that for @xmath21 @xmath23 30 k the response is indistinguishable from what is measured if the sample is cooled directly to @xmath3 .
the time the sample has been kept at @xmath21 is irrelevant for the response at @xmath3 . when the shift temperature @xmath21 is lower than 30 k , the time spent at the higher temperature starts to influence the response at @xmath3 . the first deviation from the reference ( @xmath22 0 ) curve occurs at long observation times , the maximum in the relaxation rate decreases in magnitude and on further decreasing @xmath21 it shifts towards longer times , and finally when @xmath21 approaches @xmath3 , the relaxation rate approaches the @xmath7 = 1000 s reference curve .
the important implication of this behaviour is that only the thermal history in a limited temperature region just above @xmath3 governs the reponse function . the results from positive temperature shifts shown in fig .
3(b ) give a somewhat more complicated result .
a first observation is that when the system has been cooled to and aged at temperatures well below @xmath3 , the measured curves are identical , but different from the @xmath22 0 reference curve , the maximum in the relaxation rate occurs at a shorter time than for the reference curve .
thus , the response function is somewhat different after only cooling to the measurement temperature at one specific cooling rate , compared to after substantially undercooling the sample and re - heat it with a similar heating rate .
the curves representing @xmath21= 15 and 20 k are indistiguishable from each other , but when @xmath21 is increased further , the relaxation rate gets surpressed and the maximum shifts towards longer times to finally coalesce with the reference @xmath7= 1000 s curve when @xmath24 @xmath3 .
4 shows relaxation rate curves from experiments where the sample is undercooled to @xmath21 , and either kept there 1000 s ( as in fig .
3(b ) ) or immediately re - heated to @xmath3 , where the relaxation is measured using @xmath22 0 .
the cooling / heating rates are again 0.5 k / min .
the figure shows that the wait time at the lowest temperature does not affect the results , identical curves are measured whether the sample is kept at 20 k for 1000 s or 0 s. for a shift temperature , @xmath21= 24.5 k , closer to @xmath3 , a clear difference is observed between the two curves with different wait times .
the implication from these data is again that the thermal history far enough away from the measurement temperature is irrelvant to the response function at @xmath3 , the response is in the experimental time window fully governed by the previous cooling / heating history in the very neighbourhood of @xmath3 . in fig . 5 , results for ( a ) negative and ( b ) positive temperature shifts using two different cooling rates 3 k / min and 0.5 k / min ( as in fig .
3 ) are displayed .
the cooling / heating rate from @xmath21 to @xmath3 = 27 k is always the same , 0.5 k / min . for negative temperature shifts ( fig .
5 ( a ) ) there is no influence of the different cooling rates for @xmath21 @xmath23 29 k. the rapid and slow cooling give the same response at the measurement temperature . for @xmath21 = 28
k influences of the different cooling rates start to become observable at observation times , @xmath10 @xmath25 s. at shorter observation times no sign of the different cooling rates can be resolved .
when @xmath21 gets even closer to @xmath3 , the influence from the differences in cooling rates appears earlier , but still there is a region at shorter observation times where the response is independent of the cooling rate .
for positive temperature shifts , fig . 5 ( b ) , the result is different .
there are clear cooling rate effects for all different temperatures @xmath21 .
the curve representing rapid cooling always has a larger magnitude and a sharper maximum that occurs at shorter observation times than the corresponding slowly cooled curve .
this result implies that the cooling rate in the region just above @xmath3 always remains one of the governing parameters for the response after undercooling the sample , i.e. a memory of this cooling process becomes imprinted in the spin structure and is conserved in spite of the re - structuring that occurs at lower temperatures .
our current results on the non - equilibrium response function show that the cooling rate significantly influences the measured response function , especially the reponse at short wait times is dominated by the specific cooling rate employed .
also , in experiments where the sample has been undercooled below the measurement temperature , the cooling rate above the measurement temperature remains one of the governing parameters for the non - equilibrium response .
these findings have practical importance for the design of experimental procedures and when making detailed comparisons between results obtained in different experimental set - ups and in different laboratories on similar spin glass materials .
we have , using different experimental procedures , elucidated an apparently paradoxal property of the non - equilibrium spin structure in spin glasses : the spin structure records the cooling history , this thermal history ( cooling / heating rate , wait times etc . )
remains imprinted in the configuration , and the fragment of the thermal history confined in a rather narrow region close to @xmath26 higher temperature @xmath3 within the spin glass phase , can be recovered in a relaxation experiment at @xmath3 .
a continuous memory recording occurs on cooling , in spite of the fact that the spin structure is subjected to substantial reconfiguration at all temperatures below @xmath5 as is observed from the ever present ageing behaviour ; and although the short time response appears equilibrated at all temperatures in the spin glass phase . employing the droplet scaling model @xcite and the concepts chaos and overlaplength @xcite
, the observations of non - equilibrium and ageing behaviour observed after cooling the sample to a temperature in the spin glass phase can be accounted for as an immediate consequence of the growth of the size of correlated spin glass domains at constant temperature .
interpretations of ageing in these terms have been extensively discussed in numerous articles and are recently reviewed in ref.@xcite . a somewhat modified version of the droplet scaling model that has been constructed to account for the memory behaviour@xcite is extensively discussed in ref . @xcite .
the continuous memory recording of the thermal history that is exemplified by the current experimental results does however require a comment on the paradoxal possibility that reconfiguration on different length scales are fully separable , i.e. that reconfiguration of the spin structure on short length scales allows a simultaneous stability of the old configuration on larger length scales .
the droplet model prescribes one unique equilibrium spin glass state with time reversal symmetry and that domains of spin glass ordered regions grow unrestrictedly with time at constant temperature .
if the temperature is altered , the equilibrium configuration also alters due to chaos , but there is an overlap on short length scales between the equilibrium configurations , the length scale of which rapidly decreases with increasing separation between the two temperatures .
the experimental results show that a memory of the spin structure that has developed at high temperatures remains imprinted in the non - equilibrium structure at lower temperatures ( but also that it is rapidly erased if the temperature is increased above the original aging temperature ) .
such a re - stored spin structure requires that all reconfiguration at lower temperatures must occur only on small lengths scales and in dispersed regions .
the bulk of the numerous droplet excitations of different sizes that do occur may not cause irreversible changes of the spin structure on large length scales , but are to be excited within already equilibrated regions of spin glass order .
how can this rather abstract picture of the dynamic spin structure be related to our current experimental observations ?
the processes that cause the increase of the magnetisation in a zfc magnetic relaxation experiment is a polarisation of spontaneous droplet excitations . the measured quantity , the zfc magnetisation , gives an integrated value of polarisation of all droplet excitation with relaxation time shorter than the observation time and the measured relaxation corresponds to droplets with relaxation time of the order of the observation time , @xmath10 . in an ac - susceptibility experiment , the in - phase component gives the integrated value of all droplet excitations with relaxation time shorter than the observation time , @xmath10 = 1/@xmath14 , and the out - of phase component measures the actual number of droplets of relaxation time equal to 1/@xmath14 .
the non - equilibrium characteristics imply that the distribution of droplet excitations changes with the time spent at constant temperature and that there is an excess of droplet excitations of a size that corresponds to a relaxation time of the order of the wait time . on shorter time
scales an equilibrium distribution has been attained , reflected by the equilibrium reponse always obtained in ac - susceptibility experiments at higher frequencies , and in zfc measurements at long but different wait times by the fact that a similar ( equilibrium ) reponse is approached at the shortest observation times .
the implication of the cooling rate dependence is that the actual distribution of active droplets is governed by the cooling rate and the wait time at constant temperature , and that this distribution in turn is governed by the underlying spin configuration .
the phenomenon that the sample retains a distribution of droplets that is governed by the cooling rate and any previous wait time at the measurement temperature , then implies that a closely equivalent spin configuration to the original one is also retained when the temperature is recovered .
we have shown that non - equilibrium dynamics measured at a specific temperature in spin glasses is primarily governed by the thermal history close to this temperature during the cooling sequence .
if the sample has been undercooled , the response is also partly affected by the heating rate towards the measurement temperature .
the behaviour may be incorporated in a real space picture of a random spin configuration containing fractal spin glass domains of sizes that increase through spontaneous droplet excitations .
the results emphasise the importance of well controlled experimental procedures when studying non - equilibrium dynamics of spin glasses .
financial support from the swedish natural science research council ( nfr ) is acknowledged .
numerous and useful discussions on the memory phenomenon and the non - equilibrium nature of the spin glass phase with t. jonsson , e. vincent , j .-
bouchaud and j. hammann are acknowledged .
see e.g. e. vincent , j.p .
bouchaud , j. hammann and f. lefloch , phil . mag .
b * 71 * ( 1995 ) ; j .-
bouchaud , l.f .
cugliandolo , j. kurchan and m. mzard , in _ spin glasses and random fields _ p 161 - 223 , ed .
young , world scientific ( 1998 ) . | experiments on the time dependence of the response function of a ag(11 at@xmath0mn ) spin glass at a temperature below the zero field spin glass temperature are used to explore the non - equilibrium nature of the spin glass phase . it is found that the response function is only governed by the thermal history in the very neighbourhood of the actual measurement temperature . the thermal history outside this narrow region is irrelevant to the measured response .
a result that implies that the thermal history during cooling ( cooling rate , wait times etc . ) is imprinted in the spin structure and is always retained when _ any _ higher temperature is recovered .
the observations are discussed in the light of a real space droplet / domain phenomenology .
the results also emphasise the importance of using controlled cooling procedures to acquire interpretable and reproducible experimental results on the non - equilibrium dynamics in spin glasses . 2 |
type ia supernovae ( sne ia ) are catastrophic thermonuclear explosions of massive carbon / oxygen white dwarfs ( wds ) with mass exceeding a critical value . due to the low intrinsic scatter at the peak luminosity , sne ia have been employed as `` standard candles '' for measuring distances and determining cosmological parameters ( riess et al .
1998 ; perlmutter et al . 1999 ) .
the single - degenerate ( sd ) and double - degenerate ( dd ) models are the most popular scenarios for the sn ia progenitors .
the sd models assume a wd in a close binary system accreting material from a nondegenerate companion ( whelan & iben , 1973 ) , while the dd models involve a merger of two wds ( webbink , 1984 ) .
the historical supernova remnant ( snr ) tycho ( or sn 1572 , g120.1 + 1.4 ) is thought to originate from an sn ia as evidenced by its light curve ( baade 1945 ; ruiz - lapuente 2004 ) , x - ray spectroscopic analysis ( badenes et al .
2006 ) , and light echo spectrum ( krause et al . 2008 ) . however , there is no consensus on its progenitor system .
the sd scenario is supported by the discovery of a fast - moving , type g0g2 subgiant ( tycho - g ) in the snr center , which was suggested to be the surviving companion of the sn ( ruiz - lapuente et al.2004 ) .
lu et al .
( 2011 ) found an x - ray arc inside tycho , which was probably produced by the interaction between the sn ejecta and the companion star s envelope lost due to the impact of the explosion .
the sd scenario has , however , been questioned ( badenes et al .
2007 ) . while there is a dispute over whether tycho - g is the ex - companion ( see xue & schaefer 2015 and references therein ) , other candidates such as tycho - b ( kerzendorf et al .
2013 ) and tycho - e ( ihara et al . 2007 ) were also proposed .
another independent method is needed to resolve the progenitor problem of tycho .
one key aspect of the sd and dd scenarios is the presence or absence , respectively , of outflows from the binary system during the pre - sn evolution ( badenes et al .
according to the sd scenario , a strong wind is driven to stabilize the mass transfer when the mass accretion rate exceeds a certain critical value ( e.g. , hachisu et al .
1996 ; li & van den heuvel 1997 ; hachisu et al .
the strong wind could blow a low - density bubble before the sn explodes and the shock wave interacts with the dense material of the wind cavity ( e.g , rcw 86 ; vink et al .
1997 ; broersen et al .
the size of the wind - blown bubble ( wbb ) depends not only on the wind parameters , but also the on the density of the ambient medium ( e.g. , weaver et al .
1977 ; koo & mckee 1992a , 1992b ) .
p3.cmccccc @xmath0co @xmath1=21 & ( @xmath10 , @xmath11 ) & @xmath12 & @xmath13 & - & - + @xmath0co @xmath14co @xmath1=10 & p1(@xmath15 , @xmath16 ) & - & @xmath17 & 7.3 & @xmath18 + @xmath0co @xmath14co @xmath1=10 & p2(@xmath19 , @xmath20 ) & - & @xmath17 & 5.2 & @xmath21 + @xmath0co @xmath14co @xmath1=10 & p3(@xmath22 , @xmath23 ) & - & @xmath17 & 7.7 & @xmath24 + @xmath0co @xmath14co @xmath1=10 & p4(@xmath25 , @xmath26 ) & - & @xmath17 & 9.3 & @xmath27 + @xmath0co @xmath14co @xmath1=10 & p5(@xmath28 , @xmath29 ) & - & @xmath17 & 7.3 & @xmath30 + @xmath0co @xmath14co @xmath1=10 & p6(@xmath31 , @xmath32 ) & - & @xmath17 & 8.6 & @xmath33 [ t : obs ] tycho is is proably the only known type ia snr interacting with a molecular cloud ( mc ) .
radio observations spanning a 10 yr interval revealed slow expansion in the northeast , where the shell is encountering denser material ( reynoso et al
. 1997 , 1999 ) .
lee et al .
( 2004 ) found that the snr is likely surrounded by a molecular shell at @xmath34 to @xmath7 @xmath4 with the nobeyama @xmath0co @xmath1=10 observation ( beam width of @xmath35 ) .
the mc in this velocity interval has also been studied in @xmath0co @xmath1=10 lines ( beam width of @xmath36 ; cai et al .
2009 ) , and in @xmath0co @xmath1=21 and @xmath1=32 lines ( beam width of @xmath37 and @xmath38 , respectively ; xu et al .
shock interaction with the adjacent dense gas provides a reasonable mechanism for the hadronic @xmath39-ray emission arising from tycho ( zhang et al .
the association of the mc with tycho has not been confirmed yet and still requires kinematic evidence such as line broadening and a large high - to - low excitation line ratio ( see jiang et al .
2010 for observational evidence for judging the contact of snrs with mcs ) . to confirm or refute the snr mc association and to address the progenitor of tycho , we have performed high - resolution co observations with the iram 30 m telescope .
we have performed observations of molecular lines toward tycho in 2013 august with the iram 30 m telescope .
the @xmath0co @xmath1=21 ( at 230.538 ghz ) mapping was conducted with the on - the - fly position switching mode using the 9-pixel dual - polarization heterodyne receiver array ( hera ) and the fast fourier transform spectrometer ( fts ) .
the backend fts provides a bandwidth of 0.5 ghz and a velocity resolution of 0.063 @xmath4 .
the @xmath0co @xmath1=21 observation covered a @xmath12 region centered at ( @xmath40 , @xmath41 , j2000 ) , with an antenna half - power beam width ( hpbw ) of @xmath42 .
the data were then converted to a cube with a spatial pixel size of @xmath43 . after correcting the main - beam efficiency of 59% at 230.5 ghz , the mean rms noise level of the main - beam temperature ( @xmath44 ) is 0.32 k in each pixel with a velocity resolution of 0.5 @xmath4 .
we carried out a line survey at the 3 mm band at six positions ( labeled with plus signs in the bottom - right panel of ) with the eight mixer receiver ( emir ) of the iram 30 m telescope .
the backend fts provided a 16-ghz frequency coverage in 92100 ghz and 108115 ghz .
the velocity resolution , angular resolution , and main beam efficiency were about @xmath45 , @xmath17 , and 78% , respectively .
the mean rms noise levels of @xmath44 are 0.060.10 k at 115 ghz and 0.020.05 k at 110 ghz .
the information of the observed lines is summarized in .
in addition , we retrieved the public archival @xmath14co @xmath1=10 mapping data taken during 2004 may and june with the 13.7 m millimeter - wavelength telescope of the purple mountain observatory ( pmo ) at delingha in china .
the data with a velocity resolution of 0.11 @xmath4 were convolved to an angular resolution of @xmath46 ( initial hpbw was @xmath47 ) to match the grid spacing of @xmath46 .
we used the class package of gildas to reduce the data and the karma ( gooch 1996 ) and idl software packages for the data analysis and visualization .
the field of tycho shows emission of @xmath0co @xmath1=21 in the lsr velocity range of @xmath48 to @xmath49 @xmath4 ( see ) .
the overall emission has two peaks around the @xmath50 @xmath4 and @xmath51 @xmath4 , with a few weak structures at the velocity range @xmath52 to @xmath53 .
the molecular gas near tycho is highly clumpy , as revealed in the channel maps .
the @xmath0co @xmath1=21 emission has a @xmath54 in most of the field , which is lower than the kinetic temperature of typical interstellar mc ( @xmath55 k ) .
the low main - beam temperature further supports the mc being highly clumpy ( size @xmath56 pc at an assumed distance of 2.5 kpc ) and not resolved in our observation .
the brightest @xmath0co @xmath1=21 emission ( @xmath57 ) arises from the southwestern end of a molecular strip ( @xmath58 , @xmath59 , j2000 ) , which in projection contacts the eastern concave of the radio shell ( see ) . at @xmath60
, the relatively bright @xmath0co @xmath1=21 clumps are generally distributed surrounding the shell of the snr and display similar morphology to that previously shown in the @xmath0co @xmath1=10 emission ( lee et al .
the clumps contacting the snr in the east , southeast , and north are consistent with the locations showing low expansion rate from the radio observations ( reynoso et al .
1997 ) . in order to search for kinematic evidence of the snr mc interaction , we investigate the profiles of multiple molecular lines .
only @xmath0co @xmath1=10 ( 115.271 ghz ) and @xmath14co @xmath1=10 ( 110.201 ghz ) lines were detected in the 3 mm band line survey toward the six positions at the northeastern boundary of tycho .
shows the @xmath0co @xmath1=10 , @xmath14co @xmath1=10 , and @xmath0co @xmath1=21lines at the six positions . here
the @xmath0co @xmath1=21 data were convolved to @xmath61 to match the beam size of @xmath0co @xmath1=10 data .
we searched for broad @xmath0co line profiles in order to trace the kinematic structure of the shocked molecular gas , and high temperature of @xmath0co @xmath1=21 to trace heated gas . at the positions `` p2 '' and `` p3 '' , we find @xmath0co line wings at @xmath62 where @xmath14co emission was not detected .
such high @xmath0co/@xmath14co ratios suggest that the @xmath0co lines are optically thin due to low co column density or / and large velocity gradient , as commonly seen in the line broadenings from the shocked clouds . at position ``
p2 '' , the @xmath0co @xmath1=21 emission in the velocity range @xmath6 to @xmath7 @xmath4 is stronger than @xmath0co @xmath1=10 , with a velocity - integrated intensity ratio of @xmath63 .
such a large high - to - low excitation line ratio needs high - excitation conditions and is generally regarded as evidence of shock heating and perturbation ( seta et al .
1998 ; chen et al . 2014 ) .
the azimuth angle of `` p2 '' ( @xmath64 ; counterclockwise from the north ) is consistent with the angle where the forward shocks are significantly decelerated during 2003 - 2007 ( see fig . 5 in katsuda et al .
2010 ) .
an expanding bubble is revealed in the position - velocity diagrams of @xmath0co @xmath1=21 across the center of tycho ( see ) .
the diagrams are obtained from six cuts starting from the position angle ( pa ) of @xmath65 and with a @xmath66 decrement ( labeled with `` a''``f '' in the bottom right panel of ) .
the morphology of the molecular gas revealed in the position - velocity dimensions is very similar to an expanding sphere centered at @xmath2 with an angular size of @xmath67 and a velocity of @xmath68 .
most of the gas is at the blue - shifted side ( @xmath69 to @xmath7 @xmath4 ) .
the consistency of the size and position of the expanding circular gas with those of tycho , together with the cavity morphology ( see ) and large @xmath0co @xmath1=21-to-@xmath1=10 ratio at @xmath70 to @xmath71 ( see ) , implies that tycho is associated with the expanding molecular bubble at @xmath69 to @xmath9 @xmath4 .
using the iram 30 m observations , we show that tycho is evolving in an expanding molecular bubble at @xmath72 .
the shock of the snr might not have extensively impacted the molecular gas , but only contacted partially , e.g. , in the northeast where the rim is deformed and decelerated as revealed in radio ( reynoso et al .
1997 ) and x - ray observations ( katsuda et al .
although the data do not show very broad co lines even with the best angular resolution in co observations obtained to date ( @xmath73 ) , the large @xmath0co @xmath1=21-to-@xmath1=10 ratio and @xmath0co - to-@xmath14co ratios are present in the northeast of the remnant .
this can be interpreted as the shocked clumps remaining unresolved with current observations ( low brightness temperature ; see section [ s : results ] ) , and the shock front having just reached the molecular medium in the recent past ( within a few tens of years ; reynoso et al .
1999 ) and swept up a small amount of gas .
this molecular gas can be the dense target of collision by the cosmic - ray protons accelerated by tycho s blast shock , which gives rise to @xmath39-ray emission due to hadronic interaction .
a high ambient gas density provides a possibility of the relatively low fraction of energy deposited to the shock - accelerated protons in the young tycho snr ( zhang et al .
2013 ) .
the suggested distance of tycho varies between 1.7 and 5 kpc according to different methods ( albinson et al . 1986 ; schwarz et al . 1995 ; krause et al . 2008 ; hayato et al .
2010 ) . as suggested in zhang et al .
( 2013 ; see also lee et al .
2004 ) , the system velocity of @xmath74 of the mc allows the alternative distances of 4.0 and 2.5 kpc , and the absence of an h absorption feature at @xmath75 to @xmath76 @xmath4 ( tian & leahy 2011 ) is suggestive of the nearer distance .
hence , we adopt a distance of @xmath77 kpc for tycho .
the column density of the molecular gas at the six positions at the northeastern boundary of the snr is estimated with the @xmath14co line detected at @xmath78 to @xmath7 @xmath4 .
assuming that the rotational levels of the @xmath14co molecules are in local thermodynamic equilibrium and the @xmath14co lines are optically thin , we obtain a column density of h@xmath79 @xmath80^{-1 } { \,{\rm cm}}^{-2}$ ] , where the relative abundance ratio of h@xmath79 to @xmath14co is taken as @xmath81 ( dickman 1978 ) .
a typical temperature of 10 k in mcs is adopted here for the excitation temperature @xmath82 .
the @xmath83 values at the six positions at the northeastern boundary of the tycho are derived to be in the range ( 5.2@xmath84 as tabulated in .
we use the pmo @xmath14co
@xmath1=10 line mapping to estimate the mass of the mc surrounding the snr .
the @xmath14co emission is selected in a circular region with a radius of @xmath85 and centered at point ( @xmath86 , @xmath87 , j2000 ) .
the region encloses bright @xmath0co @xmath1=21 emission present in .
the mean @xmath83 in the region is @xmath88 and the mass of the mc is @xmath89 , where @xmath90 is the distance in units of 2.5 kpc .
this mass is one order of magnitude lower than that derived with @xmath0co lines and the co - to - h@xmath79 conversion factor ( cai et al .
the standard conversion factor is a rough estimate of the global mass of a giant mc with a mass of @xmath91@xmath92 ( dame et al .
2001 ) , and has a large variability on small spatial scales ( bolatto et al .
2013 ) .
the bubble surrounding tycho has an inner radius of about @xmath93 ( corresponding to @xmath94 pc ) and a mass of @xmath89 .
such an expanding bubble could be a wbb created with a central star / system , while the shock wave of the snr is not a plausible source since the shock wave has just reached the boundary of the molecular cavity and the velocity width ( fwhm ) of the co lines in the field is smaller than the expanding velocity of @xmath95 .
first , we can exclude a massive star as the central source . if the molecular bubble here was created by a massive star , using the linear relationship between the maximum wbb sizes of main - sequence ob stars in the molecular environment @xmath96 and the initial masses of the stars @xmath97 ( chen et al .
2013 ) : @xmath98 , where @xmath99 is the interclump pressure @xmath100 scaled to @xmath101 ( @xmath102 in the interclump medium of mcs ; chevalier 1999 ; blitz 1993 ; krumholz et al .
2009 ) , the mass of the star would be @xmath103 ( b2 type or earlier ) .
such massive early - type stars have not been detected inside the bubble .
the alternative explanation is related to the progenitor system of tycho .
substantial outflow with a mass - loss rate @xmath104 has been suggested to occur when a carbon / oxygen wd accretes matter from a main - sequence or ( slightly ) evolved companion star .
the wind may have a velocity @xmath105 larger than a few hundred @xmath4 ( kato & hachisu 1999 ) .
the fast wind from the accretion disk can evacuate a low - density cavity in which radiative losses are not important ( badenes et al .
2007 ; koo & mckee 1992a , b ) .
the structure of the interaction of a strong wind with the interstellar medium is described by the four zones : ( 1 ) freely expanding wind ; ( 2 ) shocked stellar wind ; ( 3 ) a shell of shocked interstellar gas ; and ( 4 ) ambient gas ( weaver et al .
the expanding molecular bubble surrounding tycho has an inner radius @xmath106 pc and an expansion velocity @xmath107 . in the adiabatic condition ,
if the wind lasts until the sn explosion ( e.g , see fig .
1e in han & podsiadlowski 2004 ) , the radius of the bubble ( corresponding to the outer boundary of the shocked wind ) is @xmath108 , where @xmath109 is the mass density of the ambient gas , with @xmath110 the number density of h atoms ( weaver et al . 1977 ) .
the duration of the wind is @xmath111 .
the density @xmath112 is estimated from the mass , @xmath113 , of the swept - up molecular gas surrounding the progenitor of tycho .
we estimate the wind velocity as @xmath114 ( see evolutionary path a in for the case with @xmath115 ) . here
we constrain the wind velocity to be @xmath116 so that the modeled bubble is in the energy conservation stage . in this stage
, the density of adiabatic gas in the shocked wind has a flat density profile ( see eq . 16 in weaver et al . 1977 ) , which could be consistent with the suggestion that the snr is evolving in a uniform medium ( badenes et al .
2007 ; yamaguchi et al . 2014 ) .
since the mass - loss rates and durations of the wind vary in different outflow models ( han & podsiadlowski 2004 ) , if the wind stops long before the sn explosion , the observed @xmath117 and @xmath118 values can give multiple solutions of wind parameters .
( when the bubble size was @xmath119 and velocity was @xmath120 ) prior to the sn explosion , we assume the bubble will continue to expand driven by the thermal pressure ( @xmath121 ) of the shocked wind region . from the mass equation @xmath122 and the momentum equation @xmath123 , the bubble radius is govern by @xmath124 , where @xmath125 with @xmath126 ( weaver et al . 1977 ) . ] some exemplified evolutionary paths of the bubble are shown in with the parameters listed in , where a typical @xmath127 value of @xmath128 is adopted .
the gas density at the center of the shocked wind region is given by castor et al .
( 1975 ) @xmath129 @xmath130 , which is further scaled as @xmath131 if the elapse time @xmath132 between the wind stop and sn explosion can not be ignored .
the density estimates , @xmath133@xmath134 , for the interior of the bubble are consistent with the previous constraints for the snr preshock gas obtained from x - ray ( @xmath135 ; katsuda et al .
2010 ) and infrared measurements ( @xmath136@xmath137 , williams et al .
2013 ; the observed elevated density in the northeast can be related to the denser gas near the cavity wall ) .
all of the wind parameters @xmath105 , @xmath138 and @xmath127 match those expected in the sd scenario .
there is also some observational evidence to support the existence of recurrent novae in outbursts before the sn explosion ( hachisu & kato 2001 ) , which may be an alternative outflow source to affect the circumstellar medium of the snr .
lccccccc a & 250 & 0.3 & 3.9 & 0 & 0.04 & 0.04 & + b & 250 & 1.0 & 1.3 & 1.6 & 0.09 & 0.02 & + c & 280 & 1.0 & 1.1 & 1.7 & 0.10 & 0.03 & + d & 350 & 1.0 & 0.8 & 1.8 & 0.13 & 0.02 & [ tab : para ] a wbb will reach a maximum size @xmath139 when it becomes in pressure equilibrium with the ambient medium : @xmath140^{1/3}$ ] , where @xmath141 is the total mass - loss of the wind ( chevalier 1999 ) . for this expanding bubble , we have @xmath142 , and hence we set a constraint on the mean wind velocity : @xmath143 .
our results support the theoretical predictions of a strong wind ( over @xmath144 ) driven by the sd progenitor system .
an earlier wind from the post - agb star in the planetary nebula phase is unlikely to explain the current molecular bubble . assuming that the bubble is created by an ejection of mass ( @xmath145 at a velocity of @xmath146 for planetary nebulae ) before the formation of a wd
, the slowly expanding molecular bubble should have been in the radiative evolution phase with an age ( chevalier 1974 ) : @xmath147 , which is much smaller than the duration of the later evolution of either sd ( @xmath148 ; hachisu et al .
1996 ) or dd progenitor systems ( @xmath149 ; kashi & soker 2011 ) .
the disk wind and the later mass transfer should not last longer than the lifetime of the mcs ( of the order of 10 myr ; see mac low & klessen 2004 and references therein ) , or the molecular bubble would have vanished . actually , the time between the onset of the wind and the sn explosion is less than a few myr according to various sd models ( see badenes et al .
2007 and references therein ) .
compared with the core - collapse sne , sne ia have much longer delay time ( elapsed time between the formation of the star and the explosion sn ; @xmath150 myr ; maoz et al .
the mc surrounding tycho was probably newly formed in the vicinity of the progenitor binary , rather than the natal cloud .
we have performed a @xmath0co @xmath1=21 mapping and a 3-mm line survey toward the type ia snr tycho using the iram 30 m telescope .
the main results are summarized as follows : 1 .
snr tycho appears to be confined in the clumpy ( @xmath151 ) molecular gas at @xmath70 to @xmath7 @xmath4 . in this velocity range
, we have found a large @xmath0co @xmath1=21-to-@xmath1=10 ratio ( @xmath63 ) and large @xmath0co to@xmath14co ratios at the northeast of the remnant , where the shell is deformed and decelerated .
these features provide kinematic evidence for the interaction between tycho and the molecular gas .
we discover the expanding motion ( @xmath68 ) of the molecular bubble at @xmath152 .
the mass of the molecular bubble is @xmath153 and the mean column density is @xmath88 .
the snr has been expanding in the low - density wbb and the shock wave has just reached the molecular cavity wall .
the most plausible origin for the expanding bubble is the fast outflow ( with velocity hundreds of @xmath4 ) driven from the vicinity of the progenitor wd as it accreted matter from a nondegenerate companion star .
the shocked wind confined in the molecular bubble has a uniform density ( @xmath154-@xmath134 ) .
the mc surrounding tycho was probably newly formed in the vicinity of the progenitor binary , rather than the natal cloud .
this is the first unambiguous detection of the expanding bubble driven by the progenitor of the type ia snr , which provides important and independent evidence to support that the progenitor of tycho was an sd system .
p.z . is thankful to dr .
fang - jun lu , dr .
zhi - yuan li , and prof .
noam soker for helpful discussions and comments .
p.z . and y.c .
thank the support of nsfc grants 11503008 , 11233001 and 11590781 , the 973 program grant 2015cb857100 , and grant 20120091110048 by the educational ministry of china . z.z .
acknowledges support from the european research council ( erc ) in the form of an advanced grant , cosmicism . x.l . was supported by nsfc under grant numbers 11133001 and 11333004 .
s.s.h . acknowledges support from the canada research chairs program , the nserc discovery grants program , and the canadian space agency . | whether the progenitors of sne ia are single - degenerate or double - degenerate white dwarf ( wd ) systems is a highly debated topic . to address the origin of the type ia tycho s supernova remnant ( snr ) , sn 1572 ,
we have carried out a @xmath0co @xmath1=21 mapping and a 3-mm line survey toward the remnant using the iram 30 m telescope .
we show that tycho is surrounded by a clumpy molecular bubble at a local standard of rest velocity of @xmath2 which expands at a speed of @xmath3 @xmath4 and has a mass of @xmath5 ( at the distance of 2.5 kpc ) .
enhanced @xmath0co @xmath1=21 line emission relative to @xmath0co @xmath1=10emission and possible line broadenings ( in velocity range @xmath6 to @xmath7 @xmath4 ) are found at the northeastern boundary of the snr , where the shell is deformed and decelerated .
these features , combined with the morphological correspondence between the expanding molecular bubble and tycho , suggest that the snr is associated with the bubble at the velocity range @xmath8 to @xmath9 @xmath4 .
the most plausible origin for the expanding bubble is the fast outflow ( with velocity of hundreds @xmath4 ) driven from the vicinity of a wd as it accreted matter from a nondegenerate companion star .
the snr has been expanding in the low - density wind - blown bubble , and the shock wave has just reached the molecular cavity wall .
this is the first unambiguous detection of an expanding bubble driven by the progenitor of a type - ia snr , which constitutes evidence for a single - degenerate progenitor for this sn ia . |
mass loss is one of the most pronounced features of agb stars , influencing both their evolution and their observable properties in a decisive way .
the most commonly accepted scenario is that of a pulsation - enhanced dust - driven wind : stellar pulsation causes atmospheric shock waves which intermittently lift gas above the stellar surface , creating dense , cool layers where solid particles form .
the dust is accelerated away from the star by radiation pressure , dragging the gas along .
the composition of the grains is determined by the relative elemental abundances in the atmosphere , with c - rich stars forming mostly amorphous carbon grains , and o - rich objects producing silicate dust .
current knowledge suggests that winds of c - rich agb stars are rather well understood . for their numerous o - rich counterparts ,
however , recent work demonstrates that the opacities of silicate grains are insufficient for driving winds ( woitke 2006a and this volume ) , contrary to previous expectations .
this problem becomes apparent with the introduction of frequency - dependent radiative transfer in time - dependent models with a detailed description of dust formation .
ironically , this important step towards more realistic models which brought c - rich models into good agreement with observations , causes serious problems with the wind mechanism in o - rich stars .
the core of this issue can be understood with comparatively simple analytical arguments .
a significant part of this review is devoted to isolating the crucial ingredients and analyzing them with simplified analytical descriptions , in particular non - equilibrium dust formation , with special attention to the influence of non - grey effects and differences between c - rich and o - rich stars .
these simple estimates are compared to the results of existing numerical models , rounding up with an overview of the present status of such detailed models , and current trends .
detailed comparisons of the input physics of various numerical models and summaries of the historical developments in this field can be found in earlier reviews , e.g. , by willson ( 2000 ) , woitke ( 2003 ) and hfner ( 2005 ) .
dust grains forming in the extended atmospheres of agb stars condense under non - equilibrium conditions with temperature acting as a threshold , prevailing densities and abundances determining the efficiency of grain growth , and pulsation and shock waves setting the time scales . in other words
, condensation will usually be incomplete , and the dust - to - gas ratio is not a simple function of abundances .
a prerequisite for condensation or survival of existing grains are temperatures below the stability limit of a specific condensate . in principle
, a distinction between the respective temperatures of ambient gas and grains has to be made . in practice , however , gas temperatures tend to be lower than grain temperatures at a given point , and therefore the limit will usually be defined by the grain temperatures which , in turn , are given by radiative equilibrium , i.e. @xmath1 using the assumption of a planckian radiation field ( with stellar surface temperature @xmath2 at @xmath3 ) , geometrically diluted according to distance from the stellar surface @xmath4 in the equation of radiative equilibrium , the grain temperature @xmath5 as a function of distance from the star can be estimated if the wavelength - dependent opacity of the dust particles @xmath6 is known .
the latter can often be approximated in the critical wavelength region around the stellar flux maximum by @xmath7 where the exponent @xmath8 is characteristic of the grain material , leading to @xmath9 where the superscript planck denotes a planck mean of the opacity ( see , e.g. , lamers & cassinelli 1999 for details ) .
the condensation radius @xmath10 , defined as the point where the grain temperature @xmath5 is equal to the condensation temperature @xmath11 ( stability limit ) of the respective grain material , is therefore given by @xmath12 where we have used the approximation @xmath13 for @xmath14 . therefore , to determine the location where condensation may start for a particular material , we need to know its condensation temperature and the dependence of the opacity on wavelength around the flux maximum . for amorphous carbon grains we have @xmath15k and @xmath16 . assuming a stellar surface temperature of @xmath17k we obtain @xmath18 which is in good agreement with detailed frequency - dependent models where dust usually forms at @xmath19 . for silicate grains the picture is more complicated .
to illustrate the problem , we consider olivine , i.e. a material with the composition mg@xmath20fe@xmath21sio@xmath22 ( @xmath23 ) and a condensation temperature of @xmath24k in the relevant density range . on the one hand
, the lower value of @xmath25 ( compared to @xmath26 for carbon grains ) leads to a stronger dependence of @xmath27 on @xmath8 ( see fig . [ f_rcond ] ) .
on the other hand , the value of @xmath8 varies strongly with the relative fe - content ( @xmath28 ) of the material , with larger ( more positive ) values of @xmath8 for iron - rich grains . for forsterite ( mgsio@xmath22 ) ,
representing one end of the chemical spectrum , lab data indicates @xmath29 in the relevant wavelength rage ( see fig . 2 in andersen , this volume ) .
for mgfesio@xmath22 ( with equal amounts of mg and fe , corresponding roughly to solar composition ) the value is closer to @xmath30 .
therefore forsterite may condense at a distance of @xmath31 while mgfesio@xmath22 will probably not form closer to the star than @xmath32 .
already these first estimates lead us to expect that silicate grains forming in agb stars will tend to be iron - poor , a conclusion which is supported by other arguments and detailed modelling ( see below).fe@xmath33sio@xmath34 ( @xmath35 ) . ] .
the lower curve corresponds to @xmath36 ( representative of carbon grains ; full line ) and the upper curve to @xmath37 ( characteristic of silicates ; dashed ) .
the symbols show the approximate location of various materials in this diagram ( diamond : amorphous carbon ; circle : mg@xmath38sio@xmath22 ; square : mgfesio@xmath22 )
. models with grey radiative transfer ( or grey dust opacities ) correspond to @xmath39.,width=453 ] assuming for the moment that a certain dust species will actually form beyond its condensation radius , will the grains significantly contribute to driving a wind ?
this is , of course , a matter of the corresponding opacity produced by the grains .
if we can assume that gravity and radiation pressure are the only relevant forces beyond @xmath10 , the ( co - moving ) equation of motion for a matter element can be written as @xmath40 where @xmath41 is the distance from the stellar center , @xmath42 the velocity , @xmath43 denotes the gravitational acceleration at the ( arbitrary ) point @xmath44 , @xmath45 the flux mean opacity , and @xmath46 and @xmath47 are the stellar mass and luminosity , respectively ( all constants have their usual meaning ) . the material properties ( and relative abundances ) of the dust species
determine the value of @xmath48 which , in turn , determines the further dynamics of the matter element .
several special cases are immediately apparent : * if no ( or very little ) dust is forming , @xmath48 will be ( close to ) zero .
the matter element will follow a ballistic trajectory , according to its initial velocity , reaching a maximum distance @xmath49 of @xmath50 and then falling back towards the stellar surface . * if @xmath51 , the r.h.s of the equation of motion vanishes .
the matter element continues to move at a constant velocity . * if @xmath52 , @xmath53 is positive , the matter is accelerated away from the star . * if @xmath54 , the final fate of the matter element ( escape or fall - back ) depends on the velocity at which it is moving when it reaches @xmath10 .
the dynamical behavior in the regime @xmath54 can be investigated with the following simplified model : if we assume that @xmath48 is constant outside @xmath10 ( and zero inside ) , the factor @xmath55 is a constant that can be multiplied with @xmath56 , resulting in an equation of motion @xmath57 which formally looks like the ballistic equation of motion for @xmath58 , but with a re - scaled gravitational acceleration @xmath59 , and therefore a re - scaled ( lower ) escape velocity @xmath60 .
consequently , the actual velocity of the matter element when reaching @xmath10 may be above @xmath61 ( while still being below the escape velocity without dust ) , resulting in a wind . to find the critical case that divides fall - back from outflow , we assume that the velocity prior to condensation is only due to the pulsation - induced shocks propagating through the atmosphere .
we use the simplified picture that a shock passing through the matter element under consideration can be treated as an instantaneous acceleration to a velocity @xmath62 at the point @xmath44 ( which will in practice be close to @xmath3 ) , followed by a ballistic movement of the shocked gas .
we consider the solution of the ballistic equation of motion ( @xmath63 ) for @xmath64 , requiring that the velocity at the point @xmath10 matches @xmath61 , i.e. , @xmath65 for @xmath54 we can therefore distinguish two cases : * a pulsation - supported dust - driven wind regime for + @xmath66 * a parachute regime ( decelerated fall - back ) for + @xmath67 assuming that @xmath68 and , consequently , @xmath69 the condition for a pulsation - supported dust - driven outflow can be reformulated in the more convenient form @xmath70 where @xmath71 denotes the velocity of the shocked gas ( not to be confused with the shock amplitude ) . in order to determine the value of @xmath48 for various grain materials , we first need to calculate the corresponding ( flux mean ) opacities . for simplicity
, we assume that @xmath45 can be approximated with the small particle limit opacity of spherical grains at @xmath72 m ( flux maximum of the star ) .
therefore , we have @xmath73 where @xmath74 denotes the grain radius , @xmath75 the number density of grains and @xmath76 is the extinction efficiency . for grains small compared to the wavelength , the quantity @xmath77 becomes independent of the grain radius and can therefore be taken out of the integral .
the last integral represents the fraction of a given volume which is occupied by the grains , apart from a factor @xmath78 , and it can be rewritten in terms of the space occupied by a monomer ( basic building block ) in the condensed material times the number of monomers found in a certain volume , @xmath79 where we have expressed the monomer volume @xmath80 in terms of the atomic weight of the monomer @xmath81 and the density of the grain material @xmath82 , and the number of monomers in a volume @xmath83 by the abundance of the key element of the condensate @xmath84 , the degree of condensation of this key element @xmath85 , and the total number density of h atoms @xmath86 ( @xmath87 = proton mass ) . using @xmath88 ,
we finally obtain @xmath89 below , we list the properties of different grain materials and the corresponding @xmath48 , assuming @xmath90 and a degree of condensation of the key element @xmath91 ( note that the value of @xmath48 scales linearly with each of these two factors ) .
the abundance of the key element @xmath84 corresponds to either all c not bound in co for carbon grains ( c / o = 1.5 ) , or to si for silicates . the data for @xmath92 is taken from rouleau & martin ( 1991 , amorphous carbon ) and jger et al .
( 2003 , silicates ; cf . figs . 1 and 2 in andersen , this volume ) .
[ cols= " < , < , < , < , < , < " , ] as expected , the value of @xmath48 for amorphous carbon grains is well above the threshold for a dust - driven wind , even for a degree of condensation @xmath93 as typically found in detailed models .
for silicate grains , the picture is , again , more complex : for forsterite ( mg@xmath38sio@xmath22 ) , on the one hand , @xmath48 is so far below the critical value that even the scenario of a pulsation - supported dust - driven outflow as discussed above is unrealistic , since it would require @xmath94 close to unity .
silicate grains containing equal amounts of mg and fe , on the other hand , could result in a sufficiently high @xmath48 , but as we saw above they will not form sufficiently close to the star to drive a wind . at this point one might wonder about alternatives to the ballistic one - shock scenario for the trajectory prior to dust condensation .
it might be possible that a matter element gets hit by several shocks at increasing distance from the star if the fall - back is slow enough ( parachute regime , see above ) , gradually driving the gas out to a considerable distance ( see , e.g. , bowen 1988 ) .
one could even think about replacing the first step(s ) on the way with other mechanisms which on their own may not be sufficient to drive an outflow ( just like the shock waves alone will hardly do the trick , at least not within observed constraints as known today ) .
several such possibilities have been discussed in the literature earlier , e.g. , alfvn waves ( e.g. , hartmann & macgregor 1980 ) , sound waves ( e.g. , pijpers & hearn 1989 ) , or high thermal gas pressure due to heating by shock waves ( calorisphere , willson & bowen 1998 ) .
however , each of these alternatives faces the same question : will densities and dynamical timescales allow for efficient grain condensation at a certain distance from the star ?
discussions about dust in atmospheres and winds of agb stars often ignore the fact that a low enough temperature is only a necessary condition for the formation or survival of dust grains , and not a sufficient one . in dynamical atmospheres and winds where timescales are set by pulsation and wind dynamics ,
the efficiency of dust formation is strongly dependent on prevailing gas densities and abundances . as matter moves away from the star , the temperature decreases , which at first
favors grain formation . at the same time
, however , the growth of grains turns into a race against falling density which slows down the process , and more often than not leads to incomplete condensation .
simple estimates based on gas kinetics as presented in gustafsson & hfner ( 2004 ) demonstrate that the timescales for the growth of carbon grains close to the condensation radius ( i.e. at about 2 - 3 stellar radii ) are on the order of a year , i.e. comparable with the pulsation period of the star , and increasing outwards with falling density .
this is in good agreement with detailed numerical models which tend to show mean degrees of condensation well below unity ( typically 0.3 - 0.5 ) , and a rather limited zone of grain growth .
iron - poor silicates , which should have a condensation radius similar to amorphous carbon grains ( see above ) , face the problem that the abundance of the key element si is about an order of magnitude lower than that of c , increasing the estimate for the grain growth time by a corresponding factor ( the timescale is inversely proportional to this abundance ) .
this might result in a rather low dust formation efficiency , which , in combination with the low opacity of such grains , makes them a by - product of the wind , not a driver .
iron - rich silicates , on the other hand , which could in principle contribute to the total opacity , and consequently to driving the wind , will most likely not form in significant amounts , being handicapped by both the low abundance of si and the much lower densities prevailing at distances corresponding to their large condensation radius . at such distances , the densities will be at least an order of magnitude lower than in the region where carbon or iron - poor silicates may form , adding another factor of ten to the grain growth timescale .
the development of models for a particular phenomenon in this case mass loss through dusty winds often occurs in several steps : first ideas about basic processes lead to order of magnitude estimates , followed by simple ( analytical or numerical ) models .
if the basic principles seem sound the next step is an iterative improvement of numerical models by comparison with observations , leading eventually to detailed , reasonably realistic models which , finally , can be applied to study certain astrophysical phenomena in a wider context ( e.g. the role of dusty winds in stellar and galactic evolution , to pick a not - so - random example ) .
detailed time - dependent models for winds of agb stars currently come in three major groups , namely two types of spherically symmetric frequency - dependent models including non - equilibrium dust formation for c - rich and o - rich chemistry , respectively , and 2d/3d models concerned with the effects of giant convection cells and structure formation on atmospheres and mass loss .
these three types of models have reached different stages in their development , as will be discussed below .
early pioneering models in this field , exploring the general effects of pulsations and dust for cool stellar winds ( e.g. wood 1979 , bowen 1988 ) make hardly a distinction between c - rich and o - rich stars , with the possible exception of choosing appropriate values for certain input parameters .
the turning point came with the inclusion of a detailed description of dust formation ( e.g. , fleischer et al .
1992 , hfner & dorfi 1997 , winters et al .
2000 , jeong et al .
2003 ) in contrast to a parameterized description of the dust opacity , and/or the introduction of frequency - dependent radiative transfer , accounting for the complex opacities of molecules and dust ( e.g. , hfner 1999 , woitke 2006a ) . the latest generation of models for atmospheres and winds of c - rich agb stars by hfner et al . ( 2003 ) , combining a frequency - dependent treatment of radiative transfer with time - dependent hydrodynamics and a detailed description of dust formation , compares well with various types of observations , such as low - resolution nir spectra ( gautschy - loidl et al .
2004 ) or profiles of co vibration - rotation lines ( nowotny et al . 2005ab and this volume ) . with these non - grey dynamical models
it is possible for the first time to simultaneously reproduce the time - dependent behavior of fundamental , first and second overtone vibration - rotation lines of co , features originating in the outflow , dust formation region , and pulsating atmosphere , respectively , probing the dynamics from the photosphere out into the wind .
recently , mattsson et al . ( 2007 ) have applied these models to investigate the formation of detached shells in connection with a he - shell flash .
currently , a large grid of dust - driven wind models for c - rich agb stars and an accompanying library of variable synthetic spectra are being computed ( see mattsson et al .
, this volume ) .
the development of similar detailed models for winds of o - rich agb stars has been lagging behind the c - rich case , not the least due to a more complex scenario for dust formation .
jeong et al . ( 2003 ) presented wind models for m - type stars , combining a detailed description of dust formation with time - dependent dynamics and grey radiative transfer . while the stellar parameters of these models are somewhat on the extreme side ( high luminosities , low effective temperatures ) , the resulting wind characteristics are reasonably realistic . in view of the discussion in the previous section ,
it may be surprising that it is possible to drive winds with silicate grains .
this is a direct consequence of the grey radiative transfer used in these models which corresponds to an effective value of @xmath39 ( cf .
[ f_rcond ] ) , resulting in a rather small condensation radius for all types of grains , including iron - rich species .
frequency - dependent models for o - rich agb stars by woitke ( 2006a , and this volume ) clearly demonstrate that the iron content of silicate grains has to be very low ( large condensation radius for iron - rich silicates , combined with low condensation efficiency , see discussion in previous section ) , and that the wind even for quite extreme stellar parameters will consequently not be driven by silicate grains , in contrast to previous expectations . during recent years , 2d/3d dynamical atmosphere and wind models for agb stars
have emerged , in addition to the spherically symmetric models discussed above .
the computational effort behind such models is considerable , and several simplifications have to be introduced in the description of physical processes .
nevertheless , investigating the effects of intrinsically three - dimensional phenomena like convection or flow instabilities on mass loss , seems a timely project in view of recent interferometric observations which indicate deviations from spherical symmetry .
woitke ( 2006b ) presented 2d ( axisymmetric ) dust - driven wind models , including time - dependent dust formation and grey radiative transfer .
he studied how instabilities in the dust formation process create intricate patterns in the circumstellar envelope , but without taking the pulsation of the central star into consideration .
freytag & hfner ( 2003 , 2007 ) , on the other hand , investigate the effects of giant convection cells and of the resulting shock waves in the atmosphere on time - dependent dust formation in the framework of 3d rhd star - in - a - box models .
the atmospheric patterns created by convective motions are found to be reflected in the circumstellar dust distribution , due to the strong sensitivity of grain formation to temperatures and gas densities , as discussed above .
the well - known dichotomy between m - type and c - type agb stars , as observed in molecular spectra , may have an even more drastic consequence for their mass loss mechanism . while advances in modelling , in particular the introduction of frequency - dependent radiative transfer in time - dependent dynamical models , have improved agreement between models and observations of c - rich agb stars , the opposite seems to be true for the o - rich case .
recent models of m - type stars combining time - dependent dust formation with frequency - dependent radiative transfer demonstrate that silicate grains forming in such environments will be extremely iron - poor , resulting in too low opacities to drive a wind .
these qualitative differences can be understood with simplified analytical considerations , as discussed here , causing serious doubt about the validity of the dust - driven wind scenario for m - type agb stars , at least in its most simple form .
the mystery is deepened by the fact that the observed wind characteristics for both types of stars are rather similar ( see , e.g. , olofsson 2004 , ramstedt et al .
2006 ) , which hints at a common mass loss mechanism .
in addition , alternative scenarios discussed in the literature may have serious difficulties explaining the formation of considerable amounts of dust as a by - product in an outflow driven by a different force .
non - equilibrium dust condensation is very sensitive to the prevailing thermodynamical conditions , and restricted to a relatively narrow zone close to the star , putting strong constraints on potential driving forces . in this situation
the role of observers should not be underestimated .
any observations which can narrow down the possible range of conditions in the wind acceleration zone are of great importance for solving this problem . | the intricate interplay of atmospheric shock waves and a complex , variable radiation field with non - equilibrium dust formation presents a considerable challenge to self - consistent modelling of atmospheres and winds of agb stars .
nevertheless it is clear that realistic models predicting mass loss rates and synthetic spectra are crucial for our understanding of this important phase of stellar evolution . while a number of questions are still open , significant progress has been achieved in recent years . in particular , self - consistent models for atmospheres and winds of c -
stars have reached a level of sophistication which allows direct quantitative comparison with observations . in the case of stars with c / o @xmath0 ,
however , recent work points to serious problems with the dust - driven wind scenario .
this contribution analyzes the basic ingredients of this scenario with analytical estimates , focusing on dust formation , non - grey effects , and differences between c - rich and o - rich environments , as well as discussing the status of detailed dynamical wind models and current trends in this field . |
ji matouek s 2003 book `` _ _ using the borsuk ulam theorem .
lectures on topological methods in combinatorics and geometry _ _ '' @xcite is an inspiring introduction to the use of equivariant methods in discrete geometry .
its main tool is the borsuk
ulam theorem , and its generalization by albrecht dold , which says that there is no equivariant map from an @xmath5-connected space to an @xmath5-dimensional finite complex that is equivariant with respect to a non - trivial finite group acting freely .
one of the main applications of this technology in matouek s book was a proof for brny s `` topological tverberg conjecture '' on @xmath4-fold intersections in the case when @xmath4 is a prime , originally due to imre brny , senya shlosman and andrs szcs @xcite .
this conjecture claimed that for any continuous map @xmath6 , when @xmath3 , there are @xmath4 points in disjoint faces of the simplex @xmath1 that @xmath7 maps to the same point in @xmath2 .
the topological tverberg conjecture was extended to the case when @xmath4 is a prime power by murad zaydin in an unpublished paper from 1987 @xcite .
this can not , however , be achieved via the dold theorem , since in the prime power case the group actions one could use on the codomain are not free .
so more advanced methods are needed , such as the serre spectral sequence for the borel construction and the fadell
husseini index . in this paper
we present the area about and around the topological tverberg conjecture , with complete proofs for all of the results , which include the prime power case of the topological tverberg conjecture .
zaydin in 1987 had not only proven the topological tverberg theorem for prime power @xmath4 , but he had also shown , using equivariant obstruction theory , that the approach fails when @xmath4 is not a prime power : in this case the equivariant map one looks for does exist . in a spectacular recent development ,
isaac mabillard and uli wagner @xcite @xcite have developed an @xmath4-fold version of the classical `` whitney trick '' ( cf .
@xcite ) , which yields the failure of the generalized van kampen flores theorem when @xmath8 is not a prime power .
then florian frick observed that this indeed implies the existence of counterexamples to the topological tverberg conjecture @xcite @xcite by a lemma of gromov @xcite that is an instance of the constraint method of blagojevi , frick , ziegler ( * ? ? ?
* lemma4.1(iii ) and lemma4.2 ) .
( see @xcite for a popular rendition of the story . ) the tverberg theorem from 1966 @xcite and its conjectured extension to continuous maps ( the topological tverberg conjecture ) have seen a great number of variations and extensions , among them `` colored '' variants as well as versions with restricted dimensions ( known as generalized van kampen
flores theorems ) .
although many of these were first obtained as independent results , sometimes with very similar proof patterns , our presentation shows that there are many easy implications between these results , using in particular the `` constraint method '' applied to `` unavoidable complexes , '' as developed by the present authors with florian frick @xcite .
( mikhail gromov @xcite had sketched one particular instance : the topological tverberg theorem for maps to @xmath9 implies a generalized van kampen
flores theorem for maps to @xmath10 . )
thus we can summarize the implications in the following scheme , which shows that all further main results follow from two sources , the topological tverberg theorem for prime powers , and the optimal colored tverberg theorem of the present authors with benjamin matschke @xcite , which up to now even for affine maps is available only for the prime case : @xmath11\ar[ddl]|!{[dl];[dl]}\hole \ar[drr]\ar[dddl]|!{[dl];[dl]}\hole|!{[ddl];[ddl]}\hole & & & \text{optimal colored tverberg for } p\ar[dr]\ar[dl ] & & \\
\text{generalized van kampen -- flores for } p^n & & & \text{topological tverberg for } p & & \text{b\'ar\'any -- larman for } p-1\\ \text{colored van kampen -- flores for } p^n\ar[drrrr]\ar[d ] & & & & \\
\text{weak colored tverberg for } p^n\ar[drrrr ] & & & & \text{vre\'cica--\v{z}ivaljevi\'c colored tverberg for } p^n\text { type b}\ar[d]\\ & & & & \text{\v{z}ivaljevi\'c -- vre\'cica colored tverberg for } p^n\text { type a } } \ ] ] our journey in this paper starts with radon s 1921 theorem and its topological version , in section [ sec : the begining ] . here
the borsuk
ulam theorem is all that s needed . in section [ sec : the tverberg theorem ] we state the topological tverberg conjecture and first prove it in the prime case ( with a proof that is close to the original argument by brny , shlosman and szcs ) , and then for prime powers and this is where we go `` beyond the borsuk
ulam theorem . ''
implications and corollaries of the topological tverberg theorem are developed in section [ sec : corollaries of the topological tverberg theorem]so that s where we put constraints , and `` add color . '' in section [ sec : counterexamples to the topological tverberg conjecture ] we get to the counterexamples . and finally in section [ sec : the optimal colored tverberg theorem ]
we discuss the `` optimal colored tverberg conjecture , '' which is a considerable strengthening of tverberg s theorem , but up to now has been proven only in the prime case .
a summary of the main topological concepts and tools used in this paper is given at the end in the form of a dictionary , where a reference to the dictionary in the text is indicated by concept@xmath12 . _
acknowledgements_. we are grateful to alexander engstrm and florian frick for excellent observations on drafts of this paper and many useful comments .
we want to express our gratitude to peter landweber for his continuous help and support in improving this manuscript .
one of the first cornerstone results of convex geometry is a 1921 theorem of johann radon about overlapping convex hulls of points in a euclidean space .
let @xmath2 be a @xmath13-dimensional euclidean space .
let @xmath14 be points in @xmath2 and let @xmath15 be non - negative real numbers that sum up to @xmath16 , that is , @xmath17 and @xmath18 .
convex combination _ of the points @xmath14 determined by the scalars @xmath15 is the following point in @xmath2 : @xmath19 for a subset @xmath20 of @xmath2 we define the _ convex hull _ of @xmath20 , denoted by @xmath21 , to be the set of all convex combinations of finitely many points in @xmath20 : @xmath22 now radon s theorem can be stated as follows and proved using elementary linear algebra .
let @xmath2 be a @xmath13-dimensional euclidean space , and let @xmath23 be a subset with ( at least ) @xmath24 elements .
then there are disjoint subsets @xmath25 and @xmath0 of @xmath26 with the property that @xmath27 let @xmath28 .
the homogeneous system of @xmath29 linear equations in @xmath24 variables @xmath30 has a non - trivial solution , say @xmath31 .
denote @xmath32 then @xmath33 while @xmath34 and @xmath35 , and @xmath36 where @xmath37 .
consequently , the following point is in the intersection of convex hulls of @xmath25 and @xmath0 : @xmath38 in order to reformulate radon s theorem we recall the notion of an affine map .
a map @xmath39 defined on a subset @xmath40 is _ affine _ if for every @xmath41 , @xmath42 , and @xmath43 with @xmath18 and @xmath44 , we have @xmath45 here and in the following let @xmath46 be the _
standard @xmath47-dimensional simplex _ : this simplex given as the convex hull of the standard basis of @xmath48 has the disadvantage of not being full - dimensional in @xmath49 , but it has the extra advantage of being obviously symmetric ( with symmetry given by permutation of coordinates ) . with this , radon s theorem can be restated as follows .
let @xmath50 be an affine map .
then there are disjoint faces @xmath51 and @xmath52 of the @xmath53-simplex @xmath54 with the property that @xmath55 with this version of radon s theorem at hand , it is natural to ask : _ would radon s theorem still hold if instead of an affine map we consider an arbitrary continuous map @xmath56 ? _ the question we have just asked was answered in 1979 by ervin bajmczy and imre brny @xcite , using the borsuk
ulam theorem .
[ th : topological radon theorem ] let @xmath50 be any continuous map .
then there are two disjoint faces @xmath51 and @xmath52 of @xmath57 whose images under @xmath7 intersect , @xmath55 let @xmath58 be the standard simplex .
consider the subcomplex @xmath26 of the polyhedral complex @xmath59 given by @xmath60 the group @xmath61 acts freely on @xmath26 by @xmath62 .
let us assume that the theorem does not hold .
then there exists a continuous map @xmath50 such that @xmath63 for all @xmath64 .
consequently the map @xmath65 given by @xmath66 is continuous and @xmath67-equivariant , where the action on @xmath68 , the unit sphere in @xmath2 , is the standard antipodal action .
next we define a continuous @xmath67-equivariant map from a @xmath13-sphere to @xmath26 . for this we do not use the standard @xmath13-sphere , but the unit sphere @xmath69 in the hyperplane @xmath70 , that is , @xmath71 this representation of the @xmath13-sphere also has the standard antipodal @xmath67-action .
the map @xmath72 is defined by @xmath73 where @xmath74 .
this is easily checked to be well - defined and continuous ; the image point lies in the cell @xmath75 of the complex @xmath59 . the composition map @xmath76 yields a continuous @xmath67-equivariant map from a free @xmath13-sphere to a free @xmath77-sphere that contradicts the borsuk ulam theorem@xmath12 .
thus the theorem holds .
the topological radon theorem guarantees that for every continuous map @xmath78 there exist two pairwise disjoint faces whose @xmath7-images overlap .
it is natural to ask : _ is it possible to say something about the dimension of the disjoint faces whose @xmath7-images intersect ? _ in the spirit of poincar s classification of mathematical problems ( * ? ? ?
* lec.1 ) this _ binary problem _ has a quick answer _ no _ , but if understood as an _ interesting problem _ it has an answer : if we are willing to spend an extra vertex / dimension , meaning , put the simplex @xmath79 in place of @xmath57 , we get the following theorem from the 1930s of egbert r. van kampen and antonio flores @xcite @xcite .
let @xmath80 be an even integer , and let @xmath81 be a continuous map .
then there are disjoint faces @xmath82 of dimension at most @xmath83 whose images under @xmath7 intersect , @xmath55 let @xmath84 be a continuous map defined by @xmath85 where @xmath86 denotes the @xmath87-skeleton of the simplex @xmath79 , and @xmath88 is the distance of the point @xmath89 from the subcomplex @xmath86 . observe that if @xmath90 and @xmath91 , then the simplex @xmath92 belongs to the subcomplex @xmath86 .
now the topological radon theorem can be applied to the continuous map @xmath84 .
it yields the existence of points @xmath93 and @xmath94 , with @xmath95 , such that @xmath96 if one of the simplices @xmath51 , or @xmath52 , would belong to @xmath86 , then @xmath97 implying that both @xmath51 and @xmath52 belong to @xmath86 , which would concludes the proof of the theorem . in order to prove that at least one of the faces @xmath51 and @xmath52 belongs to @xmath86 , note that these are two disjoint faces of the simplex @xmath79 , which has @xmath98 vertices , so by the pigeonhole principle one of them has at most @xmath99 vertices .
the proof we have presented is an example of the _ constraint method _ developed in @xcite .
an important message of this proof is that the van kampen flores theorem is a corollary of the topological radon theorem .
it is clear that we could have considered a continuous map @xmath7 defined only on the @xmath83-skeleton .
all the results we presented so far have always claimed something about intersections of the images of two disjoint faces @xmath100 , which we refer to as @xmath101-fold overlap , or intersection . _ what about @xmath4-fold overlaps , for @xmath102 ? _
in 1964 , freezing in a hotel room in manchester , the norwegian mathematician helge tverberg proved the following @xmath4-fold generalization of radon s theorem @xcite .
it had been conjectured by bryan birch in 1954 , who had established the result in the special case of dimension @xmath103 @xcite .
the case @xmath104 is easy , see below .
( see @xcite for some of the stories surrounding these discoveries . )
let @xmath105 and @xmath106 be integers , @xmath107 , and let @xmath108 be an affine map .
then there exist @xmath4 pairwise disjoint faces @xmath109 of the simplex @xmath1 whose @xmath7-images overlap , @xmath110 any collection of @xmath4 pairwise disjoint faces @xmath109 of the simplex @xmath1 having property is called a _ tverberg partition _ of the map @xmath7 .
the dimension of the simplex in the theorem is optimal , it can not be decreased . to see this
consider the affine map @xmath111 given on the vertices of @xmath112 by @xmath113 where @xmath114 is an affinely independent set in @xmath2 , e.g. , @xmath115 .
for each vertex of the simplex @xmath116 the cardinality of its preimage is @xmath117 @xmath118 and so the map @xmath119 has no tverberg partition .
even more is true : any affine map @xmath111 that is in general position can not have a tverberg partition . as in the case of radon
s theorem it is natural to ask : _ would the tverberg theorem still hold if instead of an affine map @xmath120 we would consider an arbitrary continuous map ? _
this was first asked by brny in a 1976 letter to tverberg . in may of 1978
tverberg posed the question in oberwolfach , stating it for a general @xmath0-polytope in place of the @xmath0-simplex , see @xcite .
( the problem for a general @xmath0-polytope can be reduced to the case of the @xmath0-simplex by a theorem of grnbaum : every @xmath0-polytope as a cell complex is a refinement of the @xmath0-simplex @xcite . ) thus , the topological tverberg conjecture started its life in the late 1970s .
let @xmath105 and @xmath106 be integers , @xmath107 , and let @xmath108 be a continuous map .
then there exist @xmath4 pairwise disjoint faces @xmath109 of the simplex @xmath1 whose @xmath7-images overlap , @xmath121 the case @xmath122 of the topological tverberg conjecture amounts to the topological radon theorem , so it holds .
the topological tverberg conjecture is also easy to verify for @xmath104 , as follows .
let @xmath106 be an integer , and let @xmath123 be a continuous map .
then there exist @xmath4 pairwise disjoint faces @xmath109 of the simplex @xmath124 whose @xmath7-images overlap , @xmath125 let @xmath123 be continuous .
sort the vertices of the simplex @xmath126 such that @xmath127 .
then the collection of @xmath117 edges and one vertex of @xmath124 @xmath128 , \ \sigma_2=[{\mathbf{e}}_{\pi(2)},{\mathbf{e}}_{\pi(2r-2 ) } ] , \ldots , \ \sigma_{r-1}=[{\mathbf{e}}_{\pi(r-1)},{\mathbf{e}}_{\pi(r+1 ) } ] , \
\sigma_{r}=\{{\mathbf{e}}_{r}\}\ ] ] is a tverberg partition for the map @xmath7 . at first glance
the topological tverberg conjecture is a
_ binary problem _ in the poincar classification of mathematical problems . to our surprise
it is safe to say , at this point in time , that the topological tverberg conjecture was one of the most _ interesting problems _ that shaped interaction between geometric combinatorics on one hand and algebraic and geometric topology on the other hand for almost four decades . after settling the topological tverberg conjecture for @xmath104 and @xmath122 we want to advance .
let @xmath105 and @xmath106 be integers , and let @xmath107 .
our effort to handle the topological tverberg conjecture starts with an assumption that there is a counterexample to the conjecture with parameters @xmath13 and @xmath4 .
thus there is a continuous map @xmath108 such that for every @xmath4-tuple @xmath109 of pairwise disjoint faces of the simplex @xmath1 their @xmath7-images do not intersect , that is , @xmath129 in order to capture this feature of our counterexample @xmath7 we parametrize all @xmath4-tuples of pairwise disjoint faces of the simplex @xmath1 .
this can be done in two similar , but different ways .
@xmath4-fold @xmath101-wise deleted product@xmath12 _ of a simplicial complex @xmath130 is the cell complex @xmath131 where @xmath109 are non - empty faces of @xmath130 .
the symmetric group @xmath132 acts ( from the left ) on @xmath133 by @xmath134 for @xmath135 and @xmath136 .
this action is free due to the fact that if @xmath136 then @xmath137 for all @xmath138 .
we have seen a particular instance before : the complex @xmath26 that we used in the proof of the topological radon theorem [ th : topological radon theorem ] was @xmath139 . for more details on the deleted product construction
see for example ( * ? ? ?
* sec.6.3 ) . in the case
when @xmath130 is a simplex the topology of the deleted product @xmath133 is known from the following result of brny , shlosman and szcs ( * ? ?
* lem.1 ) .
[ th : connectivity of deleted product ] let @xmath0 and @xmath4 be positive integers with @xmath140 . then @xmath141 is an @xmath142-dimensional and @xmath143-connected cw complex .
a typical face of the cw complex @xmath141 is of the form @xmath144 , where @xmath109 are pairwise disjoint simplices .
consequently , the number of vertices of these simplices together can not exceed @xmath145 , or in the language of dimension @xmath146 the equality is attained when all the vertices are used , this is when @xmath109 is a maximal face of dimension @xmath147 thus @xmath141 is an @xmath142-dimensional cw complex .
for @xmath148 the deleted product @xmath141 is the @xmath149-dimensional simplicial complex @xmath150 $ ] and the statement of the theorem holds .
thus , we can assume that @xmath151 . for @xmath151
we establish the connectivity of the deleted product of a simplex by induction on @xmath4 making repeated use of the following classical 1957 theorem of stephen smale ( * ? ? ?
* main thm . ) : _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ * smale s theorem . * _ let @xmath26 and @xmath152 be connected , locally compact , separable metric spaces , and in addition let @xmath26 be locally contractible .
let @xmath153 be a continuous proper map , that is , any preimage of a compact set is compact .
if for every @xmath154 the preimage @xmath155 is locally contractible and @xmath5-connected , then the induced homomorphism @xmath156 is an isomorphism for all @xmath157 , and is an epimorphism for @xmath158 .
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ recall that @xmath159 denotes the standard simplex , whose vertices @xmath160 form the standard basis of @xmath161 .
the induction starts with @xmath162 and the theorem claims that the simplex @xmath1 , a contractible space , is @xmath163-connected , which is obviously true . in the case
@xmath122 consider the surjection @xmath164 given by the projection on the first factor .
any point @xmath165 of the @xmath163-skeleton @xmath166 of the simplex @xmath1 lies in the relative interior of a face , @xmath167 \}\big)\ ] ] where @xmath168 .
let us denote the complementary set of vertices by @xmath169 and its convex hull by @xmath170 .
the fiber of the projection map @xmath171 over @xmath165 is given by @xmath172 and consequently it is contractible . by smale
s theorem the projection @xmath171 induces an isomorphism between homotopy groups . since we are working in the category of cw complexes the whitehead theorem (
* thm.11.2 ) implies a homotopy equivalence of @xmath173 and @xmath166 .
the @xmath163-skeleton of a simplex is @xmath174-connected and thus the theorem holds in the case @xmath122 .
for the induction hypothesis assume that @xmath175 is @xmath176-connected for all @xmath177 .
in the induction step we want to prove that @xmath178 is @xmath179-connected .
now consider the projection onto the first @xmath47 factors , @xmath180 since @xmath181 is @xmath182-connected by induction hypothesis , its @xmath182-skeleton @xmath183 is @xmath179-connected .
for a typical point of the codomain we have that @xmath184 \}\big ) \times\cdots\times { \operatorname{relint}}\big({{\rm conv}}\{{\mathbf{e}}_{i } : i\in t_k\subseteq [ n+1 ] \}\big),\ ] ] where @xmath185 for all @xmath186 , and @xmath187 . as before , consider the complementary set of vertices @xmath188 and its convex hull @xmath189 .
the fiber of the projection map @xmath190 over @xmath191 is given by @xmath192 so it is contractible . again
smale s theorem applied to the projection @xmath190 induces an isomorphism between homotopy groups of @xmath178 and @xmath183 .
moreover , the whitehead theorem implies that these spaces are homotopy equivalent . since , @xmath183 is @xmath179-connected we have concluded the induction step and the theorem is proved .
our proof of theorem [ th : connectivity of deleted product ] may be traced back to a proof in the lost preprint version of the paper @xcite .
indeed , in the published version the first sentence of ( * ? ? ? * proof of lem.1 ) says : _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ for this elementary proof we are indebted to the referee .
our original proof used the leray spectral sequence . _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ here we used smale s theorem in place of the leray spectral sequence argument .
let @xmath130 be a simplicial complex .
@xmath4-fold @xmath47-wise deleted join@xmath12 _ of the simplicial complex @xmath130 is the simplicial complex @xmath193)\ , { { \rm card\,}}i\geq k \rightarrow \bigcap_{i\in i } \sigma_i=\emptyset \big\},\ ] ] where @xmath194 are faces of @xmath130 , including the empty face .
thus in the case @xmath195 we have @xmath196 the symmetric group @xmath132 acts ( from the left ) on @xmath197 as follows @xmath198 where @xmath199 and @xmath200 .
this action is free only in the case when @xmath195 .
[ example : deleted product and joins ] ( compare figure [ fig3 : examples ] . ) let @xmath201 be the @xmath16-simplex
. then @xmath202 while @xmath203 .
for @xmath204 we have that @xmath202 , and @xmath205 is a disjoint union of two intervals . if @xmath206 $ ] then @xmath203 . when @xmath207 $ ] , the deleted join @xmath208 is the @xmath209 chessboard complex@xmath12 , which is denoted by @xmath210 . and
@xmath205 for @xmath201 and @xmath206 $ ] . ]
the following lemma establishes the commutativity of the join and the deleted join operations on simplicial complexes .
we state it for @xmath47-wise deleted joins and prove it here only for @xmath101-wise deleted joins . for more details and insight consult the sections _ `` deleted products good '' _ and _ `` deleted joins better '' _ in matouek s book , ( * ? ? ?
* sections.5.4 and 5.5 ) .
[ lemma : commutativity of join and deleted join ] let @xmath130 and @xmath211 be simplicial complexes , and let @xmath212 and @xmath213 be integers .
there exists an isomorphism of simplicial complexes : @xmath214 we give a proof only for the case @xmath195 .
let @xmath194 and @xmath215 be simplices in @xmath130 and @xmath211 , respectively , such that @xmath216 and @xmath217 for all @xmath138 .
in addition , since the simplicial complexes @xmath130 and @xmath211 have disjoint vertex sets , we get that @xmath218 as well for all @xmath219 and @xmath220 .
thus , for all @xmath138 we obtain an equivalence : @xmath221 it induces a bijection between the following simplices of @xmath222 and @xmath223 by : @xmath224 a direct consequence of the previous lemma is the following useful fact .
[ lemma : deleted join of simplex ] let @xmath106 and @xmath225 be integers
. then @xmath226^{*(n+1)}$ ] , @xmath227 .
@xmath228^{*(n+1)})^{*r}_{\delta(k)}\cong ( [ 1]^{*r}_{\delta(k)})^{*(n+1)}\cong ( { \operatorname{sk}}_{k-2}(\delta_{r-1}))^{*(n+1)}$ ] .
thus the @xmath4-fold @xmath101-wise deleted join of an @xmath0-simplex @xmath229 is an @xmath0-dimensional and @xmath163-connected simplicial complex .
recall that , at the beginning of section [ subsec : equivariant topology steps in ] , we have fixed integers @xmath105 and @xmath106 , and in addition we assumed the existence of the continuous map @xmath108 that is a counterexample to the topological tverberg theorem .
define continuous maps induced by @xmath7 in the following way : the _ product map _ is @xmath230 the _ join map _
is @xmath231 the codomains @xmath232 and @xmath233 of the maps @xmath234 and @xmath235 are equipped with the action of the symmetric group @xmath132 given by permutation of the corresponding @xmath4 factors , that is @xmath236 for @xmath237 and @xmath238 . then both maps @xmath234 and @xmath235 are @xmath132-equivariant .
indeed , the following diagrams commute : @xmath239 ^ -{p_f}\ar[d]_{\pi\cdot } & ( f(x_1),\ldots , f(x_r))\ar[d]_{\pi\cdot}\\ ( x_{\pi^{-1}(1)},\ldots , x_{\pi^{-1}(r)})\ar[r]^-{p_f } & ( f(x_{\pi^{-1}(1)}),\ldots , f(x_{\pi^{-1}(r ) } ) ) , } \]]and @xmath240 ^ -{j_f}\ar[d]_{\pi\cdot } & ( \lambda_1 , \lambda_1 f(x_1))\oplus\cdots\oplus(\lambda_r , \lambda_r f(x_r))\ar[d]_{\pi\cdot}\\ \lambda_{\pi^{-1}(1 ) } x_{\pi^{-1}(1)}+\cdots+ \lambda_{\pi^{-1}(r)}x_{\pi^{-1}(r)}\ar[r]^-{j_f } & ( \lambda_{\pi^{-1}(1 ) } , \lambda_{\pi^{-1}(1 ) } f(x_{\pi^{-1}(1)}))\oplus\cdots\oplus(\lambda_{\pi^{-1}(r ) } , \lambda_{\pi^{-1}(r ) } f(x_{\pi^{-1}(r ) } ) ) . }
\]]the @xmath132-invariant subspaces @xmath241 of the codomains @xmath232 and @xmath233 , respectively , are called the _ thin diagonals_. the crucial property of the maps @xmath234 and @xmath235 , for a counterexample continuous map @xmath108 , is that @xmath242 indeed , the property of the map @xmath7 immediately implies that @xmath243 and @xmath244 are disjoint . for the second relation of assume that @xmath245 for some @xmath246 . then @xmath247 and consequently @xmath248 .
therefore , the maps @xmath234 and @xmath235 induce @xmath132-equivariant maps @xmath249 that , with an obvious abuse of notation , are again denoted by @xmath234 and @xmath235 , respectively .
let us denote by @xmath250 the compositions of projections and deformation retractions . here
@xmath251 denotes the orthogonal complement of the subspace @xmath252 in the relevant ambient real vector space , while @xmath253 denotes the unit sphere in the real vector space @xmath254 .
both maps @xmath255 and @xmath256 are @xmath132-equivariant maps with respect to the introduced actions .
furthermore , let @xmath257 be a vectors space with the ( left ) action of the symmetric group @xmath132 given by the permutation of coordinates .
then the subspace @xmath258 is an @xmath132-invariant subspace of dimension @xmath117 .
there is an isomorphism of real @xmath132-representations @xmath259 using this identification of @xmath132-representations the @xmath132-equivariant maps @xmath255 and @xmath256 , defined in , can be presented by @xmath260 finally we have the theorem we were looking for .
it will give us a chance to employ methods of algebraic topology to attack the topological tverberg conjecture .
[ th : criterion for topological tverberg ] let @xmath105 and @xmath106 be integers , and let @xmath107 .
if there exists a counterexample to the topological tverberg conjecture , then there exist @xmath132-equivariant maps @xmath261 if @xmath108 is a counterexample to the topological tverberg conjecture , then by composing maps from and we get @xmath132-equivariant maps @xmath262 now we have constructed our equivariant maps . the aim is to _ find as many @xmath4 s as possible such that an @xmath132-equivariant map _ @xmath263 _ can not exist_. for this we keep in mind that the @xmath132-action on @xmath141 is free , while for @xmath264 the @xmath132-action on @xmath265 is not free .
the story of the topological tverberg conjecture continues with a 1981 breakthrough of brny , shlosman and szcs @xcite .
they proved that in the case when @xmath4 is a prime , there is no @xmath266-equivariant map @xmath267 , and consequently no @xmath132-equivariant map can exist .
hence , theorem [ th : criterion for topological tverberg ] settles the topological tverberg conjecture in the case when @xmath4 is a prime .
we give a proof of this result relying on the following theorem of dold @xcite ( * ? ? ?
* thm.6.2.6 ) : _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ * dold s theorem . * _ let @xmath268 be a non - trivial finite group . for an @xmath5-connected @xmath268-space @xmath26 and at most @xmath5-dimensional free @xmath268-cw complex @xmath152
there can not be any continuous @xmath268-equivariant map @xmath269 .
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ let @xmath105 be an integer , let @xmath106 be a prime , @xmath107 , and let @xmath108 be a continuous map
. then there exist @xmath4 pairwise disjoint faces @xmath109 of the simplex @xmath1 whose @xmath7-images overlap , that is @xmath125 according to theorem [ th : criterion for topological tverberg ] it suffices to prove that there can not be any @xmath132-equivariant map @xmath267 .
let @xmath266 be the subgroup of @xmath132 generated by the cyclic permutation @xmath270 .
then it is enough to prove that there is no @xmath266-equivariant map @xmath267 .
for that we are going to use dold s theorem .
the assumption that @xmath4 is a prime implies that the action of @xmath266 on the the sphere @xmath271 is free .
now , since @xmath141 is an @xmath143-connected @xmath266-space , and @xmath271 is a free @xmath143-dimensional @xmath266-cw complex , the theorem of dold implies that a @xmath266-equivariant map @xmath267 can not exist .
the same argument yields that there can not be any @xmath132-equivariant map @xmath272 , when @xmath4 is a prime .
observe that for an application of the theorem of dold the nature of the group action on the domain is of no importance .
the next remarkable step followed a few years later . in 1987 in his landmark unpublished manuscript zaydin @xcite extended the result of brny , shlosman and szcs and proved that the topological tverberg conjecture holds for @xmath4 a prime power .
he proved even more and left the topological tverberg conjecture as a teaser for generations of mathematicians to come . but this story will come a bit later .
the first published proof of the topological tverberg theorem for @xmath4 a prime power appeared in a paper of aleksei yu .
volovikov @xcite ; see remark [ rem : volovikov ] . here
we give a proof of the topological tverberg theorem for prime powers based on a comparison of serre spectral sequences which uses a consequence of the localization theorem@xmath12 for equivariant cohomology@xmath12 ( * ? ? ?
* cor.1 , p.45 ) . for background on spectral sequences we refer to the textbooks by mccleary @xcite and by fomenko and fuchs @xcite .
[ th : topological tverberg theorem for prime power ] let @xmath105 be an integer , let @xmath106 be a prime power , @xmath107 , and let @xmath108 be a continuous map .
then there exist @xmath4 pairwise disjoint faces @xmath109 of the simplex @xmath1 whose @xmath7-images overlap , that is @xmath125 let @xmath105 be an integer , and let @xmath273 for @xmath274 a prime . by theorem
[ th : criterion for topological tverberg ] it suffices to prove that there can not be any @xmath132-equivariant map @xmath267 .
consider the elementary abelian group @xmath275 and the regular embedding @xmath276 , as explained in ( * ? ? ?
* ex.2.7 , p.100 ) .
it is given by the left translation action of @xmath275 on itself : to each element @xmath277 we associate the permutation @xmath278 from @xmath279 given by @xmath280 .
we identify the elementary abelian group @xmath275 with the subgroup @xmath281 of the symmetric group @xmath132 .
thus , in order to prove the non - existence of an @xmath132-equivariant map it suffices to prove the non - existence of a @xmath275-equivariant map @xmath267 .
our proof takes several steps ; the crucial ingredient is a comparison of serre spectral sequences . as it will be by contradiction ,
let us now assume that a @xmath275-equivariant map @xmath282 exists . *
( 1 ) * let @xmath283 denote the borel construction@xmath12 fiber bundle @xmath284 while @xmath285 denotes the borel construction fiber bundle @xmath286 then the map @xmath287 would induce the following morphism between fiber bundles @xmath283 and @xmath285 : @xmath288 ^ -{{\operatorname{id}}\times_{({\mathbb{z}}/p)^n } \varphi}\ar[d ] & & { \mathrm{e}}({\mathbb{z}}/p)^n\times_{({\mathbb{z}}/p)^n } s(w_r^{\oplus d})\ar[d]\\ { \mathrm{b}}({\mathbb{z}}/p)^n \ar[rr]^-{= } & & { \mathrm{b}}({\mathbb{z}}/p)^n . } \ ] ] this bundle morphism induces a morphism of associated cohomology serre spectral sequences : @xmath289 such that on the zero row of the second term @xmath290 is the identity . here for the morphisms we use the simplified notation @xmath291 . before calculating both spectral sequences we recall the cohomology of @xmath292 with coefficients in the field @xmath293 and denote it as follows : @xmath294 , & \deg t_{i}=1 \\ p>2 : & h^*({\mathrm{b}}\left(({\mathbb{z}}/p)^n\right);{\mathbb{f}}_p)=h^*(({\mathbb{z}}/p)^n;{\mathbb{f}}_p ) \cong { \mathbb{f}}_p[t_1,\ldots , t_n]\otimes \lambda [ e_1,\ldots , e_n ] , & \deg t_{i}=2,\,\deg e_{i}=1 , \end{array}\ ] ] where @xmath295 $ ] denotes the exterior algebra . * ( 2 ) * first , we consider the serre spectral sequence , with coefficients in the field @xmath293 , associated to the fiber bundle @xmath283 .
the @xmath296-term of this spectral sequence can be computed as follows : @xmath297 since by theorem [ th : connectivity of deleted product ] the deleted product @xmath141 is an @xmath142-dimensional , @xmath143-connected simplicial complex and consequently @xmath298 only for @xmath299 or @xmath300 .
thus , the only possibly non - zero differential of the spectral sequence is @xmath301 and therefore @xmath302 for @xmath303 . * ( 3 ) * the second serre spectral sequence , with coefficients in the field @xmath293 , we consider is associated to the fiber bundle @xmath285 . in this case
the fundamental group of the base space @xmath304 acts trivially on the cohomology @xmath305 . indeed , when @xmath306 the group @xmath307 can only act trivially on @xmath308 and on @xmath309 .
for @xmath274 an odd prime all elements of the group @xmath275 have odd order and therefore the action is trivial on the @xmath293 vector spaces @xmath310 and @xmath311 .
thus the @xmath296-term of this spectral sequence is of the form @xmath312 moreover , if @xmath313 denotes a generator then the @xmath143-row of the @xmath296-term is a free @xmath314-module generated by the element @xmath315 .
the only possible non - zero differential is @xmath316 consequently we have that @xmath296 and @xmath317 terms coincide , that is @xmath318 , @xmath143-row of the @xmath317-term is a free @xmath314-module generated by @xmath319 , @xmath320 , as all differentials , is an @xmath314-module morphism . therefore , the differential @xmath320 is zero if and only if @xmath321 furthermore , if @xmath322 , then @xmath323 .
hence , the projection map @xmath324 induces a monomorphism in cohomology @xmath325 now the following consequence of the localization theorem ( * ? ? ?
* cor.1 , p.45 ) , which in the case of finite groups holds only for elementary abelian groups , comes into play : _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ * theorem . * _ let @xmath274 be a prime , @xmath326 with @xmath327 , and let @xmath26 be a finite @xmath268-cw complex .
the fixed point set @xmath328 of the space @xmath26 is non - empty if and only if the map in cohomology @xmath329 , induced by the projection @xmath330 , is a monomorphism . _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ since the fixed point set @xmath331 of the sphere is empty , the theorem we just quoted implies that the map in cohomology @xmath325 is _ not _ a monomorphism .
consequently , the element @xmath332 is _ not _ zero . and @xmath333 and the morphism between them @xmath334 that is the identity between the @xmath149-rows up to the @xmath317-term .
] * ( 4 ) * finally , to reach a contradiction with the assumption that the @xmath275-equivariant map @xmath287 exists we track the element @xmath335 along the morphism of spectral sequences @xmath336 since the differentials in both spectral sequences are zero in all terms @xmath337 and @xmath338 for @xmath339 we have that @xmath340 is the identity for @xmath341 .
in particular , the morphism @xmath342 is still identity as it was in the second term , and so @xmath343 . passing to the @xmath142-term , with a slight abuse of notation
, we have that @xmath344)=[a],\ ] ] where @xmath345 $ ] denotes the class induced by @xmath346 in the appropriate @xmath347-term of the spectral sequences .
since @xmath348 and @xmath349 passing to the next @xmath350-term we reach a contradiction : @xmath351=a\neq 0,\ ] ] because the class of the element @xmath346 in @xmath352 vanishes ( domain of @xmath353 ) while in @xmath354 it does not vanish ( codomain of @xmath353 ) .
hence , there can not be any @xmath275-equivariant map @xmath267 , and the proof of the theorem is complete . in the language of the fadell
husseini index@xmath12 , as introduced in @xcite , we have computed that @xmath355 furthermore , we showed the existence of an element @xmath356 that has the property @xmath357 consequently @xmath358 and so the monotonicity property of the fadell
husseini index implies the non - existence of a @xmath275-equivariant map @xmath267 .
the element @xmath346 with the property can be specified explicitly .
it is the euler class of the vector bundle @xmath359 from the work of mann and milgram @xcite we get that for an odd prime @xmath274 @xmath360 where @xmath361 , while for @xmath306 we have that @xmath362 the square root in @xmath363 $ ] is not uniquely determined for an odd prime @xmath274 and @xmath13 odd : the factor @xmath364 accounts for an arbitrary square root being taken .
[ rem : volovikov ] volovikov , in his 1996 paper @xcite , proved the following extension of the topological tverberg theorem for continuous maps to manifolds : _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ * theorem . *
_ let @xmath105 be an integer , let @xmath106 be a prime power , and @xmath107
. for any topological @xmath13-manifold @xmath365 and any continuous map @xmath366 , there exist @xmath4 pairwise disjoint faces @xmath109 of the simplex @xmath1 whose @xmath7-images overlap , that is @xmath125 _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
over time many results were discovered that were believed to be substantial extensions or analogs of the topological tverberg theorem , such as the generalized van kampen flores theorem of karanbir sarkaria @xcite and aleksei volovikov @xcite , the colored tverberg theorems of rade ivaljevi and sinia vreica @xcite @xcite and pablo sobern s result on tverberg points with equal
barycentric coordinates @xcite .
it turned out only recently that the elementary idea of constraint functions together with the concept of `` unavoidable complexes '' introduced in @xcite transforms all these results into simple corollaries of the topological tverberg theorem .
well , if all these results are corollaries , is there any genuine extension of the topological tverberg theorem ?
the answer to this question will bring us to the fundamental work of brny and larman @xcite , and the optimal colored tverberg theorem @xcite from 2009 . but this will be the story of the final section of this paper . the first corollary we prove is the following generalized van kampen
flores theorem that was originally proved by sarkaria @xcite for primes and then by volovikov @xcite for prime powers .
the fact that this result can be derived easily from the topological tverberg theorem by adding an extra component to the map was first sketched by gromov in ( * ? ? ?
* sec.2.9c ) ; this can be seen as a first instance of the constraint method ( * ? ? ?
* thm.6.3 ) `` at work . ''
[ thm : genrelized_van_kampen_flores ] let @xmath367 be an integer , let @xmath4 be a prime power , let @xmath368 and @xmath369 , and let @xmath370 be a continuous map .
then there exist @xmath4 pairwise disjoint faces @xmath371 in the @xmath47-skeleton @xmath372 of the simplex @xmath1 whose @xmath7-images overlap , @xmath125 for the proof we use two ingredients , the topological tverberg theorem and the pigeonhole principle .
first , consider the continuous map @xmath373 defined by @xmath374 since @xmath375 and @xmath4 is a prime power we can apply the topological tverberg theorem to the map @xmath376 . consequently
, there exist @xmath4 pairwise disjoint faces @xmath109 with points @xmath377 such that @xmath378 , that is , @xmath379 one of the faces @xmath109 has to belong to @xmath372 .
indeed , if all the faces @xmath109 , which are disjoint , would not belong to @xmath372 , then the simplex @xmath1 should have at least @xmath380 vertices .
thus , since one of the faces is in the @xmath47-skeleton @xmath381 , and consequently @xmath382 , completing the proof of the theorem . in their 1990 study on halving lines and halving planes , brny , zoltan fredi and lszl lovsz @xcite realized a need for a colored version of the tverberg theorem .
the sentence from this paper _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ for this we need a colored version of tverberg s theorem .
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ opened a new chapter in the study of extensions of the tverberg theorem , both affine and topological . soon after , in 1992 , brny and david larman in @xcite formulated the colored tverberg problem and brought to light a conjecture that motivated the progress in the area for decades to come .
let @xmath383 be an integer and let @xmath384 be the set of vertices of the simplex @xmath1 .
coloring _ of the set of vertices @xmath384 by @xmath385 colors is a partition @xmath386 of @xmath384 , that is @xmath387 and @xmath388 for @xmath389 .
the elements of the partition @xmath386 are called _
color classes_. a face @xmath92 of the simplex @xmath1 is a _ rainbow _
face if @xmath390 for all @xmath391 .
the subcomplex of all rainbow faces of the simplex @xmath1 induced by the coloring @xmath386 will be denoted by @xmath392 and will be called the _ rainbow subcomplex_. there is an isomorphism of simplicial complexes @xmath393 .
let @xmath105 and @xmath106 be integers .
determine the smallest number @xmath394 such that for every affine map @xmath395 , and every coloring @xmath396 of the vertex set @xmath384 of the simplex @xmath397 by @xmath29 colors with each color of size at least @xmath4 , there exist @xmath4 pairwise disjoint rainbow faces @xmath371 of @xmath397 whose @xmath7-images overlap , @xmath125 a trivial lower bound for the function @xmath398 is @xmath399 .
brny and larman proved that the trivial lower bound is tight in the cases @xmath400 and @xmath401 , and presented a proof by lovsz for @xmath402 . furthermore , they conjectured the following equality .
let @xmath403 and @xmath367 be integers
. then @xmath404 .
now we present the proof of lovsz for the brny
larman conjecture in the case @xmath122 from the paper of brny and larman ( * ? ? ?
* thm.(iii ) ) .
[ th : lovasz ] let @xmath367 be an integer
. then @xmath402 .
let @xmath405 , and let @xmath406 be an affine map .
furthermore , consider a coloring @xmath396 of the vertex set @xmath384 of the simplex @xmath397 by @xmath29 colors where @xmath407 .
denote @xmath408 for @xmath409 .
the subcomplex of all rainbow faces of the simplex @xmath397 is the join @xmath410 . in this case , the rainbow subcomplex @xmath411 can be identified with the boundary of the cross - polytope @xmath412^{*(d+1 ) } \cong s^{d}$ ] . here
$ ] , as before , denotes the @xmath149-dimensional simplicial complex with two vertices .
the restriction map @xmath413^{*(d+1)}\rightarrow { \mathbb{r}}^d$ ] is a piecewise affine map , and therefore continuous .
the borsuk
ulam theorem yields the existence of a point @xmath414^{*(d+1)}\cong s^{d}$ ] on the sphere with the property that @xmath415 .
the point @xmath414^{*(d+1)}$ ] belongs to the relative interior of a unique simplex in the boundary of the cross - polytope @xmath412^{*(d+1)}$ ] , @xmath416 where @xmath417 and @xmath418 .
thus , @xmath419 . since the rainbow faces @xmath420 and @xmath421 are disjoint , and @xmath422 we have proved the theorem . in response to the work of brny and larman a modified colored tverberg problem was presented by ivaljevi and vreica in their influential paper @xcite from 1992 .
let @xmath105 and @xmath106 be integers .
determine the smallest number @xmath423 ( or @xmath424 ) such that for every affine ( or continuous ) map @xmath425 , and every coloring @xmath396 of the vertex set @xmath384 of the simplex @xmath426 by @xmath29 colors with each color of size at least @xmath427 , there exist @xmath4 pairwise disjoint rainbow faces @xmath371 of @xmath426 whose @xmath7-images overlap , that is @xmath125 observe that in the language of the function @xmath428 the brny
larman conjecture says that @xmath429 for all @xmath403 and @xmath367 .
furthermore , proving that @xmath430 does _ not _ imply @xmath431 , while proving @xmath429 would imply that @xmath432 . in order to address the modified problem ivaljevi and vreica needed to know the connectivity of chessboard complexes .
for that they recalled the following result of anders bj " orner , lov ' asz , vreica and ivaljevi ( * ?
* thm.1.1 ) .
its connectivity lower bound is best possible according to @xcite .
[ th : conn . of chessboard ]
let @xmath433 and @xmath327 be integers .
the chessboard @xmath434 is @xmath435-connected , where @xmath436 without loss of generality we can assume that @xmath437 .
the proof proceeds by induction on @xmath438 . in the case
@xmath439 the statement of the theorem is obviously true .
for @xmath440 we distinguish between two cases : if @xmath441 , then @xmath442 and @xmath443 is just a disjoint union of two edges , and if @xmath444 , then @xmath445 and @xmath446 is path connected .
let @xmath447 , and let us assume that the statement of the theorem holds for every chessboard @xmath448 where @xmath449 .
now we prove the statement of the theorem for the chessboard @xmath434 .
let @xmath450 for @xmath451 be a subcomplex of @xmath434 defined by @xmath452 the family of subcomplexes @xmath453 covers the chessboard @xmath434 .
moreover , each subcomplex @xmath450 is a cone over the chessboard @xmath454 , and therefore contractible . since , for @xmath455 $ ] we have that @xmath456,\ ] ]
the nerve@xmath12 @xmath457 of the family @xmath458 is homeomorphic to the boundary of an @xmath459-simplex @xmath460 .
thus , @xmath461 is @xmath462-connected .
furthermore , for @xmath455 $ ] with the property that @xmath463 the intersection @xmath464 is homeomorphism with the chessboard @xmath465 .
the induction hypothesis can be applied to each of these intersections .
therefore , @xmath466 now we will apply the following connectivity version of the nerve theorem@xmath12 due to bjrner , see ( * ? ? ?
* thm.10.6 ) . _ _ _ _ _ _ _ _ _ _ _ * theorem . * _ _ _ _ _ _ _ _ _ _ _ in the case of the covering @xmath458 of the chessboard @xmath434 , where @xmath467 , we have that for every face @xmath92 of the nerve @xmath457 the intersection @xmath468 is contractible when @xmath469 , and @xmath470 when @xmath463 , while the nerve @xmath457 of the family @xmath458 is @xmath462-connected with @xmath471 therefore , according to the nerve theorem applied for the cover @xmath458 the chessboard @xmath434 is @xmath435-connected .
this concludes the induction step .
the knowledge on the connectivity of the chessboard complexes was the decisive information both for the original proof of the ivaljevi and vreica colored tverberg theorem ( * ? ? ?
* thm.1 ) , which worked only for primes , and for the following version of the proof for prime powers ; see also the proof of ivaljevi ( * ? ? ?
* thm.3.2(2 ) ) .
[ th : colored tverberg of zv ] let @xmath105 be an integer , and let @xmath106 be a prime power . for every continuous map @xmath472 , and every coloring @xmath396 of the vertex set @xmath384 of the simplex @xmath426 by @xmath29 colors with each color of size at least @xmath473
, there exist @xmath4 pairwise disjoint rainbow faces @xmath371 of @xmath426 whose @xmath7-images overlap , that is @xmath125 in the language of the function @xmath474 the previous theorem yields the upper bound @xmath475 when @xmath4 is a prime power .
this bound implies the bound @xmath476 for arbitrary @xmath4 via bertrand s postulate .
let @xmath273 for @xmath274 a prime and @xmath477 .
let @xmath472 be a continuous map from a simplex @xmath426 whose set of vertices @xmath384 is colored by @xmath29 colors @xmath396 .
without loss of generality assume that @xmath478 .
in addition assume that the map @xmath7 is a counterexample for the statement of the theorem .
set @xmath479 and @xmath480 , so @xmath426 is an @xmath365-dimensional simplex .
now , the proof of the theorem will be presented in several steps . * ( 1 ) * the rainbow subcomplex of the simplex @xmath426 induced by the coloring @xmath396 in this case is @xmath481^{*(d+1)}.\ ] ] the @xmath4-fold @xmath101-wise deleted join of the rainbow subcomplex @xmath482 can be identified , with the help of lemma [ lemma : commutativity of join and deleted join ] and example [ example : deleted product and joins ] , as follows @xmath483^{*(d+1)}\big)^{*r}_{\delta(2)}\cong \big([2r-1]^{*r}_{\delta(2)}\big)^{*(d+1)}\cong ( \delta_{2r-1,r})^{*(d+1)}.\ ] ] the action of the symmetric group @xmath132 on the chessboard @xmath484 is assumed to be given by permutation of columns of the chessboard , that is @xmath485 for @xmath135 and @xmath486 a simplex in @xmath484 .
furthermore , the chessboard @xmath484 is an @xmath487-dimensional and according to theorem [ th : conn . of chessboard ]
an @xmath488-connected simplicial complex .
therefore @xmath489 * ( 2 ) * now , along the lines of section [ subsec : equivariant maps induced by f ] , the continuous map @xmath472 induces the join map @xmath490 both domain and codomain of the join map @xmath235 are equipped with the action of the symmetric group @xmath132 in such a way that @xmath235 is an @xmath132-equivariant map .
the deleted join of the rainbow complex @xmath491 is an @xmath132-invariant subcomplex of @xmath492 .
thus , the restriction map @xmath493 is also an @xmath132-equivariant map .
next consider the thin diagonal @xmath494 this is an @xmath132-invariant subspace of @xmath233 .
the key property of the map @xmath495 we have constructed for any counterexample continuous map @xmath496 , is that @xmath497 .
thus @xmath495 induces an @xmath132-equivariant map @xmath498 which we , with an obvious abuse of notation , again denote by @xmath495 .
furthermore , let @xmath499 be the composition of the appropriate projection and deformation retraction .
the map @xmath256 is @xmath132-equivariant .
recall from section [ subsec : equivariant maps induced by f ] that there is an isomorphism of real @xmath132-representations @xmath500 . here
@xmath258 and it is equipped with the ( left ) action of the symmetric group @xmath132 given by permutation of coordinates . after the identification of the @xmath132-representations the @xmath132-equivariant map @xmath256 defined in has the form @xmath501 thus we have proved that if there exists a counterexample map @xmath7 for the theorem , then there exists an @xmath132-equivariant map @xmath502 in the final step we reach a contradiction by proving that an @xmath132-equivariant map can not exist , concluding that a counterexample @xmath7 could not exist in the first place .
the proof of the non - existence of an equivariant map is following the footsteps of the proof of theorem [ th : topological tverberg theorem for prime power ] . *
( 3 ) * consider the elementary abelian group @xmath275 and its regular embedding @xmath276 .
now it suffices to prove the non - existence of a @xmath275-equivariant map @xmath503 . to prove the non - existence of such
a map assume the opposite : let @xmath504 be a @xmath275-equivariant map .
denote by @xmath283 the borel construction fiber bundle @xmath505 and by @xmath285 the following borel construction fiber bundle @xmath506 then the map @xmath287 induces the following morphism of fiber bundles @xmath507 ^ -{{\operatorname{id}}\times_{({\mathbb{z}}/p)^n } \varphi}\ar[d ] & & { \mathrm{e}}({\mathbb{z}}/p)^n\times_{({\mathbb{z}}/p)^n } s(w_r^{\oplus ( d+1)})\ar[d]\\ { \mathrm{b}}({\mathbb{z}}/p)^n \ar[rr]^-{= } & & { \mathrm{b}}({\mathbb{z}}/p)^n . } \ ] ] in turn , this morphism induces a morphism of corresponding serre spectral sequences @xmath508 with the property that on the zero row of the second term the induced map @xmath509 is the identity .
again we use simplified notation by setting @xmath291 . depending on the parity of the prime @xmath274
, we have that @xmath294 , & \deg t_{i}=1 \\ p>2 : & h^*({\mathrm{b}}\left(({\mathbb{z}}/p)^n\right);{\mathbb{f}}_p)=h^*(({\mathbb{z}}/p)^n;{\mathbb{f}}_p ) \cong { \mathbb{f}}_p[t_1,\ldots , t_n]\otimes \lambda [ e_1,\ldots , e_n ] , & \deg t_{i}=2,\,\deg e_{i}=1 , \end{array}\ ] ] where @xmath295 $ ] denotes the exterior algebra . *
( 4 ) * first we consider the serre spectral sequence , with coefficients in the field @xmath293 , associated to the fiber bundle @xmath283 . using the connectivity of @xmath491 derived in , we get that the @xmath296-term of this spectral sequence is @xmath510 consequently , the only possibly non - zero differential of this spectral sequence is @xmath511 and therefore @xmath512 for @xmath513 . and @xmath333 and the morphism between them @xmath334 that is the identity between the @xmath149-rows up to the @xmath514-term .
] * ( 5 ) * the serre spectral sequence , with coefficients in the field @xmath293 , associated to the fiber bundle @xmath285 is similar to the one appearing in the proof of theorem [ th : topological tverberg theorem for prime power ] .
briefly , the @xmath296-term of this spectral sequence is @xmath515 letting @xmath516 denote a generator , then the @xmath163-row of the @xmath296-term can be seen as a free @xmath314-module generated by @xmath517 .
thus the only possible non - zero differential is @xmath518 and it is completely determined by the image @xmath519 . as in the proof of theorem [ th : topological tverberg theorem for prime power ] , a consequence of the localization theorem implies that @xmath520 is _ not _ zero . * ( 6 ) * to reach the desired contradiction we track the element @xmath521 along the morphism of spectral sequences @xmath522 the differentials in both spectral sequences are zero in all terms @xmath337 and @xmath338 for @xmath523 .
thus , @xmath340 is an isomorphism for all @xmath524 . in particular , the morphism @xmath525 is the identity , as it was in the second term , and so @xmath526 .
when passing to the @xmath527-term , with a slight abuse of notation , we get @xmath528)=[b],\ ] ] where @xmath529 $ ] denotes the class induces by @xmath530 in the appropriate @xmath527-term of the spectral sequences .
since @xmath531 and @xmath532 we have reached a contradiction : @xmath533=b\neq 0.\ ] ] therefore , there can not be any @xmath275-equivariant map @xmath503 , and the proof of the theorem is complete .
as part of the proof of the previous theorem the following general criterion was derived .
[ cor : cstm for colored tverberg ] let @xmath534 be a coloring of the simplex @xmath426 by @xmath535 colors . if there is no @xmath132-equivariant map @xmath536 then for every continuous map @xmath472 there exist @xmath4 pairwise disjoint rainbow faces @xmath371 of @xmath426 whose @xmath7-images overlap , @xmath125 the proof of theorem [ th : colored tverberg of zv ] could have been written in the language of the fadell
husseini index @xcite .
the non - existence of an @xmath275-equivariant map @xmath503 would then follow from the observation that @xmath537 more precisely , we have computed that @xmath538 when @xmath478 .
actually we proved more : @xmath539 furthermore we have found an element @xmath540 with the property that @xmath541 and moreover @xmath542 the element @xmath530 with this property is the euler class of the vector bundle @xmath543 the work of mann and milgram @xcite allows us to specify the element @xmath530 completely : for @xmath274 an odd prime it is @xmath544 where @xmath361 , while for @xmath306 it is @xmath545 the square root in @xmath363 $ ] is not uniquely determined for an odd prime @xmath274 and @xmath13 odd .
thus we consider an arbitrary square root . combining these index computations we have that @xmath546 if a @xmath275-equivariant map @xmath503 exists , then the monotonicity property of the fadell
husseini index yields the inclusion @xmath547 which does not hold , as we just proved .
thus the @xmath275-equivariant map in question does not exist .
now observe the difference of dimensions in and compare the dimension of the element @xmath530 and the dimension of the group cohomology where the index of the join @xmath548 lives .
_ we could have proved more .
_ indeed , using the index computation we have that @xmath549 as long as @xmath550 .
we have just concluded that , if @xmath550 , then there is no @xmath275-equivariant map @xmath551 thus with corollary [ cor : cstm for colored tverberg ] we have proved the following `` colored tverberg theorem of type b '' ( * ? ? ? * thm.4 ) .
[ th : colored tverberg of zv - b ] let @xmath105 and @xmath552 be integers , @xmath107 , and let @xmath106 be a prime power .
for every continuous map @xmath472 , and every coloring @xmath553 of the vertex set @xmath384 of the simplex @xmath426 by @xmath47 colors , with each color of size at least @xmath473 and @xmath550 , there exist @xmath4 pairwise disjoint rainbow faces @xmath371 of @xmath426 whose @xmath7-images overlap , that is @xmath125 as we have just seen , the proof of the colored tverberg theorem of ivaljevi and vreica is in fact also a proof of a type b colored tverberg theorem .
is it possible that this proof hides a way to prove , for example the brny
larman conjecture ?
for this we would need to prove that for some or all @xmath4 and @xmath554 there is no @xmath132-equivariant map @xmath555 the connectivity of the chessboard @xmath556 is only @xmath557 and therefore the scheme of the proof of theorem [ th : colored tverberg of zv ] can not be used .
even worse , the complete approach fails , as the following theorem of blagojevi , matschke and ziegler ( * ? ? ?
* prop.4.1 ) shows that an @xmath132-equivariant map does exist .
[ th : exitence of equivariant map for bl ] let @xmath106 and @xmath105 be integers .
there exists an @xmath132-equivariant map @xmath558 for this we use equivariant obstruction theory , as presented by tammo tom dieck ( * ? ?
* sec.ii.3 ) .
let @xmath480 , @xmath559 , and let @xmath396 be a coloring of the vertex set of the simplex @xmath560 by @xmath29 colors of the same size @xmath4 , that is @xmath554 . as we know the deleted join @xmath491 of the rainbow complex is isomorphic to the join of chessboards @xmath561 .
the action of the symmetric group @xmath132 on the complex @xmath561 is not free .
the subcomplex of @xmath561 whose points have non - trivial stabilizers with respect to the action of @xmath132 can be described as follows : @xmath562 here for a @xmath268-space ( cw complex ) @xmath26 we we use notation @xmath563 for the subspace ( subcomplex ) of all points ( cells ) with non - trivial stabilizer , meaning that @xmath564 is a free @xmath268-space .
let @xmath565 be any continuous map .
as explained in section [ subsec : equivariant maps induced by f ] the map @xmath7 induces the join map given by @xmath566 since the rainbow complex @xmath491 is an @xmath132-invariant subcomplex of @xmath492 , the restriction @xmath493 is also an @xmath132-equivariant map .
moreover @xmath567 where , as before , @xmath568 .
thus the map @xmath495 induces an @xmath132-equivariant map @xmath569 composing this map with the @xmath132-equivariant retraction @xmath570 introduced in , we get a continuous @xmath132-equivariant map @xmath571 the @xmath487-dimensional chessboard complex @xmath556 equivariantly retracts to a subcomplex of dimension @xmath572 .
indeed , for each facet of @xmath556 there is an elementary collapse obtained by deleting all of its subfacets ( faces of dimension @xmath572 ) that contain the vertex in the @xmath4-th column . performing these collapses to all facets of @xmath556
, we get that @xmath556 collapses @xmath132-equivariantly to an @xmath488-dimensional subcomplex of @xmath556 .
consequently , the join @xmath573 equivariantly retracts to a subcomplex @xmath130 of dimension @xmath574 .
thus in order to prove the existence of an @xmath132-equivariant map @xmath575 it suffices to construct an @xmath132-equivariant map @xmath576 .
since @xmath577 , @xmath578 is @xmath163-simple@xmath12 and @xmath174-connected , and the groups where the obstructions would live are zero , the equivariant obstruction theory yields the existence of an @xmath132-equivariant map @xmath576 , provided that an @xmath132-equivariant map @xmath579 exists .
the subcomplex of all points with non - trivial stabilizer @xmath580 is a subcomplex of @xmath581 and therefore the map restricted to @xmath582 completes the argument . after this theorem an urgent question emerges : _ how are we going to handle the brny
larman conjecture ? _ an answer to this question will bring us to our last section and the optimal colored tverberg theorem . how many colored tverberg theorems can we get directly from the topological tverberg theorem without major topological machinery ? here is an answer given by ( * ? ? ?
* thm.5.3 ) .
[ th : colored_tverberg ] let @xmath367 be an integer , let @xmath4 be a prime power , @xmath583 , and let @xmath370 be a continuous map .
if the vertices of the simplex @xmath1 are colored by @xmath29 colors , where each color class has cardinality at most @xmath473 , then there are @xmath4 pairwise disjoint rainbow faces @xmath371 of @xmath1 whose @xmath7-images overlap , that is @xmath125 let @xmath384 be the set of vertices of the simplex @xmath1 and let @xmath396 be a coloring of @xmath384 where @xmath584 for all @xmath409 . to each color class @xmath585
we associate the subcomplex @xmath586 of @xmath587 defined by @xmath588 observe that the intersection @xmath589 is the subcomplex of all rainbow faces of @xmath1 with respect to the given coloring .
next consider the continuous map @xmath590 defined by @xmath591 since @xmath592 and @xmath4 is a prime power , we can apply the topological tverberg theorem to @xmath376 .
consequently there are @xmath4 pairwise disjoint faces @xmath109 with points @xmath377 such that @xmath378 , that is , @xmath593 now observe that for every subcomplex @xmath586 one of the faces @xmath109 is contained in it . indeed , if this would not hold then we would have @xmath594 , and consequently we would obtain the following contradiction : @xmath595 hence the distances , which were previously know to be equal , have to vanish , @xmath596 implying that @xmath597 for every @xmath598
. since @xmath599 are subcomplexes of @xmath1 and @xmath377 it follows that the faces @xmath600 belong to the subcomplex @xmath589 , that is , @xmath109 are rainbow faces .
a special case of the weak colored tverberg theorem we just proved , namely @xmath478 , yields @xmath601 for @xmath4 a prime power .
this is the colored tverberg theorem of ivaljevi and vreica presented in theorem [ th : colored tverberg of zv ] . along the lines of the previous theorem
we can prove the following colored van kampen flores theorem , where the number of color classes is at most @xmath29 .
[ th : colored_tverberg_b ] let @xmath367 be an integer , let @xmath4 be a prime power , let @xmath602 be an integer , and @xmath603 .
let @xmath370 be a continuous map .
if the vertices of the simplex @xmath1 are colored by @xmath47 colors , where each color class has cardinality at most @xmath473 , then there are @xmath4 pairwise disjoint rainbow faces @xmath371 of @xmath1 whose @xmath7-images overlap , @xmath125 let @xmath384 be the set of vertices of the simplex @xmath1 and let @xmath553 be a coloring where @xmath584 for all @xmath604 .
such a coloring exists because @xmath605 is equivalent to our assumption @xmath602 . to each color class @xmath585
we associate the subcomplex @xmath586 of @xmath587 defined as before by @xmath588 the subcomplex @xmath606 is a subcomplex of all rainbow faces of @xmath1 with respect to the given coloring .
consider the continuous map @xmath607 defined by @xmath608 since @xmath603 and @xmath4 is a prime power the topological tverberg theorem can be applied to the map @xmath376 .
therefore , there are @xmath4 pairwise disjoint faces @xmath109 with points @xmath377 such that @xmath378 , that is , @xmath609 now observe that every subcomplex @xmath586 contains one of the faces @xmath109 . indeed , if this would not hold then @xmath594 , and we would get the contradiction @xmath595 consequently the distances , which were previously known to be equal , have to vanish @xmath610 implying that @xmath611 for every @xmath598 .
since @xmath612 are subcomplexes and @xmath377 , it follows that @xmath613 that is , @xmath109 are rainbow faces .
the `` colored tverberg theorem of type b '' of vreica and ivaljevi , theorem [ th : colored tverberg of zv - b ] , is a particular case of this theorem , when the color classes have the same size .
the last corollary of the topological tverberg theorem that we present here is the topological version ( * ? ? ?
* thm.8.1 ) of a recent result by sobern ( * ? ? ?
* thm.1.1 ) ( * ? ? ?
* thm.1 ) .
let @xmath383 be an integer , let @xmath384 be the set of vertices of the simplex @xmath1 , and let @xmath386 be a coloring of @xmath384 .
every point @xmath89 in the rainbow subcomplex @xmath392 has a unique presentation in barycentric coordinates as @xmath614 where @xmath615 and @xmath616 for all @xmath617 .
two points @xmath614 and @xmath618 in the rainbow subcomplex @xmath392 have _ equal barycentric coordinates _ if @xmath619 for all @xmath620 .
[ theorem : equal_coeff ] let @xmath621 be an integer , let @xmath4 be a prime power , @xmath622 , and let @xmath370 be a continuous map .
if the vertices of the simplex @xmath1 are colored by @xmath623 colors where each colored class is of size @xmath4 , then there are points @xmath624 with equal barycentric coordinates that belong to @xmath4 pairwise disjoint rainbow faces @xmath371 of @xmath1 whose @xmath7-images coincide , that is @xmath625 let @xmath626 , and let @xmath386 be a coloring of the vertex set @xmath627 of the simplex @xmath1 . each point @xmath89 of the simplex @xmath1 can be uniquely presented in the barycentric coordinates as @xmath628 . for every color class @xmath585 , @xmath391
, we define the function @xmath629 by @xmath630 .
all functions @xmath631 are affine functions and @xmath632 for every @xmath633 .
now consider the function @xmath634 given by @xmath635 .
since @xmath636 , the topological tverberg theorem applied to the function @xmath376 implies that there exist @xmath4 pairwise disjoint faces @xmath109 of @xmath1 and @xmath4 points @xmath637 such that @xmath638 and @xmath639 for @xmath640 .
in addition , the equality @xmath641 implies that also @xmath642 .
assume now that @xmath643 for some @xmath644 and some @xmath391 .
then @xmath645 since @xmath646 .
consequently , @xmath647 implying that @xmath643 for all @xmath644 .
since @xmath648 and @xmath109 are pairwise disjoint it follows that each @xmath649 has precisely one vertex in the color class @xmath585 .
thus , repeating the argument for each color class we conclude that all faces @xmath109 are rainbow faces .
the immediate consequence of this fact is that @xmath650 , @xmath620 , are the barycentric coordinates of the point @xmath651 and so all the points @xmath652 have equal barycentric coordinates .
now we are going to get to a very recent piece of the topological tverberg puzzle : we show how counterexamples to the topological tverberg conjecture for any number of parts that his not a prime power were derived by frick @xcite @xcite from the remarkable works of zaydin @xcite and of mabillard and wagner @xcite @xcite , via a lemma of gromov @xcite that is an instance of the constraint method of blagojevi , frick , ziegler ( * ? ? ?
* lemmas4.1(iii ) and 4.2 ) .
first we present the second main result of zaydin s landmark manuscript ( * ? ? ?
* thm.4.2 ) .
[ th : ozaydin ] let @xmath105 be an integer , let @xmath653 be an integer , and let @xmath107 . if @xmath4 is not a prime power , then there exists an @xmath132-equivariant map @xmath654 in order to prove the existence of a continuous @xmath132-equivariant map @xmath267 we again use the equivariant obstruction theory .
since @xmath141 is an @xmath142-dimensional , @xmath143-connected free @xmath132-cw complex , and @xmath271 is a path - connected @xmath655-connected , @xmath143-simple @xmath132-space , we have that an @xmath132-equivariant map @xmath267 exists if and only if the primary obstruction@xmath12 @xmath656\in \mathcal{h}_{{\mathfrak s}_r}^{n - r+1}((\delta_n)^{\times r}_{\delta(2)},\pi_{n - r}s(w_r^{\oplus d}))\ ] ] vanishes .
the obstruction element @xmath657=[{\mathfrak o}_{{\mathfrak s}_r}^{n - r+1}(f)]$ ] does not depend on the particular @xmath132-equivariant map @xmath658 used to define the obstruction cocycle @xmath659 .
thus , in order to prove the existence of an @xmath132-equivariant map it suffices to prove that the obstruction element @xmath660 $ ] vanishes for some particular choice of @xmath7 .
let @xmath274 be a prime such that @xmath661 , and let @xmath662 denotes a @xmath274-sylow subgroup of @xmath132 . since @xmath4 is not a prime power each @xmath274-sylow subgroup of @xmath132 does not act transitively on the set @xmath150 $ ] , and hence the fixed point set @xmath663 is non - empty .
thus there exists a ( constant ) @xmath662-equivariant map @xmath267 , or equivalently the primary obstruction element with respect to @xmath662 vanishes , that is , @xmath664=[{\mathfrak
o}_{{\mathfrak s}_r^{(p)}}^{n - r+1}(f)]=0 $ ] . here
, the @xmath132-equivariant map @xmath7 is considered only as an @xmath662-equivariant map .
the restriction@xmath12 homomorphism @xmath665 is defined on the cochain level in ( * ? ? ?
* lem.5.4 ) . according to the definition of the obstruction cochain ( already on the cochain level )
the restriction homomorphism sends the obstruction cochain @xmath659 to the obstruction cochain @xmath666 .
consequently the same hold for obstruction elements @xmath667)=[{\mathfrak o}_{{\mathfrak s}_r^{(p)}}^{n - r+1}(f)].\ ] ] now , composing the restriction homomorphism with the transfer@xmath12 homomorphism @xmath668 also defined on the cochain level in ( * ? ? ?
* lem.5.4 ) , we get @xmath669\cdot [ { \mathfrak o}_{{\mathfrak s}_r}^{n - r+1}(f)]= { \operatorname{tr}}\circ{\operatorname{res}}([{\mathfrak o}_{{\mathfrak s}_r}^{n - r+1}(f)])= { \operatorname{tr}}([{\mathfrak o}_{{\mathfrak s}_r^{(p)}}^{n - r+1}(f)])= { \operatorname{tr}}(0)=0.\ ] ] finally , since @xmath670\cdot [ { \mathfrak o}_{{\mathfrak s}_r}^{n - r+1}(f)]=0 $ ] for every prime @xmath274 that divides the order of the group @xmath132 , it follows that the obstruction element @xmath660 $ ] must vanish , and the existence of an @xmath132-equivariant map is established .
[ cor : after ozaydin ] let @xmath105 be an integer , let @xmath653 be an integer that is not a prime power and let @xmath107 . for any free @xmath132-cw complex @xmath26 of dimension at most @xmath671
there exists an @xmath132-equivariant map @xmath672 the free @xmath132-cw complex @xmath26 has dimension at most @xmath671 , and the deleted product @xmath141 is @xmath143-connected , therefore there are no obstructions for the existence of an @xmath132-equivariant map @xmath673 .
next , let @xmath674 be an @xmath132-equivariant map whose existence was guaranteed by theorem [ th : ozaydin ] .
the composition @xmath675 yields the required @xmath132-equivariant map . with this theorem zaydin
_ only _ proved that the `` deleted product approach '' towards solving the topological tverberg conjecture fails in the case that @xmath4 is not a prime power .
_ what about the `` deleted join approach '' ? _
this question was discussed in ( * ? ? ?
* sec.3.4 ) .
let @xmath105 and @xmath4 be integers , and let @xmath107 . if @xmath4 is not a prime power , then there exists an @xmath132-equivariant map @xmath676 since @xmath653 is not a prime power theorem [ th : ozaydin ] implies the existence of an @xmath132-equivariant map @xmath677 now an @xmath132-equivariant map @xmath678 can be defined by @xmath679 where @xmath680 .
the function @xmath376 is well defined , continuous and @xmath132-equivariant . thus an @xmath132-equivariant map exists .
now we see that not only the `` deleted product approach '' fails if @xmath4 is not a prime power , but the `` deleted join approach '' fails as well . _ is this an indication that the topological tverberg theorem fails if the number of parts is not a prime power ? _ it is time to show that the topological tverberg conjecture fails in the case that @xmath4 is not a prime power .
this will be done following the presentation given in @xcite .
based on the work of mabillard and wagner @xcite @xcite we will prove that the generalized van kampen flores theorem for any @xmath4 that is not a prime power fails , as demonstrated by frick @xcite @xcite .
since , by the constraint method , the generalized van kampen flores theorem for fixed number of overlaps @xmath4 is a consequence of the topological tverberg theorem for the same number of overlaps @xmath4 , failure of the generalized van kampen flores theorem implies the failure of the topological tverberg theorem . [ thm : counter_van_kampen_flores ] let @xmath681 be an integer , and let @xmath653 be an integer that is not a prime power . for any integer @xmath682
there exists a continuous map @xmath683 such that for any @xmath4 pairwise disjoint faces @xmath371 from the @xmath684-skeleton @xmath685 of the simplex @xmath1 the corresponding @xmath7-images do not overlap , @xmath686 the deleted product @xmath687 is a free @xmath132-space of dimension at most @xmath688 . since @xmath4 is not a power of a prime , according to corollary [ cor : after ozaydin ] , there exists an @xmath132-equivariant map @xmath689 now we use the following result of mabillard and wagner ( * ? ? ?
* thm.3 ) ( * ? ? ?
* thm.7 ) , for which an alternative proof is given in @xcite .
skopenkov @xcite gives a user s guide .
_ _ _ _ _ _ _ _ _ _ _ * theorem . *
_ _ _ _ _ _ _ _ _ _ _ if we apply this result to the @xmath132-equivariant map @xmath119 in we get a continuous map @xmath690 with the property that for any collection of @xmath4 pairwise disjoint faces @xmath371 in @xmath685 the corresponding @xmath7-images do not overlap , @xmath691 thus we have proved that in the case when @xmath4 is not a prime power the generalized van kampen
flores theorem fails . as we have pointed out this means that the corresponding topological tverberg theorem also fails ( * ? ? ?
* thm.4.3 ) .
[ th : conterexample ] let @xmath681 and @xmath653 be integers , and let @xmath692 . if @xmath4 is not a prime power , then there exists a continuous map @xmath693 such that for any @xmath4 pairwise disjoint faces @xmath371 of @xmath1 the corresponding @xmath376 images do not overlap , @xmath694 since @xmath4 is not a power of a prime , theorem [ thm : counter_van_kampen_flores ] yields a continuous map @xmath683 such that for any @xmath4 pairwise disjoint faces @xmath371 in @xmath695 @xmath686 motivated by the proof of theorem [ thm : genrelized_van_kampen_flores ] we consider the function @xmath696 defined by @xmath697 we prove that the map @xmath376 fails the topological tverberg conjecture .
assume , to the contrary , that there are @xmath4 pairwise disjoint faces @xmath371 in @xmath1 and @xmath4 points @xmath698 such that @xmath378 .
consequently , @xmath699 next , at least one of the faces @xmath371 is in @xmath685 .
indeed , if all the faces @xmath700 would have dimension at least @xmath701 , then we would get the following contradiction : @xmath702 since one of the faces @xmath371 is in @xmath685 , all the distances vanish , meaning that @xmath703 therefore , all the faces @xmath371 belong to @xmath685 contradicting the choice of the map @xmath7 .
thus the map @xmath376 is a counterexample to the topological tverberg theorem .
the smallest counterexample to the topological tverberg theorem that can be obtained from theorem [ th : conterexample ] is a continuous map @xmath704 with the property that no six pairwise disjoint faces in @xmath705 have @xmath7-images that overlap . recently , using additional ideas , sergey avvakumov , isaac mabillard , arkadiy skopenkov and uli wagner @xcite have improved this to get counterexamples @xmath706 .
let us briefly recall the original colored tverberg problem posed by brny and larman in their 1992 paper @xcite , see section [ subsec : colored tverberg problem bl ] .
let @xmath105 and @xmath106 be integers . determine the smallest number @xmath394 such that for every affine ( continuous ) map @xmath395 , and every coloring @xmath396 of the vertex set @xmath384 of the simplex @xmath397 by @xmath29 colors with each color of size at least @xmath4 , there exist @xmath4 pairwise disjoint rainbow faces @xmath371 of @xmath397 whose @xmath7-images overlap , that is @xmath125 a trivial lower bound for the function @xmath398 is @xmath399 and it is natural to conjecture the following .
let @xmath403 and @xmath367 be integers
. then @xmath404 .
in section [ subsec : colored tverberg problem zv ] we tried to use the approach of ivaljevi and vreica to solve the brny
larman conjecture and we failed dramatically .
we hoped to prove that an @xmath132-equivariant map @xmath707 does not exist , but theorem [ th : exitence of equivariant map for bl ] gave us exactly the opposite , the existence of this map .
_ what can we do now ?
_ we change the question , and prove the non - existence of an @xmath708-equivariant map @xmath709\rightarrow s(w_{r+1}^{\oplus ( d+1)})\ ] ] instead ; here @xmath710 , and @xmath711 .
_ still , why should we be interested in such a result ?
_ [ th : non exitence > bl ] let @xmath403 and @xmath367 be integers . if there is no @xmath708-equivariant map @xmath712\rightarrow s(w_{r+1}^{\oplus ( d+1)}),\ ] ] then @xmath404 and @xmath713 .
let @xmath396 be a coloring of the vertices of the simplex @xmath426 with @xmath710 , and let @xmath714 be a continuous map . construct a simplex @xmath715 as a pyramid over @xmath426 , and let @xmath716 be the additional color class containing only the apex of the pyramid .
thus , @xmath717 is a coloring of the vertices of the simplex @xmath715 .
let us assume that an @xmath708-equivariant map @xmath718\rightarrow s(w_{r+1}^{\oplus ( d+1)})$ ] does not exist .
the non - existence of this map in combination with corollary [ cor : cstm for colored tverberg ] implies that there exist @xmath719 pairwise disjoint rainbow faces @xmath720 of the simplex @xmath715 whose @xmath7-images overlap , @xmath721 . without loss of generality
we can assume that @xmath722 .
then the faces @xmath723 are rainbow faces of the simplex @xmath426 with respect to the coloring @xmath396 and @xmath724 hence , @xmath713 and consequently @xmath404 .
the theorem we just proved tells us that in order to make an advance on the brny
larman conjecture we should try to prove the non - existence of a continuous @xmath708-equivariant map @xmath725\rightarrow s(w_{r+1}^{\oplus ( d+1)})\ ] ] at least for some values of @xmath4 .
now in order to prove the non - existence of a continuous @xmath708-equivariant map , for @xmath726 an odd prime , we will compute the fadell
husseini index of the join @xmath718=\delta_{p-1,p}^{*(d+1)}*[p]$ ] with respect to the cyclic group and compare the result with the index of the sphere @xmath727 .
let @xmath728 be an odd prime . in this section
we compute the fadell
husseini index of chessboards @xmath729 and their joins with respect to the cyclic subgroup @xmath730 of the symmetric group @xmath731 .
recall that @xmath732\otimes\lambda[e],\ ] ] where @xmath733 , @xmath734 , and @xmath295 $ ] denotes the exterior algebra .
first , we collect some simple facts about the fadell
husseini index of chessboards .
[ lem : fh index of chessboards ] let @xmath552 be an integer , and let @xmath274 be an odd prime .
then @xmath735 , @xmath736 , @xmath737 . for the statement ( i )
observe that @xmath738 $ ] and therefore @xmath739 .
since @xmath740 is a contractible space , @xmath741 in order to prove ( ii ) recall that @xmath742 is a @xmath743-dimensional @xmath744-connected free @xmath730 simplicial complex , see theorem [ th : conn . of chessboard ] . the serre spectral sequence associated to the borel construction fiber bundle
@xmath745 has the @xmath296-term given by @xmath746 thus , @xmath747 for @xmath748 .
consequently @xmath749 .
since @xmath742 is a free @xmath730 simplicial complex , we get @xmath750 , implying that @xmath751 for @xmath752 .
since the spectral sequence @xmath753 converges to the cohomology of the borel construction @xmath754 , we have that @xmath755 for @xmath756 .
in particular , @xmath757 for @xmath752 , implying that @xmath758 for ( iii ) observe that there is a sequence of @xmath730-equivariant inclusions @xmath759 given by the inclusions of the corresponding vertex sets @xmath760\times [ p]\xhookrightarrow{\quad } [ 2]\times [ p ] \xhookrightarrow{\quad } \cdots
\xhookrightarrow{\quad } [ k]\times [ p ] \xhookrightarrow{\quad } [ k+1]\times [ p ] \xhookrightarrow{\quad}\cdots .\ ] ] consequently the monotonicity property of the fadell
husseini index , combined with the fact that for @xmath761 all chessboards @xmath762 are @xmath743-dimensional free @xmath730 simplicial complexes , implies that @xmath763 in the next step we compute the index of the chessboard @xmath764 . for
that we need to establish the following fact .
[ lem : map f ] let @xmath274 be an odd prime .
there exists a @xmath730-equivariant map @xmath765 such that the induced map in cohomology @xmath766 is an isomorphism .
let @xmath767 be a standard basis of @xmath768 , let @xmath769 , and let @xmath770 for @xmath771 .
denote now by @xmath772 the simplex @xmath773 , which is invariant with respect to the action of the cyclic group @xmath730 .
moreover , its boundary @xmath774 is equivariantly homeomorphic to the representation sphere @xmath775 .
we define a continuous map @xmath776 to be the @xmath730-equivariant simplicial map given on the vertex set of @xmath764 by @xmath777 , where @xmath778\times [ p]$ ] .
it remains to be verified that @xmath766 is an isomorphism .
since @xmath779 , the chessboard complex @xmath764 is a connected , orientable pseudomanifold of dimension @xmath780 , for this see @xcite .
thus @xmath781 and an orientation class is given by the chain @xmath782 then on the chain level we have that @xmath783 for this calculation keep in mind that @xmath274 is an odd prime . the chain @xmath784 is a generator of the top homology of the sphere @xmath785 .
therefore , the induced map in homology @xmath786 is just a multiplication by @xmath787 . using the naturality of the universal coefficient isomorphism ( * ? ? ?
* cor.7.5 ) we have that the induced map in homology with @xmath293 field coefficients @xmath788 is again multiplication by @xmath789 . since @xmath789 and @xmath274 are relatively prime the multiplication by @xmath789 is an isomorphism . now using yet another universal coefficient isomorphism ( * ? ?
* cor.7.2 ) for the coefficients in a field we get that the induced map in cohomology with @xmath293 coefficients @xmath790 is an isomorphism .
now we have all ingredients needed to compute the index of the chessboard @xmath764 .
[ th : index of chessboard p-1 ] @xmath791 .
let us denote by @xmath283 the borel construction fiber bundle @xmath792 and by @xmath285 the borel construction fiber bundle @xmath793 the @xmath730-equivariant map @xmath765 constructed in lemma [ lem : map f ] induces a morphism of the borel construction fiber bundles @xmath283 and @xmath285 : @xmath794 ^ -{{\operatorname{id}}\times_{{\mathbb{z}}/p } f}\ar[d ] & & { \mathrm{e}}({\mathbb{z}}/p)\times_{{\mathbb{z}}/p } s(w_p)\ar[d]\\ { \mathrm{b}}({\mathbb{z}}/p ) \ar[rr]^-{= } & & { \mathrm{b}}({\mathbb{z}}/p ) . } \ ] ] this morphism induces a morphism of the corresponding serre spectral sequences @xmath795 with the property that on the zero row of the second term the induced map @xmath796 is the identity .
here we use simplified notation @xmath797 . in the @xmath296-term , since the homomorphism @xmath798 induces an isomorphism on the @xmath744-cohomology , and @xmath730 acts trivially on both cohomologies @xmath799 , the morphism of spectral sequences @xmath800 is an isomorphism .
the @xmath296-term of the serre spectral sequence associated to the fiber bundle @xmath285 is given by @xmath801 because @xmath730 acts trivially on the cohomology @xmath802 .
thus the only possible non - trivial differential is @xmath803 let @xmath804 denote a generator .
then the @xmath744-row of the @xmath296-term , as an @xmath805-module , is generated by @xmath806 . since the differentials are @xmath805-module maps it follows that the differential @xmath807 is completely determined by its image @xmath808 . in order to find the image of the differential notice that @xmath730 acts freely on the sphere @xmath775 and consequently @xmath809 .
since the spectral sequence @xmath810 converges to the cohomology @xmath811 we have that @xmath812 for @xmath813 .
thus , @xmath814 for some @xmath361 .
moreover , @xmath815 the @xmath296-term of the serre spectral sequence associated to the fiber bundle @xmath283 is given by @xmath816 in particular , @xmath817 and @xmath818 , because @xmath730 acts trivially on the cohomology @xmath819 .
let @xmath820 . as we have seen in the map @xmath821
is an isomorphism .
thus @xmath822 is a generator of @xmath823 , and moreover @xmath822 is a generator of the @xmath744-row of the @xmath296-term as an @xmath805-module . as in the case of the spectral sequence @xmath824 the fact that @xmath730-acts freely on the chessboard @xmath764 implies that @xmath825 for @xmath813 . since @xmath826 is a morphism of spectral sequences it has to commute with the differentials . in particular , for @xmath827 we have @xmath828 now the fact that @xmath822 is a generator of the @xmath744-row of the @xmath296-term as an @xmath805-module yields @xmath829 if in addition @xmath830 , then for every @xmath831 @xmath832 which contradicts the fact that @xmath825 for @xmath813 . in summary
we have that @xmath833 moreover , we have that @xmath834 must be an isomorphism for every @xmath831 . hence
, for @xmath831 we have that @xmath835 since , @xmath836 and @xmath837 is the identity map we conclude that @xmath838 and consequently , @xmath839 finally we claim that no non - zero differential can arrive to the @xmath149-row on @xmath840-term for @xmath827 , implying that @xmath841 and concluding the proof of the theorem .
indeed , if this is not true , then there exists a minimal @xmath842 such that @xmath827 and @xmath843 for some @xmath844 and @xmath845 . since differentials are @xmath805-module maps we have that @xmath846 for every @xmath847 .
consequently , @xmath848 for every @xmath847 contradicting the existence of the isomorphisms .
thus , no non - zero differential can arrive to the @xmath149-row before the @xmath849-term .
the proof of the previous theorem , combined together with the fact that a join of pseudomanifolds is a pseudomanifold , yields the following corollary ( * ? ? ?
* cor.2.6 ) .
[ cor : index of join of chessboard p-1 ] let @xmath850 be an integer
. then @xmath851 , @xmath852)={\operatorname{index}}_{{\mathbb{z}}/p}(s(w_p^{\oplus m})*[p])=h^{\geq m(p-1)+1}({\mathrm{b}}({\mathbb{z}}/p);{\mathbb{f}}_p)$ ] , @xmath853^{*p-1})=h^{\geq m(p-1)+p}({\mathrm{b}}({\mathbb{z}}/p);{\mathbb{f}}_p)$ ] . in the next step
we compute the index of the chessboard @xmath762 for @xmath854 .
[ th : index of chessboard k < p-2 ] @xmath855 , for @xmath856 .
let @xmath854 be an integer .
the chessboard @xmath762 is a @xmath857-dimensional free @xmath730 simplicial complex .
thus @xmath858 and consequently @xmath859 for all @xmath860 .
therefore , @xmath861 .
for @xmath862 the theorem follows from lemma [ lem : fh index of chessboards](i ) .
furthermore , for @xmath863 the statement is the content of theorem [ th : index of chessboard p-1 ] .
now let us assume that @xmath864 is even .
then @xmath865 is also even .
now consider the @xmath730-equivariant inclusion map @xmath866 from the monotonicity and join properties of the fadell
husseini index we have that @xmath867 since @xmath865 is even and , as we have seen , @xmath868 we have that @xmath869 . on the other hand , assume that there is an element @xmath870 such that @xmath871 .
then we have reached a contradiction @xmath872 because @xmath873 .
thus we have proved that for even @xmath47 @xmath874 next let us assume that @xmath875 is odd . as we observed at the start of the proof @xmath876
the @xmath730-equivariant inclusion map @xmath877 together with the computation of the index for even integers implies that @xmath878 in order to conclude the proof of the theorem it remains to prove that @xmath879 .
this would yield the equality @xmath880 for all odd @xmath47 .
indeed , assume the opposite , that is , @xmath881 .
the @xmath730-equivariant inclusion @xmath882 combined with the monotonicity and join properties of the fadell
husseini index imply that @xmath883 since @xmath884 , and we have assumed that @xmath881 , the previous relation implies that @xmath885 a contradiction .
hence @xmath879 , and the proof of the theorem is complete .
let us review the results on the fadell
husseini index of chessboards we have obtained so far : @xmath886_{= } & { \operatorname{index}}_{{\mathbb{z}}/p}\delta_{2,p}\ar[l]_-{\supseteq}\ar@{<->}[d]_{= } & \ldots\ar[l]_-{\supseteq } & { \operatorname{index}}_{{\mathbb{z}}/p}\delta_{p-1,p}\ar[l]_-{\supseteq}\ar@{<->}[d]_{= } & \ldots\ar[l]_-{\supseteq } & { \operatorname{index}}_{{\mathbb{z}}/p}\delta_{2p-1,p}\ar@{<->}[d]_{=}\ar[l]_-{\supseteq}&\ar[l]_-{=}\\ h^{\geq 1}({\mathrm{b}}({\mathbb{z}}/p);{\mathbb{f}}_p ) & h^{\geq 2}({\mathrm{b}}({\mathbb{z}}/p);{\mathbb{f}}_p)\ar[l]_-{\supseteq } & \ldots\ar[l]_-{\supseteq}&h^{\geq p-1}({\mathrm{b}}({\mathbb{z}}/p);{\mathbb{f}}_p)\ar[l]_-{\supseteq } & \ldots\ar[l]_-{\supseteq}&h^{\geq p}({\mathrm{b}}({\mathbb{z}}/p);{\mathbb{f}}_p)\ar[l]_-{\supseteq}&\ar[l]_-{= } } \ ] ] the remaining question indicated by this diagram is : _ for which chessboard @xmath762 with @xmath887 does the first jump in the index @xmath888 to @xmath889 happen ? _ [ th : index of chessboard k > p-1 ] @xmath890 , for @xmath891 .
it suffices to show that @xmath892 .
for this we are going to prove that @xmath893 .
consider the following composition of maps @xmath894 where the first map is an inclusion , the second map is the @xmath101-fold join of the map @xmath895 , @xmath896 introduced in lemma [ lem : map f ] , the third map is again an inclusion , while the last map is a deformation retraction .
all the maps in this composition are @xmath730-equivariant .
the monotonicity property of the fadell
husseini index implies that @xmath897 according to .
thus @xmath893 , and we have concluded the proof of the theorem .
now we have the answer to our question .
the jump happens in the last possible moment , that is for the index of @xmath742 .
the proof of this is due to carsten schultz .
we conclude the section with a very useful corollary ( * ? ? ?
* cor.2.6 ) , which also hides a proof for the upcoming optimal colored tverberg theorem [ th : optimal colored tverberg ] .
[ cor : index of join ] let @xmath898 .
then @xmath899 let @xmath900 , @xmath901 , and @xmath902 .
then there is a @xmath730-equivariant inclusion @xmath903 . again the monotonicity and join properties of the fadell
husseini index imply that @xmath904 furthermore @xmath905 .
the complexes @xmath130 and @xmath906 are free @xmath730-spaces and therefore , as previously observed , it follows that @xmath907 since , by corollary [ cor : index of join of chessboard p-1 ] , @xmath908 and @xmath909 is an even integer , the relation between the indexes @xmath910 implies that @xmath911 as claimed .
we have also proved that @xmath912 .
finally we will , motivated by theorem [ th : non exitence > bl ] , utilize the computation of the fadell
husseini index for the chessboards to prove the following result ( * ? ? ?
* prop.4.2 ) .
[ th : last 01 ] let @xmath367 be an integer , and let @xmath274 be an odd prime .
there is no @xmath731-equivariant map @xmath913\rightarrow s(w_{p}^{\oplus ( d+1)}).\ ] ] it suffices to prove that there is no @xmath730-equivariant map @xmath914\rightarrow s(w_{p}^{\oplus ( d+1)})$ ] , where @xmath730 is a subgroup of the symmetric group @xmath731 generated by the cycle @xmath915 .
the proof uses the monotonicity property of the fadell
husseini index . according to and
the join property for the spheres , the index of the sphere @xmath916 is @xmath917 using corollary [ cor : index of join of chessboard p-1 ] we get that @xmath918)=h^{\geq ( d+1)(p-1)+1}({\mathrm{e}}({\mathbb{z}}/p);{\mathbb{f}}_p),\ ] ] and consequently @xmath919)$ ] .
thus , @xmath920),\ ] ] implying that a @xmath730-equivariant map @xmath914\rightarrow s(w_{p}^{\oplus ( d+1)})$ ] can not exist .
a direct corollary of theorems [ th : non exitence > bl ] and [ th : last 01 ] is that the brny
larman conjecture holds for all integers @xmath4 such that @xmath719 is a prime ( * ? ? ?
* cor.2.3 ) .
[ cor : bl for prime + 1 ] let @xmath403 and @xmath367 be integers such that @xmath726 is a prime
. then @xmath404 and @xmath713 . using the pigeonhole principle and the index computation for the chessboards we can in addition
prove that in the case when @xmath274 is an odd prime the brny
larman function @xmath921 is finite .
let @xmath274 be an odd prime .
then @xmath922 .
let @xmath923 , and let @xmath396 be a coloring of the vertex set of the simplex @xmath924 by @xmath29 colors with each color class of size at least @xmath274 .
then by the pigeonhole principle at least one of the colors , let say @xmath925 , has to be of the size at least @xmath926 . according to corollary [ cor : cstm for colored tverberg ] : if we can prove that there is no @xmath731- or @xmath730-equivariant map @xmath927 then for every continuous map @xmath928 there are @xmath274 pairwise disjoint rainbow faces @xmath371 of @xmath397 whose @xmath7-images overlap , that is @xmath929 .
thus we will now prove that there is no @xmath266-equivariant map @xmath930 .
again , using and the join property for the spheres , we have that @xmath931 since @xmath932 and @xmath933 , there is a @xmath730-equivariant inclusion @xmath934 thus the monotonicity property of the fadell
husseini index and corollary [ cor : index of join of chessboard p-1](iii ) imply that @xmath935 therefore , @xmath936 and consequently there is no @xmath730-equivariant map @xmath930 .
this concludes the proof of the theorem . while focusing on the brny larman conjecture and the corresponding function @xmath398
, we almost overlooked that the index computations for the chessboards establish a considerable strengthening of the topological tverberg theorem that is known as the optimal colored tverberg theorem ( * ? ? ?
* thm.2.1 ) .
[ th : optimal colored tverberg ] let @xmath367 be an integer , let @xmath274 be a prime , @xmath937 , and let @xmath370 be a continuous map .
if the vertices of the simplex @xmath1 are colored by @xmath535 colors , where each color class has cardinality at most @xmath938 , then there are @xmath274 pairwise disjoint rainbow faces @xmath939 of @xmath1 whose @xmath7-images overlap , @xmath940
2 @xcite @xcite @xcite let @xmath268 be a finite group and let @xmath26 be a ( left ) @xmath268-space .
the _ borel construction _ of @xmath26 is the space given by @xmath941 , where @xmath942 is a free , contractible right @xmath268-space and @xmath268 acts on the product by @xmath943 .
the projection @xmath944 induces the following fiber bundle @xmath945 this fiber bundle is called the _ borel construction fiber bundle_. the serre spectral sequence associated to the borel construction fiber bundle has the @xmath296-term given by @xmath946 where the coefficients are local and determined by the action of @xmath947 on the cohomology of @xmath26 . moreover , each row of the spectral sequence has the structure of an @xmath948-module , while all differentials are @xmath948-module morphisms .
the borel construction and the associated fibration are natural with respect to equivariant maps , that is , any @xmath268-equivariant map @xmath949 between @xmath268-spaces induces the following morphism of fiber bundles @xmath950 ^ -{{\operatorname{id}}\times_g f}\ar[d ] & { \mathrm{e}}g\times_g y\ar[d]\\ { \mathrm{b}}g \ar[r]^-{= } & { \mathrm{b}}g . } \ ] ] this morphism of fiber bundle induces a morphism of associated serre spectral sequences @xmath951 such that @xmath952 is the identity .
@xcite @xcite for a finite group @xmath268 the _ classifying space _ is the quotient space @xmath953 .
the projection @xmath954 is the universal principal @xmath268-bundle , that is , the set of all homotopy classes of maps @xmath955 $ ] is in bijection with the set of all isomorphism classes of principal @xmath268 bundles over @xmath26 .
@xcite @xcite let @xmath268 be a finite group , and let @xmath365 be a ( left ) @xmath268-module .
consider a projective resolution @xmath960 of the trivial ( left ) @xmath268 module @xmath961 , that is , an exact sequence @xmath962 & p_{n+1}\ar[r]^-{d_{n } } & p_{n}\ar[r]^-{d_{n-1 } } & \cdots\ar[r]^-{d_0 } & p_0\ar[r]^-{\pi } & { \mathbb{z}}\ar[r ] & 0 , } \ ] ] where each @xmath963 is a projective ( left ) @xmath268-module .
the _ group cohomology _ of @xmath268 with coefficients in the module @xmath365 is the cohomology of the following cochain complex @xmath964_-{d_{n}^ * } \hom_{g}(p_{n},m ) & \ar[l]_-{d_{n-1}^*}\cdots & \ar[l]_-{d_{0}^ * } \hom_{g}(p_{0},m ) & \ar[l ] 0 . } \ ] ] @xcite @xcite let @xmath268 be a finite group , and let @xmath365 be a ( left ) @xmath268-module .
the _ group cohomology _ of @xmath268 with coefficients in the module @xmath365 is the cohomology of @xmath965 with local coefficients in the @xmath947-module @xmath966 , that is @xmath967 @xcite @xcite let @xmath968 be an integer . a topological space @xmath26 is _ @xmath5-connected _ if any continuous map @xmath969 , where @xmath970 , can be continuously extended to a continuous map @xmath971 , that is @xmath972 .
here @xmath973 denotes a @xmath974-dimensional closed ball whose boundary is the sphere @xmath975 .
a topological space @xmath26 is @xmath976-connected if it is non - empty ; it is @xmath149-connected if and only if it is path - connected .
if the space @xmath26 is @xmath5-connected and @xmath152 is @xmath535-connected , then the join @xmath977 is @xmath978-connected . @xcite @xcite the @xmath982 _ chessboard complex _
@xmath434 is the simplicial complex whose vertex set is @xmath983\times [ n]$ ] , and where the set of vertices @xmath486 spans a @xmath47-simplex if and only if @xmath984 .
for example , @xmath985 , @xmath986 .
the chessboard complex @xmath434 is an @xmath987-space by @xmath988 where @xmath989 , and @xmath486 is a simplex in @xmath434 .
the connectivity of the chessboard complex @xmath434 is @xmath990 for @xmath991 , the chessboard complex @xmath992 is a connected , orientable pseudomanifold of dimension @xmath993 .
therefore , @xmath994 and an orientation homology class is given by the chain @xmath995 the symmetric group @xmath996 acts on @xmath992 by the restriction action .
then @xmath997 .
@xcite let @xmath130 be a simplicial complex , let @xmath212 , @xmath213 be integers , and let @xmath998:=\{1,\ldots , n\}$ ] .
@xmath5-fold @xmath47-wise deleted join _ of the simplicial complex @xmath130 is the simplicial complex @xmath999)\ , { { \rm card\,}}i\geq k \rightarrow \bigcap_{i\in i } \sigma_i=\emptyset \},\end{gathered}\ ] ] where @xmath194 are faces of @xmath130 , including the empty face .
the symmetric group @xmath1000 acts on @xmath1001 by @xmath1002 for @xmath199 and @xmath1003 .
@xcite let @xmath130 be a simplicial complex , let @xmath212 , @xmath213 be integers , and let @xmath998:=\{1,\ldots , n\}$ ] .
@xmath5-fold @xmath47-wise deleted product _ of the simplicial complex @xmath130 is the cell complex @xmath1004)\ , { { \rm card\,}}i\geq k \rightarrow \bigcap_{i\in i } \sigma_i=\emptyset \ } , \end{gathered}\ ] ] where @xmath194 are non - empty faces of @xmath130 .
the symmetric group @xmath1000 acts on @xmath1005 by @xmath1006 for @xmath199 and @xmath1007 .
@xcite let @xmath268 be a non - trivial finite group .
for an @xmath5-connected @xmath268-space @xmath26 and an at most @xmath5-dimensional free @xmath268-cw complex @xmath152 there is no continuous @xmath268-equivariant map @xmath269 .
@xcite @xcite @xcite for a finite group @xmath268 any contractible free @xmath268-cw complex equipped with the right @xmath268 cellular action is a model for an @xmath942 space .
milnor s model is given by @xmath1008 where @xmath268 stands for a @xmath149-dimensional free @xmath268-simplicial complex whose vertices are indexed by the elements of the group @xmath268 and the action on @xmath268 is given by the right translation , and @xmath1009 is an @xmath5-fold join of the @xmath149-dimensional simplicial complex with induced diagonal ( right ) action .
@xcite @xcite @xcite let @xmath268 be a finite group and let @xmath26 be a ( left ) @xmath268-space .
the singular or ech cohomology of the borel construction @xmath1010 of the space @xmath26 is called the _ equivariant cohomology _ of @xmath26 and is denoted by @xmath1011 . here
@xmath411 denotes a group , or a ring of coefficients .
@xcite let @xmath268 be a finite group , let @xmath1012 be a relative @xmath268-cw complex with a free action on @xmath1013 , and let @xmath1014 denote the integral cellular chain complex .
the cellular free @xmath268-action on every skeleton of @xmath1013 induces a free @xmath268-action on the chain complex @xmath1014 .
thus @xmath1014 is a chain complex of free @xmath1015-modules . the @xmath268-equivariant cochain complex @xmath1019 and define the _ equivariant cohomology _ @xmath1020 of @xmath1012 with coefficients in @xmath365 to be the cohomology of the cochain complex @xmath1021 .
@xcite let @xmath268 be a finite group , let @xmath327 be an integer and let @xmath152 be a path - connected @xmath5-simple @xmath268-space . for every relative @xmath268-cw complex @xmath1012 with a free action of @xmath268 on the complement @xmath1013 , there exists the obstruction exact sequence @xmath1022_{g}\rightarrow \mathrm{im}\big([{\operatorname{sk}}_{n}x , y]_{g}\rightarrow [ { \operatorname{sk}}_{n-1}x , y]_{g}\big ) \\ \overset{[{\mathfrak o}_{g}^{n+1}]}{\longrightarrow } \mathcal{h}_{g}^{n+1}(x , a;\pi_{n}y ) , \end{gathered}\ ] ] the sequence is natural in @xmath26 and @xmath152 .
this should be understood as follows : a @xmath268-equivariant map @xmath1023 that can be equivariantly extended to the @xmath5-skeleton @xmath1024 , that is @xmath1025 , defines a unique element @xmath1026 $ ] living in @xmath1027 , called the _ obstruction element _ associated to the map @xmath7 ; the exactness of the sequence means that the obstruction element @xmath1026 $ ] is zero if and only if there is a @xmath268-equivariant map @xmath1024 whose restriction is in the @xmath268 homotopy class of the restriction of @xmath7 , that is @xmath1028 , which extends to the @xmath979-skeleton @xmath1029 . the obstruction element @xmath1030 $ ] associated with the homotopy class @xmath1031\in [ { \operatorname{sk}}_{n}x , y]_{g}$ ] can be introduced on the cochain level as well .
let @xmath1032 be an attaching map and @xmath1033 the corresponding generator .
the _ obstruction cochain _ @xmath1034 of the map @xmath7 is defined on @xmath1035 by @xmath1036\in \lbrack s^{n},y].\ ] ] the cohomology class of the obstruction cocycle coincides with the obstruction element defined via the exact sequence .
@xcite let @xmath268 be a finite group and @xmath411 be a commutative ring with unit .
for a @xmath268-space @xmath26 and a ring @xmath411 , the _ fadell husseini index _ of @xmath26 is defined to be the kernel ideal of the map in equivariant cohomology induced by the @xmath268-equivariant map @xmath1037 : @xmath1038 some basic properties of the index are : let @xmath268 be a group and let @xmath26 be a non - empty set . a _ ( left )
@xmath268-action _ on @xmath26 is a function @xmath1050 , @xmath1051 with the property that : @xmath1052 for every @xmath1053 and @xmath1054 . a set @xmath26 with a @xmath268-action
is called a @xmath268__-set__. let @xmath268 and @xmath26 in addition be topological spaces .
then a @xmath268-action is _ continuous _ if the function @xmath1050 is continuous with respect to the product topology on @xmath1055 . a topological space equipped with a continuous @xmath268-action
is called a @xmath268__-space__. @xcite @xcite let @xmath268 be a finite group .
a cw - complex @xmath26 is a _
@xmath268-cw complex _ if the group @xmath268 acts on @xmath26 by cellular maps and for every @xmath1057 the subspace @xmath1058 is a cw - subcomplex of @xmath26 . let @xmath26 be a @xmath268-cw complex , and let @xmath1059 be a subcomplex of @xmath26 that is invariant with respect to the action of the group @xmath268 and consequently a @xmath268-cw complex in its own right .
the pair of @xmath268-cw complex @xmath1012 is a _ relative @xmath268-cw complex_. @xcite @xcite the following result is a consequence of the localization theorem for elementary abelian groups : let @xmath274 be a prime , @xmath326 for @xmath327 , and let @xmath26 be a finite @xmath268-cw complex .
the fixed points set @xmath328 of the space @xmath26 is non - empty if and only the map in cohomology @xmath329 , induced by the projection @xmath330 , is a monomorphism .
let @xmath26 be a set and let @xmath1063 be a family of subsets of @xmath26 .
the _ nerve _ of the family @xmath1064 is the simplical complex @xmath1065 with the vertex set @xmath1066 , and a finite subset @xmath1067 is a face of the complex if and only if @xmath1068 .
if for every face @xmath92 of the nerve @xmath457 the intersection @xmath468 is @xmath1072-connected , then the complex @xmath130 is @xmath47-connected if and only if the nerve @xmath457 is @xmath47-connected .
@xcite @xcite let @xmath268 be a finite group , let @xmath327 be an integer and let @xmath152 be an @xmath459-connected and @xmath5-simple @xmath268-space .
furthermore , let @xmath1012 be a relative @xmath268-cw complex with the free @xmath268 action on @xmath1013 , and let @xmath1073 be a @xmath268-equivariant map . then
every two @xmath268-equivariant extensions @xmath1076 of @xmath7 are @xmath268-homotopic , relative to @xmath1059 , on @xmath1077 , that is @xmath1078_{g}\rightarrow [ { \operatorname{sk}}_{n-1}x , y]_{g}\big)=\{\mathrm{pt}\},\ ] ] if @xmath1079 is a @xmath268-equivariant homotopy between @xmath268-equivariant maps @xmath1073 and @xmath1080 , and if @xmath1081 and @xmath1082 are @xmath268-equivariant extensions of @xmath7 and @xmath1083 , then there exists a @xmath268-equivariant homotopy @xmath1084 between @xmath1085 and @xmath1086 that extends @xmath1087 . in the case when @xmath1088_{g}\rightarrow [ { \operatorname{sk}}_{n-1}x , y]_{g}\big)=\{\mathrm{pt}\}$ ] the obstruction sequence becomes @xmath1089_{g}\rightarrow \
{ \mathrm{pt}\}\overset{[{\mathfrak o}_{g}^{n+1}]}{\rightarrow } \mathcal{h}_{g}^{n+1}(x,\pi _ { n}y).\ ] ] the obstruction element @xmath1090\in \mathcal{h}_{g}^{n+1}(x,\pi _ { n}y)$ ] is called the _ primary obstruction _ and does not depend on the choice of a @xmath268-equivariant map on the @xmath5-th skeleton of @xmath26 .
@xcite @xcite let @xmath268 be a finite group and let @xmath1091 be its subgroup . consider a @xmath1015-chain complex @xmath1092 and a @xmath1015-module @xmath365 .
denote by @xmath1093 the restriction from @xmath268 to @xmath1087 .
for every integer @xmath5 there exists a homomorphism @xmath1094 that we call the _ restriction _ from @xmath268 to @xmath1087 , and a homomorphism @xmath1095 that is called the _ transfer _ from @xmath1087 to @xmath268 , with the property @xmath1096 \cdot { \operatorname{id}}.\ ] ] sergey avvakumov , isaac mabillard , arkadiy skopenkov , and uli wagner , _ eliminating higher - multiplicity intersections , iii .
codimension 2 _ , preprint , 16 pages ,
november 2015 , http://arxiv.org/abs/1511.03501[arxiv:1511.03501 ] , 2015 .
to3em , _ barycenters of polytope skeleta and counterexamples to the topological tverberg conjecture , via constraints _ , preprint , 6 pages , http://arxiv.org/abs/1510.07984[arxiv:1510.07984 ] , october 2015 .
branko grnbaum , _ convex polytopes _ , graduate texts in math .
221 , springer - verlag , new york , 2003 , second edition prepared by v. kaibel , v. klee and g. m. ziegler ( original edition : interscience , london 1967 ) .
sinia vreica and rade t. ivaljevi , _ new cases of the colored tverberg theorem _ , jerusalem combinatorics 93
( h. barcelo and g. kalai , eds . ) , contemp . math . , vol .
178 , amer .
soc . , 1994 , pp . | brny s `` topological tverberg conjecture '' from 1976 states that any continuous map of an @xmath0-simplex @xmath1 to @xmath2 , for @xmath3 , maps points from @xmath4 disjoint faces in @xmath1 to the same point in @xmath2 .
the proof of this result for the case when @xmath4 is a prime , as well as some colored version of the same result , using the results of borsuk ulam and dold on the non - existence of equivariant maps between spaces with a free group action , were main topics of matouek s 2003 book `` using the borsuk
ulam theorem . '' in this paper we show how advanced equivariant topology methods allow one to go beyond the prime case of the topological tverberg conjecture .
first , we explain in detail how equivariant cohomology tools ( employing the borel construction , comparison of serre spectral sequences , fadell
husseini index , etc . ) can be used to prove the topological tverberg conjecture whenever @xmath4 is a prime power .
our presentation includes a number of improved proofs as well as new results , such as a complete determination of the fadell
husseini index of chessboard complexes in the prime case .
secondly , we introduce the `` constraint method , '' which applied to suitable `` unavoidable complexes '' yields a great variety of variations and corollaries to the topological tverberg theorem , such as the `` colored '' and the `` dimension - restricted '' ( van kampen flores type ) versions . both parts have provided crucial components to the recent spectacular counter - examples in high dimensions for the case when @xmath4 is not a prime power . |
the standard model ( sm ) provides an excellent description of elementary particles and interactions , which has further been confirmed by the discovery of the sm - like higgs particle at the large hadron collider ( lhc ) @xcite
. nevertheless , there are still some experimental results showing the tantalizing hints to new physics beyond the sm .
especially , one of the biggest discrepancies between the experimental values and the sm predictions comes from the muon anomalous magnetic moment ( amm ) or @xmath3 for short , known as the @xmath3 anomaly . in 2001
, the e821 experiment at brookhaven national lab ( bnl ) @xcite showed that the measured value of @xmath4 exceeds the sm prediction by about @xmath5 , where the different standard deviations are originated from the various theoretical methods for estimating the hadronic contributions , by applying either the @xmath6 @xcite or the @xmath7-based @xcite data .
the value of the discrepancy between the measurement and sm prediction is given by @xcite @xmath8 to explain this anomaly , many models have been proposed ( see @xcite and references therein ) . for example , by including the mixings of the muon with some new tev - scale heavy leptons @xcite , the new contribution to @xmath3 is roughly the same order as the @xmath9- and @xmath10-boson ones due to the enhancement from the heavy lepton masses .
however , most of these models predict visible lepton flavor violating ( lfv ) processes . in particular , the decay of @xmath11 usually gives one of the most stringent constraints , due to the accuracy of its experimental searches .
for example , the meg collaboration has recently renewed the upper bound on the branching ratio of this process to @xmath12 ( @xmath13 c.l . )
@xcite . as a result , it is natural and necessary to use this updated limit to constrain the structure of new physics behind the @xmath3 anomaly .
the main purpose of the present paper is to propose some general methods of building models to not only explain the @xmath3 anomaly but also satisfy the stringent @xmath1 constraint .
we first explore this problem from the effective field theory ( eft ) perspective , finding that it is very challenging to accomplish both processes simultaneously with the same uv cutoff and the similar order of the wilson coefficients .
we then focus on some simplified models with new physics perturbatively coupled to the sm fields so that the leading order contributions to @xmath3 and @xmath14 appear at the one - loop order . by collecting all one - loop contributions with different spins ( @xmath15 ) , cp properties and electric charges of new particles ,
we find two general methods to reconcile the tension between the @xmath3 anomaly and the @xmath14 constraint : the gim mechanism and the non - universal couplings .
for the latter , we present a simple uv - complete leptoquark model as an illustration . in our study
, we assume that all of the sm contributions to the muon @xmath0 are already well understood and appropriately calculated .
the discrepancy mainly arises from new physics . for clarity ,
we will use @xmath16 to denote the new physics correction to @xmath3 .
this paper is organized as follows .
we analyze the correlation between the muon @xmath0 anomaly and @xmath1 from the eft perspective in sec .
ii . in sec .
iii , we classify the general new physics models by the leading - order one - loop diagrams for the muon @xmath0 in terms of their different structures and internal particle contents . in sec .
iv , we proposes two general methods : the gim mechanism and the non - universal couplings , which can resolve the tension between these two processes .
finally , a short summary is given in sec .
it is quite natural to consider the constraint on new physics from the lfv mode @xmath11 to explain the @xmath3 anomaly from the eft perspective since the structure of the leading effective operators for both processes is essentially the same except for that the outgoing muon is replaced by an electron , as illustrated in fig .
[ fig : fig1 ] .
[ subfig : fig1 ] the effective operators for @xmath3 and @xmath14 are given by : @xmath17 respectively , where @xmath18 and @xmath19 . at this stage , we differentiate the cutoff scales as @xmath20 and @xmath21 , and the wilson coefficients of the left and right - handed couplings as @xmath22 and @xmath23 for @xmath3 and @xmath14 , respectively .
note that we have extracted the electromagnetic coupling constant @xmath24 from the wilson coefficients for the normalization .
we can easily derive the contribution to @xmath16 from eq .
, given by @xmath25 with this formula , we find that the desired value of @xmath26 can be obtained by taking @xmath27 gev with the natural value of @xmath28 .
it is also straightforward to evaluate the branching ratio for the decay process @xmath29 , given by @xmath30 where @xmath31 denotes the fermi constant and @xmath32 is used for the normalization . from eq . and
@xmath33 , one obtains @xmath34 if we take the wilson coefficients @xmath35 to be @xmath36 , the cutoff scale @xmath21 should be at least of @xmath37 gev . from the above discussion
, we see that there is at least a five - order scale gap between the cutoff needed to solve the @xmath3 anomaly and that required by the @xmath14 constraint . in other words , if we assume that the same new physics contributes to both processes , _
i.e. _ , @xmath38 as indicated by the @xmath3 anomaly , the predicted branching ratio for the @xmath1 decay is naturally 5 orders larger than the current experimental bound , unless we choose the wilson coefficients @xmath35 of @xmath39 or even smaller , which are obviously quite unnatural from the general eft philosophy .
clearly , we come to the conclusion that it is challenging for a natural model to obtain the required contribution to @xmath40 while satisfying the current @xmath14 constraint from the eft aspect .
before providing the general methods to reconcile the tension between @xmath3 and the constraint from @xmath11 , let us first see how general new physics models perturbatively coupled to the sm part can solve the @xmath3 anomaly .
we shall work in a simplified framework in which only the relevant particles and the renormalizable parts of the lagrangian related to @xmath3 are given . in this setup , it is enough to only consider the leading one - loop contributions to @xmath41 . for simplicity , we confine the spin of the loop particles to be not larger than 1 , but we do not restrict their charges as to keep the discussion general . in this way , the leading one - loop feynman diagrams can be classified into 4 categories , as explicitly shown in fig .
[ gen ] , according to if the boson running in the loop is a ( axial-)vector or ( pseudo-)scalar and the photon is emitted from a fermion or boson .
we calculate these diagrams and present the final expressions for @xmath41 , which can be compared with those given in the literature @xcite with different gauges . in particular , we also show the analytic formulae in two useful limits : ( i ) @xmath42 and ( ii ) @xmath43 , where @xmath44(@xmath45 ) denotes the mass of the additional loop boson ( fermion ) .
however , in a complete model , the total contribution to @xmath3 is usually the summation of the two or more diagrams in fig .
[ gen ] , so the classification here is only for the convenience of the discussion . besides the new vector boson @xmath46
, we usually need to introduce some extra fermions @xmath47 with the internal index @xmath48 .
the renormalizable lepton - vector - fermion vertex can be written as follows : @xmath49 where the subscript @xmath50 is the charged flavor index and @xmath51 the ( axial-)vector coupling matrix . with this setup ,
[ subfig : nvff ] and [ subfig : cvff ] can be calculated as follows .
the contributions to @xmath3 due to the vector ( v ) and axial - vector ( a ) couplings are given by @xcite @xmath52 and @xmath53 respectively , where @xmath54 is the mass of the vector boson , @xmath55 , @xmath56 , @xmath57 and @xmath58 is the electric charge of @xmath59 .
if we further restrict to the limit of @xmath60 , the two integrals , @xmath61 and @xmath62 , can be simplified to @xmath63 as a simple check of our general formulae , eqs .
( [ fv ] ) and ( [ fa ] ) , we can identify @xmath64 as the @xmath10 boson and @xmath59 as the @xmath65 lepton , resulting in @xmath66 which agrees with the usual sm calculations in the literature @xcite . here
, we have used @xmath67 , @xmath68 , @xmath69 , @xmath70 , and @xmath71 .
furthermore , since the integral for the purely axial - vector coupling is always negative , the axial - vector contribution from negative charged fermions in the loop is always desconstructive with the sm one .
clearly , there is no hope to add a neutral vector boson with the purely axial - vector coupling to explain the @xmath3 anomaly . in order to differentiate from the previous case
, we denote the two general contributions to @xmath3 as those from the charged vector ( cv ) and charged axial - vector ( ca ) couplings , given by @xcite @xmath72 and @xmath73 respectively , where @xmath74 is the vector boson electric charge . in the limit of @xmath60 , _
e.g. _ , @xmath75 and @xmath76 have the following simple form @xmath77 we can check eqs .
( [ eq : cv ] ) and ( [ eq : ca ] ) easily by identifying the charged boson to be the w boson with @xmath78 and @xmath79 , giving @xmath80 which is in agreement with the usual sm result at one - loop level @xcite .
note that there are no cross terms of the vector and axial - vector couplings , which are proportional to @xmath81 or @xmath82 in the @xmath3 contribution , since such terms lead to the muon electric dipole operator , rather than the magnetic dipole one which we are interested in .
in addition , our derivation is based on the lagrangian of eq .
( [ veclag ] ) , in which the vector boson coupling is decomposed into the vector and axial - vector couplings . however , it is more useful to work on the basis in which the fermion field has definite chirality .
the problem is how to use our general formulae eqs .
( [ fv ] ) , ( [ fa ] ) , ( [ eq : cv ] ) and ( [ eq : ca ] ) in this chiral basis .
we find that if the two vertices involving @xmath64 are both left - handed or both right - handed , _ i_.e . , the internal fermion does not flip its chirality , the expression should be @xmath83 .
but , if one vertex is left - handed while the other is right - handed , the internal fermion flips its chirality with the result of @xmath84 . for new scalar ( s ) and pseudoscalar ( p ) bosons ,
the couplings to the sm leptons @xmath85 require to have some additional fermions @xmath86 through the following yukawa couplings : @xmath87 where @xmath88 and @xmath89 are yukawa coupling matrices for scalars and pseudoscalars , respectively , and @xmath50 denotes the lepton flavor .
the relevant feynman diagrams are depicted in figs .
[ subfig : nsff ] and [ subfig : csff ] .
in order for this diagram to contribute to @xmath3 , the internal fermions should have charge @xmath90 , no matter if the ( pseudo)scalar is charged or not .
moreover , it is convenient to further divide the @xmath41 contribution into two parts : scalar ( s ) and pseudoscalar ( p ) ones , since they lead to the different expressions . *
scalar * : the general contribution to @xmath3 from a scalar ( s ) coupling in eq .
( [ lags ] ) is @xmath91 where @xmath92 is the scalar mass , @xmath55 , @xmath93 , @xmath94 , and @xmath58 is the charge of @xmath59 as before . in the limit of @xmath95
, we can further simplify the integral @xmath96 to @xmath97 note that the integral @xmath98 in eq .
( [ eq : ns ] ) does not have the definite sign in the parameter region we are interested in . as a result , it is more useful to plot @xmath98 against @xmath99 and @xmath100 on the 3d diagram in fig .
[ subfig : ns ] .
we find that in the limits @xmath101 and @xmath102 , the value of @xmath103 is always positive .
thus , the sign of @xmath104 only depends on @xmath105 .
for example , if we take the fermion charge to be @xmath106 and the scalar to be neutral , the new contribution to the muon @xmath0 is constructive with the sm one .
we can obtain @xmath107 by choosing @xmath108 and @xmath109 tev , promising to be probed by the next run of the lhc experiments .
* pseudoscalar * : the pseudoscalar ( p ) coupling in eq .
( [ lagp ] ) can give the following general @xmath3 contribution , @xmath110 where @xmath111 denotes the pseudoscalar mass , @xmath55 , @xmath112 , and @xmath113 .
if we further restrict to the limit of @xmath114 , the integral @xmath115 can be reduced to @xmath116 as for the scalar case , we also plot @xmath117 in eq .
( [ eq : np ] ) against @xmath99 and @xmath100 on the 3d diagram in fig .
[ subfig : np ] .
it is useful to note that most of the parameter space is negative except for the @xmath118 region . as an example , if @xmath119 and the pseudoscalar is neutral , the sign of its contribution to the muon @xmath0 is positive only when @xmath120 , _
i.e. _ , @xmath121 , as seen from fig .
[ subfig : np ] .
however , if there exists such a light charged fermion , it should have already been observed at the colliders , such as the lep .
therefore , a model with a neutral pseudoscalar and a light charged fermion favored by the current @xmath3 data is already ruled out . the calculation of the feynman diagram in fig .
[ subfig : csff ] leads to the contribution to the muon amm from loops with a charged scalar ( cs ) or pseudoscalar ( cp ) .
* scalar * : when the photon is emitted from a charged scalar , the general contribution to the muon @xmath0 in fig .
[ subfig : csff ] is @xmath122 where @xmath123 is the scalar charge , @xmath55 , @xmath93 and @xmath124 as before . when @xmath95 , @xmath125 can be simplified to @xmath126 the integral in eq .
( [ eq : cs ] ) is always negative .
consequently , if @xmath123 is minus , @xmath127 will depress @xmath128 .
the simple application of this result is that when the loop in fig .
[ subfig : csff ] is enclosed by a scalar with @xmath129 and a neutral fermion , the contribution can never reduce the tension between the sm prediction and data for the muon @xmath0 .
* pseudoscalar * : when the scalar in fig .
[ subfig : csff ] is replaced by the charged pseudoscalar , the expression for the @xmath3 contribution becomes , @xmath130 where @xmath131 is the charge of the pseudoscalar , @xmath55 , @xmath112 and @xmath132 .
the integral @xmath133 in the limit @xmath114 becomes @xmath134 for a rough estimation , with a pseudoscalar with @xmath135 and a neutral fermion in the loop , the feynman diagram in fig .
[ subfig : csff ] can solve the muon @xmath0 anomaly with @xmath136 and @xmath137 tev .
similar to the ( axial-)vector case in the previous subsection , the cross terms of the scalar and pseudoscalar couplings , which are proportional to @xmath138 or @xmath139 , do not contribute to the muon @xmath0 .
rather , they give rise to the muon electric dipole moment .
moreover , when the scalar / pseudoscalar couplings in eqs .
( [ lags ] ) and ( [ lagp ] ) are decomposed into the basis in which the chiralities of the fermion fields are well defined , we can easily obtain the @xmath3 formulae from eqs .
( [ eq : ns ] ) , ( [ eq : np ] ) , ( [ eq : cs ] ) and ( [ eq : cp ] ) . concretely , when the two vertices on the fermion line are both left- and right - handed , the result would be @xmath140 .
note that the internal fermion has to flip its chirality in this case so that the final expression should be proportional to the internal fermion mass .
however , for the case that the two couplings have different chiralities , the result is @xmath141 .
it is interesting to note that our general formalism can be applied to the general supersymmetric extension of the sm .
one particular example is the minimal supersymmetric standard model ( mssm ) , in which neutralinos , charginos and various sleptons provide the additional contributions to @xmath3 and @xmath14 .
especially , one - loop diagrams related to charginos and neutralinos correspond to our case in sec . [ secfermion ] and [ secscalar ] , so the formulae in these two subsections can be used directly .
the detailed discussions of @xmath142 and its correlation to @xmath14 are summarized in ref .
now , we come back to the question how to reconcile the tension between the muon @xmath0 anomaly and the @xmath1 constraint in a generic new physics model . here , we provide two generic methods mentioned in sec .
[ intro ] .
as is well known in the sm , there is no flavor - changing neutral current ( fcnc ) at tree level due to the famous gim mechanism @xcite . even at loop level
, the gim mechanism also suppresses the fcnc greatly . of our present interest
is the sm contributions to @xmath3 and @xmath14 . in the sm with the non - zero neutrino masses ,
the one for @xmath14 can be obtained by the loops running a @xmath9-boson with different flavors of neutrinos .
but because of the gim mechanism , the unitarity of the pmns matrix results in the leading terms to be canceled .
the terms left are at least proportional to the square of the neutrino masses .
since the neutrino masses are negligibly small , it is no hope to observe the @xmath1 rate in the current experiments .
however , @xmath3 does not suffer such a huge suppression , leaving us measurable signals .
we hope that the gim mechanism may also happen in the new physics sector beyond the sm . to be specific ,
let us consider a model in which a vector @xmath64 and some fermions @xmath144 with masses @xmath145 are introduced with the following chiral couplings : @xmath146 where @xmath85 the different charged leptons in the sm with the flavor index @xmath50 .
we have extracted the overall coupling constant @xmath147 to make the mixing matrix @xmath148 to satisfy the following orthogonal normalization conditions , @xmath149 where @xmath50 and @xmath150 are the flavor indices of charged leptons ( @xmath24 , @xmath65 , @xmath7 ) . here ,
the mixing matrix @xmath148 is not necessarily unitary but can be extended to a rank - three @xmath151 one , where @xmath152(@xmath153 ) is the number of the internal fermions .
note that the lagrangian in eq .
( [ laggim ] ) is very similar to the sm w - boson couplings with the w - boson replaced by @xmath46 and the neutrinos by @xmath144 .
however , to make the discussion more general , we take the charge of the vector boson @xmath46 to be @xmath154 so that the fermion charge should be @xmath155 . we also assume a mass hierarchy @xmath156 ( @xmath157 ) for simplicity . with the lagrangian in eq .
( [ laggim ] ) , the leading - order one - loop feynman diagrams relevant to @xmath1 are essentially the same as the first two diagrams in fig . 2 except for the outgoing @xmath65 replaced by @xmath24 , resulting in the following effective operator , @xmath158 where @xmath159 and the integrals @xmath160 are the separated expressions for the two feynman diagrams in fig . 2 .
the explicit expressions of @xmath160 are not given since they are not very crucial for our discussion of the general gim mechanism .
what we should focus is the track of the coupling factors , especially the mixing matrix elements @xmath148 , which can be easily read out from the feynman diagrams .
note that the two external charged leptons in the effective vertex @xmath161 have different chiralities , while the interaction in eq .
( [ laggim ] ) only involves left - handed fermions .
therefore , an extra lepton mass insertion is needed to obtain @xmath161 by flipping the chirality of either lepton , producing two terms proportional to @xmath162 and @xmath163 . however , since @xmath164 , only the latter term is kept , which is the origin of the factor @xmath163 and the right - handed projection operator @xmath165 in eq .
( [ effgim ] ) . moreover , the mass dimensions of the integrals @xmath160 are @xmath166 , which can be made dimensionless by extracting the inverse of the largest mass scale @xmath167 squared . in this way , the total integral @xmath168 can be written in the following form : @xmath169 where @xmath170 are the coefficients for the power expansion of @xmath171 in terms of @xmath172 , and are usually expected as the @xmath173(1 ) function of @xmath174 . since only the squares of various particle masses appear in the integral @xmath168 , the expansion only has even powers of @xmath172 . by putting eq .
( [ expansion ] ) back into eq .
( [ effgim ] ) , the effective operator for @xmath14 has the form : @xmath175 note that the leading term vanishes due to the orthogonal condition in eq .
( [ unitary ] ) .
we are , then , left with the second - order term @xmath176 since , in general , we have @xmath177 for the different internal fermion masses . by considering the hierarchy @xmath178
, we see a large suppression at least in order of @xmath179 .
in contrast , for the flavor conserving muon @xmath0 correction , there is no such suppression due to the normalization condition for the outgoing and incoming leptons with the same flavor .
typically , we expect that the contribution to @xmath3 should be of order : @xmath180 with some @xmath173(1 ) coefficient @xmath181 .
thus , the lfv process of @xmath11 is naturally suppressed greatly compared with @xmath41 .
note that the gim mechanism is also applicable to the case when the vector interaction is purely right - handed , rather than left - handed as discussed above . however
, if the couplings of the new vector @xmath46 to the leptons and the internal fermions have both left - handed and right - handed parts , the gim mechanism generically breaks down .
one reason is that the couplings of different chiralities involve different mixing matrices .
for example , let us denote the left - handed vertex mixing matrix by @xmath182 , while the right - handed by @xmath183 . obviously , the products of two matrices , such as @xmath184 and @xmath185 , do not necessarily obey the normalized orthogonal conditions in eq .
( [ unitary ] ) .
furthermore , when the two vertices in the diagrams shown in figs .
2(a ) and 2(b ) have different chiralities , there should be an additional internal fermion mass @xmath145 insertion for flipping its chirality , rather than the much smaller lepton masses . as a result , the leading - order new - physics corrections to @xmath3 and @xmath1 should be of the same magnitude . consequently , we can encode this argument in terms of the following effective operators @xmath186 where @xmath187 are left- and right - handed couplings for the vector boson @xmath46 . in general , the two effective operators have the same order , except for the special case in which all the masses of the internal fermions are the same and @xmath188 .
we have discussed the gim mechanism in a general theory with massive vector bosons .
one related question is whether the gim mechanism still works in a ( pseudo)scalar theory .
unfortunately , the yukawa couplings @xmath189 in eqs .
( [ lags ] ) and ( [ lagp ] ) are general complex matrices , and do not have the orthogonal properties as in eq .
( [ unitary ] ) .
thus , the gim mechanism can not be applied to an ordinary model with some new ( pseudo)scalars .
another possible method to give a large enough @xmath16 without exceeding the @xmath143 bound is based on the observation that these two processes involve different coupling constants in a new physics model . from the diagrams in fig .
[ gen ] , we find that the muon @xmath0 contributions are proportional to either @xmath190 for the ( axial-)vectorial couplings in eq .
( [ veclag ] ) or @xmath191 for the yukawa couplings in eqs .
( [ lags ] ) and ( [ lagp ] ) , while for the @xmath1 process , the amplitudes should be proportional to @xmath192 or @xmath193 correspondingly .
since the couplings @xmath194 and @xmath195 are generically free parameters , we have the freedom to choose the coupling matrices @xmath51 and @xmath196 such that the combinations @xmath192 and @xmath193 are always smaller than @xmath190 and @xmath191 by at least 5 orders , as indicated by the effective operator analysis in sec .
[ seceft ] . in this way
, we have the possibility to suppress @xmath1 to the allowed magnitude while still giving a large enough @xmath3 correction to solve the anomaly .
although such a choice of the coupling constants has a little fine - tuning , it is still acceptable if we consider the same - level hierarchy between the top - quark and the electron yukawa couplings . in the following
, we would like to use a leptoquark model introduced in ref .
@xcite as a simple uv - complete theory to exemplify the application of this method . in ref .
@xcite , only one extra scalar leptoquark @xmath64 is introduced for phenomenological reasons . in order to eliminate the dangerous tree - level proton decay via @xmath64
, one can find that only two choices of the leptoquarks are allowed , whose quantum numbers under the sm groups of @xmath197 are ( * 3 * , * 2 * , 7/3 ) and ( * 3 * , * 2 * , 1/3 ) , respectively . in the present paper , we will concentrate on the former case and compute its muon @xmath0 contributions from the leptoquark , which was not discussed in the original paper .
for other aspects of the model , readers are recommended to refer to ref .
the relevant lagrangian for the leptoquark couplings is given by @xmath198 with @xmath199 where @xmath200 denote the yukawa couplings related to the right - handed @xmath201-type quarks and the right - handed @xmath24-type leptons , respectively , and @xmath99 is the usual antisymmetric tensor for the @xmath202 gauge group . due to the couplings in eq .
( [ laglq ] ) , various one - loop diagrams enclosed by the two leptoquark components and different quark flavors can contribute to @xmath41 and @xmath1 .
however , by assuming that the leptoquark is much heavier than any sm quarks and leptons , _
i.e. _ , @xmath203 , we find that both amplitudes for the muon @xmath0 and @xmath1 are proportional to the quark masses in the loop @xcite .
thus , the dominant contributions to both phenomena should come from the top-@xmath204 loops .
consequently , the @xmath143 decay rate in this model is given by @xmath205 with @xmath206 where , @xmath207 and @xmath208 denotes the mixing matrix that brings the left - handed ( right - handed ) fermions from the flavor to the mass eigenstates . on the other hand , from eqs .
, , and , the dominant contribution to @xmath3 with internal a top quark is @xcite @xmath209 ) \nonumber\\ & & + \frac{1}{2}{\rm re}\left [ { \tilde { \lambda } } _ { e } ^ { 23 } { \tilde { \lambda } } _ { u } ^ { 32 } \right ] \left [ { q } _ { t } \left ( { \cal i}^{s}- { \cal i}^{p } \right)+ { q } _ { x_1 } \left ( { \cal i}^{cs}- { \cal i}^{cp } \right ) \right ] \big\ } \nonumber\\ & \approx & \frac{3\lambda^\prime}{4 \pi^2 } \frac{m_\mu m_t}{m_{x_1}^2 } \left(\frac{2}{3 } \ln \frac{m_{x_1}}{m_t } - \frac{1}{12 } \right),\end{aligned}\ ] ] where @xmath210 , \end{aligned}\ ] ] @xmath211 is the number of colors , and @xmath212 is the charge of @xmath213(@xmath214 ) .
we also have taken the limit @xmath215 in eq .
( [ eq : letoquarkg-2 ] ) to simplify the final result . with the formulae in eqs .
( [ mue ] ) and ( [ eq : letoquarkg-2 ] ) , we can plot the contour of @xmath216 and the boundary of the allowed region @xmath217 in the @xmath218-@xmath219 plane , as shown in fig . [ fig : leptoquarknew ] .
note that the couplings @xmath220 and @xmath221 have totally different dependences on the more fundamental yukawa couplings @xmath222 , so that they actually have no direct relations .
we can tune these yukawa couplings to make @xmath221 positive and larger than @xmath220 by at least four orders of magnitude so as to fit the @xmath40 deviation and suppress @xmath1 to the allowed order simultaneously , which represents the essence of the method of the non - universal couplings .
note that as the process @xmath14 does not give some prominent constraint to this leptoquark interpretation of the muon @xmath0 anomaly , we need to consider other more stringent constraints .
we find that the contact coupling @xmath223 also depends on @xmath224 , so that the @xmath3 contribution can not escape from its constraints , which mainly come from the measurement of @xmath225 . as a result , the constraints for this contact operator listed in refs .
@xcite can be translated to the following bound , @xmath226 which leads to the loose limit on the @xmath41 from the present leptoquark @xmath64 @xmath227 another relevant constraint to our discussion of @xmath3 is from the muon electric dipole moment ( edm ) , @xmath228 . currently , the upper bound of @xmath229 is @xmath230 @xcite , which will be improved to be at the level of @xmath231 by the j - parc new @xmath0/edm experiment collaboration @xcite in the near future .
on the other hand , @xmath228 arises dominantly from the top - leptoquark loop in the present leptoquark model , given by , @xmath232 \right|\,,\end{aligned}\ ] ] where @xmath233 is defined in eq .
( [ f - function ] ) .
since the leptoquark contribution to @xmath14 depends on the same function @xmath233 , we can express the coupling bound from @xmath228 in terms of @xmath234 , @xmath235 \right| } { \lambda } \sqrt { { \cal b}(\mu \rightarrow e\gamma ) } \lesssim 36.2\ , .\end{aligned}\ ] ] in order to show the constraining power , if the leptoquark gave a branching ratio equal to the current experimental bound of @xmath236 , the muon edm would constrain the couplings to be @xmath235 \right| } { \lambda } \lesssim 4.8\times { 10 } ^ { 7 } .\end{aligned}\ ] ] note that @xmath228 only relates to the imaginary part of @xmath237 , while @xmath3 its real part .
thus , the muon edm can not affect our general conclusion on @xmath3 .
similarly , one has @xmath238
\right|,\end{aligned}\ ] ] which would lead to @xmath239 \right| } / { \lambda } \lesssim 2.2\times { 10 } ^ { -2 } $ ] with the most recent upper limit of @xmath240 by the acme collaboration @xcite .
however , due to the different dependence of the yukawa couplings , the limit from @xmath241 can not place a meaningful constraint on the leptoquark solution to the @xmath3 anomaly and @xmath14 .
therefore , we conclude that this leptoquark model is promising to solve the @xmath3 anomaly with the non - universal yukawa couplings to suppress @xmath1 , even if we further consider other low - energy experimental constraints .
motivated by the long - standing @xmath3 anomaly and the recently updated @xmath143 upper bound , we have examined the correlations between @xmath143 and @xmath3 .
the general eft analysis tells us that it is difficult in explaining the muon @xmath0 anomaly while satisfying the @xmath143 bound in a natural theory with @xmath2 wilson coefficients , since the cutoff scale obtained by fitting the required @xmath3 discrepancy predicts an unbearably large @xmath1 rate . after compiling all of the one - loop diagram formulae for the @xmath3 corrections ,
we have proposed two promising methods to eliminate this tension between the new physics contributions to @xmath3 and @xmath1 : the gim mechanism and the non - universality of couplings . for the latter method ,
a leptoquark model has been illustrated as a simple example .
as expected , with the appropriate choice of leptoquark yukawa couplings , it is possible to achieve the goal to understand the @xmath3 anomaly while keeping an experimentally allowed @xmath1 branching ratio . as discussed in our leptoquark model , even if the new physics accounting for the required @xmath41 can escape the @xmath1 bound with our two methods , we still need to check other constraints , such as the contact interactions , the unitarity of the ckm and/or pmns matrices , and so on , especially for those with the same coupling dependence as the explanation of the @xmath3 anomaly . finally , we hope that our investigation on the possible relation between the muon anomalous magnetic moment and the decay @xmath1 may shed light on the structure of new physics .
we would like to thank prof .
svjetlana fajfer for pointing out a mistake in our previous version .
the work was supported in part by national center for theoretical sciences , national science council ( nsc-101 - 2112-m-007 - 006-my3 ) and national tsing hua university ( 103n2724e1 ) .
unsrt j. beringer _
[ particle data group collaboration ] , phys .
d * 86 * , 010001 ( 2012 ) .
g. aad _ et al . _ [ atlas collaboration ] , phys .
b * 716 * , 1 ( 2012 ) [ arxiv:1207.7214 [ hep - ex ] ] .
s. chatrchyan _ et al . _
[ cms collaboration ] , phys .
b * 716 * , 30 ( 2012 ) [ arxiv:1207.7235 [ hep - ex ] ] .
g. w. bennett _ et al . _ [ muon g-2 collaboration ] , phys .
lett . * 92 * , 161802 ( 2004 ) [ hep - ex/0401008 ]
. g. w. bennett _ et al . _ [ muon g-2 collaboration ] , phys .
d * 73 * , 072003 ( 2006 ) [ hep - ex/0602035 ] .
m. davier , a. hoecker , b. malaescu and z. zhang , eur .
j. c * 71 * , 1515 ( 2011 ) [ erratum - ibid .
c * 72 * , 1874 ( 2012 ) ] [ arxiv:1010.4180 [ hep - ph ] ] .
r. alemany , m. davier and a. hocker , eur .
j. c * 2 * , 123 ( 1998 ) [ hep - ph/9703220 ] .
t. kinoshita and m. nio , phys .
d * 73 * , 053007 ( 2006 ) [ hep - ph/0512330 ] .
a. czarnecki , w. j. marciano and a. vainshtein , phys .
d * 67 * , 073006 ( 2003 ) [ erratum - ibid .
119901 ( 2006 ) ] [ hep - ph/0212229 ] . k. hagiwara , r. liao , a. d. martin , d. nomura and t. teubner , j. phys .
g * 38 * , 085003 ( 2011 ) [ arxiv:1105.3149 [ hep - ph ] ] .
m. davier and w. j. marciano , ann .
sci . * 54 * , 115 ( 2004 ) .
m. passera , j. phys .
g * 31 * , r75 ( 2005 ) [ hep - ph/0411168 ] .
f. jegerlehner and a. nyffeler , phys .
rept . * 477 * , 1 ( 2009 ) [ arxiv:0902.3360 [ hep - ph ] ] .
a. czarnecki and w. j. marciano , phys .
d * 64 * , 013014 ( 2001 ) [ hep - ph/0102122 ] .
k. kannike , m. raidal , d. m. straub and a. strumia , jhep * 1202 * , 106 ( 2012 ) [ erratum - ibid .
* 1210 * , 136 ( 2012 ) ] [ arxiv:1111.2551 [ hep - ph ] ] .
j. adam _ et al .
_ [ meg collaboration ] , arxiv:1303.0754 [ hep - ex ] .
t. moroi , phys .
d * 53 * , 6565 ( 1996 ) [ erratum - ibid .
d * 56 * , 4424 ( 1997 ) ] [ hep - ph/9512396 ] ; d. stockinger , j. phys .
g * 34 * , r45 ( 2007 ) [ hep - ph/0609168 ] ; z. chacko and g. d. kribs , phys .
d * 64 * , 075015 ( 2001 ) [ hep - ph/0104317 ] ; j. kersten , j. h. park , d. stckinger and l. velasco - sevilla , jhep * 1408 * , 118 ( 2014 ) [ arxiv:1405.2972 [ hep - ph ] ] .
j. m. arnold , b. fornal and m. b. wise , phys .
d * 88 * , 035009 ( 2013 ) [ arxiv:1304.6119 [ hep - ph ] ] .
i. dorner , s. fajfer , n. konik and i. niandi , jhep * 1311 * , 084 ( 2013 ) [ arxiv:1306.6493 [ hep - ph ] ] .
r. benbrik and c. -k .
chua , phys .
d * 78 * , 075025 ( 2008 ) [ arxiv:0807.4240 [ hep - ph ] ] .
k. m. cheung , phys .
d * 64 * , 033001 ( 2001 ) [ hep - ph/0102238 ] .
m. carpentier and s. davidson , eur .
j. c * 70 * , 1071 ( 2010 ) [ arxiv:1008.0280 [ hep - ph ] ] .
s. davidson and p. verdier , phys .
d * 83 * , 115016 ( 2011 ) [ arxiv:1102.4562 [ hep - ph ] ] .
g. w. bennett _ et al . _
[ muon ( g-2 ) collaboration ] , phys .
d * 80 * , 052008 ( 2009 ) [ arxiv:0811.1207 [ hep - ex ] ] .
h. iinuma [ j - parc new g-2/edm experiment collaboration ] , j. phys .
295 * , 012032 ( 2011 ) .
j. baron _ et al .
_ [ acme collaboration ] , science * 343 * , no .
6168 , 269 ( 2014 ) [ arxiv:1310.7534 [ physics.atom-ph ] ] . | while the muon @xmath0 anomaly can be successfully explained by some new physics models , most of them are severely constrained by the @xmath1 bound .
this tension is more transparent from the effective field theory perspective , in which the two phenomena are encoded in two very similar operators .
however , with the @xmath2 wilson coefficients , the current upper bound on @xmath1 indicates a new - physics cutoff scale being five orders smaller than that needed to eliminate the @xmath3 anomaly . by summarizing all the formulae from the one - loop contributions to the muon @xmath0
with the internal - particle spin not larger than 1 , we point out two general methods to reconcile the conflict between the muon @xmath0 and @xmath1 : the gim mechanism and the non - universal couplings . for the latter method
, we use a simple scalar leptoquark model as an illustration . |
inflation has been widely considered as a remarkably successful theory in explaining many problems in very early universe , such as the flatness , the horizon and the monopole problem @xcite .
it predicts that the quantum fluctuations of the inflaton field were generated to form today s large scale structure@xcite , and these fluctuations fit with current observations of cosmic microwave background very well@xcite .
it is a common viewpoint that eternal inflation@xcite can take place in a variety of inflation models .
especially , with the monomial chaotic inflaton potential , inflation generally becomes eternal in the high energy regime .
eternal inflation is not only important in concept , but also provides a possible realization for the string theory landscape@xcite . there have been an increasing number of papers investigating whether we can measure eternal inflation and calculate probabilities in the multiverse , see references@xcite .
usually , eternal inflation happens when the energy scale of the universe is extremely high .
thus we would like to take into consideration on more fundamental theories in logic , namely , the string theory . in this paper
, we focus on a universal property of string theory , i.e. , noncommutativity of spacetime . by considering the effects of spacetime noncommutativity
, we study the condition for eternal inflation and its implications .
as required by stringy spacetime uncertainty relation@xcite , the physical time @xmath0 and the physical length @xmath1 should satisfy @xmath2 where @xmath3 is a length scale given by the string theory .
there have been a lot of attempts to apply this uncertainty relation into inflationary cosmology , dubbed noncommutative inflation , see ref . @xcite .
more detailed works have been investigated in a number of literature@xcite . here
we briefly review the conception of noncommutative inflation proposed by brandenberger and ho@xcite , and then examine whether eternal inflation takes place with such noncommutativity . note that here we have slightly generalized the discussion in @xcite from power law inflation into a generally quasi - exponential inflationary scenario . in order to introduce the noncommutativity into the 4-dimensional friedmann - robertson - walker universe
, we would like to define another time coordinate @xmath4 , @xmath5 where @xmath6 is the scale factor and we have assumed a spatially flat universe(@xmath7 ) .
for the quasi - exponential expansion where we have imposed the usual slow roll approximation self - consistently , we have @xmath8 where @xmath9 is the hubble parameter with the dot represents the derivative with respect to the cosmic time @xmath10 and @xmath11 is an arbitrary parameter for the scale factor at a fixed time
. then the spacetime uncertainty relation takes the form @xmath12 where @xmath13 is the comoving radial coordinate .
this can be realized by the commutation relation of spacetime @xmath14_*=il_s^2~,\ ] ] where the @xmath15-product is defined as @xmath16 based on these considerations , for the fluctuation of the inflaton field @xmath17 , we can derive that the canonical normalized perturbation variable @xmath18 satisfies the equation of motion @xmath19 where the prime denotes the derivative with respect to @xmath20 defined as @xmath21 .
the parameter @xmath22 is an effective scale factor appeared in the dispersion relation between a mode @xmath23 and its energy defined with respect to @xmath4 , @xmath24~,\ ] ] and @xmath25 is defined as @xmath26 as the scale factor is expanding nearly exponentially , when @xmath27 is not too small compared with @xmath4 we can take the approximate form of @xmath28 as @xmath29 ^ 2~,\\ \label{b- } & \beta_k^{-}\simeq{1\over 2}a^{-2}(\tau - l_s^2k)\simeq { 1\over 2}\left[h\cdot(\tau - l_s^2k)\right]^{-2}~.\end{aligned}\ ] ] moreover , because of the relations @xmath30 and the spacetime uncertainty ( [ unc ] ) , there is an initial time @xmath31 for the perturbations to be generated , @xmath32 and the fluctuations are not allowed to exist before @xmath31
. due to this initial time for the fluctuations , the quantum fluctuations of inflaton in noncommutative environment can be generated inside or outside the hubble radius .
these fluctuations are called uv modes and ir modes respectively . in the uv mode limit ,
the effects of spacetime noncommutativity becomes very weak and hence the evolution of the fluctuations is similar with the commutative case
. therefore the spectrum of uv modes is roughly the same as in the commutative case . while in the ir region
, noncommutativity dominates the behavior of the perturbations , and some familiar pictures will be totally modified . in this paper , we shall use the analysis of noncommutativity mentioned above ( especially the ir region ) to study eternal inflation .
this paper is organized as follows . in section 2
, we calculate the fluctuations generated outside the horizon and provide the constraint for eternal inflation to exist . in section 3 , we make a conclusion and discuss some other possibilities .
in this section , we investigate the quantum fluctuations generated in noncommutative inflationary era and discuss the condition of eternal inflation . since in the uv case
the noncommutative property only contribute a few corrections on the usual perturbation theory@xcite , the condition for the eternal inflation to happen is basically the same as the commutative one .
therefore , we focus our consideration on the ir case that the quantum fluctuations are generated outside the horizon . at the time the fluctuations start to be generated , the effective scale factor is @xmath33 making use of eq .
( [ b+ ] ) and ( [ b- ] ) , and the fact that the hubble scale is larger than the noncommutativity scale in this case , we get the initial time @xmath31 and the initial scale factor @xmath34 as @xmath35 since the ir modes are generated outside the horizon , it is required that @xmath36 . by using the second relation in eq .
( [ tausol ] ) , we can see that @xmath37 in the ir case .
this verifies the physical picture that @xmath38 should be larger than the noncommutativity scale .
now we calculate the quantum fluctuation in the momentum space @xmath17 .
@xmath17 is linked to the canonical perturbation @xmath39 by @xmath18 , and when the perturbation begins to be generated the initial @xmath39 was canonically normalized as @xmath40 .
consequently , when @xmath17 is born its amplitude can be given by @xmath41 after that , the fluctuations outside the horizon are nearly frozen . by transforming to the coordinate space
, we obtain the relation @xmath42 during one hubble time , and correspondingly the ir quantum fluctuation @xmath43 per e - fold in the noncommutative spacetime is given by @xmath44 note that this result is strongly different from the commutative one @xmath45 . due to this ,
the physics of eternal inflation is greatly modified in the noncommutative case and the condition for eternal inflation to happen needs to be reconsidered . as usual
, the classical motion of the inflaton during one hubble time takes the form @xmath46 where @xmath47 denotes @xmath48 . and
as usual , the condition that inflation becomes eternal is roughly @xmath49 , or @xmath50 compared with the eternal inflation condition in the commutative case @xmath51 , the noncommutative eternal inflation is more unlikely to happen .
it is because the quantum fluctuation is generally smaller due to the suppression by the hubble parameter . in the following ,
we shall consider two explicit examples with the chaotic potentials .
firstly , let us consider the model @xmath52 . from the condition ( [ ircon ] ) we can see that the inflation is in the ir region when @xmath53 while in the uv region when @xmath54 .
consequently , the condition that inflation has become eternal in the uv region is @xmath55 interestingly , we note that from ( [ main ] ) the eternal inflation will nicely continue in the ir region if @xmath56 to substitute the well - known value @xmath57@xcite into ( [ m2phi2 ] ) , we obtain the result that eternal inflation is allowed to happen only if @xmath58 .
so we conclude that if inflation has become eternal before entering the ir region , it can continue to be eternal in the ir region .
but the amplitude of fluctuation can not grow to larger values because @xmath43 is now suppressed by a factor of @xmath59 .
this is greatly different from the commutative eternal inflation and may constrain the initial condition space for the eternal inflation . on the other hand , if inflation has not become eternal before entering the ir region , then the inequality ( [ m2phi2 ] ) does not hold . in this case inflation will never become eternal because the quantum fluctuation can not be large enough . in the noncommutative spacetime .
here we have assumed @xmath60 to be large enough so that eternal inflation can take place in the uv region . in this case
, inflation is eternal in part of the noncommutative uv region and the whole ir region.,width=355 ] as a second example , consider @xmath61 with @xmath62 .
then similarly , the condition for entering the ir region @xmath63 requires @xmath64 from the inequality ( [ main ] ) , we obtain the condition for inflation to be eternal in the ir region as follow , @xmath65 in order that ( [ p1 ] ) and ( [ p2 ] ) has overlap and inflation can be eternal in the ir region , we need @xmath66 if ( [ pcond ] ) is satisfied , the noncommutative eternal inflation is allowed to exist in the ir region , and there is an upper bound for the energy density of eternal inflation which arises from ( [ p2 ] ) . since the inflaton field can not climb higher than the bound ( [ p2 ] ) with large probability , this provides a possible initial condition for eternal inflation .
now let us examine whether there is eternal inflation at all in the @xmath62 model above .
for the eternal inflation to take place in the uv region , we again obtain the relation @xmath67 which is the same as the ir region eternal condition ( [ pcond ] ) .
so we get the similar conclusion with the @xmath68 model that if eternal inflation takes place in the uv region , it is allowed to extend into the ir region . on the other hand ,
if inflation can not be eternal in the uv region , unfortunately there will not be any eternal inflation in the whole noncommutative inflationary history . besides , there is an interesting difference between @xmath62 and @xmath69 cases that , in the @xmath62 case noncommutative eternal inflation requires an upper bound on the inflaton while this bound does not exist in the @xmath69 case .
for example , in the @xmath70 model , if we expect eternal inflation to take place , then we need the noncommutative scale @xmath71 . to apply the data @xmath72
, we obtain that this scale is around @xmath73 .
if @xmath74 is just above this scale , the upper bound ( [ p2 ] ) on @xmath75 is around @xmath76 and the corresponding energy density is about @xmath77 .
consequently , we conclude that the limitation from noncommutativity can be much stronger than that from the planck density .
such a string scale can be realized easily if we do certain compactification on a large manifold , which should be common and have a large prior probability in the string landscape . in the noncommutative spacetime .
here we have assumed @xmath60 to be large enough so that eternal inflation can take place in the uv region . in this case
, inflation is eternal in part of the noncommutative uv region and part of the ir region .
this figure is different from fig .
[ fig : primordial2 ] that eternal inflation can not happen in the green part of ir region where the energy scale is extremely high .
, width=345 ] on the other hand , such a scale @xmath78 or @xmath79 is considerably higher than the scale for the final 60 e - folds inflation .
so if we can observe the noncommutativity in future cmb experiments@xcite , then eternal inflation is not expected to have happened .
spacetime noncommutativity , predicted by string theory , has become a fundamental principle and been studied in a number of literature ( see e.g. @xcite ) .
this principle brings new physics when applied into inflation theory . in this paper
, we have seen that for the chaotic inflationary potential , the scenario of eternal inflation in noncommutative spacetime is remarkably different from the usual one . if inflation do not become eternal in the uv region , then it can never have happened .
we have also discussed that if eternal inflation happens both in the uv region and the ir region , an initial condition for eternal inflation can be provided or constrained . from the derivation in this paper , we can see that @xmath43 becomes smaller and smaller along the potential when eternal inflation enters ir region .
therefore , for the @xmath69 model the initial condition space is greatly reduced ; while for the @xmath62 model there is an upper bound for @xmath75 explicitly .
eternal inflation can not climb into higher energy regions than this bound .
this provides an initial condition for eternal inflation .
as is discussed in @xcite , initial conditions may be essential for predictions in the multiverse .
the initial condition discussed in this paper provides a possible solution for the initial condition problem and can be used in calculating the eternal inflationary probabilities . in the derivation made in section 2 ,
we have used the standard method according to a number of calculations for noncommutative and non - eternal inflation in the literature .
however , generally there are several possibilities which may change the results we obtain in this paper .
the simplest possibility is that there is no spacetime noncommutativity at all or the scale of noncommutativity is of the same order as the planck scale .
therefore , the noncommutativity will not alter the picture of the usual eternal inflation . as another possibility
, the noncommutative field theory may be not precise enough to describe the generation of the quantum fluctuations .
this saturation is somewhat similar to that inflationary fluctuations described by the common quantum field theory suffers from the transplanckian problems@xcite . to see this problem in the noncommutative field theory , we note that we have used the relation @xmath80 in the calculations of noncommutative inflation
however , this relation does not hold in theories with certain uv - ir relations while uv - ir relations arise commonly together with noncommutativity .
so there is possibility that the perturbations are generated even if @xmath81 , but it is still an open issue for us to fully understand the physics in this region . finally , the background geometry may be affected considerably by noncommutativity in the ir region .
however , up to now this case has not been carefully studied even in the non - eternal inflationary regime .
we thank bin chen , miao li , yun - song piao and xinmin zhang for helpful discussions . this work is supported in part by national natural science foundation of china under grant nos .
90303004 , 10533010 , 19925523 and 10405029 , and by the chinese academy of science under grant no .
kjcx3-syw - n2 .
the author y.w .
acknowledgments grants of nsfc .
a. d. miller _
et al . _ ,
j. * 524 * , l1 ( 1999 ) , astro - ph/9906421 ; p. de bernardis
_ et al . _ ,
nature * 404 * , 955 ( 2000 ) , astro - ph/0004404 ; s. hanany _
et al . _ ,
j. * 524 * , l5 ( 2000 ) , astro - ph/0005123 ; n. w. halverson _
et al . _ ,
j. * 568 * , 38 ( 2002 ) , astro - ph/0104489 ; b. s. mason _
et al . _ ,
j. * 591 * , 540 ( 2003 ) , astro - ph/0205384 ; a. benoit _
. astrophys . * 399 * , l25 ( 2003 ) , astro - ph/0210306 ; j. h. goldstein _ et al .
_ , astrophys
. j. * 599 * , 773 ( 2003 ) , astro - ph/0212517 ; a. d. linde , mod .
* a1 * , 81 ( 1986 ) ; a. d. linde , phys . lett . *
b175 * , 395 ( 1986 ) ; a. s. goncharov , a. d. linde , and v. f. mukhanov , int .
* a2 * , 561 ( 1987 ) ; a. d. linde , d. a. linde and a. mezhlumian , phys . lett . *
b345 * , 203 ( 1995 ) m. r. douglas , jhep * 0305 * , 046 ( 2003 ) [ arxiv : hep - th/0303194 ] ; f. denef and m. r. douglas , jhep * 0405 * , 072 ( 2004 ) [ arxiv : hep - th/0404116 ] ; m. r. douglas and s. kachru , arxiv : hep - th/0610102 ; f. denef , m. r. douglas and s. kachru , arxiv : hep - th/0701050 .
j. garriga , d. schwartz - perlov , a. vilenkin and s. winitzki , jcap * 0601 * , 017 ( 2006 ) [ arxiv : hep - th/0509184 ] ; d. schwartz - perlov and a. vilenkin , jcap * 0606 * , 010 ( 2006 ) [ arxiv : hep - th/0601162 ] ; a. vilenkin , arxiv : hep - th/0602264 ; v. vanchurin and a. vilenkin , phys . rev . * d74 * , 043520 ( 2006 ) [ arxiv : hep - th/0605015 ] ; a. vilenkin , arxiv : hep - th/0609193 .
r. bousso , phys .
lett . * 97 * , 191302 ( 2006 ) [ arxiv : hep - th/0605263 ] ; r. bousso , b. freivogel and i. s. yang , phys . rev . *
d74 * , 103516 ( 2006 ) [ arxiv : hep - th/0606114 ] .
g. w. gibbons and n. turok , arxiv : hep - th/0609095 .
a. linde , jcap * 0701 * , 022 ( 2007 ) [ arxiv : hep - th/0611043 ] ; a. linde , arxiv:0705.1160 [ hep - th ] .
d. podolsky and k. enqvist , arxiv:0704.0144 [ hep - th ] .
huang and m. li , jhep * 0306*(2003)014 , hep - th/0304203 ; q .-
huang and m. li , jcap * 0311*(2003)001 , astro - ph/0308458 ; q. g. huang and m. li , nucl .
* b713 * , 219 ( 2005 ) [ arxiv : astro - ph/0311378 ] .
s. tsujikawa , r. maartens and r. brandenberger , phys . lett . *
b574 * , 141 ( 2003 ) [ arxiv : astro - ph/0308169 ] ; s. koh and r. h. brandenberger , arxiv : hep - th/0702217 ; r. h. brandenberger , arxiv : hep - th/0703173 .
cai , phys .
* b593 * , 1 ( 2004 ) , hep - th/0403134 ; q. g. huang and m. li , nucl . phys . *
b755 * , 286 ( 2006 ) [ arxiv : astro - ph/0603782 ] ; x. zhang and f. q. wu , phys . lett . *
b638 * , 396 ( 2006 ) [ arxiv : astro - ph/0604195 ] ; qing - guo huang , phys . rev . *
d74 * , 063513 ( 2006 ) , astro - ph/0605442 ; x. zhang , jcap * 0612 * , 002 ( 2006 ) [ arxiv : hep - th/0608207 ] ; q. g. huang , jcap * 0611 * , 004 ( 2006 ) [ arxiv : astro - ph/0610389 ] .
g. calcagni , phys . rev . *
d70 * , 103525 ( 2004 ) [ arxiv : hep - th/0406006 ] ; g. calcagni , phys . lett . *
b606 * , 177 ( 2005 ) [ hep - ph/0406057 ] ; g. calcagni and s. tsujikawa , phys . rev . * d70 * , 103514 ( 2004 ) [ astro - ph/0407543 ] . | we investigate the condition for eternal inflation to take place in the noncommutative spacetime .
we find that the possibility for eternal inflation s happening is greatly suppressed in this case .
if eternal inflation can not happen in the low energy region where the noncommutativity is very weak ( the uv region ) , it will never happen along the whole inflationary history .
based on these conclusions , we argue that an initial condition for eternal inflation is available from the property of spacetime noncommutativity . |
although there is no agreed upon engine powering gamma - ray bursts(grbs ) , models based upon black hole accretion disks ( bhad models ) are attractive . the progenitors of this class of grb engine include the mergers of compact binaries ( double neutron star , black hole and a neutron star , white dwarf and a black hole ) , collapsars , and the inspiral of a neutron star or black hole into a helium star ( see fryer , woosley , & hartmann 1999a for a review ) . in all of these cases ,
these progenitors produce accretion disks around black holes which , as they accrete , use neutrinos or magnetic fields to convert part of the gravitational potential or rotational energy of the disk material ( or black hole ) into a grb outburst .
the bhad systems formed by these progenitors differ mainly in the disk mass , disk angular momentum , or amount of material above the plane of the rotation disk and these variations may well explain the diversity of the observed sample of gamma - ray bursts . for instance , classical grbs seem to divide into long and short duration subgroups ( e.g. kouveliotou et al .
1993 ; katz & canel 1996 ; tavani 1998 ) , and it is thought that these two distinct subgroups are explained by different bhad progenitors ( e.g. fryer et al .
three progenitors may produce long - duration grbs : mergers of white dwarfs with black holes , collapsars , and he - mergers .
the disks formed in the mergers of white dwarfs with black holes ( or neutron stars ) represent one extreme in disk characteristics . of the three long - duration bhad grb progenitors ,
white dwarf mergers produce disks with the most angular momentum and the least amount of material along the rotation axis .
although the high angular momentum in these disks make neutrino - mediated explosions extremely weak , white dwarf / black hole mergers may produce strong grb explosions via some magnetic field mechanism ( fryer et al . 1999b ) .
collapsars , on the other hand , represent the other extreme : they have less angular momentum and much more material along the axis of rotation ( macfadyen & woosley 1999 ; macfadyen , woosley , & heger 2000 ) . since the connection between grb 980425 and supernova 1998bw ( kulkarni et al .
1998 ) , which is naturally explained by collapsar models ( woosley 1993 ; macfadyen & woosley 1999 ; macfadyen et al .
2000 ) , the collapsar model has become the favorite progenitor for long - duration bursts .
but can all long - duration gamma - ray bursts be collapsars
? there is increasing evidence to suggest that collapsars can not explain all grbs .
first , the x - ray and optical lightcurves of grbs suggest that the environment around the grb engine is sometimes best fit by a wind - swept model , and other times is best fit by a constant density model ( chevalier & li 2000 , livio & waxman 2000 ) . based on the fact that collapsars occur only in massive stars , chevalier & li ( 2000 ) and
livio & waxman ( 2000 ) both argue that those grbs best fit by constant density environments must not be collapsars .
in addition , collapsars may have trouble explaining the iron emission lines detected in three grbs : grb 970508 ( piro et al . 1999 ) , grb 970828 ( yoshida et al . 1999 ) , and grb 991216 ( piro , et al . 2000 ) .
to explain these detections ( the detections in grb 970508 and grb 970828 are marginal at best so not all may be real ) , bttcher ( 2000 ) found that the progenitors of these bursts may need to have metallicities in excess of solar .
however , the winds of collapsar progenitors must be lower than typical winds of massive stars , and hence collapsar progenitors are likely to be low - metallicity and it is unlikely that collapsars can explain such high iron abundances . to compare the he - merger model to that of the collapsar , let us review the formation scenario of he - merger grbs .
the progenitor of this grb engine is a binary system consisting of two massive stars .
the more massive star ( primary ) collapses to form a neutron star or black hole .
the loss of matter in the supernova explosion and a possible kick on the compact remnant unbinds many of the binaries .
those that remain bound tend to be in tight orbits , and when the companion star evolves off the main sequence , it engulfs the compact remnant and the system goes into a common envelope phase . once in a common envelope , the compact remnant spirals down through the hydrogen star . if the binding energy of the envelope is too large , the compact star will spiral into the companion s helium core .
it is this inspiral which spins up the helium core and produces the conditions necessary to form a gamma - ray burst .
the merger of the compact remnant with its helium companion provides a mechanism to produce bhad system very similar to that of a collapsar ( fryer & woosley 1998 ) .
the primary difficulty with the collapsar model is producing a black hole with a rotating stellar envelope .
the helium merger model naturally forms an accreting black hole surrounded by a _ rotating _
stellar core .
the differences in the formation processes of these two grb models produce distinguishing features which can be observed .
for example , after the formation of the compact remnant , the he - merger binary system will have some net velocity with respect to its formation region , and it may travel beyond the wind and supernova ejecta of the primary .
hence , the constant density environment required by some of the lightcurves is easily explained by he - mergers .
in addition , the primary supernova s explosion may enrich the hydrogen envelope of its companion ( israelian et al . 1999 ) . when this envelope is ejected during the common envelope phase , it produces an iron rich torus which may explain the observed iron lines ( bttcher & fryer 2000 ) .
fryer & woosley ( 1998 ) argued that the engine driving he - merger grbs would be very similar to collapsars . but to truly understand the he - merger engine , we must follow the inspiral and obtain quantitative estimates of both the rotation of the helium core and the remnant mass in the core . knowing these values , we can then infer the duration and total energy of the he - merger grb . in this paper
, we present hydrodynamical simulations of the inspiral of a compact remnant into a helium core for a range of helium core masses .
we describe our numerical techniques in [ sec : nm ] and the results of our simulations in [ sec : res ] .
we conclude with a discussion of the inspiral results and of the viability of he - mergers as a grb model .
he - merger grbs are formed when , in a binary system , a giant star envelops its compact remnant ( neutron star or black hole ) companion , causing it to spiral through the hydrogen envelope in what is termed a `` common - envelope '' phase .
the details of common envelope evolution are still rather uncertain ( see sandquist et al .
1998 for a review ) , but in those systems where the orbital energy of the binary is not sufficient to eject the hydrogen envelope , the helium core of the giant and the compact remnant will merge . although it is difficult to separate the numerics from the physics in modeling the common envelope phase , modeling the merger of the helium core ( which is several orders of magnitude smaller than the hydrogen giant ) with a compact remnant is much more tractable .
we model the merger using the three - dimensional smooth particle hydrodynamics ( sph ) code developed by davies , benz , & hills ( 1991 ) . to model the physics of the merging process more accurately ,
the code has been improved ( fryer et al .
1999b ) to include an equation of state by blinnikov , dunina - barkovskaya , & nadyozhin ( 1996 ) .
to follow nuclear reactions , we include a nuclear reaction network for temperatures above @xmath11 ( timmes , hoffman , & woosley 2000 ) .
the structure of the helium core was taken from 1-dimensional helium star models by heger ( 1999 ) which were mapped into 3-dimension and then allowed to settle into a stable configuration .
for all of our helium cores , the radius settled to within 10% of the 1-dimensional model . once the helium core has stabilized , we add the compact remnant ( a point gravity source ) and place the binary in a co - rotating circular orbit with a sufficiently small separation so that roche - lobe overflow will occur . at this point ,
mass transfer from the helium core onto the compact remnant will tighten the orbit very rapidly and cause the compact remnant to merge with the helium core .
the orbital separation at which roche - lobe overflow commences is given by ( eggleton 1983 ) : @xmath12 where @xmath13 is the radius of the helium core , @xmath14 is ratio of the mass of the helium core @xmath15 to the mass of the compact remnant @xmath16 . for our simulations
, we are concerned primarily with the effects of the inspiral on the core of the helium star and these results are insensitive to our initial separation of the binary ( that is , we can decrease the separation by @xmath1720% and our results do not change ) . for most of our simulations
, we use initial separations smaller than @xmath18 so that the objects merge without detailed resolution of the outer layers of the helium star .
in addition , it is likely that the hydrogen common envelope phase would cause the compact remnant to spiral in below the @xmath18 , so using the smaller separations may even be closer to the initial conditions produced by nature . in any event ,
our simulations show that two models with different initial separation have similar results ( fig .
[ fig : avst ] ) . aside from initial separations
, we must also be cautious with two other numerical uncertainties .
first and foremost is the accretion onto the compact remnant .
as the compact remnant merges with the helium star , the accretion rate onto the compact remnant is well approximated by the bondi - hoyle accretion rate ( chevalier 1993 ; brown 1995 ; fryer , benz , & herant 1996 ; bethe & brown 1998 ) .
one might argue that if the compact remnant is a neutron star , photon pressure can slow the accretion .
however , the high densities in the helium core trap the photons and the time required for the photons produced at the hot surface of the neutron star to diffuse even 10 km from the surface would take longer than our entire simulation time ( see fryer et al .
1996 for diffusion times ) .
however , the neutron star surface is hot enough to produce copious neutrinos which can cool the accreting material in less than 0.1s and this cooling ( which however does not provide pressure support as photons would ) allows the infalling material to accrete onto the neutron star ( see fryer et al .
chevalier ( 1989 ) argued that even if the compact remnant had dipole magnetic fields in excess of @xmath19 g , these fields would be buried by the high accretion flows that occur in he - mergers .
the more conservative analysis of fryer et al .
( 1996 ) argues that the compact remnant s magnetic field would have to exceed @xmath20 g to affect the accretion .
such highly magnetized neutron stars are probably not common . hence , for neutron star and black hole remnants alike
, the inspiralling compact remnant will accrete at the bondi - hoyle rate .
however , actually modelling this accretion can be quite difficult .
we try two methods to simulate the hypercritical accretion .
in the first accretion recipe , we simply remove particles that fall within 5 - 10@xmath21 of the compact remnant and add their mass and momentum to the compact remnant .
although this is much larger than the schwarzschild radius of the compact remnant , it is the smallest scale we can achieve given our resolution ( 1 - 3 @xmath22 ) . in the second accretion method ,
when the distance between a particle and the compact remnant , @xmath23 , is less than the scale length @xmath24 of that particle , the mass of that particle @xmath25 is accreted onto the compact remnant with a rate given by @xmath26 , here @xmath27 , the free - fall timescale is @xmath28 , where @xmath29 is the gravitational constant and @xmath16 is the compact remnant mass . in this method ,
if its distance from the black hole is less than @xmath24 , the particle s motion is dominated by gravitational force of the compact remnant and it loses mass on the free - fall timescale .
we reduce the particle s mass gradually as it accretes onto the compact remnant .
the second accretion method yields a slightly longer inspiral time ( by 10 - 20% : see fig .
[ fig : avst ] ) which , in turn , yields accretion masses which differ by roughly the same amount ( see [ sec : res ] ) . to test the differences in the accretion _ rate _ , we ran a test model where we placed a @xmath0 compact remnant in the center of an @xmath30 static helium core and measured the accretion as a function of time ( fig .
[ fig : mvst ] ) .
here we see that the accretion rates for both our accretion mechanisms are nearly identical .
one might argue that the accretion timescale is not the free - fall timescale because angular momentum can slow the accretion .
however , during the inspiral stage of the merger of a helium and a compact remnant , the angular momentum of particles accreted by the compact remnant is small . for the duration of our simulation ,
the accretion timescale is dominated by the free - fall timescale , not by the viscous timescale of the accretion disk .
however , to test the differences caused by an increased accretion timescale due to the additional effects of a viscous timescale , we modified the second accretion method by assuming the accretion timescale was 10 times the free - fall timescale and ran our test model with this increased timescale ( fig .
[ fig : mvst ] ) .
the accretion rate for this factor of 10 range in timescales is nearly identical for the first 150s and does not differ by more than 20% for the entire simulation .
this is not a surprising result . by increasing the accretion timescale , we initially lower the accretion rate onto the black hole , but this does not lower the rate at which material piles up around the black hole ( which is what our hydrodynamic simulation is modeling ) .
when we raise the accretion timescale , more material piles up around the black hole , and the density of this material increases , causing the accretion rate to rise . hence an equilibrium is achieved .
the accretion rate onto the black hole is more sensitive to the rate at which the material piles up around the black hole than upon the accretion timescale of any individual particle .
this is nearly identical to the effect seen in accretion disk simulations where the accretion rate from the innermost point of the disk onto a black hole is governed more by the rate at which material is put into the disk rather than the actual value of the artificial viscosity ( popham , woosley , & fryer 1999 ; macfadyen & woosley 1999 ) . because the accretion rate onto the black hole is relatively insensitive to our accretion recipe and because the first recipe , which simply removes nearby particles around the black hole ( or neutron star ) , runs faster than other recipes , we use the first recipe in all other simulations . in both methods , we also must add the angular momentum of the particles to the compact remnant .
unfortunately , we do not resolve the accretion disk which must form around the compact remnant and so can not model the transport of angular momentum out of this disk as it accretes .
instead , we constrain the amount of angular momentum accreted to the angular momentum of the last stable orbit of the compact remnant .
we assume the rest of the angular momentum is transported out of the system . because our simulation time step is much longer than the orbital period of the last stable orbit , this is a reasonable assumption .
this angular momentum may feedback into the star , and we do not include this .
but the angular momentum that we discard in our simulation is less than 1% of the total angular momentum in the system ( fig .
[ fig : jvst ] ) , so we do not underestimate the angular momentum in the helium core by more than 1% at the end of our simulation . figure [ fig : jvst ] also shows how well the total angular momentum in our simulations is conserved . for the merger of a @xmath0 compact remnant and an @xmath31 helium star ,
the summation of total angular momentum of particles , total angular momentum of the compact remnant ( including the estimated spin angular momentum ) and the angular momentum that we discard during accretion is conserved to one part in ten thousand . during the course of the simulation , nearly all of the orbital angular momentum of the compact remnant is converted into spin angular momentum of the helium core , and yet the total angular momentum is conserved ! the other major numerical uncertainty is the artificial viscosity used in the sph .
we have varied the artificial viscosity by a factor of 5 to determine our dependence upon the numerical viscosity .
we will discuss these results in
[ sec : res ] , but the quick answer is that simulations using this wide range of artificial viscosity obtain nearly the same results and our estimates of grb energies and disk structures do not depend on the artificial viscosity .
we have run a series of models testing the initial helium star masses , initial separations , and the range of uncertainties in the numerical methods ( accretion algorithm , artificial viscosity , resolution , etc . ) .
bttcher & fryer ( 2000 ) found that the helium star masses in he - merger progenitors cover a wide range of masses : 14% have masses less than @xmath32 and 22% have masses above @xmath33 with the bulk ( 64% ) lying within @xmath34 . by varying the mass of the helium star , we can study the range of burst properties from he - mergers .
the entire set of simulations is summarized in table [ tab : par ] .
the number in the model designation denotes the helium core mass in units of solar mass .
we break up the formation of a he - merger grb engine into two stages : the inspiral of the compact remnant ( which we model with our sph code ) and a post - processing analysis of the accretion onto the compact remnant once it reaches the core of the helium star . in our simulations , we follow the compact remnant as it spirals into the helium star and accretes mass from the helium star while spinning up the star .
when the compact remnant reaches the center of the helium core , our simulations provide the mass and spin of the compact remnant as well as the structure and angular momentum distribution of the helium star .
the results from our simulations allow us to estimate the accretion of the rotating helium core around the now central black hole . we know that this material will accrete rapidly ( especially along the spin axis of the core ) , but the material along the equator has too much angular momentum to accrete immediately and will first form a disk around the compact remnant .
it is this disk accretion that can drive a grb along the relatively clear spin axis .
the mass of the compact remnant and the spin of the accreting material determine the viability of he - mergers as a grb engine .
we presume that the compact remnant has already spiralled through the hydrogen envelope of a massive giant star ( via a common envelope phase ) and begin our simulations as the helium core contacts the compact remnant in roche lobe overflow . for two of our simulations ( 8d and 8e ) we placed the compact remnant in a circular orbit with an orbital separation just low enough that the helium star overfilled its roche lobe and accreted onto the compact remnant . even if orbital angular momentum were conserved , this mass transfer ( from a more massive object onto a less massive object ) would tighten the orbits .
but , in fact , a good deal of angular momentum is lost as material is ejected from the system . in these simulations
, we have implicitly assumed that when roche - lobe overflow occurs , this mass transfer dominates the orbital evolution of the binary .
the viscosity caused by the hydrogen envelope may dominate the inspiral even after the helium core has overfilled its roche lobe and is accreting onto the compact remnant ( although we know that when the compact remnant actually enters the helium core , the effects of the helium core must dominate the inspiral ) .
fortunately , most of the accretion occurs after the compact remnant has spiralled deep into the helium star and the merger evolution below a separation of @xmath35 cm is not affected by our initial separation ( fig .
[ fig : avst ] ) .
thus , for our @xmath8 and @xmath5 helium cores , we only model the inner @xmath36 of the core and start our simulations with smaller separations .
this allows us to better resolve the core with a given number of particles .
some snapshots of the inspiral process for models 8b , 4a and 16a are shown , respectively , in figures [ fig : h8 ] , [ fig : h4 ] and [ fig : h16 ] .
some material is ejected , carrying away angular momentum and some mass . in general
, this ejecta does not play an important role on the grb explosion itself because it is likely that the grb explosion is beamed along the axis of rotation and the ejecta lies along the equator . but
this ejecta may effect the environment around the grb and may have some observational features ( e.g. iron lines - bttcher & fryer 2000 ) .
although the binary can survive many orbital periods in the roche - lobe mass - transfer phase , as soon as the compact remnant begins to merge with the helium core , the orbital separation evolves very rapidly .
the compact remnant inspirals from half its initial separation to the center of the helium star in less than @xmath17500s ( fig .
[ fig : avstall ] ) .
it is during this latter evolution that most of the accretion onto the compact remnant occurs and the conditions for a grb engine are produced . changing the artificial viscosity by a factor of 5 changes this inspiral time by less than 10% . also , results of high resolution model 8b are similar to those of low resolution model 8a ( fig . [
fig : avstall ] ) . in the initial stages of the inspiral , the accretion rate onto the compact remnant is relatively low ( less than @xmath37 ) and the compact remnant gains very little mass .
but during the rapid inspiral phase , the accretion rate increases dramatically and nearly all of the accretion occurs in the last few hundred seconds .
we stop our simulations when the compact remnant reaches the `` center '' of the helium star . after the inspiral stage
, the effect of angular momentum which can slow the accretion gradually becomes important .
however , at the end of our simulations , the effect of angular momentum is still small .
for example , at the end of the simulation for model 8b , the viscous timescale of most of the matter ( 93% of particles ) accreting onto the black hole is less than the free - fall timescale ( we estimate the viscous timescale by : @xmath38 where @xmath39 is the disk viscosity and @xmath40 is the orbital angular velocity ) .
however , the effect of angular momentum is increasing rapidly , and if we want to model the accretion beyond the inspiral phase , we would need to modify our accretion recipe in addition to increasing our mass resolution .
the mass of the compact remnant at the end of our simulations are given in table [ tab : erg ] . at this stage ,
nearly all of the orbital angular momentum of the system has been converted to spin angular momentum in the helium star ( fig .
[ fig : jr ] ) . some of the orbital angular momentum is converted into spin of the compact remnant ( table [ tab : erg ] ) .
this merged system is now very similar to that of a collapsar ( a rotating helium star collapsing onto a central black hole ) .
these conditions are believed to drive grb explosions , but to estimate the explosion energy , we must calculate the later accretion .
our simulations show that he - mergers will indeed produce rapidly rotating helium stars accreting onto a central compact remnant in much the same manner as is found in the collapsar grb engine .
however , the angular momentum in he - mergers ( fig . [ fig : jr ] ) is roughly an order of magnitude greater than that of the collapsars studied by macfadyen & woosley ( 1999 ) . with such high angular momentum , following the continued accretion
requires an understanding of disk accretion .
even if we assumed an artificial angular momentum transport algorithm , we would need much more resolution than we have in our current simulations to follow disk formation and the accretion onto the compact remnant .
however , we can analytically estimate the accretion rate onto the compact remnant by estimating the accretion time for each particle .
such a rough calculation can be obtained by assuming that pressure support is negligible and gravitational and centrifugal forces dominate the motion of the particles ( for most of the star , this is true ) .
we take the end of our sph simulations as the initial conditions of our analytic estimate . without pressure support
, each particle accelerates to nearly the free - fall velocity until centrifugal support slows its collapse and it accretes onto a disk .
we can derive the radius at which this occurs by knowing the enclosed mass ( @xmath41 ) and the angular momentum ( @xmath42 ) , and then equating forces : @xmath43 .
we estimate the timescale for the particle to accrete onto the disk is simply the free - fall timescale : @xmath44 where @xmath45 is the initial position of the particle . for matter along the pole
, the angular momentum may not be sufficient to halt the collapse , and these particles accrete directly onto the compact remnant at roughly the free - fall time .
but the bulk of the matter will accrete through a disk , and we must then estimate the disk accretion timescale . for each particle in the disk , we can calculate the viscous timescale : @xmath46 where @xmath47 is the disk viscosity parameter ( which we take to be @xmath48 ) , and @xmath40 is the orbit angular velocity .
the total accretion timescale of each particle is just the sum of its free - fall and viscous components ( @xmath49 ) .
we can then easily estimate the accretion rate by calculating the number ( and total mass ) of particles accreted at a given time ( @xmath50 ) @xmath51 where @xmath52 and @xmath53 are the mass and accretion rate of particle @xmath54 .
the accretion rates for our 3 helium star masses are shown in figures [ fig : md08 ] , [ fig : md04 ] and [ fig : md16 ] .
the accretion rate not only provides us with an estimate of the black hole mass as a function of time , but also the spin of the black hole .
there is so much angular momentum in the helium core that nearly all of the accreting material accretes through a disk , and the angular momentum of the material at its last stable orbit quickly spins up the black hole .
in figures [ fig : md08 ] - [ fig : md16 ] , the accretion rates seem to vary on short timescales , but such wide variations are likely to be an artifact of our low resolution .
higher resolution simulations will allow us to determine how much of the variation in the accretion rate is real ( due to asymmetries caused in the inspiral ) and how much is due to numerical resolution . throughout our calculation ,
we have tried to estimate an upper limit to the accretion rate , assuming no thermal pressure and a high disk viscosity .
even so , except for the @xmath8 merger , the accretion rates we obtain are much lower than those calculated in collapsars .
this is , in part , because our helium cores have much higher angular momenta .
in addition , we modeled the merger of unevolved helium cores .
if the common envelope phase occurred during case c mass - transfer , the merger would not have occurred until after helium burning , and the more compact evolved core would yield higher initial accretion rates .
in this paper , we present the first hydrodynamical simulations of the formation of he - merger grbs up to the formation of a central black hole around a rotating helium core . with the conditions produced during the inspiral ,
it is clear that the compact remnant will collapse to a black hole ( if it is not one already ) .
the densities around the central compact remnant are so high that photons are trapped in the material and radiation pressure does not halt accretion . even if the compact remnant is a neutron star , neutrino cooling will allow high accretion rates ( @xmath55 ) which will quickly cause the neutron star to collapse to a black hole .
the accreting material will also spin up the compact remnant and he - mergers produce a population of rapidly rotating black holes . with possible merger rates above @xmath56 in the galaxy ,
these black holes may make up a few percent of the total black holes formed in the galaxy .
however , they do not produce slowly accreting neutron stars ( e.g. thorne - zytkow stars ) and can not be used to explain strange neutron star systems .
but can these mergers produce gamma - ray bursts ?
the viability of the he - merger grb model depends upon the energy of the explosion produced during the disk accretion phase .
the two most - often discussed power sources for these explosions are : neutrino annihilation or some magnetic field mechanism akin to the engine behind the jets of active galactic nuclei .
the neutrino annihilation mechanism takes advantage of the profuse neutrino emission which carries off much of the binding energy of the accreting material .
popham et al .
( 1999 ) calculated the energy emitted in neutrinos for hyper - accreting black holes .
the subsequent energy generation rate from neutrino annihilation can be fitted by ( fryer et al .
1999b ) : @xmath57 where @xmath58 is the spin parameter of the black hole .
using this formula , and our derived accretion rates , black hole spin evolution , etc . , we can estimate the energy deposition rate ( @xmath59 ) for helium mergers of different initial helium star mass ( fig .
[ fig : md08]-[fig : md16 ] ) .
we stress that the variation in the energy deposition is almost certainly due to our low numerical resolution , and it will require much higher resolution simulations which can resolve any clumping structures to determine what fluctuations , if any , are produced in the accretion .
integrating @xmath59 , we get the estimates of the total explosion energy ( table [ tab : erg ] ) . with beaming into 3% of the sky , model 16a produces an equivalent isotropic energy of @xmath60ergs !
however , we should bear in mind that there are many uncertainties in these calculations : variations in the numerics ( compare models 8a , 8b , 8c ) , simplified assumptions in our accretion rates , and the uncertainties in the neutrino annihilation deposition . the errors in our calculations of the explosion energy may exceed 1 order of magnitude . because we chose high disk viscosities and assumed none of the energy produced would be dragged back into the accreting remnant , our estimates are likely to be upper limits to the burst energies .
even so , except for he - mergers involving massive helium cores ( @xmath61 ) , helium mergers powered by neutrino annihilation can not produce the most energetic grbs .
however , 22% of he - mergers involve the merger of compact remnants and massive helium stars ( @xmath62 : bttcher & fryer 2000 ) and the rate of massive helium - star mergers may well be high enough to explain the observed rate of grbs . more promising for the low accretion rates in he - mergers ,
mhd processes may extract the rotational energy of either the black hole or the accretion disk and power the grb ( blandford & znajek 1977 ; macdonald et al 1986 ; paczyski 1991 ; narayan , paczyski , & piran 1992 ; woosley 1993 ; hartmann & woosley 1995 ; thompson 1996 ; katz 1997 ; mszros & rees 1997 ; popham et al .
for example , the blandford - znajek luminosity is @xmath63 where @xmath64 is the spin parameter of the black hole and @xmath65 is the magnetic field strength in the disk .
we know the spin rate ( @xmath64 ) and mass ( @xmath16 ) of the black hole and we can get an rough upper limit to the magnetic field strength by setting the magnetic energy density equal to the kinetic energy density of accreting disk .
this yields : @xmath66 where @xmath67
. then we can estimate the maximum energy deposition rate for explosions driven by magnetic fields ( fig .
[ fig : bz08 ] ) .
when an explosion is produced , it can disrupt the disk and halt further accretion .
this will turn off the engine .
we assume the deposition continues for 200s ( a long burst ) and integrate the deposition rate to calculate maximum burst energies ( table [ tab : erg ] )
. burst energies above @xmath4ergs are possible for all our he - merger models .
these energies are most certainly overestimates , but lacking a working magnetic field mechanism , reliable calculations ( with errors less than a few orders of magnitude ) are impossible .
however , it is clear that he - mergers _ can _ produce grb explosions .
even with neutrino annihilation as the energy source , mergers of massive helium stars and compact remnants can explain long - duration bursts with inferred isotropic energies of above @xmath60ergs .
because he - mergers are similar to collapsars ( with slightly higher specific angular momenta ) , we believe they will have very similar characteristics to collapsars : they are likely to be beamed , and will produce supernovae - like explosions ( e.g. 1998bw ) .
this research has been supported by nasa ( nag5 - 2843 , mit sc a292701 , and nag5 - 8128 ) , the nsf ( ast-97 - 31569 ) , the us doe asci program ( w-7405-eng-48 ) and the los alamos feynman fellowship .
the authors gratefully appreciate stimulating conversations with andrew macfadyen and stan woosley on the subject of gamma - ray bursts and collapsars .
we thank stan woosley and tamara rogers for helpful comments on the manuscript .
ccccccc 8a & 2 & 8 & 1000 & 10000 & first & high + 8b & 2 & 8 & 1000 & 30000 & first & high + 8c & 2 & 8 & 1000 & 10000 & first & low + 8d & 2 & 8 & 1200 & 10000 & first & high + 8e & 2 & 8 & 1200 & 10000 & second & high + 4a & 2 & 4 & 650 & 10000 & first & high + 16a & 2 & 16 & 575 & 10000 & first & high cccccc 8a & 2.25 & 0.14 & 0.51 & 370 & @xmath68630 + 8b & 2.87 & 0.60 & 0.049 & 510 & @xmath68480 + 8c & 2.76 & 0.43 & 0.090 & 520 & @xmath68500 + 4a & 2.15 & 0.21 & 0.018 & 460 & @xmath68220 + 16a & 5.54 & 0.985 & 3700 & 65 & @xmath685500 + | there is growing observational evidence that gamma - ray bursts ( grbs ) are powered by black holes accreting rapidly through a disk . the supernova - like outburst that accompanies some gamma - ray bursts suggest that _ some _ long - duration grbs may be driven by the accretion of a rotating stellar core onto a central black hole .
such a system can be produced when a compact remnant spirals into the core of its binary companion . during the inspiral
, orbital angular momentum is injected into the helium core . by the time the compact remnant reaches the center of the helium core ,
it too has gained angular momentum as well as mass , producing a rapidly accreting black hole ( or neutron star ) at the center of a rotating stellar core .
whether or not such a merger ( termed he - merger ) can produce a grb depends upon the initial mass and spin of the central black hole , as well as the angular momentum in the stellar core . in this paper
, we use a 3-dimensional smooth particle hydrodynamics ( sph ) code to follow the he - merger process and make quantitative estimates of the initial mass and spin of the central compact remnant , as well as the angular momentum in the accreting helium core . during the inspiral , a @xmath0 compact remnant gains @xmath1 depending on the mass of the helium core ( more massive cores provide more accretion ) .
the accretion rates on the central remnant are initially very high , @xmath2 up to @xmath3 ( the accretion rate increases with increasing helium star mass ) , and the central remnant quickly becomes a black hole if it was not one already . from these accretion rates
, we estimate grb explosion energies . in all mergers ,
magnetically driven jets are expected to produce grb explosions with energies above @xmath4ergs .
for neutrino - annihilation - driven explosions , the grb energy increases dramatically with helium star mass : the merger of a @xmath0 compact remnant with a @xmath5 helium star only produces a @xmath6erg explosion in @xmath7s whereas the merger of a @xmath0 compact remnant with a @xmath8 helium star produces a @xmath9erg explosion in @xmath10s . |
the flavour changing neutral current ( fcnc ) decays of the @xmath0-meson are an important tool for investigating possible physics scenarios beyond the standard model ( sm ) , where such decays are forbidden at tree level .
it is for this reason that fcnc processes are very sensitive to possible small corrections that may be a result of any modification to the sm , or from some new interactions . of
the fcnc decays the radiative mode @xmath6 has been experimentally measured , with a lot of theoretical work also having gone into its study . a related decay ,
@xmath7 , has also been observed experimentally .
this latter process offers many more observables for confrontation with theory ( such as forward - backward ( fb ) asymmetries , polarizations and angular correlations between the final state particles etc . ) , where the theoretical work for this subject has spawned many investigations .
recently the radiative mode @xmath8 has also been observed with good limits also being set on the modes @xmath9 , where the @xmath10 s are the @xmath11 resonances .
the numbers for these rates are comparable to those for the @xmath12 resonance case and we may expect that with more data becoming available the related @xmath13 would be observed just as with the @xmath12 case . furthermore , as with the @xmath12 , such data will provide an independent opportunity to test the predictions of the sm .
in this paper we study the angular distribution of the rare @xmath0-decay @xmath14 , which may be expected to be observed in future b - factories .
we use the standard effective hamiltonian approach , and use the form factors that have already been estimated for the corresponding radiative decay @xmath15 .
the additional form factors for the dileptonic channel are estimated using the large energy effective theory ( leet ) , which enables one to relate the additional form factors to the form factors of the radiative mode .
our results provide , just like in the case of the @xmath12 resonance , an opportunity for a straightforward comparison of the basic theory with experimental results , which may be expected in the near future for this channel .
recall that the physical @xmath16 and @xmath17 states are the mixture of @xmath18 ( @xmath19 state ) and @xmath20 ( @xmath21 state ) states @xcite : k_1(1270 ) & = & k_1a sin+ k_1b cos , + k_1(1400 ) & = & k_1a cos- k_1b sin , [ eq:1 ] where @xmath22 is the @xmath23 mixing angle .
et al._@xcite proposed two possible solutions for this angle , namely @xmath24 , @xmath25 .
of these possibilities the negative values of the mixing angles predict the branching ratio ( br ) of @xmath26 to be more than that of @xmath27 , which is disfavoured from experimental data ( although not ruled out ) . for our work
we have taken the positive values of the mixing angles .
the paper is organized as follows : in section ii we will give the relevant effective hamiltonian and the leet form factors for the process under consideration . in section
iii we will give the expressions for the differential decay rate for the semi - leptonic decay mode under consideration .
finally we will conlude with our results in section iv .
the short distance contribution to the decay @xmath13 is governed by the quark level decay @xmath28 , and where the sm operator basis can be described by the effective hamiltonian : ^eff & = & v_ts^ * v_tb .
[ eq:2 ] we can rewrite the above effective hamiltonian in the following form : ^eff & = & v_ts^ * v_tb , + & = & v_ts^ * v_tb , [ eq:3 ] with v_&= & 1 2 ( |s _ b ) , [ eq:4 ] + a_&= & 1 2 ( |s _ _ 5 b ) , [ eq:5 ] + t _ & = & 1 2 ( |s _ b ) , [ eq:6 ] + t^5 _ & = & 1 2 ( |s _ _ 5b ) . [ eq:7 ] in equation ( [ eq:3 ] ) we have used the @xmath29 structure for the hadronic part ( except for @xmath30 ) . note that this structure does nt change under the transformation @xmath31 and @xmath32 .
furthermore , we can relate the hadronic factors of @xmath33 and @xmath34 by using the identity and that @xmath35 . ] : @xmath36 in this work we shall closely follow the notation used by kim _ et al._@xcite by defining the form factors of @xmath16 as : k_1 ( p ) | |s _
b | b ( p ) & = & - f _ ^ * - a_+ ( ^*.p ) ( p + p ) _ - a_- ( ^*.p ) ( p - p ) _ , [ eq:8 ] + k_1 ( p ) | |s _ _ 5 b | b ( p ) & = & - i g _ ^ * ( p + p)^(p - p)^ + & = & 2 i g _ ^ * p^ ( p)^[eq:9 ] , + k_1 ( p ) | |s _ _ 5 b | b ( p ) & = & g_+ _ ^ * ( p + p)^+ g_- _ ^ * ( p - p)^ + & & + h _ ( p + p)^(p - p)^ ( .p ) , [ eq:10 ] + k_1 ( p ) | |s _ b | b ( p ) & = & - i g_+ - i g_- + & & - i 2 h ( p_p_- p_p _ ) ( ^*.p ) .
[ eq:11 ] from the above equations we can observe that there are seven form factors which govern the @xmath37 transition , where we will now relate these form factors using the leet approach .
note that the advantage of writing the form factors in this form is that the expressions of the amplitudes ( @xmath38 ) are as given in equations ( 17 ) and ( 18 ) in kim _
et al._@xcite , due to the symmetry of the expressions under the exchange @xmath39 , @xmath40 , @xmath41 .
note also that cheng and chua have parameterized the tensorial form factors for the @xmath37 transition as @xcite : k_1a,1b(p ) | |s i _ q^(1 + _ 5 ) b | b(p ) & = & i _ ^ * p^q^y_a1,b1 + ( ^*_p.q - p_^*.q ) y_a2,b2 + & & + ^*.q y_a3,b3 , [ eq:12 ] where @xmath42 and @xmath43 .
the @xmath18 and @xmath20 states are the angular momentum eigenstates as defined in equation ( [ eq:1 ] ) . using equation ( [ eq:1 ] )
we can define the physical @xmath44 form factors as : y_i^b k_1(1270 ) = y_ai(q^2 ) sin+ y_bi(q^2 ) cos , i = 1 , 2 , 3 .
[ eq:13 ] the parameterizations of the form - factors @xmath45 , as used by cheng and chua , are given in appendix [ appendix : a ] .
these form factors can be related to the tensorial form factors given in equations ( [ eq:9])-([eq:12 ] ) by : y_1^b k_1(1270 ) & = & - g_+ , [ eq:14 ] + y_2^b k_1(1270 ) & = & - g_+ - g_- , [ eq:15 ] + y_3^b k_1(1270 ) & = & g_- + h ( p.q ) . [ eq:16 ] using the leet approach as given by charles _
et al._@xcite we can obtain the following relations between the form factors : f & = & 2 m e g , [ eq:17 ] + g_+ & = & - g m , [ eq:18 ] + g_- & = & g m , [ eq:19 ] + a_+ & = & - a_- = - ( g + h m ) , [ eq:20 ] where @xmath46 is the mass of the parent hadron and @xmath47 is the energy of the daughter hadron .
we now define all the form factors in terms of just two independent form factors ( @xmath48 and @xmath49 ) . using the leet relations
as given in equations ( [ eq:17])-([eq:20 ] ) and equations ( [ eq:13])-([eq:16 ] ) , the form factors for the @xmath37 transition can be related to cheng and chua s form factors as : g_+ & = & - y_1^b k_1(1270 ) , + g_- & = & y_1^b k_1(1270 ) , + g & = & , + f & = & 2 e y_1^b k_1(1270 ) , + h & = & , + a_+ & = & - a_- = , [ eq:21 ] where @xmath46 is the mass of the @xmath0-meson and @xmath50 is the mass of the @xmath10 .
in the following it is convenient to define our kinematics in terms of the following vectors : @xmath51 subsequently the decay mode @xmath52 , of the @xmath10 meson , can be parameterized by the matrix element @xcite : ( k_1 ( p ) ( p _ ) k ( p_k ) ) & = & + & = & .
[ eq:22 ] this matrix element will give the decay width @xcite : _
k_1 = q ( 1 + 2 3 ) , [ eq:23 ] with @xmath53 , and where @xmath54 , @xmath55 and @xmath56 are the momentum and polarization vectors of @xmath10 and @xmath57 respectively . in the following analysis
we shall neglect the masses of the leptons , the kaon and the @xmath57 , where in the above we have used @xmath58 and @xmath59 .
the final 4-body decay amplitude can be written as the sum of two amplitudes : = ( _ ) _ ( a_r^+ a_l^ ) , [ eq:24 ] where @xmath60 \,\,\ , , \label{eq:25}\\ { \cal a}_l^\beta & = & ( \bar \ell_l \gamma^\mu l_l ) \left ( a_l g_{\mu\nu } -b_l { p_\mu l_{\nu } } + ic_l ~\epsilon _ { \mu \nu \alpha \beta } p^\alpha l^\beta \right ) \frac{g^{\nu\alpha}-p^\nu p^\alpha / m_{k_1}^2 } { p^2-m_{k_1}^2 + i m_{k_1 } \gamma_{k_1 } } \nonumber \\ & & \hspace{1.5 cm } \times \bigg [ \left(p^2 + p.q\right ) g_{\alpha \beta } - p_\beta q_\alpha \bigg ] \,\,\ , .
\label{eq:26}\end{aligned}\ ] ] the @xmath61 , @xmath62 , @xmath63 and @xmath64 , @xmath65 , @xmath66 can be expressed as : @xmath67 + \frac{(c_9^{eff } -c_{10 } ) f}{2m_b}l^2 \,\,\ , , \label{eq:27}\\ b_l & = & -2c_7 ( g_+ - l^2 h ) -\frac{(c_9^{eff } -c_{10 } ) a_+}{m_b}l^2 \,\,\ , , \label{eq:28 } \\ c_l & = & -2c_7 g_+ -\frac{(c_9^{eff}-c_{10 } ) g}{m_b}l^2 \,\,\ , , \label{eq:29 } \\ a_r & = & -c_7 \left[2 ( p\cdot l ) g_+ + l^2 ( g_+ + g_- ) \right ] + \frac{(c_9^{eff } + c_{10 } ) f}{2m_b}l^2 \,\,\ , , \label{eq:30}\\ b_r & = & -2c_7 ( g_+ - l^2 h ) -\frac{(c_9^{eff } + c_{10 } ) a_+}{m_b}l^2 \,\,\ , , \label{eq:31}\\ c_r & = & -2c_7 g_+ -\frac{(c_9^{eff } + c_{10 } ) g}{m_b}l^2 \,\,\ , .
\label{eq:32}\end{aligned}\ ] ] -.5 in as such the decay rate can be computed , with the result : & = & | v_tb v_ts^ * ( ) |^2 ( |a_r|^2+|a_l|^2 ) , [ eq:33 ] where |a_\{l , r}|^2 = ( - g _ + ) ( a_\{l , r})^(a_\{l , r}^*)^ , [ eq:34 ] where the various angles used above are as shown in figure [ fig : kine ] .
recall that we shall present our results in terms of the vectors @xmath68 , @xmath69 , @xmath70 and @xmath71 , by use of the transformation : @xmath72 using the kinematics as prescribed in kim _
et al._@xcite , that is , where we set @xmath73 , @xmath74 and @xmath75 .
furthermore , we shall introduce various angles , namely @xmath76 as the polar angle of the @xmath57 momentum in the rest frame of the @xmath10 meson with respect to the helicity axis , _ i.e. _ the outgoing direction of @xmath10 .
similarly @xmath77 as the polar angle of the positron in the dilepton cm frame with respect to the @xmath10 momentum . and finally @xmath78 as the azimuthal angle between these planes , that is , the @xmath52 and @xmath13 planes . in this case : |a_l|^2 & = & , [ eq:35 ] |a_r|^2 & = & , [ eq:36 ] where @xmath79 .
the @xmath80 integration is performed using the narrow width approximation of the @xmath10 decay , _ i.e. _ : _ _ k_1 0 = ( p^2 - m_k_1 ^ 2 ) . as such
, the total decay width can be expressed as : & = & p^2 ( p^2 - m_k_1 ^ 2 ) _ 4 m_^2^(m_b - m_k1)^2 l^2 _
-1 ^ 1 d(cos_k ) _
-1 ^ 1 d(cos_+ ) _ 0 ^ 2 d + & & | v_tb v_ts^ * |^2 ( |a_r|^2+|a_l|^2 ) , with @xmath53 and @xmath75 .
the form factors for the radiative mode @xmath1 , as given by cheng and chua @xcite , assumed that the physical states @xmath16 and @xmath17 were mixtures of the angular momentum eigenstates @xmath18 and @xmath20 , where the mixing angle between these states is not known precisely ( though it is believed to be such as to cause the maximal mixing between the states ) . where the hypothesis of mixing between the two states naturally explains the suppression of one of the decay modes with respect to the other
. in reference @xcite the mixing angles suggested were @xmath81 .
the negative values of the mixing angles suggest the suppression of @xmath1 as compared to @xmath26 , which is disfavoured ( although not conclusively ruled out ) from the observation of the radiative decay mode @xmath82 by the belle collaboration @xcite .
although the prescription of mixing between the states helps to explain the suppression of one of the modes , as compared to the other , the form factors as given in reference @xcite predict a lower value of the branching ratio for @xmath1 as compared to experimental results .
note that there have been many attempts in references @xcite to address this issue , where these attempts essentially predict a much larger value of the zero recoil value of the form factors . for our analysis
we have used the form factors as given by cheng and chua @xcite .
our analytical results for the leet form factors and the differential decay rate retains the same form for any possible increase in the zero recoil value of the form factors . for our analysis
we have used the mixing angle between the two angular momentum eigenstates ( @xmath83 ) to be ( @xmath84 ) @xmath85 .
our sm value of the branching ratio for @xmath86 , using the input parameters as defined in appendix [ appendix : b ] , is @xmath87 .
-.5 in in figure [ fig:1 ] we have shown the variation of the differential branching ratio of @xmath86 as a function of the dileptonic invariant mass .
the two different lines in the plot correspond to the results of including and excluding the long distance charm resonances , where these long - distance effects can be included with the redefinition of @xmath88 . for this purpose
we have used the prescription as given in kruger and sehgal @xcite . note that use of the leet for obtaining the vector and axial vector form factors is justified more for relatively large values of @xmath89 , where at the moment there are no first principle determinations of these form factors similar to the ones determined for the tensor form factors by cheng and chua @xcite .
we therefore continue using the leet based form factors for the entire range . as and
when more accurate values are available for the low @xmath89-region our expressions can easily be reevaluated for a fresh plot . in figure [ fig:2 ]
we have shown the plot of the differential branching ratio as a function of the azimuthal angle between the planes @xmath90 and @xmath91 . and in our final plot we have shown the dependence of the differential branching ratio as a function of the scattering angle @xmath92 in @xmath91 , as defined in figure [ fig : kine ] .
we should like to note , at this point , that for any new physics model that can be effectively absorbed by the `` standard '' set of wilson coefficients in the effective hamiltonian , our analytic results given in previous sections can be used to obtain the corresponding change in the angular correlations .
as one final remark we may integrate our differential decay rate over the final state hadrons to get the decay rate of the process @xmath93 .
note that by using our defintion of the form factors we can relate these to the ones for @xmath12 , under the substitutions @xmath94 and @xmath95 , such that the corresponding decay rates are obtained easily ( by means of this substitution ) .
it is easily seen then that the location of the zero in the fb asymmetry of this integrated decay rate is the same as in the @xmath12 case , with the numerical value of the form factors being different . however , though the zeroes can be related , the overall shape of the fb asymmetries could be different from @xmath96 . to conclude , the mode which we have studied above , within the sm level , can in principal be measured at present @xmath0-factories .
various angular correlations of this mode can also be studied in future superb factories .
the study of the various angular correlations in @xmath86 can provide us with a very useful cross - checking tool for the sm and possible new physics in @xmath97 transitions . in this pursuit
we have given the leet form factors for @xmath98 , which could be useful in not only testing the operator structure of the sm but also the existence of possible new physics beyond it .
the work of src is supported by the department of science & technology ( dst ) , india .
src would also like to thank the theory division , kek , for their local hospitality , where the part of this work was done .
the work of ng is supported by jsps under the grant no .
firstly , it is important to note that the additional negative sign in equation ( [ eq:19 ] ) is due to the difference in our definition of @xmath99 , as compared to that of cheng and chua @xcite , charles _
et al._@xcite and kim _ et al._@xcite .
furthermore , note that the definitions of cheng _ et al .
_ are the same as that of charles _ et al . _ , but differ from that of kim _ et al .
_ by a sign , which can be taken into account by changing the sign in equation ( [ eq:19 ] ) . the form factors , as defined in cheng _
( for all the form factors except @xmath100)@xcite , are : f(q^2 ) = .
[ eq : app:1 ] for @xmath100 we use @xcite : f(q^2 ) = .
[ eq : ap:2 ] the numerical values of the factors appearing in equation ( [ eq:22 ] ) and equation ( [ eq:23 ] ) are given in table [ table:1 ] .
._the form factors for the @xmath101 and @xmath102_@xcite .
[ cols="^,^,^,^",options="header " , ]
@xmath103 gev , @xmath104 gev , @xmath105 gev , @xmath106 gev , @xmath107 gev , + @xmath108 , @xmath109 gev@xmath110 , @xmath111 , + @xmath112 c. s. kim , y. g. kim , c. d. lu and t. morozumi , phys .
d * 62 * , 034013 ( 2000 ) [ arxiv : hep - ph/0001151 ] .
j. charles , a. le yaouanc , l. oliver , o. pene and j. c. raynal , phys .
d * 60 * , 014001 ( 1999 ) [ arxiv : hep - ph/9812358 ] .
h. y. cheng and c. k. chua , phys .
d * 69 * , 094007 ( 2004 ) [ arxiv : hep - ph/0401141 ] .
l. roca , j. e. palomar and e. oset , phys .
d * 70 * , 094006 ( 2004 ) [ arxiv : hep - ph/0306188 ] .
h. yang _ et al .
_ , phys . rev
. lett . * 94 * , 111802 ( 2005 ) [ arxiv : hep - ex/0412039 ] . y. j. kwon and j. p. lee , phys .
d * 71 * , 014009 ( 2005 ) [ arxiv : hep - ph/0409133 ] ; g. nardulli and t. n. pham , phys .
b * 623 * , 65 ( 2005 ) [ arxiv : hep - ph/0505048 ] ; j. p. lee , phys .
d * 74 * , 074001 ( 2006 ) [ arxiv : hep - ph/0608087 ] .
f. kruger and l. m. sehgal , phys .
b * 380 * , 199 ( 1996 ) [ arxiv : hep - ph/9603237 ] . | flavour changing neutral current decays of the @xmath0-meson are a very useful tool for studying possible physics scenarios beyond the standard model ( sm ) , where of the many fcnc modes radiative , purely leptonic and semi - leptonic decays of the @xmath0-meson are relatively clean tests . within this context
the belle collaboration has searched for the @xmath1 process and provided an upper bound on this decay . in this work
we have used this upper bound in studying the angular coorelations for the related semi - leptonic decay mode @xmath2 , where we have used the form factors that have already been estimated for the @xmath1 mode . note that the additional form factors that are required were calculated using the large energy effective theory ( leet ) .
kek - th-1148 , + lycen 2007 - 03 .
s. rai choudhury@xmath3 , a. s. cornell@xmath4 , naveen gaur@xmath5 1.5 cm |
gravitational microlensing is now established as a powerful tool for various aspects of galactic science .
the concept of gravitational microlensing , where a background object is temporarily brightened by the gravitational lensing effect of a passing foreground object , is believed to have been initially conceived by einstein around 1912 @xcite . as he stated in a later publication @xcite
, his opinion was that `` there is no great chance of observing this phenomenon '' .
fortunately the idea was later resurrected by several authors @xcite , in particular @xcite . following the proposal of paczyski
, many experiments began to hunt for microlensing events towards the galactic bulge and the magellanic clouds : ogle @xcite , eros @xcite , macho @xcite , moa @xcite .
to date thousands of microlensing events have been detected toward several directions , including the galactic bulge @xcite , magellanic clouds @xcite , and m31 @xcite and even in surveys of bright nearby stars @xcite .
the purpose of this paper is to deal with a previously under - exploited set of targets , namely quasars . although quasar microlensing has been observed for many years ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) this has , with a few exceptions ( e.g. * ? ? ?
* ) , almost exclusively been lensing of background quasars by compact objects in the intervening ( strong ) lensing galaxy .
in contrast , we are interested in the lensing of background quasars by stars in the milky way .
this is a particularly prescient subject because imminent large - scale optical surveys , such as pan - starrs @xcite and lsst @xcite , will monitor the whole available sky every few nights down to magnitudes which are sufficient to monitor several million background quasars .
if large enough samples of such microlensing events could be accrued then it will be possible to use them to constrain models of the galaxy .
one reason why quasars are potentially important targets is that they break a number of the degeneracies known to inhibit the analysis of microlensing events from sources within our own galaxy . in general , for lensing of galactic sources it is difficult to determine the properties of the lensing population because all of the information about the event is contained in one observable , namely the event time - scale . as a consequence ,
there are a number of degeneracies which need to be broken in order to determine , for example , the lens mass or distance . for sources within the milky way , in general , we do not know the distance or transverse velocity of the star which is lensed . for quasars
the interpretation is made significantly easier because we can assume that the source is at infinity and that the transverse velocity is negligible . in this case
the lens mass is simply a function of the ( known ) time - scale and the ( unknown ) lens distance and velocity , @xmath0 where @xmath1 and @xmath2 are the lens distance and proper motion , respectively , and @xmath3 is the event time - scale .
this makes the interpretation of quasar microlensing events significantly easier than that of galactic sources and opens up the possibility of carrying out model - independent mass determinations using high precision astrometry from , for example , esa s gaia mission @xcite . even without a measurement of the lens distance / velocity
, it could also be possible to use higher - order microlensing effects such as the microlensing parallax @xcite to break the degeneracy .
clearly quasar microlensing has the potential to become an important avenue for probing both stellar and galactic astrophysics . in this paper
we carry out monte carlo simulations to address this issue , presenting an all - sky prediction of the rate and optical depth of microlensed quasars .
the paper is organised as follows . in section [ sec :
model ] , we describe the model and theory . in section [ sec : results ] , we present the main results , dealing with the time - scale distribution ( [ sec : te ] ) , the predictions for the optical depth and event rate ( [ sec : event_rate ] ) and also the prospects for determining the lens masses ( [ sec : lens_mass ] ) .
we present a summary and discussion in section [ sec : conclusion ] .
we utilise a monte carlo simulation to give an estimate of the rate and optical depth for microlensed quasar events .
the details of the simulation are described in the following subsections .
our microlensing model is constructed following @xcite and is similar to a number of previous works @xcite . in this work
we want to obtain an all sky estimate of microlensed quasars by the milky way stars , which requires a model for the stellar mass distribution . to do this we follow the model of ( * ?
* hereafter bt08 ) . unless otherwise stated all parameters are taken from model in table 2.3 of bt08 and the solar radius is taken to be 8 kpc .
it is immediately clear that the stellar halo s contribution is negligible .
thus , only bulge and disc stars are considered in this paper .
the bulge is treated as a bar with its longest axis inclined by about @xmath4 to the line from sun to the galactic center and the stellar mass distribution in the bar is given by @xmath5 where @xmath6 .
the bulge is truncated at a radius @xmath7 .
the three dimension cartesian coordinates ( @xmath8 , @xmath9 , and @xmath10 ) are shown in fig .
[ fig1 ] . for the bar parameters we adopt @xmath11 , @xmath12 , @xmath13 , @xmath14 , and @xmath15 .
we have set the normalisation of the bar mass so that the bar has the same mass as the original bulge ( @xmath16 ) .
the disc is modeled as the sum of two double - exponentials , which correspond to the thin- and thick - disc components .
the galactic disc density distribution in cylindrical coordinates is @xmath17 where @xmath18 is the central surface density , @xmath19 is the radial distance , @xmath20 is the disc scale - length ( both components are assumed to have the same scale - length ) , @xmath21 , and @xmath22 are the scale - heights for the thin and thick components , respectively , and @xmath23 and @xmath24 are the respective normalisations .
the total mass of the disc in this model is @xmath25 . for the mass function
we use the four part power - law distribution model of @xcite , @xmath26 the normalisation @xmath27 is obtained through the integral , @xmath28 where @xmath29 is the stellar density in the solar neighbourhood ( which can be derived from equation [ eq : disc ] ) .
we find that @xmath30 .
note that unlike some previous authors we do not adopt different mass functions for different populations ; as argued by @xcite , there is no clear evidence to support strong variations in the initial mass function .
we invoke the following simple approach for dealing with stellar evolution . the objects with initial masses @xmath31 and @xmath32 are assumed to become brown dwarfs and main - sequence stars , respectively .
the stars with initial mass @xmath33 are assumed to evolve into @xmath34 white dwarfs .
the stars with @xmath35 are assumed to evolve into @xmath36 neutron stars , and more massive stars are assumed to evolve into @xmath37 black holes .
we assume that the lens number density is proportional to the stellar mass density given in equations ( [ eq : bar ] ) and ( [ eq : disc ] ) ( i.e. @xmath38 ) . for convenience , we rewrite the total stellar distribution equation as follows , @xmath39 from this we can deduce the lens number density within a mass range , @xmath40 , @xmath41 and a velocity range @xmath42 , @xmath43 , as follows , @xmath44 where @xmath45 are the galactic coordinates and @xmath1 is the distance to the lens , @xmath46 is the lens mass function given in equation ( [ eq : imf ] ) and @xmath47 is the dimensionless mass distribution ( see equation [ eq : rho ] ) .
in addition to the lens mass , we also consider the lens luminosity .
this allows us to investigate the possibility of detecting the lenses and measuring their proper motion and distance , which will break the degeneracy in equation ( [ equ1 ] ) and allow a model - independent measurement of the lens mass . for the main - sequence stars we estimate the lens brightness using the mass - luminosity relation of @xcite , while we assume that all non - main - sequence lenses are dark .
the mass - luminosity relation of @xcite is given in the @xmath48-band in the johnson - cousins @xmath49 system . in order to compare with lsst observations ( see section [ sec : opdepth ] )
we need to transform the @xmath48-band luminosity into a luminosity in the lsst @xmath50 photometry system .
to do this we use the empirical colour transformations given by @xcite .
as the lenses are distributed at different distances from the sun , we need to carefully consider the galactic dust extinction . to do this we utilise the galactic dust map of ( * ? ? ?
* hereafter sfd ) , combining this with a model of the interstellar medium from bt08 , as follows .
we first construct a model to calculate the total extinction along a given line - of - sight for the photometric band @xmath8 , @xmath51 where @xmath52 is a constant ( i.e. this does not vary across the sky ) and the distribution of the interstellar medium ( @xmath53 ) is modelled using equation ( 2.211 ) of bt08 .
we estimate @xmath52 by carrying out a least - squared fit between @xmath54 and the observed value of the total extinction from sfd using @xmath55 lines - of - sight uniformly distributed across the sky .
once we have estimated @xmath52 we can calculate the extinction that a lens is subject to using the following formula , @xmath56 to calculate the event rate we need to know the velocities of the observer , lenses and sources . for simplicity
we neglect the motion of the earth around the sun , the so - called parallax effect @xcite , returning to this briefly in section [ sec : lens_mass ] . for the calculation of microlensing event rate only the relative lens - source transverse velocity ( i.e. the transverse velocity of the lens , in the lens - plane , relative to the observer - source line of sight ) , @xmath57 ,
is needed , which can be written , @xmath58 where @xmath1 and @xmath59 are the distances to the lens and source , respectively , and @xmath60 and @xmath61 are the components of the velocity tangential to the line - of - sight for the observer and lens , respectively ) for lenses residing at non - cosmological distances ( see , for example , appendix b of * ? ? ?
] we model the velocity distributions of the lens stars in the milky way using gaussian distributions , adopting the solar motion with respect to the local standard of rest from @xcite .
we take two separate tri - variate gaussians for the thin and thick discs , using the means and dispersions from table 2 of @xcite . for the bar stars ,
the random velocities are assumed to have gaussian distributions with @xmath62 @xcite along the three axes of the bar . here
, the coordinates @xmath63 are centred at the galactic centre ; @xmath64 represents the longest axis of the bar , @xmath65 is directed towards the north galactic pole .
the bar velocity dispersion needs to be computed in the galactocentric cylindrical coordinate system by , @xmath66 where @xmath67 are the velocity dispersions along the @xmath68 axis as shown in fig .
[ fig1 ] , which are calculated from @xmath69 given the assumed bar angle of @xmath4 . the resultant velocity dispersion is a function of the azimuth @xmath70 .
the rotation component of the bar velocities @xmath71 are estimated following @xcite , @xmath72 where @xmath73 is adopted . for the calculation of the relative lens - source transverse velocity , @xmath57
, we need to convert the lens velocity components from galactocentric cylindrical coordinates into solar - centric spherical coordinates .
the conversion is done as follows , @xmath74 where @xmath75 are the velocity components in solar - centric spherical coordinates corresponding to the radial , galactic longitude and galactic latitude components , respectively .
the angle @xmath76 is the angle between the lines connecting the sun , the projected position of the lens on the galactic plane , and the galactic centre ( see fig .
[ fig1]),which can be computed by , @xmath77 where @xmath78 .
are the galactic coordinates of the star , @xmath79 are the galactocentric cylindrical coordinates , and @xmath80 is the galactocentric distance of the sun .
the vectors @xmath81 , @xmath82 , and @xmath83 are the unit vectors in the solar - centric spherical coordinates . ]
quasars are very energetic and distant galaxies with an active galactic nucleus . by assuming an isotropic distribution , the distribution of quasars can be given by the so called quasar luminosity function ( qlf ) .
a lot of work has been carried out on the qlf in recent years @xcite , especially with the advent of surveys such as the two degree field ( 2df ) and sloan digital sky survey ( sdss ) . in this paper
we assume that the quasars have a constant surface number density and follow the method of @xcite to give an estimate of this number density .
we utilise the @xmath84 cold dark matter ( @xmath84 cdm ) model in a flat universe ( i.e. @xmath85 ) with @xmath86 , @xmath87 and @xmath88 , where @xmath89 is the hubble constant in units of @xmath90 .
we then simply integrate the qlf in the redshift interval @xmath91 to derive the surface number density of quasars in the @xmath92- and @xmath10-bands will have a negligible contribution to the total surface number density ( see fig
. 1 of * ? ? ?
we denote the total surface density of quasars down to a particular limiting magnitude as @xmath93 and @xmath94 for the two photometric bands . fig .
[ fig2 ] shows the resulting distributions . - and @xmath10-bands .
] since quasars are extragalactic sources , some of the light emitted from them will be absorbed by the dust in the milky way .
we take this into account using the extinction map of sfd .
the motivation for including the @xmath10-band in our analysis is to see whether the reduced extinction for this longer wavelength increases the quasar microlensing event rate compared to @xmath92-band .
the microlensing optical depth is defined as the probability that a particular background source falls into the einstein radius of any foreground lens star .
as has been mentioned above in section [ sec : qso_lf ] , we assume that the quasars are distributed uniformly across the sky . if we denote the surface number density as @xmath95 , the optical depth along a given line - of - sight can be calculated following @xcite , @xmath96^{-1}\nonumber \\ & \approx & { 4\pi g \over c^2}\int_0^{{d_{\rm s}}}d{d_{\rm l}}\chiup(l , b,{d_{\rm s}})d\int_{m_1}^{m_2 } m \xi_0(m)dm,\end{aligned}\ ] ] where @xmath97 is gravitational constant , @xmath98 for @xmath99 , and @xmath100 and @xmath101 are the mass limits of the lens stars .
note that the optical depth is independent of @xmath95 .
the event rate is the number of microlensing events that occur per unit time , which we again calculate following @xcite , @xmath102^{-1 } \nonumber \\ & \approx & { 4\sqrt{g } \over c}\int_0^{{d_{\rm s } } } d{d_{\rm l}}\chiup(l , b,{d_{\rm l}})d^{1/2 } \int_{m_1}^{m_2}dm \xi_0(m)m^{1/2 } \nonumber \\ & & \cdot \iint dv_l dv_b v_t f(v_l , v_b).\end{aligned}\ ] ] during a time interval @xmath107 , the total number of expected microlensing events is trivially , @xmath108 where @xmath109 is the total event rate integrated over all lines - of - sight and @xmath110 is the detection efficiency . for simplicity , we assume @xmath111 per cent .
this can be justified when one considers that the cadence of the surveys such as lsst will be of the order of a few days , which is suitable to detect practically all events in our time - scale range ( see section [ sec : te ] ) .
we return to this issue in the following section .
using the model described above , we are able to investigate the time - scale distributions for quasar microlensing events . fig . [
fig : te ] shows the predicted event rate as a function of time - scale for the different galactic components .
the average time - scales for the bar , thin disc , thick disc and total ( bar plus discs ) are @xmath112 , @xmath113 , @xmath114 and @xmath115 , respectively .
the longest time - scales are obtained for thin - disc events towards @xmath116 and @xmath117 degrees . in these regions
the mean time - scale can reach as high as 130 days ; these events will be very well sampled and are likely to display microlensing parallax signatures ( see section [ sec : lens_mass ] ) .
note that the average time - scale for disc - lensing events reaches a minimum around the central parts of the galaxy .
the average in this region is reduced due to lenses on the far - side of the bar ; these have large transverse velocities owing to their rotational velocity acting in the opposite direction to the sun s .
the events with shortest time - scales are those caused by lenses in the galactic bar , because bar lenses have larger velocity dispersions and hence larger transverse velocities .
furthermore , unlike nearby disc - lenses , they do not share the disc s rotation velocity .
the combined map in the lower - panel of fig [ fig : te ] closely resembles the time - scale map for the thin disc , since the majority of events are caused by thin - disc lenses ( see section [ sec : event_rate ] ) .
arguably the most important quantity which we can calculate is the total event rate , i.e. the number of events which we expect future surveys will be able to detect .
this is given in table [ tab : event_rate ] for various apparent magnitude limits .
not surprisingly the total event rate increases rapidly with increasing depth , reflecting the fact that the number density of quasars rises steeply for fainter limiting magnitudes ( as shown in fig . [ fig2 ] ) .
the contributions from the various milky way components are shown in table [ tab : event_rate ] .
we can clearly see that the total event rate is dominated by the disc component , including the thin and thick disc components .
the thick disc component contributes about @xmath118 per cent of the whole disc events and therefore plays a non - negligible role in quasar microlensing .
although the total mass of the thin disc is @xmath119 times more massive than the thick disc , the fact that the thick disc is less concentrated in regions of high extinction helps to boost its lensing signal .
the bar component , with only around 10 per cent of the total mass of the disc , has significantly fewer events . as has been stated in section [ mod - mass - distribution ] , the stellar halo will have a negligible contribution .
lcccc + @xmath120 ( mag ) & bar & thin disc & thick disc & total + + + 21 & 0.07 & 0.55 & 0.17 & 0.79 + 23 & 0.51 & 3.53 & 0.86 & 4.90 + 24 & 0.97 & 6.94 & 1.56 & 9.54 + 27 & 5.13 & 34.75 & 6.92 & 46.80 + + + 21 & 0.12 & 0.91 & 0.23 & 1.26 + 23 & 0.77 & 5.40 & 1.09 & 7.26 + 24 & 1.50 & 10.31 & 1.95 & 13.76 + 27 & 7.30 & 48.82 & 8.26 & 64.38 + the all - sky distributions of average optical depth and the event rates are presented in fig .
the different rows of this figure show the average optical depth and the event rate for different apparent magnitude limits . as remarked above , the event rate increases with the increase of the apparent magnitude limit owing to the increase in source density .
the event rate also increases with decreasing galactic latitude , tracing the scale - height of the discs .
however , as one reaches close to the galactic plane extinction becomes important , effectively ruling out the ability to detect events with @xmath121 .
although one might expect this problem to be alleviated by operating at longer wavelengths , the gradient of the extinction ( as a function of @xmath122 ) is so great that even in the @xmath10-band the event rate for the bar and disc only increases by around 30 - 50 per cent ( see table [ tab : event_rate ] ) .
the most crucial question which we can pose is how many quasar microlensing events do we expect future surveys to detect ?
there are a number of upcoming surveys which will map the entire visible sky every few nights .
two important ones are pan - starrs @xcite and lsst @xcite , both of which operate with a cadence of every few nights .
we assume a 100 per cent detection efficiency , which should be a fair approximation since practically all events will have durations longer than this cadence ; from fig .
[ fig : te ] we find that 95 per cent of all events have time - scales greater than 7.5 days ( possible limitations to this assumption are discussed in section [ sec : conclusion ] ) .
pan - starrs will reach a @xmath92-band ( @xmath10-band ) magnitude limit of about @xmath123 ( @xmath124 ) mag and will cover about @xmath125 square degrees , while lsst will reach an @xmath92-band ( @xmath10-band ) magnitude limit of about @xmath126 ( @xmath127 ) mag per visit and cover about @xmath128 square degrees . in the @xmath92-band we estimate that pan - starrs will detect three events per year , while lsst will detect around five events per year .
clearly these numbers depend on the detectability of events , which is an issue we will return to in the discussion . as mentioned above ,
conducting the survey in the @xmath10-band does not increase the number of detected events significantly .
furthermore , for both of these surveys the @xmath10-band limiting magnitude is around one magnitude shallower than in the @xmath92-band , which cancels out any advantage gained from the lower extinction . as a consequence of these two factors ,
the total number of events which would be detected in this band is only 1 and 4 for pan - starrs and lsst , respectively , meaning that such a survey is optimally carried out in the @xmath92-band .
in practice we believe that these are lower - limits to the number of detected events , because the above calculation is based on events for which the lens and source are aligned to within one einstein radius .
this corresponds to a minimum amplification of 0.32 magnitudes , which is easily within reach of these surveys with their milli - magnitude ( or better ) photometric precision .
for example , if we allow for events within two einstein radii ( corresponding to a minimum amplification of around 0.06 mag ) then the event rate will increase by a factor of 2 . although it will be challenging to identify events with such small amplification , it does means that surveys like lsst and pan - starrs could potentially observe around 16 events per year .
it is important to investigate the characteristics of the lens population , particularly with respect to potential follow - up work on the lenses . in section [ mod - mass - function ]
we described our simplistic implementation of stellar evolution .
from equation ( [ eq : imf ] ) it is trivial to obtain the relative contribution of the different lens populations .
this is summarised in table [ tab : lens_population ] .
we see that by far the majority of lenses are main - sequence stars ( @xmath129 per cent ) , although a significant fraction are white / brown dwarfs .
around 1 per cent of events will be caused by neutron stars , which raises the remote , yet interesting , possibility of quasar lensing from pulsars .
as has been highlighted by previous authors ( e.g. * ? ? ?
* ) , pulsar microlensing is of great interest because for these events it may be possible to determine both the distance and transverse velocity through radio observations ( and hence , via equation [ equ1 ] , a model independent mass determination ) .
however , the most promising avenue for carrying out model - independent mass determinations is for main - sequence stars .
this is important because microlensing is the only method which can be applied to single stars ( as opposed to all other methods of mass determination which are applicable only to stars in binary systems ) .
as has been stated in the introduction , to break the microlensing mass degeneracy requires a measurement of both the distance to the lens and its transverse velocity . however , to measure these lens properties requires the lens to be both bright and nearby .
we investigate this by plotting maps of the event rate for various lens properties in figs .
[ fig : lens_luminosity_marg ] to [ fig : lens_propermotion ] . in the near future instruments such as the eso gaia mission
@xcite will measure parallaxes and proper motions to unprecedented accuracy for many millions of relatively bright stars .
its accuracy degrades rapidly for fainter stars , reaching accuracies of around 0.1 - 0.2 mas for stars at the magnitude limit of around @xmath130 mag @xcite .
this corresponds to a parallax ( and hence distance ) error of around 20 per cent at a distance of 2 kpc . unfortunately , as can be seen in fig .
[ fig : lens_luminosity_marg ] , we predict that very few lenses will be brighter than 20th magnitude in @xmath92-band . in fig .
[ fig : lens_mag ] we have plotted the map of event rate as a function of lens brightness and distance . from this we can estimate the total event rate for lenses bright enough and near enough to be within reach of gaia . equation ( [ equ1 ] ) shows that the error on the lens mass will be entirely dominated by the uncertainty in the parallax ; if we scale the gaia errors linearly from the bright end to the faint end ( i.e. from 0.01 mas at @xmath131 mag to 0.1 mas at @xmath132 mag ) and consider events for which we will be able to obtain a parallax error of 20 per cent , we find that we will obtain around 1 event every two years over the whole sky for quasars down to @xmath133 . if we allow for weakly magnified events ( as above ) , this means that lsst will be able to detect around one event per year for which we will be able to robustly determine the distance ( and hence lens mass ) to within 20 per cent accuracy using a parallax measurement from gaia .
another potential source of accurate parallax measurements is the proposed sim - lite space mission @xcite , which , if it proceeds , will enable significantly better mass determinations . with an astrometric accuracy of 0.01 mas down to @xmath134 mag , this will be able to follow - up a similar number of events to gaia but with a ten - fold improvement in the accuracy . even though the precision of the mass recovery will be greatly improved , the total number of events will not increase dramatically for sim - lite since the limiting magnitude will be similar to gaia ; the only advantage to the event rate is the fact that the increased accuracy means it will be able to detect parallaxes for more distant lenses , which results in a doubling of the event rate compared to gaia ( i.e. around 2 events per year for lsst using @xmath133 quasars ) .
even if we are unable to obtain accurate trigonometric parallax measurements , it will still be possible to accurately determine the proper motion for a significant number of lenses using ground - based astrometry .
for example , lsst will be able to carry out astrometry at around 3 mas for @xmath135 mag . as can be seen from the lens luminosity function ( fig .
[ fig : lens_luminosity_marg ] ) this covers a significantly larger number of events , many of which will have measurable proper motion ( see fig .
[ fig : lens_propermotion_mag ] ) .
this would almost entirely break the lens mass degeneracy , but to obtain a unique mass measurement requires additional information , for example the distance ( from photometric / spectroscopic techniques ) or through higher - order microlensing effects .
one potential higher - order effect is microlensing parallax @xcite , where the motion of the earth as it orbits the sun induces perturbations on the standard microlensing signal .
simulations of parallax frequency show that for distant sources the rate of parallax events towards the galactic bulge can reach around 10 per cent ( e.g. fig
. 7 of * ? ? ?
this fraction will probably increase for lines - of - sight away from the bulge , since the velocity dispersion of disc stars is smaller than bulge stars and hence the event time - scale will increase ( and hence the parallax fraction will also increase as this is related to time - scale ; e.g. see fig . 6 of * ? ? ?
* ) . as is shown in fig .
[ fig : te ] , we find that 10 per cent of all events are predicted to have time - scales greater than three months .
.the fraction of different lenses by population , following our simplistic implementation of stellar evolution ( see section [ mod - mass - function ] ) .
these fractions are obtained trivially via the imf ( equation [ eq : imf ] ) . [
cols="<,^ " , ] -band lens magnitudes , after taking into account extinction .
note that this corresponds only to the events caused by main - sequence lenses , i.e. a further 22 per cent of events have non - main - sequence lenses which we classify as ` dark ' ( see table [ tab : lens_population ] ) . ] . note that this corresponds only to the events caused by main - sequence lenses , i.e. a further 22 per cent of events have non - main - sequence lenses which we classify as ` dark ' ( see table [ tab : lens_population ] ) . ] , but for lens magnitude and proper motion . ] , but for lens distance and proper motion .
note that unlike figs .
[ fig : lens_mag ] and [ fig : lens_propermotion_mag ] , this plot also includes the non - main - sequence lenses ( see table [ tab : lens_population ] ) . ]
in this paper we have carried out monte carlo simulations to make predictions for the microlensing of quasars by stars ( and stellar remnants ) in the milky way . using a simple model of the galaxy
we have predicted the all - sky microlensing optical depth and have calculated the time - scale distributions and event rates for future large - area sky surveys , both for @xmath92- and @xmath10-band .
we found that the total event rate increases rapidly with increasing magnitude limit , and predict that around 10 to 20 events will be detected per year from surveys like pan - starrs and lsst ( see section [ sec : event_rate ] ) .
most events are caused by lenses in the thin disc , even though extinction hampers the detectability of low - latitude events . if we could identify and monitor large numbers of quasars at low latitudes ( e.g. * ? ? ?
* ) , then it may even be possible to increase the number of detected events .
we have also highlighted how such quasar microlensing can be used to carry out robust and model - independent lens mass determinations , predicting that with accurate astrometry from gaia or sim - lite it will be possible to accurately recover the lens mass for around one event per year .
microlensing is uniquely placed to provide mass measurements for individual stars ( as opposed to those in binary systems ) and as such will undoubtedly play a crucial role in stellar astronomy .
although these results are very promising , there are a couple of technical challenges which must be overcome in order to detect these microlensing events . in this paper
we have assumed a detection efficiency of 100 per cent , which we have justified by noting that practically all events should be reasonably well - sampled by surveys like lsst . however , there are practical difficulties which will potentially reduce the efficiency .
firstly , it is well - known that quasars themselves are intrinsically variable ( e.g. * ? ? ? * ) . they can exhibit both long - term low - level variability as well as outburst phenomena .
indeed , one recent paper claimed to have detected a spectacular uv flare in a quasar behind m31 @xcite , although the authors admit that an alternative explanation could be microlensing .
some authors have claimed that nearly all quasars are microlensed by intervening ( cosmological ) compact objects ( e.g. * ? ? ?
* ) , although more recent studies have challenged this hypothesis ( e.g. * ? ? ?
this latter paper analyses a variability diagnostic ( the structure function ) of around 40,000 quasars , concluding that the most - likely cause of quasar variability is flares due to disc instabilities .
they remark that the variability is , in general , both chromatic and asymmetric .
this is an important issue since the microlensing events which we are considering in this paper should be both achromatic and symmetric .
therefore one avenue for disentangling microlensing from intrinsic variability would be to utilise the multi - band photometry from surveys such as lsst .
in addition , the amplitude of the quasar variability is generally at a lower level than our microlensing amplitude ( e.g. @xcite show that the rms scatter for quasars is @xmath136 mag in the @xmath137-band and , because the rms decreases for longer wavelengths , the variatbility will be even less for microlensing surveys carried out in @xmath92- or @xmath10-bands ) .
although most microlensing events are symmetric and achromatic , this is not always the case .
there are higher - order microlensing effects which can cause deviations from the standard symmetric and chromatic form , such as binarity in the lens or finite - source effects
. both of these will probably be of only limited importance ; for standard galactic microlensing the fraction of events which show clear binary signatures is of the order 3 per cent ( e.g. * ? ? ?
* and references therein ) , while for finite - source effects it is even less .
although it is safe to assume that the fraction of binary lens events for quasar sources will be similar to that for bulge sources , there is no reason to assume this will also hold for finite - source effects . if we assume that a typical quasar accretion disc is of the order @xmath138 cm , then the angular size will be less than a few @xmath139as .
this angular size is comparable to a giant star in the galactic bulge , which means that the finite - source effect for quasar microlensing should be no - more important than for standard galactic microlensing ; for these events finite - source effects only come into play when the amplification reaches around one hundred or more ( e.g. * ? ? ?
therefore for the vast majority of quasar microlensing events the light - curves will indeed be symmetric and chromatic . for the limited number of events
for which the magnification is sufficient to detect finite - source signatures , microlensing will enable constraints to be placed on the profiles of quasar accretion discs , although this could only be done in very fortuitous circumstances ( e.g. when a low - redshift quasar is very highly magnified ) .
a second obstacle to detecting microlensing of quasars is the difficulty of identifying the quasars in the presence of a luminous lens .
the quasars will be identified via cuts in colour - colour space , and so if the lens is sufficiently bright it could compromise the quasar selection .
fortunately in most cases the lenses will be significantly fainter than the sources ( see fig .
[ fig : lens_luminosity_marg ] ) .
we calculate that for only a quarter of lsst events will the lens be brighter than the source and for just over half of all events the lenses will be fainter than the single - epoch magnitude limit of @xmath133
. however , this issue will be particularly important for the events from which we wish to determine the lens mass , since for these we @xmath140 the lens to be luminous so that its distance and proper motion can be determined .
for these events it will be important to spectroscopically confirm the presence of a quasar behind the lens , or to obtain high - resolution imaging a number of years after the event so that the lens and quasar can be resolved .
in contrast to the above challenges that hamper the detection of quasar microlensing events , there are avenues which could enhance our ability to detect events .
if we could build up an accurate astrometric map of the foreground lenses and their proper motion ( using , for example , data from the gaia mission ) then it could be possible to @xmath141 which quasars will be lensed in advance of the amplification .
this method has been suggested as an approach to detect lensing by pulsars @xcite but it is equally applicable to any lens with an accurate proper motion measurment .
furthermore , this could be applied to fainter quasars , such as those provided by the complete co - added lsst survey , which will reach a depth of around @xmath142 . at this magnitude
the number density of quasars climbs to around 3000 per square degree ( see fig .
[ fig2 ] . )
another avenue is that of astrometric microlensing where , instead of detecting the photometric lensing signatures , we identify the astrometric shift of the image centroid as the quasar is lensed ( e.g. * ? ? ?
* ; * ? ? ?
this will be a demanding task as it requires very high precision astrometry for our faint source population ; the maximum astrometric shift is 0.354 times the einstein radius @xcite and for our events typical einstein radii are around 1 mas . as above
we could envisage targeting specific quasars based on predictions from lens proper motions , but this would still require astrometry at a level of at least tens of @xmath139as . in this work
we have only considered lenses within our own galaxy , yet one might wonder whether quasars could be lensed by stellar mass objects at cosmological distances . however , this is a remote possibility as it would require a very high density of such objects and even then the time - scales for such events are a couple of orders of magnitude longer than what we find for galactic lenses ( see , for example , * ? ? ? * ) ] . the notable exception to this , as noted in the introduction ,
is the phenomena of microlensing of macro - lensed ( i.e. multiple - imaged ) quasars by stellar mass lenses in the intervening lensing galaxy . here
the density of lenses is sufficient to produce an optical depth of order unity and hence detecting microlensing becomes feasible .
surveys such as lsst and pan - starrs will have an immense impact on this field as they will monitor the light - curves for many thousands of multiply - imaged quasars .
although we are unlikely to find events from lenses at cosmological distances ( with the exception of macro - lensed quasars ) , it may be possible to detect lenses from populations just outside our own galaxy .
for example , one could identify quasars behind the lmc ( e.g. * ? ? ? * ) or m31 @xcite and study their lensing by foreground stars in these galaxies . as we mentioned above
, there may have already been an event of this kind detected in a quasar behind m31 @xcite . in conclusion
, we have shown that imminent ground - based surveys will be able to detect a number of quasar microlensing events every year .
the efforts of pan - starrs and lsst will , after five years of operation , detect up to one hundred events .
although this is still too few events to constrain galactic models due to the overwhelming numbers of parameters , it will allow for valuable consistency checks to be carried out on these models and could lead to exciting and unexpected discoveries .
the authors are deeply indebted to shude mao for advice and guidance throughout this work and to a helpful referee who enabled us to improve the clarity of a number of issues .
jw is supported by the national basic research program of china ( grant no .
2009cb824800 ) .
mcs acknowledges support from the peking university one hundred talent fund ( 985 ) and nsfc grants 11043005 and 11010022 ( international young scientist ) .
alcock c. et al .
, 1993 , nature , 365 , 621 anguita t. , faure c. , yonehara a. , wambsganss j. , kneib j .-
, covone g. , alloin d. , a&a , 2008 , 481 , 615 aubourg e. et al . , 1993 , nature , 365 , 623 bailer - jones c.a.l . , 2009 , in j. andersen , j. bland - hawthorn , & b. nordstrm , eds , proc .
254 , the galaxy disk in cosmological context , p. 210
bastian n. , covey k.r . , meyer m.r . , 2010 , ara&a , 48 , arxiv:1001.2965 binney j. , tremaine s. , 1987 , galactic dynamics ( princeton ; princeton univ . press ) binney j. , tremaine s. , 2008 , galactic dynamics ( 2nd ed . ; princeton univ . press )
( bt08 ) bond i. a. et al . , 2001 ,
mnras , 327 , 868 boyle b. j. et al . , 2000 , mnras , 319 , 1014 byalko a. v. , 1970 , soviet astronomy , 13 , 784 cassan a. et al .
, 2006 , a&a , 460 , 277 calchi novati s. et al . , 2005 ,
a&a , 443 , 911 chang k. , refsdal s. , 1979 , nature , 282 , 561 coles j. , saha p. , schmid h. m. , 2009 , arxiv:0912.0515 cox a. n. , 1999 , allen s astrophysical quantities , fourth edition , ( springer - verlag : new york ) , 489 croom s. m. et al . , 2009 , mnras , 399 , 1755 dominik m. , sahu , k.c . , 2000 , apj , 534 , 213 dai s. , xu r.x .
, esamdin a. , 2010 , mnras , in press ( arxiv:0912.1167 ) de jong j. t. a. et al . , 2006 ,
a&a , 446 , 855 , apj , 714 , 663 de vries w.h . , becker r.h . , white r.l . ,
loomis c. , 2005 , aj , 129 , 615 einstein a. , 1936 , sci , 84 , 506 fontanot f. et al . , 2007 , a&a , 461 , 39 fukui a. et al . , 2007 , apj , 670 , 423 gaudi b. s. et al . , 2008 , apj , 677 , 1268 grski k. m. et al . , 2005 , apj , 622 , 759 gould a. , 1992 , apj , 392 , 442 gould a. , 2000 , apj , 535 , 928 griest k. , 1991 , apj , 366 , 412 gwinn c. r. et al . , 1997 ,
apj , 485 , 87 hamadache c. et al .
, 2006 , a&a , 454 , 185 han c. , 2008 , apj , 681 , 806 han c. , gould a. , 1995 , apj , 447 , 53 hawkins m.r.s . , 1996 , mnras , 278 787 hennawi j. f. , dalal n. , bode p. , 2007
, apj , 654 , 93 hg e. , novikov i.d . , polnarev a.g . , 1995 , a&a , 294 , 287 huo z .- y .
et al . , 2010 ,
raa , 10 , 612 irwin m. j. et al . , 1989 , aj , 98 , 1989 ivezi . , tyson j. a. , allsman r. , andrew j. , angel r. , 2008 , arxiv:0805.2366 jordi k. , grebel e. k. , ammon k. , 2006 , a&a , 460 , 339 kaiser n. et al . , 2002 , spie , 4836 , 154 kayser r. , refsdal s. , stabell r. , 1986 , a&a , 166 , 36 kiraga m. , paczyski b. , 1994 , apj , 430 , l101 kozowski s. , kochanek c. s. , 2009 , apj , 701 , 508 kozowski s. et al . , 2010 , apj , 708 , 927 kroupa p. , 2002
, sci , 295 , 82 lee i. et al .
, 2008 , apjs , 175 , 116 li g. l. et al . , 2007 , mnras , 378 , 469 liebes s. , 1964 , phys .
, 133 , 835 mao s. , 2008 , arxiv:0811.0441 meusinger h. , 2010 , a&a , 512 , 1 paczyski b. , 1986 , apj , 304 , 1 paczyski b. , 1991 , apj , 371 , l63 paczyski b. , 1995 , acta astronomica , 45 , 345 paczyski b. , 1996 , ara&a , 34 , 419 palanque - delabrouille n. et al .
, 1998 , a&a , 332 , 1 paulin - henriksson s. et al .
, 2002 , apj , 576 , l121 perryman m.a.c .
et al . , 2001 , a&a , 369 , 339 popowski p. et al . , 2005 , apj , 631 , 879 refsdal s. , 1964 , mnras , 128 , 295 renn j. , sauer t. , stachel j. , 1997 , sci , 275 , 184 richards g. et al . , 2006 , aj , 131 , 2766 richards g. et al . , 2009 ,
apj , 180 , 67 sesar b. et al . , 2007 ,
aj , 134 , 2236 schlegel d. , finkbeiner d. p. , and davis m. , 1998 , apj , 500 , 525 ( sfd ) , mnras , in press ( arxiv:0912.3693 ) skowron j. et al . , 2008 , acta astron . , 57 , 281 smith m. c. , belokurov v. , evans n.w . , mao s. , an j.h .
, 2005 , mnras , 361 , 128 sumi t. et al . , 2006 , apj , 636 , 240 smith m. c. et al . , 2007 , mnras , 379 , 755 tisserand p. et al . , 2007 , a&a , 469 , 387 thomas c. l. et al . , 2005 , apj , 631 , 906 tyson j.a . , 2002 , spie , 4836 , 10 udalski a. et al . , 1992 , acta astron . , 42 , 253 udalski a. et al
. , 2000 , acta astron .
, 50 , 1 unwin s.c .
et al . , 2008 , pasp , 120 , 38 uglesich r. r. , 2004 , apj , 612 , 877 walker m.a . , 1995 , apj , 453 , 37 wambsganss j. , 2006 , in g. meylan , p. jetzer , p. north , p. schneider , c. s. kochanek , & j. wambsganss , eds , saas - fee advanced course 33 : gravitational lensing : strong , weak and micro wood a. , mao s. , 2005 , mnras , 362 , 945 wyrzykowski .
et al . , 2009 ,
mnras , 397 , 1228 wyrzykowski .
et al . , 2010 , mnras , in press ( arxiv:1004.5247 ) | this paper presents an investigation into the gravitational microlensing of quasars by stars and stellar remnants in the milky way .
we present predictions for the all - sky microlensing optical depth , time - scale distributions and event rates for future large - area sky surveys .
as expected , the total event rate increases rapidly with increasing magnitude limit , reflecting the fact that the number density of quasars is a steep function of magnitude .
surveys such as pan - starrs and lsst should be able to detect more than ten events per year , with typical event durations of around one month .
since microlensing of quasar sources suffers from fewer degeneracies than lensing of milky way sources , they could be used as a powerful tool for recovering the mass of the lensing object in a robust , often model - independent , manner . as a consequence , for a subset of these events it will be possible to directly ` weigh ' the star ( or stellar remnant ) that is causing the lensing signal , either through higher order microlensing effects and/or high - precision astrometric observations of the lens star ( using , for example , gaia or sim - lite ) .
this means that such events could play a crucial role in stellar astronomy .
given the current operational timelines for pan - starrs and lsst , by the end of the decade they could potentially detect up to 100 events .
although this is still too few events to place detailed constraints on galactic models , consistency checks can be carried out and such samples could lead to exciting and unexpected discoveries .
gravitational lensing : micro , galaxy : structure , galaxy : kinematics and dynamics |
flavour changing neutral current ( fcnc ) processes have attracted renewed attention since the recent cleo measurement of the fcnc decays of the type @xmath2 . in the standard model ( sm )
these processes are forbidden at the tree level and are strongly suppressed by the gim mechanism , in particular for up type quarks .
hence they offer a unique possibility to test the ckm sector of the sm and possibly open a window to physics beyond the sm . from the experimental side
there is some data on the exclusive decay @xmath3 and the inclusive decay @xmath4 with branching fractions @xcite @xmath5 which is at present compatible with the sm .
however , the experimental as well as the theoretical uncertainties are still sizable ; in particular an improvement of the theoretical prediction involves a next - to - leading order calculation which has recently become available @xcite .
various scenarios have been used to calculate the influence of physics beyond the sm on the decay @xmath2 and also model independent analyses have been suggested @xcite . however , based on the decays of @xmath6 mesons it will not be possible to analyse the helicity structure of the effective hamiltonian mediating the decay @xmath2 , since the information on the handedness of the quarks is lost in the hadronization process . the only way to access
the helicity of the quarks is to consider the decay of baryons . from the experimental side
the decay @xmath0 is a good candidate , since the subsequent @xmath7 decay @xmath8 is self analyzing .
the only experimental drawback of this mode is that the production rate of @xmath9 baryons in @xmath10 quark hadrionization is about an order of magnitude smaller than the rate for @xmath6 mesons , and hence the analysis suggested here has to wait for more data on heavy quark decays from colliders . from the theoretical point of view
one may make use of the heavy quark symmetry for the @xmath10 quark .
the @xmath11 transition is of the heavy - to - light type for which heavy quark symmetries restrict the number of form factors for the baryonic transition @xmath12 light spin-1/2 baryon to only two .
these heavy quark symmetries are expected to work best at the point of maximal momentum transfer @xmath13 ( corresponding to small recoil to the light degrees of freedom ) , whereas in the decay under consideration we are at the opposite side of phase space , namely at @xmath14 .
we shall still use heavy quark symmetry , at least it will be a reasonable model assumption for the form factors .
once we use heavy quark symmetries over the whole phase space , one may obtain the form factors from the semileptonic decay @xmath1 , for which some data exists @xcite .
unfortunately , data are still not good enough to allow a measurement of the two form factors as a function of @xmath15 , and consequently one has to model the form factor shape using some ansatz .
this makes the predicted total rate to some extent uncertain , while the polarization of the @xmath7 is practically unaffected by the choice of a model . another theoretical difficulty with the process suggested are possible long distance contributions which will dilute the effects of a non - sm - contribution to the short distance effective hamiltonian for @xmath16 .
there are two effects of this kind : a vector dominance contribution from the process @xmath17 and the internal @xmath18 exchange .
both contributions are hard to estimate , since one has to refer to models . in the next section
we shall discuss the short distance contribution and its hadronic matrix elements . in section 3 we shall use simple models to estimate the long distance contributions . in section 4
we shall use the input of the semileptonic @xmath19 decay and two simple models for the form factors to obtain a prediction for the total rate of @xmath0 .
section 5 is devoted to a discussion of the polarization variable and its connection to the parameters of the short distance part of the effective hamiltonian .
the standard model and the sensitivity to non - standard model effects are considered .
finally we discuss our results and conclude .
in the sm as well as in many non - sm scenarios , the hamiltonian relevant for @xmath11 transitions consists of ten operators @xmath20 , of which only one ( @xmath21 ) mediates the decay @xmath2 .
the effective hamiltonian has the form @xmath22 with @xmath23 f^{\mu \nu}\ ] ] where the renormalization scale @xmath24 is usually taken to be @xmath25 .
it is well known that the leading - log calculation exhibits a strong residual dependence on the renormalization point @xmath24 , since the qcd corrections encoded in @xmath26 are sizable .
recently the complete next - to - leading order calculation has been performed and we shall use the value for @xmath26 from this calculation . a more general form of the operator @xmath21 allowing for non sm couplings is _ 7 = m_b |s _ ( g_v - g_a_5 )
b f^ where @xmath27 and @xmath28 are the vector and axial vector couplings , respectively . in the standard model g_v=1+m_sm_b ,
g_a=-1+m_sm_b the correction to the integer values of @xmath28 and @xmath27 due to the non - vanishing @xmath29 quark mass is of the order of 3% .
hence in general at the level of dimension - six operators there are two operators relevant for the decay under discussion , namely the one with right - handed @xmath10 quark ( implying left - handed @xmath29 quark ) and the one with right - handed @xmath29 quark ( implying left - handed @xmath10 quark ) .
in other words , in general there are two parameters , which may be taken to be @xmath30 and @xmath31 .
once the short distance structure of the effective hamiltonian is known , it remains to calculate matrix elements of the operators . in the present case
we shall make use of the fact that the @xmath10 quark is heavy , while the @xmath29 quark is taken to be a light quark .
in general , heavy quark symmetries restrict the number of possible form factors quite significantly ; in the present case there are only two , parametrized by @xcite [ heavytolightdecay ] = |u_(p , s ) \ { f_1(p.v)+v f_2(p.v ) } u__b ( v , s ) where @xmath32 is the velocity of the heavy @xmath9baryon which in hqet is equal to the velocity of the heavy @xmath10quark , @xmath33 is the spin of the heavy @xmath9 , which due to heavy quark spin symmetry equals the spin of the @xmath10 quark , @xmath34 is the momentum of the @xmath7baryon and @xmath29 its spin .
furthermore , @xmath35 is an arbitrary dirac matrix , such that any transition between a @xmath36 ( @xmath37 ) and a light spin @xmath38 baryon is given by the same form factors . in particular , this allows us to compare the exclusive semileptonic decay of a @xmath19 with the process under consideration here , although the helicity structure of the relevant currents is completely different .
heavy quark symmetries are expected to work best at the point where the final state light hadron is slow ( in the rest frame of the decaying hadron ) , such that the momentum transfer to the light degrees of freedom is small . at the other end of phase space , @xmath14
, it is easy to see that the the momentum transfer to the light degrees of freedom scales with the mass of the heavy quark , it is not clear whether heavy quark symmetries are applicable at this point .
however , we shall still make use of heavy quark symmetries even for @xmath39 , taking this in the worst case as a model assumption .
keeping this in mind , the matrix element of the operator @xmath40 between the initial and final state is & & ( p , s ) , ( k,)| o_7 | _ b(v , s ) + & & = e16 ^ 2 m_b ( p , s ) ||s _ 12 ( g_v - g_a5 ) b | _ b(v , s ) ( k,)|f^|0 the quantity of interest is the decay rate of unpolarized @xmath9 baryons into @xmath7 baryons with a definite spin directions @xmath29 . to obtain from this expression this rate as a function of the form factors taken at @xmath14 is a matter of algebra , and we obtain & = & c_7 ^ 2 g_f^2 ( v_ts^*v_tb)^2 ( e16 ^ 2)^2 m_b^2 ^3 ( 1- x^2 ) + & & \ { g_v^2 + g_a^2 2 [ ( 1 - 2x^2+x^4 ) |f_1|^2 + & & + ( x-2x^3+x^5)(f_1 f_2^ * + f_1^ * f_2 ) + & & + ( x^2 - 2x^4+x^6)|f_2|^2 ] + & & + g_v g_a ( v.s ) [ ( 2x-2x^3 ) |f_1|^2 + & & + ( 2x^2 - 2x^4)(f_1 f_2^ * + f_1^ * f_2 ) + & & + ( 2x^3 - 2x^5)|f_2|^2 ] } [ gammashortdistance ] where @xmath41 .
as expected , the terms which depend on the spin of the @xmath7 are all proportional to @xmath42 , while the rest of the rate has the factor @xmath43 .
thus a polarization analysis of the final state @xmath7 allows a determination of the ratio @xmath44 . in order to comply with the standard definitions we rewrite the rate in terms of the polarization variables as defined in @xcite = _ 0 where @xmath45 is the momentum vector of the @xmath7 and @xmath46 is its spin vector . using expression ( [ gammashortdistance ] )
one finds for @xmath47 &
= & ( 1-x^2)(1+x^2)(2x-6x^3)+2(2x^2 - 6x^4)r+ 2x^3 r^2 ( 1 - 2x^2+x^4)+2(x-2x^3)r_(x^2 - 2x^4)r^2 2g_v g_a g_v^2 + g_a^2 + & = & 0.378 [ alphadef ] where we have only used the central value of the cleo measurement @xcite r = f_2 f_1= -0.25 0.14 0.08 and the ratio of the baryon masses @xmath48 . note that @xmath49 vanishes in the limit @xmath50 ; the leading order is given by
= 2x 2g_v g_a g_v^2 + g_a^2 which for @xmath51 already is a reasonable approximation to the value obtained from ( [ alphadef ] ) .
in order to obtain a realistic estimate of the branching ratio and of the sensitivity on the coupling constants @xmath27 and @xmath28 one also has to take into account the long distance contributions .
one may distinguish between two types of contributions : the vector - meson - dominance like contributions , where there is a vector meson intermediate state and the internal @xmath18-boson exchange , which corresponds to the weak transition of the heavy @xmath9 into excited @xmath7 states decaying subsequently into @xmath7 and a photon .
the vector - dominance like contributions are dominated by the ckm allowed contributions where one has a @xmath52 as the internal vector meson .
they are governed by a different part of the @xmath53 hamiltonian given by h_eff^c|c = g_f v_cs^*v_cb(c_1 o_1 + c_2 o_2 ) [ heffshortandcc ] with _ 1 & = & ( |s__(1 - 5 ) b_)(|c _ ^(1 - 5)c _ ) , + o_2&=&(|s__(1 - 5 ) b_)(|c _ ^(1 - 5)c _ ) , where @xmath54 and @xmath55 are @xmath56 indices and @xmath57 and @xmath58 are the corresponding wilson coefficients .
the decays @xmath59 ( @xmath60 being the excited states with the quantum numbers of the @xmath52 ) are exclusive non - leptonic decays and the result for this part will be strongly dependent on model assumptions . following the usual folklore we keep from @xmath61 only the part in which @xmath62 and @xmath63 are in a color singlet state and perform factorization with this part
. thus we model @xmath64 by ^(n ) | h_eff^c|c | _ b = g_f v_cs^*v_cb(c_1 + ) | |s _ ( 1 - 5 ) b | _
b ^(n ) | ( |c ^(1 - 5)c | 0 it is well known that the combination of wilson coefficients @xmath65 ( @xmath66 ) comes out very small due to an accidental cancellation between the two contributions , yielding a too small rate for e.g. the decay @xmath67 . hence it is more realistic to use for this combination the value implied by the data and to put @xmath68 @xcite . furthermore , we consider a process in which we are dealing with a virtual @xmath60 which is quite far off shell .
it has been suggested @xcite to take into account this effect by assuming a @xmath15 dependent decay constant for the @xmath60 according to ^(n ) ( q , ) | |c ^(1 - 5)c | 0 = i
g_^(n ) ( q^2 ) ^*_(q ) from the comparison between the leptonic width of the @xmath52 ( yielding @xmath69 ) with the data from @xmath70-photo - production ( yielding @xmath71 ) one obtains the ratio _ = = 0.12 0.04 corresponding to a suppression by roughly a factor of ten . unfortunately
, there is no data for excited @xmath70 s , so we shall simply assume the same suppression factor @xmath72 for all of them . assembling all the pieces , summing over all @xmath70 resonances with the appropriate quantum numbers and employing the gordon identity to isolate the transverse part one finds |
h_eff^c|c | _ b = - 13 a_2 _
n g_^(n)^2(m_^(n)^2 ) m_^(n)^2 m_b | |s _ ( 1 + 5 ) b | _
[ heffshortandlong ] which takes the same form as the short distance contribution .
the second kind of long distance contribution is the one originating from internal @xmath18 exchange .
the relevant feynman diagrams are depicted in fig.[internalwex ] .
this contribution is even more difficult to calculate and we shall perform a relatively crude estimate .
the idea we are going to employ is along the lines of brodsky and lepage @xcite . in this picture the decay amplitude is decomposed into a hard piece which is calculated in terms of perturbative qcd while the soft contribution is described in terms of wave functions .
since in this picture a baryon is still a relatively complicated object , we shall simplify further by gathering the up and the down quark in the @xmath9 and the down and strange quark of the @xmath7 into diquarks .
this picture is in fact to some extent motivated from the phenomenological side , since it has been advertised as an explanation of the @xmath73 rule in hyperon decays @xcite .
hence we shall describe the baryons effectively as a bound state of a diquark and a @xmath10 quark or an @xmath74 quark , respectively .
the hard process in the sense of brodsky and lepage is a @xmath75 transition in which the weak boson is absorbed by the @xmath76 diquark , transforming it into a @xmath77 diquark . in order to conserve energy and momentum a photon
is radiated off , making this in total a @xmath0 transition .
= 2.12 cm = 2.12 cm = 2.12 cm = 2.12 cm to perform the calculation one has to define feynman rules for the radiation and the weak transition of the diquarks , which are simply obtained from the corresponding interactions of scalar particles in a straightforward way .
note that in order to ensure gauge invariance one also has to include a `` seagull '' term corresponding to a diquark-@xmath18-@xmath78 coupling .
the feynman rules are given in fig.[frules ] and these allow us a calculation of the hard part of the decay amplitude .
the soft contributions are parametrized in terms of wave functions of the two constituents of the baryons . in the spirit of @xcite
we neglect the transverse components of the momenta of the constituents .
the @xmath9 is moving with the velocity @xmath32 and hence we write for the for the momentum @xmath34 of the @xmath10 quark @xmath79 and @xmath80 for the momentum of the @xmath76 diquark .
similarly , we have @xmath81 for the momentum of the @xmath74 quark in the @xmath7 and @xmath82 for the momentum of its @xmath83 diquark .
the wave functions describing the soft part are @xmath84 and @xmath85 for the @xmath9 and the @xmath7 , which are functions of the momentum fractions @xmath86 and @xmath87 . for the @xmath9
we assume that the @xmath10 quark carries almost all momentum , which means that @xmath86 fiffers from one only by terms of the order @xmath88 , such that @xmath89 is independent of @xmath25 in the heavy mass limit .
hence we use @xmath90 for the wave functon of the @xmath9 . for the @xmath7
all the quarks have comparable masses and we use a simple ansatz such that the average of @xmath87 is @xmath91 .
motivated by the corresponding ansatz for the pion wave function we use @xmath92 where we have assumed that the analogue of the decay constant is simply @xmath93 . using this picture for the internal @xmath18 exchange it turns out that its contribution is much smaller than the long distance contribution of the vector dominance type .
this is not a surprise , since the internal @xmath18 exchange is on one hand suppressed by the small combination @xmath94 of ckm elements , on the other hand it contains two decay constants @xmath93 and @xmath95 , yielding another suppression by a factor @xmath96 in total , in the way we have estimated the long distance contributions they turn out to be small , such that the decay @xmath0 is dominated by the short distance piece , which may be sensitive to new physics .
quantitatively , the long distance corrections are dominated by the vector - dominance contribution . summing over the lowest six @xmath60 we find that the long distance contributions may be absorbed into a redefinition of the coupling constants @xmath28 and @xmath27 according to = g_v + = g_a - , = 0.035 [ rho ]
since we are dealing with an exclusive decay , we eventually need some input for the form factors in order to calculate numbers . however , not much is known about these functions , except that they have to obey the heavy quark symmetry implied by the heavy - to - light transition we are considering . in particular , at least at maximum momentum transfer the form factors appearing in @xmath97 transition and @xmath98 decays have to be the same , independent of the spin structure of the currents involved , see ( [ heavytolightdecay ] ) .
again we shall assume heavy quark symmetry over all phase space including @xmath14 in @xmath97 and extract the relevant form factors from @xmath1 , which is according to ( [ heavytolightdecay ] ) given by the same two form factors as @xmath0 .
for the semileptonic @xmath19 decay there is some experimental data @xcite ; the total rate and the form factor ratio @xmath99 has been measured , where the brackets indicate a particular average over phase space .
nothing is known about the @xmath15 dependence of the form factors and at that point model assumptions enter the game .
we shall start from a simple pole model for both form factors @xmath100 where @xmath101 is the mass of the nearest resonance with the correct quantum numbers for @xmath102 and @xmath103 rspectively and @xmath104 is a normalization factor .
the form factors in ( [ heavytolightdecay ] ) are given in terms of the energy @xmath105 of the outgoing light baryon which has a simple relation with the momentum transfer @xmath106 .
inserting this into the denominator of our form factor parametrization we have @xmath107 where we have neglected the mass of the light baryon .
the relevant resonances contain a heavy quark and some light degrees of freedom and hence the mass of the resonance is @xmath108 , where we shall in the following identify @xmath109 .
the parameter @xmath7 is small compared to the heavy quark mass and does not scale with the heavy mass ; it remains a constant as @xmath110 .
using this the denominator of the pole fomula for the form factors becomes @xmath111 for the @xmath98 transition the relevant resonance for the vector current is the @xmath112 and since in the heavy mass limit @xmath113 , we have @xmath114 mev , while for the bottom decay @xmath97 we have the @xmath115 in the vector channel , and hence here we obtain @xmath116 mev .
thus we have a fairly consistent picture and shall use @xmath117 mev , it turns out that the dependence of our results on @xmath7 is weak . putting the pieces together we use as a parametrization of the form factors in the heavy mass limit @xmath118 whre we assume for simplicity the same @xmath119 dependence for both form factors . in this simple case
the ratio of the form factors @xmath120 is independent of @xmath119 and we may use the value measured by cleo without worrying about the phase space average going into the measurement .
we shall use @xcite @xmath121 the large uncertainty in this value does not imply a large uncertainty in our results since any appearance of the form factor @xmath103 is suppressed by a factor @xmath122 . using the measured branching ratio for @xmath123 and the lifetime of @xmath19 from @xcite one obtains the normalization @xmath124 @xmath125 ) for the extrapolation to @xmath14 in the decay @xmath0 we obtain for the form factors @xmath126
this may be inserted into ( [ gammashortdistance ] ) with the corrections from ( [ rho ] ) to obtain the total rate for the decay @xmath0 in the standard model . in the heavy mass limit
the mass of the @xmath9 is equal to the @xmath10 quark mass and hence the total rate depends on @xmath127 , where some of the mass dependence comes from phase space ( and is actually @xmath128 ) and some is due to the @xmath25dependence of the effective hamiltonian .
to get rid of this @xmath127dependence we take as a reference the inclusive semileptonic decays of the @xmath6 meson and write : = br(_b ) br(b x_c l ) [ brlbtolg ] the inclusive semileptonic decay rate of @xmath6 mesons depends also on @xmath127 and is given by ( b x_c l)=m_b^5 g_f^2 v_cb^2 192 ^3 _ qcd r_c , [ btoxlnu ] where @xmath129 is a qcd correction and @xmath130 is a phase space factor . using the lifetimes and the inclusive semileptonic branching fraction @xmath131 from @xcite we obtain br(_b ) = ( 1- 4.5 ) 10 ^ -5 , where the lower and upper value corresponds to a dipole and monopole @xmath15 evolution of the form factors , respectively . for values of @xmath28 and @xmath27 different from the standard model ones one has to multiply the above equation by the factor @xmath132 , neglecting the strange quark mass and the long distance contributions .
in this section we shall discuss the possibility of using the decay @xmath0 as a test of the standard model and to what extent it could be a possible window to new physics . as opposed to its mesonic counterpart , the decay @xmath3
, the baryonic rare decay @xmath0 allows access to the helicity structure of the short distance piece of the effective hamiltonian , since the helicity of the final state @xmath7 can be measured .
the two relevant dimension-5 operators entering the short distance part are parametrized in terms of two coupling constants @xmath31 and @xmath30 and the polarization of the final state @xmath7 is sensitive to these couplings .
the first piece of information is the total rate of @xmath0 which contains ( as does the mesonic decay @xmath3 ) information on the combination @xmath133 , while the polarization variable @xmath54 defined in ( [ alphadef ] ) contains information on @xmath134 .
as we have discussed above , our rough estimates of the long distance contributions indicate that they are very small and will not significantly influence the sensitivity of the @xmath47 measurement on the parameters of the short distance part of the effective hamiltonian . in the following
we shall use the measurement of the inclusive rate @xmath4 to fix @xmath133 ; in the @xmath30@xmath31plane a measurement of any total rate ( meaning any of the processes @xmath4 , @xmath3 or @xmath0 ) will correspond to a circle , since all these decay rates are proportional to @xmath133 .
if in future experiments a measurement of @xmath47 in @xmath0 is performed one may use the relation between @xmath47 and @xmath134 . including our estimate of the long distance contributions , one obtains @xmath135 which corresponds to straight lines in the @xmath31@xmath30 plane . neglecting the long distance contribution ( i.e. @xmath136 ) these lines would go through the origin ; including our estimate for these long distance contributions we also need an input for the wilson coefficient @xmath26 , for which we shall use the standard model value as obtained form the nllo calculation in @xcite . in fig.[gagvfig ]
we plot the @xmath31@xmath30 plane .
the central circle corresponds to the central value of the measurement of @xmath4 and the thin circles indicate the experimental uncertainty . assuming the standard model values for @xmath31 and @xmath30 we have also plotted the corresponding lines .
the width of these lines is given by our estimate for the long distance contributions ; we have taken this estimate as an additional uncertainty meaning that the width of these lines is @xmath137 .
a measurement of @xmath47 yields in general two lines which have in total four intersections with the circle . the two intersections of each line correspond to a sign interchange @xmath138 and at the same time @xmath139 which is an unobservable phase .
the remaining ambiguity corresponds to the interchange @xmath140 and @xmath141 , since the polarization variable and the total rate are symmetric functions of @xmath28 and @xmath27 .
graphically this means that the two lines for a measured value of @xmath49 are mirror images of each other with respect to the lines @xmath142 . in order to resolve this ambiguity additional measurements
would be necessary . in the sm @xmath143 ( @xmath144 ) up to corrections from the not - vanishing @xmath29 quark mass ,
therefore the two solid lines almost coincide for the sm value of @xmath47 .
we also have plotted two dashed lines for a hypothetical measurement of @xmath145 .
with the advent of the second generation @xmath10 physics experiments , bottom baryons will provide a wealth of additional information on @xmath10 quarks .
in particular , specific aspects of fcnc decays can be tested which are not accessible in @xmath6 mesons decays . as its mesonic counterpart , @xmath0
is dominated by the short distance part of the effective hamiltonian , although there are additional long distance effects such as internal @xmath18 exchange .
we have estimated the long distance effects using standard methods and found a negligibly small contribution .
thus one expects a good sensitivity to test the short distance part , the relevant piece of which is given in terms of two operators , which differ by the handedness of the @xmath10 quark .
it is a particular property of the decays of the @xmath9 that they allow one to test the handedness of the effective interaction ; this is impossible in the corresponding meson decays .
heavy quark symmetries imply relations between the form factors which are thought to hold best close to the point where the outgoing hadron is almost at rest . for the transition @xmath146 light spin-1/2 baryon
there are only two independent form factors due to heavy quark symmetry .
measuring these form factors in @xmath1 then allows us to predict other processes .
unfortunately , in the process @xmath0 we are considering the end of the @xmath15 spectrum , where the @xmath7 receives a large recoil of the order of the @xmath10 quark mass , and hence heavy quark symmetries might fail .
we still used the relations implied by them , which in the worst case is simply a model assumption . using the input for the form factors obtained from @xmath1 we made predictions for the decay @xmath0 within the standard model and beyond .
it turns out that a measurement of the polarization of the final state @xmath7 is quite sensitive to the handedness of the underlying effective short distance hamiltonian , up to a twofold ambiguity , which can not be resolved by a measurement of the rate and the polarization alone . with a branching ratio for @xmath0 of the order @xmath147 one needs @xmath148 @xmath10 quarks to have about one hundred events , without applying cuts for efficiencies .
clearly this will be feasible at dedicated @xmath10 physics experiments at colliders such as the one at tevatron or lhc , and possibly also at fixed target experiments like hera - b .
t.m . thanks a. ali and a. schenk for useful conversations .
this work was supported by deutsche forschungsgemeinschaft . | we investigate the rare decay @xmath0 which receives both short and long distance contributions .
we estimate the long distance contributions and find them very small .
the form factors are obtained from @xmath1 using heavy quark symmetry and a pole model .
the short distance piece opens a window to new physics and we discuss the sensitivity of @xmath0 to such effects .
@sh[#1]#2 fmsl@sh fmsl@sh fmsl@sh fmsl@sh#1#2#3@th 5_5 ps .
* .s * _ hep - ph/9701399 + ttp9706 + january 28 , 1997 * flavour changing neutral current decays of heavy baryons : the case @xmath0 * thomas mannel and stefan recksiegel + _ institut fr theoretische teilchenphysik , d
76128 karlsruhe , germany . |
the discovery of quantum error - correcting codes and the theory of fault - tolerant quantum computation have greatly improved the long - term prospects for quantum computing technology @xcite . to implement fault - tolerant quantum computation for a given quantum error - correcting code , protocols for performing fault - tolerant operations are needed .
the basic design principle of a fault - tolerant operation protocol is that if only one component in the procedure fails , then the failure causes at most one error in each encoded block of qubits output from the procedure .
the most straightforward protocol is to use transversal gates whenever possible .
a transversal operation has the virtue that an error occurring on the @xmath5th qubit in a block can only ever propagate to the @xmath5th qubit of other blocks of the code , no matter what other sequence of gates we perform before a complete error - correction procedure @xcite .
unfortunately , it is widely believed in the quantum information science community that there does not exist a quantum error correcting code , upon which we can perform universal quantum computations using just transversal gates @xcite , and recently this belief is proved @xcite .
we therefore have to resort to other techniques , for instance quantum teleportation @xcite or state distillation @xcite .
the @xmath0 hierarchy is introduced by gottesman and chuang to implement fault - tolerant quantum computation via teleportation @xcite .
the starting point is , if we can perform the pauli operations and measurements fault - tolerantly , we can then perform all clifford group operations fault - tolerantly by teleportation .
we can then use a similar technique to boot - strap the way to universal fault - tolerant computation , using teleportation , which gives a @xmath0 hierarchy of quantum teleportation , as defined below : the sets @xmath0 are defined in a recursive way as sets of unitary operations @xmath6 that satisfy : @xmath7 where @xmath8 is the pauli group .
we call a unitary operation an @xmath1-qubit @xmath0 gate if it belongs to the set @xmath0 and acts nontrivially on at most @xmath1 qubits
. note by definition @xmath9 is the clifford group , which takes the pauli group into itself . and
@xmath10 , but @xmath0 for @xmath11 is no longer a group . denotes an * epr * pair
, @xmath12 represents bell - basis measurement , @xmath13 , where @xmath14 is a pauli operator .
the double wires carry classical bits and single wire carries qubits .
any gate in the @xmath0 hierarchy can be implemented fault - tolerantly using this teleporation scheme.,width=288 ] -teleportation , @xmath15 , @xmath16 , @xmath17 , and @xmath18 is a cnot gate with the first qubit as its target . for @xmath19-teleportation ,
@xmath20 , @xmath21 , @xmath22 , and @xmath18 is a cnot gate with the first qubit as its control .
all semi - clifford @xmath0 gates can be implemented fault - tolerantly using this scheme.,width=288 ] all the gates in @xmath0 can be performed with the two - bit teleportation scheme ( fig .
[ 2bit ] ) in a fault - tolerant manner . because , as proved in @xcite , it is possible to fault - tolerantly prepare the ancilla state @xmath23 , apply the classically controlled correction operation @xmath24 , and measure in bell basis on a stabilizer code .
however the precise set of gates which form @xmath0 is unknown , even for a fixed number of qubits .
it is demonstrated in @xcite that a subset of @xmath0 gates could be implemented by a different architecture than the standard teleportation , called one - bit teleportation , as shown in fig .
[ 1bit ] . those gates
adopt the form @xmath25 , where @xmath26 is a diagonal gate in @xmath0 and @xmath27 are two clifford operations .
gates of this form are recently studied in literature and are called the semi - clifford operations @xcite . in the following we will denote the @xmath1-qubit pauli group as @xmath28 and a semi - clifford operation is defined to be a gate which sends at least one maximal abelian subgroup of @xmath28 to another maximal abelian one under conjugation . due to the fact that one - bit teleportation needs only half the number of ancilla qubits per teleportation than the standard two - bit teleportation , it is important to understand the difference of capabilities between one and two - bit teleportation for the practical implementations of fault - tolerant architecture .
it is conjectured in @xcite that those two capabilities coincide for @xmath29 , which means that all the @xmath30 gates for two qubits are semi - clifford operations . in this paper
, we prove this conjecture for a more general situation where @xmath31 , and @xmath3 .
we then disprove it for parameters @xmath4 by explicit construction of counterexamples .
we leave open the question for the parameters @xmath32 , and a more general problem of fully characterizing the structure of @xmath0 : we conjecture that all gates in @xmath0 are something we refer to as generalized semi - clifford operations , i.e. a natural generalization of the concept of semi - clifford operation to the case including classical permutations .
our results about this semi - clifford operations versus @xmath0 gates relation can be visualized in fig .
[ cksemi ] . gates . a : all gates ; b : generalized semi - clifford gates ; c : semi - clifford gates ; d : @xmath0 gates ; e : @xmath30 gates .
c is strictly contained in b and e is strictly contained in d. the two question marks indicate two open problems we have : whether @xmath33 is a subset of @xmath12 ; and whether @xmath18 is a subset of @xmath34.,width=288 ] just as in the usual circuit model , different gates are implemented with different levels of complexity using this teleportation scheme .
it is then natural to ask the questions , how to characterize this concept of gate complexity with concrete physical quantities , how does this measure based on teleportation schemes compare with the usual circuit depth , and what it implies for the practical construction of quantum computation architecture . to answer these questions ,
we introduce a quantity as a measure of gate complexity for fault - tolerant quantum computation based on the @xmath0 hierarchy , called the teleportation depth , which characterizes how many teleportation steps are necessary , on average , to implement a given gate .
we demonstrate the effect of the existence of non semi - clifford operations in @xmath0 on the estimation of the upper bound for the teleportation depth , as well as some quantitative difference between the capabilities of one and two - bit teleportation .
the paper is organized as follows : section ii gives definition and basic properties of semi - clifford operations and generalized semi - clifford operations ; in section iii we study the structure of @xmath0 hierarchy in terms of semi - clifford and generalized semi - clifford operations ; section iv is devoted to the discussion of teleportation depth and how it depends on the structure of @xmath0 ; and with section v , we conclude our paper .
the concept of semi - clifford operations was first introduced in @xcite , to characterize the property of gates transforming pauli matrices acting on a single qubit . here
we generalize it to the @xmath1-qubit case , through the following definition .
an @xmath1-qubit unitary operation is called semi - clifford if it sends by conjugation at least one maximal abelian subgroup of @xmath28 to another maximal abelian subgroup of @xmath28 . that is ,
if @xmath6 is an @xmath1-qubit semi - clifford operation , then there must exist at least one maximal abelian subgroup @xmath35 of @xmath28 , such that @xmath36 is another maximal abelian subgroup of @xmath28 .
the most basic property of a semi - clifford operation is , if @xmath37 is a semi - clifford operation , then there exist clifford operations @xmath27 such that @xmath38 is diagonal . * proof * : @xmath39 represents the pauli @xmath40 operation on the @xmath41 qubit .
if @xmath37 is an @xmath1-qubit semi - clifford operation , then there must exist @xmath1-qubit operations @xmath42 such that @xmath43 , @xcite , i.e. @xmath44 holds for any @xmath45 .
therefore , @xmath46 , i.e. the @xmath1-qubit gate @xmath38 is diagonal.@xmath47 in other words , semi - clifford operations are those gates diagonalizable ` up to clifford multiplications ' .
thus the structure problem of the whole set of semi - clifford operations is reduced to that of the diagonal subset within it . as we shall see later ,
the notion of semi - clifford operations is useful in characterizing some but not all gates in the @xmath0 hierarchy .
more generally , we might also consider those gates with properties of transforming the span , or in other words the group algebra over the complex field , of a maximal abelian subgroup of @xmath28 .
a generalized semi - clifford operation on @xmath1 qubits is defined to send by conjugation the span of at least one maximal abelian subgroup of @xmath28 to the span of another maximal abelian subgroup of @xmath28
. denote @xmath48 the group generated by a set of operators @xmath49 , and denote the span of the group @xmath48 as @xmath50 . then in a more mathmatical form we can write the above definition as : if @xmath6 is a generalized semi - clifford operation on @xmath1 qubits , then there must exist at least one maximal abelian subgroup @xmath51 of @xmath28 , such that for all @xmath52 , @xmath53 , where @xmath36 is another maximal abelian subgroup of @xmath28 .
then the basic property of a generalized semi - clifford operation is , if @xmath37 is a generalized semi - clifford operation , then there exist clifford operations @xmath27 , and a classical permutation operator @xmath54 such that @xmath55 is diagonal . * proof *
: if @xmath37 is a generalized semi - clifford operation , then there must exist @xmath42 such that @xmath56 , i.e. @xmath57 .
that is , @xmath38 maps all the diagonal matrices to diagonal matrices , therefore @xmath38 must be a monomial matrices , i.e. there exist a permutation matrix @xmath54 and a diagonal matrix @xmath26 , such that @xmath58 is diagonal.@xmath47 note for the single qubit case , i.e. @xmath59 , the concepts of semi - clifford operation and generalized semi - clifford operation coincide .
in this section we study the structure of gates in @xmath0 . to begin with , we study some basic properties of @xmath0 gates . then we give our main results as structure theorems , which state that all the @xmath0 gates are semi - clifford when @xmath31 and @xmath3 , but for @xmath4 there are examples of @xmath0 gates which are non - semi - clifford .
we then discuss the open question for the parameters @xmath32 , and based on the constructed counterexamples we conjecture that all @xmath0 gates are generalized semi - clifford operations .
it should be noted that the set of @xmath1-qubit @xmath0 gates is always strictly contained in the set of @xmath1-qubit @xmath60 gates . in @xcite ,
explicit examples are given to support this statement .
if we denote as @xmath61 the @xmath1-qubit gate which applies @xmath6 to the @xmath1th qubit only if the first @xmath62 qubits are all in the state @xmath63 , then @xmath64 is in @xmath65 .
we first state an important property of gates in @xmath0 , which reduce the problem of characterizing the structure of @xmath0 into a problem of characterizing a certain subset of gates in @xmath0 . if @xmath66 , then @xmath67 , where @xmath68 , @xmath69 .
* proof * : we prove this proposition by induction .
\i ) it is obviously true for @xmath70 ; \ii ) assume it is true for @xmath5 ; \iii ) for @xmath71 , @xmath72 implies @xmath73 , where @xmath74 . if we conjugate @xmath75 by @xmath38 , we get @xmath76 since @xmath68 , @xmath77 are in @xmath9 also . and because @xmath78 , @xmath79 . according to assumption ii ) , @xmath80 .
finally as we can see from eqn ( 2 ) , @xmath81.@xmath47 according to proposition 3 , in order to characterize the full structure of @xmath0 , we only need to characterize the structure of a subset of it which generates the whole set with clifford multiplications .
it is known that @xmath0 is not a group for @xmath82 and its structure is in general hard to characterize . however , if we denote all the diagonal gates in @xmath0 as @xmath83 , then we have the following : @xmath84 is a group .
if we can characterize the group structure of @xmath83 , then the structure of the @xmath0 subset @xmath85 is known to us ( @xmath86 , @xmath87 ) . according to proposition 1 ,
this is just the set of all semi - clifford operations in @xmath0 . in the next section
, we will repeatedly use this fact to gain knowledge about semi - clifford @xmath0 gates from the group structure of @xmath83 and for now we will give a brief proof of the above proposition . * proof * : we prove by induction .
\i ) it is of course true for @xmath70 ; \ii ) assume it is true for @xmath5 , i.e. @xmath84 is group ; \iii ) then for @xmath71 , note for any @xmath88 , @xmath89 , for non - diagonal @xmath90 , where @xmath91 .
\a ) if @xmath88 , then @xmath92 , since @xmath93 , which is in @xmath84 by assumption ii ) .
\b ) if @xmath94 , then @xmath95 , since @xmath96.@xmath47 according to this proposition , all semi - clifford @xmath0 gates can be characterized by the group structure of diagonal @xmath0 gates .
our main results about the structure of @xmath0 are the following three theorems , which state that all the @xmath0 gates are semi - clifford when @xmath31 and @xmath3 , but it is no longer true for @xmath4 .
all gates in @xmath0 are semi - clifford operations for @xmath97 .
* proof * : here we prove the case of @xmath98 .
the proof of the @xmath59 case is similar but can also be checked by direct calculation and lead to a complete classification of all 1-qubit @xmath0 gates according to the group structure of diagonal 1-qubit @xmath0 gates .
we give details for the @xmath59 case in appendix . for @xmath98
, we prove this theorem by induction : \i ) it is obviously true for @xmath99 ; \ii ) assume it is true for @xmath5 ; \iii ) for @xmath71 : \a ) we calculate the set @xmath100 for all @xmath101 , where @xmath102 . note by assumption ii ) , @xmath103 gives us all the elements in @xmath0 up to clifford conjugation .
\b ) note in general @xmath104 for some angles @xmath105 and @xmath106 . by exhaustive calculation with all @xmath101
we show that if there exists an element @xmath107 such that @xmath108 is trace zero and hermitian , then @xmath109 for some @xmath110 and @xmath111 .
furthermore , we can again show by exhaustive calculation with all @xmath101 that the only trace zero and hermitian @xmath107 is of the following form up to clifford conjugation : @xmath112 \c ) we calculate the set @xmath113 for all @xmath101 , which by assumption ii ) and fact b ) gives all the elements in @xmath0 which are trace zero and hermitian .
\d ) we show that for any two - qubit gate @xmath6 such that @xmath114 and @xmath115 , there exist @xmath42 such that @xmath116 is diagonal .
this can be started from studying the eigenvectors of @xmath108 , which can be chosen of the form @xmath117 and carefully considering the possible phase of each eigenvector and the possible superposition of the eigenvectors due to the degeneracy of the eigenvalues , similar to the process shown in appendix a.@xmath47 all gates in @xmath0 are semi - clifford operations for @xmath3 . * proof * : we prove this theorem exhaustively using the following proposition : an @xmath1-qubit @xmath0 gate @xmath6 is semi - clifford if and only if the group @xmath118 contains a maximally abelian subgroup of @xmath28 .
* proof * : suppose @xmath119 , then @xmath120 . on the contrary ,
if @xmath118 contains a maximal abelian subgroup of @xmath28 , then there must exist @xmath121 such that @xmath122 , i.e. @xmath123 holds for any @xmath45 .
therefore , @xmath124 , @xmath125 is diagonal . if we denote this diagonal gate as @xmath26 , @xmath126 .
therefore , by exhaustive study with the subgroups of the three - qubit clifford group which are isomorphic to @xmath127 , we complete the proof of this theorem .
more detailed analysis about this is given in appendix b. the calculation is done using gap @xcite.@xmath47 not all gates in @xmath0 are semi - clifford operations for @xmath128 .
* proof * : actually we only need to prove this theorem for @xmath129 then it naturally holds for all the other parameters of @xmath4 .
however we would like to explicitly construct examples for all @xmath130 .
define @xmath131 as in fig .
[ fig4 ] .
gate @xmath132 .
@xmath133.,width=192 ] the gate @xmath134 is a @xmath60 operation but not a semi - clifford operation , where @xmath135 is toffoli gate with the 1st and 2nd qubit as its control and the 3rd qubit as its target , @xmath136 is single qubit operator @xmath137 on the 3rd qubit . * proof * : to prove that @xmath132 is in @xmath60 , \i ) when @xmath70 , @xmath138 .
@xmath139 is of the form @xmath140 , where @xmath141 is a clifford operation and @xmath37 is the toffoli gate . according to proposition 3 ,
@xmath139 and the toffoli gate are both in @xmath30 .
\ii ) for @xmath142 , direct calculation shows that @xmath143 , @xmath144 .
@xmath145 , @xmath146 , @xmath147 .
the images of @xmath148 s under the conjugation of @xmath132 can all be written in the form @xmath149 , where @xmath150 , @xmath151 , @xmath152 , @xmath153 are diagonal gates in @xmath0 and @xmath154 , @xmath155 are diagonal single qubit gates in @xmath156 acting on the third qubit .
the image of the whole 3-qubit pauli group @xmath157 is generated by the six elements shown above .
as multiplication by clifford gates preserves the @xmath0 hierarchy , we only need to check the images of pauli operations which are composed of two or more @xmath148 s and see if their images are still in @xmath0 .
this is obviously true considering the special form of @xmath158 .
multiplication of any two of them is of the form @xmath159 .
this is in @xmath0 as the diagonal @xmath0 gates form a group .
further more , multiplication of all of them takes the form @xmath160 . as @xmath155 is a single qubit operation on the third qubit , @xmath161
this is again a @xmath0 gate because of the group structure of diagonal @xmath0 gates .
therefore , we have checked explicitly that @xmath162 .
to prove that @xmath132 is not semi - clifford , we can exhaustively calcultate @xmath163 and find its intersection with @xmath127 .
the fact that @xmath164 does not contain a maximally abelian subgroup of @xmath127 implies that @xmath132 is not semi - clifford , due to proposition 5 .
with this example we have directly proved theorem 3 .
@xmath47 let us try to understand more about the structure theorems we have in the previous section .
first recall from @xcite that the controlled - hadamard gate @xmath165 , which is a @xmath30 gate , is explicitly shown to be semi - clifford .
we can also view this from the perspective of proposition 5 , by noting that @xmath166 , @xmath167 , which means that the maximal abelian subgroup of the pauli group generated by @xmath168 is in the image of @xmath165 .
however , if we consider @xmath169 from the perspective of proposition 5 , we get @xmath170 , @xmath171 , @xmath172 . note this do not give us a maximal abelian subgroup of the pauli group @xmath173 , due to the effect of @xmath174 caused by conjugating through the toffoli gate .
this intuitively explains why theorem 3 could be true , but no counterexample to theorem 2 exists .
note that @xmath132 is actually a generalized semi - clifford operation , which is apparent from its form .
also , the construction of the series of gates @xmath132 , as well as their extensions to @xmath175 qubits , can not give any non - semi - clifford @xmath30 gate .
we then have the following conjectures on the open problem of the structure of @xmath0 hierarchy in general .
all gates in @xmath30 are semi - clifford operations .
all gates in @xmath0 are generalized semi - clifford operations .
teleportation , as a computational primitive , is a crucial element providing universal quantum computation to fault - tolerant schemes based on stabilizer codes .
however , not all gates are of equal complexity in this scheme . to actually incorporate this technique in the construction of practical computational architecture ,
it is useful to know which gates are easier to implement and which are harder , so that we could achieve optimal efficiency in performing a computational task . in the circuit model of quantum computation
, we face the same problem and in that case ` circuit depth ' was introduced @xcite to characterize the number of simple one and two - qubit gates needed to implement an operation .
while this provides a good measure of gate complexity , it does not take into consideration of fault - tolerance .
it is interesting to have measures quantifying fault - tolerant gate complexity to be compared with ` circuit depth ' to give us a better understanding of the computational tasks at hand .
based on the @xmath0 hierarchy introduced in @xcite and the knowledge of its structure gained in previous section , we define a measure of gate complexity for the teleportation protocol , called the teleportation depth , which characterizes how many teleportation steps are necessary , on average , to implement a given gate .
since any teleportation unavoidably causes randomness , we need to figure out a certain point to start with , i.e. we should assume in advance that some kind of gates can be performed fault - tolerantly .
we know that a fault - tolerant protocol is usually associated with some quantum error - correcting codes .
self - dual css codes , such as the @xmath176-qubit steane code , admit all gates in the clifford group to be transversal @xcite . in such a situation
, we only need to teleport the gate outside the clifford group , and in the following , we will assume this as a starting point .
the advantage of doing this , in practice , is that due to proposition 1 , we have the freedom of preparing the ancilla states up to some clifford multiplications . with the standard two - bit teleportation scheme ( fig .
[ 2bit ] ) in mind , it is easy to see that all gates in the @xmath0 hierarchy can be teleported fault - tolerantly as a whole in a recursive manner .
suppose @xmath6 is an @xmath1-qubit @xmath0 gate .
the ancilla state can be fault - tolerantly prepared and all the elements in the teleportation circuit of @xmath6 are in @xmath9 and can be performed fault - tolerantly , except the classically controlled operation @xmath177 , where @xmath14 is an operator in @xmath8 which depends on the ( random ) bell - basis measurement outcomes @xmath178 .
however , as @xmath6 is in @xmath0 , @xmath179 is in general a @xmath156 operation and can be implemented again by teleportation . in this way , after each teleportation step , a @xmath0 gate is mapped to another gate one level lower .
this recursive procedure terminates when @xmath180 is in @xmath9 .
based on the above picture we give a more formal definition of teleportation , which characterizes its randomness nature .
the teleportation map @xmath181 takes an @xmath1-qubit operator @xmath75 to a set of operators via the following manner : @xmath182 where @xmath183 are elements of the @xmath1-qubit pauli group @xmath28 .
note @xmath184 and @xmath185 each element of image of the map @xmath186 on @xmath75 is associated with a set @xmath187 denote @xmath188 as the element in image of the map @xmath186 on @xmath75 associated with the set @xmath189 .
each element in the image occurs with equal probability .
@xmath188 terminates if @xmath190 .
if @xmath191 terminates , then @xmath192 terminates for any @xmath193 , and @xmath194 .
therefore , for each @xmath188 that terminates , there must exist a set @xmath195 with the minimal size such that @xmath196 terminates , where @xmath197 ( @xmath198 ) . in our following discussions
, we will only consider sets @xmath189 which are minimal in this sense .
this mapping procedure works directly on @xmath0 gates .
if @xmath199 is an @xmath1-qubit @xmath0 gate , then there is no need to decompose it into consecutive application of several other gates and we say we can ` direct teleport ' @xmath199 .
@xmath199 is in @xmath0 iff @xmath200 , @xmath201 , and @xmath202 , s.t . @xmath203 . among all @xmath0 gates ,
the set of semi - clifford operations have the special property that they can be teleported with only half the ancilla resources as in a standard teleportation scheme .
this ` one - bit teleportation scheme ' is illustrated in fig .
this scheme also complies with the mapping description given above . instead of bell basis measurement ,
randomness in one - bit teleportation scheme comes from single qubit measurement and @xmath204 belongs to a maximal abelian subgroup of the whole @xmath1-qubit pauli group in general . to teleport an arbitrary @xmath1-qubit gate @xmath75
, we can first decompose @xmath75 into the @xmath0 hierarchy , @xmath205 , where @xmath206 , because we only know how to teleport @xmath0 gates fault - tolerantly .
we call this procedure ` decomposition of @xmath75 into @xmath207 ' .
suppose that to teleport each gate @xmath208 , @xmath209 maps are needed on average , with average taken over all possible set @xmath210 .
then the teleportation depth of @xmath75 is defined as follows .
the teleportation depth of a gate @xmath75 , denoted as @xmath211 , is the minimal sum of all @xmath209the average number of teleportation steps needed to implement each component gate of @xmath75where the minimum is taken over all possible decompositions of @xmath75 into @xmath207.[depth ] due to definition [ depth ] , in order to calculate the teleportation depth of a given gate @xmath75 , one needs to find all possible decompositions of @xmath75 into @xmath207 gates and calculate the corresponding depth , then minimize over all of them .
this is generally intractable , but one may expect to upper bound the depth with some particular decomposition of @xmath75 into @xmath207 gates .
let us first consider the case of an @xmath1-qubit @xmath0 gate .
@xmath212 is the teleportation depth of an @xmath1-qubit @xmath0 gate .
as such a gate can be teleported directly , @xmath212 is upper bounded by the average number of steps needed in this direct teleportation scheme to terminate the teleportation procedure .
@xmath213 where the summation is over all possible ( minimal ) sets @xmath189 and @xmath214 is the number of such sets .
when @xmath215 , it is not obvious that the above summation will converge .
we will show that this is true .
then for an arbitratry gate @xmath75 , by decomposing @xmath75 into a finite series of @xmath0 gates , we can see that the teleportation depth of @xmath75 turns out to be finite
. then we do not actually require the procedure to terminate within a finite number of steps .
different teleportation schemes , for example one - bit and two - bit teleportation , give different upper bounds on teleportation depth for a certain circuit .
while for some circuits one scheme is obviously more efficient than others , the comparison among different schemes in other case may not be so straightforward and may depend sensitively on various parameters in the circuit . in the following sections , we study such dependence and present surprising results beyond our usual expectation with examples from important quantum circuits .
we first calculate explicitly an upper bound for the teleportation depth of semi - clifford @xmath1-qubit @xmath0 gates .
we know from @xcite that this kind of gate can be teleported directly with the architecture of one - bit teleporation and we denote the upper bound calculated with this ` * * one - bit * * ' ` * * direct * * ' teleportation procedure as @xmath216 . for a general @xmath1-qubit gate , if it is possible to decompose it into a series of semi - clifford @xmath0 operations , the upper bound of teleportation depth obtained by teleporting each part separately using one - bit teleportation scheme is in general denoted as @xmath217 .
@xmath217 is the average total number of teleportation steps needed to teleport separately each semi - clifford @xmath0 component of a quantum circuit using the one - bit teleportation scheme , if such a decomposition is possible .
more specifically , @xmath216 is the average number of teleportation steps needed to teleport an @xmath1-qubit semi - clifford @xmath0 gate directly ( i.e. without decomposition ) using the one - bit teleportation scheme .
apparently we have @xmath218 in general . the probability that the teleportation process terminates immediately after one teleportation step equals the percentage weight of a maximal abelian subgroup in the whole pauli group , which is @xmath219 for an @xmath1-qubit pauli group
now each teleportation step may have two possible endings : i ) with probability @xmath220 , @xmath221 and the process terminates ; ii ) with probability @xmath222 , @xmath223 is a general @xmath1-qubit @xmath156 gate and the process goes on .
the upper bound of teleportation depth calculated with this process is then @xmath224 it is clearly seen from eq .
( [ depthck ] ) that @xmath216 converges to @xmath225 when @xmath215 , which means that @xmath212 is in general bounded .
for instance , when @xmath98 , eq .
( [ depthck ] ) tells us @xmath226 .
the behavior of @xmath227 is shown in fig .
[ depthtwoqubits ] .
however , since @xmath228 , we find that teleporting two single - qubit semi - clifford @xmath0 gates together using the one - bit teleportation scheme needs fewer teleportation steps than to teleport each of them separately .
, width=288 ] since @xmath229 , @xmath216 quickly reaches @xmath225 as @xmath5 grows
. therefore , generally , the upper bound of the teleportation depth of a @xmath0 gate given by ` direct teleportation ' is not determined by @xmath5 , but by the number of qubits @xmath1 it actually acts on . moreover , since @xmath230 , i.e. the upper bound of teleportation depth increases exponentially with @xmath1 , in generally , when @xmath231 are large , it is better to decompose an @xmath1-qubit @xmath0 gate into some one and two - qubits gates to get a lower upper bound . however , if @xmath232 , where @xmath233 is a polynomial in @xmath1 , then @xmath216 scales as @xmath233 .
now we give two examples as applications of the above upper bounds , through which we obtain some idea about the order of teleportation depth in comparison with the usual circuit depth .
-qubit quantum fourier transform , width=326 ] the first example is the @xmath1-qubit quantum fourier transform ( qft ) circuit , as shown in fig .
@xmath234 denotes the unitary transformation @xmath235 .
the circuit depth of @xmath1-qubit qft goes as @xmath236 and we will soon find that the teleportation depth of this circuit is of the same order . each block of gates within a single dashed box ( hadamard plus controlled @xmath237-rotations on the @xmath238 qubit)is a semi - clifford @xmath239-qubit @xmath240 gate , @xmath241 and can be teleported directly using the one - bit scheme .
therefore the whole circuit can be teleported piece by piece by one - bit teleportation .
note that @xmath242 for large @xmath1 .
actually , numerical data shows that even when @xmath1 is small , @xmath243 is almost also true .
therefore , the teleportation depth of the @xmath1-qubit qft is upper - bounded by @xmath244 numerial calculation shows that @xmath245 is almost @xmath246 .
note the probability for the teleportation process to terminate is @xmath247 for teleporting an @xmath1-qubit @xmath248 gate @xmath249 times .
this means that the upper bound we got for this block teleportation scheme of qft is just slightly lower than naively assuming that we need @xmath250 teleportation steps to teleport a @xmath0 gate .
the reason we do not benefit from the avarage is that for qft , @xmath5 is generally comparable with @xmath1 .
-fold uniformly controlled rotation of a qubit about the axis @xmath251,width=288 ] now we consider another example , the uniformly controlled rotations , which are widely used in analyzing the circuit complexity of an arbitrary @xmath1-qubit quantum gate @xcite @xcite .
this circuit in general needs @xmath252 cnot gates and @xmath253 one - qubit elementary rotations to implement . for complexity analysis of this circuit
see for example @xcite .
the teleportation depth of this rotation is in general upper bounded by @xmath225 .
however , if each @xmath254-qubit - controlled gate is in @xmath0 , we might expect to do better .
for instance , when @xmath255 , for any positive constant @xmath1 , the teleportation depth scales as @xmath256 , i.e. linear in @xmath1 . moreover , if @xmath232 , where @xmath233 is a polynomial in @xmath1 , then the teleportation depth scales as @xmath233 .
now recall our series of examples of non - semi - clifford @xmath0 gates given in fig .
[ fig4 ] . we know
that if @xmath257 , then @xmath258 . and
the group @xmath259 does not contain a maximally abelian subgroup of @xmath127 , i.e. @xmath258 is not directly one - bit teleportable .
therefore , we know that there are some @xmath132 gates in the @xmath0 hierarchy which can only be teleported directly by the standard two - bit teleportation scheme . using this scheme , we can calculate another upper bound for teleportation depth , which we denote as @xmath260 .
@xmath261 is the average total number of teleportation steps needed to teleport separately each @xmath0 component of a quantum circuit using two - bit teleportation scheme , if such a decomposition is possible .
more specifically , @xmath260 is the average number of teleportation steps needed to teleport an @xmath1-qubit @xmath0 gate directly ( i.e. without decomposition ) using the two - bit teleportation scheme . for a general @xmath1-qubit @xmath0 gate ,
@xmath260 can be calculated by replacing @xmath262 with @xmath263 in eq .
( [ depthck ] ) @xmath264which then converges to @xmath265 when @xmath215 .
one may guess that in general to teleport @xmath132 directly using the two - bit scheme will give a lower bound for teleportation depth than to teleport the toffoli gate and @xmath266 separately using the one - bit scheme .
surprisingly , this is not generally true . when @xmath267 , this is indeed true .
teleporting @xmath132 directly gives a bound of @xmath268 , which is less than @xmath269 , i.e. the bound given by teleporting the toffoli gate and @xmath270 separately with the one - bit scheme .
however , when @xmath215 , teleporting @xmath132 directly gives a bound of @xmath271 , which is greater than @xmath272 , i.e. the bound given by teleporting the toffoli gate and @xmath270 separately .
this means that there exists a critical value @xmath5 that determines which way is more efficient for teleporting @xmath132 , directly or separately .
note if @xmath257 , we also have @xmath273 . calculating the bounds of teleportation depth for @xmath274
shows a similar behavior as that of @xmath132 , however of a slightly different value .
for instance , when @xmath267 , teleporting @xmath274 directly gives a bound of @xmath275 , which is less than 2 , the bound given by teleporting separately .
however when @xmath215 , teleporting @xmath274 directly gives a bound of @xmath276 , but teleporting separately gives only a bound of @xmath277 .
up to now , our discussion is entirely based on the @xmath0 hierarchy . to summarize the capacity of @xmath0 for fault - tolerant quantum computation and provide basis for comparison with non-@xmath0 schemes discussed below ,
we introduce another notion of @xmath278 .
@xmath278 is the minimum number of total teleportation steps needed to teleport separately each @xmath0 component of a quantum circuit using either one - bit or two - bit teleportation scheme .
@xmath278 is defined in a way that represents the maximum capacity of teleportation based on @xmath0 hierarchy . in general @xmath279 ,
@xmath280 . to understand exactly how they compare for a given circuit ,
a full characterization of the structure of @xmath0 is necessary .
here based on the structure theorems given in section iii , we gave a simple example where @xmath217 or @xmath261 could be strictly larger than @xmath278 .
the next question to ask is then whether we can go beyond @xmath0 and this will be discussed in the following section . in the definition of teleportation depth , we require that @xmath75 be decomposed into a set of @xmath207 gates .
this is due to the fact that @xmath207 are the only gates that we know so far how to perform fault - tolerantly by teleportation . in general ,
if we do not require the decomposition to be in @xmath207 , then we might get a better upper bound on teleportation depth than the one defined previously , i.e. there might exist upper bound @xmath281 of teleportation depth that is strictly less than @xmath278 .
we give two such examples below .
we leave open the problem of how to implement teleportations fault - tolerantly for a general @xmath1-qubit gate .
+ * example 1 * for a general one - qubit gate @xmath6 , we know that @xmath6 can be decomposed into three @xmath207 gates , each of which has @xmath282 . hence through the decomposition we can bound its total teleportation depth by @xmath283 .
however , to teleport @xmath6 directly without decomposition via two - bit teleportation gives a bound of @xmath284 less than @xmath278 .
+ * example 2 * consider a classical reversible circuit given in fig .
we denote this series of three toffoli gates as @xmath285 .
gate three toffoli gates in series , width=96 ]
this gate @xmath285 is not in @xmath0 hierarchy as can be shown below : suppose that @xmath286 is at certain level of the hierarchy , @xmath287 must be a gate in @xmath156 .
calculating explicitly as in fig .
[ fig6 ] we have by @xmath285,width=336 ] the non - clifford part of the right hand side of the equation is a series of two toffoli gates , and we denote it as @xmath288 . due to proposition 3 , @xmath288 is also in @xmath156 . as shown in fig .
[ fig7 ] , conjugating @xmath289 by @xmath288 results in @xmath290 , where @xmath141 is a clifford operation .
however , by exchanging the second and third qubit in fig .
[ fig7 ] , we find that @xmath291 , i.e. conjugating @xmath289 by @xmath292 gives back @xmath288 .
therefore , @xmath288 can not be in the @xmath0 hierarchy and we can conclude that @xmath285 is not a @xmath0 gate either .
@xmath47 by @xmath288,width=288 ] if we leave aside the problem of how to teleport gates beyond @xmath0 fault - tolerantly , we can teleport @xmath285 directly and obtain an upper bound of @xmath293 , which is less than @xmath294 , the bound given by teleporting the three toffoli gates separately .
in this paper we address the following questions : what is the capacity of the teleportation scheme in practical implementation of fault - tolerant quantum computation and what is the most efficient way to make use of the teleportation protocol . to answer these questions
we first notice that one - bit and two - bit teleportation schemes require different resources to implement and are of different capabilities . to understand what kind of gates can be teleported fault - tolerantly with these two schemes respectively
, we study the structure of @xmath0 hierarchy and its relationship with semi - clifford operations .
we show for @xmath2 , all the @xmath0 gates are semi - clifford operations , which is also true for @xmath3 . however , this is no longer true for parameters @xmath4 .
based on the counterexamples we constructed for @xmath295 , we conjecture that all @xmath30 gates are semi - clifford and all @xmath0 gates are generalized semi - clifford .
such an understanding of the @xmath0 structure has great implications on the optimal design of fault - tolerant architectures .
while all @xmath0 gates can be teleported fault - tolerantly , the semi - clifford subset of it requires less resources to implement than others . to quantify this notion of gate complexity in fault - tolerant quantum computation based on the @xmath0 hierarchy
, we introduce a measure called the teleportation depth @xmath211 , which characterizes how many teleportation steps are necessary , on average , to implement a given gate . using different teleportation schemes
, we can give different upper bounds on @xmath211 , for example @xmath217 , @xmath261 and @xmath278 .
general assumption was that @xmath296 .
however we showed in this work that , surprisingly for certain series of gates @xmath217 could be strictly greater than @xmath278 and @xmath278 could also be strictly greater than @xmath211 .
the ultimate understanding of the structure of @xmath0 will provide a clearer clue on how to teleport circuits most efficiently . to achieve this goal
, some results from other branches of mathematics might be helpful .
it is noted that the barnes - wall lattices , whose isometry group is a subgroup of index 2 in the real clifford group , have been extensively studied and recently their involutions have been classified @xcite .
it is our hope that the @xmath30 structure might be further understood once we have a better understanding of the clifford group . for @xmath59
, we fully characterize the structure of @xmath0 by further study on the diagonal gates in @xmath0 , which form a group .
it is interesting to note some evidence that @xmath0 gates might be the only non - clifford gates which could be transversally implemented on a stabilizer code @xcite .
we also fully characterize the structure of @xmath30 for @xmath297 , but this seems not directly related to allowable transversal non - clifford gates on stabilizer codes .
it is shown that those transversal non - clifford gates are allowed only if they are generalized semi - clifford @xcite , therefore we might expect some generalized semi - clifford @xmath0 gates transversally implementable on some stabilizer codes .
we believe such kind of exploration on the relationship between transversally implementable gates and teleportable gates will shed some light on further understanding of practical implementation of fault - tolerant architectures .
we thank daniel gottesman , debbie leung , and carlos mochon for comments .
in this section we discuss what kind of single qubit unitary gates could have eigenvalues @xmath298 apart from an overall phase factor , i.e. if @xmath299 denote the two eigenvalues of a single qubit unitary @xmath6 , then what is the condition under which @xmath300 .
this information is useful since only the unitary of this kind can be transformed into elements in pauli group under conjugation , i.e. there exits a unitary operator @xmath37 , such that @xmath301 , where @xmath74 .
we ll see that those kind of unitary has very restricted form which is given by the following proposition .
: the single qubit unitary gates which have eigenvalues @xmath298 apart from an overall phase factor could only be of the following two forms : @xmath302 $ ] or @xmath303 $ ] * proof * : we begin to prove this proposition by writing down a general form of single qubit unitary gate as the following : @xmath304\ ] ] direct calculation gives @xmath305 therefore @xmath300 gives @xmath306 if @xmath307 , the unitary must adopt the form of @xmath308 ; if @xmath309 , then apart from an overall phase , we can simply choose @xmath310 which leads to the form of @xmath311.@xmath47 note @xmath312 could be viewed as a special situation of @xmath313 for the case @xmath307 .
however , we list @xmath312 separately for future convenience . in this section
we investigate the gate seises associated with @xmath308 and @xmath311 .
it is obvious that if @xmath314 , then the unitary @xmath315 whose columns are the eigenvectors of @xmath308 or @xmath311 might be in @xmath60 , given that @xmath316 . for @xmath308
, the two normalized eigenvectors can be chosen as latexmath:[\ ] ]
let @xmath369 denote the clifford group for @xmath1 qubit . define the quotient group @xmath370 . since @xmath367 is a normal subgroup of @xmath371 , we could further define a quotient group @xmath372 . note @xmath373 and @xmath374
denote the set @xmath375 , i.e. @xmath376 are the set of all involutions of the symplectic group @xmath377 . for any @xmath1-qubit @xmath30 gate
@xmath6 , if @xmath407 , where @xmath408 , then the quotient group @xmath409 is abelian . because elements of @xmath410 are either commute or anticommute , the corresponding elements in @xmath403 should commute . and starting from elements in @xmath417 and calculate their eigenvectors , we understand that @xmath418 can only be of the values @xmath419 for @xmath420 .
this directly leads to the fact that the conjecture is true for any @xmath5 when @xmath59 .
see ck.pdf for the details of this . since @xmath374 , @xmath421 .
hence @xmath412 contains at least one element in @xmath422 .
however , this is not enough to claim @xmath423 holds for any two - qubit @xmath6 .
we need to examine the structure of @xmath412 in more detail .
consider the maximal abelian subgroup in @xmath424 of order @xmath425 , and its corresponding elements in @xmath371 , direct calculation shows it does not contain a subgroup of structure @xmath426 .
hence we need to further consider abelian subgroup in @xmath424 of order @xmath427 . due to lemma 3
, we result in @xmath428 holds for any two - qubit @xmath30 gate .
since a(3)=64 , and direct calculation of this group shows that not all the elements could be in @xmath431 , hence @xmath432 contains at least one element in @xmath433 .
again , this is not enough to claim @xmath434 holds for any three - qubit @xmath6 .
we need to examine the structure of @xmath432 in more detail to dig out 2 more elements in @xmath433 . * proof * : we first prove for @xmath443 .
note we choose @xmath439 to get rid of the overall phase of @xmath75 .
denote @xmath444 , and @xmath445 .
note @xmath446 , @xmath447 are also diagonal . since @xmath438
, @xmath447 must be in pauli apart from an overall phase .
and we also have @xmath448 , @xmath449 .
hence we must have @xmath450 , i.e. @xmath451 for some integer @xmath452 . | teleportation is a crucial element in fault - tolerant quantum computation and a complete understanding of its capacity is very important for the practical implementation of optimal fault - tolerant architectures . it is known that stabilizer codes support a natural set of gates that can be more easily implemented by teleportation than any other gates .
these gates belong to the so called @xmath0 hierarchy introduced by gottesman and chuang ( nature * 402 * , 390 ) .
moreover , a subset of @xmath0 gates , called semi - clifford operations , can be implemented by an even simpler architecture than the traditional teleportation setup ( phys .
rev . * a62 * , 052316 ) .
however , the precise set of gates in @xmath0 remains unknown , even for a fixed number of qubits @xmath1 , which prevents us from knowing exactly what teleportation is capable of . in this paper
we study the structure of @xmath0 in terms of semi - clifford operations , which send by conjugation at least one maximal abelian subgroup of the @xmath1-qubit pauli group into another one .
we show that for @xmath2 , all the @xmath0 gates are semi - clifford , which is also true for @xmath3 .
however , this is no longer true for @xmath4 . to measure the capability of this teleportation primitive
, we introduce a quantity called ` teleportation depth ' , which characterizes how many teleportation steps are necessary , on average , to implement a given gate .
we calculate upper bounds for teleportation depth by decomposing gates into both semi - clifford @xmath0 gates and those @xmath0 gates beyond semi - clifford operations , and compare their efficiency . |
extrinsic geometry deals with the study of the geometry of subspaces relative to that of an ambient space . given a riemannian manifold @xmath0 and a submanifold @xmath1 , classical ( differential ) extrinsic geometry studies infinitesimal changes in the riemannian metric on @xmath1 induced from @xmath0 .
this involves an analysis of the second fundamental form or shape operator @xcite . in coarse geometry local or infinitesimal machinery
is absent .
thus it does not make sense to speak of tangent spaces or riemannian metrics .
however , the large scale notion of metric continues to make sense . given a metric space @xmath2 and a subspace @xmath3 one
can still compare the intrinsic metric on @xmath3 to the metric inherited from @xmath2 .
this is especially useful for finitely generated subgroups of finitely generated groups . to formalize this , gromov introduced the notion of distortion in his seminal paper @xcite .
( @xcite,@xcite)if @xmath4 is an embedding of the cayley graph of @xmath5 into that of @xmath6 , then the _ distortion _ function is given by @xmath7 , where @xmath8 is the ball of radius @xmath9 around @xmath10 .
the definition above differs from the one in @xcite by a linear factor and coincides with that in @xcite .
* note*the above definition continues to make sense when @xmath11 and @xmath12 are replaced by graphs or ( more generally ) path - metric spaces ( see below for definition ) @xmath2 and @xmath3 respectively .
a _ path - metric space _ is a metric space @xmath13 such that for all @xmath14 there exists an isometric embedding @xmath15 \rightarrow x$ ] with @xmath16 and @xmath17 . if the distortion function is linear we say @xmath12 ( or @xmath3 ) is _ quasi - isometrically _ ( often abbreviated to _ qi _ ) embedded in @xmath11 ( or @xmath2 )
this is equivalent to the following : a map @xmath18 from one metric space @xmath19 into another metric space @xmath20 is said to be a _
@xmath21quasi - isometric embedding _
if @xmath22 if @xmath18 is a quasi - isometric embedding , and every point of @xmath23 lies at a uniformly bounded distance from some @xmath24 then @xmath18 is said to be a _
quasi - isometry_. a @xmath25quasi - isometric embedding that is a quasi - isometry will be called a@xmath25quasi - isometry .
we collect here a few other closely related notions : a subset @xmath23 of @xmath2 is said to be _ @xmath26quasiconvex _ if any geodesic joining @xmath27 lies in a @xmath26neighborhood of @xmath23 .
a subset @xmath23 is _ quasiconvex _ if it is @xmath26quasiconvex for some @xmath26 .
@xmath21quasigeodesic _ is a @xmath21quasi - isometric embedding of a closed interval in @xmath28 .
a @xmath29quasigeodesic will also be called a @xmath30quasigeodesic . for hyperbolic metric spaces ( in the sense of gromov @xcite ) the notions of quasiconvexity and qi embeddings coincide .
this is because quasigeodesics lie close to geodesics in hyperbolic metric spaces @xcite , @xcite , @xcite .
distortion can be regarded , in some sense , as the simplest extrinsic notion in coarse geometry .
however a complete understanding of distortion is lacking even in special situations like subgroups of hyperbolic groups or discrete ( infinite co - volume ) subgroups of higher rank semi - simple lie groups .
one of the aims of this survey is to expose some of the issues involved .
this is done in section 2 .
a characterisation of quasi - isometric embeddings in terms of group theory is another topic of extrinsic geometry that has received some attention of late .
this will be dealt with in section 3 .
a different perspective of coarse extrinsic geometry comes from the asymptotic point of view .
the issue here is behavior ` at infinity ' . from this perspective it seems possible to introduce and study finer invariants involving distortion along specified directions .
section 4 deals with this in the special context of hyperbolic subgroups of hyperbolic groups . finally in section 5
, we discuss some other invariants of extrinsic geometry that have come up in different contexts .
it goes without saying that this survey reflects the author s bias and is far from comprehensive .
* acknowledgements*i would like to thank the organizers of david epstein s birthday fest where the idea of this survey first came up .
i would also like to take this opportunity to thank benson farb for several inspiring discussions on coarse geometry over the years .
if a finitely generated subgroup @xmath5 of a finitely generated group @xmath6 is qi embedded we shall refer to it as undistorted . otherwise @xmath5 will be said to be distorted .
we shall also have occasion to replace the cayley graph of @xmath6 by a symmetric space ( equipped with its invariant metric ) or more generally a path metric space @xmath13 . in the latter case
, distortion will be measured with respect to the metric @xmath31 on @xmath2 .
distorted subgroups of hyperbolic groups or symmetric spaces are somewhat difficult to come by .
this has resulted in a limited supply of examples .
brief accounts will be given of some of the known sources of examples .
an aspect that will not be treated in any detail is the connection to algorithmic problems , especially the magnus problem .
see @xcite or ( for a more detailed account ) @xcite for a treatment .
one of the earliest classes of examples of distorted hyperbolic subgroups of hyperbolic groups came from thurston s work on @xmath32manifolds fibering over the circle @xcite .
let @xmath0 be a closed hyperbolic @xmath32manifold fibering over the circle with fiber @xmath33 .
then @xmath34 is a hyperbolic subgroup of the hyperbolic group @xmath35 .
the distortion is easily seen to be exponential .
it follows from work of bonahon @xcite and thurston @xcite that if @xmath5 is a closed surface subgroup of the fundamental group @xmath35 of a closed hyperbolic @xmath32manifold @xmath0 then the distortion of @xmath5 is either linear or exponential .
this continues to be true if @xmath5 is replaced by any freely indecomposable group .
in fact exponential distortion of a freely indecomposable group corresponds precisely ( up to passing to a finite cover of @xmath0 ) to the case of a hyperbolic @xmath32manifold fibering over the circle .
the situation is considerably less clear when we come to freely decomposable subgroups of hyperbolic @xmath32manifolds .
the tameness conjecture ( attributed to marden @xcite ) asserts that the covering of a closed hyperbolic @xmath32manifold corresponding to a finitely generated subgroup of its fundamental group is topologically tame , ie is homeomorphic to the interior of a compact @xmath32manifold with boundary .
if this conjecture were true , it would follow ( using a theorem of canary @xcite ) that any finitely generated subgroup @xmath5 of the fundamental group @xmath36 is either quasiconvex in @xmath6 or is exponentially distorted
. moreover , exponential distortion corresponds precisely ( up to passing to a finite cover of @xmath0 ) to the case of a fiber of a hyperbolic @xmath32manifold fibering over the circle .
much of this theory can be extended to take parabolics into account .
this class of examples can be generalized in two directions .
one can ask for distorted discrete subgroups of @xmath37 or for distorted hyperbolic subgroups of hyperbolic groups ( in the sense of gromov ) . we look first at discrete subgroups of @xmath37 .
a substantial class of examples comes from geometrically tame groups .
in fact the simplest surface group , the fundamental group of a punctured torus ( the puncture corresponds to a parabolic element ) , displays much of the exotic extrinsic geometry that may occur .
these examples were studied in great detail by minsky in @xcite .
the distortion function was calculated in @xcite .
let @xmath38 be a hyperbolic punctured torus so that the two shortest geodesics @xmath39 and @xmath40 are orthogonal and of equal length .
let @xmath41 denote @xmath38 minus a neighborhood of the cusp .
let @xmath42 and @xmath43 be regular collar neighborhoods of @xmath39 and @xmath40 in @xmath41 .
for @xmath44 , define @xmath45 if @xmath46 is even and equal to @xmath40 if @xmath46 is odd .
let @xmath47 be the open solid torus neighborhood of @xmath48 in @xmath49 given by @xmath50 and let @xmath51 .
let @xmath52 be a sequence of positive integers greater than one .
let @xmath53 and let @xmath54 be an oriented meridian for @xmath55 with a single positive intersection with @xmath56 .
let @xmath0 denote the result of gluing to each @xmath55 a solid torus @xmath57 , such that the curve @xmath58 is glued to a meridian .
let @xmath59 be the mapping class from @xmath41 to itself obtained by identifying @xmath41 to @xmath60 , pushing through @xmath0 to @xmath61 and back to @xmath41 .
then @xmath62 is given by @xmath63 , where @xmath64 denotes dehn twist along @xmath65 , @xmath26 times .
matrix representations of @xmath66 are given by @xmath67 and @xmath68 recall that the metric on @xmath69 is the restriction of the product metric .
the @xmath57 s are given hyperbolic metrics such that their boundaries are uniformly quasi - isometric to @xmath70 .
then from @xcite , @xmath0 is quasi - isometric to the complement of a rank one cusp in the convex core of a hyperbolic manifold @xmath71 .
let @xmath72 denote the shortest path from @xmath73 to @xmath74 .
let @xmath75 denote @xmath72 with reversed orientation .
then @xmath76 is a closed path in @xmath0 of length @xmath77 .
further @xmath78 is homotopic to a curve @xmath79 on @xmath41 .
then @xmath80 hence @xmath81 since @xmath0 is quasi - isometric to the complement of the cusp of a hyperbolic manifold @xcite and @xmath82 s lie in a complement of the cusp , the distortion function of @xmath83 is of the same order as the distortion function above . in particular ,
functions of arbitrarily fast growth may be realized .
this answers a question posed by gromov in @xcite page 66 .
a closely related class of examples ( the so called ` drill holes ' examples of which the punctured torus examples above may be regarded as special cases ) appears in work of thurston @xcite and bonahon and otal @xcite .
let us now turn to finitely generated subgroups of hyperbolic groups .
if we restrict ourselves to hyperbolic subgroups there is a considerable paucity of examples .
the chief ingredient for constructing distorted hyperbolic subgroups of hyperbolic groups is the celebrated combination theorem of bestvina and feighn @xcite .
this theorem was partly motivated by thurston s hyperbolization theorem for haken manifolds @xcite , @xcite and continues to be an inevitable first step in constructing any distorted hyperbolic subgroups .
the following proposition summarizes these examples .
the proof follows easily from normal forms .
let @xmath6 be a hyperbolic group acting cocompactly on a simplicial tree @xmath84 such that all vertex and edge stabilizers are hyperbolic .
also suppose that every inclusion of an edge stabilizer in a vertex stabilizer is a quasi - isometric embedding .
let @xmath5 be the stabilizer of a vertex or edge of @xmath84 .
then the distortion of @xmath5 is linear or exponential .
[ bf ] based on bestvina and feighn s combination theorem and work of thurston s on stable and unstable foliations of surfaces @xcite , mosher @xcite constructed a class of examples of normal surface subgroups of hyperbolic groups where the quotient is free of rank strictly greater than one .
this idea was used by bestvina , feighn and handel in @xcite to construct similar examples where the normal subgroup is free .
thus one has examples of exact sequences @xmath85 of hyperbolic groups where @xmath1 is a free group or a surface group .
owing to a general theorem of mosher s regarding the existence of quasi - isometric sections of @xmath86 @xcite the distortion of any normal hyperbolic subgroup @xmath1 of infinite index in a hyperbolic group @xmath6 is exponential .
further , it follows from work of rips and sela @xcite , @xcite that a torsion free normal hyperbolic subgroup of a hyperbolic group is a free product of free groups and surface groups . however , the only known restriction on @xmath86 is that it is hyperbolic @xcite .
it seems natural to wonder if there exist examples where the exact sequence does not split or at least where @xmath86 is not virtually free .
we now describe some examples exhibiting higher distortion @xcite .
start with a hyperbolic group @xmath6 such that @xmath87 is exact , where @xmath33 is free of rank 3 .
let @xmath88 denote the normal subgroup .
let @xmath89 denote a section of the quotient group .
let @xmath90 be @xmath46 distinct copies of @xmath6 .
let @xmath91 and @xmath92 denote copies of @xmath93 and @xmath94 respectively in @xmath95 .
let + @xmath96 where each @xmath97 is a free group of rank 3 , the image of @xmath97 in @xmath95 is @xmath92 and the image of @xmath97 in @xmath98 is @xmath99 .
then @xmath6 is hyperbolic .
let @xmath100 .
then the distortion of @xmath5 is superexponential for @xmath101 .
in fact , it can be checked inductively that the distortion function is an iterated exponential of height @xmath46 .
starting from bestvina , feighn and handel s examples above , one can construct examples with distortion a tower function .
let @xmath102 be generators of @xmath93 and @xmath103 be generators of @xmath94
. then @xmath104 where @xmath105 are words in @xmath106 s .
we add a letter @xmath65 conjugating @xmath106
s to ` sufficiently random ' words in @xmath107 s to get @xmath108 . thus , @xmath109 , where @xmath110 s are words in @xmath107 s satisfying a small - cancellation type condition to ensure that @xmath108 is hyperbolic .
see @xcite , page 151 for details on addition of ` random ' relations .
it can be checked that these examples have distortion function greater than any iterated exponential .
the above set of examples were motivated largely by examples of distorted cyclic subgroups in @xcite , page 67 and @xcite ( these examples will be discussed later in this paper ) .
so far , there is no satisfactory way of manufacturing examples of hyperbolic subgroups of hyperbolic groups exhibiting arbitrarily high distortion .
it is easy to see that a subgroup of sub - exponential distortion is quasiconvex @xcite .
not much else is known .
one is thus led to the following question : given any increasing function @xmath111 does there exist a hyperbolic subgroup @xmath5 of a hyperbolic group @xmath6 such that the distortion of @xmath5 is of the order of @xmath112 ?
note that the above question has a positive answer if @xmath6 is replaced by @xmath113 .
if one does not restrict oneself to hyperbolic subgroups of hyperbolic groups , one has a large source of examples coming from finitely generated subgroups of small cancellation groups .
these examples are due to rips @xcite .
let @xmath114 be any finitely presented group .
construct a small cancellation ( @xmath115 ) group @xmath6 with presentation as follows : @xmath116 + for @xmath117 , @xmath118 and @xmath119 .
@xmath120 where @xmath121 , @xmath122 , @xmath123 are words in @xmath124 , @xmath125 satisfying @xmath115 .
then one has an exact sequence @xmath126 where @xmath5 is the subgroup of @xmath6 generated by @xmath127 and @xmath86 is the given finitely presented group .
the distortion of @xmath5 can be made to vary by varying @xmath86 ( one basically needs to vary the complexity of the word problem in @xmath86 ) .
however the subgroups @xmath5 are generally not finitely presented .
a remarkable example of a finitely presented normal subgroup @xmath5 of a hyperbolic group @xmath6 has recently been discovered by brady @xcite .
this is the first example of a finitely presented non - hyperbolic subgroup of a hyperbolic group .
the distortion in this example is exponential as the quotient group is infinite cyclic .
now let @xmath6 be a semi - simple lie group .
cyclic discrete subgroups generated by unipotent elements are exponentially distorted .
this is because discrete subgroups of the nilpotent subgroup @xmath1 in a @xmath128 decomposition of @xmath6 is distorted in this way .
this is the most well known source of distortion .
other known examples seem to have their origin in rank 1 phenomena . given any lie group @xmath6 containing @xmath129 as a discrete subgroup one has distorted subgroups coming from a construction due to mihailova @xcite , @xcite , @xcite ( see below ) .
in some sense these examples are ` reducible ' .
truly higher rank phenomena are hard to come by .
one has the following basic question : are there examples of distorted finitely generated discrete subgroups @xmath5 of irreducible lattices in higher rank semi - simple lie groups @xmath6 such that @xmath5 has no unipotent element ? ( see @xcite also ) .
note that thurston s construction of normal subgroups can not possibly go through here on account of the following basic theorem of kazhdan margulis : @xcitelet @xmath83 be an irreducible lattice in a symmetric space of real rank greater than one .
then any normal subgroup @xmath130 of @xmath83 is either finite or the quotient @xmath131 is finite .
another non - distortion theorem has recently been proven by lubotzky moses
raghunathan @xcite answering a question of kazhdan : any irreducible lattice in a symmetric space @xmath2 of rank greater than one is undistorted in @xmath2 .
the above theorems indicate the difficulty in obtaining distorted subgroups of higher rank lie groups .
similar questions may be asked for rank one symmetric spaces also eg for complex hyperbolic , quaternionic hyperbolic and the cayley hyperbolic planes . here ,
too there is a dearth of examples . in real hyperbolic spaces ,
the situation is slightly better owing to thurston s examples of @xmath32manifolds fibering over the circle . based partly on thurston s examples , bowditch and mess
@xcite have described an example of a finitely generated subgroup of a uniform lattice in @xmath132 that is not finitely presented .
abresch and schroeder @xcite have given an arithmetic construction of this lattice , too .
one wonders if this arithmetic description can be used to give similar examples in @xmath133 or @xmath134 .
such infinitely presented subgroups are necessarily distorted .
related examples have also been discovered by potyagailo and kapovich @xcite , @xcite .
a natural question is whether thurston s construction goes through in higher dimensions or not : does there exist a uniform lattice in a rank one symmetric ( other than @xmath135 ) space containing a finitely presented ( or even finitely generated ) infinite normal subgroup of infinite index
? one should note that any such normal subgroup can not be hyperbolic ( by @xcite ) .
there are certain special classes of distorted subgroups of finitely presented groups that do not fall into any of the above categories .
a basic class of examples comes from the baumslag solitar groups @xmath136 where the cyclic group generated by @xmath39 has exponential distortion for @xmath101 .
a class of examples with higher distortion have appeared in work of gersten @xcite .
we briefly describe these .
take @xmath137 for @xmath138
. then the cyclic subgroup generated by @xmath139 has distortion an iterated exponential function of height @xmath46 .
next consider @xmath140 .
then the cyclic group generated by @xmath39 has distortion greater than any iterated exponential .
another class of subgroups with distortion a fractional power occurs in work of bridson @xcite : + let @xmath141 where @xmath142 is the unipotent matrix with ones on the diagonal and superdiagonal and zeroes elsewhere . for @xmath143 , @xmath144 has infinite cyclic center . given two such groups @xmath145 , @xmath146 amalgamate them along their cyclic center @xmath147 to get @xmath148 .
then the distortion function of @xmath146 in @xmath149 is of the form @xmath150 .
a large class of examples of distortion arise from subgroups of nilpotent and solvable groups @xcite .
finally we describe a class of examples due to mihailova @xcite which give rise to non - recursive distortion ( see also @xcite @xcite ) .
let @xmath151 be any finitely presented group with defining presentation @xmath152 .
then @xmath153 maps @xmath154 to @xmath155 .
the pull - back @xmath5 under this map of the ` diagonal subgroup ' @xmath156 is generated by elements of the form @xmath157 , @xmath117 and @xmath158 , @xmath159 .
if @xmath6 has unsolvable word problem , then the distortion of @xmath5 in @xmath160 is non - recursive .
it was seen in the previous section that construction of distorted subgroups usually involves some amount of work .
in fact for subgroups of hyperbolic groups , gromov @xcite describes ` length angle ' relationships between generators that would ensure quasiconvexity of the subgroup .
this can be taken as a genericity result . in another setting
, one could ask for examples of groups all whose finitely generated subgroups are undistorted .
this is known for free groups , surface groups and abelian groups
. however , a general group - theoretic characterization of quasiconvexity seems far off .
gersten has recently described a functional analytic approach to this problem .
we briefly describe this .
later we shall discuss a more group - theoretic approach .
we shall restrict ourselves to finitely generated subgroups of hyperbolic groups ( in the sense of gromov ) in this section .
the following discussion appears in @xcite , @xcite , @xcite .
let @xmath161 be a complex of type @xmath162 with finite @xmath163 skeleton @xmath164 and let @xmath2 be the universal cover of @xmath161 .
the vector space of cellular chains @xmath165 is equipped with the @xmath166 norm for a basis of @xmath167cells .
then the boundary maps @xmath168 are bounded linear and ( owing to the finiteness of the @xmath169skeleton ) one gets quasi - isometry invariant homology groups @xmath170 for @xmath171 . since these homology groups are quasi - isometry invariant it makes sense to define @xmath172 for @xmath171 for any such @xmath2 .
the following theorem of gersten s occurs in @xcite .
the finitely presented group @xmath6 is hyperbolic if and only if @xmath173 .
moreover , if @xmath5 is a finitely generated subgroup of @xmath6 then @xmath5 is quasiconvex if and only if the map @xmath174 induced by inclusion is injective . earlier results along these lines had been found in @xcite , @xcite , @xcite . in a different direction
, one would like a purely group - theoretic characterization of quasiconvexity .
we start with some definitions .
let @xmath5 be a subgroup of a group @xmath6 .
we say that the elements @xmath175 of @xmath6 are essentially distinct if @xmath176 for @xmath177 .
conjugates of @xmath5 by essentially distinct elements are called essentially distinct conjugates . note
that we are abusing notation slightly here , as a conjugate of @xmath5 by an element belonging to the normalizer of @xmath5 but not belonging to @xmath5 is still essentially distinct from @xmath5 .
thus in this context a conjugate of @xmath5 records ( implicitly ) the conjugating element .
we say that the height of an infinite subgroup @xmath5 in @xmath6 is @xmath46 if there exists a collection of @xmath46 essentially distinct conjugates of @xmath5 such that the intersection of all the elements of the collection is infinite and @xmath46 is maximal possible .
we define the height of a finite subgroup to be @xmath178 . the main theorem of @xcite states : if @xmath5 is a quasiconvex subgroup of a hyperbolic group @xmath6,then @xmath5 has finite height .
[ gmrs ] the following question of swarup was prompted partly by this result : ( swarup)suppose @xmath5 is a finitely presented subgroup of a hyperbolic group @xmath6 .
if @xmath5 has finite height is @xmath5 quasiconvex in @xmath6 ?
so far only some partial answers have been obtained .
the first result is due to scott and swarup : @xcite let @xmath179 be an exact sequence of hyperbolic groups induced by a pseudo anosov diffeomorphism of a closed surface with fundamental group @xmath5 .
let @xmath180 be a finitely generated subgroup of infinite index in @xmath5 .
then @xmath180 is quasiconvex in @xmath6 . in @xcite
an analogous result for free groups was derived
. the methods also provide a different proof of scott and swarup s theorem above : @xcite let @xmath181 be an exact sequence of hyperbolic groups induced by a hyperbolic automorphism @xmath182 of the free group @xmath5 .
let @xmath183 be a finitely generated distorted subgroup of @xmath6 .
then there exist @xmath184 and a free factor @xmath30 of @xmath5 such that the conjugacy class of @xmath30 is preserved by @xmath185 and @xmath180 contains a finite index subgroup of a conjugate of @xmath30 .
another special case where one has a positive answer is the following : @xcite let @xmath6 be a hyperbolic group splitting over @xmath5 ( ie @xmath186 or @xmath187 ) with hyperbolic vertex and edge groups .
further , assume the two inclusions of @xmath5 are quasi - isometric embeddings .
then @xmath5 is of finite height in @xmath6 if and only if it is quasiconvex in @xmath6 .
[ mit5 ] swarup s question is therefore still open in the following special case , which can be regarded as a next step following the theorems of @xcite and @xcite above .
suppose @xmath6 splits over @xmath5 satisfying the hypothesis of theorem [ mit5 ] above and @xmath180 is a quasiconvex subgroup of @xmath5 .
if @xmath180 has finite height in @xmath6 is it quasiconvex in @xmath6 ? more generally ,
if @xmath180 is an edge group in a hyperbolic graph of hyperbolic groups satisfying the qi embedded condition , is @xmath5 quasiconvex in @xmath6 if and only if it has finite height in @xmath6 ?
a closely related problem can be formulated in more geometric terms : * question*let @xmath188 be a finite 2 complex with fundamental group @xmath6 .
let @xmath189 be a cover of @xmath188 corresponding to the finitely presented subgroup @xmath5 .
let @xmath190 be the injectivity radius of @xmath189 at @xmath191 .
does @xmath192 as @xmath193 imply that @xmath5 is quasi - isometrically embedded in @xmath6 ? a positive answer to this question for @xmath6 hyperbolic would provide a positive answer to swarup s question .
the answer to this question is negative if one allows @xmath6 to be only finitely generated instead of finitely presented as the following example shows : let @xmath194 denote the free group on four generators .
let @xmath195 and @xmath196 for some function @xmath197 .
introducing a stable letter @xmath198 conjugating @xmath199 to @xmath110 one has a finitely generated hnn extension @xmath6 .
the free subgroup generated by @xmath200 provides a negative answer to the question above for suitable choice of @xmath18 .
in fact one only requires that @xmath18 grows faster than any linear function .
if @xmath18 is recursive one can embed the resultant @xmath6 in a finitely presented group by higman s embedding theorem .
but then one might lose malnormality of the free subgroup generated by @xmath200 . if one can have some control over the embedding in a finitely presented group
, one might look for a counterexample .
a closely related example was shown to the author by steve gersten .
so far the following question ( attributed to bestvina and brady ) remains open : let @xmath6 be a finitely presented group with a finite @xmath162 .
suppose moreover that @xmath6 does not contain any subgroup isomorphic to @xmath201 .
is @xmath6 hyperbolic ?
a malnormal counterexample to swarup s question would provide a counterexample for the above question ( observed independently by m. sageev ) .
the purpose of this section is to take an asymptotic rather than a coarse point of view and expose some of the problems from this perspective .
since virtually all the work in this area involves actions of hyperbolic groups on hyperbolic metric spaces we restrict ourselves mostly to this . roughly speaking
, one would like to know what happens ` at infinity ' .
we put this in the more general context of a hyperbolic group @xmath5 acting freely and properly discontinuously by isometries on a proper hyperbolic metric space @xmath2 .
then there is a natural map @xmath202 , sending the vertex set of @xmath12 to the orbit of a point under @xmath5 , and connecting images of adjacent vertices in @xmath12 by geodesics in @xmath2 .
let @xmath203 denote the gromov compactification of @xmath2 .
the basic question discussed in this section is the following : does the continuous proper map @xmath204 extend to a continuous map @xmath205 ?
a measure theoretic version of this question was asked by bonahon in @xcite . a positive answer to the above
would imply a positive answer to bonahon s question .
related questions in the context of kleinian groups have been studied by cannon and thurston @xcite , bonahon @xcite , floyd @xcite and minsky @xcite .
much of the work around this problem was inspired by a seminal ( unpublished ) paper of cannon and thurston @xcite .
the main theorem of @xcite states : @xcitelet @xmath0 be a closed hyperbolic @xmath32manifold fibering over the circle with fiber @xmath33 .
let @xmath206 and @xmath207 denote the universal covers of @xmath33 and @xmath0 respectively . then @xmath206 and @xmath207 are quasi - isometric to @xmath208 and @xmath135 respectively .
let @xmath209 and @xmath210 denote the standard compactifications .
then the usual inclusion of @xmath206 into @xmath207 extends to a continuous map from @xmath211 to @xmath212 .
[ ct ] the proof of the above theorem involved the construction of a local ` sol - like ' metric using affine structures on surfaces coming from stable and unstable foliations . coupled with thurston s hyperbolization of @xmath32manifolds fibering over the circle one has a very explicit description of the boundary extension . using these ( local ) methods minsky
@xcite generalized this theorem to the following : @xcite let @xmath83 be a kleinian group isomorphic ( as a group ) to the fundamental group of a closed surface , such that @xmath213 has injectivity radius uniformly bounded below by some @xmath214
. then there exists a continuous map from the gromov boundary of @xmath83 ( regarded as an abstract group ) to the limit set of @xmath83 in @xmath215 .
[ minsky ] finally klarreich @xcite generalized the above theorem to the case of freely indecomposable kleinian groups .
a different proof was given by the author @xcite ( see below ) .
let @xmath83 be a freely indecomposable kleinian group , such that @xmath213 has injectivity radius uniformly bounded below by some @xmath214 .
then there exists a continuous map from the gromov boundary of @xmath83 ( regarded as an abstract group ) to the limit set of @xmath83 in @xmath215 .
[ locconn ] klarreich proved theorem [ locconn ] by combining her theorem [ klarreich ] below with theorem [ minsky ] above .
@xcitelet @xmath2 and @xmath3 be proper , geodesic gromov
hyperbolic spaces , @xmath216 a collection of closed , disjoint path - connected subsets of @xmath2 , and @xmath217 a quasi lipschitz map such that for every @xmath218 , @xmath219 restricted to @xmath218 extends continuously to the boundary at infinity .
suppose that the following hold : 1 .
the complement in @xmath2 of the sets @xmath218 is open and path - connected as also the complement of @xmath220 in @xmath3 .
2 . there is some real number @xmath221 such that the sets @xmath218 are all @xmath26quasiconvex in @xmath2 and @xmath220 s are @xmath26quasiconvex in @xmath3 .
3 . there is a real number @xmath222 such that @xmath223 and such that @xmath224 for all @xmath225 and @xmath226 .
then if the map @xmath219 induced on the electric spaces is a quasi - isometry , @xmath219 extends continuously to a continuous map from the boundary of @xmath2 to the boundary of @xmath3
. here the electric spaces are the spaces obtained from @xmath2 and @xmath3 by collapsing each space @xmath218 ( or @xmath227 ) to points : they inherit path metrics from @xmath2 and @xmath3 .
[ klarreich ] one should note that since cannon and thurston s theorem [ ct ] deals with asymptotic behavior it might well be regarded as a theorem in coarse geometry .
the above theorems are all of this form . but
the proof techniques in @xcite , @xcite are local as they rely on thurston s theory of singular foliations of surfaces . in @xcite and @xcite a different approach
was described using purely large - scale techniques giving generalized versions of theorems [ ct ] [ minsky ] and [ locconn ] .
@xcite let @xmath6 be a hyperbolic group and let @xmath5 be a hyperbolic subgroup that is normal in @xmath6 .
let @xmath228 be the continuous proper embedding of @xmath12 in @xmath11 described above .
then @xmath167 extends to a continuous map @xmath229 from @xmath230 to @xmath231 .
a more useful generalization of theorem [ ct ] is : @xcite let ( x , d ) be a tree ( t ) of hyperbolic metric spaces satisfying the quasi - isometrically embedded condition .
let @xmath232 be a vertex of @xmath84 .
let @xmath233 denote the hyperbolic metric space corresponding to @xmath232 .
if @xmath2 is hyperbolic then the inclusion of @xmath234 in @xmath2 extends continuously to the boundary .
[ tree ] a direct consequence of theorem [ tree ] above is the following : let @xmath6 be a hyperbolic group acting cocompactly on a simplicial tree @xmath84 such that all vertex and edge stabilizers are hyperbolic .
also suppose that every inclusion of an edge stabilizer in a vertex stabilizer is a quasi - isometric embedding .
let @xmath5 be the stabilizer of a vertex or edge of @xmath84 .
then an inclusion of the cayley graph of @xmath5 into that of @xmath6 extends continuously to the boundary . in @xcite ,
bestvina and feighn give sufficient conditions for a graph of hyperbolic groups to be hyperbolic .
vertex and edge subgroups are thus natural examples of hyperbolic subgroups of hyperbolic groups .
these examples are covered by the above corollary .
using thurston s pleated surfaces technology one then gives a ` coarse ' proof of theorem [ locconn ] . with some further work and using a theorem of minsky @xcite
, one can give @xcite a ` partly coarse ' proof of another result of minsky @xcite : thurston s ending lamination conjecture for geometrically tame manifolds with freely indecomposable fundamental group and a uniform lower bound on injectivity radius .
@xcitelet @xmath235 and @xmath236 be homeomorphic hyperbolic @xmath32mani - folds with freely indecomposable fundamental group .
suppose there exists a uniform lower bound @xmath214 on the injectivity radii of @xmath235 and @xmath236 .
if the end invariants of corresponding ends of @xmath235 and @xmath236 are equal , then @xmath235 and @xmath236 are isometric .
one should note here that the coarse techniques referred to circumvent only the building of a ` model manifold ' a local construction in @xcite .
it might be worthwhile to obtain a coarse proof of the main theorem of @xcite .
a positive answer to the following coarse question will do the job ( as can be seen from @xcite ) : let @xmath237 be a map . for @xmath238 a closed curve on @xmath38 , let @xmath239 denote the length of the shortest curve freely homotopic to @xmath238 on @xmath240 .
suppose there exists @xmath241 such that for all closed curves @xmath238 on @xmath38 one has @xmath242 for all @xmath243 .
then does @xmath244 lie in a bounded neighborhood of a teichmuller geodesic ?
the above question was motivated in part by the ` hallways flare ' condition of @xcite and a recent relative hyperbolicity result of masur minsky @xcite . since a continuous image of a compact locally connected set
is locally connected @xcite theorem [ locconn ] also shows that the limit sets of freely indecomposable kleinian groups with a uniform lower bound on the injectivity radius are locally connected .
the issue of local connectivity has received a lot of attention lately due to some recent foundational work of bowditch and swarup @xcite , @xcite , @xcite , @xcite , @xcite following earlier work by bestvina and mess @xcite .
let @xmath5 be a one - ended hyperbolic group
. then itsboundary is locally connected .
next assume @xmath5 does not split over any two - ended group and acts on a proper hyperbolic metric space @xmath2 with limit set @xmath245
. then @xmath130 is locally connected .
[ bow ] the existence of continuous boundary extensions in general would thus imply ( using theorem [ bow ] ) local connectivity of limit sets of hyperbolic groups acting on proper hyperbolic metric spaces
. one wonders if some kind of a converse exists .
such speculations are prompted on the one hand by theorem [ bow ] and by the following observation .
let @xmath83 be a simply degenerate kleinian group isomorphic to a surface group .
further assume @xmath83 has no parabolics .
let @xmath130 be the limit set of @xmath83 , @xmath246 its domain of discontinuity and @xmath2 the boundary of the convex hull of @xmath130 .
then ` nearest point projections ' give a natural homeomorphism between @xmath246 and @xmath2 .
from this it is easy to conclude that a continuous boundary extension exists if and only if a neighborhood of @xmath130 in @xmath247 deformation retracts onto @xmath130 . in this special case
therefore local connectivity is equivalent to continuous boundary extensions . before concluding this section it is worth pointing out that one needs finer invariants than distortion to understand asymptotic extrinsic geometry
. one way of approaching the problem is to consider extrinsic geometry of rays ( starting at @xmath248 ) and describe those which are not quasigeodesics in the ambient space @xmath2 .
if one looks at bi - infinite geodesics instead of rays one gets ` ending laminations ' . for
@xmath32manifolds fibering over the circle with fiber @xmath33 and monodromy @xmath182 one can think of these as the stable and unstable foliations of @xmath182 .
motivated by this , the author gave a more group theoretic description in @xcite in the special case of a hyperbolic normal subgroup of a hyperbolic group .
recall that for a hyperbolic @xmath32manifold @xmath0 fibering over the circle with fiber @xmath33 cannon and thurston show in @xcite that the usual inclusion of @xmath206 into @xmath207 extends to a continuous map from @xmath211 to @xmath212 .
an explicit description of this map was also described in @xcite in terms of ` ending laminations ' [ see @xcite for definitions ] .
the explicit description depends on thurston s theory of stable and unstable laminations for pseudo - anosov diffeomorphisms of surfaces @xcite . in the case of normal hyperbolic subgroups of hyperbolic groups ,
though existence of a continuous extension @xmath249 was proven in @xcite , an explicit description was missing . in @xcite some parts of thurston s theory of ending laminations were generalized to the context of normal hyperbolic subgroups of hyperbolic groups . using this an explicit description of the continuous boundary extension
@xmath250 was given for @xmath5 a normal hyperbolic subgroup of a hyperbolic group @xmath6 .
in general , if @xmath251 is an exact sequence of finitely presented groups where @xmath5 , @xmath6 and hence @xmath86 ( from @xcite ) are hyperbolic , one has ending laminations naturally parametrized by points in the boundary @xmath252 of the quotient group @xmath86 . corresponding to every element @xmath253 there exists an automorphism of @xmath5 taking @xmath219 to @xmath254 for @xmath255 .
such an automorphism induces a bijection @xmath256 of the vertices of @xmath12 .
this gives rise to a map from @xmath12 to itself , sending an edge [ @xmath257 linearly to a shortest edge - path joining @xmath258 to @xmath259 .
fixing @xmath260 for the time being ( for notational convenience ) we shall define the set of ending laminations corresponding to @xmath261 .
let @xmath262 be a semi - infinite geodesic ray in @xmath263 starting at the identity @xmath264 and converging to @xmath265 .
let @xmath244 be a single - valued quasi - isometric section of @xmath86 into @xmath6 .
let @xmath266 be the vertex on @xmath262 such that @xmath267 and let @xmath268 .
given @xmath269 let @xmath270 be the ( @xmath5invariant ) collection of all free homotopy representatives ( or shortest representatives in the same conjugacy class ) of @xmath271 in @xmath12 . identifying equivalent geodesics in @xmath270
one obtains a subset @xmath272 of ( unordered ) pairs of points in @xmath273 . the intersection with @xmath274 of the union of all subsequential limits ( in the chabauty topology ) of @xmath275
will be denoted by @xmath276 .
the set of _ ending laminations corresponding to @xmath265 _ is given by @xmath277 the set @xmath130 of all _ ending laminations _ is defined by @xmath278 it was shown in @xcite that the continuous boundary extension @xmath229 identifies end - points of leaves of the ending lamination .
further if @xmath229 identifies a pair of points in @xmath279 , then a bi - infinite geodesic having these points as its end - points is a leaf of the ending lamination .
similar descriptions of laminations have been used by bestvina , feighn and handel for free groups @xcite .
using these two descriptions in conjunction gives further information eg about subgroup structure @xcite .
to fix notions consider a finitely generated group @xmath5 acting on a path - metric space @xmath2 . as mentioned in the introduction distortion
arises out of comparing the intrinsic metric on @xmath12 to the metric inherited from the ambient space @xmath2 .
alternately this can be regarded as arising out of comparing filling functions , where one fills a copy of @xmath280 in @xmath12 and @xmath2 and compares the sizes of the chains required .
we will define relative versions of the above two notions .
since the definitions of these invariants require @xmath46connectedness of the spaces we shall assume that whenever these invariants are defined , the spaces in question are quasi - isometric to ( or admit thickenings that are ) @xmath46connected
. it will be clear that one gets quasi - isometry invariants in the process .
reference to an explicit quasi - isometry may at times be suppressed .
distortion of @xmath284 and @xmath285 can be defined in a somewhat more general context .
fix classes @xmath286 and @xmath287 of @xmath46cycles in @xmath2 , @xmath3 respectively ( eg one might restrict to connected cycles or images of spheres ) such that @xmath288 .
let @xmath289 be one be one of the functions @xmath290 or @xmath285 .
define for @xmath296 , @xmath297 the set of maps of the @xmath178sphere @xmath280 and @xmath298 or @xmath299 we get back the original distortion function . note that @xmath299 is approximately half of @xmath300 . for @xmath301 , @xmath302 the set of maps of @xmath303 and @xmath304 we get _ area distortion _ in the sense of gersten @xcite .
distortion has been surveyed in section 2 .
we give a brief sketch of gersten s results on area distortion . in @xcite gersten
shows that in extensions of @xmath306 by finitely presented groups @xmath6 area distortion of @xmath6 is at most an exponential of an isoperimetric function for the extension .
moreover , he describes examples of undistorted ( in the usual sense of length ) subgroups that exhibit area distortion .
he observes further that for torus bundles over the circle with sol geometry , area in the fiber subgroup is undistorted whereas length is exponentially distorted .
gersten showed further that area is undistorted for finitely presented subgroups of finitely presented groups of cohomological dimension 2 . from this
it follows that finitely presented subgroups @xmath5 of hyperbolic groups @xmath6 are finitely presented provided @xmath6 has cohomological dimension 2 or @xmath6 is a hyperbolic small cancellation group @xcite .
the remaining distortion functions are yet to be studied systematically . the first class of examples where @xmath307 seem interesting and tractable are examples coming from extensions of @xmath306 by @xmath308 , ie for @xmath309 where @xmath310 .
such examples have been studied by bridson @xcite and bridson and gersten @xcite .
much less is known about @xmath311 .
these functions are related to topology of balls in groups ( chapter 4 of @xcite ) . for a group @xmath83
admitting a uniformly @xmath26connected thickening @xmath2 ( see @xcite for definitions ) gromov defines @xmath312 to be the infimal radius @xmath313 such that the inclusion of balls @xmath314 is @xmath26connected . , * t brady * , * d cooper * , * v ferlini * , * m lustig * , * m mihalik * , * m shapiro * , * h short * , _ notes on word hyperbolic groups _
, from : `` group theory from a geometrical viewpoint '' , e ghys , a haefliger , and a verjovsky ( editors ) , world scientific ( 1991 ) 363 , _ isoperimetric and isodiametric functions of finitely presented groups _ , from : `` geometric group theory i '' , ga niblo and ma robler ( editors ) , lms lecture notes 182 , cambridge university press ( 1993 ) | this paper is a survey of some of the developments in coarse extrinsic geometry since its inception in the work of gromov .
distortion , as measured by comparing the diameter of balls relative to different metrics , can be regarded as one of the simplist extrinsic notions .
results and examples concerning distorted subgroups , especially in the context of hyperbolic groups and symmetric spaces , are exposed .
other topics considered are quasiconvexity of subgroups ; behaviour at infinity , or more precisely continuous extensions of embedding maps to gromov boundaries in the context of hyperbolic groups acting by isometries on hyperbolic metric spaces ; and distortion as measured using various other filling invariants . |
compressed sensing @xcite is a fundamental idea in mathematics , which utilizes the _ a priori _ property of signal @xmath3 of length @xmath4 being @xmath5 sparse in some domain , where @xmath6 , along with an appropriately constructed sensing matrix @xmath7 , to establish a unique solution for an otherwise undetermined system of linear equations : @xmath8 the actual solution , is the vector from the solution set , which has the minimum @xmath9 norm . since this is a np - hard problem , so we choose the solution which minimizes the @xmath10 norm . it is observed that minimizing @xmath10 norm will give an accurate solution @xcite provided the sensing matrix @xmath7 satisfies the restricted isometry property ( @xmath11 ) property @xcite .
however , if the sparsity of the input signal is not precisely known , but known to lie within a specified range , traditional compressed sensing as such can not exploit this fact and would need to use the same number of measurements for all sparsity values in this range . in this case , the compressed sensing algorithm has to work taking into account the worst case , which corresponds to the signal being least sparse .
for example if the input signal is a discrete - time digital signal of length @xmath4 and can have sparsity anywhere between @xmath12 to @xmath13 in the frequency domain , for compressed sensing to work , one has to design the sensing matrix keeping in mind the sparsity value @xmath14 .
for this case , there are @xmath15 frequencies in the signal which will correspond to @xmath16 complex coefficients ( depending upon the locations of those 25 frequency coefficients ) , it was experimentally observed to take about @xmath1 ( @xmath17 ) measurements for an accurate reconstruction by minimizing the @xmath10 norm . thus if the input signal had sparsity @xmath18 , conventional compressed sensing would take @xmath1 measurements ( since it has been designed for sparsity @xmath14 ) whereas only 40 measurements would have sufficed .
thus , we have unnecessarily used 135 more measurements than needed in this case . in this paper
, we propose a novel method called compressed shattering to address this particular issue .
the central idea of compressed shattering is to adapt compressed sensing to the specified sparsity range by creating shattered signals @xcite which have fixed sparsity using a filter - bank .
our primary aim is to reduce the number of measurements .
the problem is stated as follows . the input signal is a discrete - time digital signal of length
@xmath4 which needs to be sensed .
it is sparse within a range @xmath19 $ ] in the frequency domain , where @xmath20 . here ,
@xmath5-sparse means there are only @xmath5 non - zero coefficients in the @xmath21 of the input signal , without considering the symmetric complex conjugate parts .
the proposed algorithm is described in fig . [ forward_compresss_filtering ] .
first the input signal is permuted .
this results in a permutation of the spectrum ( @xmath21 ) in order to remove any clusters and to spread it out .
the permuted spectrum is then passed through a filter - bank , which is a set of @xmath22 band - pass filters , where @xmath23 .
an inverse permutation operation is done on all the filter outputs to put the spectrum back in its original position .
the compressed sensing algorithm is applied on the output of the filters ; using the same sensing matrix on each of them . depending upon the sparsity level of the original signal
, the filter outputs might be zero or a very sparse signal , and the level of sparsity in the output of each of the filters can be controlled by adjusting the characteristics and number of filters . in the succeeding subsections , we describe each block in detail .
the permutation block performs a mapping operation in which the indices of input signal are rearranged .
it is given by : @xmath24 where @xmath3 is the input discrete - time signal of length @xmath4 and @xmath25 .
it is to be noted that all the operations performed on the indices are modulo @xmath4 operations .
the parameter @xmath26 should be relatively prime to @xmath4 to ensure that the resultant permutation matrix is invertible .
considering @xmath4 to be power of @xmath27 , any odd number belong to set @xmath28 would suffice . permutation done using @xmath26 will ensure that the spectrum of the signal also gets permuted but with @xmath29 as the permutation parameter @xcite , in accordance with the equation : @xmath30 where @xmath31 is the @xmath4-@xmath21 of @xmath3 and @xmath32 and @xmath29 is defined as @xmath33 .
[ permutation_unclustering ] depicts the permutation on an example , it can also be seen that this permutation helps to de - cluster the signal spectrum .
the permuted signal is then passed through a filter - bank of @xmath22 non - overlapping ideal filters .
it should be noted that the filter design and the algorithm that follows in this paper is done by considering @xmath4 to be even and the number of filters @xmath22 divides @xmath34 to give an integer .
the frequency response of the filter banks are : @xmath35[c]{ls } 1,\;\;\;\;\;\ ; & $ \frac{(b-1)\times n}{2\times t}\leq k < \frac{n\times b}{2 \times t}$\\ 1,\;\;\;\;\;\ ; & $ a_1 < k\leq a_2$\\ 0,\;\;\;\;\;\;&elsewere% \end{ieeeeqnarraybox}\right.,\ ] ] where , @xmath36 @xmath37 for @xmath38 . when @xmath39 , we have : @xmath40 , where @xmath41 .
when @xmath42 , we have : @xmath43 out of the @xmath22 filters , only @xmath44 of them will have significant outputs , where @xmath45 .
this filter bank will do a circular convolution as opposed to the normal linear convolution .
it is done to preserve the length of the signal and it will be a perfect element wise multiplication in the fourier domain without having to pad any zeros .
preservation of length is necessary for the inverse permutation block that comes next .
the signal is then passed through it for reversing the permutation operation , thereby putting the spectrum back in its original position .
[ filter_performance_mega ] gives an example of filtering and inverse permutation operations .
the output signals obtained at the end of filtering and inverse permutation are known as _ shattered _
signals @xcite .
as can be seen , the shattered signals are relatively more sparse than the input signal and the sparsity can be controlled by suitably changing @xmath26 and @xmath46 .
it is also possible to obtain shattered signals which are at most 1-sparse as the outputs .
the shattered signals are now ready for compressed sensing . each of the output signals are sensed by the same sensing matrix @xmath7 , designed for specific level of sparsity , which will preferably be much less than the original minimum sparsity @xmath47 of the signal specified in the range . note that more the number of frequencies present in the signal , the less sparse is its spectrum .
although it is possible to obtain different level of sparsity for the shattered signal , in this paper , we have ensured that the shattered signals are all 0 or 1-sparse .
in other words , the filter outputs have at most a single frequency .
hence the number of filters @xmath22 should be at least @xmath47 .
this is sensed by a @xmath48 sensing matrix @xmath7 specifically designed to sense such 1-sparse data , taking into account the symmetry of the @xmath21 . by this
, we ensure that each of the non - zero shattered signals can be sensed in just @xmath27 measurements . in total that will amount to at most @xmath49 measurements . as opposed to use of a sensing matrix with random values , we propose a simple but efficient deterministic sensing matrix @xmath7 .
we make use of the information that shattered signals are either 0 or 1-sparse .
the number of unknowns are just two for each output ( position and value of complex @xmath21 coefficient ) .
we also make use of the fact that the @xmath21 , for real signals , is conjugate symmetric . @xmath50 @xmath51
@xmath52 @xmath53[c]{,c / c / c / c / c / c / c / c , } \cos(\theta_{0})&\cos(\theta_{1})&$ ... $&\cos(\theta_{\frac{n}{2}})&0&0& ... &0\\ \sin(\theta_{0})&\sin(\theta_{1})&$ ...
$&\sin(\theta_{\frac{n}{2}})&0&0& ... &0 \end{ieeeeqnarraybox*}\right),\ ] ] where @xmath54 ( @xmath55 ) , @xmath56 is the signal of length @xmath4 at output of the @xmath57 filter ( @xmath58 ) which has at most a single frequency ( 0 or 1-sparse ) . @xmath59 is the @xmath60 matrix , @xmath7 is the sensing matrix of which any two columns of the first @xmath34 columns of a are linearly independent , @xmath61 is @xmath62 measurement vector which is complex valued .
this sensing matrix will ensure that the sensing happens only for the first half of the spectrum .
further , at most only @xmath44 of the filters will have significant output , namely @xmath63 .
so , only the measurements , @xmath64 ( @xmath65 ) , corresponding to those @xmath44 filters needs to be stored . for this reason
all the measurements @xmath66 , is passed through a threshold block
* th * ( refer to fig . [ forward_compresss_filtering ] ) , where insignificant measurements are discarded by choosing an appropriate threshold for the @xmath67 norm of the shattered signals .
[ cols="^,^,^,^,^,^,^,^,^ " , ] in compressed shattering , since we are using a deterministic sensing matrix @xmath7 as described above , we can make use of the inherent structure in the matrix to design a very fast reconstruction algorithm .
we directly calculate the position and the value of the frequency coefficient of the signal by the following set of equations : ( because @xmath68 ) @xmath69 @xmath70 @xmath71 where @xmath72 represents the complex coefficient and @xmath73 represents the position of the coefficient . from the above equation we can reconstruct the spectrum of the signal in the following way .
( when @xmath74 ) @xmath75[c]{ls } \beta_{j},\;\ ; & $ k = \alpha_{j}$\\ \beta_{j}',\;\ ; & $ k = n-\alpha_{j}$\\ 0,\;\;\;\;\;\;\;\;\;\;\;\;\;\;\;\;\;&elsewere% \end{ieeeeqnarraybox}\right.\;\;\;,\ ] ] when @xmath76 , @xmath75[c]{ls } \beta_{j},\;\ ; & $ k = 0$\\ 0,\;\;\;\;\;\;\;\;\;\;\;\;\;\;\;\;\;&elsewere% \end{ieeeeqnarraybox}\right.,\ ] ] where @xmath77 is the complex conjugate of @xmath72 and @xmath78 is the reconstructed spectrum of the output of the filter @xmath79 .
summing up all respective reconstructed spectrums of the @xmath44 significant filters will give the reconstructed version of the original signal spectrum represented by @xmath80 ( refer to fig .
[ reconstruction_compress_filtering ] ) . @xmath81 @xmath82 by taking the inverse @xmath21 of @xmath83 we get @xmath84 which represents the reconstructed version of the original time domain signal .
to summarize , compressed shattering has four steps in the following order .
the input signal is 1 ) permuted , 2 ) passed through a filter - bank , 3 ) de - permuted , and 4 ) finally sensed by a sensing matrix @xmath7 .
there will be @xmath22 such paths corresponding to @xmath22 filters , however only @xmath44 will be significant ( refer to fig . [ forward_compresss_filtering ] ) . since every block is a linear transformation ( up to the thresholding block ) , we can reduce the entire compressed shattering procedure to one single matrix ( for each of the @xmath22 paths ) .
this is given by : @xmath85 here @xmath56 can be replaced with the following : @xmath86 where @xmath87 is the permutation matrix , @xmath88 is the inverse permutation matrix and @xmath89 is an @xmath90 circular convolution matrix corresponding to the @xmath57 filter .
these matrices can be multiplied to form a single matrix @xmath91 , of size @xmath92 , with complex entries , that takes the input @xmath93 and transforms it into the measurements @xmath61 corresponding to the @xmath57 filter : @xmath94 @xmath95_{2t\times n}\cdot\;\;\;\;\ ; x_{n \times 1}\;\ ; = \left [ \begin{array}{c } y_{1}\\y_{2}\\.\\.\\.\\y_{t}\end{array } \right]_{2t\times 1}.\ ] ]
in this section , we perform numerical simulations to test our proposed algorithm and compare it with conventional compressed sensing .
the parameters for comparison will be number of measurements stored and number of computations .
the input signal to the system is a discrete - time real signal of length @xmath96 and will have sparsity anywhere in the range @xmath12 to @xmath14 frequencies .
we report results for both the extreme cases of sparsity : @xmath97 and @xmath47 .
the input signal and its dft spectrum corresponding to sparsity @xmath98 are shown in fig .
[ input_time_f25 ] and fig .
[ input_spectrum_f25 ] respectively owing to space constraints . ] .
table [ tab : tableresults ] shows the comparison between compressed sensing and compressed shattering in terms of number of measurements to be stored and number of additions and multiplications .
although @xmath99 filters are used in the compressed shattering algorithm ( @xmath26=11 ) , very few shattered signals have significant energy indicating that most of them are 0-sparse . by choosing a threshold of 0.01 for the @xmath100 ,
only very few shattered signals are retained as 1-sparse output signals .
the measurements for compressed shattering are complex values whereas as compressed sensing yields real measurement values . however , in the table we have indicated number of real measurements which implies that we have multiplied the number of measurements for compressed shattering by 2 . in all cases , we obtained near - perfect reconstruction since the maximum absolute reconstruction error was @xmath101 . from the table , we can infer that there is a tradeoff between the number of measurements that have to be stored and the computational complexity involved in taking the initial measurement .
only half the number of real values have to be stored in the case of compressed shattering compared to the conventional compressed sensing method , but the computational complexity of the former is a little more than twice that of the latter in terms of both number of addition and multiplication .
this is the price we pay for the reduction in number of measurements .
it also should be noted that the algorithm , as of now , is heavily dependent on the @xmath26 we choose .
so if we choose the wrong @xmath26 the algorithm might fail because one of the filters might pick up more than one frequency . a plot of number of measurements stored versus the sparsity @xmath5 is shown in fig
. [ measurment_vs_m_n16384 ] , for @xmath102 .
the flexibility of compressed shattering to the sparsity range is evident when compared to traditional compressed sensing and thus results in huge gains , especially when sparsity @xmath5 is small .
we have proposed compressed shattering - a novel way of extending compressed sensing when the sparsity of the input signal is within a specified range .
the idea of using a linear congruential generator on the discrete - time indices helps to randomize the frequency components , and thus in de - clustering the spectrum .
this is then exploited by creating 1-sparse signals by means of a filter - bank .
reconstruction is very fast owing to a simple deterministic sensing matrix that we have proposed for 1-sparse signals .
it is conceivable that a more sophisticated prng could be used to efficiently de - cluster the spectrum .
compressed shattering outperforms traditional compressed sensing in terms of number of measurements that needs to be stored but at the cost of increased computational cost .
future research directions include studying compressed shattering in the presence of noise , finding optimal choices for @xmath26 , an enhanced prng , and a faster algorithm for generating shattered signals . a. c. gilbert , s. muthukrishnan , m. strauss , `` improved time bounds for near - optimal sparse fourier representations , '' _ in proc .
spie wavelets xi _ , m. papdakis , a. f. laine , and m. a. unser , eds .
, san diego , ca , 2005 , pp . | the central idea of compressed sensing is to exploit the fact that most signals of interest are sparse in some domain and use this to reduce the number of measurements to encode .
however , if the sparsity of the input signal is not precisely known , but known to lie within a specified range , compressed sensing as such can not exploit this fact and would need to use the same number of measurements even for a very sparse signal . in this paper , we propose a novel method called compressed shattering to adapt compressed sensing to the specified sparsity range , without changing the sensing matrix by creating shattered signals which have fixed sparsity .
this is accomplished by first suitably permuting the input spectrum and then using a filter bank to create fixed sparsity shattered signals . by ensuring that all the shattered signals are utmost 1-sparse , we make use of a simple but efficient deterministic sensing matrix to yield very low number of measurements . for a discrete - time signal of length 1000 , with a sparsity range of @xmath0 ,
traditional compressed sensing requires @xmath1 measurements , whereas compressed shattering would only need @xmath2 measurements . |
on february 15 , 2013 , at 9:22 am , an exceptionally bright and long - duration fireball was observed by many eyewitnesses in the chelyabinsk region , russia @xcite . a large - sized object with a relatively low mass - loss rate and shallow atmospheric entry angle led to a very long trail .
the fireball was observed in the chelyabinsk , kurgan , orenburg , tyumen , ekaterinburg , kostanay , and aktobe regions , and in the republic of bashkortostan , russia , and in kazakhstan .
the event was recorded by numerous video cameras from the ground and was also imaged from space by the meteosat and fengyun satellites @xcite .
a strong shock wave associated with the fireball caused significant damage including broken windows and partial building collapses in chelyabinsk and the surrounding territories .
two days later the first fragments of the chelyabinsk meteorite were reported to be found around pervomaiskoe , deputatsky , and yemanzhelinka , located approximately 40 km south of chelyabinsk .
successful search campaigns in this area were organized by the scientists from the ural federal university as well as by the laboratory of meteoritics of the vernadsky institute , russian academy of science .
numerous meteorite samples were also recovered and collected out of snow by local residents before a subsequent extensive snowfall .
the majority of fragments are composed of relatively small pieces , with the largest officially reported being 3.4 kg as of this publication .
the largest piece in the ural federal university collections is a 1.8 kg sample no .
a50 ( fig .
[ fig : one ] ) .
a large amount of additional material was collected on the ground along the fireball track after the snow melted at the end of april and in may 2013 .
the total collected mass of the chelyabinsk meteorites exceeds 100 kg .
as more meteorite samples were collected , it became apparent that multiple types of material are present .
the recovered meteorites are composed of either light - colored or dark - colored ( shock - darkened ) lithology @xcite . additionally , dark impact - melt is being present in varying amounts within light- and dark - colored stones .
impact - melt is distinct in appearance from the dark - colored ( shock - darkened ) lithology and is distinguished as a third lithology type .
some meteorites contain significant portion of this impact - melt lithology ( fig .
[ fig : two ] ) .
however , the light- and dark - colored ( shock - darkened ) lithologies were ( up to our knowledge ) not found together within one meteorite . according to the computations by yeomans and chodas @xcite , the initial diameter of the meteoroid ( or a small asteroid ) was approximately 1720 m with a mass of approximately @xmath0 kg .
this object is the largest observed celestial body colliding with the earth since the tunguska event in 1908 . in this study , we report physical properties ( bulk and grain density , porosity , and magnetic susceptibility ) of 49 individual chelyabinsk meteorites from the ural federal university collections , ekaterinburg , russian federation
. detailed mineralogical and spectral characterization of two samples ( representing the light - colored and dark - colored lithologies , both with impact - melt veins ) is presented to determine and explain the differences between the lithologies . finally , the origin of the chelyabinsk meteorites and their parent body history are briefly discussed .
the mineral composition was determined at the department of geosciences and geography , university of helsinki , with a jeol jxa-8600 superprobe using standard - calibrated energy - dispersive spectrometry ( eds ) on a nitrogen - cooled si(li ) detector .
two samples ( the light - colored and dark - colored lithologies ) were polished and carbon - coated prior to the measurements .
matrix correction was performed with the xmas software using the pap correction method .
a combination of natural and synthetic silicate and oxide standards was used for calibration .
the accuracy was determined by cross - analyses of standard materials and was better than 2% for the major elements ( > 10 wt% ) and better than 5% for the minor elements ( 110 wt% ) .
oxygen was calculated by stoichiometry .
the analysis conditions were 15 kv accelerating voltage , 1 na beam current and 100 s acquisition time per spot with a focused beam .
altogether , 116 spots were analyzed , 58 in the light - colored lithology sample ( no .
vg1 ) and 58 in the dark - colored lithology sample ( no .
vg4b ) . because only major and minor elements were analyzed , the eds method was considered accurate enough and faster than the commonly used more accurate wavelength - dispersive spectrometry ( wds ) method in quantitative microanalysis .
subsequently , two polished thin sections from these two samples ( 27 mm @xmath1 15 mm from the light - colored lithology sample no .
vg1 and 12 mm @xmath1 14 mm from the dark - colored lithology sample no .
vg4b ) were prepared .
texture , mineralogical characteristics , and shock features were studied using a leica dmlp petrographic microscope at the czech geological survey .
the spectral characterization was carried out using an ol 750 automated spectroradiometric measurement system by gooch & housego at the department of physics , university of helsinki .
the spectroradiometer is equipped with a quartz tungsten - halogen light source , a monochromator , an integrating sphere with a polytetrafluoroethylene ( ptfe ) coating , and two high sensitivity detectors .
the incident beam has an angle of 10@xmath2 with the sample holder normal vector .
there is a light trap for specular reflection at the mirror angle .
the detector is positioned so that there is no direct reflection from the sample to the detector and all the detected signal is multiply scattered in the ptfe coated integrating sphere .
the system is closed during the measurement and there is no stray light .
the detectors use either a silicon detector head ( for wavelengths from 0.21.1 ) or a cooled pbs detector head ( for wavelengths up to 3.2 ) . in the diffuse spectral measurement procedure with the spectroradiometer , the spectra of the ptfe coating in the integrating sphere
is always measured before the actual target while the target is already placed in the sample holder .
by comparing these reflectances and correcting with the known calibration values of the ptfe material , the reflectance spectra , normalized against an ideal diffuser , are obtained .
the reflectance of the light - colored ( sample no .
vg1 ) , dark - colored ( sample no .
vg4b ) , and impact - melt ( vein in the sample no .
m18 ) lithologies was acquired in the wavelength range of 5002500 nm from saw - cut ( vg1 and vg4b ) or broken ( m18 ) surfaces .
the surfaces were rough enough preventing a significant direct specular reflection and were identical to those used for mineralogical characterization .
the size of the incident light beam footprint on the target can be varied using slits and apertures in the monochromator .
the light - colored lithology sample ( no .
vg1 ) had a slightly larger surface area available for the spectral measurement , and therefore a 5 mm slit and a 5 mm aperture were used for this target .
thus , the resulting collimated beam footprint on the target had a diameter of 5 mm .
the dark - colored lithology sample ( no . vg4b ) and impact - melt lithology ( vein in sample no .
m18 ) were measured with a 2.5 mm slit and a 3 mm aperture , and the beam footprint on the target was approximately 2.5 mm .
the density and porosity values were obtained using the mobile laboratory described in @xcite deployed at the ural federal university , ekaterinburg .
the bulk volume was determined using a modified archimedean method @xcite incorporating glass beads @xmath3 mm in diameter .
ten sets of measurements per sample were carried out . the method was thoroughly tested and calibrated prior to the measurements using volume standards and the absolute resolution and precision were determined to be @xmath4 @xmath5 .
the grain volume of the meteorites was measured using a quantachrome ultrapyc 1300e he pycnometer .
the absolute resolution and the precision of this device are better than @xmath6 @xmath5 .
thus , the relative errors of both volumetric methods increase with decreasing sample size .
however , compared to the bulk volume , the grain volume is determined with higher resolution and precision .
meteorite masses were determined using a digital ohaus navigator scale with 0.1 g resolution and precision .
the scale was calibrated prior to the measurements using internal calibration .
the magnetic susceptibility of the samples smaller than 2.5 cm was measured using a zh instruments sm-100 susceptibility meter operating at 8 khz frequency and 320 a / m rms field amplitude . for larger samples ,
a zh instruments sm-30 portable surface susceptibility meter with a large 5 cm coil operating at the same 8 khz frequency was used .
the values were corrected for the actual sample size following the procedure described in gattacceca et al .
@xcite .
the susceptibility of the samples was measured three times along three perpendicular directions .
subsequently , a logarithm of the average apparent magnetic susceptibility ( in @xmath7 m@xmath8/kg ) was calculated as described in rochette et al . @xcite . the relative error in the determined value of the magnetic susceptibility logarithm
is below 3% . for the statistical calculations of the density , porosity , and susceptibility ,
only samples larger than 5 g were considered to be homogeneous enough to be representative of the whole - rock material .
figure [ fig : three ] shows the overall optical and detailed sem images of the light - colored ( no . vg1 ) and dark - colored ( no .
vg4b ) lithology meteorites . the computed olivine fayalite ( fa ) content and orthopyroxene ferrosilite ( fs ) content determined by means of eds ( fig .
[ fig : four ] ) from the main meteorite compounds ( matrix and chondrules , excluding melt veins ) reveal that the light - colored and dark - colored lithologies are almost identical in composition and , based on the classification scheme in brearley and jones ( @xcite , p. 38 ) , can be classified as ll ordinary chondrites . the homogeneity of olivine and pyroxene as indicated by a standard deviation of the fa and fs compositions ( standard deviation @xmath9 5% ) , texture showing recrystallized matrix with readily distinguished chondrules , and an absence of igneous glass in the chondrules are typical for petrographic type 5 .
abundant impact - melt veins are found in both light - colored and dark - colored lithologies .
their silicate composition is almost identical to the surrounding meteorite material pointing again on their common source .
average , @xmath10 median , @xmath11 standard deviation , @xmath12 number of analyzed spots . ] as revealed by optical microscopy and sem imaging , the difference between the light - colored and dark - colored lithologies is in the level of shock .
the light - colored lithology is slightly fractured and represents monomict ll5 breccia ( fig .
[ fig : three ] ) .
individual clasts show slightly different shock features indicating a different intensity of shock metamorphism .
the shock features observed in olivine for individual clasts vary from undulose extinction to strong mosaicism which is indicative of shock levels s3 and s4 @xcite .
the silicate grains show microscopic fracturing .
the large inter - granular cracks are filled with dark melt veins containing also fine - grained mineral fragments .
the silicate composition of this melt is close to that of the surrounding material .
thus , we interpret this melt as an impact - melt originating from the light - colored lithology .
the dark - colored lithology is shocked to a larger extent .
the material is extensively fractured and abundant impact - melt is present ( fig .
[ fig : three ] ) .
two populations of melt exist within the dark - colored lithology .
the inter - granular space is filled with impact - melt veins containing fine - grained mineral fragments ( first population ) similar to impact - melt observed in the light - colored lithology .
the silicate composition of these melt veins also closely resembles the main meteorite material with the additional presence of clinopyroxene .
additionally , the inter - granular as well as intra - granular fractures are impregnated with a dense network of thin , fine - grained melt veins composed predominantly of troilite and metal ( the second melt population ) .
this metallic melt is abundant through the dark - colored lithology and is the main darkening agent responsible for its optical appearance .
the abovementioned findings are similar to these in independent study by galimov et al .
@xcite .
the spectrum of the light - colored lithology shows broad 1 and 2 absorption bands of olivine and pyroxene ( fig .
[ fig : five ] ) .
the overall normalized reflectance is between 0.10 and 0.14 .
the reflectance of the dark - colored lithology is lower ( reflectance in the range of 0.060.08 ) and the 1 and 2 absorption bands are almost indistinguishable in the spectrum .
the reflectance of the impact - melt is being intermediate between the light - colored and dark - colored lithologies .
the results of the density and porosity measurements of the chelyabinsk meteorites are summarized in tables [ tab : one ] and [ tab : two ] .
the bulk density of the individual meteorites ranges from 3.14 to 3.48 g/@xmath5 with the mean value of 3.32 g/@xmath5 and with s.d .
( standard deviation ) of 0.09 g/@xmath5 ( light - colored lithology , 18 samples included ) , and 3.27 g/@xmath5 s.d .
0.08 g/@xmath5 ( dark - colored , 7 samples included ) .
the grain density ranges from 3.22 to 3.60 g/@xmath5 with the mean value of 3.51 g/@xmath5 s.d .
0.07 g/@xmath5 ( light - colored , 14 samples included ) and 3.42 g/@xmath5 s.d .
0.10 g/@xmath5 ( dark - colored , 7 samples included ) .
the dark - colored lithology appears slightly less dense .
however , the light - colored and dark - colored meteorite density ranges significantly overlap and neither bulk nor grain density can be used as a reliable criterion to distinguish between the two lithologies .
porosity is almost identical for both lithologies ranging from 1.5 to 11.4% with the mean value of 6.0% s.d .
3.2% ( light - colored , 13 samples included ) and 5.7% s.d .
1.7% ( dark - colored , 6 samples included ) .
.comparison of the mean bulk and grain density , porosity , and magnetic susceptibility together with their standard deviations ( s.d . ) of the chelyabinsk light - colored and dark - colored lithologies . [ cols="<,<,<",options="header " , ] the measured physical properties are in good agreement with the bulk ( average 3.22 g/@xmath5 s.d
. 0.22 g/@xmath5 ) and grain ( 3.54 g/@xmath5 s.d .
0.13 g/@xmath5 ) density and porosity ( 8.2% s.d .
5.5% ) of other ll chondrite falls reported in consolmagno et al .
@xcite and no correlation with the meteorite masses is observed ( fig .
[ fig : six ] ) .
smaller samples show higher density and porosity scatter .
this is most likely due to a higher relative uncertainty in the measured volume values of the smaller samples ( as described in the materials and methods section ) and increasing inhomogeneity of the material at smaller scales . ) and porosity ( @xmath13 ) of the chelyabinsk meteorites as a function of their mass .
no correlation is observed . ]
the results of the susceptibility measurements of the chelyabinsk meteorites are summarized in tables [ tab : one ] and [ tab : two ] .
the logarithm of the apparent magnetic susceptibility ( in 10@xmath14 m@xmath8/kg , fig .
[ fig : seven ] ) ranges from 4.33 to 4.79 with the mean value of 4.49 s.d .
0.07 ( light - colored lithology , 20 samples included ) and 4.52 s.d .
0.15 ( dark - colored , 8 samples included ) .
this is within s.d . of the results from an independent study by bezaeva et al .
@xcite on samples from the collections at the vernadsky institute , russian academy of sciences , moscow .
however , the magnetic susceptibility values are higher compared to the other ll chondrite falls ( average 4.10 s.d .
0.30 ) reported in rochette et al .
@xcite and are closer to the transitional l / ll class ( 4.60 s.d .
0.11 @xcite ) .
the susceptibility values are quite uniformly distributed among samples of various masses ( fig .
[ fig : eight ] ) . as magnetic susceptibility in ordinary chondrites
mainly reflects the amount of ferromagnetic metallic iron , it is apparent that , compared to typical ll chondrites , the chelyabinsk meteorites are enriched in metallic iron . in 10@xmath14 m@xmath8/kg ) of the chelyabinsk meteorites as a function of their mass .
no correlation is observed . ]
the dark - colored lithology shows slightly higher magnetic susceptibility ( similarly to that reported by bezaeva et al .
@xcite ) and thus is slightly richer in metallic iron content . however , similarly to the density values , the light - colored and dark - colored material magnetic susceptibility ranges significantly overlap and susceptibility can not be used as a reliable criterion to distinguish between the two lithologies .
the chelyabinsk meteorite fragments represent fresh ll5 type chondrite material originating from a small , @xmath15 m sized meteoroid ( small asteroid ) .
the meteorite samples are of petrographic type 5 and were subjected to recrystallization at approx .
this thermal metamorphism evidence suggests that their parent body must have been originally much larger .
therefore the meteoroid , which collided with the earth and delivered the chelyabinsk meteorites , was only a fragment of the ll chondrite parent body .
moreover , the chelyabinsk meteorites are composed of three lithologies : the light - colored , dark - colored , and impact - melt .
the mineralogical analysis revealed that these lithologies are derived from the same ll5 chondritic source .
the difference between them is in the shock .
this gives us a unique opportunity to study effects of high shock on the mineralogy , reflectance spectra , and physical properties of chondritic material .
the light - colored lithology shows lower level of fracturing and the impact - melt lithology is localized to inter - granular space only .
in contrast , the dark - colored lithology is shocked up to higher shock level causing higher level of fracturing .
in addition to the presence of silicate impact - melt lithology similar to the one in the light - colored lithology , another metal and sulfide - rich melt population impregnating the inter- as well as intra - granular space is present .
this metal and sulfide - rich melt is the darkening agent .
the silicate impact - melt lithology rather resembles the characteristics of whole - rock melt and it is present in both light- and dark - colored lithology meteorites .
some meteorites are even dominated by the impact - melt and thus it is distinguished as a third lithology type and should be not confused with the dark - colored lithology .
in contrast , the metal and sulfide - rich melt present in dark - colored lithology is rather partial melt and appears uniformly distributed .
the higher shock level observed in the dark - colored lithology can be explained by a single shock event identical to the one in the light - colored lithology , only being stronger due to the darkened material being closer to the impact point .
alternatively , increased shock darkening can be caused by a secondary event affecting a part of the parent body only .
both shock scenarios seem to be too strong to occur on the @xmath15 m chelyabinsk meteoroid .
thus , the shock most likely took place earlier on the parent body .
it is possible that the last high - shock impact event responsible for the darkening and other shock features observed in both lithologies may have also caused the disintegration of the parent body and the origin of the chelyabinsk meteoroid .
most likely , numerous smaller asteroid fragments similar to the chelyabinsk meteoroid body were created during this event @xcite .
the spectrum of the light - colored lithology is typical for an ll ordinary chondrite with the presence of 1 and 2 olivine and pyroxene absorption bands .
such a spectrum is similar to fresh s or q type asteroids .
in contrast , the impact - melt and dark - colored ( shock - darkened ) lithology spectra are darker and the silicate absorption bands are almost invisible .
this characteristic is similar to those reported earlier on other shock - darkened meteorites ( e.g. @xcite ) . due to these characteristics
it was suggested earlier ( e.g. @xcite ) that shock - darkened chondritic material may be present on the surfaces of some dark asteroids ( e.g. c type ) and our results support this suggestion . compared to impact melting
, shock darkening seems to be a more efficient process in altering the reflectance spectra and causes small slope change .
shock also affects the albedo and silicate absorption bands of chondrite material in a way similar to space weathering .
however , unlike space weathering ( or at least unlike its lunar - type form with the presence of nanophase iron in the surface coatings of silicate minerals ) there is negligible ( impact - melt lithology ) to minor ( dark - colored lithology ) change in the spectral slope observed .
the bulk and grain densities and the porosity of the light - colored and dark - colored lithologies are almost identical .
the dark - colored lithology has a slightly smaller average grain density of 3.42 g/@xmath5 compared to 3.51 g/@xmath5 for the light - colored one .
however , the slight difference between the two lithologies is smaller compared to the s.d . of these values .
the dark - colored lithology is more fractured .
thus , one would expect a significantly lower bulk density and higher porosity .
however , the fractures are filled by melt veins .
this explains the observed negligible density differences between the light - colored and dark - colored lithologies .
similarly , the magnetic susceptibility logarithm is slightly higher ( pointing on a slightly higher amount of metal related to the presence of troilite and metal - rich impact - melt veins ) for the dark - colored lithology , but the difference is again within the s.d . and the susceptibility ( as well as the density and porosity ) values can not be used to distinguish between the two lithologies .
it was not possible to measure densities , porosity and magnetic susceptibility of impact - melt lithology because it appears only in association with light - colored and dark - colored lithology .
however , values measured on meteorites with significant portion of impact - melt lithology ( e.g. meteorite no . a2 ,
table [ tab : one ] ) do not deviate from others .
thus , despite lithological variety in meteorite material , the chelyabinsk meteorite shower is rather homogeneous in its physical properties , in resemblance to other meteorite showers reported in consolmagno et al .
the chelyabinsk fireball with an almost instant recovery of a large number of fresh meteorites is one of the most spectacular meteorite fall events in the recent history .
the recovered meteorite material was characterized as brecciated ll5 ordinary chondrite .
based on the magnetic susceptibility being in the intermediate range between ll and l chondrites , the chelyabinsk meteorites are richer in metallic iron as compared to other ll chondrites .
the measured bulk and grain densities and the porosity closely resemble other ll chondrites .
three lithologies , the light - colored , dark - colored , and impact - melt lithologies , were found within the recovered meteorites .
the impact - melt is silicate - rich and is present within the light - colored and dark - colored lithology stones as dark inter - granular veins . in the dark - colored lithology , additional fine - grained metal and sulfide - rich melt forming a dense network of fine veins impregnating the inter- and intra - granular pore space
is present . this metal and sulfide - rich melt
is related to a higher shock experienced by the dark - colored lithology and is the main darkening agent .
the presence of the ll5 lithologies of distinct shock levels enables us to study shock - generated changes in the spectral and physical properties of chondritic material . both impact - melt generation and shock darkening cause a decrease in reflectance and a suppression of the silicate absorption bands in the reflectance spectra .
shock darkening seems to be a more efficient process in altering the reflectance spectra .
such spectral changes are similar to the space weathering effects observed on asteroids . however , space weathering of chondritic materials is often accompanied with significant spectral slope change ( reddening ) . in our case , only negligible ( impact - melt lithology ) to minor ( dark - colored lithology ) change in the spectral slope is observed .
thus , it is possible that some dark asteroids with invisible silicate absorption bands may be composed of relatively fresh shock - darkened chondritic material .
the main spectral difference of asteroid surfaces dominated by impact - melt or shock darkening , or space weathering , is a significant slope change ( reddening ) in the latter case .
shock darkening does not have a significant effect on the material physical properties .
the increased shock causes fracturing with expected porosity increase and bulk density decrease .
however , this is compensated by impregnation of the fractures by fine - grained sulphide and metal - rich melt having an opposite effect on the porosity and bulk density .
the presence of sulphide and metal - rich melt is responsible for the slight increase of the magnetic susceptibility observed in the dark - colored lithology .
the work was supported by academy of finland projects no .
257487 and no . 260027 ( sironta ) ,
ministry of education , youth and sports czech republic grant lh12079 , federal grant - in - aid program gc no .
14.740.11.1006 , and the erc advanced grant no .
320773 ( saempl ) .
authors would like to thank to ilya weinstein for help with the measurements arrangements , to pasi heikkil , alevtina maksimova , razilya gizzatullina , albina zainullina , and anastasia uryvkova for help with the laboratory work , and to hanna pentikinen for language corrections , and to to three anonymous reviewers for constructive comments on the manuscript .
galimov , e. m. , kolotov , v. p. , nazarov , m. a. , kostitsyn , yu .
a. , kubrakova , i. v. , kononkova , n. n. , i. a. roshchina , v. a. alexeev , l. l. kashkarov , d. d. badyukov , and v. s. sevastyanov ( 2013 ) .
analytical results for the material of the chelyabinsk meteorite .
_ geochemistry international _ * 51*:580598 .
doi : 10.1134/s0016702913070100 yeomans d. and chodas p. ( 2013 ) .
additional details on the large fireball event over russia on feb .
15 , 2013 .
nasa / jpl near - earth object program , http://neo.jpl.nasa.gov/news/fireball_130301.html , read on 25.6.2013 kohout , t. , kletetschka , g. , elbra , t. , adachi , t. , mikula , v. , pesonen , l.j .
, et al . (
physical properties of meteorites applications in space missions to asteroids .
_ meteoritics & planetary science _ * 43*:10091020 .
doi : 10.1111/j.1945 - 5100.2008.tb00689.x macke r. j. , britt d. t. , and consolmagno g. j. ( 2010 ) .
analysis of systematic error in bead method
measurements of meteorite bulk volume and density . _ planetary and space science _ * 58*:421426 .
doi : 10.1016/j.pss.2009.11.006 rochette p. , sagnotti l. , bourot - denise m. , consolmagno g. , folco l. , gattacceca j. , osete m. l. , and pesonen l. j. ( 2003 ) .
magnetic classification of stony meteorites : 1 .
ordinary chondrites .
_ meteoritics & planetary science _ * 38*:251268 bezaeva , n. s. , badyukov , d. d. , nazarov , m. a. , rochette , p. , and feinberg , j. ( 2013 ) .
magnetic properties of the chelyabinsk meteorite : preliminary results .
_ geochemistry international _ * 51*:629635 .
doi : 10.1134/s0016702913070082 de la fuente marcos c. and de la fuente marcos r. ( 2013 ) .
the chelyabinsk superbolide : a fragment of asteroid 2011 eo40 ? _ monthly notices of the royal astronomical society _ ( in press ) .
doi : 10.1093/mnrasl / slt103 britt , d. t. , pieters , c. m. , petaev , m. i. , and zaslavaskaya , n. i. ( 1989a ) . the tsarev meteorite petrology and bidirectional reflectance spectra of a shock - blackened l chondrite . in : lunar and planetary science conference ,
19th , houston , tx , mar . 1418 , 1988 , proceedings ( a89 - 36486 1591 ) .
cambridge / houston , tx , cambridge university press / lunar and planetary institute , 1989 , p. 537545
britt , d. t. and pieters , c. m. ( 1994 ) .
darkening in black and gas - rich ordinary chondrites : the spectral effects of opaque morphology and distribution .
_ geochimica et cosmochimica acta _
* 58*:39053919 .
doi : 10.1016/0016 - 7037(94)90370 - 0 britt , d. t. , pieters , c. m. , webb , r. s. , and pratt , s. f. ( 1989b ) .
relationship of c - type asteroids to dark meteorites : evidence for optical alteration by asteroidal regolith processes .
abstracts of the lunar and planetary science conference 20:111112 consolmagno g. j. , macke r. j. , rochette p. , britt d. t. , and gattacceca j. ( 2006 ) .
density , magnetic susceptibility , and the characterization of ordinary chondrite falls and showers .
_ meteoritics & planetary science _ * 41*:331342 | the mineralogy and physical properties of chelyabinsk meteorites ( fall , february 15 , 2013 ) are presented .
three types of meteorite material are present , described as the light - colored , dark - colored , and impact - melt lithologies . all
are of ll5 composition with the impact - melt lithology being close to whole - rock melt and the dark - colored lithology being shock - darkened due to partial melting of iron metal and sulfides .
this enables us to study the effect of increasing shock on material with identical composition and origin .
based on the magnetic susceptibility , the chelyabinsk meteorites are richer in metallic iron as compared to other ll chondrites .
the measured bulk and grain densities and the porosity closely resemble other ll chondrites .
shock darkening does not have a significant effect on the material physical properties , but causes a decrease of reflectance and decrease in silicate absorption bands in the reflectance spectra .
this is similar to the space weathering effects observed on asteroids .
however , compared to space weathered materials , there is a negligible to minor slope change observed in impact - melt and shock - darkened meteorite spectra .
thus , it is possible that some dark asteroids with invisible silicate absorption bands may be composed of relatively fresh shock - darkened chondritic material .
+ cite as : t. kohout et al .
mineralogy , reflectance spectra , and physical properties of the chelyabinsk ll5 chondrite insight into shock induced changes in asteroid regoliths .
icarus * 228 * , 7885 , 2014 , doi:10.1016/j.icarus.2013.09.027 . |
the dynamical interaction between newly formed massive stars , their h@xmath9ii regions , and the surrounding neutral material has been a topic of great interest . based on simple ideas ,
many have argued that h@xmath9ii regions should expand quickly into the ambient molecular cloud , expanding beyond the ultra - compact phase in 10@xmath1010@xmath11 yr ( e.g. dreher & welch , 1981 , dreher et al .
it has been recognized that a lifetime paradox exists for ultra - compact h@xmath9ii regions , in that there are far too many given the current birth rate of massive stars ( wood & churchwell , 1989 ) .
numerous authors have suggested resolutions for the h@xmath9ii region lifetime paradox ( see review of hollenbach et al .
1994 ) : a ) confinement by infalling material ( e.g. reid et al .
1981 ) , b ) champagne flows ( e.g. tenorio tagle , 1979 ) , c ) moving star bow shock ( e.g. van buren et al . 1990 ) , d ) disk photoevaporation model ( hollenbach et al . 1994 ) , and e ) mass loaded winds ( dyson et al . , 1995 ; lizano et al . , 1996
these models have been developed to various degrees but all exhibit serious shortcomings when it comes to matching current observational data .
de pree et al .
( 1995 ) have suggested a simple resolution for the lifetime paradox . revisiting the lifetime arguments , but assuming confinement by hot ( 100@xmath9k ) and high density ( 10@xmath12@xmath6 ) ambient material , de pree et al .
( 1995 ) find that the ultra - compact h@xmath9ii region phase can persist for @xmath5yr . to study the interfaces between the h@xmath9ii regions and the surrounding material
, one must find the appropriate tracers .
the usual high density tracers such as cs are not useful because they are not specific to the interfaces of the h@xmath9ii regions : there are also high column densities of this molecule in hot cores . from theoretical arguments and observations
it is known that the abundances of some molecules are enhanced by uv radiation from the h@xmath9ii region .
based on the results of the photon dominated regions ( pdrs ) in ngc@xmath92023 and ngc@xmath97023 ( fuente et al .
1993 , 1995 ) and on pdr chemical models ( sternberg and dalgarno 1995 , jansen et al .
1995 ) , we have selected cn as one of the best tracer of the material at the h@xmath9ii / h@xmath4 boundary layer affected by the uv photons from the ob stars .
franco et al . (
1998 ) have used the 30-m telescope to map the @xmath1 and the @xmath3 lines of cn towards m@xmath942 and m@xmath943 . modeling the cn emission , rodrguez
franco et al . (
1998 ) have produced h@xmath4 density maps in the orion m@xmath942 and m@xmath943 regions .
the cn emission reveals a number of high density ( @xmath13@xmath6 ) bars ( north bar , optical bar , molecular filaments , and molecular ridge ) that surround the trapezium cluster .
the cn bars represent the interfaces between the molecular cloud and the major ionization fronts of m@xmath942 , confining this h@xmath9ii region in basically all directions except along the line of sight .
surprisingly , the largest h@xmath4 densities , @xmath14@xmath6 are found towards the trapezium region where the emission from co is relatively weak .
this cn emission from the trapezium arises from the thin layer of molecular gas that is related with the material that interacts with the rear ionization fronts of m@xmath15 confining the h@xmath9ii region in this direction .
the cn layer is likely to be the remnant of the high density layer ( @xmath16@xmath6 ) that also confined the h@xmath9ii region in the ultra- and hyper - compact phase . the critical densities to excite the @xmath3 and @xmath1 lines of cn are @xmath17@xmath6 .
therefore from the analysis of these transitions one can not constrain the h@xmath4 densities if they are larger than @xmath18@xmath6 . to estimate the h@xmath4 densities in the trapezium region one needs to measure a transition with higher einstein coefficients . in this paper , we present observations of the cn @xmath0 transition taken with an angular resolution similar to that of the @xmath3 line and analyze the new results combined with 30-m data .
the new data show that the h@xmath4 densities in the trapezium layer are @xmath19@xmath6 .
the observations of the @xmath0 and the @xmath1 lines of cn , at @xmath20ghz and @xmath21ghz respectively , and the @xmath1 of @xmath2cn at @xmath22ghz were taken between 27 april and 3 may 1999 and april 2000 with the heinrich hertz telescope ( hht ) in arizona ( usa ) . at the rest frequencies of the @xmath0 and the @xmath1 , the full width to half power of the telescope was 26@xmath23 and 34@xmath23 respectively .
the pointing was checked by continuum cross scans of venus .
the rms pointing error was always less than @xmath243@xmath23 .
the 345@xmath9ghz receiver is a facility instrument consisting of a dual channel superconducting mixer provided by the max plank institut fr radioastronomie ( mpifr ) .
the double sideband receiver noise temperatures were typically @xmath24150 - 200@xmath9k and @xmath24100@xmath9k for the 345@xmath9ghz and 230@xmath9ghz receivers respectively . at the elevation of orion ,
the single sideband system noise temperature was typically @xmath24800 to 1400@xmath9k for the 345@xmath9ghz receiver and @xmath25 to @xmath26k for the 230@xmath9ghz receiver .
the data were calibrated using the chopper wheel method .
we have used position switching , with the reference at ( @xmath27 , 0@xmath23 ) from the on- source position .
we established the final temperature scale by comparisons of our orion - kl results with the data of groesbeck et al .
( 1994 ) and schilke et al .
( 1997 ) . for some tunings of the 345@xmath9ghz receiver
, there is a frequency - dependent variable ratio between the signal and image sidebands with a frequency of @xmath24500@xmath9mhz . to check for the presence of such variability
, we shifted each spectrum five times by up to @xmath28km@xmath9s@xmath29 .
this also allowed us to measure all the hyperfine components with the narrow band spectrometers , and to search for the presence of intense lines in the image spectrum .
the spectra were analyzed using filter banks and acoustic optical spectrometers ( aos s ) .
the wide band aos s with a bandwidth of 983@xmath9mhz had a resolution of 480@xmath9khz and the narrow band aos s with a bandwidth of 64@xmath9mhz has a resolution of 250@xmath9khz .
the @xmath30 , @xmath31 , and @xmath32 hyperfine components of the @xmath0 , @xmath33 multiplet of cn are separated by 790@xmath9mhz .
this range of frequencies was in the analyzing band of the wide - band aos s .
for the @xmath1 line of cn the @xmath34 , @xmath35 , @xmath36 , @xmath37 , @xmath38 , and @xmath39 hf components of the fine @xmath40 transition was observed with the 64@xmath9mhz narrow band aos .
the observations were made toward selected positions in the ridge and in the trapezium region obtained from the cn @xmath3 line map of rodrguez
franco et al .
[ fig : mapa ] shows superimposed on the cn @xmath3 map from rodrguez franco et al . (
1998 ) the positions in the trapezium region ( t1 t6 ) and in the molecular ridge ( r1 r6 ) where we took the spectra in the @xmath0 line of cn .
the @xmath1 line of cn and @xmath2cn were measured toward the positions r4 and r5 .
[ fig : espec ] shows the spectra of the @xmath0 line toward all the observed positions , and fig .
[ fig : iso ] the @xmath1 spectra of cn and @xmath2cn towards r4 and r5 .
table [ tab : obs ] gives the observational parameters derived from gaussian fits to the @xmath30 , @xmath31 , and @xmath32 hyperfine ( hereafter hf ) components of the @xmath0 , @xmath33 multiplet .
table [ tab : obs ] gives also the observed parameters of the @xmath1 line of cn and @xmath2cn the line towards r4 and r5 ( see next section for a discussion ) .
usually one can use the intensity ratio between the fine and hf components of the cn transitions to estimate the optical depth of the lines ( see e.g. rodrguez franco et al .
this method is widely used for molecules like nh@xmath41 ( pauls et al .
1983 ) and gives reliable results when all the hf components have the same excitation temperature . in some cases , like for nh@xmath41 and cn , it is well known that the different hf components might have different excitation temperatures ( gaume et al .
we now discuss the problems associated with the use the hf ratios to estimate the cn line opacities for the submillimeter lines of cn .
the @xmath3 rotational transition results in 9 hf components grouped in two main fine - structure groups , the @xmath1 and the @xmath0 transitions are split in 18 hf lines which are grouped in three fine - structure groups , and the @xmath2cn @xmath1 line is split into 71 hf components . in the wide band aos we have observed the two most intense fine structure groups of the @xmath0 line simultaneously .
we have also measured the @xmath40 fine structure group in the @xmath1 line .
we have derived the hf ratio for the observed cn lines by fitting a `` comb '' of gaussian profiles centered at the frequencies and the relative intensities of the hf components for every cn line .
fig [ fig : iso ] shows the fits of the hf components for the cn and @xmath2cn @xmath1 line in the two positions of the ridge .
similar good fits are also obtained for the cn @xmath0 hf components in fig .
[ fig : espec ] . for both the ridge and the trapezium positions ,
we derive a ratio between the @xmath33 and the @xmath42 fine components of the @xmath0 line ( rf32 hereafter ) close to @xmath43 ( see table [ tab : obs ] ) , which would indicate optically thin emission .
this is surprising in view of the relatively strong cn @xmath0 lines in the ridge and the detection of the @xmath2cn @xmath1 line towards two positions in the ridge .
these facts indicate that the cn lines in the ridge can not be optically thin in contrast to the rf32 indicating optically thin emission .
these discrepancies are due to different excitation temperatures for different hf components . in order to study to what extent
the assumption of equal excitation temperature for all the hf components can be applied to the @xmath0 line to derive the opacities , we have studied the excitation of cn by combining the cn and @xmath2cn line intensities measured in this paper with those measured by rodrguez
franco et al .
( 1998 ) . for this , we have used the model based on the large velocity gradient ( lvg ) approximation described by rodrguez
franco et al . ( 1998 ) and a constant kinetic temperature of @xmath44k ( wilson et al .
2000 ) . in this model
only collisional excitation has been considered .
the effects of the excitation of the first vibrationally excited levels by infrared radiation at 4.8 microns has been considered negligible for the typical physical conditions in photodissociation regions ( fuente et al .
1995 ) . the data of rodrguez
franco et al . (
1998 ) have been smoothed to have the same angular resolution as the hht cn @xmath0 data
[ fig : lvg]a and b show the results obtained from the lvg model calculations . fig .
[ fig : lvg]a shows the dependence of the h@xmath4 densities ( n@xmath45 ) with the cn column density ( n(cn ) ) as a function of the intensity of main group of components of the @xmath0 multiplet , @xmath46(@xmath47 ) ( thick line ) observed in this paper and the ratio between the intensities of the @xmath0 and @xmath1 lines , r31 ( dotted lines ) . in fig .
[ fig : lvg]b we plot the h@xmath4 density versus the cn column density as a function of the optical depth of the @xmath0 , @xmath48 , @xmath49 line , @xmath50 , and the ratio between the fine components rf32 .
[ fig : lvg]b shows that model predicts for a rather wide range of h@xmath4 densities and cn column densities , line intensity ratios around 1.5 ( see table 1 ) . even for the case of optically thick emission with opacities between 1 and 5 the predicted value of rf32 will be of @xmath43
this is clearly the case for the two positions in the ridge r4 and r5 . in order to fit the @xmath2cn and cn lines simultaneously in the two ridge positions r4 and
r5 where @xmath2cn was observed , @xmath2cn column densities of @xmath51@xmath52 are required . assuming a standard cn/@xmath2cn ratio of 89 ( wilson 1999 )
, the cn column density must be close to @xmath53@xmath52 .
for this column density , the hf component of cn @xmath0 line presented in this paper is optically thick with opacities @xmath54 ( see fig .
[ fig : lvg]b ) and the expected hf ratio rf32 of @xmath43 .
this indicates , as previously mentioned , that the hf ratio does not necessarily give a measure of the cn line optical depths .
this is because the excitation temperature is not the same for all the observed hyperfine components .
since the excitation is mainly collisional , the excitation temperature is expected to be higher in the hyperfine components with higher opacities .
in fact , in the simple isothermal case considered in this paper , the lvg model gives @xmath55k and @xmath56k for the @xmath42 and @xmath33 fine components of the @xmath0 transition respectively .
because of these differences in the excitation temperature , the ratio between the hf structure components remains quite uniform and close to 1.5 for a large range of opacities . in summary
, the hf line intensity ratio of the @xmath0 cn line does not necessary give a reliable estimate of the line opacity even if the ratios are close to the expected optically thin case .
observations of the @xmath2cn line and/or a number of hf components with a wide range of relative intensities are needed to estimate the opacities of the strongest hf lines of cn .
as shown in fig . [ fig : lvg]a one can use the line intensity ratio of the @xmath0 , and the @xmath3 lines of cn to derive the h@xmath4 density and the cn column density for the observed position in the ridge and in the trapezium once the kinetic temperature is known . in table
[ tab : derived ] we present the results for the h@xmath4 densities and cn column densities for the trapezium and the ridge for a kinetic temperature of @xmath44k .
these results are basically independent of the assumed kinetic temperatures for kinetic temperatures larger than 25@xmath9k ( rodrguez franco et al .
1998 ) . for the positions in the ridge we can not fit the intensity of the cn @xmath0 line and the ratio r31 for the optically thin solution .
therefore we have considered the solution for the optically thick case found in the previous section for r4 and r5 . for the trapezium positions , the cn @xmath0 line is weaker than in the ridge , and the optically thin solution can explain both , the line intensities and the ratio r31 .
the trapezium and the ridge regions present very different physical conditions ( see table [ tab : derived ] ) . in the trapezium region , the densities are between @xmath57 and @xmath58@xmath59@xmath6 and cn column densities around @xmath60@xmath52 . with these physical conditions
the opacity of the main hyperfine component of the cn @xmath0 line is @xmath61 as expected from the hyperfine ratios of the cn @xmath622 and @xmath630 lines . for the ridge
we obtain densities between @xmath64 and @xmath65@xmath6 , cn column densities between @xmath66 and @xmath67@xmath52 , and opacities between 1 and 5 .
the derived h@xmath4 densities for the ridge from the cn emission are in good agreement with those derived from hc@xmath41n ( rodrguez@xmath68franco et al .
1992 ) . in summary ,
the @xmath0 cn data confirm the h@xmath4 densities for the trapezium region , typically of a few @xmath18@xmath6 .
these are larger , on average , by a factor of @xmath69 than those in the molecular ridge ( @xmath70@xmath6 )
. however , the cn column densities in the ridge ( @xmath71@xmath52 ) are larger by more than a factor of 10 than in the trapezium ( @xmath72@xmath52 ) .
we have also derived the relative abundance of cn in both regions by combining the cn column densities with the h@xmath4 column density estimated from the @xmath2co and c@xmath73o data of white & sandell ( 1995 ) and wilson et al .
the h@xmath4 column densities have been derived by using the lvg approximation to model the co excitation and assuming a @xmath2co / c@xmath73o ratio of 5.6 .
the results are given in table [ tab : derived ] .
we find h@xmath4 column densities of @xmath74@xmath52 for the trapezium region and of @xmath75@xmath52 for the ridge .
> from the h@xmath4 and cn column densities , we estimate the cn / h@xmath4 ratios shown in column 5 of table [ tab : derived ] .
the cn abundance ratios in the trapezium region and molecular ridge are similar , although there is a systematic trend for the cn abundance to be somewhat larger in the trapezium ( average of @xmath76 ) than in the ridge ( average of @xmath77 ) . combining the h@xmath4 densities derived for the multi - transitional analysis of cn with the h@xmath4 column densities derived from @xmath2co and c@xmath73o
, we can also estimate the thickness of the molecular gas layer along the line of sight .
in column 6 of table [ tab : derived ] we show the results .
we derive a thickness for the ridge of typically @xmath78 cm .
this is consistent with the emission arising from spherical condensations with a size of @xmath79 .
the thickness of the molecular layer in the trapezium region is @xmath80 times smaller than the thickness of the ridge .
the small thickness of the cn emission toward the trapezium region makes it very unlikely that this emission arises from condensations with spherical geometry as for the ridge , since this geometry would require condensations with sizes of only @xmath81 . to explain the observed cn line intensities one would require a large number of these condensations filling the telescope beam .
since the cn emission is extended we conclude that the cn emission in the trapezium region arises from a thin , @xmath8 cm , layer of dense ( few @xmath82@xmath6 ) molecular gas ( hereafter the trapezium molecular layer ) .
rodrguez franco et al . ( 1998 ) have discussed in detail the implications of the cn emission in the confinement of the orion a h@xmath9ii region in the perpendicular direction to the line of sight .
the geometry of the trapezium molecular layer , and the chemistry , with large cn abundance as compared with other molecules like hc@xmath41n , confirms that the cn emission is related with the material which confines the rear side of orion a. the presence of a layer of material behind the trapezium confining the orion a h@xmath9ii region is required to explain the kinematics of the ionized gas ( zuckerman , 1973 ) .
the confining material will show the typical multilayer ( ionized , neutral atomic and molecular ) structure expected when molecular material is exposed to uv radiation ( see eg .
hollenbach & tielen 1997 ) .
[ fig : model ] shows an sketch of the distribution and kinematics of the ionized and molecular gas towards the trapezium .
a model of the location of the different layers along the line of sight has been obtained by odell et al .
the ionized gas is blueshifted with respect to the neutral molecular gas , indicating that the ionized gas is flowing towards the observer . to prevent the ionized gas from escaping from the h@xmath9ii region in the direction away from the observer
, the rear side of the h@xmath9ii region must be confined by a dense layer of gas and dust .
this is confirmed by the radial velocities of o@xmath9i and h@xmath9i . both the atomic layer probed by o@xmath9i and h@xmath9i and the molecular layer probed by cn show velocities close to 9@xmath9km@xmath9s@xmath83 , the ambient cloud velocity , suggesting that this neutral gas forms the front face of the neutral confining cloud ( odell et al .
1993 ; van der wef & goss 1989 ) .
the cn emission towards the trapezium has similar radial velocities than that of the warm co ( howe et al .
1993 ; wilson et al .
2000 ) heated by the uv radiation .
as expected for a dense photodissociation region ( sternberg & dalgarno 1995 ) , both molecular emission arise from basically the same layer which located close to the principal ionization front ( see eg .
odell et al . 1993 ) .
at the opposite side of the ionization from the dense and warm molecular layer will be surrounding by an envelope of molecular and atomic material ( see van der wef & goss 1989 ) .
the origin of the thin confining layer in the trapezium and its present morphology is unclear since the evolution of h@xmath9ii regions is far from understood ( see e.g. churchwell 1990 ) .
one would think that the dense thin layer has been built up by material swept by the expansion of the h@xmath9ii region in the last @xmath18yr . in this case
the molecular component of the confining material should be expanding . expanding molecular gas with a velocity of at least @xmath84km@xmath9s@xmath29
has been observed from the cn emission in the molecular molecular bars ( northern bar , optical filament ) which are believed to belong to the material that confine the h@xmath9ii region in the direction close to the plane of the sky ( rodrguez@xmath68franco et al .
1998 ) . however , the observe thin cn layer has a radial velocity of @xmath85km@xmath9s@xmath29 , typical of the molecular gas unperturbed by the effects of the h@xmath9ii region .
these arguments point towards the possibility that either the cn trapezium layer is not related to the material that confines the h@xmath9ii region or to the fact that the evolution of m@xmath942 does not fit the simple picture of an h@xmath9ii region expanding at a velocity of @xmath86km@xmath9s@xmath29 .
the first possibility is unlikely since the uv radiation seems to dominate the chemistry of the thin molecular layer indicating small visual extinction ( @xmath87mag ) between the ionizing front and the molecular layer ( rodrguez@xmath68franco et al .
1998 ) . furthermore , the line widths of the cn are broader than those found for the ridge , revealing a kinematic interaction between the h@xmath9ii region and the trapezium molecular layer . if the trapezium molecular layer is indeed interacting with the h@xmath9ii region , the lack of velocity shifts between co ( from the bulk of the ridge ) and cn from the trapezium layer indicates that the h@xmath9ii region is not expanding away from the observer .
the trapezium molecular layer would therefore indicate that the presence of `` static '' layers not only occurs at the early phases of the evolution of the h@xmath9ii regions as shown by wood & churchwell ( 1989 ) , but also in more evolved objects .
although several suggestions have been made to explain the lifetime , morphology and kinematics of the ultra - compact h@xmath9ii regions ( see e.g.tenorio-tagle 1979 ; reid et al .
1980 ; van buren et al . 1990 ; hollenbach et al . 1994 ; dyson et al . 1995 ; lizano et al . 1996 ) , none of these models adequately explain all the observational facts ( jaffe & martn@xmath68pintado , 1999 ) . most of the models developed to explain the lifetime paradox such as the bow shocks , mass - loaded winds and confinement by infalling material predict that the confining material is not `` static '' with respect to the ambient molecular clouds .
the disk evaporated model can not explain the morphology and kinematics of evolved h@xmath9ii regions such us m@xmath15 .
the champagne flow could explain both the kinematics of the ionized gas in m@xmath942 and the presence of the `` static '' and the expanding molecular layers confining m@xmath942 .
in the scenario of the champagne model , the massive stars are formed within a molecular cloud with a large density gradient .
the h@xmath9ii region confined by material with a density gradient will expand in some directions but depending on the density and temperature of of the surrounding material , can be `` static '' in other directions . as suggested by de pree et al .
( 1995 ) very dense ( @xmath88@xmath6 ) , and warm ( 100@xmath9k ) molecular material would slow down the expansion of an h@xmath9ii region in the early phases and provide an explanation for the presence of `` static '' layers in more evolved h@xmath9ii regions . in the case of m@xmath15
the h@xmath4 density gradient is along the line of sight with the larger density in the rear side of the h@xmath9ii region , just where the trapezium layer is found . in the trapezium layer
the h@xmath4 density is larger than @xmath18@xmath6 and the kinetic temperatures are @xmath89k ( wilson et al .
the cn layer will be in pressure equilibrium with the ionized gas for the typical densities of a few @xmath90@xmath6 and the electron temperature of @xmath91k measured in m42 ( wilson et al . 1997 ) .
the small thickness of the the cn layer , as compared to the ridge , can also be understood in the framework of the champagne flow since the gas flowing towards the observer has been photo evaporated from the confining layer as sketched in fig .
[ fig : model ] .
if the ionized material is flowing away from the ionization front at a speed of @xmath92km@xmath9s@xmath29 the thin molecular layer will be photo evaporated in a time scale of @xmath93 years .
we have used the hht telescope to carry out observations of the @xmath0 and new observations of the @xmath1 lines of cn toward selected positions in the trapezium region and in the molecular ridge of the orion a molecular cloud .
the main results are summarized as follow . 1
. the hyperfine intensity ratio of just one transition of cn does not give reliable estimate of the optical depth of the lines due to different excitation temperatures for different hf components .
the ratios between hf components of the @xmath0 line can be close to optically thin values even in the case of optically thick emission observations of several hf groups with very different intensities or / and observations of isotopic substitute of cn are required to derive reliable opacities .
2 . our measurements of @xmath0 line of cn
confirm the presence of a high density ( a few @xmath18@xmath6 ) molecular gas toward the trapezium region with also relatively large cn abundances ( @xmath94 ) .
the h@xmath4 density in the trapezium is a factor of 10 larger than in the ridge .
the high density molecular gas towards the trapezium is located in a thin layer ( @xmath95 cm ) .
this thin molecular layer is related with the material that confines the far side of the h@xmath9ii region in the trapezium region , and it is `` static '' with respect to the molecular cloud .
the kinematics of the ionized gas and the small thickness of the trapezium layer can be understood in the framework of the champagne flows .
the `` static '' molecular layer is in equilibrium with the h@xmath9ii region suggesting that evolved h@xmath9ii regions are not expanding in all directions as occurs for uc h@xmath9ii regions .
lccccccc t1 ( 60,@xmath96 ) & 0.9 & 9.5 & 5.9 & & & 0.4 & 1.6 + t2 ( 108,24 ) & 0.6 & 9.9 & 3.8 & & & @xmath97 & 1.4 + t3 ( 180,@xmath96 ) & 2.1 & 10.4 & 3.2 & & & 1.2 & 1.3 + t4 ( 48,@xmath98 ) & 1.1 & 9.2 & 4.1 & & & 1.3 & 1.6 + t5 ( 96,@xmath99 ) & 0.3 & 9.7 & 5.3 & & & @xmath100 & 0.9 + t6 ( 38,@xmath101 ) & 2.6 & 8.8 & 3.9 & & & 0.6 & 1.3 + r1 ( 41,186 ) & 5.4 & 9.6 & 2.5 & & & 10.9 & 1.3 + r2 ( 0,113 ) & 2.8 & 9.2 & 2.4 & & & 12.0 & 1.6 + r3 ( @xmath102,15 ) & 3.1 & 7.8 & 3.2 & & & 4.9 & 1.5 + r4 ( @xmath102,@xmath102 ) & 4.7 & 7.6 & 3.6 & 0.46 & 0.12 & 7.5 & 1.3 + r5 ( @xmath103,@xmath104 ) & 4.7 & 7.2 & 3.7 & 0.43 & 0.13 & 9.0 & 1.5 + r6 ( @xmath105,@xmath106 ) & 3.1 & 8.6 & 1.7 & & & 6.2 & 1.4 + lcccccc t1 ( 60,@xmath96 ) & 8.0 & 2.59 & 6 & 3.1 & 4.31 & + t2 ( 108,24 ) & 2.2 & 0.86 & 12.5 & 2.5 & 0.69 & + t3 ( 180,@xmath96 ) & 3.0 & 0.86 & 50 & 3.5 & 0.17 & + t4 ( 48,@xmath98 ) & 3.3 & 3.46 & 6 & 0.9 & 5.77 & + t5 ( 96,@xmath99 ) & 1.7 & 0.43 & 10 & 3.9 & 0.43 & + t6 ( 38,-84 ) & 4.3 & 0.86 & 70 & 4.9 & 0.12 & + r1 ( 41,186 ) & 245 & & 1 & & & + r2 ( 0,113 ) & 240 & 6.06 & 0.31 & 39 & 195 & + r3 ( @xmath102,15 ) & 30 & 10.4 & 2 & 2.8 & 52.0 & + r4 ( @xmath102,@xmath102 ) & 97 & 12.1 & 1 & 8.0 & 121 & 1.3 + r5 ( @xmath103,@xmath107 & 75 & 12.1 & 2 & 6.2 & 60.7 & 1.0 + r6 ( @xmath105,@xmath106 ) & 90 & 11.2 & 0.7 & 8.0 & 160 & + | we present hht observations in the @xmath0 rotational transition of the cn radical toward selected positions of the trapezium region and of the molecular ridge in the orion molecular cloud .
two of the positions in the ridge were also observed in the @xmath1 line of cn and @xmath2cn .
the @xmath0 cn lines have been combined with observations of the @xmath1 and @xmath3 transitions of cn , and of the @xmath1 of @xmath2cn to estimate the physical conditions and cn abundances in the molecular gas .
we analyze in detail the excitation of the cn lines and find that the hyperfine ratios of the @xmath0 line are always close to the local thermodynamic equilibrium ( lte ) optically thin values even in the case of optically thick emission .
this is due to different excitation temperatures for the different hyperfine lines . from the line intensity ratios between the different cn transitions we derive h@xmath4 densities of @xmath5@xmath6 for the molecular ridge and of @xmath7@xmath6 for the trapezium region .
the cn column densities are one order of magnitude larger in the ridge than in the trapezium region , but the cn to h@xmath4 ratio is similar both in the trapezium and in the ridge .
the combination of the low cn column densities , high h@xmath4 densities and relatively high cn abundances toward the trapezium region requires that the cn emission arises from a thin layer with a depth along the line of sight of only @xmath8 cm .
this high density thin layer of molecular gas seems to be related with material that confines the rear side of the h@xmath9ii region orion a. however the molecular layer is not moving as expected from the expansion of the h@xmath9ii region , but it is `` static '' with respect to the gas in the molecular cloud .
we discuss the implication of a high density `` static '' layer in the evolution of an h@xmath9ii region . |
the decay width of the process @xmath112 is proportional to the following phase space , @xmath113 where @xmath114 , @xmath115 .
furthermore , in the rest frame of @xmath3 boson ( @xmath116 ) , we have @xmath117 where @xmath118 and @xmath119 , @xmath120 and @xmath121 .
the step function @xmath122 is determined by ensuring @xmath123 , where @xmath124 with all these step function above , we can get the integration ranges : @xmath125 where @xmath126 . furthermore , we can obtain the @xmath69 distribution @xmath127 where the extra jacobian @xmath128 and there are two @xmath63 in different range of @xmath69 and @xmath62 , i.e. @xmath129 \nonumber\\ & & \pm m_z|\vec{q_2}|\cos\theta_{23}[m_{b_c}^4 - 2m_{b_c}^2(m_c^2 + 2(m_z - q_2 ^ 0)^2 -s_1 - 2|\vec{q_2}|^2 \cos^2\theta_{23})+(m_c^2-s_1)^2]^\frac{1}{2}\bigg\ } , \end{aligned}\ ] ] where @xmath130 is obtained when @xmath131 $ ] and @xmath132 , s_{1\min}[{\cos\theta_{23}=0}]]$ ] . and @xmath133 is obtained when @xmath134 $ ] and @xmath135,s_{1\max}]$ ] or @xmath136 $ ] and @xmath137,s_{1\max}]$ ] .
the @xmath122 function determines the boundary of @xmath62 : @xmath138&=&\frac{m_b^{2 } ( \cos^2\theta_{23}-1 ) m_{b_c}^2 + m_z^{2}(m_c^2 + m_{b_c}^2 \cos^2\theta_{23 } ) + m_{b_c } m_z\sqrt{y } } { ( \cos^2\theta_{23}-1 ) m_{b_c}^2 + m_z^2}\end{aligned}\ ] ] with @xmath139 the distribution for @xmath68 can be obtained in a similar way .
the amplitude @xmath140 of the process @xmath25 has the general structure @xmath141 where @xmath142 can be read from eqs.([a1])-([a4 ] ) . to derive analytical expression for the process and to make its form simpler as much as possible
, we adopt the ` new trace amplitude approach ' suggested by refs.@xcite to do our calculation .
detailed process of the approach can be found in refs.@xcite , and here , we shall only list our main results . after summing up the spin states ,
the square of the amplitude can be divided into four parts , @xmath143 where by introducing a light - like momentum @xmath144 and a spacelike vector @xmath145 that satisfies the relations , @xmath146 and @xmath147 , the four amplitude @xmath148 can be written as @xmath149 , \nonumber\\ m_2 & = & \frac{n}{{\sqrt 2 } } tr\left [ { ( { \slashed{q}_1 } - { m_c}){\gamma _ 5}{\slashed{k}_0}({\slashed{q}_2 } + { m_b})a } \right ] , \nonumber\\ m_3 & = & \frac{n}{{\sqrt 2 } } tr\left [ { ( { \slashed{q}_1 } - { m_c}){\slashed{k}_0}{\slashed{k}_1}({\slashed{q}_2 } + { m_b})a } \right]\nonumber\end{aligned}\ ] ] and @xmath150 , \nonumber\ ] ] where @xmath151 is the normalization constant . @xmath144 and @xmath145
are arbitrary momenta , and in order to write down @xmath152 as explicitly and simply as possible : + 1 ) we set @xmath153 , where the coefficient @xmath154 is determined by the requirement that @xmath144 be a lightlike vector : @xmath155 2 ) we set @xmath156 , where @xmath157 ensures @xmath146 .
it is found that @xmath158 can be expressed as , @xmath159 . \\\ ] ] and then the resultant @xmath148 can be simplified as : @xmath160 , \\
m_2 & = & { l_2 } \times tr [ ( { \slashed{q}_1 } - m_c){\gamma _ 5}(\slashed{q}_{2}+{m_b})a ] , \\
m_3 & = & m_{3 ' } - { n_0}[{m_b } ( { q_1 } \cdot { k } ) + { m_c}({q_2 } \cdot { k})]m_2 , \\ m_4 & = & m_{4 ' } + { n_0}[{m_b } ( { q_1 } \cdot { k } ) - { m_c}({q_2 } \cdot { k})]m_1 , \end{aligned}\ ] ] where @xmath161 , \\ m_{4 ' } & = & -\frac{n_0}{4l_1 } tr\left [ { ( { \slashed{q}_1 } - { m_c}){\slashed{k}}({\slashed{q}_2 } + { m_b})a } \right ] .\end{aligned}\ ] ] furthermore , the amplitudes @xmath148 can be expanded over some basic lorentz structures : @xmath162 and @xmath163 where @xmath164 is the number of basic lorentz structure @xmath165 , whose value dependents on the @xmath78-quarkonium state @xmath166 : e.g. @xmath167 for @xmath168_1 $ ] , @xmath169 for @xmath170_1 $ ] . as for @xmath171 and @xmath172 ,
they can be expressed by @xmath173 a^2_j(n ) , \nonumber\\ a^4_j(n ) & = & a^{4'}_j(n)+{n_0}[{m_b } ( { q_1 } \cdot { k } ) - { m_c}({q_2 } \cdot { k } ) ] a^1_j(n ) .
\nonumber\end{aligned}\ ] ] the explicit expression for @xmath174 and @xmath175 of each state shall be listed in the following subsections . to shorten the notation , we set @xmath176 and @xmath177 . and , we define some dimensionless parameters @xmath178 and @xmath179 where @xmath66 , @xmath67 , and @xmath180 , which satisfy the relation : @xmath181 .
and the short notations for the denominators are @xmath182 furthermore , the following relations are useful to short the expressions : @xmath183 | the programming new @xmath0 collider with high luminosity shall provide another useful platform to study the properties of the doubly heavy @xmath1 meson in addition to the hadronic colliders as lhc and tevatron . under the ` new trace amplitude approach ' , we calculate the production of the spin - singlet @xmath1 and the spin - triplet @xmath2 mesons through the @xmath3 boson decays , where uncertainties for the production are also discussed .
our results show @xmath4 kev and @xmath5 kev , where the errors are caused by varying @xmath6 and @xmath7 within their reasonable regions .
+ * pacs numbers : * 12.38.bx , 12.39.jh , 14.40 lb , 14.40.nd the @xmath1 meson is a double heavy quark - antiquark bound state and carries flavors explicitly . since its first discovery at tevatron @xcite , @xmath1 physics is attracting more and wide interests .
recently , many progresses have been made for the hadronic production of @xmath1 meson at high energy colliders as lhc and tevatron .
a computer program bcvegpy for the direct hadronic production of @xmath1 meson has been presented in refs.@xcite . and it has been found that the indirect production of @xmath1 via top quark decays can also provide useful information on @xmath1 meson @xcite . comparing with the hadronic colliders ,
an @xmath8 collider has its own advantages , mainly because of its lower background . as for the previous lep - i experiment ,
no @xmath1 events have been found due to its lower collision energy and low luminosity @xcite .
however , if the luminosity of the @xmath9 collider can be raised up to @xmath10 or even higher as programmed by the internal linear collider ( ilc ) @xcite , then there might have enough events .
moreover , if the @xmath8 collider further runs at the @xmath3-boson energy , the resonance effects at the @xmath3 peak may raise the production rate up to several orders .
it has been estimated by ref.@xcite that more than @xmath11 @xmath3-events can be produced at ilc per year , which is about @xmath12 orders higher than that collected by lep - i .
such a high luminosity collider is called as gigaz @xcite or a @xmath13-factory @xcite .
then it will open new opportunities not only for high precision physics in the electro - weak sector , but also for the hadron physics .
the production of @xmath1 through @xmath3 decays has been studied in refs @xcite with various methods .
since the process is very complicated , it would be helpful to have a cross check of these results .
furthermore , considering the forthcoming @xmath13-factory , it may be interesting to know the theoretical uncertainties in estimating of @xmath1 production . .
, scaledwidth=40.0% ] for the purpose , we need to calculate the process @xmath14 , whose feynman diagrams are shown in fig.([feyn ] ) . according to the nrqcd factorization formula @xcite , the decay width for the process @xmath14
can be written in the following factorization form : @xmath15+b+\bar{c})\langle{\cal o}^h\rangle,\ ] ] where the matrix element @xmath16 is proportional to the inclusive transition probability of the perturbative state @xmath17 $ ] into the bound states of @xmath1 . as for the two color - singlet @xmath18-wave states
@xmath19 $ ] and @xmath20 $ ] , their matrix elements @xmath16 are related with the bethe - salpeter wave function at the origin that can be determined by the potential model @xcite .
@xmath21+b+\bar{c})$ ] stands for the short - distance decay width , i.e. @xmath22+b+\bar{c})= \frac{1}{2k^0 } \overline{\sum } |m|^{2 } d\phi_3,\ ] ] where @xmath23 means we need to average over the spin states of initial particles and to sum over the color and spin of all the final particles . and in @xmath3 rest frame , the three - particle phase space can be written as @xmath24 the hard scattering amplitude for the process @xmath25 can be written as : @xmath26 where @xmath27@xmath28 .
the gamma structure @xmath29 ( @xmath30 , @xmath31 , @xmath32 ) corresponds to the four feynman diagrams in fig.([feyn ] ) .
more explicitly , @xmath29 can be written as @xmath33_{q=0 } , \label{a1}\\ a_2 & = & \left[\gamma_{\rho}\frac{\slashed{k}-\slashed{q}_{32}+{m_b}}{(k - q_{32})^2- m_b^2}{\slashed{\epsilon}(k)}{\gamma_{z\bar b } } \frac{\chi^{s{s_z}}_{q_3}(q ) } { ( q_{31 } + { q_1})^2}{\gamma_\rho}\right]_{q=0 } , \label{a2}\\ a_3 & = & \left[\gamma_{\rho}\frac{\chi^{s{s_z}}_{q_3}(q)}{({q_{32}}+ { q_2})^2 } \slashed{\epsilon}(k ) \gamma_{zc } \frac{\slashed{q}_{31 } - \slashed{k } + m_c}{(q_{31 } - k)^2- m_c^2}{\gamma_\rho}\right]_{q=0 } \label{a3}\end{aligned}\ ] ] and @xmath34_{q=0 } , \label{a4}\end{aligned}\ ] ] where @xmath35 , @xmath36 and @xmath37 is the relative momentum between the two constitute quarks of @xmath38-quarkonium . in the nonrelativistic approximation , the @xmath18-wave projector @xmath39 takes the following form @xmath40 where @xmath41 and @xmath42 for @xmath43 meson respectively .
@xmath44 is the polarization vector , @xmath45 and @xmath46 are the momenta of the two constitute quarks of the meson , @xmath47 where @xmath48 is implicitly adopted to ensure the gauge invariance of the hard scattering amplitude . by using the conventional trace technique ,
we need to derive the squared amplitude , which is very complicated and lengthy for the present case . to derive analytical expression for the process @xmath49 and to make its form simpler as much as possible
, we adopt the ` new trace amplitude approach ' suggested and developed by refs.@xcite to do our calculation . under the approach , we first arrange each of the four amplitudes listed in eqs.([a1],[a2],[a3],[a4 ] ) into four orthogonal sub - amplitudes according to the four spin combinations of the outgoing @xmath50-quark and @xmath51-antiquark , and then do the trace of the dirac-@xmath52 matrix strings at the amplitude level by properly dealing with the massive spinors , which results in explicit series over several independent lorentz - invariant structures . and then , our task left is to determine the coefficients of these lorentz - invariant structures . to make the paper more compact , we present the detailed formulae for dealing with the process in appendices a and b , where appendix a gives the phase - space integration formulae and appendix b gives the ` new trace amplitude approach ' , which presents all the necessary coefficients for the lorentz - invariant structures . as a cross check of the present obtained results , by taking the same parameters , we can obtain consistent numerical results as that of ref.@xcite within reasonable numerical errors .
in doing the numerical calculation , we take @xmath53 gev and @xmath54 @xcite . to be consistent with the present leading - order calculation
, we adopt the leading - order @xmath55 running , and by taking the normalization scale to be @xmath56 , which leads to @xmath57 .
the two constitute quark masses are taken as @xmath58 gev and @xmath59 gev . with the above parameter values , it can be found that the total decay width @xmath60 kev and @xmath61 kev .
the differential distributions of the invariant masses @xmath62 and @xmath63 , i.e. @xmath64 and @xmath65 are shown in fig.([diss1s2 ] ) , where @xmath66 and @xmath67 . and
the differential distributions of @xmath68 and @xmath69 , i.e. @xmath70 and @xmath71 are shown in fig.([discos ] ) , where @xmath72 is the angle between @xmath73 and @xmath74 , and @xmath75 is the angle between @xmath76 and @xmath74 respectively .
it can be found that the largest differential decay width of @xmath70 is achieved when @xmath77 , i.e. the @xmath78-quarkonium and @xmath79-quark moving in the same direction . while the largest differential decay width of @xmath71 is achieved when @xmath80 , i.e. the @xmath78-quarkonium and @xmath50-quark moving back to back .
next , it would be interesting to show the theoretical uncertainties for the production .
main uncertainty sources include the matrix elements ( or the wavefunction at the origin of the binding system @xmath81 ) , the renormalization scale @xmath82 , the constitute quark masses @xmath6 and @xmath7 .
@xmath81 and @xmath83 are overall parameters for the present case , and their uncertainties can be easily figured out .
for example , one can set @xmath82 to be @xmath56 or @xmath84 , since the intermediate gluon as shown in fig.([feyn ] ) should be hard enough so as to produce a @xmath85-quark pair or a @xmath86-quark pair , which inversely ensures the pqcd applicability of the process . by setting these two scales to calculate the process , we obtain the ratio @xmath87 . in the following discussion ,
we fix @xmath88 and @xmath89 gev@xmath90 @xcite . for clarity
, we present the uncertainties of @xmath7 and @xmath6 in ` a factorizable way ' .
when focussing on the uncertainties from @xmath7 , we let it be a basic ` input ' parameter varying in a possible range @xmath91 with all the other factors , including the @xmath50-quark mass and _ etc .
_ being fixed to their center values .
similarly , when discussing the uncertainty caused by @xmath6 , we vary the @xmath50-quark mass @xmath6 within the region of @xmath92 . .decay
width for the production of @xmath93 through @xmath3 decay with varying @xmath7 , where @xmath6 is fixed to be @xmath94 gev . [ cols="^,^,^,^",options="header " , ] the decay width for the production of @xmath93 through @xmath3 decay with varying @xmath7 or @xmath6 are presented in tab.[tabmc ] and tab.[tabmb ] .
it shows that the decay width is more sensitive to @xmath7 , which decreases with the increment of @xmath7 . by adding these two uncertainties caused by @xmath6 and @xmath7 in quadrature
, we obtain @xmath95 and @xmath96 the shaded bands in figs.([sun],[cosun ] ) show the corresponding uncertainty more clearly , where the contributions from @xmath97 and @xmath98 are summed up .
the center solid line is for @xmath99 and @xmath100 , the upper edge of the band is obtained by setting @xmath101 and @xmath102 , while the lower edge of the band is obtained by setting @xmath103 and @xmath104 . as a summary : by using the ` new trace amplitude approach ' , we calculate the @xmath1 production process , @xmath105 . the decay widths together with their uncertainties caused by the @xmath50 and @xmath79 quark masses are @xmath106 kev and @xmath5 kev , where the errors are caused by varying @xmath6 and @xmath7 within their reasonable regions @xmath107 $ ] gev and @xmath108 $ ] gev .
further more , the differential decay width for @xmath109 and @xmath110 together their uncertainties are drawn in figs.([sun],[cosun ] ) .
considering the advantage of the clean environment in @xmath111 collider , it will provide another useful platform in studying the @xmath1 production .
acknowledgements : this work was supported in part by natural science foundation project of cq cstc under grant no.2008bb0298 , by natural science foundation of china under grant no.10805082 and no.11075225 , and by the fundamental research funds for the central universities under grant no.cdjzr101000616 . |
alkali doped iron selenide materials@xcite a@xmath1fe@xmath2se@xmath3 ( a = alkali element ) are high temperature superconductors with t@xmath6k@xcite that also undergo a transition to a iron vacancy ordered structure at t@xmath7@xcite followed closely by a transition to a block anti - ferromagnetic state ( bafm ) with neel temperatures t@xmath8.@xcite the bafm develops within the @xmath9 arrangement of iron vacancies with a magnetic moment of @xmath10 .
this is the highest reported magnetic moment among all iron pnictide and iron chalcogenide superconductors .
the existence of the above phases within the superconducting state has generated intense debate about the interplay between these seemingly competing phases , and about the origin of superconducting pairing leading to high temperature superconductivity in this unusual environment.@xcite the presence of phase separation between superconducting regions and the bafm has been confirmed using a variety of experiments including spectroscopy,@xcite microscopy,@xcite and x - ray diffraction@xcite techniques .
the phase separation has been observed along the @xmath11-axis and also in the fese plane as nanoscale filamentary superconducting domains existing within the vacancy ordered bafm regions.@xcite though the morphology and physical properties of the metallic nano domains can depend on the thermal history@xcite and stoichiometry,@xcite a number of alkali iron selenide samples have been observed to be free from iron vacancies in the superconducting regions.@xcite angular resolved photoemission ( arpes ) experiments investigating the metallic regions observe a fermi surface with two electron pockets around the m points and a small electron pocket developing around the z point.@xcite the absence of a hole pocket around the @xmath12 point has reinvigorated the discussion of the origin of superconductivity in the fe - based materials .
specifically , the absence of @xmath13 nesting between hole and electron fermi surface sheets prevents the leading @xmath14 pairing instability found e.g. within spin fluctuation mediated pairing.@xcite theoretical calculations for the superconducting ground state are currently inconclusive and predictions have been made for both @xmath4-wave @xcite and @xmath5-wave @xcite pairing symmetry .
measurements within the superconducting state find a nodeless superconducting gap.@xcite although arpes detects a nearly isotropic gap structure,@xcite stm measurements have observed a double gap feature in the local density of states ( ldos).@xcite this does not identify the gap symmetry in alkali iron selenides since neither @xmath4-wave nor @xmath5-wave symmetry possess any symmetry enforced nodes at @xmath15 for this fermi surface .
however , arpes measurements in ref .
find an isotropic superconducting gap structure at the z point where a electron pocket exists around the @xmath12 point indicating an @xmath5-wave gap symmetry . from the above discussing
, it is evident that new theoretical methods to determine the pairing symmetry of these materials are desirable . in this paper
we study the in - plane interface states near boundary regions of a bafm with ordered arrangement of iron vacancies and a next nearest neighbor ( nnn ) @xmath5-wave or nearest neighbor ( nn ) paired @xmath4-wave superconductor .
note that in the absence of a fermi pocket around the @xmath12 point the nnn @xmath14 state does not have a sign change between fermi pockets .
we denote this superconducting state by @xmath5-wave with the understanding that it represents nnn superconducting pairing interaction in real space .
we use a five orbital tight binding hamiltonian and include superconductivity by generating pairing from a spin fluctuation exchange mechanism within an rpa weak coupling theory .
the model is able to reproduce the electronic structure of k@xmath1fe@xmath2se@xmath3 and stabilise a bafm structure as well as a superconducting state with @xmath5-wave or @xmath4-wave symmetry .
it is found that the presence of an interface between these two ordered phases leads to the formation of in - gap bound states which should be detectable by future stm measurements .
bound interface states are generated for both @xmath5-wave and @xmath4-wave symmetry due to the magnetic nature of the bafm . a qualitative difference between @xmath5- and @xmath4-wave order is shown to arise near e.g. ( 110 ) non - magnetic scattering boundaries such as cracks , grain boundaries or free surfaces within the superconducting phase .
the paper is organized as follows , in the next section we provide details of the model used for our calculations .
this is followed by an analysis of the electronic structure , magnetic order , bulk superconducting order and finally we discuss the results for the ldos in the simplified inhomogeneous situation of an interface between a bafm and a superconductor .
the five - orbital model hamiltonian is given by @xmath16 the first term is determined by the tight - binding band @xmath17 here the operators @xmath18 ( @xmath19 ) create ( annihilate ) an electron at the @xmath20-th site in the orbital @xmath21 and with spin projection @xmath22 , and @xmath23 is the chemical potential . the indices @xmath21 and @xmath24 correspond to the @xmath25 , @xmath26 , @xmath27 , @xmath28 and @xmath29 iron orbitals .
the hopping parameters @xmath30 are obtained by slightly modifying the tight - binding parameters of kfe@xmath3se@xmath3 in ref . .
this is similar to the prescription used by maier _
et al._@xcite and consists of shifting the @xmath25/@xmath26 orbital nearest neighbor hopping by -0.08 ev and @xmath28 orbital hoppings by 0.08 ev . the resulting band structure and fermi surface
are shown in fig .
[ fig : normal ] .
it can be seen by comparing with the kfe@xmath3se@xmath3 bands in ref . that the modification of the hopping parameters shifts down the hole band near the @xmath12 point and leads to a fermi surface with only electron sheets .
the fermi surface shown in fig .
[ fig : normal](b , c ) corresponds to an electron doping of k@xmath1fe@xmath2se@xmath3 with @xmath31 , and @xmath32 ( 0.15 electrons per fe ) which may be contrasted to the undoped compound with @xmath33 , and @xmath32 ( 0.5 electrons per fe ) .
the fermi surface calculated in fig .
[ fig : normal](c ) contains a small electron pocket around the z point in addition to the fermi pockets around the m point .
the ldos calculated at @xmath15 in fig.[fig : normal](d ) contains mainly @xmath25 , @xmath26 and @xmath28 orbitals near the fermi energy whereas the contribution from the @xmath27 and @xmath34 orbitals near the fermi energy are negligible .
fe@xmath2se@xmath3 . here
the color codes represent green : @xmath25 , red : @xmath26 , cyan : @xmath27 , blue : @xmath28 , yellow : @xmath35 .
( b ) fermi surface calculated at @xmath15 .
( c ) fermi surface at @xmath36 .
( d ) orbital resolved ldos at @xmath15 in the normal state at an electron doping of @xmath37 e@xmath38/fe . ]
fe@xmath2se@xmath3 . here
the color codes represent green : @xmath25 , red : @xmath26 , cyan : @xmath27 , blue : @xmath28 , yellow : @xmath35 .
( b ) fermi surface calculated at @xmath15 .
( c ) fermi surface at @xmath36 .
( d ) orbital resolved ldos at @xmath15 in the normal state at an electron doping of @xmath37 e@xmath38/fe . ]
fe@xmath2se@xmath3 . here
the color codes represent green : @xmath25 , red : @xmath26 , cyan : @xmath27 , blue : @xmath28 , yellow : @xmath35 .
( b ) fermi surface calculated at @xmath15 .
( c ) fermi surface at @xmath36 .
( d ) orbital resolved ldos at @xmath15 in the normal state at an electron doping of @xmath37 e@xmath38/fe . ]
fe@xmath2se@xmath3 . here
the color codes represent green : @xmath25 , red : @xmath26 , cyan : @xmath27 , blue : @xmath28 , yellow : @xmath35 .
( b ) fermi surface calculated at @xmath15 .
( c ) fermi surface at @xmath36 .
( d ) orbital resolved ldos at @xmath15 in the normal state at an electron doping of @xmath37 e@xmath38/fe . ] the second term of eq.([eq : h ] ) describes the intra - site coulomb interaction @xmath39 which includes the intra - orbital ( inter - orbital ) interaction @xmath40 ( @xmath41 ) , the hund s rule coupling @xmath42 and the pair hopping energy @xmath43 .
we will assume orbital and spin rotational invariance where the relations @xmath44 and @xmath45 hold . the third term in eq.([eq : h ] )
is the superconducting pairing term @xmath46,\ ] ] with the order parameter @xmath47 . here
, @xmath48 is the strength of an effective attraction that is generated from a spin fluctuation exchange mechanism . to this end
, we follow the procedure given e.g. by graser _
et al._@xcite and calculate the singlet pairing vertex @xmath49_{ps}^{tq}.\end{aligned}\ ] ] note that we have retained only those pairing vertices of the form @xmath48 .
this is done only for computational speed since we do not find any qualitative changes in our results by including additional inter - orbital interactions as they are quite small in magnitude .
the pairing interaction strengths are fixed by requiring that @xmath40 and @xmath42 generate a superconducting gap magnitude of @xmath50mev for both the @xmath4-wave and @xmath5-wave symmetry which agrees reasonably well with gap magnitude observed e.g. in arpes experiments.@xcite the obtained pairing symmetry is determined simply by the fermi surface used as input to eq.([eq : rpa ] ) , i.e. the fermi surface in fig.[fig : normal](b ) [ fig.[fig : normal](c ) ] generates @xmath4 [ @xmath5 ] pairing , respectively . in table [ table : gapvals ] we show the self - consistent gap magnitudes @xmath52 on each orbital .
it can be seen that the gap primarily resides on the @xmath28 orbital which we attribute to the intra orbital pairing interaction of the @xmath28 orbital @xmath53 being at least twice that of the @xmath25/@xmath26 orbital @xmath54 .
.real space self - consistent gap values @xmath52 on each orbital for nearest neighbor and next nearest neighbor pairings .
units are in mev . [ cols="<,<,<,<,<,<,<",options="header " , ] we stress that the present calculation is 2d , and for the results discussed below we use a tight - binding band with the fermi surface shown in fig .
[ fig : normal](b ) .
it is only for generation of the pairing couplings @xmath12 that we utilise the fact that @xmath4- or @xmath5-wave pairing is preferred within the rpa spin - fluctuation approach by fermi surfaces corresponding to fig .
[ fig : normal](b ) and fig .
[ fig : normal](c ) , respectively .
this is simply a way to simulate an @xmath5-wave state in the presence of the fermi surface of fig .
[ fig : normal](b )
. such a superconducting phase could be stabilised by additional pair scattering processes than the ones allowed by fig .
[ fig : normal](b ) caused by e.g. grazing hole pockets near the @xmath12 point , or important inter - band processes between hole and electron processes at @xmath36.@xcite the last term in the hamiltonian ( [ eq : h ] ) incorporates the iron vacancies .
we model the vacancies as impurities with sufficiently large onsite potentials that remove any electronic states from their location @xmath55 which adds the potential @xmath56 at the vacancy sites @xmath57 .
after a standard mean - field decoupling of the onsite interaction term we arrive at the following multiband bogoliubov de - gennes equations@xcite @xmath58 where @xmath59.\end{aligned}\ ] ] the local densities and the sc order parameters are obtained self - consistently through iteration of @xmath60 in the following section , we discuss the results of the above procedure for the magnetic , and superconducting properties applicable to iron chalcogenide superconductor k@xmath1fe@xmath2se@xmath3 .
the band structure properties of the iron based superconductors generally include contributions from all five @xmath4-orbitals near the fermi surface .
a simple electron count and arpes results reveals that the 122 iron selenides are strongly electron doped .
we impose a @xmath61 electron doping on the metallic region by adjusting the chemical potential to @xmath62ev . the evaluated fermi surface shape for @xmath15 and @xmath36 are shown in fig . [ fig : normal](b ) and [ fig : normal](c ) , respectively .
they agree well with arpes observations which find only electron pockets at the m points for @xmath15 and a small additional electron pocket around the z point.@xcite in our calculations for the normal metallic state , the onsite coulomb term has been fixed by @xmath63ev and @xmath64 . in fig .
[ fig : normal](d ) we show the orbitally resolved ldos .
it can be seen that the primary contributions to the density of states near the fermi level are from the @xmath28 and @xmath65 orbitals .
though the orbital content can depend on the doping level as well as the presence of iron vacancies , observations on similar alkali doped iron selenide systems identify the above t@xmath66 orbitals to be the dominant contributors of the fermi surface at @xmath15.@xcite .
( b ) real - space map of the magnetic moment in the bafm state .
white : vacancy sites , green : 3.63@xmath67/fe , blue : -3.63@xmath67/fe .
( c ) same as ( b ) but for orbital order @xmath68-@xmath69 .
blue : -0.011 , green : 0.011 . ] + .
( b ) real - space map of the magnetic moment in the bafm state .
white : vacancy sites , green : 3.63@xmath67/fe , blue : -3.63@xmath67/fe .
( c ) same as ( b ) but for orbital order @xmath68-@xmath69 .
blue : -0.011 , green : 0.011 . ] .
( b ) real - space map of the magnetic moment in the bafm state .
white : vacancy sites , green : 3.63@xmath67/fe , blue : -3.63@xmath67/fe .
( c ) same as ( b ) but for orbital order @xmath68-@xmath69 .
blue : -0.011 , green : 0.011 . ]
the presence of an ordered arrangement of iron vacancies with a @xmath9 arrangement in the lattice structure of k@xmath1fe@xmath2se@xmath3 stabilises a block antiferromagnetic state ( bafm ) .
stm measurements , for example , find that these regions are insulating.@xcite additionally , experiments find an electron occupation of 6e@xmath70fe.@xcite in our model calculations , the chemical potential for the bafm state is the same as in the normal metal discussed above . an electron density of 6 e@xmath38/fe is obtained by a coulomb interaction strength of @xmath71ev and @xmath72 in the bafm state .
the stoichiometry for the bafm with @xmath9 can be considered to be k@xmath73fe@xmath74se@xmath3 from an electron and vacancy concentration count .
it has been previously argued by luo _
et al . _
@xcite that a correct description of the magnetic state requires a larger coulomb interaction term compared to what is typically used for pnictides .
in addition , we note that in the phase separated region the normal metal is likely to have a smaller effective on - site coulomb repulsion as otherwise the normal region would undergo a sdw transition to a @xmath75 magnetic state and no such magnetism has been observed . using the self - consistency approach discussed in section 2
we find a stable bafm state with a magnetic moment of 3.63 @xmath67/fe .
the magnetic moment obtained is shown in a real - space map in fig .
[ fig : bafm](b ) .
calculation of the associated ldos yields an insulating gap of around 0.35 ev as shown in fig .
[ fig : bafm](a ) .
this gap agrees reasonably well with the gap magnitude obtained in stm measurements of the bafm regions.@xcite another result that comes out naturally from this calculation is the presence of orbital order defined as @xmath76 as shown in fig .
[ fig : bafm](c ) .
the origin of the orbital order is the orbitally selective suppression of hopping of an electron in a @xmath25 ( @xmath26 ) orbital in the horizontal ( vertical ) direction next to a vacancy location.@xcite -wave superconducting state .
black : total ldos , blue : @xmath28 , red : @xmath65 .
the inset shows the gap anisotropy around the fermi surface centered at @xmath77 .
the @xmath78-axis is angle measured with respect to the @xmath79 direction .
( b ) same as ( a ) but for the @xmath5- wave symmetry . ]
-wave superconducting state .
black : total ldos , blue : @xmath28 , red : @xmath65 .
the inset shows the gap anisotropy around the fermi surface centered at @xmath77 .
the @xmath78-axis is angle measured with respect to the @xmath79 direction .
( b ) same as ( a ) but for the @xmath5- wave symmetry . ]
as described in the model section , superconducting order is stabilised by including the pairing interactions generated within a standard weak coupling rpa method . before turning to the discussion of the ldos near boundary regions between superconducting and bafm regions ,
it is instructive to briefly display the properties of the homogeneous superconductor . in fig .
[ fig : sc ] we show the orbitally resolved ldos for the superconducting state with @xmath4-wave and @xmath5-wave pairing symmetry , respectively . due to the same gap amplitude in the two cases , and
the fact that the ldos is not phase sensitive , the final ldos of the homogeneous phase is very similar in the two states as expected .
the calculated ldos are typical for a nodeless superconducting state and display an apparent `` two - gap '' behavior similar to what has been observed in recent experiments.@xcite the two sets of coherence peaks close to @xmath80mev and @xmath81mev in fig .
[ fig : sc ] results from a significant gap anisotropy on the electron pockets as seen from the insets in figs .
[ fig : sc ] .
the superconducting pairing interaction has been chosen to result in minimum gap amplitudes of about 12mev which is of the order of experimentally observed amplitudes that have been observed to be around 10mev .
-wave superconductor .
the system size is @xmath82 with a bafm of @xmath83 sites placed between site @xmath84 and @xmath85 .
( a ) map of the magnetic moment ( @xmath86fe ) shown between sites @xmath87 to @xmath88 and @xmath89 to @xmath90 .
( b ) same as ( a ) but for the orbital order ( @xmath76 ) .
( c ) orbital resolved averaged superconducting gap magnitude shown between @xmath87 , @xmath91 to @xmath88 , @xmath91 .
the color scheme is the same as in fig .
[ fig : normal ] . ] to @xmath92 for @xmath5-wave gap symmetry .
( b ) same as ( a ) but for @xmath4-wave symmetry of the superconducting order parameter . ]
to @xmath92 for @xmath5-wave gap symmetry .
( b ) same as ( a ) but for @xmath4-wave symmetry of the superconducting order parameter . ]
as mentioned in the introduction , the alkali doped iron selenide materials a@xmath1fe@xmath2se@xmath3 are phase separated , and interfaces between bafm and superconducting regions are abundant and an important situation to understand in detail .
for such an interface , it can be seen from fig .
[ fig : interface1](a ) that the magnetism penetrates only a few lattice spacings into the superconducting region due to the significant amplitude of the magnetisation in the bafm regions .
since the bafm is insulating , the superconducting order parameter also does not exhibit any significant leaking into the bafm regions as seen from fig .
[ fig : interface1](c ) .
this short - range proximity effect is qualitatively consistent with the existence of filamentary superconducting channels near these magnetic regions.@xcite the orbital order near the interface is shown in fig .
[ fig : interface1](b ) .
as seen , at particular sites near the interface , the orbital order undergoes significant enhancement by a factor of about ten compared to the size of the orbital order in the bulk bafm .
this takes place adjacent to the vacancy sites bordering the superconductor causing significant local ldos modulations at these sites .
note that the absolute change in the orbital order near the interface is still weak with @xmath93 . what are the low - energy states near the bafm and superconductor interfaces ? in fig .
[ fig : interface ] we show the evolution of the ldos upon approaching the interface for both the @xmath5- and @xmath4-wave cases .
as seen in both cases , prevalent low - energy in - gap bound states clearly exist , and there is no qualitative difference in the ldos caused by the pairing symmetry of the bulk superconductor .
this is contrary to the case of a single nonmagnetic impurity where zhu _
et al._@xcite showed that the @xmath4-wave state leads to in - gap bound states as opposed to the nnn @xmath5-wave state ( see however ref .
for a more general discussion ) .
the magnetisation of the bafm is crucial for the generation of in - gap states in both fig .
[ fig : interface](a ) and fig .
[ fig : interface](b ) since it breaks time - reversal symmetry and generates bound states for both @xmath5- and @xmath4-wave superconductors .
certainly , in a picture where one thinks of the interface as a row of magnetic and non - magnetic impurities the resulting interference of the multiple generated bound states should generate an in - gap band of states leading to ldos results similar to those shown in fig .
[ fig : interface].@xcite -wave nnn superconductor and a bafm with vacancies forming a ( 110 ) wall . here , @xmath63 , and @xmath64 which leads to a nonmagnetic interface .
( a ) ldos plots for @xmath94 to @xmath84 for @xmath5-wave gap symmetry .
( b ) same as ( a ) but for @xmath4-wave gap symmetry .
( c ) incident wave vector @xmath95 at the ( 110 ) interface and reflected wave vector @xmath96 for @xmath5-wave symmetry .
( d ) same as in ( c ) but for @xmath4-wave symmetry .
blue : @xmath97 , red : -@xmath97 . ] -wave nnn superconductor and a bafm with vacancies forming a ( 110 ) wall . here , @xmath63 , and @xmath64 which leads to a nonmagnetic interface .
( a ) ldos plots for @xmath94 to @xmath84 for @xmath5-wave gap symmetry .
( b ) same as ( a ) but for @xmath4-wave gap symmetry .
( c ) incident wave vector @xmath95 at the ( 110 ) interface and reflected wave vector @xmath96 for @xmath5-wave symmetry .
( d ) same as in ( c ) but for @xmath4-wave symmetry .
blue : @xmath97 , red : -@xmath97 . ] -wave nnn superconductor and a bafm with vacancies forming a ( 110 ) wall . here , @xmath63 , and @xmath64 which leads to a nonmagnetic interface .
( a ) ldos plots for @xmath94 to @xmath84 for @xmath5-wave gap symmetry .
( b ) same as ( a ) but for @xmath4-wave gap symmetry .
( c ) incident wave vector @xmath95 at the ( 110 ) interface and reflected wave vector @xmath96 for @xmath5-wave symmetry .
( d ) same as in ( c ) but for @xmath4-wave symmetry .
blue : @xmath97 , red : -@xmath97 . ] -wave nnn superconductor and a bafm with vacancies forming a ( 110 ) wall . here , @xmath63 , and @xmath64 which leads to a nonmagnetic interface .
( a ) ldos plots for @xmath94 to @xmath84 for @xmath5-wave gap symmetry .
( b ) same as ( a ) but for @xmath4-wave gap symmetry .
( c ) incident wave vector @xmath95 at the ( 110 ) interface and reflected wave vector @xmath96 for @xmath5-wave symmetry .
( d ) same as in ( c ) but for @xmath4-wave symmetry .
blue : @xmath97 , red : -@xmath97 . ] in the case of cuprate @xmath4-wave superconductors it is well - known that a ( 110 ) diagonal surface / interface or grain boundary leads to the generation of a zero - energy resonant state due to a sign change of the gap function between the incoming and scattered momentum vector within a semiclassical picture.@xcite in the present case , a similar scattering geometry which probes the different gap signs of the electron pockets should provide a qualitatively different low - energy ldos between the @xmath5- and @xmath4-wave superconductors . as shown in fig . [
fig : diagint ] , a ( 110 ) wall placed at 45 degrees to the @xmath78-axis indeed leads to a bound state solutions for the @xmath4-wave state unlike the case for the @xmath5-wave gap symmetry .
the origin of this difference is exactly the same as for the standard one - band @xmath4-wave materials.@xcite thus , using scanning probes to locate such dislocations in the fese planes , local stm measurements near such regions could provide important clues to the pairing symmetry of these materials .
we have provided a microscopic real - space study of the interplay between bafm and superconducting domains relevant to alkali doped iron selenide materials a@xmath1fe@xmath2se@xmath3 which have been shown to exhibit significant in - plane and out - of - plane phase separation .
our modelling includes all five iron @xmath4-orbitals and the onsite coulomb interaction treated at the mean field level .
superconductivity is incorporated through the spin - fluctuation rpa approach .
a main result is that low - energy in - gap bound states should exist near interfaces between bafm and superconducting regions irrespective of whether the pairing symmetry of the superconductor is @xmath4- or @xmath5-wave .
by contrast , a non - magnetic ( 110 ) scattering plane only causes low - energy bound states for the @xmath4-wave pairing and should provide a means to distinguish between @xmath4 and @xmath5-wave pairing symmetry of these materials .
b.m.a . and m.n.g .
acknowledge support from the danish council for independent research @xmath98 natural sciences , and the lundbeckfond fellowship ( grant a9318 ) .
100 p. j. hirschfeld , m. m. korshunov , and i. i. mazin , rep .
. phys . * 74 * , 124508 ( 2011 ) .
e. dagotto , arxiv:1210.6501 ( 2012 ) .
j. guo , s. jin , g. wang , s. wang , k. zhu , t. zhou , m. he , and x. chen , phys .
b * 82 * , 180520 ( 2010 ) .
p. zavalij , w. bao , x. f. wang , j. j. ying , x. h. chen , d. m. wang , j. b. he , x. q. wang , g. f. chen , p .- y .
hsieh , q. huang , and m. a. green , phys .
b * 83 * , 132509 ( 2011 ) .
w. bao , q. huang , g. f. chen , m. a. green , d. m. wang , j. b. he , x. q. wang , and y. qiu , chin .
* 28 * , 086104 ( 2011 ) . v. yu pomjakushin , e. v. pomjakushina , a krzton - maziopa , k conder , and z. shermadini , j. phys .
condens . matter , * 23 * , 156003 ( 2011 ) .
f. ye , s. chi , wei bao , x. f. wang , j. j. ying , x. h. chen , h. d. wang , c. h. dong , and m. fang , phys .
107 * , 137003 ( 2011 ) .
i. , mazin physics * 4 * , 26 ( 2011 ) .
f. chen , m. xu , q. q. ge , y. zhang , z. r. ye , l. x. yang , juan jiang , b. p. xie , r. c. che , m. zhang , a. f. wang , x. h. chen , d. w. shen , j. p. hu , and d. l. feng , phys . rev .
x * 1 * , 021020 ( 2011 ) .
n. lazarevi , m. abeykoon , p. w. stephens , hechang lei , e. s. bozin , c. petrovic , and z. v. popovi , phys .
b * 86 * , 054503 ( 2012 ) .
r. h. yuan , t. dong , y. j. song , p. zheng , g. f. chen , j. p. hu , j. q. li , and n. l. wang , sci
. rep . * 2 * , 221 ( 2012 ) .
x. ding , d. fang , z. wang , h. yang , j. liu , q. deng , g. ma , c. meng , y. hu , and h.- h. wen , arxiv:1301.2668 . w. li , h. ding , p. deng , k. chang , c. l. song , k. he , l. l. wang , x. c. ma , j. p. hu , x. chen , and q. k. xue , nature phys . * 8 * 126 ( 2012 ) . w. li , h. ding , z. li , p. deng , k. chang , k. he , s. ji , l. wang , x. ma , j .-
hu , x. chen , and q .- k .
xue , phys .
lett . * 109 * , 057003 ( 2012 ) .
y. j. yan , m. zhang , a. f. wang , j. j. ying , z. y. li , w. qin , x. g. luo , j. q. li , j. hu , and x. h. chen , sci .
* 2 * , 212 ( 2012 ) .
s. landsgesell , d. abou - ras , t. wolf , d. alber , and k. prokes , phys .
b * 86 * , 224502 ( 2012 ) .
a. charnukha , a. cvitkovic , t. prokscha , d. prpper , n. ocelic , a. suter , z. salman , e. morenzoni , j. deisenhofer , v. tsurkan , a. loidl , b. keimer , a. v. boris , phys .
. lett . * 109 * , 017003 ( 2012 ) .
a. ricci , n. poccia , g. campi , b. joseph , g. arrighetti , l. barba , m. reynolds , m. burghammer , h. takeya , y. mizuguchi , y. takano , m. colapietro , n. l. saini , and a. bianconi , phys .
b. * 84 * 060511 ( 2011 ) .
y. liu , q. xing , k. w. dennis , r. w. mccallum , and t. a. lograsso , phys .
b * 86 * , 144507 ( 2012 ) .
y. texier , j. deisenhofer , v. tsurkan , a. loidl , d. s. inosov , g. friemel , j. bobroff , phys .
lett . * 108 * , 237002 ( 2012 )
. y. zhang , x. yang , m. xu , z. r. ye , f. chen , c. he , h. c. xu , j. jiang , b. p. xie , j. j. ying , x. f. wang , x. h. chen , j. p. hu , m. matsunami , s. kimura , and d. l. feng , nature mat . * 10 * , 273 ( 2011 ) .
y. zhang , l. x. yang , m. xu , z. r. ye , f. chen , c. he , h. c. xu , j. jiang , b. p. xie , j. j. ying , x. f. wang , x. h. chen , j. p. hu , m. matsunami , s. kimura , and d. l. feng , nature mat . * 10 * , 273 ( 2011 ) .
t. qian , x .-
wang , w .- c .
jin , p. zhang , p. richard , g. xu , x. dai , z. fang , j .-
chen , h. ding , phys .
lett . * 106 * , 187001 ( 2011 ) .
i. i. mazin , d. j. singh , m. d. johannes , and m. h. du , phys .
057003 ( 2008 ) . t. a. maier , s. graser , p. j. hirschfeld , and d. j. scalapino , phys .
b * 83 * , 100515 ( 2011 ) . f. a. wang , f. yang , m. gao , z .- y .
lu , t. xiang , and d.- h. lee , europhys .
* 93 * , 57003 ( 2011 ) . c. fang ,
wu , r. thomale , b. a. bernevig , and j. p.
hu , phys .
x * 1 * , 011009 ( 2011 ) .
i. i. mazin phys .
b * 84 * , 024529 ( 2011 ) .
b. zeng , b. shen , g. f. chen , j. b. he , d. m. wang , c. h. li , and h. h. wen , phys .
b * 83 * , 144511 ( 2011 ) .
m. xu , q.q .
ge , r. peng , z. r. ye , j. jiang , f. chen , x. p. shen , b. p. xie , y. zhang , a. f. wang , x. f. wang , x. h. chen , and d. l. feng , phys .
b. * 85 * , 220504 ( 2012 ) .
d.- y. liu , y.- m. quan , z. , zeng and l.- j. zou , physica b * 407 * , 1139 ( 2012 ) .
s. graser , t. a. maier , p. j. hirschfeld , and d. j. scalapino , new j. phys . * 11 * , 025016 ( 2009 ) . w. yu , l. ma , j. b. he , d. m. wang , t .- l .
xia , g. f. chen , and w. bao , phys .
* 106 * , 197001 ( 2011 ) .
zhu , r. yu , a. v. balatsky , and q. si , phys . rev
. lett . * 107 * , 167002 2011 . j .-
x . zhu , and a. r. bishop , europhys .
100 * , 37004 ( 2012 ) .
y. k. li , c. y. shen , h. j. guo , c. lv , x. j. yang , l. zhang , y. k. luo , g. h. cao , and z. a. xu j. phys . : condens .
matter * 24 * , 232202 ( 2012 ) .
m. n. gastiasoro and b. m. andersen , j. supercond .
( 2013 ) . c. fei , g. qingqin , x. min , z. yan , s. xiaoping , li wei , m. masaharu , k. shin - ichi , h. j. hu , and f. d. lai , chin .
sci . bull . * 57 * 30 ( 2012 ) . c. cao , and j. dai , phys .
rev . lett . * 107 * , 056401 ( 2011 ) .
w. li , h. ding , p. deng , k. chang , c. song , k. he , l. wang , x. ma , j .- p .
hu , x. chen , and q .- k .
xue , nat .
* 8 * , 126 ( 2012 ) .
q. luo , a. nicholson , j. riera , d.- x. yao , a. moreo , and e. dagotto , phys .
b * 84 * , 140506 ( r ) ( 2011 ) .
w. lv , w.- c. lee , and p. phillips , phys .
b * 84 * , 155107 ( 2011 ) .
r. beaird , i. vekhter , and j.- x. zhu , phys .
b. * 86 * , 140507 ( r ) ( 2012 ) .
d. k. morr and n. a. stavropoulos , phys .
b * 66 * , 140508 ( 2002 ) . b. m. andersen and p. hedegrd , phys .
b * 67 * , 172505 ( 2003 ) . c. r. hu , phys . rev .
* 72 * , 1526 ( 1994 ) .
y. tanaka and s. kashiwaya , phys .
lett . * 74 * , 3451 ( 1995 ) .
l. j. buchholtz , m. palumbo , d. rainer , and j. a. sauls , j. low temp
* 101 * , 1099 ( 1995 ) .
b. m. andersen , i. v. bobkova , p. j. hirschfeld , and yu .
s. barash , phys .
b * 72 * , 184510 ( 2005 ) . | the high - t@xmath0 alkali doped iron selenide superconductors k@xmath1fe@xmath2se@xmath3 have been recently shown to be intrinsically phase separated into fe vacancy ordered block antiferromagnetic regions and superconducting regions at low temperatures . in this work , we use a microscopic five orbital hubbard model to obtain the electronic low - energy states near the interfaces between block antiferromagnets and superconductors .
it is found that abundant low - energy in - gap bound states exist near such interfaces irrespective of whether the superconductor has @xmath4- or @xmath5-wave pairing symmetry .
by contrast , it is shown how nonmagnetic scattering planes can provide a natural means to distinguish between these two leading pairing instabilities of the k@xmath1fe@xmath2se@xmath3 materials . |
there are several known data structures that answer distance queries in planar graphs .
we survey them below .
all of these data structures use the following basic idea .
they split the graph into _ pieces _ , where each piece is connected to the rest of the graph only through its _
boundary vertices_. then , every path can go from one piece to another only through these boundary vertices .
the different data structures find different efficient ways to store or compute the distance between two boundary vertices or between a boundary vertex and a non - boundary vertex .
frederickson @xcite gave the first data structures that can answer distance queries in planar graphs fast .
he gave a data structure of linear size with @xmath1 preprocessing time that can find the shortest path tree rooted at any vertex in @xmath2 time , where @xmath3 is the number of vertices in the graph .
this leads also to an @xmath4 solution to the all - pairs shortest - paths problem , and implies a distance query data structure of size @xmath4 with @xmath5 query time .
feuerstein and marchetti - spaccamela @xcite modified the data structure of @xcite and showed how to decrease the time of a distance query by increasing the preprocessing time .
they do not provide an analysis of their data structure in terms of preprocessing time , storage space , and query time , but they show the total running time of @xmath6 queries , which is @xmath7 , @xmath8 , @xmath9 , @xmath10 for @xmath11 , @xmath12 , @xmath13 , @xmath14 , respectively .
this solution actually consists of three different data structures for the three cases @xmath11 , @xmath15 and @xmath16 , where the data structure for the first case is the one of @xcite .
henzinger , klein , rao and subramanian @xcite gave an @xmath2 algorithm for the single - source shortest path problem .
this implies a trivial distance query data structure , which uses the algorithm , and takes @xmath2 space and query time .
djidjev @xcite gave three data structures .
we will use the specific section number in @xcite 3 ,
4 , or 5 , to refer to each one of them .
the first one ( * ? ? ?
* ( 3 ) ) works for @xmath17 $ ] and has size @xmath18 , @xmath18 preprocessing time , and @xmath19 query time .
the same data structure was also presented by arikati et al .
this data structure is similar to the two data structures of @xcite , but takes advantage of the algorithm of @xcite to get a better preprocessing time .
the second ( * ? ? ?
* ( 4 ) ) works for @xmath20 $ ] and has size @xmath18 , @xmath21 preprocessing time , and @xmath19 query time .
the third data structure ( * ? ? ? * ( 5 ) ) works for @xmath22 $ ] , has size @xmath18 , @xmath21 preprocessing time , and @xmath23 query time.$ ] with @xmath24 query time , however the same data structure works for a larger range of @xmath25 , and the running time is actually @xmath23.[ftnt : d96 ] ] chen and xu @xcite presented a data structure with the same time and space bounds .
, the bounds here are derived by setting @xmath26 in the bounds that appear below lemma 28 ( page 477 ) of @xcite ; the bounds stated in @xcite depend on the minimum number @xmath27 of faces required to cover all vertices of the graph ( @xmath27 is called _ face - on - vertex covering _ ) , these bounds are obtained using hammock decomposition @xcite , which can be applied to any planar distance data structure.[ftnt : cx00 ] ] fakcharoenphol and rao @xcite gave a data structure with @xmath1 space and @xmath28 preprocessing time and @xmath29 query time .
klein @xcite improved the preprocessing time of the data structure to @xmath30 .
cabello @xcite presented a data structure that uses @xmath18 space and can be constructed in @xmath18 time for @xmath31 $ ] , and answers a query in @xmath32 time .
this data structure can answer @xmath6 queries in a total of @xmath33 time .
if the queries are known in advance , the algorithm of @xcite avoids storing the entire structure , and uses only @xmath34 space .
other data structures exist for outerplanar graphs ( some use the fact that these graphs have treewidth @xmath35 ) and planar graphs with small face - on - vertex covering @xcite , for planar graphs with bounded edge lengths @xcite , and for approximate distance queries in planar graphs @xcite .
in this paper we present three new data structures for the problem : * section [ sec : a ] : a data structure with @xmath2 space , @xmath1 preprocessing time , and @xmath36 query time , for a constant @xmath37 .
this data structure has the best known query time achievable with linear space .
this also improves the total running time for answering @xmath6 distance queries when @xmath6 is @xmath38 and @xmath39 , and if we limit ourselves to data structures with @xmath34 space then the upper bound on the range of @xmath6 grows to @xmath40 .
the data structure is based on the data structure of @xcite , by replacing the _ recursive decomposition _ of @xcite with recursive _ @xmath41-decomposition _ as in @xcite .
as the data structure of @xcite , our data structure also generalizes to graphs embedded in a surface of bounded genus .
* section [ sec : b ] : for @xmath42 $ ] , a data structure with @xmath18 space , @xmath43 preprocessing time , and @xmath23 query time .
this data structure matches the storage space and query time of ( * ? ? ?
* ; * ? ? ?
* ( 5 ) ) , which is the best query time for this range of storage space , with a better preprocessing time for @xmath44 . the data structure is obtained by combining a preprocessing algorithm similar to @xcite with a data structure similar to ( * ? ? ?
* ( 5 ) )
. * section [ sec : c ] : for @xmath42 $ ] , a data structure with @xmath18 space , requiring @xmath45 preprocessing time , and @xmath46 query time .
this data structure has the fastest query time for the same range of storage space as the previous data structure , but with a longer preprocessing time .
the fast query time is obtained using an efficient minimum search in a monge matrix .
the different data structures are summarized in table [ table : sum ] .
.comparison of distance query data structures for planar graphs .
time bounds are expressed as a function of the storage space .
the data structures are ordered by decreasing storage space and then by decreasing query time . [ cols="^,^,^,^",options="header " , ]
we consider a directed simple planar graph @xmath47 .
we let @xmath48 , and by euler s formula @xmath49 .
we assume that @xmath50 is given with a fixed planar embedding , in other words it is a _
plane graph_. without loss of generality we assume that @xmath50 is a triangulated , bounded degree graph ; this assumption is required by algorithms that we use and are described in sect .
[ sec : decomp ] and [ sec : dense ] .
we assume that @xmath50 is connected , since we can handle each connected component separately .
every edge in @xmath51 has a non - negative _
length_. the length of a _ path _ is the sum of lengths of all of its edges .
the _ distance _ from a vertex @xmath52 to a vertex @xmath53 is the minimum length of a path from @xmath52 to @xmath53 . with additional @xmath54 preprocessing time we can allow negative edge lengths as well ,
see @xcite for details .
let @xmath55 be subgraphs of @xmath50 .
we write @xmath56 to denote the distance from @xmath52 to @xmath53 in @xmath57 .
the graph @xmath58 is the subgraph induced by @xmath59 . for short we denote @xmath60 .
a _ decomposition _ of a planar graph @xmath50 is a set of subgraphs of @xmath50 , such that each edge is in exactly one subgraph and each vertex of @xmath50 is in at least one subgraph .
each of the subgraphs which define the decomposition is called a _ piece_.
a vertex @xmath53 is a _ boundary vertex _ of a piece @xmath61 , if @xmath62 and @xmath53 is incident to some edge not in @xmath63 .
the set of all boundary vertices of @xmath61 is the _ boundary _ of @xmath61 , denoted by @xmath64 .
a _ hole _ is a face of @xmath61 ( including the external face ) that is not a face of @xmath50 . for a hole @xmath57
we denote by @xmath57 also the subgraph of @xmath50 inside @xmath57 .
boundary walk _ of @xmath61 is a facial walk of @xmath61 around a hole @xmath57 . for a piece @xmath61 with hole @xmath57
we denote @xmath65 = \partial b \cap v(h)$ ] .
a vertex of @xmath66 is an _ internal vertex _ of @xmath61 .
all distance query data structures mentioned in the introduction decompose the planar graph .
they take advantage of the fact that a path can go from one piece to another only through boundary vertices . a _ recursive decomposition _
@xcite is obtained by starting with @xmath50 itself being the only piece in level 0 of the decomposition . at each level
, we split each piece @xmath61 with @xmath67 vertices and @xmath68 boundary vertices that has more than one edge into two pieces , each with at most @xmath69 vertices and at most @xmath70 boundary vertices .
we require that the boundary vertices of a piece @xmath61 are also boundary vertices of the subpieces of @xmath61 .
each piece @xmath61 in the decomposition has @xmath71 boundary vertices .
an _ @xmath41-decomposition _
@xcite is a decomposition of the graph into @xmath72 pieces , each of size at most @xmath41 with @xmath73 boundary vertices .
fakcharoenphol and rao @xcite showed how to find a recursive decomposition of @xmath50 , such that each piece is connected and has at most a constant number of holes .
they use these two properties for their distance algorithm .
the construction of the decomposition takes @xmath1 time using @xmath1 space , and is done by recursively applying the separator algorithm of miller @xcite .
frederickson @xcite showed how to find an @xmath41-decomposition in @xmath1 time and @xmath2 space by recursively applying the separator algorithm of lipton and tarjan @xcite .
thus , an @xmath41-decomposition is a limited type of recursive decomposition where we stop the recursion earlier ( when we get to pieces of size @xmath41 ) , and do not store all the levels of the recursion ( we store only the leaves ) .
cabello @xcite combined the two constructions of @xcite ( using @xcite instead of @xcite ) and constructed an @xmath41-decomposition with the properties that the number of holes per piece is bounded by a constant , and that each piece is connected . in sect .
[ sec : a ] we use a combination of recursive decomposition and @xmath41-decomposition we decompose the graph recursively , but we decompose each piece into @xmath72 pieces instead of 2 . in sect .
[ sec : b ] we use @xmath41-decomposition . in sect .
[ sec : c ] we use @xmath41-decomposition as well , there we take advantage of the fact that the construction of an @xmath41-decomposition is the same as of a recursive decomposition , which was stopped earlier . fakcharoenphol and rao @xcite define the _ dense distance graph _ of a recursive decomposition . for each piece
@xmath61 in the recursive decomposition they add a piece to the dense distance graph that contains the vertices of @xmath64 and for every @xmath74 an edge from @xmath52 to @xmath53 whose length is @xmath75 .
the multiple - source shortest paths algorithm of klein @xcite finds @xmath6 distances where the sources of all of them are on the same face in @xmath76 time .
therefore , using @xcite it takes @xmath77 time to find the part of the dense distance graph that corresponds to a piece @xmath61 ( recall that @xmath78 and @xmath61 has a constant number of holes ) .
it thus takes @xmath30 time to construct the dense distance graph over all pieces of the recursive decomposition .
every single edge defines a piece in the base of the recursion , so it is clear that the distance from @xmath52 to @xmath53 in the dense distance graph is the same as the distance between these two vertices in the original graph .
fakcharoenphol and rao noticed that in order to find the distance from @xmath52 to @xmath53 we do not have to search the entire dense distance graph , but that it suffices to consider only edges that correspond to shortest paths between boundary vertices in a limited number of pieces .
the pieces are these containing either @xmath52 or @xmath53 , and their siblings in the recursive decomposition .
there are @xmath79 such pieces with a total of @xmath80 boundary vertices .
fakcharoenphol and rao gave an implementation of dijkstra s algorithm that runs over a subgraph of the dense distance graph with @xmath81 vertices , defined by a partial set of the pieces in the recursive decomposition , in @xmath82 time .
this gives the @xmath83 query time of their data structure .
we use dense distance graphs in two of our data structures ( sect .
[ sec : a ] and [ sec : c ] ) . in both cases
it is on a variant of recursive decomposition , as discussed above .
a @xmath84 matrix @xmath85 satisfies the _ monge property _ if for every two rows @xmath86 and two columns @xmath87 , @xmath85 satisfies @xmath88
. we can find the minimum element of @xmath85 by transposing , negating and reversing @xmath85 , and using the smawk algorithm @xcite for row - maxima on the resulting totally monotone matrix . if we do not store @xmath85 explicitly , but are able to retrieve each entry in @xmath5 time this takes @xmath89 time . note that if we add a constant to an entire row or to an entire column of a matrix with the monge property , then the property remains .
consider two disjoint sets @xmath90 and @xmath91 of consecutive boundary vertices on a boundary walk of some piece @xmath61 .
rank the vertices of @xmath90 from @xmath92 to @xmath93 according to their order around the boundary walk , and rank the vertices of @xmath91 from @xmath94 to @xmath95 according to their order in the opposite direction around the boundary walk . for
@xmath86 and @xmath87 , the shortest path from @xmath96 to @xmath97 inside @xmath61 and the shortest path from @xmath98 to @xmath99 inside @xmath61 must cross each other .
let @xmath100 be a vertex common to both paths .
then , @xmath101 ( see fig . [
fig : monge ] ) .
therefore , the matrix @xmath85 such that @xmath102 has the monge property . the monge property was first used explicitly for distance queries in planar graphs by @xcite .
@xmath103 a _ partial matrix _ is a matrix that may have some blank entries . in a _ falling staircase matrix _
the non - blank entries are consecutive in each row starting not before the first non - blank entry of the previous row and ending at the end of the row ( see fig .
[ fig : stair ] ) , _ inverse falling staircase matrix _ is defined similarly by exchanging the positions of the non - blanks and the blanks .
aggarwal and klawe @xcite find the minimum of an ( inverse ) falling staircase matrix whose non - blank entries satisfy the monge property in @xmath104 time by filling the blanks with large enough values to create a monge matrix . in sect .
: c ] we use this tool for finding the minimum of two staircase matrices whose non - blank entries satisfy the monge property .
in this section we present a data structure with linear space , almost linear preprocessing time , and query time faster than any previous data structure of linear space .
we generalize the data structure of fakcharoenphol and rao @xcite by combining recursive decomposition of the graph with @xmath41-decomposition .
this is similar to the way that mozes and wulff - nilsen @xcite improved the shortest path algorithm of klein , mozes and weimann @xcite .
mozes and sommer @xcite have independently obtained a similar result .
we find an @xmath41-decomposition of @xmath50 into @xmath105 pieces , and then we recursively decompose each piece into @xmath27 subpieces , until we get to pieces with a single edge .
the depth of the decomposition is @xmath106 where at level @xmath107 we have @xmath108 pieces , each of size @xmath109 and with @xmath110 boundary vertices .
constructing this recursive decomposition takes @xmath111 time .
an alternative way to describe this decomposition is to perform a recursive decomposition on @xmath50 while storing only levels @xmath112 for @xmath113 of the recursion tree and the leaves of the recursion ( the pieces containing single edges ) .
we compute the dense distance graph for the recursive decomposition , in the same way as in the data structure of @xcite .
that is , we compute the distance between every pair of boundary vertices in each piece . using the algorithm of klein @xcite this takes @xmath1 time for each level , and a total of @xmath111 time .
the size of dense distance graph over our recursive decomposition is @xmath114 .
when a distance query from @xmath52 to @xmath53 arrives , we use the dijkstra implementation of @xcite to answer it .
we run the algorithm on the subgraph of the dense distance graph that includes all the pieces that contain either @xmath52 or @xmath53 , and the @xmath115 siblings in the recursive decomposition of each such piece .
we require the sibling pieces because the shortest path can get out of a piece @xmath61 into a sibling of @xmath61 without getting out of any piece that contains @xmath61 .
therefore , the number of boundary vertices involved in each distance query is @xmath116 .
hence the query time using the algorithm of @xcite is @xmath117 .
we conclude that for a planar graph with @xmath3 vertices and any @xmath118 $ ] , we can construct in @xmath119 time a data structure of size @xmath120 that computes the distance between any two vertices in @xmath117 time .
if we set @xmath121 we get exactly the data structure of @xcite .
if we set @xmath122 for a constant @xmath123 we get : [ col : a ] for a planar graph with @xmath3 vertices and any constant @xmath37 , we can construct in @xmath1 time a data structure of size @xmath2 that computes the distance between any two vertices in @xmath36 time .
the total time for @xmath6 distance queries is @xmath124 and the required space is @xmath34 .
this improves the fastest time for @xmath6 distance queries for @xmath125 and @xmath126 simultaneously . among data structures that require only @xmath34 space ,
the upper bound on the range of @xmath6 is @xmath40 .
fakcharoenphol and rao @xcite noted that smith suggested that their algorithm can be generalized to graphs of bounded genus . if a graph @xmath50 with bounded vertex degree
is embedded in an orientable surface of genus @xmath127 , then @xcite showed how to find a _ planarizing set _ of @xmath128 edges whose removal from the graph makes the graph planar , in @xmath129 time .
we use the planarizing set for the first decomposition of the graph , and combine the dijkstra implementation of @xcite with standard implementation using a heap for the topmost pieces in the recursion .
we get that the bounds of theorem [ col : a ] apply also to graphs embedded in an orientable surface of a fixed genus .
in this section we present a data structure that matches the space - query time tradeoff of the data structures of djidjev ( * ? ? ? * ( 5 ) ) and chen and xu @xcite with the preprocessing time of the data structure of cabello @xcite .
our data structure combines parts of the data structures of ( * ? ? ?
* ( 5 ) ) and of @xcite .
first , we construct an @xmath41-decomposition of @xmath50 in @xmath1 time , for some parameter @xmath130 . for each piece
@xmath61 our data structure has three parts : a. the distances @xmath131 and @xmath132 for every @xmath133 and @xmath62 .
b. a data structure that reports @xmath75 in @xmath73 time for @xmath134 . c. for each hole @xmath57 of @xmath61 we store @xmath135 for every
@xmath136 $ ] and @xmath137 such that @xmath53 is a boundary vertex of some piece contained in @xmath57 .
part ( i ) is from the data structure of cabello @xcite .
the construction of this part requires @xmath138 time and @xmath139 space per piece @xcite . part ( ii ) was used both by djidjev ( * ? ? ? * ( 5 ) ) and by cabello @xcite .
this is the data structure of ( * ? ? ?
* ; * ? ? ?
* ( 3 ) ) with @xmath140 , its construction takes @xmath141 time and space per piece . part ( iii ) is from the data structure of ( * ? ? ? * ( 5 ) ) , but we construct it more efficiently .
we find the distances for this part using the multiple - source shortest paths algorithm of klein @xcite for every boundary walk .
the required space per piece for part ( iii ) is @xmath2 and the preprocessing time is @xmath1 .
since there are @xmath72 pieces , each with a constant number of holes and @xmath73 boundary vertices , constructing the three parts takes @xmath142 time and @xmath143 space .
let @xmath144 be a query pair .
we use the data structure of this section to find @xmath131 in @xmath145 time .
if @xmath52 and @xmath53 are in the same piece then we find the distance from @xmath52 to @xmath53 using parts ( i ) and ( ii ) of the data structure with the query algorithm of @xcite in @xmath73 time ( see details in sect .
[ sec : cq ] below ) .
if @xmath52 and @xmath53 are in different pieces then we find the distance using parts ( i ) and ( iii ) with the query algorithm of ( * ? ? ?
* ( 5 ) ) in @xmath145 time ( see details in appendix [ apx : b ] ) .
we conclude that for a planar graph with @xmath3 vertices and any @xmath146 , we can construct in @xmath147 time a data structure of size @xmath148 that computes the distance between any two vertices in @xmath149 time .
the sum @xmath150 minimizes at @xmath151 , and for @xmath152 we get : for a planar graph with @xmath3 vertices and @xmath42 $ ] , we can construct in @xmath43 time a data structure of size @xmath18 that computes the distance between any two vertices in @xmath23 time .
in this section we present a data structure with an improved query time , for the same range of space bounds as in the previous section . in return
, the preprocessing time is higher . for this purpose
we use minimum search in monge matrices .
while previous planar distance data structures have taken advantage of the monge property before , this is the first to use fast minimum search in a monge matrix with the smawk algorithm @xcite .
again , we construct an @xmath41-decomposition of @xmath50 .
assume that we want to find the distance from a vertex @xmath52 to a vertex @xmath53 that are in two different pieces .
let @xmath61 and @xmath153 be the different pieces that contain @xmath52 and @xmath53 respectively , let @xmath57 and @xmath154 be the holes of @xmath61 and @xmath153 that contain @xmath53 and @xmath52 respectively , and let @xmath155 $ ] and @xmath156 $ ] .
let @xmath157 be the subgraph of @xmath50 contained both in @xmath57 and in @xmath154 .
we assume without lost of generality that @xmath158 contains the infinite face .
see fig .
[ fig : subgraphj ] . .
the vertex @xmath52 is in the piece @xmath61 , and @xmath53 is in @xmath153 .
the subgraph @xmath158 is between @xmath61 and @xmath153 .
every path from @xmath52 to @xmath53 must contain a vertex from the boundary walks @xmath90 of @xmath61 and a vertex from the boundary walk @xmath91 of @xmath153 . ] the shortest path from @xmath52 to @xmath53 must contain a vertex @xmath159 and a vertex @xmath160 ( it is possible that @xmath161 ) .
we assume that there is no internal vertex of @xmath61 or @xmath153 between @xmath162 and @xmath163 in this path , since otherwise we can replace @xmath162 with a later vertex .
therefore , @xmath164 .
our goal then is to find @xmath165 that minimize @xmath166 . for a particular order of @xmath90 and of @xmath91 , which we specify below
, let @xmath85 be the matrix such that @xmath167 , and @xmath168 be the matrix such that @xmath169 .
we show how to order the members of @xmath90 and @xmath91 such that @xmath85 decomposes into two staircase matrices , each with the monge property .
since @xmath170 is fixed for a fixed @xmath96 , and @xmath171 is fixed for a fixed @xmath99 , then @xmath168 also consists of two staircase matrices with the monge property .
thus we can use the algorithm of aggarwal and klawe @xcite to find the minimum entry of @xmath168 , which is the desired distance . for every @xmath159 we define the _ leftmost shortest path from @xmath162 to @xmath91 _ , denoted by @xmath172 as follows .
we add to the embedding of @xmath158 a vertex @xmath173 inside @xmath61 and connect it with an edge to @xmath162 , and a vertex @xmath174 inside @xmath153 and connect it with edges from every vertex of @xmath91 ( recall the internal vertices of @xmath61 and @xmath153 are not in @xmath158 ) .
we set the length of all new edges to be @xmath175 .
an edge @xmath176 is called _ tight _ if the length of @xmath177 is equal to @xmath178 , that is if @xmath177 is on some shortest path from @xmath173 to @xmath179 .
we remove all non - tight edges from the graph , and perform a _ left - first search _ from @xmath173 until we find @xmath174 ( i.e. we perform a depth - first search from @xmath173 , and visit the edges outgoing from a specific vertex according to their left - to - right order , see also @xcite ) .
let @xmath172 be the path we obtain by removing the first and the last edges of the leftmost path we found from @xmath173 to @xmath174 , and let @xmath180 be the last vertex of @xmath172 .
note that @xmath172 is a shortest path from @xmath162 to @xmath181 .
the reason we added @xmath173 is to decide between two paths that diverge at @xmath162 itself , the reason we added @xmath174 is to decide between two paths such that one is a prefix of the other . note that @xmath172 may contain more than one vertex of @xmath91 .
moreover , even if @xmath182 it is not necessarily true that @xmath183 . see fig .
[ fig : leftfs ] . .
only tight edges of @xmath158 are shown , @xmath61 and @xmath153 are _ shaded_. the leftmost path from @xmath173 to @xmath174 is _ bold _ , @xmath184 , @xmath172 is its subpath between @xmath162 and @xmath163 . directions of the edges are not shown . ] fix some arbitrary vertex of @xmath90 to be @xmath92 and rank the other vertices of @xmath90 in a clockwise order .
let @xmath185 , and rank the vertices of @xmath91 in a counterclockwise order .
let @xmath186 .
[ lem : c ] let @xmath187 and @xmath188 .
there is a shortest path @xmath189 in @xmath158 from @xmath96 to @xmath99 such that either @xmath189 does not cross @xmath190 , or every prefix of @xmath189 crosses @xmath190 from the left side of @xmath190 to its right side at most once more than it crosses @xmath190 from its right side to its left side .
assume that every shortest path from @xmath96 to @xmath99 in @xmath158 crosses @xmath190 , and let @xmath189 be such a path .
first , we can assume that at the first time that @xmath189 emanates from @xmath190 it emanates from its right side
if @xmath189 emanates from the left side of @xmath190 at a vertex @xmath100 , we can replace the suffix of @xmath190 that starts at @xmath100 with the suffix of @xmath189 and get a path from @xmath92 to @xmath99 which is to the left of @xmath190 , contradicting its definition as @xmath172 ( see fig . [
fig : qxp](a ) ) .
second , we may assume that if @xmath189 meets @xmath190 at a vertex @xmath100 , and then again at a vertex @xmath191 such that @xmath191 is after @xmath100 also in @xmath190 , then the subpath of @xmath189 between @xmath100 and @xmath191 is the same subpath as in @xmath190 , since otherwise we can replace the subpath of @xmath189 between @xmath100 and @xmath191 with the subpath of @xmath190 ( see fig .
[ fig : qxp](b ) ) .
last we may assume that in two consecutive times that @xmath189 crosses @xmath190 it does so from different directions , since if the same direction is repeated twice , and by the previous observation the second crossing precedes the first crossing in their order in @xmath190 , then @xmath189 must cross itself ( see fig .
[ fig : qxp](c ) ) . from these three observations
the lemma follows .
+ let @xmath192 be the partial matrix of @xmath85 where @xmath193 is non - blank if there is a shortest path in @xmath158 from @xmath96 to @xmath99 that does not cross @xmath190 , and let @xmath194 be the partial matrix of @xmath85 where @xmath195 is non - blank if every shortest path from @xmath96 to @xmath99 crosses @xmath190 .
the partial matrix @xmath192 has the monge property , we get this by cutting open @xmath158 along @xmath190 and using the claim from sect .
[ sec : monge ] ( see @xmath196 in fig . [
fig : qxp](d ) ) .
a similar argument ( by taking two copies of @xmath158 open at @xmath190 and `` gluing '' the right side of one of them to the left side of the other ) shows that @xmath194 also has the monge property ( see @xmath197 in fig .
[ fig : qxp](d ) ) .
the non - blank entries of row @xmath107 in @xmath192 are from @xmath198 to @xmath185 , and the rest of the row is in @xmath194 .
the partial matrix @xmath192 is a falling staircase matrix and @xmath194 is an inverse falling staircase matrix , since @xmath199 , the leftmost path from @xmath200 to @xmath91 , can not cross the path @xmath201 from its right side to its left ( this is similar to the illustration in fig .
[ fig : qxp](a ) ) .
let @xmath202 and @xmath203 be the corresponding staircase matrices partial to @xmath168 ( with same blank entries as @xmath192 and @xmath194 respectively ) , both of them have the monge property .
in order use the insights above , we take parts ( i ) and ( ii ) of the data structure of the previous section together with the following two new parts , where @xmath204 and @xmath158 are as defined above ( part ( iii ) is not necessary ) : a. for each two pieces @xmath61 and @xmath153 we store @xmath205 for every @xmath159 , @xmath160 .
b. for each two pieces @xmath61 and @xmath153 , and every @xmath159 , we store @xmath181 .
notice that @xmath205 and @xmath181 depend on the specific pieces @xmath61 and @xmath153 .
since there are @xmath206 pieces , each piece has @xmath207 boundary vertices , and each vertex is on the boundary of a constant number of pieces , the total space required for part ( iv ) is @xmath208 and for part ( v ) is @xmath209 .
this does not increase the total space complexity of the data structure which remains @xmath148 .
we have an @xmath41-decomposition of @xmath50 obtained by a recursive decomposition , where we decomposed each piece into two pieces using a separator and stopped the recursion at pieces of size @xmath210 , from which we took only the pieces that correspond to leaves of the recursion tree .
now we will take all the pieces of the entire recursive decomposition which defined the @xmath41-decomposition .
we build a dense distance graph for @xmath50 based on this recursive decomposition using the algorithm of fakcharoenphol and rao @xcite with the improvement of klein @xcite in @xmath30 time and @xmath1 space . for two fixed pieces @xmath61 and
@xmath153 we use a subgraph of the dense distance graph to compute the distances from vertices of @xmath90 to vertices of @xmath91 in @xmath158 .
we should choose carefully a set @xmath25 of pieces of the recursive decomposition to use , we must obey three rules we can not take any piece of the @xmath79 pieces that contain either @xmath61 or @xmath153 , we should cover all the paths from @xmath90 to @xmath91 , and the total number of boundary vertices of pieces in @xmath25 should not be too large .
we start with the entire graph @xmath50 , which is the root of the recursive decomposition , as the single piece in @xmath25 .
as long as there is a piece @xmath211 in @xmath25 containing either @xmath61 or @xmath153 in it ( @xmath211 is an ancestor of @xmath61 or @xmath153 in the recursive decomposition ) , we replace @xmath211 with both of its two children in the recursion tree . when we get to @xmath61 or @xmath153
, we remove them from @xmath25 . at the end of this process
, @xmath25 contains @xmath79 pieces with @xmath80 boundary vertices .
all the vertices of @xmath90 and @xmath91 are boundary vertices of pieces of @xmath25 ( because otherwise such a vertex will be internal vertex of some member of @xmath25 , which is an ancestor of @xmath61 or @xmath153 ) , and the paths between them in @xmath158 are covered by @xmath25 ( since @xmath25 only misses internal vertices of @xmath61 and @xmath153 ) , as required .
denote the subgraph of the dense distance graph that include exactly the pieces of @xmath25 by @xmath212 . for @xmath159 , we compute part ( iv ) , by computing @xmath205 for every @xmath160 using the dijkstra implementation of @xcite on @xmath212 in @xmath83 time .
we use the distances of vertices of @xmath212 from @xmath162 to find @xmath181 for part ( v ) as well .
we add to @xmath212 a vertex @xmath173 inside @xmath61 and connect it to @xmath162 , and a vertex @xmath174 inside @xmath153 and connect every vertex of @xmath91 to it , as described before . since we added an edge with length @xmath175 from @xmath173 to @xmath162 , for every vertex @xmath179 in @xmath212 , @xmath213 , so we can determine for each edge of @xmath212 whether it is tight in a constant time . even though @xmath212 is not planar
, we can define a cyclic order on the edges incident to each vertex using the embedding of @xmath50 .
the order of edges incident to a specific vertex of @xmath212 is defined by the order of the shortest paths in @xmath50 that they represent . due to space constraint
we give the complete details of this process which takes total time of @xmath28 in appendix [ apx : c ] .
then , we ignore all non - tight edges in @xmath212 and find @xmath181 by finding the leftmost path from @xmath173 to @xmath174 .
this takes time proportional to the number of vertices in @xmath212 , which is @xmath80 .
we perform the computation of parts ( iv ) and ( v ) for every @xmath61 and @xmath153 , and @xmath159 . for every @xmath162 this computation takes @xmath83 time , and we repeat it @xmath209 times .
the total time required for constructing the data structure is @xmath214 ( note that the term @xmath215 is dominated by this bound ) .
we note that it is also possible to perform the computation in @xmath216 time using the algorithm of klein @xcite for every pair of pieces , however this does not improve the time bound for our range of @xmath41 .
let @xmath144 be a query pair .
we use the data structure of this section to find @xmath131 in @xmath73 time . again , let @xmath61 , @xmath153 be the pieces that contain @xmath52 , @xmath53 respectively .
if @xmath217 then we can answer the query in @xmath73 time in same way as in the previous data structure , which uses the query algorithm of @xcite .
either the shortest path from @xmath52 to @xmath53 is inside @xmath61 or the shortest path contains some vertex @xmath218 . in other words , @xmath219 .
we retrieve the distance @xmath75 from part ( ii ) of the data structure in @xmath73 time . for each @xmath218
we retrieve @xmath220 from part ( i ) in @xmath5 time .
since @xmath221 , it takes @xmath73 time to go over all vertices of @xmath64 and find @xmath218 that minimizes @xmath220 . then , @xmath131 is the minimum between @xmath75 and @xmath220 . for the case where @xmath52 and @xmath53 are in different pieces , let @xmath90 and @xmath91 be as before .
fix @xmath92 to be an arbitrary vertex of @xmath90 and let @xmath185 .
let the matrices @xmath168 , @xmath202 , @xmath203 be as before .
we compute an entry @xmath222 in @xmath5 time using parts ( i ) and ( iv ) of the data structure .
we determine whether an entry of @xmath168 is in @xmath202 or in @xmath203 in @xmath5 time using part ( v ) .
therefore , we can use the smawk algorithm @xcite as in @xcite to find the minimum value in @xmath168 in @xmath73 time .
this value is the requested distance .
we conclude that for a planar graph with @xmath3 vertices and any @xmath146 , we can construct in @xmath223 time a data structure of size @xmath148 that computes the distance between any two vertices in @xmath73 time . as in sect .
[ sec : b ] , we set @xmath152 and get : for a planar graph with @xmath3 vertices and @xmath42 $ ] , we can construct in @xmath45 time a data structure of size @xmath18 that computes the distance between any two vertices in @xmath46 time . 10 aggarwal , a. , klawe , m. m. , moran , s. , shor , p. , wilber , r. : geometric applications of a matrix - searching algorithm .
algorithmica 2 , 195 - 208 ( 1987 ) .
aggarwal , a. , klawe , m. : applications of generalized matrix searching to geometric algorithms .
discrete appl .
27 , 3 - 23 ( 1990 ) .
arikati , s. r. , chen , d. z. , chew , l. p. , das , g. , smid , m. h. , zaroliagis , c. d. : planar spanners and approximate shortest path queries among obstacles in the plane . in : proceedings of the fourth annual european symposium on algorithms .
lecture notes in computer science , vol . 1136 , pp . 514 - 528
springer - verlag ( 1996 ) .
bodlaender , h. l. : dynamic algorithms for graphs with treewidth 2 . in : proceedings of the 19th international workshop on graph - theoretic concepts in computer science .
lecture notes in computer science , vol .
790 , pp .
112 - 124 .
springer - verlag ( 1994 ) .
cabello , s. : many distances in planar graphs .
algorithmica ( to appear ) .
doi : 10.1007/s00453 - 010 - 9459 - 0 .
chaudhuri , s. , zaroliagis , c. d. : shortest paths in digraphs of small treewidth .
part i : sequential algorithms .
algorithmica 27 , 212 - 226 ( 2000 ) .
chen , d. z. : on the all - pairs euclidean short path problem . in : proceedings of the sixth annual acm - siam symposium on discrete algorithms , pp . 292 - 301 .
siam , philadelphia ( 1995 ) .
chen , d. z. , xu , j. : shortest path queries in planar graphs . in : proceedings of the thirty - second annual acm symposium on theory of computing , pp .
469 - 478 .
acm , new york ( 2000 ) .
djidjev , h. n. : efficient algorithms for shortest path queries in planar digraphs . in : graph - theoretic concepts in computer science .
lecture notes in computer science , vol . 1197 , pp
. 151 - 165 .
springer - verlag ( 1997 ) .
djidjev , h. n. , pantziou , g. e. , zaroliagis , c. d. : computing shortest paths and distances in planar graphs . in : proceedings of the 18th international colloquium on automata , languages and programming .
lecture notes in computer science , vol .
510 , pp .
327 - 338 .
springer - verlag ( 1991 ) .
djidjev , h. n. , pantziou , g. e. , zaroliagis , c. d. : on - line and dynamic algorithms for shortest path problems , in : proceedings of 12th stacs .
lecture notes in computer science , vol .
900 , pp .
193 - 204 .
springer - verlag ( 1995 ) .
dijdjev , h. n. , venkatesan , s. m. : planarization of graphs embedded on surfaces . in : proceedings of the 21st international workshop on graph - theoretic concepts in computer science .
lecture notes in computer science , vol . 1017 , pp .
springer - verlag ( 1995 ) .
eppstein , d. : subgraph isomorphism in planar graphs and related problems . j. graph algorithms appl . 3 , 1 - 27 ( 1999 ) .
fakcharoenphol , j. , rao , s. : planar graphs , negative weight edges , shortest paths , and near linear time .
j. comput .
72 , 868 - 889 ( 2006 ) .
feuerstein , e. , marchetti - spaccamela , a. : dynamic algorithms for shortest paths in planar graphs .
116 , 359 - 371 ( 1993 ) .
frederickson , g. n. : fast algorithms for shortest paths in planar graphs , with applications .
siam j. comput .
16 , 1004 - 1022 ( 1987 ) .
frederickson , g. n. : using cellular graph embeddings in solving all pairs shortest paths problems .
j. algorithms 19 , 45 - 85 ( 1995 ) .
frederickson , g. n. : searching among intervals and compact routing tables .
algorithmica 15 , 448 - 466 ( 1996 ) .
henzinger , m. r. , klein , p. , rao , s. , subramanian , s. : faster shortest - path algorithms for planar graphs .
j. comput .
55 , 3 - 23 ( 1997 ) .
hutchinson , j. p. , miller , g. l. : deleting vertices to make graphs of positive genus planar . in
: discrete algorithms and complexity theory , pp .
81 - 98 . academic press , boston ( 1986 ) .
klein , p. : preprocessing an undirected planar network to enable fast approximate distance queries . in : proceedings of the thirteenth annual acm - siam symposium on discrete algorithms , pp .
820 - 827 .
siam , philadelphia ( 2002 ) .
klein , p. n. : multiple - source shortest paths in planar graphs . in : proceedings of the sixteenth annual acm - siam symposium on discrete algorithms , pp .
145 - 155 .
siam , philadelphia ( 2005 ) .
klein , p. n. , mozes , s. , weimann , o. : shortest paths in directed planar graphs with negative lengths : a linear - space @xmath30-time algorithm .
acm trans .
algorithms 6 , 1 - 18 ( 2010 ) .
kowalik , . ,
kurowski , m. : oracles for bounded - length shortest paths in planar graphs .
acm trans .
algorithms 2 , 335 - 363 ( 2006 ) .
lipton , r. j. , tarjan , r. e. : a separator theorem for planar graphs .
siam j. on appl .
36 , 177 - 189 ( 1979 ) .
miller , g. l. : finding small simple cycle separators for 2-connected planar graphs .
j. comput .
32 , 265 - 279 ( 1986 ) .
mozes , s. , sommer , c. : exact distance oracles for planar graphs .
arxiv:1011.5549 ( 2010 ) .
mozes , s. , wulff - nilsen , c. : shortest paths in planar graphs with real lengths in @xmath54 time . in : algorithms - esa 2010 ,
18th annual european symposium .
lecture notes in computer science , vol . 6347 , pp . 206 - 217
springer ( 2010 ) .
sleator , d. d. , tarjan , r. e. : a data structure for dynamic trees . j. comput .
26 , 362 - 391 ( 1983 ) .
thorup , m. : compact oracles for reachability and approximate distances in planar digraphs . j. acm 51 , 993 - 1024 ( 2004 ) .
consider the data structure of sect .
[ sec : b ] .
let @xmath144 be a query pair , such that @xmath52 is in a piece @xmath61 and @xmath53 is in another piece @xmath153 .
the query algorithm that we describe here is similar to the one of ( * ? ? ? * ( 5 ) ) , and the complete details are given there .
let @xmath57 be the hole of @xmath61 that contains @xmath53 and let @xmath154 the hole of @xmath153 that contains @xmath52 .
denote @xmath155 $ ] and @xmath156 $ ] .
we assume without loss of generality that @xmath224 contains the infinite face .
a shortest path from @xmath52 to @xmath53 contains some vertex @xmath159 and some vertex @xmath160 ( it is possible that @xmath161 ) .
we may assume that there is no internal vertex of @xmath61 between @xmath162 and @xmath163 in the shortest path ( since otherwise we can replace @xmath162 with another vertex of @xmath90 ) .
therefore , @xmath225 .
next we show how to find @xmath226 for every @xmath159 in @xmath145 time . for a fixed vertex @xmath187 it is easy to find @xmath227 that minimizes @xmath228 in @xmath73 time ( the same member of @xmath91 may be @xmath229 for different members of @xmath90 ) , by going over all vertices of @xmath91 and using parts ( iii ) and ( i ) of the data structure
. let @xmath94 and @xmath230 be the vertices that minimizes @xmath228 for @xmath92 and @xmath231 .
there is a shortest path from @xmath92 to @xmath53 that contains @xmath94 , and similarly a shortest path from @xmath231 to @xmath53 that contains @xmath230 .
let @xmath232 be a vertex between @xmath92 and @xmath231 in the clockwise order of @xmath90 starting at @xmath92 .
there is a vertex @xmath233 that minimizes @xmath234 located between @xmath94 and @xmath230 in the counterclockwise order of vertices of @xmath91 starting at @xmath94 . since otherwise
, every shortest path from @xmath232 to @xmath53 must cross the shortest path from @xmath92 or from @xmath231 to @xmath53 that contains @xmath94 or @xmath230 , respectively .
assume without loss of generality that the shortest path from @xmath232 to @xmath53 crosses the shortest path from @xmath92 to @xmath53 , and let @xmath100 be the vertex in which the two shortest paths meet .
then , if we replace the suffix of the shortest path from @xmath92 that begins at @xmath100 with the suffix of the shortest path from @xmath232 we get a shorter path , this is a contradiction ( see fig . [
fig : x3y3 ] ) .
this gives the following algorithm for finding @xmath227 for every @xmath187 . .
for @xmath232 between @xmath92 and @xmath231 , the vertex @xmath235 is between @xmath94 and @xmath230 , since otherwise we get that every path from @xmath232 to @xmath53 ( _ dashed _ ) crosses either the shortest path from @xmath92 to @xmath53 or from @xmath231 to @xmath53 at a vertex @xmath100 . ]
let @xmath236 be two arbitrary vertices of @xmath90 .
find @xmath94 and @xmath230 for @xmath92 and @xmath231 by going over all vertices of @xmath91 .
let @xmath232 be the middle vertex between @xmath92 and @xmath231 in the clockwise order of @xmath90 starting at @xmath92 .
find @xmath235 by going over all vertices of @xmath91 between @xmath94 and @xmath230 in the counterclockwise order of vertices of @xmath91 starting at @xmath94 .
continue recursively for the vertices of @xmath90 between @xmath92 and @xmath232 and the vertices of @xmath91 between @xmath94 and @xmath235 , and also for the vertices of @xmath90 between @xmath232 and @xmath231 and the vertices of @xmath91 between @xmath235 and @xmath230 .
similarly , find @xmath229 for every @xmath96 between @xmath92 and @xmath231 in the counterclockwise order of @xmath90 starting at @xmath92 .
we conclude that we can find @xmath226 for every @xmath159 in @xmath145 time . now , we go over all vertices of @xmath90 , and using part ( i ) of the data structure we find @xmath237 in @xmath73 time .
the total query time is @xmath145 .
in this appendix we define a cyclic order on the edges incident to a specific vertex in the graph @xmath212 , which is a subgraph of the dense distance graph .
we use this order in the preprocessing algorithm of sect .
[ sec : cp ] , in order to find @xmath181 for a boundary vertex @xmath162 .
we define the order of the edges such that the leftmost shortest paths from @xmath162 to @xmath91 in @xmath50 , and in @xmath212 , both end at the same vertex of @xmath91 ( @xmath238 and @xmath212 are defined in sect . [
sec : c ] ) .
a vertex @xmath100 of @xmath212 is a boundary vertex of more than one piece , however the order between two edges in two different pieces is clear from the embedding of @xmath50 ( the pieces of @xmath212 are pairwise edge disjoint ) .
therefore , here we define the left - to - right order of the edges inside each piece . the left - to - right order of the edges , is in fact a left - to - right order of the boundary vertices , because the edges of a piece in the dense distance graph connect a vertex on the boundary of the piece to all other vertices on the boundary .
we define the left - to - right order from @xmath100 to the other boundary vertices of the piece according to the left - to - right order of the leftmost shortest paths from @xmath100 to the other vertices .
this order allows us to find @xmath181 as required .
the order that we define does not depend on the specific graph @xmath212 , so we perform the procedure described here only once for every boundary vertex of every piece .
let @xmath100 be a boundary vertex of a piece @xmath211 .
when we compute the distances from @xmath100 to the other vertices of @xmath239 for the dense distance graph , we use the algorithm of klein @xcite , which maintains a _ dynamic tree _ @xcite that contains the rightmost shortest path from @xmath100 to every vertex of @xmath211 . since we are interested in leftmost shortest paths we use a symmetric version of @xcite , by replacing the roles of left and right .
denote this leftmost shortest path tree rooted at @xmath100 by @xmath240 .
let @xmath179 and @xmath241 be two vertices of @xmath239 different from @xmath100 .
we show how to decide in @xmath242 time which of the two vertices is to the left of the other , with respect to @xmath100 .
let @xmath243 be the nearest common ancestor of @xmath179 and @xmath241 in @xmath240 .
we can find @xmath243 and the two edges that lead from it to @xmath179 and to @xmath241 in @xmath242 time from the dynamic tree @xcite .
first assume that @xmath244 .
consider the following three edges incident to @xmath243 in @xmath240 the edge that connects @xmath243 to its parent ( if @xmath243 = @xmath100 then we add a dummy edge inside the hole that @xmath100 lies on its boundary for this purpose ) , the edge that leads from @xmath243 to @xmath179 , and the edge that leads from @xmath243 to @xmath241 .
the order of these edges around @xmath243 determine the order between @xmath179 and @xmath241 ( see fig . [
fig : compare](a ) ) . now assume without loss of generality that @xmath245 .
the vertex @xmath179 lies on the boundary of some hole of @xmath211 , denote this hole by @xmath57 .
there are two edges incident to @xmath179 on the boundary of @xmath57 .
we can find the two edges when we find the piece @xmath211 .
consider the edge that connects @xmath179 to its parent in @xmath240 , the edge that leads from @xmath179 to @xmath241 , and the place of @xmath57 among the edges incident to @xmath179 .
if in the clockwise order of edges around @xmath179 starting at the edge that connects @xmath179 to it parent , the edge that leads to @xmath241 is before @xmath57 , then @xmath241 is to the left of @xmath179 , otherwise @xmath179 is to the left of @xmath241 ( see fig . [
fig : compare](b ) ) .
since we compare two vertices in @xmath242 time , we can use comparison sort to sort the @xmath246 vertices of @xmath239 around the vertex @xmath100 from left to right in @xmath247 time .
we repeat the process for each vertex of @xmath239 in a total of @xmath248 time . for the pieces of a single layer of the recursive decomposition ,
the total time is @xmath249 . and
for all the pieces of the dense distance graph the process takes @xmath28 time . | there are several known data structures that answer distance queries between two arbitrary vertices in a planar graph .
the tradeoff is among preprocessing time , storage space and query time . in this paper
we present three data structures that answer such queries , each with its own advantage over previous data structures .
the first one improves the query time of data structures of linear space .
the second improves the preprocessing time of data structures with a space bound of @xmath0 or higher while matching the best known query time .
the third data structure improves the query time for a similar range of space bounds , at the expense of a longer preprocessing time .
the techniques that we use include modifying the parameters of planar graph decompositions , combining the different advantages of existing data structures , and using the monge property for finding minimum elements of matrices . |
it is notable that while the exact solutions in general relativity fill already ample handbooks , the exactly solvable problems of quantum theory form still an exceptional area . yet
, it seems that they are just the `` top of an iceberg '' , an opinion recently confirmed by advances in the exact methods @xcite .
one of the simplest ways of obtaining new exactly solvable spectral problems in schrdinger s quantum mechanics is to look for pairs of quantum hamiltonians coupled `` supersymmetrically '' by an _ intertwining operator _
@xmath5 h_1 a_1 = a_1 h_0 .
[ i.1 ] this equation should hold on a certain dense domain of the hilbert space of states @xmath6 , assuring that the spectrum of @xmath7 yields some information about the spectrum of @xmath8 and _ vice versa_. the best known cases of ( [ i.1 ] ) occur in @xmath9 , where @xmath7 and @xmath8 are the schrdinger hamiltonians h_0 = - + v_0(x ) , h_1 = - + v_1(x ) , [ i.2 ] and @xmath5 , @xmath10 are the first order differential operators a_1= + _
1(x ) , a_1^ = - + _
1(x ) . [ i.3 ] by introducing ( [ i.2 ] ) and ( [ i.3 ] ) into ( [ i.1 ] ) , and after an elementary integration
, one gets the riccati equation for @xmath11 -_1(x ) + _ 1 ^ 2(x ) = 2[v_0(x ) - ] , [ i.4 ] where @xmath12 is an arbitrary integration constant called the _ factorization energy _ ( a terminology which does not imply that @xmath12 must be a physical energy ) .
simultaneously , v_1(x ) = v_0(x ) + _1(x ) .
[ i.5 ] equations ( [ i.4 ] ) and ( [ i.5 ] ) are necessary and sufficient conditions for the hamiltonians to be factorized as h_0 - = 12 a_1^ a_1 , h_1 - = 12 a_1 a_1^. [ i.6 ] the susy partners ( [ i.5])([i.6 ] ) appear in all quantum problems solvable by factorization @xcite .
an element of arbitrariness in this construction was explored in 1984 @xcite ; its presence in witten s susy quantum mechanics was noticed by nieto @xcite ; links with the historical darboux theory @xcite were established by andrianov _
et al _ @xcite .
the higher order intertwining was considered by crum @xcite ( see also krein @xcite , adler and moser @xcite ) . in physical applications though , for a long time , the attention was focused mostly on the first order intertwining operator @xmath5 ( see @xcite and references quoted therein ) .
the idea of the higher order @xmath5 reemerged in 80tieth @xcite and it was systematically pursued since 1993 ( see the works of veselov and shabat @xcite , andrianov _ et al _ @xcite , eleonsky _ et al _ @xcite , bagrov and samsonov @xcite , fernndez _ et al _ @xcite , and one of us @xcite ) . as it turns out , the higher order susy partners of an arbitrary @xmath7 can be generated just by iterating the darboux transformation with different factorization constants @xmath12 ( see interesting theorems by bagrov and samsonov in @xcite ) .
formally , this can be done by introducing a higher order intertwiner constructed by means of the _ determinant formula _ of crum @xcite , a method fundamentally important , though not very easy to apply .
the purpose of this note is to explore a faster method based on the bcklund transformation @xcite which operates very simply at the level of the @xmath13-functions ( _ superpotentials _ ) in equations ( [ i.3])([i.5 ] ) .
it consist just in applying the finite difference calculus , without caring about wave functions , differential equations , etc . moreover ,
if applied to a sufficiently rich initial family of solutions to ( [ i.4 ] ) , it permits to recover the complete parametric dependence in all subsequent susy steps without the need of integrating any differential equations @xcite .
we also show that the differential case of the algorithm gives an entire susy hierarchy for a fixed @xmath12 just in terms of quadratures and algebraic operations .
the solutions of the riccati equation ( [ i.4 ] ) , and the potentials ( [ i.5 ] ) , depend implicitly on the factorization energy @xmath14 .
we shall make explicit this dependence by writing @xmath15 for the solutions to ( [ i.4 ] ) , and @xmath16 for the corresponding susy partners of @xmath17 .
hence , the intertwiner in ( [ i.1 ] ) reads @xmath18 , while the corresponding hamiltonians are @xmath19 , and @xmath20 .
notice that @xmath7 does not depend on @xmath12 , yet it admits many different factorizations for many factorization energies @xmath12 , each one defining a different intertwined hamiltonian @xmath21 , with a different potential @xmath22 .
let us fix now @xmath23 , and consider one of these potentials : @xmath24 since @xmath25 is a susy partner of @xmath17 , the eigenfunctions @xmath26 of the hamiltonian @xmath27 can be generated from the eigenfunctions @xmath28 of @xmath7 by using the operator @xmath29 .
it means simply that if @xmath30 is a sufficiently smooth function and if @xmath31 , then : h_0 _ 0(x , ) = _ 0(x , ) h_1(_1 ) _ 1(x,)= _ 1(x , ) .
[ fd2 ] notice that ( [ fd2 ] ) does not require the square integrability of either @xmath28 or @xmath26 .
nevertheless , if @xmath12 corresponds to one of the discrete spectrum eigenvalues @xmath32 of @xmath7 , then @xmath28 can be chosen square integrable .
a question arises now , can @xmath25 be a convenient point of departure for the next susy step ?
the answer is positive and admits an explicit construction . in geometric form
it was described by adler as a mapping of complex polygons corresponding to the bcklund transformations @xcite ( see also veselov and shabat @xcite ) .
its presence in the supersymmetric algorithm can be shown directly , using only the riccati equation ( [ i.4 ] ) .
indeed , if @xmath15 is a solution of ( [ i.4 ] ) , the function @xmath33}\ ] ] satisfies @xmath34 .
therefore , the function defined as u_2(x,_1 , ) = a_1(_1 ) u_1(x , ) = [ _ 1(x,_1 ) - _ 1(x,)][udos ] satisfies @xmath35 , and the function @xmath36 defined by u_2(x , _ 1 , ) = [ fd3 ] must fulfill the new riccati equation : -_2(x , _ 1 , ) + _ 2 ^ 2(x , _ 1 , ) = 2[v_1(x,_1 ) - ] .
[ rdos ] reading back ( [ fd3 ] ) and using ( [ udos ] ) one has @xmath37 and using again ( [ i.4 ] ) one ends up with an algebraic expression _ 2(x , _ 1 , ) = -_1(x,_1 ) - , [ fd4 ] which is the basic element of the auto - bcklund transformation for the korteweg - devries ( kdv ) equation ( as reported in @xcite , see also @xcite and @xcite ) .
thus , the @xmath13-functions of the second susy step are algebraically determined by a _ finite difference _ operation performed on the @xmath13-functions of the previous step .
they give the next operator @xmath38 intertwining the hamiltonian @xmath27 with a new one @xmath39 , h_2(_1 , ) a_2(_1 , ) = a_2(_1 , ) h_1(_1 ) , [ idos ] where v_2(x,_1 , ) = v_1(x,_1 ) + _ 2(x , _ 1 , ) .
[ 2susy ] quite obviously ( [ i.1 ] ) and ( [ idos ] ) imply @xmath40 where @xmath41 . in @xcite
the method has been applied to deform the cyclic supersymmetric chains , but we shall show that it can replace completely all other techniques in constructing quite arbitrary susy sequences . indeed , the method can be now repeated by induction , leading to the sequence of first order intertwining operators a_k ( ) + _ k(x , ) , k=1,2, ... ,n , [ fd6 ] where each @xmath42 is simply the result of a finite difference operation performed on @xmath43 : _ k ( x , ) = - _ k-1 ( x,_k-1 ) - . [ fd7 ] we have adopted here a shortcut notation making explicit only the dependence of @xmath44 and @xmath45 on the factorization constant introduced in the very last step , keeping implicit the dependence on the previous factorization constants ( henceforth , the same criterion will be used for any other symbol depending on @xmath46 factorization energies ) : @xmath47 the @xmath45 functions constructed at each next susy step automaticaly solve the riccati equation with the potential @xmath48 of the previous step -_k(x , ) + _ k^2(x , ) = 2[v_k-1(x,_k-1 ) - ] , [ fd8 ] or equivalently -_k(x , ) + _ k^2(x , ) = _ k-1(x,_k-1 ) + _ k-1
^ 2(x,_k-1 ) + 2(_k-1 - ) [ a.0 ] permitting to define the new potential v_k(x , ) = v_k-1(x,_k-1 ) + _ k ( x , ) .
[ fd9 ] in the resulting sequence of the new hamiltonians h_k ( ) = -12 + v_k(x , ) = a_k ( ) a^_k ( ) + , k=1,2, ... ,n [ hk ] each next is intertwined with the previous one , h_k ( ) a_k ( ) = a_k ( ) h_k-1 ( _ k-1),h_0 ( _ 0 ) h_0 .
[ fd12 ] all of them are the higher order susy partners of the initial @xmath7 h_k ( ) a_k = a_k h_0 , [ 0k ] ( compare with @xcite ) where @xmath49 is the higher order intertwining operator obtained in the factorized form : _ k a_k ( )
a_k-1 ( _ k-1 ) a_2 ( _ 2 ) a_1 ( _ 1 ) .
[ tocho1 ] ( here we follow our notation , making explicit the dependence of each @xmath44 on its last factorization energy . ) once having the sequence of superpotentials @xmath50 , constructed by the algorithm ( [ fd7 ] ) , it is easy to determine the new eigenfunctions @xmath51 injected by the process into the hamiltonians @xmath52 .
indeed , the @xmath53 are defined by the first order differential equations a_k ( ) u_k(x , ) = u_k(x , ) + _ k(x , ) u_k(x , ) = 0 , k=1,2, ... ,n .
[ fd15 ] up to now we have maintained implicit the element of arbitrariness in the @xmath13-functions at every recurrence step .
however , the true advantages of the method can be seen if @xmath54 is wider than just a particular solution of ( [ i.4 ] ) . indeed ,
for each fixed @xmath12 , we have to our disposal the general solution of the riccati equation ( [ fd8 ] ) , _ i.e. _ , an entire one - parametric family of functions .
e.g. _ , the general solution of ( [ i.4 ] ) , for a given @xmath55 , depends on an additional integration parameter @xmath56 .
given any particular solution @xmath57 , the general one is obtained by an elementary transformation : _
1(x,_1 ) = _ 1p ( x , _ 1)- ( x , _ 1 , _ 1 ) .
[ fd16 ] now , the algorithm ( [ fd4 ] ) can be interpreted as an operation on two _ classes of functions _ @xmath15 and @xmath58 : an arbitrary member of the class @xmath15 subtracted from an arbitrary member of @xmath58 in the denominator of ( [ fd4 ] ) .
quite similarly , the general iteration step ( [ fd7 ] ) can be applied for @xmath59 and @xmath60 meaning any two superpotentials of the previous step , the only condition stating that they must solve the corresponding riccati equation with the same potential @xmath61 ( a fact which is automaticaly assured in the iterative process ) .
if so , the method accumulates the integration constants , permitting to recover the full parametric dependence on the @xmath62-th susy potentials in a purely algebraic way .
e.g. _ , @xmath63 with the dependence on @xmath64 inherited from the previous step ( @xmath56 and @xmath65 selecting two solutions among the elements of the families @xmath58 and @xmath15 respectively ) .
in general , our algorithm yields @xmath66 where the parameters @xmath67 will be varied to perform the next step , whereas @xmath68 will stay fixed . in
what follows , we shall employ the symbols @xmath69 whenever wanting to abbreviate , but we shall hold the dependence on the accumulated integration constants @xmath70 if necessary .
let us now introduce the functions : _ k ( x , ) = observe their _ fractal _ structure accumulating the _ finite difference _ derivatives ( compare with @xcite ) _
k(x,_k ) = - , k2 .
[ fd17 ] now , ( [ fd7 ] ) can be rewritten as @xmath71 . for example , for @xmath72
we get ( [ fd4 ] ) , while for @xmath73 we have @xmath74 \!\!&\!\ ! \!\ !
& \!\!\ ! + \frac{\epsilon_{2}-\epsilon_{3 } } { \displaystyle \frac{\epsilon_{1}-\epsilon_{2 } } { \beta(x;\epsilon_1;\lambda_{1})-\beta(x;\epsilon_2 ; \lambda_{2 } ) } - \frac{\epsilon_{1}-\epsilon_{3 } } { \displaystyle \beta(x;\epsilon_1;\lambda_{1})- \beta(x;\epsilon_3 ; \lambda_{3})}}. \label{fractal}\end{aligned}\ ] ] as one can also notice , the `` @xmath62-susy potentials '' @xmath75 , present the different fractal structures for @xmath62 even and odd . summarizing : [ fd18 ] v_n(x,_n ) = v_0 ( x ) + _ k=1^n _ k(x , _ k ) = v_0 ( x ) + _ k=1^n _ k ( x,_k ) .
this expression has been already applied to simplify the study of @xmath62-susy partners of the harmonic oscillator @xcite and the coulomb potential @xcite .
notice now that when @xmath76 , and taking the limit @xmath77 in ( [ fd7 ] ) , the method produces a differential algorithm ( a _ confluent susy _ operation ) ^conf_k(x , ) = - _ k-1(x , ) - 2 ( ) ^-1 , [ 2.15 ] which can be freely iterated .
likewise , the finite differences can be replaced by the @xmath12-derivatives in one or more places of the fractal formulae ( [ fractal ] ) .
the mechanism has been implicitly applied by stalhoffen ( see section iii of @xcite ) , in the context of crum s determinant .
the difference - differential algorithms ( [ fd7 ] ) , ( [ 2.15 ] ) simplify remarkably the construction of transparent wells @xcite .
in fact , the riccati equation ( [ i.4 ] ) for @xmath78 has the general solution _
1(x;;)= - , [ free1 ] with four different real 1-susy branches : s ( _ singular _ ) , r ( _ regular _ ) , p ( _ periodic _ ) , and n ( _ null _ ) , for different ( real or complex ) values of @xmath12 and @xmath79 ( see table i ) . for example
, the regular case ( r ) leads to the modified pschl - teller potential @xmath80 $ ] , while the null case ( n ) corresponds to the potential barrier @xmath81 .
all of them are particular cases of the weierstrass @xmath82-function : [ pfunction ] v_1(x;;)= ( x+,3(4/3)^2,(4/3)^3)+ .
0.3 cm table i. the four different real superpotentials @xmath83 coming out from ( [ free1 ] ) , depending on the values of @xmath12 and the integration parameter @xmath79 . in each case
s means singular , r regular , @xmath84 periodic , and n null .
the parameters @xmath85 and @xmath86 are arbitrary real numbers . ccccc + [ -2ex ] & & & & + [ -1ex ] case & @xmath12 & @xmath87 & @xmath88 & @xmath15 + [ 1ex ] & & & & + [ -1ex ] s & @xmath89 & @xmath90 & @xmath85 & @xmath91 $ ] + [ 2ex ] r & @xmath89 & @xmath92 & @xmath93 & @xmath94 $ ] + [ 2ex ] @xmath84 & @xmath95 & @xmath96 & @xmath85 & @xmath97 $ ] + [ 2ex ] n & @xmath98 & @xmath98 & @xmath85 & @xmath99 + [ 2ex ] + [ -2ex ] 0.5 cm by applying the algorithm ( [ fd4 ] ) to the superpotentials r and s as given in table i , one immediately gets the transparent double wells in form of the _ quadratic bargmann potentials _ @xcite ( see figure 1 ) [ free2 ] v_2 ( x,_2 ) = - ( _ 1 ^ 2 - _ 2 ^ 2 ) . the double wells ( [ free2 ] ) are well known in supersymmetry , but the use of the bcklund transformation makes the derivation considerably simpler , explaining also their place in the theory of the kdv equation , @xmath100 . as a matter of fact
, all the transparent @xmath62-susy wells can be now reinterpreted as the _ instantaneous forms _ of the multisoliton solutions propagating according to kdv ( thus , _
e.g. _ , the 2-susy potentials ( [ free2 ] ) originate the propagating soliton pairs @xcite ) .
presumably , the susy partners of an arbitrary @xmath101 generate as well the traveling solitonic deformations " on a more general , non - null background .
the operational advantages of the differential algorithm ( [ 2.15 ] ) deserve as much attention .
the application of the finite difference algorithm to the null wells is so easy , because for @xmath102 one knows the solutions of the riccati eq .
( [ i.4 ] ) for any @xmath12 . for non - null @xmath103 , ( [ fd4 ] ) can be applied efficiently only if one knows at least two solutions of ( [ i.4 ] ) for two different factorization energies " @xmath12 ( compare @xcite , @xcite ) . for a general @xmath103 ,
however , this is not granted .
it is therefore interesting , that the difficulty is circumvented by the differential algorithm ( [ 2.15 ] ) which gives an easy access to higher order susy steps even if one starts from just one superpotential @xmath69 for a fixed @xmath12 . indeed , by taking the limit @xmath104 , one reduces ( [ a.0 ] ) to the following equation for @xmath105 :
k-1 ^ 2 , k=2,3 , ... , [ a.1 ] which has the particular solution @xmath106 and integrates easily yielding : @xmath107 .
\label{a.2}\end{aligned}\ ] ] note that the right hand side of ( [ a.2 ] ) depends essentially just on one integration constant .
the recurrence ( [ a.1 ] ) acquires even simpler form in terms of the new ` key functions ' @xmath108 ( the integration constants @xmath109 are superfluous to determine @xmath110 ) . in fact , integrating both sides of ( [ a.2 ] ) , taking to the exponent , differentiating again , and choosing properly the ( inessential ) constants @xmath111 , one has : ( b_nb_n-1)= - b_n-1 ^ 2 .
[ a.3 ] when this method is applied to @xmath112 with @xmath113 , the first step recovers the abraham - moses oscillator @xcite , while the second step leads to the new 2-parametric family v_2(x)=2- , [ oscilconf ] which are different from the potentials discussed in @xcite , because they appear for a fixed @xmath12 and do not add 2 new spectral levels to @xmath103 ; we therefore propose to call them the _
2-nd order abraham - moses potentials_. the detailed analysis shows that ( [ oscilconf ] ) are singular unless @xmath114 and @xmath115 , or @xmath116 and @xmath117 . a nonsingular case with @xmath114 and @xmath118 is shown in figure 2 .
the higher order abraham - moses functions can be as easily obtained in terms of ( [ a.2 ] ) . generically , our algorithm is close to the nice idea of leble @xcite ( though leble did not construct the non - trivial contributions for @xmath119 ) .
an equally interesting application of the confluent formula ( [ 2.15 ] ) arises for the periodic superpotential @xmath84 in table i. in this case , the first order susy partner is given by [ 1conf ] v_1^conf(x , ) = v_1(x , ) = , while the corresponding second order susy step leads to [ free3 ] v_2^conf(x , ) = 8 k^2 .
notice that ( [ 1conf ] ) is a periodic potential with singularities at the points @xmath120 , @xmath121 , while @xmath122 has only one singularity .
the _ confluent susy step _ ( [ 2.15 ] ) has removed all the singularities from @xmath123 except that placed on @xmath124 .
the potential ( [ free3 ] ) has been plotted on figure 3 ( dashed curve ) for @xmath125 .
observe that we have rederived the stahlhofen function , but the method is simpler ( see equations ( 50 ) and ( 40 ) of @xcite ) .
moreover , some other potentials with analogous properties can be immediately obtained .
our figure 3 reports the fourth order confluent potential @xmath126 , with @xmath127 .
notice the appearance of two new symmetric singularities .
as becomes obvious , for the periodic superpotentials @xmath83 in table i , the method leads to two distinct classes of @xmath62-th order confluent potentials . for @xmath62
even , the periodic superpotential @xmath83 does not contribute directly to the sum ( [ fd18 ] ) ( the term with @xmath128 vanishes ) .
the resulting susy partners have only a finite number of singularities . on the other hand , for @xmath62 odd
, the function @xmath83 contributes to the first term in the sum ( [ fd18 ] ) and its global effect is never canceled .
the corresponding susy partners @xmath129 therefore have infinite sequences of singularities .
this difference reflects the distinct fractal structure of ( [ fd18 ] ) for @xmath62 even and odd .
the possibility of mixed applications of ( [ fd7 ] ) and ( [ 2.15 ] ) is open .
this work has been supported by the spanish dges ( pb94 - 1115 and pb98 - 0370 ) and by junta de castilla y len ( co2/199 ) .
bm and oro acknowledge the kind hospitality at departamento de fsica terica , univ
. de valladolid .
oro acknowledges support from conacyt ( mexico ) .
lmn thanks prof .
glasser for useful comments . | the higher order supersymmetric partners of the schrdinger s hamiltonians can be explicitly constructed by iterating a simple finite difference equation corresponding to the bcklund transformation .
the method can completely replace the crum determinants .
its limiting , differential case offers some new operational advantages .
= 1ex = 0.55 cm = 0.55 cm = 1.5em = 23.0 cm = -1.0 cm = eufm10 scaled1 # 1 = msbm10 scaled 1 # 1 b. mielnik@xmath0 , l.m .
nieto@xmath1 , and o. rosas
ortiz@xmath2 0.5 cm _ @xmath3departamento de fsica , cinvestav - ipn , ap 14 - 740 + 07000 mxico df , mexico + [ 2ex ] @xmath4institute of theoretical physics , warsaw university , hoa 69 warsaw , poland + [ 2ex ] @xmath1departamento de fsica terica , universidad de valladolid + 47011 valladolid , spain _ 1.5 cm pacs : : : 03.65.ge , 03.65.fd , 03.65.ca key - words : : : supersymmetry , factorization , difference equations * accepted for publication in phys .
lett .
a ( 2000 ) * |
_ oblivious transfer _ @xcite , _ ot _ for short , is a functionality of great importance @xcite in cryptography or , more precisely , _ secure two - party computation _
, where two parties , who mutually distrust each other , want to collaborate with the objective of achieving a common goal , e.g. , evaluate a function to which both hold an input|but without revealing unnecessary information about the latter . in ( chosen one - out - of - two bit ) ot , one of the parties , the _ sender _ , inputs two bits @xmath0 and @xmath1 , whereas the other party has a _ choice _ bit @xmath2 .
the latter then learns @xmath3 , but remains ignorant about the other message bit @xmath4 .
the sender , on the other hand , does not learn any information about @xmath2 . various ways , based on public - key encryption , for instance , have been proposed for realizing ot , where the security for one of the parties , however , is only computational .
in fact , oblivious transfer is impossible to achieve in an unconditionally secure way for both parties|even when they are connected by a quantum channel @xcite . on the other hand
, it has been shown that unconditionally secure ot can be reduced to weak information - theoretic primitives such as simply a noisy communication channel @xcite , @xcite or so - called _ universal ot _
@xcite .
a recent result @xcite shows that ot can be stored : given one realization of ot , a sample of distributed random variables @xmath5 ( known to @xmath6 ) and @xmath7 ( known to @xmath8 ) can be generated , where the joint distribution @xmath9 is such that @xmath5 and @xmath7 can be used to realize an instance of ot .
we will call the distributed pair of random variables @xmath10 an _ oblivious key _ or ok for short ; in some sense , as we will see , this is the _ local _ ( hidden - variable ) part of ot ( as opposed to non - local systems and behavior , see section 1.2 ) .
another consequence|observed in @xcite|is that , since ok is symmetric , ot is , too .
this solved a long - standing open problem posed in @xcite .
_ entangled _ but possibly distant two - partite quantum systems can show a joint behavior under measurements that can not be explained by `` locality '' or hidden variables , i.e. , distributed classical information ; such behavior is called _ non - local_. there exists , for instance , a so - called _ maximally entangled _
state @xmath11 with the following properties .
if the parties @xmath6 and @xmath8 controlling the two parts of the system both choose between two fixed possible bases for measuring their system in ( where this pair of bases is not the same for the two parties ) , where the measurement outcome can be @xmath12 or @xmath13 in both cases , then the following statistics are observed .
( here , the two possible bases for each party are called @xmath12 and @xmath13 , too . ) @xmath14\ \ = \ \ 0\ , \\ p_{01 } & : = & { \rm prob\ , } [ \ , \mbox{outcome $ a$}=\mbox{outcome $ b$}\ , |\ , \mbox{basis $ a$}=0\ , , \
\mbox{basis $ b$}=1\ , ] \ \ = \ \ 1/4\ , \\ p_{10 } & : = & { \rm prob\ , } [ \ , \mbox{outcome $ a$}=\mbox{outcome $ b$}\ , |\ , \mbox{basis $ a$}=1\ , , \
\mbox{basis $ b$}=0\ , ] \ \ = \ \ 1/4\ , \\ p_{11 } & : = & { \rm prob\ , } [ \ , \mbox{outcome $ a$}=\mbox{outcome $ b$}\ , |\ , \mbox{basis $ a$}=1\ , , \
\mbox{basis $ b$}=1\ , ] \ \ = \ \ 3/4\ .\end{aligned}\ ] ] it has been shown that such statistics are impossible to achieve between two parties who can not communicate when they share arbitrary _ classical _ information only ( i.e. , agree on a classical strategy beforehand ) .
more precisely , the so - called chsh _ bell inequality _ is violated since @xmath15 holds .
it is , on the other hand , important to note that this non - local behavior is `` weaker '' that communication between @xmath6 and @xmath8 and does not allow for such|fortunately , since such a possibility would be in contradiction with relativity . with the objective of achieving a better understanding of such `` non - local behavior , '' popescu and
rohrlich @xcite defined a `` non - locality primitive '' behaving in a similar way , but where the probabilities @xmath16 are @xmath17 in other words , both parties have an input bit ( corresponding to the choice of the basis in the quantum model ) @xmath18 and @xmath19 and get an output bit @xmath5 and @xmath7 , respectively , where @xmath5 and @xmath7 are random bits satisfying @xmath20 it is important to note , however , that the behavior of this `` pr primitive '' can not , although it does not allow for communication either , be obtained from any quantum state|it violates a `` quantum bell inequality '' that is even valid for the behavior of quantum states . on the other hand , the primitive _
does _ allow for perfectly simulating the behavior of a maximally entangled quantum bit pair under _ any _ possible measurement @xcite .
the latter has been shown possible , for instance , also between parties who may communicate _
one _ classical bit @xcite @xcite , but the possibility of achieving the same with the pr primitive is of particular interest since this functionality does not allow for any communication .
the three _ information - theoretic primitives _ or two - party functionalities described in the previous sections can be modeled by their mutual input - output behavior , i.e. , by a conditional probability distribution @xmath21 , where @xmath18 , @xmath19 , @xmath5 , and @xmath7 are the two parties input and output , respectively ( see figure 1 ) .
( 0,0 ) # 1#2#3#4#5 @font ( 22815,4260)(376,-3994 ) ( 16651,-2386)(0,0)[lb ] ( 16651,-1186)(0,0)[lb ] ( 16501 , 14)(0,0)[lb ] ( 22351,-1186)(0,0)[lb ] ( 22351,-2386)(0,0)[lb ] ( 17776,-3886)(0,0)[lb ] ( 19276,-1786)(0,0)[lb ] ( 22576 , 14)(0,0)[lb ] ( 8776 , 14)(0,0)[lb ] ( 6226,-1186)(0,0)[lb ] ( 6226,-2386)(0,0)[lb ] ( 526,-1711)(0,0)[lb ] ( 526,-1186)(0,0)[lb ] ( 376 , 14)(0,0)[lb ] ( 10726,-3661)(0,0)[lb ] ( 3151,-1786)(0,0)[lb ] ( 11626,-1786)(0,0)[lb ] ( 8026,-2386)(0,0)[lb ] ( 14551,-2386)(0,0)[lb ] ( 6451 , 14)(0,0)[lb ] ( 14776 , 14)(0,0)[lb ] in section 2 , we will show simple perfect and single - copy information - theoretic reductions between the three primitives|in some sense , they are , provocatively speaking , all the same . more precisely , a single - copy reduction of a primitive @xmath22 to another @xmath23 means that the functionality @xmath22 can be realized given one instance of @xmath23 .
hereby , no computational assumptions have to be made . _ perfect
_ means that no non - zero failure probability has to be tolerated .
note , however , that the reduction protocol may use communication ; of course , because from a `` communication and locality viewpoint , '' the three primitives are very different : ot allows for communication , pr does not , but is non - local , whereas ok is simply distributed classical information , i.e. , `` local . ''
although we keep an eye on this communication in the reductions|all our reductions minimize the required amount of communication| , our interest is _ privacy _ : when @xmath22 is obtained from @xmath23 , say , then both parties must not obtain more information than specified for @xmath22 .
in other words , our viewpoint is the one of _ cryptography _ rather than of communication - complexity theory .
note that our reductions have the property that a party who is misbehaving in the protocol _ can not _ obtain more information than specified ( but possibly violate the privacy of her proper inputs ) .
[ lemma : ot2nl ] using one instance of @xmath24 , we can simulate @xmath25 .
@xmath8 chooses @xmath26 .
@xmath6 chooses @xmath27 at random and sends @xmath28 and @xmath29 with @xmath24 .
@xmath8 receives @xmath3 and outputs @xmath30 .
@xmath6 outputs @xmath27 .
we have @xmath31 .
[ lemma : nl2ok ] using one instance of @xmath25 , we can simulate @xmath32 .
@xmath6 and @xmath8 choose @xmath33 and @xmath34 at random .
@xmath8 outputs @xmath35 and @xmath36 .
@xmath6 outputs @xmath37 and @xmath38 .
we have @xmath39 .
[ lemma : ot2ok ] using one instance of @xmath24 , we can simulate @xmath32
. follows directly from lemmas [ lemma : ot2nl ] and [ lemma : nl2ok ] .
we get the following protocol : @xmath6 and @xmath8 choose all their input at random .
@xmath6 outputs her inputs , @xmath8 his input and his output .
[ lemma : nl2ot ] using one instance of @xmath25 , we can simulate @xmath24 using one bit of communication .
@xmath6 inputs @xmath40 .
@xmath8 inputs @xmath41 .
@xmath6 gets @xmath27 and @xmath8 gets @xmath42 .
@xmath6 sends @xmath43 to @xmath8 .
@xmath8 outputs @xmath44 .
we have @xmath45 .
since @xmath6 does not receive any message from @xmath8 , she gets no information about @xmath2 . @xmath8 only receives one bit , which is equal to @xmath3 . in @xmath25 ,
no communication takes place , but we are able to send one bit using @xmath24 .
hence , at least one bit of communication is needed to simulate @xmath24 by @xmath25 .
[ lemma : ok2nl ] using one instance of @xmath32 , we can simulate @xmath25 using two bits of communication .
@xmath6 sends @xmath46 to @xmath8 .
@xmath8 sends @xmath47 to @xmath6 .
@xmath6 outputs @xmath48 .
@xmath8 outputs @xmath49 .
we have @xmath50 . hence @xmath51 they both send their inputs `` xored '' with @xmath52 and @xmath53 , respectively .
since the other party has no information about these values , this is a one - time pad , and they receive no information about the other s input .
we show that the two bits of communication are optimal in this case .
let us assume that there exists a protocol using only one - way communication from @xmath6 to @xmath8 . since @xmath8 can calculate his output @xmath54 for both inputs for @xmath55 and @xmath56 for @xmath57 , we have @xmath58 , and , therefore , @xmath59 . using one instance of @xmath32 , we can simulate @xmath24 using three bits of communication
. follows directly from lemmas 4 and 5 .
alternatively , we can use the bbcs protocol @xcite , which requires three bits of communication as well .
here , @xmath8 sends @xmath60 to @xmath6 , whereas @xmath6 sends @xmath61 and @xmath62 to @xmath8 .
@xmath8 outputs @xmath63 .
we have @xmath64 .
@xmath8 s message does not give any information about @xmath2 to @xmath6 , since it is `` one - time padded '' with the value @xmath53 about which @xmath6 has no information .
@xmath8 knows either @xmath65 or @xmath66 but has no information about the other value .
so , either @xmath0 or @xmath1 gets `` one - time padded , '' and @xmath8 obtains information about that value , even if he is given the other value .
three bits of communication are optimal : first of all , two - way communication is needed .
if @xmath6 would send less than two bits , but still in such a way that @xmath8 would get the bit he wants , then @xmath6 would have to know which bit @xmath8 has chosen .
ot is _ a priori _ an asymmetric functionality , and the possibility of inverting its orientation has been investigated , for instance , in @xcite , where a protocol was given using @xmath67 realizations of ot from @xmath8 to @xmath6|called to|in order to obtain one realization from @xmath6 to @xmath8 , where a failure probability exponentially small in @xmath67 has to be tolerated . since , however , pr is a symmetric functionality , our reductions imply that ot is as well .
more precisely , the reductions of ot to pr and _ vice versa _ can be put together to the following protocol inverting ot .
this reduction of ot to to given in @xcite , is single - copy , information - theoretic , perfect , and minimizes the required additional communication .
using one instance of @xmath68 , we can simulate @xmath24 using one bit of communication .
@xmath6 inputs @xmath69 to @xmath68 .
@xmath8 chooses a random bit @xmath70 and inputs @xmath70 and @xmath71 to @xmath68 .
@xmath6 receives @xmath27 and sends @xmath72 to @xmath8 .
@xmath8 outputs @xmath73 .
we have @xmath74 .
@xmath6 does not get any message from @xmath8 , so she does not get any information about @xmath2 . @xmath8 get one message by @xmath6 , which is either equal to @xmath54 , if the xor of his input values is @xmath12 , and @xmath56 otherwise .
if he does not choose @xmath70 at random , @xmath6 might be able to get the value @xmath2 , but there is no advantage for @xmath8 .
the protocol is obviously optimal since @xmath6 can communicate one bit with @xmath8 using @xmath24|which she can not using @xmath68 .
finally , we show that an @xmath32 can easily be reversed , without any communication .
using one instance of @xmath75 , we can simulate @xmath32 without any communication .
@xmath6 gets @xmath65 and @xmath66 from @xmath75 , and @xmath8 gets @xmath53 and @xmath7 .
@xmath6 outputs @xmath76 and @xmath77 , and @xmath8 outputs @xmath78 and @xmath79 .
we have @xmath80 and @xmath81 .
the @xmath32 primitive can also be defined in a symmetric way : it is the distribution that we get when both @xmath6 and @xmath8 input a random bit to @xmath25 .
we have shown a close connection between the important cryptographic functionality of oblivious transfer and quantum non - locality , more precisely , the `` non - locality machine '' of popescu and rohrlich : they are , modulo a small amount of ( classical ) communication , the same|one can be reduced to the other . as a by - product
, we have obtained the insight that _ ot is symmetric _
: one instance of ot from @xmath8 to @xmath6 allows for the same functionality from @xmath6 to @xmath8 in a perfect information - theoretic sense .
figure 2 shows the reductions between the different functionalities discussed above .
the ( optimal ) numbers of bits to be communicated are indicated .
in has been shown in @xcite that the behavior of an epr pair can be perfectly simulated without any communication if one realization of the pr primitive is available .
however , this reduction , although it yields the correct statistics with respect to the two parts behavior , is not `` cryptographic '' or `` private '' in the sense of our reductions : the parties are tolerated to obtain more information about the other party s outcome than they would when actually measuring an epr pair .
we state as an open problem to simulate , in this stronger sense , the behavior of an epr pair using the pr primitive . c. h. bennett , g. brassard , c. crpeau , and h. skubiszewska , practical quantum oblivious transfer , _ advances in cryptology | proc . of eurocrypt
91 _ , lncs , vol . 576 , pp .
351366 , springer - verlag , 1992 . | _ oblivious transfer _ , a central functionality in modern cryptography , allows a party to send two one - bit messages to another who can choose one of them to read , remaining ignorant about the other , whereas the sender does not learn the receiver s choice .
oblivious transfer the security of which is information - theoretic for both parties is known impossible to achieve from scratch . | the joint behavior of certain bi - partite quantum states is _ non - local _ ,
i.e. , can not be explained by shared classical information . in order to better understand such behavior , which is classically _
explainable _ only by communication , but does not _ allow _ for it , popescu and rohrlich have described a `` non - locality machine '' : two parties both input a bit , and both get a random output bit the xor of which is the and of the input bits . | we show a close connection , in a cryptographic sense , between ot and the `` pr primitive . ''
more specifically , unconditional ot can be achieved from a single realization of pr , and _ vice versa_. our reductions , which are single - copy , information - theoretic , and perfect , also lead to a simple and optimal protocol allowing for inverting the direction of ot . |
we begin this paper with the geometry of distributions .
the main idea here is the various notions of symmetry and their use in solving a given differential equation . in section 2
, we introduce the basic notions and definitions . in section 3
, we describe the relation between differential equations and distributions . in section 4 ,
we present the geometry of distributions by their symmetries , and find out the symmetries of @xmath0gordon equation by this machinery . in section 5 , we introduce a simplified proof of the frobenius theorem and some related corollaries . in section 6 , we describe the relations between symmetries and solutions of a distribution . in all steps
, we study the @xmath0gordon equation as an application ; a partial differential equation which appears in differential geometry and relativistic field theory .
it is a generalized form the klein - gordon equation `` @xmath1 '' .
it is a relativistic version of schrodinger equation , which is used to describe spinless particles .
it was named after walter gordon and oskar klein @xcite .
let @xmath2 be an @xmath3dimensional smooth manifold .
a map @xmath4 is called an _ @xmath5dimensional tangent - distribution on @xmath2 _ , or briefly _ @xmath6distribution _ , if @xmath7 is an @xmath5dimensional subspace of @xmath8 .
the smoothness of @xmath9 means that : + for each @xmath10 , there exists an open neighborhood @xmath11 of @xmath12 ; and smooth vector fields @xmath13 such that : @xmath14 a map @xmath15 is called an _ @xmath16dimension cotangent - distribution on @xmath2 _ , or briefly _ @xmath17distribution _ , if @xmath18 is an @xmath16dimensional subspace of @xmath19 .
the smoothness of @xmath20 means that : + for each @xmath21 , there exists an open neighborhood @xmath11 of @xmath12 and smooth @xmath22forms @xmath23 such that : @xmath24 [ [ section ] ] in the sequel , without loss of generality , we can assume these definitions are globally satisfied .
there is a correspondence between these two types of distributions . for @xmath6distribution @xmath9
, there exist nowhere zero smooth vector fields @xmath25 on @xmath2 such that : @xmath26 ; and similarly , for @xmath27distribution @xmath20 , there exist global smooth @xmath22forms @xmath28 on @xmath2 such that @xmath29 .
[ exa:1 ] * ( cartan distribution ) * @xmath30 let @xmath31
. denote the coordinates in @xmath2 by @xmath32 and given a function @xmath33 consider the following differential @xmath22forms : @xmath34 and the distribution @xmath35 .
this is the @xmath22dimensional distribution , called the _
cartan distribution_. this distribution can also be described by a single vector field @xmath36 , @xmath37 , where @xmath38 [ exa:2 ] ( * @xmath0gordon equation * ) let @xmath39 be a differentiable function .
the corresponding @xmath0gordon pde is @xmath40 .
we construct @xmath41dimensional sub - manifold @xmath2 defined by @xmath42 , of @xmath43 consider the @xmath22forms @xmath44 this distribution can also be described by the following vector fields @xmath45 let @xmath46 be a @xmath6distribution and set @xmath47 it is clear that @xmath48 .
an @xmath22form @xmath49 _ annihilates _
@xmath9 on a subset @xmath50 , if and only if @xmath51 for all @xmath21 .
the set of all differential @xmath22forms on @xmath2 which annihilates @xmath9 , is called _ annihilator of @xmath9 _ and denoted by @xmath52 .
therefore , for each @xmath53-distribution @xmath54 we can construct a @xmath27distribution @xmath55 and viceversa ; in the other words , for each @xmath56distribution @xmath57 , we can construct a @xmath27distribution @xmath58 , and viceversa .
[ pro:1 ] a ) @xmath9 and its annihilator , are modules over @xmath59 .
+ b ) let @xmath36 be a smooth vector field on @xmath2 and @xmath60 , then @xmath61 _ proof : _ ( a ) is clear , and for ( b ) , if @xmath62 belongs to @xmath63 , then @xmath64 , and + @xmath65 = -\omega[x.y ] = -(\omega\circ { \cal l}_x)y.$ ] @xmath66
let @xmath9 be a distribution . a bijective immersed sub - manifold @xmath50 , is called an _ integral manifold of @xmath9 _ if one of the following equivalence conditions is satisfied : * @xmath67 , for all @xmath68 . *
@xmath69 .
@xmath50 is called _ maximal integral manifold _ if for each @xmath68 , there exists an open neighborhood @xmath11 of @xmath12 such that there is no integral manifold @xmath70 containing @xmath71 .
it is clear that , the dimension of maximal integral manifold does not exceed , the dimension of the distribution .
@xmath72 is called a _ completely integrable distribution _ , or briefly cid , if for all maximal integral manifold @xmath73 , one of the following equivalence conditions is satisfied : * @xmath74 . *
@xmath75 for all @xmath68 * @xmath76 , and if @xmath70 be an integral manifold with @xmath77 , then @xmath78 .
[ [ section-1 ] ] in the sequel , the set of all maximal integral manifolds is denoted by @xmath79 .
@xmath80 ; that is @xmath81 for @xmath82 . *
( continuation of example [ exa:1 ] ) * @xmath30 [ exa:1c ] if @xmath73 is an integral curve of the distribution then @xmath12 can be chosen as a coordinate on @xmath73 , and therefore @xmath83 conditions @xmath84 imply that @xmath85 or that @xmath86 for some function @xmath87 .
the last equation @xmath88 gives us an ordinary differential equation @xmath89 .
the existence theorem shows us once more that the integral curves do exist , and therefore the cartan distribution is a cid . *
( continuation of example [ exa:2 ] ) * @xmath30 [ exa:2c ] this distribution in not cid , because there is no @xmath90dimensional integral manifold , and @xmath91 . for ,
if @xmath73 be a @xmath90dimensinal integral manifold of the distribution , then @xmath92 can be chosen as coordinates on @xmath73 , and therefore @xmath93 condition @xmath94 imply that @xmath95 , which is impossible . by the same reason ,
we conclude that there is not any @xmath96dimensional integral manifold .
now , if @xmath73 be a @xmath97dimensinal integral manifold of the distribution , then @xmath98 can be chosen as coordinates on @xmath73 , and therefore @xmath99 conditions @xmath94 and @xmath100 imply that @xmath101 , @xmath102 , @xmath103 , and @xmath104 .
the last equation @xmath105 implies that @xmath106 .
this distribution is not cid .
in this section , we consider a distribution @xmath107 on manifold @xmath108 .
a diffeomorphism @xmath109 is called a _ symmetry of _ @xmath9 if @xmath110 for all @xmath21 .
therefore , [ pro:3 ] _ the following conditions are equivalent : _ * @xmath111 is a symmetry of @xmath9 ; * @xmath112s determine the same distribution @xmath63 ; that is @xmath113 ; * @xmath114 for @xmath115 ; * @xmath116 , where @xmath117 ; * @xmath118 for all @xmath21 and @xmath115 ; and * @xmath119 , where @xmath120 .
if @xmath111 be a symmetry of @xmath9 and @xmath73 be an integral manifold , then @xmath121 is an integral manifold .
_ proof : _
@xmath111 is a diffeomorphism , therefore @xmath121 is a sub - manifold of @xmath2 . from other hand , if @xmath68 , then @xmath122 for all @xmath115 , therefore @xmath123 is an integral manifold .
@xmath66 let @xmath79 be the set of all maximal integral manifolds , and @xmath109 be a symmetry , then @xmath124 .
_ proof : _ if @xmath125 , then @xmath122 for all @xmath115 , therefore @xmath126 ; and @xmath127 .
now if @xmath128 , then there exists @xmath21 such that @xmath129 , since @xmath111 is a diffeomorphism .
therefore @xmath130 for all @xmath115 ; thus @xmath125 and @xmath131 .
@xmath66 a vector field @xmath36 on @xmath2 is called an _ infinitesimal symmetry of distribution _
@xmath9 , or briefly a _ symmetry of @xmath72 _ , if the flow @xmath132 of @xmath36 be a symmetry of @xmath9 for all @xmath133 . _
a vector field @xmath134 is a symmetry if and only if @xmath135 _ _ proof : _ let @xmath36 is a symmetry . if @xmath136 , then @xmath137 , by the ( 3 ) of theorem [ pro:3 ] ; moreover by the definition @xmath138 , one gets @xmath139 therefore @xmath140 . in converse ,
let @xmath141 or @xmath142 for @xmath115 and @xmath143 .
now , if @xmath144 , then @xmath145 and @xmath146 where @xmath147 , and @xmath148 therefore , @xmath149 is a solution of the linear homogeneous system of odes ( [ eq:2 ] ) with initial conditions ( [ eq:1 ] ) , and @xmath150 must be identically zero . @xmath66 @xmath36 is symmetry if and only if for all @xmath151 then @xmath152\in{\cal d}$ ] _ proof : _ by above theorem .
@xmath36 is symmetry if and
only if ; for all @xmath153 then @xmath154 . the theorem comes from the theorem [ pro:1]-(b ) .
@xmath155 on @xmath72 . in other words ,
@xmath156 $ ] for all @xmath151 .
denote by @xmath157 the set of all symmetries of a distribution @xmath72 . *
( continuation of example [ exa:1c ] ) * @xmath30 [ exa:1cc ] let @xmath158 . a vector field @xmath159 is an infinitesimal symmetry of @xmath9 if and only if @xmath160mod@xmath161 , for @xmath162 .
these give two equations @xmath163 * ( continuation of example [ exa:2c ] ) * @xmath30 [ exa:2cc ] we consider the point infinitesimal transformation : @xmath164 then , @xmath165 is an infinitesimal symmetry of @xmath9 if and only if @xmath166mod@xmath161 , for @xmath167 .
these give ten equations @xmath168 complicated computations using maple , shows that : @xmath169 and @xmath170 , @xmath171 and @xmath172 must satisfy in pde : @xmath173
let @xmath174 , and @xmath73 maximal integral manifold .
then @xmath36 is tangent to @xmath73 . _
proof : _ let @xmath175 . then exists open set @xmath11 of @xmath12 and sufficiently small @xmath176 .
such that @xmath177 is a smooth sub manifold of @xmath2 . since @xmath178
, so @xmath179 is an integral manifold . since @xmath180
, so tangent to @xmath181 belongs to @xmath72 , for all @xmath182 . on the other hand , tangent spaces to @xmath179
are sums of tangent spaces to @xmath183 and the @xmath22dimensional subspace generated by @xmath36 , but both of them belongs to @xmath72 . and there means @xmath184 if @xmath185 and @xmath73 be a maximal integral manifold , then @xmath186 for all @xmath133 .
a distribution @xmath72 is completely integrable , if and only if it is closed under lie bracket . in other words , @xmath152\in{\cal d}$ ] for each @xmath187 . _
proof : _ let @xmath73 be a maximal integral manifold with @xmath75 .
therefore , for all @xmath187 there @xmath36 and
@xmath62 are tangent to @xmath73 and so @xmath152 $ ] , is also tangent to @xmath73 .
on the other hand , let for all @xmath187 there @xmath152\in{\cal d}$ ] . by theorem
, all @xmath188 is symmetry too . and so all @xmath178 is tangent to @xmath73 . and
this means @xmath189 , for all @xmath68 .
a distribution @xmath72 is completely integrable if and only if @xmath190 .
let @xmath191 be completely integrable distribution and @xmath178 . then the differential @xmath22forms @xmath192 vanish on @xmath72 for all @xmath133 .
_ proof : _
@xmath72 is completely integrable , then @xmath36 is symmetry . hence @xmath193
if an ( infinitesimal ) symmetry @xmath36 belongs to the distribution @xmath72 , then it is called a _
characteristic symmetry_. denote by @xmath194 the set of all characteristic symmetries .
any symmetry @xmath200 generates a flow on @xmath79 ( the set of all maximal integral manifolds of @xmath9 ) , and , in fact the characteristic symmetries generate trivial flows . in other words , classes @xmath201 mix or `` shuffle '' the set of all maximal manifolds . *
( continuation of example [ exa:2cc ] ) * @xmath30 [ exa:2ccc ] in this case , we have @xmath205 in @xmath199 . therefore @xmath199 spanned by @xmath206 where @xmath207 and @xmath170 , @xmath171 and @xmath172 must satisfy in pde : @xmath208 * ( quasilinear klein - gordon equation ) * in this example , we find the shuffling symmetries of quasilinear klein - gordon equation : @xmath209 , as an application of previous example , where @xmath210 , @xmath211 , and @xmath150 are real constants . the equation can be transformed by defining @xmath212 and @xmath213 .
then , by the chain rule , we obtain @xmath214 .
this equation reduce to @xmath215 by @xmath216 , @xmath217 and @xmath218 . by solving the pde ( [ eq:2 - 1 ] ) ,
we conclude that @xmath219 spanned by the three following vector fields : @xmath220 for example , we have @xmath221 and if @xmath222 be a solution of ( [ eq:3 ] ) , then @xmath223 is also a new solution of ( [ eq:3 ] ) , for sufficiently small @xmath224 . 9 d.v .
alekseevskij v.v .
lychagin a.m. vinogradov basic ideas and concepts of differential geometry , geometry vol .
i springer - verlag , new york , heidelberg , berlin 1991 a. barone f. esposito c.j .
magee a.c .
scott theory and applications of the sine - gordon equation riv .
nuovo cim .
1 227 - 267 1971 h. kragh equation with the many fathers .
the klein - gordon equation in 1926 american jornal of physics 52 , issue 11 , 1024 - 1033 1984 a. kushner v.v .
lychagin v. rubtsov contact geometry and non - linear differential equations cambridge university press 2007 | in this paper we describe the geometry of distributions by their symmetries , and present a simplified proof of the frobenius theorem and some related corollaries
. then , we study the geometry of solutions of @xmath0gordon equation ; a pde which appears in differential geometry and relativistic field theory . |
mechanical ventilation is the most frequently used life - sustaining intervention in the intensive care unit ( icu ) , where approximately 50% of patients receive ventilatory support @xcite . at some point in their management , many patients on mechanical ventilation ( mv ) are described as fighting their ventilator " .
this jargonistic expression is used to indicate a mismatch between patient respiratory efforts and ventilator breaths " .
this form of disharmony between patient and ventilator results in an increased work of breathing and is a major source of discomfort for the patient with some deleterious effects such as dyspnea and anxiety .
dyspnea and anxiety are major drivers of post - traumatic stress disorders frequently observed in patients who survive the icu @xcite .
therefore , it is crucial to detect patient - ventilator disharmony as early as possible .
currently , this relies on monitoring of physiological signals generated indirectly ( pressure , air flow ) or directly ( electromyography ) by the respiratory muscles in response to the descending neural drive to breathe .
it is generally assumed that ventilatory support should be adapted to the neural drive to breathe @xcite .
some approaches address this issue by using the diaphragmatic emg ( e.g. the neurally adjusted ventilatory assist or nava @xcite ) .
nevertheless , these techniques fail to take into account the fact that , under certain circumstances , the automatic respiratory activity of the brain stem is supplemented by respiratory - related cortical circuits .
indeed , inspiratory loading in awake humans elicits a cortical response that can be observed in eeg signals @xcite .
such responses have been found to correlate with respiratory discomfort in healthy subjects fighting a ventilator @xcite .
these observations give rise to the prospect of an effective brain - ventilator interface ( bvi ) that would target the neural correlates of respiratory discomfort rather than the automatic drive to breathe @xcite .
current neuroscience research attempts to understand how brain functions result from dynamic interactions in large - scale cortical networks , and to further identify how cognitive tasks or brain diseases contribute to reshape this organization @xcite .
covariance analyses of brain data are widely used to elucidate the functional interactions between brain regions during different brain states .
the relevance of covariance matrices as a feature for bci has been already assessed @xcite and they constitute a very appropriate choice given their ability to reflect spatio - temporal dynamics in eeg . in this paper , we use elements of differential geometry to evaluate the ability of eeg covariance matrices to characterize changes in respiratory states in healthy subjects . the use of brain computer interfaces ( bci ) is increasingly common in clinical environments as a technology to improve patient communication and rehabilitation using brain signals @xcite . however , the application of bci in the respiratory context has not yet been explored . in this work ,
we propose a framework which provides the basis of a brain - ventilator interface ( see figure [ fig : plan_bvi ] ) .
a first possible implementation consists of an open loop configuration that generates an output signal to trigger an alarm in the case of breathing disharmony .
the second implementation is a more advanced version that could generate a continuous output signal in a closed loop to directly adapt ventilator parameters to the patient s needs .
the different blocks in the proposed bvi are as follows : 1 .
acquisition : set of electrodes , amplifiers and a / d converter providing digitized eeg signals .
pre - processing : improves signal - to - noise ratios in eeg signals by applying artifact correction / rejection and/or filters .
feature extraction : sample covariance matrices ( cms ) are obtained from segmented , pre - processed signals .
4 . classification : cms are labelled according to two possible classes : normal and altered breathing .
detection of anomalous respiratory states is achieved by one - class learning , measurement of the distance between a number of reference matrices learned during reference condition and the cm corresponding to a particular signal epoch .
since cms do not lie in vector space , appropriate distance metrics must consider their natural geometry .
translation : external application that converts the binary signal from the classifier to an alarm or ventilator command .
the present paper focuses on the signal processing aspects of the bvi pre - processing , feature extraction and classification blocks as a detector of respiratory - related activities compatible with breathing discomfort .
the framework is validated with eeg from healthy subjects under two breathing constraints to emulate patient - ventilator disharmony .
the reliability and performances of our method are also compared with those obtained by a common spatial patterns ( csp ) method in combination with linear discriminant analysis ( lda ) . the experimental protocol and data
are detailed in section [ database ] .
section [ meth ] describes the different signal processing blocks of the bvi , including our riemann geometry based classifier and other standard classification methods that use eeg and breathing signals .
section [ setup ] studies bvi settings for optimal detection of breathing discomfort and section [ res ] provides the experimental results and evaluation of the bvi .
finally , we conclude the paper with a discussion in section [ concl ] .
the database is composed of nine healthy subjects ( 21 - 29 years ; 5 women ) with no prior experience with respiratory or neurophysiology experiments ( for more details see @xcite ) . according to the declaration of helsinki , written informed consent
was obtained from each subject after explanation of the study , which was approved by the local institutional ethics committee ( comit de protection des personnes ile - de - france vi , groupe hospitalire piti - salptrire , paris , france ) .
subjects were sitting in a comfortable chair and breathed continuously through a mouthpiece .
they were asked to avoid body and head movements .
they were distracted from the experimental context by watching a movie during the entire experiment , on a screen placed in front of them . to minimize emotional interference ,
the movie was a neutral animal documentary .
it is worthy to note that , even though many icu patients are supine , clinical practice guidelines recommend that mechanically ventilated patients in the icu should be semi - supine , or semi - seated , to decrease the risk of respiratory infections and increase patient comfort @xcite .
electroencephalographic activity was recorded via surface electrodes ( acticap , brainproducts gmbh , germany ) using 32 electrodes according to the standard 10 - 20 montage and sampled at 2500 hz .
impedance between electrodes and skin were set below 5 k@xmath0 .
the mouthpiece was connected to a pneumotachograph ( hans ruldoph inc .
, mo , usa ) and a two - way valve to measure air flow and attached , when required , an inspiratory load ( range 18 - 25 cmh20 ) . the experiment was designed to activate cortical regions by altered breathing and consisted of three parts : 1 .
normal , spontaneous ventilation ( sv condition ) .
breathing is controlled automatically by the autonomous nervous system without cortical contribution .
2 . voluntary brisk inhalations or sniffs ( sn condition ) .
breathing movements are planned before execution , thus motor and pre - motor cortical regions are solicited .
3 . inspiratory loaded breathing ( ld condition ) .
ventilatory muscles perform a supplementary effort to overcome an inspiratory threshold load to maintain adequate air flow , a condition known to engage cortical networks @xcite .
in contrast to the sv condition , where breathing is comfortable , ld is associated with respiratory discomfort . in a clinical context
, sv would correspond to patient - ventilator harmony , whereas ld condition would correspond to patient - ventilator disharmony .
sn can be considered as a positive control condition where cortical control is expected .
for all subjects , 10 minutes of eeg was recorded for each condition , i.e. 10 min of sv , 10 min of sn and 10 min of ln .
the experiments were well tolerated by all subjects and no intervention was necessary to modify the amount of load or sniff pattern during the recordings . hence , during 10-minute recordings for each condition , no relevant changes occurred within each of these 10-min blocks regarding experimental conditions or subject behavior .
the purpose of this block is to enhance motor cortical activity , whose main rhythms are within @xmath1 and @xmath2 bands @xcite .
therefore , signals were band - pass filtered by a linear phase fir filter ( see section [ ch_freq ] for more details regarding frequency selection ) .
data segments with artifacts due to repetitive eye blinks and ocular movements were visually detected and removed from the original eeg dataset .
then , signals were down - sampled to 250 hz and segmented in 5 second sliding , 50% overlapped windows to reduce computational cost in subsequent blocks .
this time interval is in concordance with the slow breathing dynamics ( a breath every 2.5 to 5 seconds ) . as mentioned above ,
the basis to classify the breathing state in the bvi are sample covariance matrices .
the feature extraction block processed eeg data in epochs of @xmath3 samples and @xmath4 channels , as a matrix @xmath5 , and then transformed to a sample covariance matrix @xmath6 .
the latter was computed by the unbiased estimator : @xmath7 where the superscript @xmath8 denotes the matrix transpose . by construction , @xmath9
are symmetric and positive - definite matrices ( spd ) that do not lie in a vector space but on a riemannian manifold @xcite .
therefore , previous methods defined on a euclidean structure are not longer adequate .
the correct manipulation of these spd matrices requires the application of riemannian geometry concepts , described in section [ riemann ] .
thanks to the ability of covariance matrices to capture eeg spatial dynamics , a one - class approach can be chosen to solve the classification problem in a bvi . in one - class algorithms ,
the availability of the labels from only one class is enough to classify instances from a second class , as the latter can be mapped in a different ( distant ) region of the space representation . in our framework ,
motor cortex activation provoked during uncomfortable breathing should be different to that underlying normal , comfortable breathing ( absence of motor cortical activity ) .
the objective of the classifier is to learn from @xmath10 data samples of the reference class sv ( @xmath11 ) in order to label new trials @xmath12 in two classes , @xmath13 , which correspond to breathing comfort ( sv condition ) and discomfort ( sn or ld conditions ) , respectively . during the learning process ,
the algorithm first finds @xmath14 matrices , subsequently called prototypes , serving as reference to perform classification .
those prototypes constitute @xmath14 centers from the ensemble of covariance matrices @xmath15 and they are estimated by means of a general @xmath14-means clustering algorithm : 1 .
initialize the prototypes @xmath16 , @xmath17 by random selection of @xmath14 matrices from @xmath15 , 2 . for each sample covariance matrix ,
compute its distance @xmath18 to all the prototypes and assign it to the closest one in order to form the cluster @xmath19 : @xmath20 3 . update each prototype by averaging points in @xmath19 as @xmath21 for points lying in a vector space
, @xmath16 corresponds to the arithmetic mean of points in @xmath19 4 .
go to step ( 2 ) until convergence ( e.g. the assignments no longer change ) is achieved .
the resulting prototypes @xmath16 represent the class @xmath22 , and are then used to classify the new unlabelled data @xmath23 ( sub - index @xmath24 denotes the number of the @xmath3-sample window ) according to : @xmath25 where @xmath26 is the distance to the closest reference prototype , @xmath27 ; and @xmath28 is a scalar that can be adjusted to a performance criterion , like a statistical significance or a desired specificity / sensitivity value , for instance .
classification performance was measured by the area under the curve ( auc ) of the receiver operating characteristic ( roc ) .
auc values range from 0.5 ( a random classification ) to 1 ( perfect classification ) .
all auc values were computed by applying 10-fold cross validation , excluding the learning period in the classification . within an euclidean framework ,
the @xmath14-means algorithm divides the dataset into groups and attempts to minimize the euclidean distance between samples labeled to be in a cluster and a point designated as the arithmetic mean of that cluster .
nevertheless , the spd manifold is not a linear space with the conventional matrix addition operation .
a natural way to measure closeness on a manifold is by considering the geodesic distance between two points on the manifold @xcite .
such distance is defined as the length of the shortest curve connecting the two points . as an example , consider the earth s surface as a manifold : the riemannian distance between the two poles is given by a meridian , while the euclidean distance corresponds to the straight line going through the earth s core from pole to pole . in the space of sdp matrices ,
the clustering algorithm must minimize the geodesic distances between each point of the manifold ( the covariance matrices @xmath29 ) and the reference matrix @xmath16 .
each cluster center can be then obtained by an averaging process that employs the intrinsic geometrical structure of the underlying set . derived from different geometrical , statistical or information - theoretic considerations , various distance measures
have been proposed for the analysis of spd matrices @xcite .
although many of these distances try to capture the non - linearity of spd matrices , not all of them are geodesic distances .
the space of spd matrices , @xmath30 , constitutes a differentiable riemannian manifold @xmath31 of dimension @xmath32 . at any point @xmath33
there is a tangent euclidean space , @xmath34 .
let two points @xmath35 and @xmath36 be two points on the tangent space ( e.g. the projection of two spd matrices ) , the scalar product in the tangent space at @xmath37 is defined by @xmath38 that depends on point @xmath37 .
the logarithmic map that locally projects a covariance matrix @xmath9 onto the tangent plane is given by : @xmath39 where @xmath40 is the matrix logarithm operator .
projecting sdp matrices onto @xmath34 is advantageous because this tangent space is euclidean and distance computations in the manifold can be well approximated by euclidean distances in @xmath34 .
the inverse operation that projects a point @xmath41 of the tangent space back to the manifold @xmath31 is given by the exponential mapping : @xmath42 in this paper , we have employed the two most widely used distance measures in the context of riemannian manifolds : the affine - invariant distance @xcite and the log - frobenius distance , also referred to as log - euclidean distance @xcite . for comparative purposes ,
we have also used an euclidean metric . on the space of real square matrices
, we have the frobenius inner product @xmath43 and the associated metric : @xmath44 given a set of @xmath45 real square matrices @xmath46 , their arithmetic mean is given by : @xmath47 using ( [ e1 ] ) , the riemannian distance between two spd matrices can be computed as : @xmath48 this metric has several useful theoretical properties : it is symmetric and satisfies the triangle inequality . furthermore , it is scale , rotation and inversion invariant @xcite . to find the mean of a set of @xmath45 covariance matrices @xmath46 ,
the distance @xmath49 needs to be applied in the following expression @xcite : @xmath50 although no closed - form expression exists , this ( geometric ) mean can be computed efficiently using an iterative algorithm .
the geometric mean in @xmath30 converges into a unique solution @xcite and can be computed efficiently by the algorithm described in @xcite .
the log - euclidean distance between two spd matrices is given by @xcite : @xmath51 this metric maps spd matrices to a flat riemannian space ( of zero curvature ) so that classical euclidean computations can be applied @xcite . under this metric
, the geodesic distance on the manifold corresponds to a euclidean distance in the tangent space at the identity matrix .
this metric is easy to compute and preserves some important theoretical properties , such as scale , rotation and inversion invariance @xcite . given a set of @xmath45 covariance matrices @xmath46 , their log - euclidean mean exists and is uniquely determined by @xcite : @xmath52 in this paper , we have also evaluated the detection of altered respiratory states using eeg and air flow - based features as inputs for one - class support vector machine ( ocsvm ) classifiers .
although one - class classification , or novelty detection , is more appropriate for the clinical monitoring of physiological condition in ventilated patients ( i.e. altered breathing epochs may be detected as departures from `` normal '' breathing state ) , we have compared our results with those obtained from a common spatial pattern ( csp ) method in combination with linear discriminant analysis ( lda ) to detect altered breathing from eeg .
csp is a two - class oriented approach that needs a - priori labelled data from two classes previously defined ( the normal and the altered breathing conditions ) to train the classifier , while our one - class approach learns from a single training set containing only normal respiratory epochs .
however , we have included these methods in the comparative study as they constitute a standard in brain computer interfacing @xcite .
one - class support vector machines is a very popular machine learning technique used as outlier or novelty detector for a variety of applications @xcite . in ocsvm ,
the support vector model is trained on data that has only one class , which represents the _ normal _ class .
this model attempts to learn a decision boundary that achieves the maximum separation between the points and the origin .
a ocsvm firstly uses a transformation function defined by a kernel function to project the data into a higher dimensional space .
the algorithm then learns a decision boundary that encloses the majority of the projected data , and that can be applied to the outliers detection problem .
more details on the algorithmic aspects of one - class svms can be found in @xcite . to guarantee the existence of a decision boundary we have used here a gaussian kernel : @xmath53 where @xmath54 and @xmath55 denote two feature vectors and the parameter @xmath56 is set to be the median pairwise distances among training points @xcite . for this classifier
, feature vectors @xmath57 were extracted by computation of covariance matrices @xmath58 in epoched data @xmath23 followed by determination of the vector form of the upper triangular part of @xmath58 .
air flow features were computed , as for eeg classifiers , in 50% overlapped @xmath3-sample windows . in this case ,
feature vectors @xmath57 are composed of six air flow descriptors computed in each window @xcite : peak value ( l / s ) , average flow ( l / s ) , total volume ( l ) , flow variance , skewness and kurtosis .
after an exhaustive feature selection procedure ( testing the classifier with all @xmath59 possible combinations ) using 10-fold cross validation , the greatest averaged aucs was provided by the combination of 3 features : air flow peak value , variance and skewness .
csp is a widely used spatial filtering technique in bcis to find linear combinations of electrodes such that filtered eeg maximizes the variance difference between two classes @xcite .
the computation of csp yields a projection matrix @xmath60 whose column vectors @xmath61 ( @xmath62 ) , called spatial filters , can be considered as an operator that transforms eeg to `` source '' components .
the matrix @xmath63 contains the so - called spatial patterns @xmath64 in its columns which can be considered as eeg source distribution vectors . a standard procedure to extract features in eeg epoched data ( @xmath23 ) by csp consists of computation of the log - variances of projected @xmath23 on the spatial filters @xcite .
following the recommendations in @xcite , we used the 4 most relevant spatial filters to obtain the feature vectors @xmath57 , which were then used as inputs in a lda classifier .
following bci designs , the bvi should exploit the oscillatory modulations on @xmath1 and @xmath2 bands ( 8 - 30 hz ) following motor and somato - sensorial cortical activations @xcite .
very low frequency potentials associated with voluntary movements @xcite may also be present during respiratory tasks as evidenced by previous work @xcite .
to study the impact of specific frequency bands on the efficacy of the bvi to detect altered breathing , we compared several frequency ranges ( from 0 to 30 hz ) and bandwidths ( 4 to 22 hz ) to tune the bandpass filter used in the preprocessing block ( see section [ res : freq ] ) . in bci ,
the choice of the adequate number of sensors and their location is fundamental . reducing the number of irrelevant electrodes avoids over - fitting and optimizes computational costs , but also improves patient comfort and reduces installation time .
several channel selection methods can be employed to simplify the implementation of bcis , being the most popular those based on common spatial pattern ( csp ) @xcite and svms @xcite .
in addition to csp based channel selection , we propose an iterative procedure to rank electrodes according to their discriminating power , the common highest order ranking ( chorra ) .
results obtained by both methods are compared in section [ elect_rankings ] ) .
chorra is performed in two steps .
firstly , an electrode ranking for each subject ( intra - subject rankings ) is found .
then , a general ranking of the most significant electrodes is computed over all subjects ( inter - subject rankings ) . to compute intra - subject ranks ,
a recursive backward elimination of electrodes was applied until the remaining number was 2 , the minimal dimension to compute a covariance matrix ( see algorithm 1 ) .
initialise the subset @xmath65 with the initial @xmath66 electrodes @xmath67 = classification performance without @xmath68 @xmath69 = @xmath68 that provides the minimal @xmath70 remove the electrode @xmath68 from the list @xmath65 * return * @xmath71 at the end of this procedure , the list @xmath71 contains the indices to the @xmath66 electrodes sorted from more to less relevant in terms of classification performances for a particular subject . combining ranking lists from all subjects , a general configuration for a ready - to - use bvi device can be chosen .
we have tested two possible rank combinations : 1 .
* ranking aggregation * : this method determines electrode position in the ranked list , compared to a null model where all the lists are randomly shuffled @xcite . based on order statistics
, this algorithm assigns a @xmath72-value to each electrode in the aggregated list describing the rank improvement compared to expected ( a score of @xmath73 is assigned to channels that are preferentially ranked at the top of all lists ) .
the procedure only takes into account the best rank , thus providing a robust re - ordering of elements @xcite .
* averaged ranking * : this is a classical positional method where the total score of an electrode is simply given by the arithmetic mean of the positions in which the channel appears in each ranking list . to increase rank robustness ,
several learning periods were used .
hence , after obtaining a set of @xmath74 lists ( @xmath74-fold cross validation ) per subject , the intra - subject ranking was computed according to one of the above rank combination applied to the @xmath74 lists of electrode rankings . as the electrode positions leading to the maximal auc in a particular subject is nt necessarily the same for another , we studied a common electrode configuration that could be shared by any user .
this general electrode set - up is more pertinent in a clinical context as an eeg headset could be interfaced promptly .
therefore , we applied the above rank combination methods to all intra - subject rankings to obtain common and unique electrode lists .
since csp patterns @xmath64 ( @xmath62 ) can be considered as eeg source distribution vectors , their associated weights can be used to find the most relevant electrodes with regard to the discrimination of the two classes @xcite . to compute intra - subject csp - based ranks , we retained the first pattern ( @xmath75 ) as it explains the largest variance for class 0 ( and the lowest for class 1 ) and the last pattern ( @xmath76 ) , that explains the largest variance for class 1 ( and the lowest for class 0 ) . within each pattern , the largest weights ( absolute values )
are associated with the most relevant electrodes .
hence , the list of electrode ranks was generated by selecting , alternately , in decreasing order , the first weight in @xmath75 , then the first weight in @xmath76 , the second weight in @xmath75 and so on as proposed by wang et al @xcite .
the final list of inter - subject csp ranks were obtained by normalizing ( by their maximal value ) @xmath75 and @xmath76 in each subject , then finding the average of these patterns across subjects as in @xcite , and finally by selecting the weights as done above for intra - subject ranks . for consistency with chorra ranks ,
a similar @xmath74-fold procedure was applied : csp was computed using @xmath74 different mean covariance matrices in class 0 .
the average of the resultant @xmath74 projection matrices yielded the final @xmath77 .
as mentioned above , the prototypes are local centers that represent the space defined by @xmath15 . to this end
, we adopted the @xmath14 prototypes provided by the general @xmath14-means algorithm as the centers @xmath16 introduced in section [ classif ] .
the structure of the manifold defined by @xmath15 is unknown .
covariance matrices may be organized in a compact , uniform manner so a few centers are enough to correctly represent class 0 . on the contrary
, @xmath15 may have a complex distribution and need more centers to be correctly represented .
therefore , in order to achieve good classification , there may be an optimal number of prototypes . to satisfactorily exploit the bvi ,
previous knowledge about the learning time needed to train the classifier is necessary .
limited learning time may result in a small number of covariance matrices @xmath10 and hence , in a poor representation of the space defined by normal breathing .
moreover , the clustering algorithm would provide a less accurate center when few cms are available . on the other hand ,
the learning time should meet clinical requirements , as long set - up times in these environments are impractical .
since the number of covariance matrices needed to represent class 0 may impact on the optimal number of prototypes , the classification performance has to be computed by modifying @xmath10 and @xmath14 . in next section ,
these parameters are tested experimentally to find the optimal values .
this section shows and discusses the results of applying the bvi to detect altered respiratory condition ( ld or sn ) after learning a reference period of normal breathing ( sv ) . to find the best settings in the classification block
, several methods estimated the distances ( @xmath26 ) to the reference condition sv .
distances were then employed to estimate the areas under roc curves ( auc ) as a measure of the classifier s performance .
plots in figure [ fig : alpha_delta ] illustrate the feature distances and corresponding aucs for detection of ld condition in one single subject . for the eeg - based classifier , the optimal frequency band for preprocessing signals was first selected .
then , euclidean , riemannian and log - euclidean distances were compared to select the best metric used for evaluating the optimal electrode configuration .
finally , the impact of learning time and the number of prototypes were also assessed .
as seen in figure [ fig : auc_freq ] , the frequency bands that provided the highest discriminant rates in terms of auc were 8 - 24 hz and 8 - 30 hz . since the latter is more susceptible any muscular artifacts @xcite , we preferred to set the bandpass cut - off frequencies between 8 - 24 hz .
low frequencies ( @xmath784 hz ) were discarded not only for their moderate discrimination power , but also because in a real implementation of the bvi this spectral range would be corrupted by electromagnetic interferences inherent in clinical environments .
other frequency combinations were not superior to 8 - 24 hz classification rates .
the first test used the 14 most central electrodes , thus covariance matrices were obtained from eeg signals @xmath79 .
the classifier settings were @xmath80 and @xmath81 .
figure [ fig : auc_ini ] shows that the best performing metric is the log - euclidean , which provided slightly better auc values than riemannian distances in ld condition ( @xmath82 versus @xmath83 ) and almost equal auc values ( @xmath84 ) for the sn condition .
the classification performances for sn had the largest variability probably due to the discontinuous occurrence of sniffs ( one every two breaths ) .
although the performances of log - euclidean are similar to those obtained by the affine invariant riemannian metric , the former is advantageous from the computational point of view due to its reduced algorithmic complexity ( cpu times were , on average , 3 time faster ) .
euclidean distances provided the lowest average classification rates and the most unstable auc values , demonstrating the limitation of linear matrix operators ( i.e. the arithmetic mean ) for eeg covariance matrix classification .
the analysis of @xmath85 values also proves that riemannian geometry is a better framework to classify eeg covariance matrices , than the outlier detector based on one - class svms .
we notice , however , that the latter provided better results than the classifier with euclidean structure ( mean values are @xmath86 versus @xmath87 ) . in ld condition , classification performances obtained with air flow ( @xmath88 ) remained below than those obtained using log - euclidean and affine invariant distances , and almost equal to the one - class svms eeg - based detector ( @xmath89 ) .
because of the characteristic large pattern of air flow during sniffs , air flow signals provided better classification in sn than svm eeg - based detector ( @xmath90 and @xmath88 , respectively ) and similar performances to our approach .
finally , if we considered the bvi as a two - class detector ( unsuitable for on - line detection as discomfort classes are not known a priori ) , the classification by csp and lda yielded @xmath91 for ld and @xmath88 for sn , values clearly below our riemannian geometry based approach .
we assessed the most relevant electrodes in the proposed bvi , starting from a set of 14 central electrodes from the original 32-electrode montage .
retained electrodes included positions on primary sensorimotor cortices ( c3 , c4 ) , higher - level secondary and association cortices ( fz and cz ) , pre - parietal ( cp1 and cp2 ) and pre - frontal cortices ( f3 , f4 ) , pre - motor and supplementary motor areas ( fc1 , fc2 ) , central ( fc5 and fc6 ) and parieto - occipital areas ( cp5 and cp6 ) .
the choice of these areas follows results of earlier experiments describing the cortical networks elicited during respiratory load compensation @xcite . according to the chorra procedure to find the best channels from the initial set - up ( c.f algorithm 1 in section [ ch_selelct ] ) , we first found intra - subject rankings and then applied a global rank aggregation .
as described above , we employed auc as the measure of classification performance and performed the rankings from @xmath74=10 lists ( i.e. 10 different learning periods ) of each subject .
the classification results are shown in figure [ fig : auc_elect]-(a ) for intra - subject auc optimized ranks and figure [ fig : auc_elect]-(b ) for intra - subject csp electrode ranks , where auc values are expressed as a function of the number of removed electrodes . in both cases ,
the evolution of the curves during the optimization process display a slight increasing tendency of auc for the log - euclidean metric when electrodes contributing negatively to classifications are removed .
classification performances decrease with configurations smaller than six electrodes .
results show that customized individual selection of electrodes can provide an optimal auc within a single subject . using an intra - subject auc based optimization ,
classification rates can be improved up to 0.95 using the six most significant electrodes . on the other hand ,
selecting the electrodes according to csp patterns provides lower aucs , with the best 6-electrode reaching an average auc value of 0.85 in ld . as shown in different plots of figure [ fig : auc_elect ] , for a configuration with more than three electrodes , the log - euclidean metric performs better than the euclidean distance at every optimization step .
these results support the idea that large spatio - temporal information of the eeg ( as reflected by the sample covariance matrices ) is optimally captured when the intrinsic geometry structure of the underlying data is taken into account .
we applied the two rank combinations proposed by chorra to the lists of sorted electrodes of all subjects to find a general ranking of electrodes for each method .
following the rank of the final lists , we computed the auc values by reducing the initial 14-electrode set to two sensors ( see figure [ fig : auc_elect]-(c - d ) ) .
classifications provided by inter - subject csp ranks are depicted in figure [ fig : auc_elect]-(e ) .
results from both chorra procedures indicate that a good compromise between a reduced number of electrodes and reasonable classifications can be reached by selecting the best 6 electrodes obtained by the averaged ranks ( @xmath92 for both ld and sn conditions ) . in line with intra - subject ranks ,
the general electrode list obtained by csp patterns resulted in poorer classification rates , providing @xmath93 for a set of 6 electrodes .
results indicate that , for general configurations with more than three electrodes , log - euclidean metric still performs better than the euclidean distance .
the aggregated chorra scores displayed in figure [ fig : heads ] indicate scalp zones with more influence for the classification of both sn and ld conditions .
although both optimization procedures ( robust rank aggregation and rank averaging ) provide similar classification performance , the spatial concentration for discrimination of scalp regions obtained by the robust aggregation method is larger than those obtained by a classical averaging rank procedure .
the spatial distribution of scores and the rankings ( see figure [ fig : heads ] ) indicates that discrimination of ld condition is better if the electrode configuration contains the pre - motor and supplementary motor area ( fc1 ) , the fronto - central region ( fc6 ) and the supramarginal gyrus ( the part covered by cp6 electrode ) .
this agrees with previous findings suggesting that during ld , the supplementary motor area ( sma ) is recruited @xcite , most likely within the frame of a cortico - subcortical cooperation allowing compensation of the inspiratory load @xcite . during sn ,
the conscious preparation of a breath activates pre - motor areas and the execution of the breath activates motor areas @xcite .
indeed , our results show that sn condition is better discriminated if electrodes include the supplementary motor area ( fc1 ) , the central motor area cz and part of the somatosensory association cortex ( covered by the electrode cp1 ) .
figure [ fig : heads ] also shows the relative weights of electrode positions related to the most relevant csp patterns . in both ln and sn conditions ,
topographic plots reveal that highly ranked positions reside in similar scalp regions . for loaded breathing ,
csp weights suggest a bilateral activation of the cortex that includes fronto - central , central ( motor cortex ) and some centro - parietal positions ( somato - sensory cortex ) .
on the other hand , during voluntary sniffs , highly scored electrodes match fronto - central positions on the left hemisphere ( motor and pre - motor areas ) .
electrode selection based on csp coefficients converges to a great extent with chorra selection .
nevertheless , for a given number of electrodes our proposed approach provides better classification performances than the counterpart selection in csp ranks ( see figure [ fig : auc_elect ] ) . to confirm that changes in brain activity during ld and sn epochs detected by the eeg - based classifier are related to breathing , correlations between eeg and air flow were also computed after correcting by autocorrelations and time trends present in time series @xcite .
results show that mean correlations in central areas ( including c3 , c4 , fz and fc1 electrodes ) increase about @xmath94 .
the increase of correlation values between eeg and air flow signals confirms that changes in brain state detected by our classifier are related to respiration .
our results suggest that , in general , eeg signals might provide better discriminant features than air flow to detect breathing discomfort .
once the effect of different metrics and channels configurations were assessed , we tested the influence of the learning parameters , i.e. the number of covariance matrices used as a reference ( @xmath10 ) and the number of prototypes ( @xmath14 ) used to characterize the reference class . for this , we assessed classification performance for different values of @xmath14 , ranging from 1 to 7 and and for different number of learning matrices , @xmath10 , from 20 to 40 in steps of 5 .
this procedure was also repeated with 10-fold cross validation for ld and sn conditions .
three different electrode configurations were tested : 1 ) the initial 14-electrode configuration , 2 ) a selection of 6 electrodes optimized for every subject and 3 ) a general configuration with the best 6 electrodes after applying the global channel selection procedure .
results are depicted in the color maps of figure [ fig : auc_kl ] , where average auc values are expressed as a function of @xmath14 and @xmath10 .
importantly , this figure shows a small effect of both parameters on overall classification rates , regardless of the electrode configuration . in view of these findings
, the choice of a small number of prototypes ( @xmath95 ) and a short learning time ( @xmath96 cms ) is an advantageous trade - off between computational complexity and classification performances , even for a common electrode configuration setting .
compared to classification based on air flow , our findings demonstrate a better discriminatory power of the covariance patterns of eeg signals to detect a patient - ventilator disharmony ( @xmath97 versus @xmath98 , in average , for the loaded breathing condition ) .
the analysis of auc values suggests that riemannian geometry is a better framework to manipulate covariance matrices , even with a few number of channels , than classical matrix operators with an euclidean structure .
we notice that one - class svms detector also provides better results than a classifier employing euclidean metrics .
in contrast with other classical methods used in bci , ( as csp which assumes a - priori labelled data from two classes ) , the results support the better discriminant capacity of our approach to identify anomalous respiratory periods , by learning from a single training set containing only normal respiratory epochs .
the proposed recursive channel selection procedure may provide a subject - customized bvi setting with a reduced number of six channels and maintained performance ( @xmath99 ) .
if the channel configuration includes the most significant electrodes across subjects , classification rates are reduced compared to customized optimization .
nevertheless , this general setting provides a good compromise between a reduced number of electrodes ( 6 channels ) and reasonable classification performances ( @xmath100 ) , which is advantageous in general clinical practice where ready - to - use devices are necessary .
interestingly , overall classification performance do not significantly depend on the other parameters of the classifier ( learning time and number of prototypes ) .
similar to our approach , there have been previous attempts to use divergences from information theory ( e.g. kl divergence , jensen - shannon divergences or bregman matrix divergence ) for developing distances on covariance matrices @xcite .
the use of such dissimilarity measures could improve the performance of our algorithm .
an optimization of control parameters for other svms models might also increase classification performances but this subject is beyond the scope of our paper . one major limitation of our approach results from the difficulty to properly estimate the covariance matrices .
poor estimates do not accurately reflect the underlying neural process and thus will directly affect the classifier s performances .
the estimation of a covariance matrix in an eeg segment assumes a multivariate , stationary stochastic process so it can be strongly affected by artifacts such as eye blinks , muscle activity or swallowing .
changes over longer time scales such as changes in electrode impedance , lose of electrodes or sudden electrode shifts , would also deteriorate classification performances .
the use of regularization or adaptive learning approaches ( see @xcite and references therein ) could improve the robustness of the algorithm .
this work provides the first extensive evaluation of what could be a brain - ventilator interface ( bvi ) designed to detect altered respiratory conditions in patients on mechanical ventilation .
the novelty of our proposal is a riemannian geometry approach to identify a cortical signature of breathing discomfort from spatio - temporal eeg patterns . in general , characterization of brain networks
provides meaningful insights into the functional organization of cortical activities underlying breathing control and respiratory diseases .
our study supports the hypothesis of a strong correlation between voluntary and compensatory respiratory efforts and the activation of cortical circuits @xcite .
correlation increase between eeg and air flow during breathing discomfort epochs also corroborates that changes in brain state detected by our classifier are related to respiration .
a hybrid bci combining eeg and airflow signals could be explored in future studies .
the introduction of a bvi is a first step toward a critical class of interfaces for respiratory control applications in a variety of clinical conditions where the use of mechanical ventilation is required to decrease the work of breathing in the patients .
for instance , in patients suffering amyotrophic lateral sclerosis , we have recently reported a respiratory - related cortical activity @xcite .
effective translation of our approach to a suitable device for long - term monitoring of these patients faces difficult challenges that arise from the nature of the heterogeneous population ( e.g. different respiratory problems ) and the numerous sources of artifacts in clinical units . nevertheless , due to its technical simplicity ( portable , non - invasive , with few electrodes and fast computation ) the proposed bvi can be highly operable in clinical environments as well as in custom - designed systems .
indeed , our algorithm @xcite is currently assessed by a large clinical study undertaken at the icu of our group . beyond short - term applications where the bvi would prompt clinicians to run a dyspnea check - list , future work
should also integrate the bvi in a feedback scheme to automatically set ventilator parameters with minimal physician intervention .
this work was supported by the program investissement davenir anr-11-emma-0030 and anr-10-aihu 06 of the french government and by the grant legs poix from the chancellerie de luniversit de paris , france .
x. navarro - sune is financially supported by air liquide medical systems s.a .
hudson was supported by an nhmrc ( australia ) early career fellowship .
10 a. carlucci et al .
srlf collaborative group on mechanical ventilation , `` noninvasive versus conventional mechanical ventilation .
an epidemiologic survey , '' _ american journal of respiratory and critical care medicine _ , vol .
163 , no . 4 , pp .
874880 , 2001 .
p. leung et al .
`` comparison of assisted ventilator modes on triggering , patient effort , and dyspnea , '' _ american journal of respiratory and critical care medicine _ , vol . 155 , no . 6 , pp .
19401948 , 1997 .
m. raux et al .
`` cerebral cortex activation during experimentally induced ventilator fighting in normal humans receiving noninvasive mechanical ventilation , '' _ the journal of the american society of anesthesiologists _ , vol .
107 , no . 5 , pp .
746755 , 2007 .
a. l. hudson et al . , `` electroencephalographic detection of respiratory - related cortical activity in humans : from event - related approaches to continuous connectivity evaluation , '' _ journal of neurophysiology _ , vol . 115 , pp . 22142223 , 2016 .
a. cherian et al .
, `` jensen - bregman logdet divergence with application to efficient similarity search for covariance matrices , '' _ ieee transactions on pattern analysis and machine intelligence _ , vol .
35 , no . 9 , pp .
21612174 , 2013 .
p. t. fletcher and s. joshi , `` principal geodesic analysis on symmetric spaces : statistics of diffusion tensors , '' in _ computer vision and mathematical methods in medical and biomedical image analysis _
3117 , pp . 8798 , 2004 .
m. h. quang et al . , `` a unifying framework for vector - valued manifold regularization and multi - view learning , '' in _ proceedings of the 30th international conference on machine learning ( icml-13 ) _ , pp . 100108 , 2013 .
j. farquhar et al . , `` regularised csp for sensor selection in bci , '' in _ proceedings of the 3rd international brain - computer interface . workshop and training course 2006 .
graz university of technology , austria _ , pp .
1415 , 2006 .
g. macefield and s. c. gandevia , `` the cortical drive to human respiratory muscles in the awake state assessed by premotor cerebral potentials , '' _ the journal of physiology _ , vol .
439 , pp . 545558 , 1991 .
m. raux et al .
, `` functional magnetic resonance imaging suggests automatization of the cortical response to inspiratory threshold loading in humans , '' _ respiratory physiology & neurobiology _ , vol . 189 , no . 3 , pp .
571580 , 2013 .
t. similowski et al.,``method for characterising the physiological state of a patient from the analysis of the cerebral electrical activity of said patient , and monitoring device applying said method , '' 2013 . wo patent wo2013164462 ( a1 ) . | * abstract * during mechanical ventilation , patient - ventilator disharmony is frequently observed and may result in increased breathing effort , compromising the patient s comfort and recovery .
this circumstance requires clinical intervention and becomes challenging when verbal communication is difficult . in this work
, we propose a brain computer interface ( bci ) to automatically and non - invasively detect patient - ventilator disharmony from electroencephalographic ( eeg ) signals : a brain - ventilator interface ( bvi ) .
our framework exploits the cortical activation provoked by the inspiratory compensation when the subject and the ventilator are desynchronized .
use of a one - class approach and riemannian geometry of eeg covariance matrices allows effective classification of respiratory states .
the bvi is validated on nine healthy subjects that performed different respiratory tasks that mimic a patient - ventilator disharmony .
classification performances , in terms of areas under roc curves , are significantly improved using eeg signals compared to detection based on air flow .
reduction in the number of electrodes that can achieve discrimination can often be desirable ( e.g. for portable bci systems ) . by using an iterative channel selection technique , the common highest order ranking ( chorra ) , we find that a reduced set of electrodes ( n=6 ) can slightly improve for an intra - subject configuration , and it still provides fairly good performances for a general inter - subject setting .
results support the discriminant capacity of our approach to identify anomalous respiratory states , by learning from a training set containing only normal respiratory epochs .
the proposed framework opens the door to brain - ventilator interfaces for monitoring patient s breathing comfort and adapting ventilator parameters to patient respiratory needs .
* keywords : * biomedical signal processing , brain - computer interfaces ( bci ) , biomedical monitoring , medical signal detection , electroencephalography ( eeg ) at ( current page.south ) ; |
in 1970 , zeldovich suggested that the large - scale structure of the universe such as sheets and filaments form through the anisotropic collapse of large - scale density fluctuations along the principal axes of the gravitational tidal fields @xcite .
although his original model assumed a hot dark matter cosmology , this picture can be well accommodated even by the currently popular @xmath14-dominated cold dark matter ( @xmath14cdm ) cosmology @xcite .
the dark matter halos that condense out first on the smallest scales merge anisotropically along the principal axes of the tidal fields into larger and larger structures .
the inherent anisotropic nature of the gravitational merging process leads to the formation of filament - like and sheet - like structures , as envisioned by zeldovich s model .
the nonlinear evolution of the tidal shear fields sharpens the anisotropic interconnection of the large - scale structure ( which is often called the cosmic web ) , developing a hierarchical system where small filaments are threaded with the larger parent filaments ( e.g , * ? ? ?
to understand the formation and evolution of the large - scale structure , an important question to answer is how frequently the halos merge into filaments and under what circumstances the halo - to - filament merging event occurs efficiently .
@xcite have for the first time evaluated analytically the merging rates of bound halos by extending the original press - schechter formalism @xcite .
their analytic work was later refined and complimented by numerical calculation of merger trees from n - body simulations @xcite .
recently , @xcite constructed merger trees from the large millennium run halo catalogs @xcite and found that the mean merging rate per halo can be well modeled by a nearly universal fitting formula .
in addition , @xcite showed that the prediction of the extended press - schechter formalism for the halo merging rates disagrees with the numerical results from the millennium run halo catalog by up to a factor of a few .
the halo - to halo merging rates were originally derived under the assumption that the merging events occur in a hierarchical way .
if the matter content of the universe were composed purely of cdm particles , then the structure formation would proceed in a strictly hierarchical way . however , if the non - cdm particles ( either warm or hot ) coexist with the cdm particles , the pure hierarchy in the build - up of the large - scale structure would break down at some level , depending on their mass fraction @xcite .
true as it is that the current observations put a tight constraint on the amount of the possible non - cdm particles , they have not been completely excluded . as a matter of fact , the existence of non - cdm particles have been suggested as one possible solution to alleviate the apparent mismatches between observations and the predictions of @xmath14cdm cosmology on the subgalactic scales ( see * ? ? ?
* ; * ? ? ?
* for a recent discussion ) . among the various candidates for the non - cdm particles ,
the massive neutrinos have so far attained the most serious attentions . ever since it was discovered that at least one flavor of the neutrinos must be massive and that the neutrino masses would be best constrained from the astronomical data ( see * ? ? ?
* for a review ) , a flurry of research has been conducted to quantify the effect of massive neutrinos on the formation and evolution of the large - scale structure ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
? * ; * ? ?
? * ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) . in previous studies of the structure formation in a @xmath14mdm ( @xmath14+cdm+massive neutrinos ) universe ,
what has been largely focused is the suppression of the small - scale powers in the density power spectrum due to massive neutrino s free - streaming and its effect on the structure formation . since the massive neutrinos would decrease the merging rates of the small - scale halos due to their free streaming
, one would naively think that the formation of large scale structures would be deferred in a @xmath14mdm universe .
meanwhile , the recent discovery of high-@xmath13 quasars at @xmath15 @xcite has suggested that the first galaxies of mass @xmath16 must have formed at redshifts @xmath17 . at first glance
the existence of such massive first galaxies at such early epochs might look difficult to reconcile with a @xmath14mdm model . even in a @xmath14cdm cosmology , it is not so easy to explain the existence of the high-@xmath13 quasars since it requires very high merging / accretion rate in the early universe .
yet , what has not been noted in previous works is that the presence of massive neutrinos may actually open a new channel for the rapid structure formation . since the massive neutrinos would behave as effective warm dark matter ( wdm ) particles
, they would fasten the collapse of matter along the major axes of the gravitational tidal fields on large scales .
in other words , while the free - streaming of massive neutrinos slows down the formation of the zero - dimensional bound halos on small scales , it speeds up the formation of one - dimensional filaments on larger scales .
the filaments are the most optimal environments for the rapid accretion of gas and matter . furthermore ,
when the filaments collapse along their longest axes , they would provoke the violent merging of the constituent halos . here
, assuming a @xmath14mdm universe , we study how the first filaments form through the clustering of the first star - forming halos and how they evolve after the longest - axis collapse . the most idealistic approach for the study of the first filaments would be to use high - resolution n - body simulations for a @xmath14mdm universe .
but , it is still a daunting task to implement the dynamics of the massive neutrinos into the current n - body simulations . instead , we choose an analytic approach based on the extended press - schechter formalism @xcite for this study . throughout this paper , we assume flat @xmath14mdm models where the neutrinos have mass and three species @xmath18 .
the key cosmological parameters are set at @xmath19 ( for the @xmath14cdm part ) , @xmath20 and @xmath21 , to be consistent with wmap7 results @xcite .
the outline of this paper is as follows . in 2 , we derive analytically the rates of merging of the first star forming halos into the first filaments and determine their characteristic mass scales . in
3.1 , we evaluate the epochs of the longest - axis collapse of the first filaments and the rates of merging of the first galaxies into the second - generation filaments , and determine their characteristic mass scales . in 4 , we explore the possibility of constraining the neutrino mass from the observable mass distribution of the high-@xmath13 supermassive blackholes . in 5 ,
the results are summarized and a final conclusion is drawn .
recent observations indicate that the first galaxies must have formed in the halos of mass @xmath22 at redshifts @xmath23 ( e.g. , * ? ? ?
* ; * ? ? ?
what has been inferred from the current high - resolution hydrodynamic simulation is that these first galaxies may have formed through rapid merging of the first - star forming minihalos of mass @xmath24 at @xmath25 ( * ? ? ?
* and references therein ) . in this section ,
we investigate how the presence of massive neutrino would change the evolution channel of the first - star forming halos . in the extended press - schechter theory developed by @xcite ,
the key quantity is the conditional probability , @xmath26 , that a halo of mass @xmath27 existing at a given redshift @xmath13 will merge to form a more massive halo of mass @xmath28 at some lower redshift @xmath29 . adopting the spherical collapse condition that a halo of mass @xmath27 forms at @xmath13 when its linear density contrast @xmath30 on the mass scale @xmath27 reaches a redshift - dependent threshold @xmath31 where @xmath32 and @xmath33 is the linear growth factor , @xcite equated this key conditional probability to the differential fraction of the volume in the linear density field occupied by the regions satisfying @xmath34 on the mass scale @xmath35 at @xmath36
provided that it also satisfies the condition of @xmath37 on the mass scale @xmath27 at earlier epoch @xmath13 .
this key conditional probability @xmath38 can now be written as @xmath39^{3/2 } \frac{{\omega}{\prime}({\omega}- { \omega}{\prime})}{{\omega } } \exp\left[-\frac{({\omega}{\prime}s-{\omega}s{\prime})^2}{2ss{\prime}(s - s{\prime})}\right]\,ds{\prime } , \label{eqn : con}\end{aligned}\ ] ] where @xmath40 with @xmath41 , @xmath42 , @xmath43 , and @xmath44 . here
@xmath45 and @xmath46 are the rms fluctuation of the linear density field smoothed on the mass scale @xmath27 and @xmath35 , respectively .
normalization of @xmath47 is done for the @xmath14mdm linear power spectrum to be identical with the @xmath14cdm linear power spectrum at large scale . here
, of course , with the normalized @xmath14cdm linear power spectrum @xmath47 at @xmath48 has the value of @xmath49 given by wmap7 .
using equation ( [ eqn : con ] ) , @xcite calculated the ( instantaneous ) rate of merging of a dark halo with mass @xmath27 at a given redshift @xmath13 into a larger halo with mass @xmath35 by taking the derivative of @xmath50 with respect to @xmath36 : @xmath51^{3/2}\exp\left[-\frac{{\omega}^2}{2 } \left(\frac{1}{s{\prime}}-\frac{1}{s}\right)\right ] , \label{eqn : mrateh}\end{aligned}\ ] ] where @xmath52 and @xmath53 is the linear density power spectrum for a @xmath14mdm cosmology .
we use the analytic approximation given by @xcite for the evaluation of @xmath53 . with the help of the above analytic prescriptions , we determine the effect of massive neutrinos on the merging rate of the first star - forming mini - halos of mass @xmath54 at redshift @xmath55 for the range of the neutrino mass fraction , @xmath56 .
we use the analytic formula given by @xcite for the @xmath14mdm linear density power spectrum @xmath57 where @xmath58 is the neutrino mass density parameter .
the @xmath14mdm linear growth factor @xmath33 is approximated as @xmath59 where @xmath60 is the linear growth factor for a @xmath14cdm universe @xcite .
the range of the neutrino mass fraction is chosen to be @xmath56 where the upper limit of @xmath61 comes from the recent wmap7 constraint @xcite .
it is also worth mentioning here that for this range of @xmath2 the mini - halos of mass @xmath24 at @xmath55 correspond to the @xmath62 peaks of the initial density field .
the rates of the merging of the first star - forming mini - halos are plotted as thin lines in figure [ fig : merg1 ] for the five different cases of @xmath2 , which reveals that the halo - to - halo merging rates decrease rapidly as @xmath2 increases .
this result implies that in a @xmath14mdm universe the free streaming of the massive neutrinos have an effect of deferring the formation of larger halos through the major mergers of the first star - forming mini - halos .
the result of 2 indicate that the direct merging of the first mini - halos into larger halos are slowed down by the free streaming of massive neutrinos .
but , what about the merging of the first star - forming mini - halos into the first mini - filaments ? to derive the rate of merging of the first star - forming mini - halos into the mini - filaments , we adopt the classification scheme proposed by @xcite , according to whom a filament of mass @xmath35 forms at @xmath36 when the two of the three eigenvalues , @xmath63 , of the linear deformation tensor ( defined as the second derivative of the linear peculiar potential ) smoothed on the scale of @xmath35 exceed the threshold value , @xmath64 with @xmath65 .
using the n - body results , they demonstrated that the structures identified by this dynamical classification scheme exhibit indeed filamentary shapes .
their filament - finding scheme is also consistent with the statistical results obtained by @xcite . adopting this condition for the formation of filaments ,
we modify the lacey - cole formalism to determine the halo - to - filament merging rates . in this modified formalism
, the key quantity is the conditional probability , @xmath26 , that a halo of mass @xmath27 observed at @xmath13 will merge into a filament of mass @xmath35 formed at @xmath36 .
we equate this conditional probability to the differential volume fraction occupied by the regions satisfying the filament formation condition , @xmath66 on the mass scale @xmath35 at redshift @xmath36 provided that the linear density contrast of the same regions , @xmath30 , on the smaller mass scale @xmath27 at earlier epoch @xmath13 satisfies the spherical collapse condition @xmath37 . here
@xmath30 equals the sum of the three eigenvalues of the deformation tensor smoothed on the mass scale @xmath27 .
now , this key conditional probability can be expressed as @xmath67 with @xmath68 . to evaluate @xmath69 in equation ( [ eqn : ffh ] )
, it requires us the joint conditional probability distribution of @xmath70 , which has been already derived by @xcite from the original work of @xcite as @xmath71 \exp\left[-\frac{5}{2{{\sigma}{\prime}}^2}({i{\prime}_1}^2 - 3i{\prime}_2)\right ] , \label{eqn : lee06}\end{aligned}\ ] ] where @xmath72,@xmath73 , @xmath74 , @xmath75 and @xmath76 .
it is now straightforward to calculate the ( instantaneous ) rate of merging of the first star forming halos of mass @xmath54 at @xmath55 into the early mini - filaments of mass @xmath35 as @xmath77 figure [ fig : merg1 ] plots the halo - to - filament merging rates ( thick lines ) .
in contrast to the halo - to - halo merging rates , the halo - to - filament merging rates increase with @xmath2 . as @xmath2 changes from @xmath78 to @xmath61 , the maximum halo - to - filament merging rates increase by a factor of @xmath79 .
we call the scale on which the halo - to - filament merging rate reaches the maximum value the characteristic filament mass scale and denote it as @xmath80 . for @xmath4 , the halo - to - filament merging rate begins to exceed @xmath5 at the characteristic mass scale around @xmath81 . whereas for @xmath3 , the filament - to - halo merging rate is below @xmath5 at all filament mass scale .
note that for @xmath82 the mini - halos of mass @xmath83 at @xmath55 tend to merge more rapidly into the mini - filaments of mass @xmath84 rather than into the larger halos of mass @xmath85 .
figure [ fig : mf_fnu ] plots the characteristic mass @xmath80 of the first filaments versus the neutrino mass fraction @xmath2 .
as one can see , the value of @xmath86 decreases almost linearly with @xmath2 .
in the previous section , it is shown that the first star - forming mini - halos would merger rapidly into the first filament of characteristic mass @xmath6 in the presence of massive neutrinos .
as the universe evolves , these first filaments would ultimately collapse along their longest - axes that are in the direction of the minor principal axes of the local tidal field . to determine the distribution of the redshift @xmath87 when
the first filaments collapse along their longest axes , we calculate the conditional probability density that a region satisfying the filament - formation condition ( @xmath88 ) at @xmath55 on the mass scale @xmath80 will meet the halo - formation condition ( @xmath89 ) at the same mass scale . ] but at some lower redshift , @xmath90 : @xmath91 where @xmath92 is the linear density contrast on the mass scale @xmath80 at @xmath87 , @xmath93 are the three eigenvalues of the deformation tensor smoothed on the mass scale of @xmath80 at @xmath55 , @xmath94 and @xmath95 . using the joint probability density distribution of @xmath96 derived by @xcite , it is also straightforward to evaluate equation ( [ eqn : epoch ] )
. figure [ fig : epoch1 ] plots this distribution @xmath97 of the epochs of the longest axis collapse of the first filaments for the five different cases of @xmath2 .
as can be seen , @xmath97 has a maximum value at @xmath98-@xmath99 for all five cases , which implies that the first filaments retain their structures for a long period from @xmath55 to @xmath9-@xmath99 without collapsing into zero - dimensional halos .
as @xmath2 increases , however , the distribution @xmath97 becomes narrower and its peak position moves slightly to the lower value of @xmath87 .
that is , the massive neutrinos play a role of delaying the longest - axis collapse of the first filaments .
now that the halo - to - filament merging rates are found to increase with the neutrino mass fraction , we would like to see how the presence of the massive neutrinos alters the merging rates of the first galaxies .
it is naturally expected that some fraction of the first galaxies would also merge first in the filaments ( called the second - generation filaments ) .
a question is what the characteristic masses of the second generation filaments are and when they would undergo the longest - axis collapse . to answer this question
, we recalculate everything through equations [ eqn : start]-[eqn : epoch ] , but for the first galactic halos of mass @xmath8 at @xmath9 .
figure [ fig : merg2 ] plots the halo - to - halo and halo - to - filament merging rates of the first galactic halos of mass @xmath84 at @xmath9 as thin and thick lines , respectively . a similar trend to that shown in figure [ fig : merg1 ]
is found for the first galactic halos : as the neutrino mass fraction increases , the halo - to - halo merging rate decreases while the halo - to - filament merging rate increases .
the comparison of figs.[fig : merg1 ] and [ fig : merg2 ] also reveals that the halo - to - filament merging rates increases with redshift and with halo mass . for @xmath100 ,
the rate of merging of the first galaxies into the second generation filaments exceeds @xmath5 at all mass scales .
the characteristic mass of the second generation filaments is found to lie in range of @xmath101 .
figure [ fig : epoch2 ] plots the distribution of the redshifts at which the second generation filaments of mass @xmath102 formed at @xmath9 experience the longest axis collapse .
a similar trend to that shown in figure [ fig : epoch1 ] is also found : @xmath97 has a maximum value at @xmath103-@xmath104 for all five cases , which implies that the second - generation filaments last from @xmath9 to @xmath105-@xmath104 without collapsing into zero - dimensional halos . the increase of @xmath2 results in narrowing down the shape of @xmath97 and shifting it slightly to the low-@xmath13 section .
note that these second generation filaments correspond to filaments hosting galactic halos and are expected to be interconnected to the larger - scale parent filaments of mass @xmath106 ( e.g. , * ? ? ?
the massive galaxies formed through the longest - axis collapse of the second generation filaments would grow through the anisotropic accretion of matter and gas along the larger - scale parent filaments in the cosmic web .
once the first filaments form through the merging of the first star - forming halos at @xmath107 , the first stars comprising the first filaments would grow more rapidly as the accretion of matter and gas occur more efficiently along the bridges of the first filaments @xcite .
when the first filaments collapse along their longest axes at @xmath7 , the strong collisions between stars and gas particles in the filaments would result in seeding the supermassive blackholes in the massive first galaxies that are believed to power the ultra - luminous high-@xmath13 quasars @xcite . in this scenario ,
the high - end slope of the mass distribution of the supermassive blackholes inferred from the observable luminosity function of the ultra - luminous high-@xmath13 quasars would reflect the mass distribution of the first filaments .
since the mass distribution of the first filaments depend sensitively on the mass of massive neutrinos , it might be possible to constrain the mass of neutrinos by comparing the mass distributions of the first filaments with that of the high-@xmath13 supermassive blackholes .
the mass distribution function of the first filaments ( @xmath108 ) at the epoch of longest - axis collapse ( @xmath87 ) can be derived by employing the press - schechter approach @xcite : @xmath109 where @xmath110 is the mean mass density . here
, @xmath111 represents the volume fraction occupied by the first filaments of mass larger than @xmath27 at @xmath87 . following the press - schechter approach ,
this volume fraction can be expressed as @xmath112,\ ] ] where @xmath113 is the normalization factor , @xmath114 is the value of the @xmath115 at the epoch of the longest - axis collapse . since the first filaments formed at the threshold of @xmath116 , the value of @xmath115 increases gradually till they collapse eventually along their longest axes .
it has been found that at the epoch of the longest - axis collapse , @xmath115 reaches the critical value of @xmath117 . using the joint distribution , of @xmath118 , derived by @xcite , it is straightforward to evaluate equation ( [ eqn : mf ] ) . figure [ fig : mf ] plots the mass distribution of the first filaments in logarithmic scales at the epoch of the longes - axis collapse for five different cases of the neutrino mass fraction , @xmath2 . as we have already shown in 2 , that the longest - axis collapse of the first filaments occur around @xmath9 , we set @xmath87 at @xmath119 for the evaluation of equation ( [ eqn : mf ] ) .
as it can be seen , in the high - mass section ( @xmath120 ) , the abundance of the first filaments decreases sharply with @xmath27 for all cases .
the rate of the decrease of the mass function of the first filaments , however , depends sensitively on the value of @xmath2 . the larger the value of @xmath2 is , the more rapidly the mass function drops in the high - mass section . to examine quantitatively
how the rate of the decrease of the mass function of the first filaments in the high - mass section changes with @xmath2 , we approximate it as a power law @xmath121 and determine the value of the high - end slope , @xmath122 , in the mass range of @xmath123 .
figure [ fig : slope ] plots the high - end slope , @xmath122 , of the mass distribution of the first filaments versus the neutrino mass fraction , @xmath2 as solid line .
as it can be seen , the absolute value of the high - end slope of the mass function of the first filaments increases sharply as @xmath2 increases .
this result suggests that the value of @xmath2 could be constrained by using the high - end slope of the mass function of the first filaments . in practice , of course , it is not possible to measure directly the high - end slope of the mass function of the first filaments .
as mentioned above , however , we can infer it from the observable high-@xmath13 blackhole mass function , under the assumption that the longest - axis collapse of the first filaments with mass larger than the characteristic mass ( @xmath124 ) would result in the first galaxies hosting the supermassive blackholes that power the ultra - luminous high-@xmath13 quasars .
now , we would like to compare the high - end slope of this theoretically derived mass distribution of the first filaments with that of the observationally obtained mass function of the massive galaxies hosting the high-@xmath13 supermassive blackholes .
very recently , @xcite have derived the blackhole mass function at @xmath125 from the observed luminosity function of the ultra - luminous high-@xmath13 quasars ( see figure 8 in willott et al .
2010 ) . converting the mass of the high-@xmath13 supermassive blackholes into the total mass of their host galaxies through the following relation given by @xcite , @xmath126 ,
we find the best - fit high - end slope is in the range of @xmath127 , which are plotted as two dashed lines in figure [ fig : slope ] . comparing this range of @xmath122 against the theoretical curve plotted in figure [ fig : slope ] , it is found that the neutrino mass fraction is constrained as @xmath128 which corresponds to the neutrino mass range of @xmath129 .
starting with the first star - forming halos of mass @xmath24 at @xmath55 , we have shown analytically that the presence of massive neutrinos opens a new channel for the rapid formation of massive structures by speeding up the halo - to - filament merging process .
once the first filaments of mass @xmath130 form through the clustering of some of the first star - forming halos at @xmath131 , matter and gas would accrete more efficiently along the first filaments , resulting in the fast growth of the constituent halos in the first filaments . as the universe evolves and the tidal forces increases
, the first filaments would undergo the ultimate collapse into the zero - dimensional halos ( corresponding to the first massive galaxies ) in the directions parallel to the minor principal axes of the local tidal field .
what has been found here is that the longest - axis collapses of the first filaments end up forming the first massive galaxies of mass @xmath130 at @xmath17 .
the longest axis collapses of the first filaments would accompany violent mergers among the constituent halos , which in turn would trigger the vigorous star - formation and feed the central blackholes in the resulting first galaxies .
our result suggests that the ultraluminous high-@xmath13 quasars detected at @xmath15 might correspond to the first massive galaxies formed through the longest - axis collapses of the first filaments .
it has been also found that the second - generation filaments of mass @xmath8 form through the clustering of some of the first galaxies of mass @xmath8 at @xmath7 .
the first galaxies in the second generation filaments would evolve fast through the filamentary accretion of matter and gas @xcite .
when the second generation filaments collapse along their longest axes , they would also induce violent mergers among the constituent galaxies , which would lead to the formation of massive galaxies of mass larger than @xmath102 with high star - formation rates and large central blackholes at redshifts as high as @xmath105 .
we test the possibility of constraining the neutrino mass , @xmath132 , from the mass function of the high-@xmath13 supermassive blackholes that power the ultraluminous high-@xmath13 quasars , under the assumption that the longest axis collapses of the first filaments would seed the supermassive blackholes . comparing the high - end slope of the theoretically derived mass function of the first filaments at the epoch of their longest - axis collapse with that of the observed mass function of the massive first galaxies inferred from the luminosity function of the ultra - luminous high-@xmath13 quasars
, we find that the neutrino mass is constrained to be @xmath133 .
yet , it is still only a feasibility study .
a more comprehensive work should be done to put a robust constraint on @xmath132 from observational data .
first of all , more refined theoretical models for the halo - to - filament merging rates and the abundance of the first filaments should be constructed than the crude approximations based on the extended press - schechter formalism , since the formalism has been shown to fail in accurately predicting the merging rates and abundance of bound halos @xcite .
second , the observational results from the luminosity function of the ultra - luminous high-@xmath13 quasars still suffer from small - number statistics and thus more quasar data are required .
third , we use the relation given by @xcite to convert the blackhole mass into the halo mass
. however , the relation of @xcite has been obtained from the blackholes detected at @xmath134 , and thus it may not be applicable to the higher-@xmath13 blackholes . a more accurate relation between the high-@xmath13 blackhole mass and host halo mass should be found .
our results also hints that the massive neutrinos may have something to do with the early metal enrichment of the intergalactic medium ( igm ) .
@xcite showed that the early outflows driven by supernovae ( sn ) ejecta from the galaxies of mass @xmath16 at @xmath135 would enrich the igm with the products of stellar nucleosynthesis provided that those galaxies possess relatively high star - formation efficiencies .
it is interesting to note that the halo mass and epoch required for the sn - driven outflows are more or less coincident with our estimates for the galactic halos formed through the longest - axis collapse of the first filaments in a @xmath14mdm model . assuming that the longest - axis collapse of the first filaments trigger star formation rapidly in the first massive galaxies in a @xmath14mdm model
, the star - formation efficiency of the first mass galaxies may be as high as required for the early metal enrichment by sn - driven outflows .
furthermore , given our speculation that the longest - axis collapse of the first filaments would feed the supermassive blackholes , the anisotropic outflows from the active galactic nuclei ( agn ) powered by the supermassive blackholes of the first massive galaxies might also contribute to the early metal enrichment of the igm ( e.g. , * ? ? ?
* ; * ? ? ?
. it will be intriguing to explore how the metal enrichment of the igm depends on the neutrino mass .
our future work is in this direction .
we thank an anonymous referee for helpful comments .
this work was supported by the national research foundation of korea ( nrf ) grant funded by the korea government ( mest , no.2010 - 0007819 ) .
support for this work was also provided by the national research foundation of korea to the center for galaxy evolution research .
bandara , k. , crampton , d. , & simard , l. 2009 , , 704 , 1135 barai , p. , martel , h. , & germain , j. 2011 , , 727 , 54 boyanovsky , d. , de vega , h. j. , & sanchez , n. g. 2008 , , 77 , 043518 bond , j. r. , efstathiou , g. , & silk , j. 1980 , , 45 , 1980 bond , j. r. , kofman , l. , & pogosyan , d. 1996 , , 380 , 603 bromm , v. , yoshida , n. , hernquist , l. , & mckee , c. f. 2009 , nature , 459 , 49 bromm , v. , & loeb , a. 2003 , , 596 , 34 de vega , h. j. , & sanchez , n. g. 2010 , , 404 , 885 doroshkevich , a. g. 1970 , astrofizika , 6 , 581 doroshkevich , a. g. , khlopov , m. i. , sunyaev , r. a. , szalay , a. s. , & zeldovich , i. b. 1981 , annals of the new york academy of sciences , 375 , 32 dressler , a. 1980 , , 236 , 351 eisenstein , d. j. , & hu , w. 1999 , , 511 , 5 fakhouri , o. , & ma , c .- p . 2008 , , 386 , 577 fan , x. 2003 , bulletin of the american astronomical society , 35 , 1415 fan , x. et al .
, 2001 , , 122 , 2833 fan , x. et al . , 2004 , agn physics with the sloan digital sky survey , 311 , 431 fan , x. et al . , 2006 , , 132 , 117 forero - romero , j. e. , hoffman , y. , gottlber , s. , klypin , a. , & yepes , g. 2009 , , 396 , 1815 fujita , a. , mac low , m .-
m . , ferrara , a. , & meiksin , a. 2004 , , 613 , 159 germain , j. , barai , p. , & martel , h. 2009 , , 704 , 1002 hu , w. 1998 , , 506 , 485 hu , w. , & eisenstein , d. j. 1998 , , 498 , 497 jiang , l. et al . , 2009 , , 138 , 305 kauffmann , g. et al .
, 2004 , , 353 , 713 komatsu , e. , et al .
2011 , , 192 , 18 kuzio de naray , r. , martinez , g. d. , bullock , j. s. , & kaplinghat , m. 2010 , , 710 , l161 lacey , c. , & cole , s. 1993 , , 262 , 627 lacey , c. , & cole , s. 1994 , , 271 , 676 lee , j. 2006 , arxiv : astro - ph/0605697 lee , j. , & shandarin , s. f. 1998 , , 500 , 14 lesgourgues , j. , & pastor , s. 2006 , , 429 , 307 lewis , a. , & challinor , a. 2002 , , 66 , 023531 li , y. , et al .
2007 , , 665 , 187 madau , p. , ferrara , a. , & rees , m. j. 2001 , , 555 , 92 park , d. , & lee , j. 1998 , , 123 , 14 pinsonneault , s. , martel , h. , & pieri , m. m. 2010 , , 725 , 2087 postman , m. & geller , m. j. 1984 , , 281 , 95 press , w. h. , & schechter , p. 1974
, , 187 , 425 saito , s. , takada , m. , & taruya , a. 2009 , , 80 , 083528 seljak , u. , slosar , a. , & mcdonald , p. 2006
, , 10 , 14 shen , y. 2009 , , 704 , 89 shoji , m. , & komatsu , e. 2010 , , 81 , 123516 somerville , r. s. , lemson , g. , kolatti , t. s. , & dekel , a. 2000 , , 316 , 479 song , h. , & lee , j. 2009 , , 703 , 14 springel , v. , et al .
2005 , , 435 , 629 tegmark , m. et al . , 2006 , , 74 , 123507 valdarnini , r. , kahniashvili , t. , & novosyadlyj , b. 1998 , , 336 , 11 whitmore , b. c. , gilmore , d. m. , & jones , c. 1993 , , 407 , 489 willott , c. j. et al .
, 2009 , , 137 , 3541 willott , c. j. et al . , 2010 , , 139 , 906 springel , v. , et al .
2005 , , 435 , 629 yoshida , n. , sokasian , a. , hernquist , l. , & springel , v. 2003 , , 591 , 1 zeldovich , y. b. 1970 , , 5 , 84 into the first filaments ( thick lines ) and larger halos ( thin lines ) at @xmath55 for the five different values of the neutrino mass fraction : @xmath136 and @xmath137 ( solid , dotted , dashed , dot - dashed , and dot - dot - dashed , respectively ) . ] | we study the effect of massive neutrinos on the formation and evolution of the first filaments containing the first star - forming halos of mass @xmath0 at @xmath1 . with the help of the extended press - schechter formalism ,
we evaluate analytically the rates of merging of the first star - forming halos into zero - dimensional larger halos and one - dimensional first filaments .
it is shown that as the neutrino mass fraction @xmath2 increases , the halo - to - filament merging rate increases while the halo - to - halo merging rate decreases sharply . for @xmath3 ,
the halo - to - filament merging rate is negligibly low at all filament mass scales , while for @xmath4 the halo - to - filament merging rate exceeds @xmath5 at the characteristic filament mass scale of @xmath6 .
the distribution of the redshifts at which the first filaments ultimately collapse along their longest axes is derived and found to have a sharp maximum at @xmath7 .
we also investigate the formation and evolution of the second generation filaments which contain the first galaxies of mass @xmath8 at @xmath9 as the parent of the first generation filaments .
a similar trend is found : for @xmath4 the rate of clustering of the first galaxies into the second - generation filaments exceeds @xmath10 at the characteristic mass scale of @xmath11 .
the longest - axis collapse of these second - generation filaments are found to occur at @xmath12 .
the implications of our results on the formation of massive high-@xmath13 galaxies and the early metal enrichment of the intergalactic media by supernova - driven outflows , and possibility of constraining the neutrino mass from the mass distribution of the high-@xmath13 central blackholes are discussed . |
this work was made possible by the computational facilities of dr .
o. rubel and the shared hierarchical academic research computing network ( sharcnet : www.sharcnet.ca ) and compute / calcul canada . financial support of ontario ministry of research and innovation through a research excellence program ontario network for advanced medical imaging detectors is highly acknowledged .
preliminary results : epr spectroscopy performed on undoped pbo samples has revealed the paramagnetic centers in the triplet state at room temperature .
lee p.a . , nagaosa n. and wen x .-
mod . phys . * 17 * 2006 , 78 . | a new route to @xmath0 magnetism is established with help of the first principles methods .
non - magnetic elements in groups 13 and 14 of the periodic table are found to act as the magnetic centers upon embedding in polycrystalline @xmath1-pbo structure .
thus , the local magnetic moment is generated on the impurity site ( 1.0@xmath2 and 2.0@xmath2 for elements in group 13 and 14 , respectively ) due to @xmath3-orbitals partially filled with electrons whose on - site spin ordering is governed by the first hund s rule .
the magnetic interactions between impurities are controlled by occupation of the @xmath3-orbitals such as antiferromagnetic ordering ( afm ) occurs between impurities of 2.0@xmath2 while ferromagnetic ( fm ) between impurities possessing 1.0@xmath2 . in respect to the strength of the magnetic interactions ,
the atomic radius of impurity is found to be a key to tune the wave function tails of localized electrons : with reduction of the atomic radius the on - site stability of the spin polarized state grows while losing in the long - rang order interactions .
however , it has been shown that a suppression of the long - rang order interactions can be compensated by higher impurity concentration that is allowed by shift of the solubility limit to higher magnitude .
= 1.5truecm the research on magnetic semiconductors has intensified in recent years due to requirements imposed by the rapidly developing field of spintronics @xcite .
originally , magnetic semiconductors were created by doping of the conventional semiconductors with magnetic ions whose @xmath4 or @xmath5 orbitals are partially filled @xcite . the enormous attention given to so - called `` diluted magnetic semiconductors '' has been rewarded with the discovery of a mechanism of the intrinsic magnetism in semiconductors - defect - induced magnetism .
initially , in the semiconductors doped with magnetic ions , the intrinsic defects were considered only to mediate the magnetic coupling between localized spins occupying the partially filled @xmath4 or @xmath5 orbitals of ions thus contributing to the collective magnetism effect @xcite . however
, a better understanding of the defect properties has revealed that the defects with their @xmath6 localized spins are able to generate the magnetic phenomenon themself @xcite .
the discovery of defect - induced magnetism dubbed as @xmath0 magnetism , i.e. magnetism which occurs not due to partially filled @xmath4 orbitals , brought new impetus into field of magnetic semiconductors and more importantly in spintronics . in order to pursue the spintronics applications ,
the collective magnetic ordering is required to be established between the magnetic centers provided the spin - polarization energy of the localized state is large enough for the local magnetic moment to appear above the room temperature .
the stability of the local magnetic moment is defined by the impurity wave function localization that unfortunately results in suppression of its tails thus precluding the collective ordering .
therefore , success of @xmath0 magnetism in spintronics is recognized to be defined by the proper combination of defect / host @xcite enabling both components . in practice ,
the weak magnetic interactions between the magnetic centers can be compensated by their high concentration @xcite . however , raising a defect concentration is not always a straightforward solution @xcite . for those defects known to induce magnetism : the vacancy @xcite and substitutional defects @xcite , the low limit to the defect concentration even at the most favorable growth conditions is often applied as defined by their formation energy @xcite .
moreover , defects especially in high concentration are not always mechanically tolerated by the crystal lattice , not to mention that the defect - induced lattice perturbation may lead to unwanted changes in the electronic properties @xcite .
therefore , an idea of intrinsic magnetism requires some efforts to bring it to a level of practical applications .
our recent finding of new route to @xmath0 magnetism offers an elegant solution and , therefore , promises a breakthrough in development of magnetic semiconductor @xcite . instead of crystalline systems ,
the layered materials are proposed to be applied as the semiconductor host . in crystalline solids ,
the vacancies only have been considered to establish @xmath0 magnetism @xcite because the formation energy of other defects is too high to reach the concentrations required for the magnetic percolation to occur . in respect to the layered systems ,
the interstitial defects become a feasible source of unpaired electrons because they are incorporated between layers that significantly lowers their formation energy . in this work
we consider the polycrystalline @xmath1-pbo to be semiconductor host for @xmath0 magnetism .
we found that the pb interstitial defect in @xmath1-pbo induces the local magnetic moment of 2.0 @xmath2 @xcite .
the origin of the local magnetic moment upon bonding of the impurity with the host is unique ; the pb interstitial of pb@xmath7:@xmath8 valence shell utilizes its only pb@xmath7:@xmath9 electrons to be attached to the host ( through pb:@xmath9 electrons as well ) while leaving two unperturbed @xmath10 electrons on the defect site .
the hund s rule dictates the spin alignment of the @xmath10 electrons ( the triplet ground state ) that manifests in on - site magnetic moment of 2.0 @xmath2 and provides the high stability of the spin polarized state defined by the spin - polarization energy @xmath11=0.235 ev @xcite . as a result
, the pb atom gains magnetization upon embedding as the interstitial defect into the @xmath1-pbo crystal lattice ( an appearance of the magnetic moment is verified experimentally @xcite ) . in analogy with magnetic ions
, magnetism occurs due to partially filled orbitals , but here it is due to the @xmath3 orbital .
the unique mechanism of bonding which utilizes only @xmath12 valence electrons allow to extend a choice of host and the impurity to several candidates . because family of the @xmath1-pbo lattice shows the lone pair @xmath12 valence shell electrons as a common feature , in principal any of those systems can be used as the host . in fact , @xmath1-pbo crystal structure is well recognized in superconductivity @xcite : fe - pnistides ( basics are fese , feas ) , cuprate ( basics are cuo , cus ) , and lanthanum compounds ( basics are laf , lao )
. however , among others the @xmath1-pbo compound is seems to be the best candidate due to its wide band gap . for the wide band systems , impurity induces the localized defect states inside the band gap which location may vary with impurity choice .
any chemical elements possessing partially filled @xmath3 valence shell can gain magnetic properties upon embedding into the @xmath1-pbo crystal lattice .
thus , elements of the same valence shell as pb atom , i.e. belonging to group 14th of the periodic table ( see fig . [
fig : fig1](b ) ) , are expected to induce the local magnetic moment 2.0@xmath2 .
following the same principal , the chemical elements of @xmath13 valence shell from group 13 of the periodic table would work as magnetic centers characterized by the local magnetic moment 1.0@xmath2 . since we can expect both , a difference in the atomic radius of impurity and occupation of the @xmath3 orbital , to control the magnetic behavior , focus is on possibility to tune magnetism with different impurities looking for a proper combination impurity / host allowing to reach d@xmath14 ferromagnetism .
if our hypothesis is proven true , this approach can open a wide perspective to design the desired magnetic behavior in the @xmath1-pbo semiconductor by generating a network of the interstitial defects acting as the magnetic centers .
the feasibility to generate such network is considered through the thermodynamics of the defect formation .
( a ) the @xmath1-pbo crystal structure which contains the impurity interstitial atom .
( b ) the list of impurities from group 13 and 14 in periodic table used by us to create the local magnetic moment @xmath15=1.0@xmath2 and @xmath15=2.0@xmath2 , respectively ( the atomic number on the top and the standard atomic weight at the bottom ) .
( c ) the pb self - interstitial : the spin density map is plotted with isovalues of @xmath160.003 e / @xmath17 in xcrysden for the energy range ( @xmath18)@xmath160.15 ev .
it demonstrates alignment of electrons at the impurity site for which defect tails can be traced up to the last oxygen atom shown . ] in our study we applied the generalized gradient approximation ( gga ) with the pbe parametrization @xcite provided by wien2k package for the density functional calculations @xcite ( augmented plane wave + local orbitals approach ) .
the pb:@xmath19 and o:@xmath20 electrons have been treated as the valence electrons ( the energy cutoff was -8 ry ) . the supercell approach ( rkmax=7 ) with sufficiently large supercell of 108-atom size ( 3@xmath213@xmath213 array of the primitive unit cells ) has been used for single impurity calculation while the 190-atom size ( 4@xmath214@xmath213 ) supercell for the interacting defects . for integration of the brillouin - zone , the monkhorst - pack scheme of the 5@xmath215@xmath214 ( or 4@xmath214@xmath212 ) k - mesh was applied . in application to the unpaired electrons , gga often fails to perform the localization of the defect wave function due to an electron self - interaction error @xcite that has been examined here with the hartree - fock ( hf ) approach applied directly to the unpaired electrons .
moreover , it is known that when the band gap size is underestimated by gga , the hole - carrying impurity orbital may appear above the bottom of the conduction band thus inducing the spurious long - range order interactions @xcite . for the lattice parameters optimized with gga ,
the band gap is 1.8 ev ( very close to the experimental value @xcite ) while it found to shrink by 0.22 ev when the experimental lattice parameters are considered @xcite .
such gap deviation originates as a result of the interlayer distance mismatch to occur upon lattice optimization performed with gga @xcite . since in the @xmath1-pbo crystal structure
the band gap size is controlled by the interlayer interactions of the pb:@xmath9 electrons , application of gga to optimization of the lattice parameters through overestimation of the interlayer distance causes the band gap to increase . in order to prevent the `` spurious '' effect ,
calculations of the electronic property are performed for the lattice parameters optimized with gga as it gives the better agreement of the band gap size with the experimental data .
on other hand , to preclude the defect formation energies to be underestimated , the experimentally determined interlayer distance has been used for those calculations .
the formation energies of the interstitials have been evaluated for the vacuum conditions ( details on the formation energy simulations are presented in ref .
@xcite ) .
the band diagram for the @xmath1-pbo crystal structure containing the pb self - interstitial defect : @xmath22 and @xmath23 are the bands formed by the @xmath3-localized electrons pb:@xmath24 , while @xmath25 and @xmath26 are the antibonding orbitals of the im - pb bond @xcite . ] in respect to origin of the local magnetic moment in the @xmath1-pbo compound on site of the pb interstitial @xcite , our study had revealed that the pb interstitial combines the advantages of the vacancies @xcite and magnetic ions @xcite . the high spin - polarization energy is observed for pb interstitial due to the spin ordering of @xmath24 to be governed by the hund s rule as for magnetic ions .
at the same time a hybridization of impurity state with the host lattice results in the extended defect tails promising to induce the long - range order interactions . as shown in fig .
[ fig : fig1](c ) , the defect tails appear in upper and lower layers , they are extended up to seven nearest - neighbors and show the higher spin localization at the oxygen atoms .
although the impurity interacts with the top layer through bonding while with bottom layer only through the hybridization interactions , the defect tails are observed to be more pronounced at the bottom layer .
it occurs because the pb interstitial is tightly sandwiched between layers that results in its strong hybridization with the bottom layer .
the redistribution of the spin density from pb interstitial site to the host lattice explains the on - site stability of the triplet state ( @xmath11=0.235 ev ) to be lowered in comparison to the magnetic impurities of @xmath4 or @xmath5 types @xcite known to exhibit the localized nature of the unpaired electrons .
we expect a hybridization with the host to be a key to tune the magnetic behavior when the atomic radius of the impurity is reduced . in this work
we track an alteration in the electronic properties of the host upon replacement of the pb interstitial with different impurities through a behavior of the impurity associated bands depicted in fig .
[ fig : fig2 ] as @xmath22 , @xmath23 , @xmath25 and @xmath26 .
the @xmath22 and @xmath23 bands are those induced by @xmath3-localized electrons ( the spin - up band @xmath22 is occupied by @xmath3 electrons from the impurity valence shell , while spin - down band @xmath23 is empty ) , the @xmath25 and @xmath26 bands are antibonding orbitals of the impurity - host bond ( im - pb ) .
an appearance of all four bands inside the band gap upon changing the impurity type is shown in fig .
[ fig : fig3 ] ( absence of @xmath25 and @xmath26 bands is referred to the antibonding orbitals outside of the band gap ) .
other important parameters such as impurity atomic radius @xmath27 , length of the impurity - host bond
im - pb , spin - polarization energy @xmath11 and splitting of the @xmath22 and @xmath23 bands ( @xmath28 ) are disclosed in table [ tab : table1 ] .
the common trends on formation of the local magnetic moment as a function of the impurity atomic radius are investigated based on elements from group 14 with @xmath29 valence shell generating the local magnetic moment 2.0@xmath2 .
it was found that reduction in atomic radius of impurity leads to shift of both impurity bands @xmath22 and @xmath23 ( @xmath22 is occupied by two electrons ) towards the valence band ( @xmath30 ) but in the same time the gradual enhancement of the splitting of these bands occurs .
the @xmath25 band is also shifted towards the lower energy due to shortening of the im - pb bond , while the energetic position of the @xmath26 band deviates in narrow energy range .
the enhanced splitting of the @xmath22 and @xmath23 bands ( see ( @xmath28 ) in table [ tab : table1 ] ) indicates a gain in stability of the spin - polarized state and , indeed , a raise in the spin - polarization energy @xmath11 is observed .
we found that for the si and c impurities which electronic interactions with the bottom layer are reduced because of shortening of the im - pb bond , the on - site stability already approaches the magnitudes known for the magnetic ions @xcite .
the effect of the spin - orbit coupling ( + so ) on the defect bands splitting has been investigated with gga+so for the heavier elements ( see fig . [
fig : fig3 ] ) .
it is found that for pb impurity the spin - orbit coupling breaks degeneracy of the @xmath22 and @xmath23 levels resulting in their splitting by 0.52 ev .
although , such large splitting is responsible for reduction of ( @xmath28 ) , the triplet state remains stable that is confirmed by exhibition of the local magnetic moment in experiment at room temperature @xcite .
the spin - orbit effect is less pronounced for the sn impurity for which the spin - orbit splitting is reduced to 0.19 ev . for elements with smaller atomic radius ,
the spin - orbit coupling can be neglected : it is 0.07 ev for ge atom and decreases further down for impurities of smaller radius .
we also have examined an effect of the electron self - interaction error @xcite on magnetic behavior of the pb impurity through application of the hf approach directly to the unpaired electrons .
the spin - polarization energy is found to increase more than twice to @xmath11=0.490 ev as a result of enhancement of the splitting of the @xmath22 and @xmath23 bands to 1.12 ev : the @xmath22 orbital is shifted towards the valence band while @xmath23 towards the conduction band by @xmath31 0.4 ev each .
however , the opposite effect is observed when the experimental lattice parameters are taken into account because a reduction in the interlayer distance results in enhancement of impurity hybridization with the opposite host layer .
when both are applied , the compensation effect is developed causing a reduction of the band splitting to ( @xmath28)=0.68 ev .
this value is very close to that found with gga ( see table [ tab : table1 ] ) that indicates the reliability of gga for this task .
the chemical elements from group 13 in the periodic table are also found to be able to act as the magnetic impurity forming the local magnetic moment 1.0@xmath2 induced by @xmath3 electron occupying the @xmath13 valence shell .
since the @xmath25 and @xmath26 bands appears in the conduction band , they are not presented in fig .
[ fig : fig3 ] .
in contrast to @xmath29 impurities , the @xmath22 orbital occupied by single unpaired electron is found to appear very close to the conduction band .
this causes the significant defect wave function delocalization and , therefore , much weaker splitting of the @xmath22 and @xmath23 bands .
thus , the in atom possesses almost zero splitting .
the expected growth of ( @xmath28 ) is observed for the ga impurity for which splitting reaches 0.17 ev . because of small splitting , electron from @xmath22 leaks to @xmath23 inducing reduction of the local magnetic moment to 0.96@xmath2 .
the spin - orbit coupling effect is weakly pronounced for the ga impurity causing a negligible reduction of ( @xmath28 ) by 0.015 ev .
although further increase in the splitting of the @xmath22 and @xmath23 bands has been expected for the al impurity , the modification of the bonding mechanism has discontinued such trend .
the al atom is attached to the pb atom from the bottom layer instead of top layer shown in fig .
[ fig : fig1 ] ( a ) .
it results in shift of the im - pb bonding orbital from valence band into the band gap .
the splitting of the @xmath22 and @xmath23 bands is found to decrease to 0.13 ev that induces a further reduction of the local magnetic moment to 0.78@xmath2 . for the b atom as the magnetic impurity ,
a suppression of the ( @xmath28 ) is even stronger leading to disappearance of the local magnetic moment .
therefore , for new bonding mechanism , the opposite trend is observed : the spin - polarization energy @xmath11 defining the stability of the local magnetic moment decreases with reduction of the atomic radius . the general conclusion , among chemical elements from group 13 in the periodic table only ga impurity promises some advantages for d@xmath14 magnetism .
.[tab : table1 ] the stability of the spin - polarized state determined within gga calculation as a function of impurity atomic radius @xmath27 @xcite : the spin - polarization energy @xmath11 and the energy splitting of the @xmath22 and @xmath23 bands ( @xmath28 ) .
im - pb is a length of the bond to be formed between impurity and the host . [
cols="^,^,^,^,^",options="header " , ] in respect to formation energy of defect , reduction in the atomic radius of impurity leads to less distortion within the @xmath1-pbo crystal structure : in order to accommodate the pb interstitial the layers of the host move apart while for impurity of smaller radius the interlayer distance is preserved . as a result
, we have observed a reduction of the defect formation energy : 1.23 ev is found for pb interstitial ( for details on calculation of the formation energy see @xcite ) , 0.79 ev for the sn impurity , just above zero for ge , si and ga , and it is becoming negative for c impurity .
the low formation energy promises not only the better mechanical tolerance of the host lattice to defects but also the higher finite defect concentration to be reached .
thus , thermodynamically granted defect concentration for the negative formation energy can be as high as number of the sites available for bonding @xmath31@xmath32 @xmath33 .
the formation energies are found to drop down when impurity is placed on surface of single crystal that opens a way to perform a doping of the nearest - neighbors sites ( for example by almost 1.0 ev for the pb interstitial @xcite ) .
although the impurity of small atomic radius shows a better on - site stability of the local magnetic moment and the low formation energy , but stronger localization of the defect wave function implies shorter defect tails that would influence the long - order interactions .
the collective magnetic ordering may only occur when two impurities are close to each other to establish the magnetic coupling of their localized spins . in this respect
the long range order interactions play the essential role .
the magnetic coupling between impurities has been simulated for the system containing two interstitials of @xmath29 valence shell .
the @xmath34 electrons localized on impurities have been aligned on - site , while their inter - site ordering has been switched from antiferromagnetic to ferromagnetic in order to evaluate @xmath35 .
the @xmath24 state is exactly half filled and , therefore , if the localized electrons of two interacting impurities are ferromagnetically coupled ( the total magnetic moments is 4 @xmath2 ) , the inter - site virtual hopping is not allowed @xcite .
the virtual hopping is supported only for afm coupling ( the total magnetic moment for two interacting impurities is zero ) and because it lowers the total energy , afm becomes the ground state . for two pb interstitials placed on distance 4.0 we found that @xmath36=-0.96 ev ( a negative sign indicates the afm ground state ) while for two interacting c interstitials it is reduced to @xmath37=-0.38 ev .
the electronic interactions and the magnetic coupling between defects exponentially decrease with defect separation .
thus , for two defects placed on a distance 12.5 ( for this calculation the size of the supercell was 4@xmath214@xmath213 ) , the energy difference between afm and fm states is drastically supressed to @xmath36=-0.0056 ev and @xmath37=-0.0023 ev for the pb and c interstitials , respectively .
these data prove that the larger is the atomic radius of impurity , the stronger is the hybridization of the impurity state with the host lattice being responsible for extension of the defect tails . to establish the magnetic percolation , a reduction in the inter - impurity coupling to occur for the impurities of smaller atomic radius
can be compensated by the impurity concentration which can be increased due to a shift of the thermodynamic limit of defect formation to the higher magnitudes .
for example , since the c interstitial possesses the negative formation energy , potentially it can be induced on the nearest - neighbouring sites .
for this case , the theoretical limit of the exchange interaction strength defined by @xmath37=-0.38 ev can be achieved .
although the impurities of @xmath29 valence shell show the interesting physics , but their inter - site afm ordering and appearance of the defect states deep inside the band gap ( the coupling of the impurity state to the band like state is essential to support the spin - polarized carrier transport ) make them inappropriate for spintronics applications .
the impurities from group 13 have a better fit to the requirements imposed by spintronics : the @xmath22 and @xmath23 bands both couples to the conduction band and the fm ground state should be granted for interacting impurities . since the ga impurity has shown the highest potential due to its high spin - polarization energy , here we focus on development of the magnetic interactions between two ga impurities .
we found that for two defects placed on a distance 4.0 , the strength of the magnetic coupling is defined by @xmath38=1.08 ev ( a positive sign indicates the fm ground state ) .
the distance between impurities has been increased twice that corresponds to the realistic defect concentration @xmath39=2.5@xmath40 . to simulate the worse - case scenario ,
the impurities have been attached to the opposite layers . in this case , the overlap of the defects tails is weakest as the defect tails for small atomic radius impurity are stronger pronounced in the layer the impurity is attached to .
secondly , instead of straight line location , the impurities have been placed obliquely that also reduces the interaction of their defect tails ( oblique line involves the lead atoms while the defect tails are stronger on the oxygen atoms to be on the straight line shown in fig .
[ fig : fig1](c ) ) . for such impurity location ,
the exchange interaction strength are accounted by @xmath38=0.04 ev ( a distance between impurities was 8.66 ) .
in order to roughly estimate the curie temperature @xmath41 , the simplified mean - field approximation for the heisenberg model can be applied as @xmath42 @xcite ( @xmath43 is the boltzmann constant ) .
the curie temperature is found to be just above 300 k for 2.5@xmath40 of impurity but this value is rather underestimated as the worse - case scenario on impurity location has been applied . in summary
, we propose to generate the local magnetic moment in compounds of @xmath1-pbo crystal structure by its doping with non - magnetic impurities belonging to group 13th or 14th of the periodic table . in analogy with magnetic ions ,
the magnetic moment origin is due to partially filled orbitals , but instead of @xmath4 or @xmath5 types it is being of @xmath3-type .
the partial occupation of @xmath3 orbital appears as a result of the unique bonding of impurities with the host lattice : the original partial occupation of @xmath3 orbital of impurity is preserved after its bonding to the host .
we found that the magnetic behavior of the dopants depends on their atomic radius : dopants with the smaller radius are found to establish the higher on - site stability of the localized spins . for the si and c impurities , the on - site stability reaches such a high magnitude that it becomes comparable with that for the magnetic ions of @xmath4 or @xmath5 types @xcite .
another benefit of the small atomic radius of impurity is its near zero defect formation energy that shifts the solubility limit to magnitudes as high as @xmath31@xmath32 @xmath33 . for the spintronics applications ,
the chemical elements from group 13 of the periodic table are found to induce @xmath44 ferromagnetism . among all ,
in particular the ga impurity shows the characteristics required to establish the magnetic percolation above the room temperature .
overall , it has been shown that the @xmath1-pbo crystal structure is a good candidate to become the semiconductor host for @xmath0 magnetism as it offers a flexibility to tune magnetism by the changing the impurity type . |
entropy is applied in information theory and statistics for characterizing the diversity or uncertainty in a probability distribution . for a continuous distribution @xmath3 with density @xmath4 ,
the rnyi entropy is defined ( rnyi , 1970 ) as @xmath5henceforth we use @xmath6 to denote the natural logarithm of @xmath7 .
the rnyi entropy @xmath8 is a generalization of the shannon entropy ( shannon , 1948 ) : @xmath9 from the statistical point of view , the quadratic rnyi entropy @xmath10 is the simplest point on the rnyi spectrum @xmath11 , where @xmath12 is subset of @xmath13 , such that entropies exist .
note that @xmath14 , where we assume that the quadratic functional @xmath15 is well defined , and hence the point @xmath16 belongs to the set @xmath17 .
more entropy generalizations are known in information theory , e.g. , the tsallis entropy ( tsallis , 1988 ) : @xmath18 the rnyi entropy ( or information ) for stationary processes can be understood as that of the corresponding ergodic or marginal distributions , see , e.g. , gregorio and iacus ( 2003 ) , where the rnyi entropy is computed for a large class of ergodic diffusion processes .
numerous applications of the rnyi entropy in information theoretic learning , statistics ( e.g. , classification , distribution identification problems , statistical inference ) , computer science ( e.g. , average case analysis for random databases , pattern recognition , image matching ) , and econometrics are discussed , e.g. , in principe ( 2010 ) , kapur ( 1989 ) , kapur and kesavan ( 1992 ) , pardo ( 2006 ) , escolano et al . ( 2009 ) , neemuchwala et al .
( 2005 ) , ullah ( 1996 ) , baryshnikov et al .
( 2009 ) , seleznjev and thalheim ( 2003 , 2010 ) , thalheim ( 2000 ) , leonenko et al .
( 2008 ) , and leonenko and seleznjev ( 2010 ) . various estimators for the quadratic functional @xmath19 and the entropy @xmath20 for _ independent _ samples have been studied .
leonenko et al .
( 2008 ) obtain consistency of nearest - neighbor estimators for @xmath21 , see also penrose and yukich ( 2011 ) and the references therein .
bickel and ritov ( 1988 ) and gin and nickl ( 2008 ) show rate optimality , efficiency , and asymptotic normality of kernel - based estimators for @xmath19 in the one - dimensional case .
laurent ( 1996 ) builds an efficient and asymptotically normal estimator of @xmath19 ( and more general functionals ) for multidimensional distributions using orthogonal projection .
see also references in these papers for more studies under the independence assumption . in our paper
, we study @xmath1-statistic estimators for @xmath22 and @xmath21 based on the number of @xmath2-close vector observations ( or the number of small inter - point distances ) in a sample from a stationary @xmath0-dependent sequence with marginal distribution @xmath23 .
this extends further the results and approach in leonenko and seleznjev ( 2010 ) ( see also kllberg and seleznjev , 2012 ) , where the same estimators are studied under independence .
the number of small inter - point distances in an independent sample exhibits rich asymptotic behaviors , including , e.g. , poisson limits and asymptotic normality ( see jammalammadaka and janson , 1986 , and references therein ) .
we show that some of the established limit results for this statistic are still valid when the sample is from a stationary @xmath0-dependent sequence .
it should be noted that our normal limit theorems do not follow from the general theory developed for degenerate variable @xmath1-statistics under dependence , see , e.g. , kim et al . , 2011 , and references therein .
note that the class of stationary @xmath0-dependent processes is quite large , see , e.g. , the book of joe ( 1997 ) , where there are numerous copula constructions for @xmath0-dependent sequences with given marginal distribution , or harrelson and houdre ( 2003 ) , where the class of stationary @xmath0-dependent infinitely divisible sequences is studied .
first we introduce some notation . throughout this paper , it is assumed that the sequence @xmath24 of random @xmath25-vectors is strictly stationary and @xmath0-dependent , i.e. , @xmath26 and @xmath27 are independent sets of vectors when @xmath28 .
let @xmath3 be the ( marginal ) distribution of @xmath29 with density @xmath30 , and entropy @xmath31 .
we write @xmath32 for the euclidean distance in @xmath33 and define @xmath34 to be an @xmath2-ball in @xmath33 with center at @xmath7 and radius @xmath2 .
denote by @xmath35 , the volume of the @xmath2-ball .
let @xmath36 and @xmath37 be independent and with distribution @xmath3 and introduce the @xmath2-ball probability as @xmath38two vectors @xmath7 and @xmath39 are said to be @xmath2__-close _ _ if @xmath40 , for some @xmath41 .
the @xmath2__-coincidence _ _ probability for independent vectors is written @xmath42 .
then the rnyi @xmath2-entropy @xmath43 can be used as a measure of uncertainty in @xmath3 ( see seleznjev and thalheim , 2008 , leonenko and seleznjev , 2010 ) . in what follows , let @xmath44 .
denote by @xmath45 the cardinality of the finite set @xmath46 and let @xmath47 be the random number of @xmath2-close observations in the sample @xmath48 , @xmath49where @xmath50 is the indicator of an event @xmath51 .
then @xmath52 is a @xmath1-statistic of hoeffding with varying kernel .
for a short introduction to @xmath1-statistics techniques , see , e.g. , serfling ( 2002 ) , koroljuk and borovskich ( 1994 ) , lee ( 1990 ) .
denote by @xmath53 and @xmath54 convergence in distribution and in probability , respectively . for a sequence of random variables
@xmath55 , we write @xmath56 as @xmath57 if for any @xmath58 and large enough @xmath59 , there exists @xmath60 such that @xmath61 . moreover , for a numerical sequence @xmath62 , let @xmath63 as @xmath57 if @xmath64 as @xmath57 . the developed technique can also be used for estimation of the corresponding entropy - type characteristics for discrete distributions ( see , e.g. , leonenko and seleznjev , 2010 ) and stationary @xmath0-dependent sequences . in this case , the applied estimator is a @xmath1-statistic with fixed kernel and so the problem is simplified in the way that some already established general results yield the limit properties , including consistency and asymptotic normality ( see appendix ) .
an approach to statistical estimation of the shannon entropy for discrete stationary @xmath0-dependent sequences can be found in vatutin and mikhailov ( 1995 ) .
the remaining part of the paper is organized as follows . in section
[ sec : main ] , the main results for the number of small inter - point distances @xmath65 and the estimators of @xmath22 and @xmath21 are presented .
numerical experiments illustrate the rate of convergence in the obtained asymptotic results . in section [ sec : app ] , we discuss applications of these results to @xmath2-keys in time series databases and distribution identification problems for dependent samples . section [ sec : proofs ] contains the proofs of the statements in section [ sec : main ]
. some asymptotic properties of entropy estimation for the discrete case are given in appendix .
we formulate the following assumption about finite dimensional distributions of the stationary sequence @xmath66 . 1
. the marginal density fulfills @xmath67 .
moreover , for each @xmath68-tuple of distinct positive integers @xmath69 , the distribution of the random vector @xmath70 has a density @xmath71 in @xmath72 that satisfies @xmath73 * remark 1 . *
( i ) the integrability ensures that the dependence among @xmath74 is weak enough .
in fact , provided @xmath75 , it holds for an independent sequence , so assumption @xmath76 is a generalization of the condition @xmath67 used for studying the same estimators under independence ( kllberg and seleznjev , 2012 ) .
\(ii ) if the density @xmath77 is bounded for each distinct @xmath78 , the following condition is sufficient for @xmath76 : for each distinct pair @xmath79 , let the density @xmath80 of @xmath81 satisfy @xmath82 let in the following examples @xmath83 be a sequence of independent identically distributed ( i.i.d . )
normal @xmath84-random variables .
* example 1 . *
( i ) assumption @xmath76 holds for all vector gaussian sequences @xmath85 .
in particular , it is satisfied for the @xmath0-dependent moving average time series ma@xmath86 generated by @xmath87 , i.e. , @xmath88 \(ii ) an exponential transformation of time series gives a non - linear sequence @xmath89 the finite dimensional distributions of @xmath66 are the multivariate log - normal distributions ( see , e.g. , kotz et al . , 2000 ) and thus @xmath76 is fulfilled in this case .
\(iii ) let @xmath90 .
then @xmath66 is a stationary @xmath91-dependent sequence , say , a cauchy sequence .
it can be shown that for @xmath92 , @xmath93 that is is valid in this case and similarly for other @xmath94 . since also @xmath95 , condition @xmath76 is satisfied . let the expectation and variance of the number of small inter - point distances @xmath65 be @xmath96 and @xmath97 , respectively . for @xmath98
, we introduce the characteristic @xmath99 .
let @xmath100 [ prop : nn1 ] suppose that @xmath76 holds .
* then the expectation and variance of @xmath65 fulfill @xmath101 * if @xmath102 , and @xmath103 when @xmath104 , then @xmath105 the asymptotic distribution for @xmath65 depends on the rate of decrease of @xmath106 . some results for @xmath107 under the i.i.d .
assumption ( i.e. , @xmath108 ) are obtained in jammalamadaka and janson ( 1986 ) ( see also leonenko and seleznjev , 2010 ) . with only additional weak conditions ,
we show that these results are still valid when @xmath66 is stationary and @xmath0-dependent .
let @xmath109 for @xmath110 .
[ th : nn2 ] suppose that @xmath76 holds . * if @xmath111 , then @xmath112 . * if @xmath113 , @xmath114 , then @xmath115 and @xmath116 * if @xmath117 and @xmath118 , @xmath119 , and @xmath120 when @xmath121 , then @xmath122 note that definition implies @xmath123 with equality , e.g. , if @xmath23 is uniform .
* remark 2 .
* the following inference procedure is discussed in leonenko and seleznjev ( 2010 ) for i.i.d . sequences .
let @xmath124 . by applying theorem [ th : nn2]_(ii ) _ to the minimum inter - point distance @xmath125 and @xmath126 for a fixed @xmath127 , i.e. ,
@xmath128 , we get @xmath129 hence @xmath130 has asymptotically exponential distribution @xmath131 and an asymptotic confidence interval for the quadratic functional @xmath19 can be written @xmath132\ ] ] for certain positive @xmath133 .
we consider an estimator for the quadratic functional @xmath19 based on the normalized statistic @xmath134 , defined as @xmath135 let @xmath136 be the corresponding estimator for the entropy @xmath20 .
the asymptotic behavior of @xmath137 and @xmath138 depends on the rate of decreasing for @xmath106 . in the following theorem for consistency , we give two versions for different asymptotic rates of @xmath106 , with significantly weaker distribution assumptions in _
( ii)_. [ th : p ] * if @xmath76 holds and @xmath117 , then @xmath139 * let @xmath67 and assume that @xmath140 , for all @xmath141 .
if @xmath142 , then @xmath139 let @xmath143 and @xmath144 .
next we show asymptotic normality properties for the estimators @xmath137 and @xmath145 when @xmath146 and @xmath2 vary accordingly .
let @xmath147 and recall definition @xmath148 of @xmath149 .
[ th : prelim ] suppose that @xmath76 holds and @xmath150 . *
if @xmath151 , and @xmath120 when @xmath121 , then @xmath152 * if @xmath153 , then @xmath154 to evaluate the quadratic functional @xmath19 and the entropy @xmath20 , we introduce smoothness conditions for the marginal density @xmath155 .
denote by @xmath156 , a linear space of functions in @xmath157 satisfying a @xmath158-hlder condition in @xmath159-norm with constant @xmath160 , i.e. , if @xmath161 and @xmath162 , then @xmath163 note that holds , e.g. , if for some function @xmath164 , @xmath165 there are different ways to define the density smoothness , e.g. , by the conventional or pointwise hlder conditions ( leonenko and seleznjev , 2010 , kllberg et al . , 2012 ) or the fourier characterization ( gin and nickl , 2008 ) . the rate of convergence in probability can now be described in terms of the smoothness of @xmath155 .
let @xmath166 , be a slowly varying function as @xmath57 .
[ th : cons ] let @xmath76 hold and assume that @xmath167 . * then the bias @xmath168 . * if @xmath169 and @xmath170 , then @xmath171 * if @xmath172 and @xmath173 and @xmath174 , then @xmath175 * remark 3 . *
since the aim of this paper is to provide asymptotic properties for estimation under dependence , we leave questions regarding efficiency and optimality of the obtained convergence rates for further research .
nevertheless , it could be mentioned that , in the independent one - dimensional case ( @xmath176 ) , bickel and ritov ( 1988 ) show that the rates in theorem [ th : cons ] are optimal in a certain sense ( see also laurent , 1996 , gin and nickl , 2008 ) . in order to make the normal limit results of theorem [ th : prelim ] practical , e.g. , to calculate approximate confidence intervals ,
the asymptotic variances have to be estimated .
in particular , we need a consistent estimate of the characteristic @xmath149 . assuming that @xmath0 is exactly known might be too strong in our non - parametric setting
. however , note that @xmath177 for @xmath178 , so under the less restrictive assumption that a bound @xmath179 for @xmath0 is known , we can use a consistent estimator of @xmath180 . to construct this estimator , for @xmath181 , consider the following estimator of @xmath182 , @xmath183 where @xmath184 and the number of summands @xmath185 .
let @xmath186 .
[ std ] if @xmath76 holds and @xmath187 , then @xmath188 * remark 4 .
* under the conditions of proposition [ std ] , we have @xmath189 and thus @xmath190 is a consistent estimator of the _ cubic _ rnyi entropy @xmath191 . a consistent plug - in estimator for @xmath180 can be set up according to @xmath192 where it is assumed that the sequence @xmath193 satisfies @xmath194 .
now we construct asymptotically pivotal quantities by using theorem [ th : prelim ] , the smoothness of the marginal density , and variance estimators .
to achieve @xmath195-rate of convergence , an upper bound @xmath178 has to be available .
let @xmath196 be the corresponding consistent estimator of @xmath197 when @xmath151 .
[ th : norm1 ] let @xmath76 hold and assume that @xmath198 , and @xmath199 . if @xmath173 and @xmath151 , and @xmath120 when @xmath121 , then @xmath200 next we apply theorem [ th : prelim]_(ii ) _ to weaken the condition @xmath201 and therefore get asymptotic normality for less smooth cases .
additionaly , asymptotically pivotal quantities can be built even without a bound @xmath179 available for @xmath0 .
note , however , that the obtained rate of convergence is slower than @xmath195 .
define a consistent estimator for @xmath202 as @xmath203 .
let @xmath204 .
[ th : norm2 ] let @xmath76 hold and assume that @xmath205 and @xmath117 . * if @xmath206 , for some @xmath207 , and @xmath208 , then @xmath209 * if @xmath210 , then @xmath211 the practical applicability of the results in this paper relies on an accurate choice of the parameter @xmath2 .
one possibility is to use the cross - validation techniques for choosing the optimal bandwidth for density estimation , see , e.g. , hart and vieu ( 1990 ) .
however , the problem of finding a suitable @xmath2 is a topic for future research . in the following examples ,
let @xmath87 be a sequence of i.i.d .
@xmath84-variables .
* example 2 .
* we consider estimation of the quadratic rnyi entropy @xmath31 for the 2-dependent moving average ma@xmath212 process @xmath213 , where @xmath214 . in this case @xmath215 and @xmath216 .
we simulate @xmath217 independent and normalized residuals @xmath218 , @xmath219 , with @xmath220 , @xmath221 , and @xmath222 .
the histogram and normal quantile plot in figure [ fig1 ] illustrate the performance of the normal approximation for @xmath223 implied by theorem [ th : norm1 ] .
the p - value ( 0.60 ) for the kolmogorov - smirnov test also supports the hypothesis of standard normality for the residuals .
-dependent ma@xmath212 time series ; sample size @xmath220 , @xmath221 , and @xmath224 .
standard normal approximation for the normalized residuals ; @xmath225.,scaledwidth=85.0% ] * example 3 .
* consider the sequence @xmath226 , where @xmath227 , i.e. , @xmath66 is a 1-dependent log - normal sequence , @xmath228 . in this case
the quadratic entropy @xmath229 .
figure [ fig3 ] shows the accuracy of the normal approximation in theorem [ th : norm1 ] for the residuals @xmath230 , @xmath231 , where @xmath220 , @xmath232 , and @xmath233 .
the histogram , normal quantile plot , and p - value ( 0.36 ) of the kolmogorov - smirnov test imply that the normality hypothesis can not be rejected .
-dependent log - normal sequence ; sample size @xmath220 , @xmath232 , and @xmath234 .
standard normal approximation for the normalized residuals ; @xmath217.,scaledwidth=85.0% ] * example 4 .
* estimation of the quadratic functional @xmath235 for the 1-dependent cauchy sequence @xmath236 .
here @xmath3 is the cauchy distribution with @xmath237 .
we simulate residuals @xmath238 , @xmath231 , where @xmath220 , @xmath239 , and @xmath217 .
figure [ fig2 ] illustrates the performance of the normal approximation of @xmath240 indicated by theorem [ th : norm2 ] .
the histogram , normal quantile plot , and p - value ( 0.47 ) for the kolmogorov - smirnov test allow to accept the hypothesis of standard normality .
-dependent cauchy sequence ; sample size @xmath220 and @xmath241 .
standard normal approximation for the normalized residuals ; @xmath217.,scaledwidth=85.0% ]
@xmath2-keys in time series databases _
let a time series database @xmath242 be a matrix with @xmath146 random records ( or tuples ) @xmath243 , and @xmath244 attributes , @xmath245 , with continuous tuple distribution with density @xmath246 .
as contrast to conventional static databases , the ordering of records in @xmath242 is significant , i.e. , the timestamp @xmath247 can be associated with an additional attribute for @xmath248 .
for example , time series databases are used for modelling stock market , environmental ( e.g. , weather ) or web usage data ( see , e.g. , last et al . , 2001 ) .
then the database @xmath242 can be considered as a sample from a vector time series @xmath249 .
assume additionally that @xmath250 is a stationary @xmath0-dependent time series .
a subset @xmath251 , @xmath252 , is called an @xmath2-_key _ if @xmath253 , i.e. , there are no @xmath2-close in attributes @xmath254 sub - records @xmath255 .
the distribution of @xmath256 characterizes the capability of @xmath254 to distinguish records in @xmath242 and can be used to measure the complexity of a database design for further optimization , e.g. , for optimal @xmath2-key selection or searching dependencies between attributes ( or _ association rules _ ) ( see , e.g. , thalheim , 2000 , seleznjev and thalheim , 2008 , leonenko and seleznjev , 2010 ) .
now theorem [ th : nn2]_(ii ) _ gives an approximation of the probability that @xmath254 is an @xmath2-key , @xmath257 , where @xmath258 i.e. , asymptotically optimal @xmath2-key candidates are amongst @xmath259 , sets with minimal value of the quadratic functional @xmath19 and the corresponding estimators of @xmath19 are applicable with various asymptotics for @xmath2 and @xmath146 ( remark 2 and theorems [ th : p ] , [ th : cons ] , th : norm1 , and [ th : norm2 ] ) . + _ entropy maximizing distributions for stationary _
@xmath0-_dependent sequences _ note that conditions of consistency for our estimate of the quadratic rnyi entropy @xmath260 ( see theorem [ th : p]_(ii ) _ ) are rather weak and can be easily verified for many statistical models .
hence , one can use these consistent estimators to build goodness of fit tests based on the maximum entropy principle , see , e.g. , goria et al .
( 2005 ) and leonenko and seleznjev ( 2010 ) , where similar approaches were proposed for shannon and rnyi entropies , respectively .
let us remind some known facts about the maximum entropy principle , see , e.g. , johnson and vignat ( 2007 ) . consider the following maximization problem : given a symmetric positive definite matrix @xmath261 , for all densities @xmath155 with mean @xmath262 and such that@xmath263the quadratic rnyi entropy @xmath260 is uniquely maximized by the distribution @xmath264 with density @xmath265that is for all other densities support @xmath266 , mean @xmath267 , and covariance matrix @xmath268 , see ( [ me0 ] ) , we have @xmath269 , with equality if and only if @xmath270 almost everywhere with respect to the lebesgue measure in @xmath33 .
the distribution @xmath271 belongs to the class of multivariate pearson type ii distributions ( or _ student-@xmath179 distributions _ ) and its quadratic rnyi entropy@xmath272 for an i.i.d .
sample , the goodness of fit test based on the maximum quadratic rnyi entropy principle was proposed by leonenko and seleznjev ( 2010 ) . to generalize this test for @xmath0-dependent data ,
we need to show that there exists a stationary @xmath0-dependent sequence with marginal distribution ( me1 ) . for one - dimensional processes
, one can apply some results from joe ( 1997 ) .
henceforth , we use some definitions and notation from this book .
it is known that for continuous multivariate distributions , the univariate marginals and the multivariate or dependent structures can be separated by copula .
let @xmath273 be a bivariate copula with conditional distribution @xmath274 the inverse conditional distribution is denoted by @xmath275 let @xmath276 be a continuous univariate distribution function and let @xmath277 be a sequence i.i.d .
uniformly distributed @xmath278-random variables .
a @xmath91-dependent sequence with marginal distribution @xmath276 is @xmath279 , where @xmath280 $ ] .
the marginal distribution of @xmath281 is @xmath282 , see joe ( 1997 ) , p. 253 , and the joint distribution of @xmath283 is of the form@xmath284 in the same spirit , one can construct a stationary @xmath0-dependent sequence with given marginal distribution for any @xmath285 , see again joe ( 1997 ) , p. 255 .
in particular , for any @xmath286 there exists a one - dimensional stationary @xmath0-dependent sequence with marginal distribution @xmath276 , which has density ( [ me1 ] ) for @xmath287 . for @xmath288 ,
the above copula construction seems difficult , but one can use again the results from johnson and vignat ( 2007 ) , reformulated for quadratic entropy as follows : if @xmath289 is a sequence of i.i.d .
@xmath84-variables , then @xmath290has density ( [ me1 ] ) , where @xmath291 . now a stationary @xmath0-dependent sequence with the marginal density ( [ me1 ] ) can be defined for all @xmath292 , e.g. , @xmath293 .
consequently , for any fixed @xmath25 and @xmath294 , there exists a stationary @xmath0-dependent vector sequence with marginal maximum quadratic entropy distribution ( [ me1 ] ) .
next , from ( [ me2 ] ) we get @xmath295 let @xmath296 be a class of @xmath25-dimensional density functions @xmath297 , with support @xmath298 , which satisfy condition ( [ me0 ] ) .
note that the density @xmath299 belongs to this class .
let @xmath300 , be an @xmath0-dependent sample from a member of @xmath296 .
consider @xmath301 , the sample covariance matrix , as a consistent estimate of @xmath302 as @xmath303 , where the sample mean @xmath304 .
the consistency of the sample covariance matrix is a simple consequence of the asymptotic properties of stationary @xmath0-dependent processes , see , e.g. , lee , ( 1990 ) , anderson , ( 1994 ) .
then under the null hypothesis @xmath305 is a sample from _ _ _ _ @xmath0-dependent data with density @xmath299 , we obtain from the slutsky theorem and theorem th : p_(ii ) _ , that @xmath306where @xmath307 is defined in ( [ me3 ] ) , and @xmath138 is the consistent estimator of the quadratic rnyi entropy . under the alternative
@xmath308 is a sample from any other member @xmath309 of _ _ @xmath296 _ _ , we find that@xmath310 in other words , the above mentioned test is consistent against such alternatives .
note that @xmath311 .
the following lemmas are used in the subsequent proofs . [ lemma : lee ] let @xmath312 .
then the number of k - tuples of integers @xmath313 that satisfy @xmath314 for @xmath315 is @xmath316 .
[ bill95 ] if , for each k , @xmath317 , if @xmath318 as @xmath319 , and if @xmath320 for positive @xmath321 , then @xmath322 . [ lemma ]
* assume that @xmath67 and let @xmath323 .
then , for @xmath324 , @xmath325 and thus @xmath326 .
* if @xmath76 is satisfied , then @xmath327 * if @xmath76 is satisfied , then @xmath328 _ proof .
( i ) _ we use the following result from lemma 1 in kllberg and seleznjev ( 2012 ) : for random vectors @xmath36 and @xmath37 with densities @xmath329 , @xmath330 , @xmath331 note that immediately implies @xmath332 . furthermore ,
if @xmath37 is defined to have density @xmath333 , we obtain from that @xmath334 now consider the decomposition @xmath335 by the stationarity of @xmath66 and hlder s inequality , the last two terms in are in absolute value bounded by @xmath336 which by and tends to zero as @xmath337 .
hence , @xmath338 and the assertion follows . _
( ii ) _ first note that the indices @xmath339 and @xmath340 can be chosen in such a way that @xmath341 and @xmath342 are independent and also independent of @xmath343 . for the corresponding density of @xmath344 , we have @xmath345 , where @xmath346 is the density of @xmath81 . consequently , from assumption @xmath76 , @xmath347 integrating @xmath348 with respect to @xmath349
gives @xmath350 therefore , by hlder s inequality , @xmath351 where @xmath352 since @xmath353 and @xmath354 as @xmath337 , the dominated convergence theorem yields @xmath355 as @xmath337 , and hence from we obtain @xmath356 for each distinct pair @xmath357 .
finally , the stationarity and @xmath0-dependence of the sequence @xmath66 imply that @xmath358 attains a finite number of values as @xmath359 and @xmath360 vary .
thus , the rate of convergence in remains valid if we take the supremum over distinct @xmath357 as desired , so the statement follows . _
( iii ) _ since the argument is similar to that of _ ( ii ) _ , we show the main steps only .
first assume that @xmath361 and @xmath362 .
under @xmath76 , it can be shown in a similar way as above that for each @xmath363-tuple of distinct positive integers @xmath364 , the density @xmath365 of @xmath366 satisfies @xmath367 now introduce the random vectors @xmath368 and @xmath369 in @xmath370 and note that @xmath371 where @xmath372 is the euclidean norm in @xmath370 . define the coordinate vectors @xmath373 and @xmath374 . by hlder s inequality , @xmath375 where @xmath376 by the dominated convergence theorem , @xmath377 as @xmath337 and therefore , since @xmath378 , we get from and that @xmath379 where @xmath380 and @xmath381 . by a similar argument , this also valid when @xmath382 and @xmath383 .
finally , from the stationarity and @xmath0-dependence of @xmath66 , the rate of convergence in still holds if we take the supremum over distinct pairs of pairs .
this completes the proof .
@xmath384 _ proof of proposition [ prop : nn1 ] .
( i ) _ define the index set @xmath385 and the reduced form of @xmath65 : @xmath386 since @xmath387 when @xmath388 , lemmas [ lemma : lee ] and [ lemma ] yield the claim for the expectation : @xmath389 for the variance of @xmath65 , we first study the variance of @xmath390 . to this end , we need to consider @xmath391 terms of the form @xmath392 where @xmath393 . to count the various types of such terms , we use results from ch .
2.4.1 , theorem 1 , in lee ( 1990 ) .
it should be noted that these results are stated for @xmath1-statistics based on random variables , rather than random vectors , but the argument is merely combinatorial and thus also valid here . 1 .
when @xmath394 , the random variables in @xmath395 are independent and the covariance is zero . 2 .
only two random variables are involved , i.e. , @xmath396 and @xmath397 , so the covariance is @xmath398 as @xmath337 .
the number of such terms is @xmath399 .
exactly one of the four possible differences @xmath400 is zero and the rest are greater than @xmath0 . by conditioning , the covariance is @xmath401 .
there are @xmath402 terms of this type .
4 . for @xmath403 ,
@xmath404 for one of the differences @xmath400 and the others are greater than @xmath0
. then the covariance is @xmath405 .
the number of terms of this type is @xmath406 .
the number of the remaining terms is @xmath407 , so lemma [ lemma]_(iii ) _ indicates that their sum is @xmath408 . from the above and lemma [ lemma ] ,
we obtain @xmath409 next it is verified that @xmath410 and @xmath411 are close .
note that @xmath412 where @xmath413 .
we have @xmath414 where the number of pairs @xmath415 in the last sum for which @xmath416 and @xmath417 are _ not _ independent is @xmath418 . from lemma
[ lemma ] , we get @xmath419 , and consequently leads to @xmath420 furthermore , observe that @xmath421 and hence gives @xmath422 the next step is to show that @xmath423 to this end , first note that lemma [ lemma]_(ii ) _ implies @xmath424 since the number of terms in this sum is @xmath418 , so we only need to prove that @xmath425 we have @xmath426 there are @xmath407 terms in this double sum .
moreover , @xmath418 of these have the form @xmath427 and the remaining are of the type @xmath428 where @xmath429 , so follows from lemma 3 .
therefore holds true , and combining with and , we get @xmath430 the assertion is proved._(ii ) _ if @xmath431 , the claim follows directly from lemma [ lemma ] and . when @xmath432 , we have @xmath433 by assumption .
hence , from and the condition @xmath434 @xmath435 this completes the proof .
@xmath384 _ proof of theorem [ th : nn2 ] .
( i ) _ from the assumption @xmath111 and proposition [ prop : nn1]_(i ) _ , we get @xmath436 , @xmath437 .
consequently , @xmath438 and the assertion follows._(ii ) _ by the condition @xmath439 , we have @xmath153 and thus proposition [ prop : nn1]_(i ) _ and lemma [ lemma]_(i ) _ yield @xmath440 .
moreover , since and @xmath439 , imply @xmath441 , it is enough to verify the poisson convergence for @xmath390 . for this
we apply the stein - chen poisson approximation method ( barbour et al . ,
1992 ) . to measure the deviation between two probability distributions @xmath442 and @xmath443 of integer - valued random variables
, we use the _ total variation distance _ @xmath444 recall the definition @xmath445 and let @xmath446 .
further , using the notation @xmath447 , we define @xmath448 since @xmath449 for all @xmath450 , corollary 2.c.5 in barbour et al .
( 1992 ) yields @xmath451 where @xmath452 denotes the distribution of @xmath390 and @xmath453 .
the @xmath0-dependence of @xmath66 gives that the number of terms in @xmath454 , for each @xmath455 , has the bound @xmath456 so for the first two sums in , lemma [ lemma ] and the condition @xmath457 , lead to @xmath458 for the last sum in , we see that the various types of terms can be counted in a similar way as the covariances in . with the notation @xmath459
, it follows from @xmath460 , and lemma [ lemma]_(i ) _ that @xmath461 note also that @xmath462 by combining , , , and , @xmath463 finally , if @xmath464 and @xmath465 , then yields @xmath322 , so from we obtain @xmath466 for each @xmath467 , and thus @xmath468 .
the statement follows . _
( iii ) _ the idea of the proof is to apply lemma [ bill95 ] to the reduced form @xmath390 of @xmath65 .
in fact , by , , and proposition [ prop : nn1]_(ii ) _ , for the last two terms in the decomposition @xmath469 we have @xmath470 and @xmath471 ( in quadratic mean ) .
consequently , it is enough to show @xmath472 the weak convergence follows from lemma [ bill95 ] if there exist successive approximations @xmath473 of @xmath474 such that 1 .
@xmath475 and @xmath476 .
2 . for every @xmath58 , @xmath477 in order to construct such @xmath478 , for each @xmath479
, define an integer @xmath480 $ ] and consider the set of @xmath244-subsets of @xmath481 @xmath482 now , based on the subset @xmath483 of @xmath484 , let @xmath485 first we prove c2 ) .
denote by @xmath486 , @xmath487 , and @xmath488 , the numbers of terms of types 2 , 3 , and 4 , respectively , defined in proposition [ prop : nn1]_(i)_. furthermore , let @xmath489 , @xmath490 , and @xmath491 be the numbers of these terms that also appear in @xmath492 since @xmath493 , we observe that @xmath494 first , for @xmath495 , we obtain a lower bound for @xmath496 . assume without loss of generality that @xmath497 .
there are @xmath498 elements @xmath499 in @xmath500 that also satisfy @xmath501 , where @xmath502 .
for every element of this type , an index @xmath359 with @xmath503 or @xmath504 can be chosen in 4 different ways . moreover , for each such alternative we can choose @xmath360 in at least @xmath505 different ways .
we get that @xmath506 is bounded from below by @xmath507 further , by proposition [ prop : nn1]_(ii ) _ , lemma [ lemma]_(i ) _ , and the limits @xmath508 and , we get @xmath509 for some @xmath510 . by a similar argument
, we have lower bounds @xmath511 and @xmath512 for @xmath513 and @xmath514 , respectively , such that , for some @xmath515 and @xmath516 , @xmath517 now , from , , , and proposition [ prop : nn1]_(ii ) _ , @xmath518 and hence , since @xmath519 has zero mean , c2 ) is implied by chebyshev s inequality . next we prove that c1 ) .
let @xmath520 \dfrac{k^3 } { ( k+m)^3 } \dfrac{\zeta^{(k)}_{1,m}}{\zeta_{1,m } } , & \quad a = \infty,\end{array } \right.\ ] ] where @xmath521 note that , since @xmath522 as @xmath523 , we have @xmath524 as @xmath523 , so c1 ) follows if it can be verified that @xmath525 to prove this , we apply the corresponding result of jammalamadaka and janson ( 1986 ) for independent samples .
in fact , if we introduce the pooled random vectors in @xmath526 @xmath527 then the @xmath0-dependence of @xmath74 implies that @xmath528 is an _
independent _ sequence in @xmath526 .
thus @xmath529 can be represented as a @xmath1-statistic with respect to the independent sample @xmath530 , @xmath531 furthermore , let @xmath532 and define @xmath533 since @xmath117 yields @xmath534 , we obtain from lemma [ lemma ] that @xmath535 and therefore @xmath536 moreover , jammalamadaka and janson ( 1986 ) show that @xmath537 as @xmath337 , and hence the stationarity of @xmath66 and give @xmath538 since @xmath57 implies @xmath539 , we get from and that the conditions of theorem 2.1 in jammalamadaka and janson ( 1986 ) are satisfied .
consequently @xmath540 further , using proposition [ prop : nn1]_(ii ) _ and definition , it is straightforward to show @xmath541 so the desired limit of @xmath478 follows from and the slutsky theorem .
this completes the proof .
@xmath384 _ proof of theorem [ th : p ] .
( i ) _ from lemma lemma , proposition [ prop : nn1]_(i ) _ , and the condition @xmath542 , @xmath543 hence , the assertion follows._(ii ) _ in order to avoid condition @xmath76 , we repeat the argument of proposition [ prop : nn1]_(i ) _ with the convergence rates in lemma [ lemma]_(ii)-(iii ) _ replaced by the weaker limits @xmath544 which follow from the stationarity and @xmath0-dependence of @xmath66 and since @xmath545 .
first we obtain from that @xmath546 moreover , if we use in place of lemma [ lemma ] in the derivation of , , , and thus finally , it follows that @xmath547 from , the last statement in , and the condition @xmath151 , @xmath548 so the claim holds true .
this completes the proof .
@xmath384 _ proof of theorem [ th : prelim ] .
( i ) _ let @xmath549 where , by proposition [ prop : nn1]_(i ) _ and the assumption @xmath550 , @xmath551 furthermore , from proposition [ prop : nn1]_(ii ) _ , @xmath552 finally , combining , , , theorem [ th : nn2]_(iii ) _ , and the slutsky theorem gives the assertion for @xmath137 .
the statement about @xmath145 follows from proposition 2 in leonenko and seleznjev ( 2010 ) . + _
( ii ) _ the details are omitted , since the argument is similar to that of _
( i ) _ , using the decomposition corresponding to with the @xmath553-scaling .
this completes the proof .
@xmath384 _ proof of theorem [ th : cons ] .
( i ) _ as in leonenko and seleznjev ( 2010 ) , the density smoothness condition yields @xmath554 this bound and proposition [ prop : nn1]_(i ) _ imply the assertion @xmath555 _ ( ii ) _ note that , by the assumptions @xmath556 and @xmath557 , @xmath558 and hence , from , @xmath559 further , since @xmath560 , we get from _
( i ) _ and that the bias fulfills @xmath561 consequently , for some @xmath562 and any @xmath563 , @xmath564 and the desired convergence for @xmath137 follows .
moreover , combining this with proposition 2 in leonenko and seleznjev ( 2010 ) proves the statement for @xmath145 . + _
( iii ) _ the argument is similar to that of _ ( ii ) _ and therefore is left out .
this completes the proof .
@xmath384 _ proof of proposition [ std ] .
_ first we study the expectation of @xmath565 . by lemma [ lemma : lee ] ,
the number of 3-tuples @xmath566 that satisfy @xmath567 , is @xmath568 .
furthermore , observe that @xmath569 of the elements @xmath570 are permutations of @xmath566 .
for the corresponding variables , @xmath571 and @xmath572 are mutually independent and also independent of @xmath573 , so for these we obtain @xmath574 thus , from lemma [ lemma ] and the assumption @xmath575 , @xmath576 next we consider the variance of @xmath565 . using the notation @xmath577 , we have @xmath578 we count the number of terms in this sum that are zero .
lemma lemma : lee implies that the number of 6-tuples @xmath579 with @xmath580 , is @xmath581 .
each such 6-tuple can be divided and permuted into @xmath582 pairs @xmath583 .
the @xmath0-dependence of @xmath66 yields that the corresponding random variables are independent , and hence at least @xmath584 summands in are zero . for each of the @xmath585 non - zero terms
, @xmath586 so lemma [ lemma]_(iii ) _ gives that the sum of the non - zero terms in is @xmath587 .
combining this with the condition @xmath588 , we get @xmath589 finally , from and it follows that @xmath590 , which completes the proof .
@xmath384 _ proof of theorem [ th : norm1 ] .
_ the argument is similar to that of theorem 6 in leonenko and seleznjev ( 2010 ) , so we show the main steps only . from the decomposition
@xmath591 we see that the assertion for @xmath137 is implied by the slutsky theorem if @xmath592 and @xmath593 .
the asymptotic normality follows straight away from theorem [ th : prelim]_(i)_. furthermore , the conditions for @xmath158 and @xmath2 together with bound lead to the desired convergence of @xmath594 .
finally , proposition 2 in leonenko and seleznjev ( 2010 ) proves the claim for @xmath145 .
this completes the proof .
@xmath384 _ proof of theorem [ th : norm2 ] .
( i ) _ we use the decomposition corresponding to : @xmath595 note that the condition @xmath596 gives @xmath153 and @xmath597 , so the asymptotic normality @xmath598 follows from theorem [ th : prelim]_(ii ) _ and the slutsky theorem .
further , the assumptions @xmath206 , @xmath596 , and bound imply @xmath599 since @xmath600 .
thus , from , , , and the slutsky theorem , we obtain the statement for @xmath137 .
the assertion for @xmath145 follows by an argument similar to that of proposition 2 in leonenko and seleznjev ( 2010 ) .
+ + _ ( ii ) _ the argument follows the same steps as that of _
( i ) _ and consequently is omitted .
this completes the proof .
@xmath384 anderson , t.w .
( 1971 ) , _ the statistical analysis of time series _ , new york : john wiley and sons .
barbour , a.d . , holst , l. , janson , s. ( 1992 ) , _ poisson approximation _
, oxford : oxford university press .
baryshnikov , y. , penrose , m.d .
, yukich , j.e .
( 2009 ) , gaussian limits for generalized spacings , _ ann .
_ , 19 , 158185 .
bickel , p.j . and ritov , y. ( 1988 ) ,
estimating integrated squared density derivatives : sharp best order of convergence estimates , _ sankhy : the indian journal of statistics _ , series a , 381393 .
billlingsley , p. ( 1995 ) , _ probability and measure _ , new york : wiley .
escolano , f. , suau , p. , bonev , b. ( 2009 ) , _ information theory in computer vision and pattern recognition _ , new york : springer .
gin , e. , nickl , r. ( 2008 ) , a simple adaptive estimator for the integrated square of a density , _ bernoulli _ , 14 , 4761 .
goria , m.n .
, leonenko , n.n . ,
mergel , v.v . ,
inverardi , p.l.n .
( 2005 ) , a new class of random vector entropy estimators and its applications in testing statistical hypotheses , _ j. nonparam .
_ , 17 , 277297 .
gregorio , a. , iacus , s.m .
( 2009 ) , on rnyi information for ergodic diffusion processes , _ inform .
_ , 179 , 279291 .
harrelson , d. , houdr , c. ( 2003 ) , a characterization of @xmath0-dependent stationary infinitely divisible sequences with applications to weak convergence , _ ann .
_ , 31 , 849881 . hart , j.d . , vieu , p. ( 1990 ) , data - driven bandwidth choice for density estimation based on dependent data , _ ann .
_ , 18 , 873890 .
jammalamadaka , s.r . ,
janson , s. ( 1986 ) , limit theorems for a triangular scheme of @xmath1-statistics with applications to inter - point distances , _ ann .
_ , 14 , 1347 - 1358 .
johnson , o. , vignat c. ( 2007 ) , some results concerning maximum rnyi entropy distributions , _ ann .
. h. poincar probab .
_ , 43 , 339351 .
joe , h. ( 1997 ) , _ multivariate models and dependence concepts _ , london : chapman and hall .
kapur , j.n .
( 1989 ) , _ maximum - entropy models in science and engineering _
, new york : wiley .
kapur , j.n . ,
kesavan , h.k .
( 1992 ) , _ entropy optimization principles with applications _ ,
new york : academic press .
kim , t.y . , luo , z.m . ,
kim , c. ( 2011 ) , the central limit theorem for degenerate variable @xmath1-statistics under dependence , _ j. nonparam .
_ , 23 , 683 - 699 .
koroljuk , v.s . , borovskich , y.v .
( 1994 ) , _ theory of @xmath1-statistics _ , dordrecht : kluwer .
kotz , s. , balakrishnan , n. , johnson , n.l .
( 2000 ) , _ continuous multivariate distributions : models and applications _ , new york : wiley .
kllberg , d. , seleznjev , o. ( 2012 ) , estimation of entropy - type integral functionals , preprint arxiv:1209.2544 .
kllberg , d. , leonenko , n. , seleznjev , o. ( 2012 ) , statistical inference for rnyi entropy functionals , _ lecture notes in comput .
_ , 7260 , 36 - 51 .
last , y. , klein , m. , kandel , a. ( 2001 ) , knowledge discovery in time series databases , _ ieee trans . systems . man . and cybernetics - part b _ , 31 , 160169 .
laurent , b. ( 1996 ) , efficient estimation of integral functionals of a density , _ ann .
_ , 24 , 659681 .
lee , a.j .
( 1990 ) , _ @xmath1-statistics : theory and practice _
, new york : marcel dekker .
leonenko , n. , pronzato , l. , savani , v. ( 2008 ) , a class of rnyi information estimators for multidimensional densities , _ ann .
_ , 36 , 2153 - 2182 .
corrections , ( 2010 ) , _ ann .
_ , 38 , 3837 - 3838 .
leonenko , n. , seleznjev , o. ( 2010 ) , statistical inference for the @xmath2-entropy and the quadratic rnyi entropy , _ j. multivariate anal .
_ , 101 , 1981 - 1994 .
neemuchwala , h. , hero , a. , carson , p. ( 2005 ) , image matching using alpha - entropy measures and entropic graphs , _ signal processing _ , 85 , 277 - 296 .
pardo , l. ( 2006 ) , _ statistical inference based on divergence measures _ , boca raton : chapman & hall .
penrose , m. , yukich , j.e .
( 2011 ) , limit theory for point processes in manifolds , _ annals of applied probability _ , to appear , see also preprint arxiv:1104.0914v1 .
principe , j.c .
( 2010 ) , _ information theoretic learning _ , new york : springer .
rnyi , a. ( 1970 ) , _ probability theory _
, amsterdam : north - holland .
seleznjev , o. , thalheim , b. ( 2003 ) , average case analysis in database problems , _ methodol .
_ , 5 , 395 - 418 .
seleznjev , o. , thalheim , b. ( 2010 ) , random databases with approximate record matching , _ methodol .
_ , 12 , 6389 . serfling , r.j . ( 2002 ) , _ approximation theorems of mathematical statistics _ , new york : wiley .
shannon , c.e .
( 1948 ) , a mathematical theory of communication , _ bell syst . tech .
_ , 27 , 379 - 423 , 623656 .
thalheim , b. ( 2000 ) , _ entity - relationship modeling .
foundations of database technology _ , berlin : springer - verlag .
tsallis , c. ( 1988 ) , possible generalization of boltzmann - gibbs statistics , _ j. stat
_ , 52 , 479487 .
ullah , a. ( 1996 ) , entropy , divergence and distance measures with econometric applications. _ j. statist .
inference _ , 49 , 137162 .
vatutin , v .
a. , mikhailov , v.g .
( 1995 ) , statistical estimation of the entropy of discrete random variables with a large number of outcomes , _ russian math .
surveys _ , 50 , 963 - 976 .
wang , q. ( 1999 ) , on berry - esseen rates for @xmath0-dependent @xmath1-statistics , _ stat .
letters _ , 41 , 123 - 130 .
consider a stationary @xmath0-dependent sequence @xmath601 with discrete @xmath25-dimensional ( marginal ) distribution @xmath602 .
we present some results on the estimation of quadratic rnyi entropy for discrete distributions @xmath603 and the corresponding quadratic functional @xmath604 where @xmath36 and @xmath37 are independent vectors with distribution @xmath3 . similarly to the continuous case , let @xmath605 define the normalized statistic @xmath606to be an estimator for @xmath22 . let @xmath607 be the corresponding estimator for @xmath21 . for @xmath608 , we also introduce the following estimator for @xmath182 , @xmath609where @xmath610 and @xmath611 are defined as in section sec : main . by an argument similar to that of proposition [ std ] , we get @xmath612 . hence , for @xmath199 , a consistent estimator for @xmath613 is given by @xmath614 some asymptotic properties for the estimators of @xmath22 and @xmath21 follow by combining the results of ch . 2 in lee ( 1990 ) , wang ( 1999 ) , and the slutsky theorem . | the rnyi entropy is a generalization of the shannon entropy and is widely used in mathematical statistics and applied sciences for quantifying the uncertainty in a probability distribution .
we consider estimation of the quadratic rnyi entropy and related functionals for the marginal distribution of a stationary @xmath0-dependent sequence .
the @xmath1-statistic estimators under study are based on the number of @xmath2-close vector observations in the corresponding sample .
a variety of asymptotic properties for these estimators are obtained ( e.g. , consistency , asymptotic normality , poisson convergence ) .
the results can be used in diverse statistical and computer science problems whenever the conventional independence assumption is too strong ( e.g. , @xmath2-keys in time series databases , distribution identification problems for dependent samples ) .
= 3.4 ex _
ams 2010 subject classification : _
62g05 , 62g20 , 62m99 , 94a17 _ keywords : _
entropy estimation , quadratic rnyi entropy , stationary @xmath0-dependent sequence , inter - point distances , @xmath1-statistics |
recently , very bright objects which may be undergoing supercritical ( or super - eddington ) accretion flows have successively been found .
good examples are ultraluminous x - ray sources ( ulxs ; @xcite @xcite ; @xcite @xcite ; @xcite @xcite ) .
these are pointlike off - center x - ray sources whose x - ray luminosity significantly exceeds the eddington luminosity of a neutron star @xcite . because of substantial variations , it is reasonable to assume that the ulxs are single compact objects powered by accretion flows @xcite . if so , there are two possibilities to account for large luminosities exceeding the eddington luminosity for a mass of 100 @xmath13 : sub - critical accretion onto an intermediate - mass black hole ( imbh ) and supercritical accretion onto a stellar - mass black hole .
we support the latter possibility , since through the fitting to several _ xmm - newton _ epic data of ulxs , which have been claimed as good imbh candidates , we have found evidence of supercritical flows @xcite .
another interesting group is narrow - line seyfert 1 galaxies ( see * ? ? ?
* for a review ) .
because of their relatively small black hole masses , they have in general large eddington ratios ( @xmath14 with @xmath15 and @xmath1 being the luminosity and the eddington luminosity , respectively ) , and some of them seem to fall in the slim - disk regimes ( @xcite @xcite ; @xcite @xcite ; see also @xcite @xcite ) despite growing evidence indicating the existence of supercritical accretion flows in the universe , theoretical understanding is far from being complete .
it is well known that any spherically accreting object , irrespective of the nature of the central source , can not emit above the eddington luminosity , since otherwise significant radiative force will prevent accretion of the gas .
if we examine detailed radiation - matter interactions in the interior , however , we notice that the situation is not so simple .
actually , radiation produced at the very center can not immediately reach the surface ( i.e. , photosphere ) , since photons generated at the center should suffer numerous scatterings with accreting material and thus take a long time to reach the surface . if the matter continuously falls , and if the mass accretion timescale is shorter than the mean travel time for photons to reach the surface ( the diffusion timescale ) , photons at the core may not be able to go out .
this is the so - called photon - trapping effect ( e.g. , * ? ? ?
* ; * ? ? ?
here we define the trapping radius , inside which radiation - matter interaction is so frequent that photons are trapped within the accretion flow . inside this trapping radius , therefore , radiative flux can be negative ( i.e. , inward ) .
thus , the apparent luminosity is reduced as compared with the case without photon trapping .
nevertheless , the concept of the eddington luminosity is still valid , since far outside the trapping radius radiative flux should be outward and its absolute value should be less than @xmath16 , where @xmath4 is the distance from the central black hole , in the quasi - steady state .
radiation - hydrodynamic ( rhd ) simulations of spherically symmetric supercritical accretion flows have been performed by @xcite .
the situation may differ in the case of disk accretion , since the radiation field is not isotropic due to inhomogeneous matter distribution .
that is , matter can fall predominantly along the disk plane , whereas radiation can go out along its rotating axis , where matter density is low .
in other words , the main directions of the inward matter flow and the outward radiative flux are not parallel to each other in disk accretion , leading to a situation in which the radiative force does not completely counteract the gravitational force .
there is room for the possibility of a supercritical accretion flow with super - eddington luminosity .
based on such an argument , many researchers have speculated about the occurrence of supercritical flows in disk accretion systems . @xcite
discussed the possibility of supercritical disk accretion based on a one - dimensional steady model ( see also * ? ? ?
they mentioned that the mass accretion rate would not be steady but oscillate if the mass accretion rate exceeded the critical rate ; otherwise , a part of the accreting matter might be ejected from the disk as a disk wind .
radiatively driven outflows from supercritical disks were investigated by @xcite , @xcite , and @xcite . despite a long history in the study of supercritical disk accretion flows , the occurrence of steady supercritical disk accretion has not yet been accurately verified .
similar simulations have been performed since the 1980s , but all of them calculated only the initial transient phase .
their conclusions then were not general , since the back - reactions ( i.e. , enhanced radiation pressure ) , which may inhibit steady flow , were not accurately evaluated .
although the radiative force predominantly has an effect in the vertical direction , it should also have an effect on the material within the disk .
hence , in the direction parallel to the disk plane , the situation may be similar to or more severe than the case of spherical accretion .
this is because the radiative force , together with the centrifugal force , may possibly overcome the gravitational force . to study the possibility of supercritical disk accretion flows , precise and quantitative research treating both supercritical disks and outflows , and which also takes into consideration multi - dimensional effects ,
is needed .
@xcite ( ( * ? ? ?
* hereafter paper i ) ) have confirmed the occurrence of quasi - steady supercritical disk accretion onto a black hole by two - dimensional rhd simulations .
the motivation of the present study is to investigate the physical mechanisms which make supercritical disk accretion possible . for this purpose
we examine quantitatively the flow motion and force fields via the radiative flux , rotation , and gravity of a supercritical disk accretion flow onto a central black hole , based on the two - dimensional rhd simulation data from paper i. through detailed inspection of the results , it will be possible to clarify the physics behind supercritical disk accretion flows . in 2 we plot the spatial distributions of several key quantities which control flow dynamics .
a discussion is given in 3 .
here we adopt the spherical coordinates @xmath17 , with @xmath18 corresponding to the disk plane and the origin being set at the central black hole . the momentum equation for the radial component of the flow
is @xmath19 here @xmath20 is the radial component of the flow velocity , @xmath21 is the lagrangian derivative , @xmath22 is the gravitational force by the central black hole , @xmath23 is the radiative force with @xmath24 being absorption and scattering coefficients , @xmath2 being the speed of light , and @xmath25 representing the radial component of the radiative flux in the comoving frame , @xmath26 is the radial component of the gas pressure force , and @xmath27 is the radial component of the centrifugal force .
we adopt a pseudo - newtonian potential , thus having @xmath28 with @xmath29 being the mass of the black hole and @xmath30 @xcite .
to proceed , it is important to distinguish two different views : the view from an observer comoving with the accreting gas ( i.e. , the comoving frame ) and the view from an observer standing at infinity ( the rest frame ) .
the @xmath31th component of radiative flux in the comoving frame , @xmath32 , is related to that in the rest frame , @xmath33 , on the order of @xmath34 , @xmath35 where @xmath10 and @xmath36 are the radiation energy density and the radiation stress tensor in the comoving frame , respectively . here
the final term in equation ( [ flux ] ) is equal to @xmath37 in the optically thick diffusion limit . in 2.3
we show that the gas accretes toward the black hole in the disk region . in this region ,
even if the radiative flux in the comoving frame is positive , @xmath38 ( i.e. , outward flux ) , @xmath39 can be negative , @xmath40 ( inward flux ) , because @xmath41 .
the term @xmath42 thus represents photon trapping . in
2.5 we explicitly show this effect in the simulation data .
in contrast , the flow dynamics in the direction perpendicular to the disk plane is distinct . in particular
, the centrifugal force has no effect in this direction .
in addition , the matter density is significantly smaller along the rotation axis . hence , the radiative flux can be much more effective in the vertical direction , thus driving strong outflows , as we see in 2.3 .
our present analysis is based on the two - dimensional rhd simulation data from paper i. these are the first simulations of supercritical accretion flow in the quasi - steady state .
although the research history of such simulations stems back to the late 1980s , when @xcite performed numerical simulations for the first time , their calculations were restricted to the first few seconds ( see also @xcite @xcite ; @xcite @xcite ; @xcite @xcite ; @xcite @xcite ) . back - reactions were not fully taken into account in their simulations . in our simulations presented in paper i , matter is added continuously from the outer disk boundary at @xmath43 to the initially ( nearly ) empty space at the rate of @xmath44 .
the injected mass has a small specific angular momentum so that it first free falls but then forms a rotating disk at @xmath45 .
mirror symmetry is assumed with respect to the disk ( equatorial ) plane .
the @xmath46-prescription of the viscosity is employed , in which the viscosity is set to be proportional to the total pressure in the optically thick limit and the gas pressure only in the optically thin region .
hence , the viscosity is more effective in the disk region than in the outflow regions above and below the disk .
the viscosity parameter is assumed to be @xmath47 .
except for the radial - azimuthal component , all components of the viscous stress tensor are set to be zero .
the radiative transfer was solved under the flux - limited diffusion approximation @xcite .
the central object is taken to be a non rotating stellar - mass black hole ( @xmath48 ) , generating a pseudo - newtonian potential .
only @xmath49 of the inflowing material finally reaches the inner boundary ( @xmath50 ) ; i.e. , @xmath51 of the mass input gets stuck in the dense , disk like structure around the equatorial plane , or transforms into the known collimated high - velocity outflows perpendicular to the equatorial plane or into low - velocity outflows with wider opening angles .
the resulting luminosity is about 3 times larger than the eddington luminosity .
figure [ rhoe ] indicates the time - averaged contours of the matter density ( _ left panel _ ) and the radiation energy density ( _ right panel _ ) on the @xmath52-@xmath53 plane ( @xmath54 and @xmath55 ) , where both axes are normalized by the schwarzschild radius ( @xmath12 ) .
note that these panels are similar to figures 3 and 5 of paper i , except that figure [ rhoe ] of the present paper has contours which are time - averaged over @xmath56 s , whereas those in paper i are snapshots at @xmath57 s ( i.e. , in a quasi - steady phase ) .
the high concentration of gas and radiation in the region near the black hole on the equatorial plane is clear .
first we plot the time - averaged radiation energy density normalized by @xmath58 in figure [ e0 ] .
we find that this value is larger than unity in most regions .
in particular , we have @xmath59 in the disk region .
this means that there exists so large a number of photons that if we simply assume @xmath60 as in the optically thin region , the radiative force should highly exceed the gravity , which inevitably prevents the inflow motion .
the simulation data , however , do show inflow motion ( discussed below ) .
figure [ 2dforce ] illustrates the time - averaged force balance in the radial ( @xmath4 ) directions with various @xmath61-values from the origin .
( it should be noted that since force is a vector , we need to specify the direction to draw such contours .
here we only plot the radial component of the forces . )
figure [ 2dforce ] ( _ left _ ) represents the ratio of the radial component of the radiative force to the gravitational force . especially surprising is that most regions are turquoise , indicating that the radiative force is outward and less than the gravitational force .
why is the radiative force so weak , even though the radiation energy density is very large , @xmath62 ( see fig . [ e0 ] ) , and even though the total luminosity exceeds the eddington luminosity , @xmath63 ?
this is because the disk is very optically thick .
although the absolute value of the radiative flux is equal to @xmath64 in the optically thin limit , it is roughly given by @xmath65 , and hence , @xmath66 can be very small , @xmath67 , in the optically thick region ( @xmath68 ) , where @xmath69 is the optical thickness . approximately , we find @xmath70 , where @xmath71 is the gas density .
thus , flux is attenuated a great deal .
we thus have a relatively small radiative flux within the disk in spite of a large @xmath10 ( see 2.4 for more detail ) .
there are two exceptions : the color of the region in the vicinity of the black hole is blue , which means that the radiative force is inward . in this region , a large number of photons are swallowed by the black hole , @xmath72 , so that the gas is accelerated inward .
the region near the rotation axis , in contrast , has a white color , which means that the radiative force exceeds the gravitational force . to discuss how matter flows
, we also need to consider the centrifugal force .
( note that the gas pressure is negligible in the present case . )
figure [ 2dforce ] ( _ right _ ) illustrates the ratio of the sum of the radial components of the radiation and centrifugal forces to the gravitational force .
certainly , the repulsing force is enhanced along the disk plane .
however , this force ratio is still around unity in most regions except in the very vicinity of the black hole .
that is , the gas is nearly in balance of forces in wide regions .
the regions near the rotation axis again have white colors , which means that the radiative force dominates , as in figure [ 2dforce ] ( _ left _ ) ( note that the centrifugal force is not effective near the rotation axis . ) since there are regions with a negative effective force in the radial direction , we expect regions with negative velocity ( i.e. , inflow ) .
figure [ vr ] shows the spatial distribution of the radial component of the matter velocity in units of the free - fall velocity .
interestingly , the flow is inward in the disk region , but the infall velocity is much smaller than the free - fall velocity .
along the rotation axis , conversely , the flow is outward , and its magnitude is larger than the free - fall velocity . let us have a more quantitative discussion based on the one - dimensional distributions of these quantities , since it is difficult to obtain their precise values from the color contours .
we calculate several quantities along two radial lines : one at 0.45@xmath73 ( close to the equatorial plane ) and the other at @xmath74 ( nearly along the rotation axis ) .
they are illustrated in the top and bottom panels of figure [ 1dforce ] , respectively .
we see in the top panel that the radiative force is not negligible , although it is smaller than the centrifugal force .
the sum of the radiative and centrifugal forces is nearly balanced with the gravitational force .
radiation does work as a `` radiation cushion , '' which decelerates the accretion of the gas in cooperation with the rotation ( centrifugal force ) .
deceleration via radiative flux was also reported by @xcite .
figure [ 1dforce ] ( _ top _ ) also shows that the radial velocity is much smaller than the free - fall velocity and only barely below zero ( inflow ) , although there also exist regions with @xmath75 ( no inflow ) .
in contrast , figure [ 1dforce ] ( _ bottom _ ) shows that the ratio of the radiative force to the gravitational force grows as the distance increases .
the outward velocity thus grows rapidly upward .
we confirm again that the strong outflow around the rotation axis is produced by the strong radiative force .
however , the velocity is negative in the very vicinity of the hole .
this means that the radiation which produces the huge radiative force in the vertical direction does not originate from the region just outside the event horizon but from the region with @xmath10 at maximum [ at @xmath76 .
here we note that the angular momentum is very small around the rotation axis , leading to a negligible centrifugal force ( _ dotted line _ ) . this is because a significant amount of material exists even inside the radius of the marginally stable orbit , @xmath77 , in the supercritical flow ,
so the matter can lose angular momentum there @xcite .
this situation contrasts with that of the subcritical flow , in which there is a sharp density drop at @xmath78 .
our computational domain covers exactly down to the rotation axis .
numerical simulations show that most of the outflow matter is blown away into an oblique direction by the radiative force in cooperation with the centrifugal force . here
we consider the reason why the radiative force in the radial direction does not dominate over the gravitational force inside the disk , even though @xmath79 is larger than @xmath80 as shown in figure [ e0 ] .
the radial component of the radiative flux is evaluated as @xmath81 in the disk region .
it is attenuated since the disk is dense and very optically thick ( @xmath68 ) .
that is , even though the radiation energy density itself is large , the high matter density significantly suppresses the radiative flux inside the disk . however , the high density ( large optical thickness ) is not the only condition for the occurrence of supercritical accretion . in the case of a spherical system
, the radiative flux should be @xmath82 even if the matter is very optically thick .
the radiative force dominates over the gravitational force when @xmath83 , preventing the accretion of matter .
hence , there should be another reason that supercritical disk accretion is realized in spite of @xmath83 .
whereas the radiation field is isotropic in the spherical case by definition , it can be anisotropic in the disk case . indeed , the simulated profile of the radiation energy density is highly anisotropic ( fig .
[ rhoe ] , _ right _ ) .
unlike in spherical geometry , photons can escape from the less dense region around the rotation axis without thrusting through the dense disk region .
thus , the effective radiative force is attenuated in the disk region .
here we note that , even in the presence of an anisotropic matter distribution , @xmath15 can not exceed @xmath1 if the medium is optically thin . to sum up , supercritical disk accretion with super - eddington luminosity
is realized due to a large optical thickness and anisotropy of the radiation field .
photon - trapping effects assist the occurrence of supercritical accretion .
figure [ flux2 ] ( _ top _ ) indicates the radiative fluxes in the comoving and rest frames inside the disk . as shown in this figure ,
the radial component of the radiative flux in the rest frame ( _ dotted line _ ) is negative in the region of @xmath84 , whereas that in the comoving frame ( _ solid line _ ) is positive except in the vicinity of the black hole .
this means that the radiation energy is transported inward via photon trapping , @xmath85 ( see equation [ [ flux ] ] ) .
some part of this radiation energy is swallowed by the black hole , whereby the luminosity is reduced .
this effect attenuates the radiative force , supporting the accretion .
the anisotropy of the radiation field is enhanced by photon trapping .
this is another important role of the photon trapping .
the radiation energy , which is advected inward with the matter by the photon trapping but not swallowed by the black hole , is transported to the vertical direction and finally released from the disk surface .
it is shown in figure 5 ( _ bottom _ ) , which plots the @xmath61-components of the radiative fluxes in the comoving ( _ solid line _ ) and the rest ( _ dotted line _ ) frames along a radial line at @xmath86 , that the entire region is below the photosphere .
in this figure we can see that the radiation energy is transported toward the disk surface , @xmath87 , in most of the region .
we also find @xmath88 in this figure .
this implies that the gas motion contributes to the transport of radiation energy toward the disk surface , @xmath89 . to conclude , the photon - trapping effect enhances the anisotropy of the radiation field , assisting the occurrence of supercritical accretion . here
we note that the gas motion in the vertical direction seems to be driven by convection in cooperation with the radiative force .
in fact , we find that the adiabatic conditions for radiation - pressure - dominated matter are roughly satisfied along the vertical line in the disk region , @xmath90 and @xmath91 , where @xmath92 is the total pressure and @xmath93 is the temperature .
in this figure the radiative flux in the comoving frame , @xmath25 , is positive ( outward ) except in the vicinity of the black hole .
however , the radiative flux in the rest frame , @xmath94 , is negative ( inward ) via photon trapping , @xmath85 , in the trapping region . in the vicinity of the black hole ,
both @xmath25 and @xmath39 are negative .
matter slowly accretes inside the disk , since the sum of the radiative and centrifugal forces is nearly balanced with the gravitational force . here
we stress again that the radiative force counteracts the gravitational force in spite of the trapping region because @xmath95 .
radiatively driven high - velocity outflows appear above and below the disk . in the very vicinity of the black hole ,
the gas is accelerated inward by the radiative force and the gravitational force . as far as the physical quantities around the equatorial plane are concerned , the simulated profiles of the density , temperature , and radial as well as rotational velocities roughly agree with the prediction of the slim - disk model @xcite .
such features have already been shown in figure 11 in paper i. however , only about @xmath49 of the injected mass can accrete onto the black hole , and an almost equal amount of matter is ejected as high - velocity outflows .
the mass accretion rate is not constant in the radial direction and decreases near the black hole ( see fig . 6 in paper
thus , we conclude that the simulated flows do not perfectly agree with the slim - disk model with regard to the whole structure of the flow .
@xcite have investigated local maximum values of the accretion rate in the supercritical disk accretion flows . in their paper
they focused only on the force balance in the vertical and radial directions around the equatorial plane .
multi - dimensional effects were not taken into consideration .
they revealed that the vertical radiative force limits the maximum accretion rate at the inner disk region , leading to a decrease of the accretion rate with a decrease of the radius .
their results imply that the disk loses mass via the radiatively driven outflows and the mass accretion rate decreases at the inner region .
such tendencies agree with our results .
as shown in figure [ vr ] , our simulations show that radiatively driven outflows form above and below the disk .
the mass accretion rate decreases with a decrease of the radius ( see fig . 6 in paper i ) . in the present study , focusing on numerical simulations in which the mass input rate at the outer boundary is set to be @xmath96
, we show that the radiative force is attenuated in the disk region via large optical thickness , which makes supercritical disk accretion possible .
such dilution of the radiative force would effectively operate even if the mass input rate ( mass accretion rate ) varied .
in fact , simulations with mass input rates of @xmath97 and @xmath98 show that @xmath99 is almost independent of the mass input rate , although the radiation energy density goes up as the mass input rate increases .
that is , the dynamics is not sensitive to the precise value of @xmath100 , as long as it largely exceeds the critical value .
so far , we have studied steady accretion flows .
although a highly supercritical disk ( @xmath101 ) is quasi - steady , it has been revealed that a moderately supercritical disk [ @xmath102 is unstable and exhibits limit - cycle behavior ( @xcite @xcite , @xcite ; see also @xcite @xcite ; @xcite @xcite ) .
the luminosity goes up and down around the eddington luminosity .
@xcite has reported that the time - averaged mass , momentum , and kinetic energy output rates via the outflow , the mass accretion rate , and the disk luminosity increase as the mass input rate increases , @xmath103 for @xmath104 and @xmath105 for @xmath106 . as we have already mentioned in 3.1 ,
the sum of the accreting matter and the matter ejected as high - velocity outflows is @xmath107 of the injected mass , and @xmath108 of the injected matter is ejected from the computational domain as low - velocity outflows , whose velocities do not exceed the escape velocity at the outer boundary . since such outflowing matter tends to be accelerated by the radiative force even at the outside of the computational domain , it would be blown away from the system . however
, a part of the outflowing matter might return to the vicinity of the black hole through the disk region , since the radiative force does not exceed the gravity near the equatorial plane .
numerical simulations with larger computational domains would make this point clear . whereas the resulting mass accretion rate onto the black hole is around @xmath109 in the present simulations ,
@xcite have indicated that the mass accretion rate can increase up to @xmath110 .
they have reported that the vertical force balance breaks down via a strong radiative force if the mass accretion rate exceeds this limit . however , even in such a case , the matter might accrete onto the black hole , although the strong radiative force would produce powerful outflows . to investigate
the maximum value of the accretion rate is an outstanding issue .
we should perform numerical simulations with larger computational domains , since the trapping region is expected to expand with the increase of the mass accretion rate .
we reveal in the present paper that photons generated deep inside the disk are effectively trapped in the flow , leading to supercritical disk accretion .
although the magnetic fields are not solved in our simulations , magnetic buoyancy might play an important role in the transportation of matter , as well as photons , toward the disk surface ( @xcite @xcite ; @xcite @xcite ; @xcite @xcite ) .
magnetic buoyancy might lead to photon generation near the disk surface if the magnetic fields rise quickly without dissipation deep inside the disk .
thus , magnetic buoyancy would dilute the photon - trapping effect .
a photon bubble instability , which is induced in the magnetized , radiation - pressure - dominated region , might also suppress the photon trapping @xcite . in these cases
the enhanced radiative force would more effectively accelerate the matter around the disk surface , working to decrease the mass accretion rate .
however , the magnetic fields might prevent such acceleration if they strongly tie the matter near the disk surface with the disk matter . in the disk region
the matter might easily accrete toward the black hole , since the radiation energy density decreases .
global radiation - magnetohydrodynamic ( rmhd ) simulations would make these points clear .
local rmhd simulations of accretion flows have been performed by @xcite and @xcite .
in addition , it is thought that disk viscosity has magnetic origins ( @xcite @xcite ; @xcite @xcite ; for a review see @xcite @xcite ) .
hence , we should stress again that rmhd simulations are very important to more realistically investigate viscous accretion flows , although an @xmath46-viscosity model is employed in the present study .
we would like to thank the anonymous reviewer for many helpful suggestions .
the calculations were carried out by a parallel computer at rikkyo university and the institute of natural science , senshu university .
this work was supported in part by a special postdoctoral researchers program in riken ( k. o. ) , by a research grant from the japan society for the promotion of science ( 17740111 ; k. o. ) , by grants - in - aid from the ministry of education , science , culture , and sports ( 14079205 , 16340057 ; s. m. ) , and by the grant - in - aid for the 21st century coe `` center for diversity and universality in physics '' from the ministry of education , culture , sports , science , and technology of japan .
abramowicz , m. a. , czerny , b. , lasota , j. p. , & szuszkiewicz , e. 1988 , , 332 , 646 balbus , s. a. 2003 , , 41 , 555 begelman , m. c. 1978 , , 184 , 53 begelman , m. c. 2002 , , 568 , l97 boller , t. 2004 , pthps , 155 , 217 burger , h. l. & katz , j. i. 1980 , , 236 , 921 burger , h. l. & katz , j. i. 1983 , , 265 , 393 ebisawa , k. , zycki , p. , kubota , a. , mizuno , t. , & watarai , k. 2003 , , 597 , 780 eggum , g. e. , coroniti , f. v. , & katz , j. i. 1987 , , 323 , 634 fabbiano , g. 1989 , , 27 , 87 hawley , j. f. , balbus , s. a. & stone , j. m. 2001 , , 554 , l49 heinzeller , d. , & duschl , w. j. 2007 , , 374 , 1146 hirose , s. , krolik , j. h. , & stone , j. m. 2006 , , 640 , 901 houck , j. c. & chevalier , r. a. 1991 , , 376 , 234 icke , v. 1980 , , 85 , 329 jones , b. c. & raine , d. j. 1979 , , 76 , 179 kato , s. , fukue , j. , & mineshige , s. 1998 , black - hole accretion disks ( kyoto : kyoto univ .
press ) kawaguchi , t. 2003 , , 593 , 69 kley , w. 1989 , , 222 , 141 kley , w. & lin , d. n. c. 1999 , , 518 , 833 levermore , c. d. & pomraning , g. c. 1981 , , 248 , l321 makishima , k. , et al .
2000 , , 535 , 632 machida , m. , matsumoto , r. , & mineshige , s. 2001 , , 53 , l1 maraschi , l. , reina , c. , & treves , a. 1976 , , 206 , 295 meier , d. l. 1979 , , 233 , 664 mineshige , s. , kawaguchi , t. , takeuchi , m. , & hayashida , k. 2000 , , 52 , 499 ohsuga , k. 2006 , , 640 , 923 ohsuga , k. 2007 , , 659 , 205 ohsuga , k. , mori , m. , nakamoto , t. , & mineshige , s. 2005 , , 628 , 368 ( paper i ) okajima , t. , ebisawa , k. , & kawaguchi , t. 2006 , , 652 , l105 okuda , t. 2002 , , 54 , 253 okuda , t. , fujita , m. , & sakashita , s. 1997 , , 49 , 679 paczysky , b. & wiita , p. j. 1980 , , 88 , 23 parker , e. n. 1975 , , 198 , 205 sakimoto , p. j. , & coroniti , f. v. 1989 , , 342 , 49 shakura , n. i. & sunyaev , r. a. 1973 , , 24 , 337 shakura , n. i. , & sunyaev , r. a. 1976 , , 175 , 613 shibazaki , n. & hshi , r. 1975 , pthph , 54 , 706 stella , l. , & rosner , r. 1984 , , 277 , 312 turner , n. j. , blaes , o. m. , socrates , a. , begelman , m. c. , & davis , s. w. 2005 , , 624 , 267 turner , n. j. , stone , j. m. , krolik , j. h. , & sano , t. 2003 , , 593 , 992 vierdayanti , k. , mineshige , s. , ebisawa , k. , & kawaguchi , t. 2006 , , 58 , 915 wang , j .- m . ,
szuszkiewicz , e. , lu , f .- j .
, & zhou , y .- y . 1999 , , 522 , 839 watarai , k. , & mineshige , s. 2003 , , 55 , 959 watarai , k. , mizuno , t. , & mineshige , s. 2001 , , 549 , l77 | although the occurrence of steady supercritical disk accretion onto a black hole has been speculated about since the 1970s , it has not been accurately verified so far . for the first time
, we previously demonstrated it through two - dimensional , long - term radiation - hydrodynamic simulations . to clarify why this accretion is possible , we quantitatively investigate the dynamics of a simulated supercritical accretion flow with a mass accretion rate of @xmath0 ( with @xmath1 and @xmath2 being , respectively , the eddington luminosity and the speed of light ) .
we confirm two important mechanisms underlying supercritical disk accretion flow , as previously claimed , one of which is the radiation anisotropy arising from the anisotropic density distribution of very optically thick material .
we qualitatively show that despite a very large radiation energy density , @xmath3 ( with @xmath4 being the distance from the black hole ) , the radiative flux @xmath5 could be small due to a large optical depth , typically @xmath6 , in the disk .
another mechanism is photon trapping , quantified by @xmath7 , where @xmath8 is the flow velocity . with a large @xmath9 and @xmath10 , this term significantly reduces the radiative flux and even makes it negative ( inward ) at @xmath11 , where @xmath12 is the schwarzschild radius . due to the combination of these effects ,
the radiative force in the direction along the disk plane is largely attenuated so that the gravitational force barely exceeds the sum of the radiative force and the centrifugal force . as a result
, matter can slowly fall onto the central black hole mainly along the disk plane with velocity much less than the free - fall velocity , even though the disk luminosity exceeds the eddington luminosity . along the disk rotation axis , in contrast ,
the strong radiative force drives strong gas outflows . |
the study of penetrating probes constitutes a key aspect of the relativistic heavy ion program . in the hadronic sector ,
jet quenching , for example , has been a striking revelation , and has contributed to expose the qualitatively different physics that appeared in the transition from the sps to rhic . in this context , electromagnetic radiation also defines a privileged class of observables owing mainly to the absence of final state interactions .
we briefly review some recent developments in the measurement of low- and intermediate - mass lepton pairs , and then address real photon measurements at rhic , together with the observed supression in the hard parton spectrum .
at sps energies , the measurement of low - mass lepton pairs had previously been made by the helios/3 @xcite and by the ceres @xcite experimental collaborations . as a reminder of the potential of such measurements for the discovery of new physics , it is useful to write the emission rate of lepton pairs from a finite - temperature interacting system .
it is @xcite @xmath0 where @xmath1 is a bose - einstein distribution function , @xmath2 is a lepton tensor , and @xmath3 is the imaginary part of the in - medium , retarded , finite - temperature self - energy of the photon .
furthermore , in the nonperturbative sector , vector meson dominance ( vmd ) relates the photon self - energy to the in - medium vector spectral density .
it is therefore clear that the measurement of the low - mass spectrum of lepton pairs can reveal pristine features of the interacting many - body system . in this regard ,
the situation before this conference was summarized in a presentation of the ceres collaboration @xcite , where three theoretical approaches were shown to be consistent with the data .
those were ( i ) a many - body calculation where the in - medium spectral densities are altered owing to their coupling with a variety of states accessed through interactions with a hot and dense hadronic ensemble @xcite ; ( ii ) an effective chiral model , where strong precursor effects already manifest themselves in a shifting of hadronic masses at intermediate baryonic densities @xcite ; ( iii ) a simple thermal parametrization of the quark - antiquark annihilation born rates , justified by a duality argument @xcite .
finally and importantly , approaches solely based on vacuum properties have difficulty providing an interpretation of the ceres measurements .
it is probably fair to write that one of the highlights of the quark matter 2005 meeting has been the disclosure of the new na60 data @xcite for in - in collisions at the cern sps .
the low - mass dilepton component of this experiment s measurements is of exceptional quality and statistics @xcite .
the experimental collaboration also has shown a comparison with their data with the results of approaches ( i ) and ( ii ) , above .
on the basis of that comparison , the experimental data strongly suggest that large mass shifts of the vector mesons are not observed . this same comparison also shows consistency with the many - body calculations of ( i ) above .
an immediate conclusion is the following : na60 data are now precise enough to distinguish between different approaches , or at least between some of the more extreme scenarios .
this is important progress , and represents a great stride forward .
whether this rules out or not entire classes of models will remain to be seen , but one point is clear : the approach shown by the experimental collaboration which is based on in - medium mass shifts _ is the same one that was in agreement with ceres results _ @xcite .
it is entirely possible that theoretical refinements are needed in order to be in line with evolving theoretical paradigms , but the fact remains that any single theory now has to deal with two separate experimental results .
the low - mass results do signal an unambiguous many - body effect . to go beyond this to a state of deeper theoretical understanding
is not an easy talk .
for example , one of the original aims of this whole experimental program was to isolate a signal ( precursor or not ) of chiral symmetry restoration .
this goal has unfortunately remained elusive , owing largely to the difficulty of directly extracting an axialvector correlator from relativistic heavy ion data . in the chiral
limit , vector and axialvector correlators are constrained by sum rules of the weinberg - type @xcite , and these may be realized in several different ways @xcite . the ability to chart a path to correlator degeneracy ( in the weinberg sense ) would definitely represent a significant breakthrough .
however , the precise data shown at this meeting will surely fuel many investigations .
the study of lepton pair production at intermediate masses ( @xmath4 ) are especially interesting in the context of searches for the quark - gluon plasma , as kinematical arguments combined with the original high temperature of the qcd plasma would designate the intermediate invariant mass region as a window of opportunity for the observation of plasma radiation @xcite . in this context
, considerable interest was generated by the fact that an excess over sources expected from pa measurements has been confirmed in the intermediate mass region by the helios/3 @xcite and na50 @xcite collaborations .
this excess could in turn signal an increase in @xmath5 abundances , which would then manifest itself through the correlated semileptonic decays of open charm mesons . alternatively
, thermal lepton pairs need to be ruled out as viable scenario before other explanations be invoked .
those thermal sources are akin to the ones identified in the low dilepton mass sector by the many - body calculations .
showing that they also shine at higher invariant masses would go a long way in providing a consistent picture of electromagnetic radiation in heavy ion collisions . in
the theoretical interpretation of such data , a potential caveat is lurking , and is related to the use of effective lagrangian techniques .
these rely on physical parameters which were essentially all fitted in regions characterized by soft energy scales , and had mostly to do with strong and electromagnetic decay widths . when moving over to larger invariant masses
, off - shell effects will set in , making controlled extrapolations a problem @xcite .
fortunately , there exists a wealth of data of the type @xmath6 hadrons , which cover precisely the same range in invariant mass as the heavy ion data @xcite .
these data can be analyzed channel - by - channel and have been used , together with @xmath7-decay measurements , to construct the vector and axial vector spectral densities that can be related to the lepton pair spectrum @xcite .
summing all kinematically - relevant channels , one arrives at a source which may be compared to the data via the space - time modeling of the nuclear collision .
the result of one such exercise is shown in figure [ na50dilep ] .
it is fair to write that the specific values of the temperature evolution , for example , depend somewhat on the particularities of the space - time modeling .
however , a fairly robust conclusion still emerges : the intermediate - mass na50 data does not demand a large contribution from plasma radiation ( it is about 20% here ) , nor does it need a large enhancement of the initial charm content .
even though specific details do differ , this conclusion is shared by other theoretical studies of a similar nature @xcite . however , in order to bring this situation to some degree of closure , a direct measurement of the strangeness component would go a long way .
this has now been done and has been reported at this conference by the na60 collaboration @xcite .
the first measures taken by the collaboration was to experimentally confirm the presence of an enhancement in the intermediate - mass continuum in nuclear collisions , over what is seen in proton - nucleus events .
however , with the new information provided by the muon offset measurement , na60 can now assert that the excess is not linked with open charm enhancement .
this new data is also compatible with the excess that had previously been observed by na50 .
the analysis will proceed further , but it already reveals that the signal that exceeds what is associated with pa sources increases faster than linearly with the number of participants .
the statistics will improve , the 2004 proton - nucleus data will be analyzed and such measurements are important for the determination of qcd effects beyond leading - twist @xcite .
it is however clear that these results do represent a great leap in our quest for a complete quantitative understanding of the electromagnetic radiation being produced in relativistic nuclear collisions .
one of the most striking findings of the rhic program is the strong apparent modification of jet characteristics , following a passage through an interacting , dense medium @xcite .
a compelling theoretical interpretation of these results is that of jet absorption , signaling in effect the existence of a hot and dense partonic phase .
several models of jet - quenching through gluon bremsstrahlung have been elaborated @xcite . here
, we use that of amy @xcite , because of its potential to handle consistently jet energy loss and photon emission . in this approach ,
fokker - planck equations are solved to obtain the time - evolution of the initial hard gluon , @xmath8 and hard quark plus antiquark distributions , @xmath9 .
the coupled equations are @xmath10 the kernels @xmath11 are the transition rates , and they contain the resummation effects typical of interactions with a thermal medium @xcite .
the solution of the joint equations for the time - evolution of the parton distribution functions permits the modeling in real time of the partonic spectra .
the hard parton can then fragment into the different varieties of observed particles .
up to suppressed corrections , the cross section for produced pions in nucleon - nucleon collisions can be written in a factorized form as @xmath12 where @xmath13 is the parton distribution function in a nucleon , @xmath14 is the pion fragmentation function , @xmath15 is the parton - parton cross section at leading order , and @xmath16 accounts for higher order effects . here , `` jet '' essentially means a fast parton with @xmath17 1 gev . this procedure , with the factorization scale @xmath18 and the fragmentation scale @xmath19 set equal to @xmath20 , the cteq5 parton distribution functions , and @xmath21 , does a very good job of reproducing the measured @xmath22 above @xmath23 5 gev @xcite , in nucleon - nucleon collisions at rhic .
this is shown in figure [ pi0 ] . to obtain the high @xmath20 @xmath22 cross section in aa collisions ,
the pp calculation must be modified in two important ways .
first , the parton distribution function of a nucleus differs from that of a proton : @xmath24 the nuclear modification factor of the structure function @xmath25 takes into account shadowing and anti - shadowing @xcite . also , the hard parton looses energy between the initial hard scattering and its hadronization , and this information in contained in the time - evolution of the entire partonic profile we have described earlier .
the @xmath22 spectrum obtained in aa collisions is then @xmath26 where @xmath27 is the average number of binary collisions , @xmath28 is the inelastic nucleon - nucleon cross section , and the medium - modified fragmentation function is @xmath29 where @xmath30 takes into account the geometry of the emitting source , and @xmath31 where @xmath32 , and @xmath33 .
note also that @xmath34 and @xmath35 are the solutions to the fokker - planck equation and represent the probability to get a given parton with final momentum @xmath36 , given that the initial configuration is a particle of type @xmath37 and momentum @xmath38 .
information on the initial temperature sensitivity , on the geometry and other details is in ref .
@xcite . a quantitative measure of in - medium modifications
is contained in the so - called @xmath39 profile @xmath40 when plotted as a function of the transverse momentum .
clearly , if a nucleus - nucleus collision is nothing but a superposition of nucleon - nucleon collisions , then @xmath39 should be unity .
this variable is now available for a variety of particles , over a wide range of transverse momenta . for @xmath22 , as a function of transverse momentum .
the full line is the calculation described in this text , and in @xcite , with @xmath41 = 0.34 .
the initial temperature is 370 mev , and is consistent with the study in @xcite .
data are from @xcite .
[ raa],width=264 ] figure [ raa ] reveals new preliminary data , where the behavior of @xmath42 over an impressive range of almost 20 gev / c is shown .
also there , is the result of a calculation with the formalism described here . in this approach ,
the strong coupling constant , @xmath41 , is a free parameter . in his plot , a value of @xmath43 yields a good fit .
this value is kept fixed for the rest of this work .
further note there are uncertainties associated with the determination of @xmath44 , as well as inherent systematic measurement uncertainties .
this combination roughly is of the order of 10% . the apparent agreement ( for @xmath45 gev )
is at least superficially satisfying , even if many questions remain on details of the opacity of the medium , and on the actual sensitivity of this variable on bulk properties .
the answer to those questions is at present scheme - dependent . with data of this quality
however , the next meeting in this series is bound to see progress on these issues
. nevertheless , the flatness of these spectra appears a robust feature .
the softer part of the spectrum is not reproduced by a qcd fragmentation framework and requires additional ingredients possibly related to parton recombination @xcite . with the machinery at hand , the spectrum of photons produced in nucleus - nucleus collisions can be calculated , including those originating for jet - plasma interactions @xcite , but consistently taking account energy - loss systematics
the sources include those active in pp collisions : the direct photons produced by parton compton and annihilation events , and the fragmentation photons produced by bremsstrahlung from final state partons . in aa collisions , the sources above still operate but the fragmenting jets are now subject to energy - loss considerations .
in addition , hard partons traveling in the medium can also produce photons through medium - induced bremsstrahlung .
finally , the conversion of leading partons to photons @xcite was found to be a significant contribution and should be treated consistently with the other channels enumerated here .
the different sources for au + au collisions at rhic are shown in figure [ rhic_phot ] .
the agreement with the recent phenix photon data is very good .
furthermore , it is satisfying to note that the modeling parameters were fixed prior to its release @xcite .
further note that when the jet - photon conversions are omitted , the total photon production is reduced by up to 45% around @xmath20 = 3 gev / c , showing the importance of this process .
the total plasma contribution appears important for @xmath46 6 gev / c . here
also , the quality of this data opens the door to additional investigations .
some have been done @xcite and more will follow , but importantly , electromagnetic signals and hard hadronic probes are no longer disjoint observables .
it is not an overstatement to write that the measurement of electromagnetic probes in relativistic has produced some very exciting results .
all expectations are that this will continue , with larger statistics samples and the measurement of lepton pairs at rhic . with the data shown at this meeting
, we are entering an era of precision measurements and modeling .
for example , measurements of photon azimuthal asymmetry are now at reach .
i am grateful to all of my collaborators for their involvement with any aspect of the work presented here . of those ,
i also thank simon turbide for a critical reading of this manuscript .
this work was supported in part by the natural sciences and engineering research council of canada , and in part by the fonds nature et technologies of the government of quebec .
99 m. a. mazoni _
et al . _ ,
. phys . * a566 * ( 1994 ) 95c ; m. masera _
_ , nucl . phys .
* a590 * ( 1995 ) 93c ; a. l. s. angelis _
j. c * 13 * ( 2000 ) 433 .
d. mikowiec , these proceedings .
e. scomparin , these proceedings s. damjanovic , these proceedings .
a. weldon , phys .
d * 42 * ( 1990 ) 2384 ; charles gale and joseph i. kapusta , nucl . phys . * b357 * ( 1991 ) 65 .
r. rapp and j. wambach , adv .
nucl . phys .
* 25 * ( 2000 ) 1 .
see , for example , g. e. brown and mannque rho , phys .
* 363 * ( 2002 ) 85 .
k. gallmeister _
et al . _ ,
a715 * ( 2003 ) 705 . s. weinberg , phys . rev
. lett . * 18 * ( 1987 ) 507 .
j. i. kapusta and e. v. shuryak , phys .
d * 49 * 46944704 ( 1994 ) .
e. v. shuryak , phys .
b * 78 * ( 1978 ) 150 .
k. kajantie , j. kapusta , l. mclerran , and a. mekjian , phys .
d * 34 * ( 1986 ) 2746 .
s. gao and c. gale , phys .
c * 57 * ( 1998 ) 254 .
see , for example , s. i. dolinsky _
et al . _ ,
* 202 * ( 1991 ) 99 , and references therein .
z. huang , phys .
b * 361 * ( 1995 ) 131 , and private communication .
m. c. abreu _ et al .
j. c * 14 * ( 2000 ) 443 .
g. q. li and c. gale , phys .
* 81 * ( 1998 ) 1572 ; phys .
c * 58 * ( 1998 ) 2914 .
r. rapp and e. v. shuryak , phys .
b * 473 * ( 2000 ) 13 .
k. gallmeister , b. kmpfer , and o. p. pavlenko , phys .
b * 473 * ( 2000 ) 20 .
r. shahoyan , these proceedings .
wang , these proceedings . see , for example , jan - wei qiu , eur .
j. c * 43 * ( 2005 ) 239 .
r. baier , y. l. dokshitzer , a. h. mueller , s. peign , and d. schiff , nucl . phys . *
b484 * ( 1997 ) 291 ; r. baier , y. l. dokshitzer , a. h. mueller , and d. schiff , jhep * 0109 * ( 2001 ) 33 .
e. wang and x .-
wang , phys .
* 89 * ( 2002 ) 162301 .
i. vitev and m. gyulassy , phys .
* 89 * ( 2002 ) 252301 .
a. kovner and u. a. wiedemann , _ qgp3 _ , r. hwa and x .-
n . wang , editors ( world scientific , singapore , 2003 ) .
b. g. zakharov , jetp lett .
* 63 * ( 1996 ) 952 ; _ ibid_. * 65 * ( 1997 ) 615 ; _ ibid_. * 70 * ( 1999 ) 176 .
p. arnold , g. d. moore , and l. g. yaffe , jhep * 0111 * ( 2001 ) 57 ; jhep * 0206 * ( 2002 ) 30
. s. turbide , c. gale , s. jeon , and g. d. moore , phys .
c * 72 * , 014906 ( 2005 ) .
k. j. eskola , v. j. kolhinen , and c. a. salgado , eur . phys .
j c * 9 * , 61 ( 1999 ) .
s. turbide , r. rapp , and c. gale , phys . rev .
c * 69 * , 014903 ( 2004 ) .
b. cole , these proceedings .
d. molnr , these proceedings .
r. j. fries , b. mller , and d. k. srivastava , phys .
lett . * 90 * , 132301 ( 2003 ) .
h. buesching , these proceedings .
d. denterria and d. peressounko , nucl - th/0503003 . | we discuss the status of a subset of penetrating probes in relativistic nuclear collisions . thermal photons and dileptons
are considered , as well as the electromagnetic signature of jets . |
* this is the preliminary version of invited lecture for nato - asi material science , fundamental properties and future electronic applications of high - tc superconductors , albena , bulgaria , september 1998 , kluwer academic publishers , dortrecht , pp .
323 - 360 ( 2001 ) * the interpretation of the normal and superconducting state far infrared region ( fir ) properties of metals is still rather ambiguous . in this lecture
i will restrict myself to the consideration of the of isotropic single wide band ( isb ) model based on the nam formalism @xcite . * from the optical conductivity @xmath8 one derives the so called transport spectral function @xmath9 .
it has the same spectral structure , i.e. the same number of peaks with similar relative positions ( see fig.4 in @xcite and fig.1 in @xcite ) as a standard @xmath0 which enters the eliashberg equations .
but their amplitudes can be lower , and as a result the transport coupling constant @xmath10 is less than the standard @xmath11 . in this lecture
i will not distinct the @xmath0 and @xmath9 . *
if the electronic band is very narrow , the forward scattering could dominate . in this case the optical properties manifests themselves by the negligible coupling , while the tunnelling spectroscopy gives reasonable finite coupling strength .
a similar problem arises when the quasiparticles from the flat or nested parts of the fermi surface are mainly responsible for the superconductivity .
since they have small fermi velocities , their influence in the transport response function which is proportional to @xmath12 is weak .
such a scenario was recently proposed for borocarbides @xcite . * in the frame of the eliashberg theory it is impossible to elucidate the nature of the bosons which are responsible for the superconductivity . in this context
i will use the term boson , keeping for short the standard electron - phonon notation @xmath0 . *
i will restrict myself to the consideration of the local ( london ) limit case , that is , to the normal skin effect scenario .
the generalisation of results obtained to the nonlocal ( pippard ) limit is possible but will not be considered here due to the lack of space . *
the power of the proposed methods is illustrated analysing available experimental data for a single - domain yba@xmath4cu@xmath5o@xmath13 crystal @xcite .
note , that to the best of my knowledge , the experimental htsc superconducting state reflectivity was not been considered from the inverse problem viewpoint at all , despite the single early attempt @xcite .
* theoretical analysis of any physical quantity , as optical conductivity , is performed in frame of some model which includes a theoretical formalism and some assumptions about the band structure , the anisotropy of the coupling function and so on .
the considered here isb model is the standard default . nevertheless , in frame of its _
s-_wave version the question : what is the value of the gap ? `` has an unique answer , but for _ d-_wave case the correct questions are either : ' ' what are the values of the gaps ? `` or ' ' what is the value of the maximum gap ? .
in other words , the absolute values and the physical meaning of the quantities are model dependent . *
if the model is reach enough , one can calculate many physical properties using the same _ set of input ( material ) parameters_. a solution of the inverse problem means the determination of these material parameters from the experimental data .
a short list of the isb model input parameters is given in section [ isbsub ] .
* as a rule , the inverse problem is ill - posed .
the experimental data contain less information , than we would like to know .
the reasons for are the temperature induced broadening and the finite signal to noise ratio s / n . * in this lecture we will mainly be interested in the amount of the remaining information about the spectral function @xmath0 , which can be obtained from the optical data . in spectroscopy the unit of information is a peak .
a single peak needs four parameters for its description : the upper and the low frequency bounds @xmath14 and @xmath15 , the position of the maximum and its amplitude .
an narrow symmetric peak within the broad band needs three parameters : the amplitude , the halfwidth , and the position .
the sum of all peaks yields the upper and lower frequency bounds @xmath14 and @xmath15 of the spectral band .
since the @xmath15 of the @xmath0 is usually equal to zero , the remaining three band parameters of @xmath0 can be determined even from the normal state data @xcite .
the standard isb model @xcite is the most developed part of the modern theory of superconductivity .
it describes _ quantitatively _ the renormalization of the physical properties of metals due to the electron - boson interaction .
the input quantities of the isb model are the density of states at the fermi energy @xmath16 , the fermi velocity @xmath17 , the impurity scattering rate @xmath18 , the coulomb pseudopotential @xmath19 , and the spectral function @xmath0 of the electron - boson interaction .
the scale of transport and optical properties is fixed by the plasma frequency @xmath20 . in this section
i present only the final expressions of the nam theory , more detailed information can be found in @xcite . in order to calculate any physical property in the superconducting state one
needs a solutions of eliashberg equations ( ee ) @xcite . in my opinion ,
the following ee representation is most suitable for numerical solution by iterations @xmath21 \mbox{re}\frac{\tilde{\delta}(y)}{\sqrt{\tilde{\omega}^2(y ) -\tilde{\delta}^2(y ) } } \label{eq1 } \\
im\tilde{\omega}(\omega ) & = & \frac{i\gamma_{imp}}{2}\frac{\tilde{\omega}(\omega ) } { \sqrt { \tilde{\omega}^2(\omega)-\tilde{\delta}(\omega ) } } + \frac{\pi}{2}\int dy \alpha^2f(\omega -y ) , \nonumber \\ & & \left [ \coth{\left(\frac{\omega - y}{2t}\right ) } -\tanh{\left ( \frac{y}{2 t } \right ) } \right ] \mbox{re}\frac{\tilde{\omega}(y)}{\sqrt{\tilde{\omega}^2(y)-\tilde{\delta}^2(y ) } } , \label{imel}\end{aligned}\ ] ] where @xmath22 and @xmath23 are the renormalized gap function and the renormalized frequency respectively , and @xmath18 denotes the impurity scattering rate within the born approximation .
the real and the imaginary parts of the eliashberg functions @xmath22 and @xmath23 are connected by the kramers - kronig relations . hence , they have the same fourier images .
this reasoning yields the fast solution procedure .
the convolution type integrals ( [ eq1]-[imel ] ) should be calculated by the fast fourier transform ( fft ) algorithm .
the inverse _ complex _ fourier transformations of the results obtained give _
complex _ values of @xmath22 and @xmath23 . if one would like to use this efficient fft method also for _ d-_wave pairing symmetry , the process of arithmetic - geometric mean @xcite for the evaluation of the complex elliptic integrals ( see below eqs .
( [ nscm1],[nscm2 ] ) is strongly recommended .
the density of states re@xmath24 and the density of pairs re@xmath25 @xmath26 could be approximated by step functions .
the convolution of any spectrum with the step function sign@xmath27 results in the shift by @xmath28 at @xmath29 and by @xmath30 at @xmath31 and its integration over y. it is important , that due to the presence of singularities in @xmath24 and @xmath25 at @xmath32 the more complicated functions such as @xmath33 and @xmath34 still contain a jump ( now at @xmath35 ) . since the convolution is a quite general property of the green functions approach , one can expect that the first derivatives from a spectrum ( of different nature ) over frequency will reproduce the input spectral function @xmath0 ( may be with the appropriate phase distortion ) . when applied to the optical conductivity
, this approach gives visual accessibility ( va ) procedure discussed in the sec .
[ super ] . in the normal state
the set ( [ imel ] ) is reduced to the following simple formula @xmath36 , \label{nel}\ ] ] which however looks unwieldy in comparison with the same formula presented in matsubara formalism @xmath37 where @xmath38 are kernels of the spectral function , @xmath39 and @xmath40 are fermion ( @xmath41 ) and boson ( @xmath42 ) matsubara energies , respectively , and @xmath43 . the normal state isb optical conductivity takes the following forms @xmath44 and @xmath45 in the real and imaginary axes techniques , respectively . the tedious expression for the superconducting state isb optical conductivity has the form @xcite @xmath46}}{\left(\sqrt{\tilde{\delta}^2(x ) -\tilde{\omega}^2(x)}+\sqrt{\tilde{\delta}^2(x+\omega ) -\tilde{\omega}^2(x+\omega)}\right)^ * } \nonumber \\ & & + \frac{g_{ar}\left\{\tanh{[(x+\omega)/2t]}-\tanh{(x/2t)}\right\ } } { \sqrt{\tilde{\delta}^2(x ) -\tilde{\omega}^2(x)}+\left(\sqrt{\tilde{\delta}^2(x+\omega ) -\tilde{\omega}^2(x+\omega)}\right)^ * } , \label{sigsup } \end{aligned}\ ] ] where @xmath47 are _ coherent factors_.
it is amusing , that the eq .
( [ signm ] ) is solvable for @xmath48 and eq .
( [ ommats ] ) for @xmath49 .
hence , at least theoretically , an exact analytical solution exists .
it takes three or four steps as shown below .
\i ) one has to perform the analytical continuation of the conductivity @xmath8 from the real energy axis , where we are living , to the poles of the bose distribution function ( matsubara boson energies ) @xmath50 .
this procedure is similar to the kramers - kronig analysis , which is nothing other than the analytical continuation from the real frequency axis to itself .
for example , if one fits the data by a sum of drude and lorentz terms or by formula ( [ padx ] ) , one has simply to substitute the complex values @xmath51 for the frequency .
note , that the non - metallic ( ir active direct phonon contribution and interband transition ) part of the dielectric permeability @xmath52 have to be subtracted from the total one @xmath53 before starting of the analysis . in the genuine far infrared region
one could use the real constant @xmath54 , instead of @xmath52 , but if the spectral range is wide , the subtraction of @xmath52 is not trivial .
\ii ) the renormalized frequencies should be calculated from ( [ signm ] ) as follows @xmath55 \iii ) inverting eq .
( [ ommats ] ) , one could evaluate the values of the spectral function @xmath56 we looked for @xmath57 \iv ) using , for example , pade polynoms @xcite one has to continue the electron - boson interaction function back to the real axis . in practice ,
if the signal to noise ratio is not extremely large , the last step makes no sense due to the dramatic increase of the uncertainty in the @xmath58 with elevation of n. the point is that the small quantities @xmath58 eq . ( [ eslam2 ] ) and @xmath59 eq . ( [ esom3 ] ) are the difference of large quantities .
note , that the uncertainty in the @xmath58 values has to be small , not in comparison with @xmath58 itself , but with the difference between @xmath58 and its high energy asymptotic value @xmath60 where @xmath61 if the average boson energy has the order of @xmath62 , the last condition ( @xmath63 ) may be violated even for n=3 , since @xmath64 and @xmath58 have similar values .
for example , at @xmath65100 k for the spectral function , shown by the dotted line in fig .
[ josfig ] , @xmath66=1.37 , 1.12 , 1.06 , 1.03 for n=1 , 2 , 3 , 4 respectively . three or four @xmath58 values only form a too small parameter set to recover correctly the shape of the spectral function . actually , the possibility to obtain at least small amount of reliable information , is the additional advantage provided by the matsubara solution in comparison with the real axis one where _ all _ values are unreliable @xcite . fig .
[ josfig ] shows the good agreement between the experimental normal state 1 - 2 - 3 single crystal reflectivity data @xcite and the calculated ones using for _ various _ spectral functions .
unfortunately , the lack of a unique solution promoted unreasonable speculations on the mechanisms in several novel superconducting systems in connection with proposed hand - made electron - boson interaction functions . in my opinion , twelve years after the discovery of the superconductivity in cuprates it is timely to return to a self - consistent description of the superconductivity in this compounds .
the first matsubara value of the conductivity @xmath67 ) gives the value of the coupling constant @xmath68 , if the plasma frequency @xmath69 and the impurity scattering rate @xmath18 are known _ a priori _ @xmath70 the reflectivity r@xmath71 can also be continued to the imaginary frequency axis and included into the solution of the inverse problem .
following this way , one can derive from ( [ sig1 ] ) the following formula @xmath72 where @xmath73 is the low - frequency value of the non - metallic ( phonon and interband transition ) part of dielectric permeability , and k@xmath74 is the electromagnetic kernel @xmath75 the first @xmath76 and the second @xmath77 allow ( for example ) to estimate the average boson energy and _ very approximately _ the width of the spectral function .
if there is the possibility to combine this incomplete information about @xmath0 with the results of other spectral measurements , for example , tunnelling or neutron data , one can qualify the transport spectral function .
the so called model 2 electron - boson interaction function @xcite was recovered assuming , that the solution has to be resemble the phonon density of states , measured by inelasitc neutron spectroscopy .
in ref.@xcite we shown , that it is useful to describe the optical conductivity in terms of the so called `` extended '' drude formula @xmath78 here the optical relaxation @xmath79 , the optical mass @xmath80 and the optical scattering rate @xmath81 as the complex , real and imaginary parts of an inverse normalised conductivity @xmath82 have been introduced . this presentation is old and popular in spectroscopy , but the practical value of the eq .
( [ eqw ] ) has not been exploited much .
the optical relaxation will be compared with the effective mass @xmath83 and the effective scattering rate @xmath84 , which by definition are the real and imaginary parts of the renormalized frequency @xmath85 when the frequency dependencies of @xmath81 and @xmath86 are plotted , the coupling constant , the average boson energy and the upper boundary of spectral function are visually accessible ( see fig . [ fig1 ] ) .
the appropriate visual accessibility rules can be derived from the properties of the renormalized frequency @xmath87 using the allen formula eq .
( [ allg ] ) @xcite .
since such useful expression was originally obtained in the weak coupling approximation ( @xmath88 ) only , below i rederive it for the general case , including strong coupling . for any @xmath89
the values of @xmath49 are positive and are of the same order of magnitude , except may be @xmath90 .
it means , that in the zero order the series @xmath91 defined by eq .
( [ ommats ] ) is an arithmetical progression .
consequently , according to the gauss rule for the arithmetical series the denominators @xmath92 in ( [ signm ] ) at a given @xmath93 do not depend on @xmath89 , that is , they coincide . in view of eq .
( [ eqw ] ) , eq .
( [ signm ] ) takes the form @xmath94 hence we can expand the denominator in ( [ signm ] ) @xmath95}\ ] ] to the first order in power of @xmath96 .
after some simple algebra we arrive at the sought - for allen formula @xmath97 the real axis versions of eq .
( [ allg ] ) at @xmath650 look similar @xmath98 and justify the following simple description of the interaction of light with the conducting subsystem of the normal state metal .
when a photon with an energy @xmath99 has been absorbed , _ two _ excited _ virtual _ quasiparticles , an electron `` and a ' ' hole `` are created .
if the first particle s frequency is @xmath100 , the second ones frequency should be @xmath101 . the exited ' ' electron `` and ' ' hole relax to the fermi level according to the quasiparticle laws ( see eqs .
[ nel],[ommats ] ) . since @xmath100 can varies from 0 to @xmath99 , the optical relaxation w(@xmath99 ) is the frequency - averaged ( @xmath102 ) , double ( electron+hole @xmath103 factor 2 ) renormalized frequency @xmath87 . at @xmath104
the isb effective scattering rate @xmath105 @xmath106 has a clear physical meaning .
the exited quasiparticles with the energy @xmath99 can relax by the emission of virtual bosons with the energy varying the range from 0 to @xmath99 or by the impurity scattering . at finite temperature the allen formula
@xmath107 and the effective scattering rate ( [ nel ] ) have exactly the same interpretation , since the only difference between them and eqs .
( [ maindt0],[maindt0],[goptt0],[gameft0 ] ) is the presence of the fermi distributions instead of step functions at @xmath108=0 .
figure [ fig1 ] shows the frequency dependencies of the normal state optical and effective masses and scattering rates calculated using the model spectral function @xcite with @xmath109=2 .
the results are plotted for the low temperature @xmath108=1 k and for @xmath108=100 k which is close to @xmath110 and has the order of @xmath111 , where @xmath112=330 @xmath7 is the averaged boson frequency .
both @xmath113 and @xmath114 show the characteristic features corresponding the peculiarities of the phonon spectrum .
when the frequency @xmath99 exceeds the upper energy bound of the boson spectrum @xmath14 , the quasiparticle becomes undressed .
its mass @xmath115 and effective scattering rate @xmath116 , but of course , the absolute value of @xmath117 could be huge .
the optical mass @xmath118 and scattering rate @xmath119 have no features at @xmath120 due to the energy averaging ( [ maindt0],[moptt0],[goptt0],[maind ] ) .
one can only _ calculate _ the approximate value of @xmath14 as the frequency where the @xmath121 and @xmath119 reach the halves of their maximum magnitudes .
[ fig1 ] shows , that this my rule can not be applied even to the calculated frequency dependence of the optical scattering rate @xmath119 , the more to the experimental curve , where _ the high energy _
behaviour is distorted by the interband transition contribution to the dielectric permeability .
in contrast the frequency dependence of the optical mass @xmath118 is suitable for the application of the simple criterion @xmath122 moreover its _ low energy _ value @xmath123 itself contains a useful piece of information .
[ fig1]a shows , that the approximate values of the transport coupling constant @xmath124 and the upper frequency bound of the electron - boson coupling function @xmath125 $ ] are visually accessible .
if the spectral function is wide and its peak position ( or averaged frequency ) coincide roughly with the @xmath14/2 , the optical relaxation w(@xmath99 ) is smooth and the elucidation of @xmath112 at high temperature is impossible . when an exotic superconductor with the narrow , @xmath126-function like spectral function will be discovered , following this way one can easily construct the recipe for the evaluation of the strength and the position of the single peak .
the pade polynom analytical continuation @xcite is a capricious and not completely correct procedure . nevertheless , if the temperature is high enough in comparison with the energy under consideration it works splendidly .
the numerical experiment has given to my great surprise the following ( and to some extent obvious ) result ( see fig . [ 4p ] ) .
the good agreement between the real axis calculations and the matsubara ones takes place every time when the approximation using 1000 @xmath127 points gives the same result as the _ four _ points pade approximation .
this remarkable pade polynom with only four parameters @xmath128 , @xmath129 , @xmath130 and @xmath131 @xmath132 fits well the majority of the experimental data at @xmath133 and itself is the demonstration of the result obtained in section [ exact ] : the few first matsubara values @xmath134 are responsible for the frequency dependence of @xmath8 .
the pade polynom ( [ padx ] ) is a formula of merit . at first , it has the correct analytical properties .
its two poles are located in the lower half - plane of the complex energy .
the high ( @xmath135 ) and low ( @xmath136 ) energy asymptotic behaviour are reasonable @xcite . at second
, it is ideal for the least square fits of experimental r(@xmath99 ) @xcite , as well as for simple analytical estimations . having the exact solution ( see sec.[exact ] ) we have to substitute the complex values @xmath51 for energy , perform easily the analytical continuation to the imaginary axis and analyse the values of @xmath137 . in the way we obtain the interpretation of a , b , c. another possibility is a fast and easy analysis of the data in terms of the optical mass @xmath80 and the optical scattering rate @xmath81 @xmath138 at third , if the low - frequency reflectivity r(@xmath99 ) measured at high temperature can not be fitted by eq .
( [ padx ] ) or the obtained value of @xmath90 contradicts the generally accepted values , it might indicate , that the material under consideration is _ not _ a standard isotropic single wide band ( isb ) metal and its properties have to be analysed if terms of _ another , _ unconventional model .
in order to illustrate the utility of the this approach , let us consider another superconducting compound with relatively high transition temperature @xmath139 @xcite .
since the ratio of the experimental reflectivities in the superconducting and normal states @xmath140 above 100 @xmath7 @xcite , in frame of the isb model it means , that the high energy bound of the electron - boson interaction function @xmath141 100 @xmath7 .
the low energy part of the normal state reflectivity r(@xmath99 ) , measured at t=40 k , was fitted ( with the adopted fixed @xmath69=10000 @xmath7 ) by ( [ padx ] ) ( see fig . [
4p]b ) , @xmath80 is shown in the inset .
one can see , that the optical mass at small @xmath99 is huge itself and gives @xmath142 .
this values points to heavy fermion or polaron models , rather than to the generally accepted weak or intermediate coupling scenario . on the other hand , the analysis of the normal state specific heat data gives an extremely small value of the coupling constant . as a result we concluded that rb@xmath5c@xmath143 is not a isb metal and its properties should be treated in terms of a more complicated model which takes into account the self - energy and conductivity vertex corrections , the small value of the bandwidth @xmath144100 mev , possible influence of the electron - electron correlation , the strong anisotropy of the coupling function and the fermi velocity , and etc . and the last but not least , if the data can be fitted by ( [ padx ] ) , that is by the formula with four parameters , it is impossible to recover from this _ four _ parameters the detailed shape of the electron - boson interaction function @xmath0 . in conclusion
, it is naturally to define the amount of the available information by the number of poles and zeros of the function which fits well the experimental curve or the number of the determinate matsubara values . in frame of the isb model we assume , that all quasiparticles have the same fermi velocity , but in real metals there is some anisotropy of @xmath17 .
if the impurity scattering rate @xmath18 is small in comparison with the complete optical scattering rate @xmath145 and the electron - boson coupling is isotropic , we can use the average value @xmath146 and keep the isb model . on the other hand ,
if the impurity scattering , given by the mean free path @xmath147 , dominates , the quasiparticles from different parts of the fermi surface will have different scattering rates @xmath148 .
let us accept for simplicity , that we can divide the fermi surface between two parts with the approximately uniform properties .
the optical conductivity of this two - band system is the sum of two drude terms @xmath149 one can see , that eq .
( [ 2drude ] ) and eq .
( [ padx ] ) are equivalent and four parameters @xmath150,@xmath151 , @xmath152 , @xmath153 in ( [ 2drude ] ) connected with @xmath154 , @xmath129 , @xmath130 , @xmath131 in ( [ padx ] ) as follows @xmath155 it means , that even if there is no electron - boson interaction in the two - band system , the frequency dependence of its optical conductivity ( [ 2drude ] ) will reproduce a isb model one ( [ signr ] ) with some frequency dependent mass `` and ' ' scattering rate .
the multiband ersatz of the electron - boson effects could be revealed by the investigation of the temperature dependence of @xmath80 and @xmath81 .
it originates mainly from the presence of the bose factor in eqs.([nel ] ) ( @xmath156 sampling in ( [ ommats ] ) ) and to a less degree from the fermi distributions in eqs .
( [ nel]-[signm ] ) . in subsection [ omasgam ]
the low ( @xmath157 ) temperature behaviour of @xmath158 and @xmath119 was discussed in detail .
we found , that at a given quasipartical frequency @xmath99 in accord with the frequency conservation principle only virtual bosons with @xmath159 can contribute to the relaxation . at high temperature
the bosons lose their individuality . at @xmath160 the expression ( [ nel ] ) can be substantially simplified , since the bose factor @xmath161 dominates in comparison with the fermi one @xmath162 $ ] . keeping the leading term @xmath163 we arrive at @xmath164 in matsubara technique the same result can be obtained even more easily . at @xmath160 all kernels of the spectral function @xmath165 , defined by eq .
( [ aaa2 ] ) , vanish as @xmath166 for all k , except k=0 .
therefore , the eq .
( [ ommats ] ) for the renormalized frequency becomes trivial @xmath167 if one substitutes the obtained @xmath23 in eq .
( [ signr ] ) or @xmath168 into eq .
( [ signm ] ) after some algebra one arrives at the drude type expression for the optical conductivity @xmath169 note , that the relation @xmath170 is correct , if and only if the self energy @xmath171 does not depend on frequency .
otherwise , the relation @xmath172 does not satisfy the kramers - kronig ( k .- k . )
relations , that is , it violates the causality .
the proof is very simple . due to the charge conservation law , when the photon has been absorbed , _ two _ quasiparticles , the electron `` and ' ' the hole
come into being .
the optical conductivity is a bosonic function defined by its values at @xmath42 , while the quasiparticle self energy @xmath173 is a fermionic function defined by the values at @xmath41 . as a result ,
the right and left hand sides of eq .
( [ varma ] ) have different statistics , what is very strange . in frame of the isb model considered here
the proof is a little bit more complicated .
the optical conductivity itself and its inverse @xmath174 are response functions .
it means , that they have no poles on the upper half - plane of the complex frequency or , in other words , their real and imaginary parts are connected by kramers - kronig relations .
the same is valid for the self energy . from eq .
( [ varma ] ) after k .- k .
transformation one obtains @xmath175 the optical and effective masses have the same asymptotic value 1+@xmath109 at @xmath176 .
due to the presence of the factor of 2 in ( [ varmam ] ) it would mean , that 1=2 . let us turn to the temperature dependence of the optical properties of isb metals , which are shown in fig .
[ gammasott ] .
all curves were calculated using model 2 spectral function and @xmath177 .
panels [ gammasott]a presents the frequency dependence of the optical scattering rate @xmath81 ( solid lines ) for t=100 k , 200 k , 300k .
the corresponding optical masses are shown in panel [ gammasott]b .
one can see , that the absolute value of @xmath81 increases with t , but the frequency dependent component decreases .
it leads to diminishing of @xmath80 when t is rising .
the temperature dependence of the minimum ( @xmath178,solid line ) and maximum ( @xmath179,solid line ) values of @xmath119 are shown in panel [ gammasott]d in comparison with their common asymptote @xmath180 ( dotted line ) . the corresponding function @xmath181 , named sometimes @xmath182 is presented in panel 5c .
the position of maximum of the small hump at low t is approximately equal to @xmath183 , where @xmath15 is the frequency position of the lowest boson peak in @xmath0 . at the same temperature the dc resistivity @xmath184 starts the linear rise . in subsection [ omasgam ]
the simple rule for determination of the _ upper _ frequency bound of the spectral function @xmath14 was proposed . at @xmath120 the optical mass without unity
@xmath185 is approximately the half of its value at @xmath178 .
the temperature dependence of the quantity @xmath186 defined by @xmath187= m^*_{opt}(\omega=0,t)/m_b-1\ ] ] is shown by solid line on inset in panel [ gammasott]c in comparison with the phenomenological line @xmath188 .
one can see , that @xmath189 the normalised optical scattering @xmath190 is shown in fig([gammasott]a ) ( dotted lines ) . in our opinion
@xcite this quantity is far useless in comparison with @xmath119 and @xmath80 . with the exception of the low frequency region where its value is constant , @xmath191 is a quasilinear function @xmath192 up to max@xmath193 .
even the simple analysis by the favourite polynom eq .
( [ padx ] ) allows to determine the four parameters , while @xmath191 depends on two .
it is interesting , that the gilbert transformation from @xmath191 is not equal to 0 . in this connection
we note that the use of the bsc normal state conductivity expression @xmath194 is correct , only if @xmath195 . here
@xmath196 is the frequency - independent renormalised plasma frequency .
note , that according to the allen formula ( [ maind ] ) , the optical relaxation is approximately _ linear _ with respect to the spectral function @xmath0 .
it means , that the conversion from the low temperature scenario to the asymptotic high temperature one takes place for each boson with frequency @xmath197 separately at @xmath198 , that is , when @xmath197 becomes of the order of the _ first _ matsubara energy @xmath199 . since the isotropic single wide band model is the subject of the present lecture , discussing the properties of two - band systems we have to pay special attention to possible artefact results .
let us start our consideration from the temperature independent two - drude terms simulator ( [ 2drude ] ) .
since it is equivalent to the favourite polynom ( [ padx ] ) , we can extract the information about spectral function `` including the temperature dependence of the coupling constant @xmath200 .
it is known , that if the material parameters are temperature independent , the optical scattering rate @xmath81 is a monotonically increasing function . since the simulator s @xmath119 does not changes , when t is rising , one may conclude , that the coupling ' ' constant
@xmath200 is a monotonically decreasing function with the high temperature asymptote 1/t . on the other hand ,
the analysis of the optical mass , which is usually a monotonically decreasing function , gives the opposite result : @xmath200 dramatically increases when temperature is growing .
the study of the reflectivity @xmath201 gives the third fitted function @xmath200 .
the reasonable way to resolve this conflict is the replacement of the model .
if one restricts oneself to the consideration of the scattering rate only , the contradiction can be overlooked .
the fact is that the strong temperature dependence of material parameters exists , for example , in heavy fermion systems .
another interesting scenario was realised in fullerides , where the dc resistivity ( measured at the constant pressure ) changes with temperature as t@xmath202 and the same dc resistivity ( measured at the constant volume ) follows conventional linear law @xmath203 , as it should be , since the theory assumes the v = const condition .
the second example is the clean limit of anisotropic systems . as usually , one can divide the fermi surface between two ( for simplicity ) parts with the approximately uniform properties .
this disjoined representation of the fermi surface is known as the multiband model .
we accept also , that the intra- and interband interactions are governed by the electron - boson spectral functions having the same model 2 shape , but different coupling constants @xmath204 .
the scales of transport and optical properties of the bands are fixed by the plasma frequencies @xmath205 . in the normal state
is possible to diagonalise the equations by the substitution @xmath206 and @xmath207 .
the resulting two - band optical conductivity is the sum of two isb terms ( [ signr ] ) . for the model calculations the arbitrary values @xmath208 , @xmath209 and @xmath210=2 was chosen . the frequency and temperature dependencies of the two - band optical mass , scattering rate , reflectivity and dc resistivity
are shown by solid lines in fig .
( [ twoband])a , b , c , d respectively .
the _ by - eye - analysis _ can not reveal the difference between clean limit two - band optical properties and the standard isb ones .
fortunately , the isb model can pose several possibilities for a quantitative study .
i simply fit the two - band curves , shown on panels ( [ twoband])a , b , c , by the appropriate isb ones ( dotted lines ) , calculated using the same model 2 spectral function .
the coupling constant @xmath109 was used as fitting parameter .
one can see , that the temperature dependencies of @xmath80 , @xmath81 and @xmath201 of the anisotropic system are yet weaker , that it should be according to the isb rules .
the fitted functions @xmath200 grow or drop , when t is rising , depending on which quantity has been chosen for the fit .
for the few metals , where the mutual anisotropy of @xmath211 and of @xmath212 was investigated , the following regularity was found .
the quasiparticles with smaller @xmath212 have larger @xmath211 .
usually this information is unavailable .
one can estimate the anisotropy , if one compares the material parameters obtained from the optical spectra with the ones revealed by the analysis of thermodynamic , or point - contact tunnelling data .
the scale of the optical conductivity is fixed by @xmath213 .
the normal state specific heat data are proportional to @xmath214 .
the point - contact tunnelling spectra are weighted by the factor @xmath215 . as a result ,
if one treats optical spectra of anisotropic metals in terms of the isb the model , one reveals the electron - boson coupling function @xmath216 , which distinguish itself by smaller value of the coupling function @xmath217 in comparison with lambdas , obtained by other physical methods .
this difference is usually assigned to the kinetic coefficient ( 1-cos ) @xcite or , in another language , to the vertex corrections to the quasipartical green function@xcite .
it is interesting , that if the system under consideration is completely isotropic , and @xmath218 do not depend on * k * and * k*@xmath219 , the vertex corrections vanish as soon as considered in this subsection usual anisotropy effects . what kind of corrections are more important in each particular case , first order vertex ones or discussed here zero order distortions caused by the interplay of @xmath220 and @xmath211 , it is the question _ for _ the gallop band structure calculations . in conclusion ,
the temperature dependence of the optical properties of isb metals really exists and can be well quantitatively described .
it originates mainly from the presence of the bose factor in eqs.([nel ] ) ( @xmath156 sampling in ( [ ommats ] ) ) and to a lesser extent from the presence of the fermi distributions in eqs.([nel]-[signm ] ) .
the principal deviation from the predicted frequency and temperature behaviour manifests that the chosen isb model is not complete and has to be replaced .
the formal textbook relation between the conductivity @xmath8 and permeability @xmath53 @xmath222 is not suitable for a practical use , since the entering eq .
( [ idiotsigma ] ) @xmath8 reflects the total response of all excitations . the simplest reasonable decomposition of ( [ idiotsigma ] ) , valid over the spectral region up to frequencies just above plasma edge , is @xmath223 where @xmath224 and @xmath225 are the contributions of interband transition(s ) and direct ir active phonons centred below plasma edge .
the response of high energy excitations and wide interbands , overlapping the plasma edge is presented by the complex constant @xmath221 .
@xmath8 is _ the quasiparticle _
conductivity , defined by eqs.([signr],[sigsup ] ) .
the direct phonon band contribution will not be considered , since it is usually weak and plays some role only in the selected compounds containing light elements and having small value of @xmath69 , for example , in doped fullerenes .
the used here model interband contribution @xmath226 is the simple lorentzian with @xmath227=16000 @xmath7 , @xmath228=4000 @xmath7 , and @xmath229=6000 @xmath7 .
the overlooking of the interband contribution was simulated as follows .
the sum of two terms given by eqs .
( [ epsinter ] ) and ( [ signr ] ) was treated as the quasiparticle response in terms of the optical mass and scattering rate @xmath230 for simplicity i set @xmath231 and following to the results are shown in figs .
[ inter]a , [ inter]b and [ inter]c by solid lines .
the true
masses and scattering rates @xmath232 and the _ chi - by - eye @xcite _ fit by the formula ( [ padx ] ) are plotted in figs .
[ inter]a , [ inter]b and [ inter]c by dotted and dashed lines respectively .
note , that due to the frequency averaging , the quasipartical optical conductivity ( [ signr],[sigsup ] ) is a slowly varying function in the mid - infrared region .
in contrast , the interband permeability near transition edge ( here , near the center of the lorentzian ) can changes rapidly .
the second important feature of @xmath233 is its very weak temperature dependence .
the mixture of the standard drude term ( @xmath234=1000 @xmath7 ) with the lorentzian ( [ epsinter ] ) is shown in panel [ inter]a .
this simple artefact can be easily recovered , since it is temperature independent and the optical mass changes too quickly .
the second and third examples are the half - by - half blends of the model 2 isb conductivities ( t=100 k , panel [ inter]b;t=300 k , panel [ inter]c ) with the same @xmath233 eq .
( [ epsinter ] ) .
it is more complicated case , since the nonmonotonic behaviour could be screened by forthcoming transitions . in this case
detailed quantitative analysis of both the frequency and the temperature dependenc_ies _ is needed .
one has to pay the special attention to the temperature dependencies of @xmath235 , @xmath186 and @xmath236 .
if the interband transitions are wide spectral bands without unqualified features , their contribution to the permeability is usually included in @xmath221 .
the effects caused by the underestimation of its real part was considered by us in @xcite . here
we study the problems produced by the use of eq .
[ idiotsigma ] and by the underestimation of the imaginary part of @xmath221 . since the last quantity can be approximated by the simple constant only in the limited frequency range , for the sake of the analytical correct the model function
was chosen in the form @xmath237
\mbox{\ \ \ \ at $ \omega>0$}\nonumber , \\
im(\delta\epsilon_{\infty}(-\omega ) ) & = & - im(\delta\epsilon_{\infty}(\omega ) ) \nonumber , \\
im(\delta\epsilon_{\infty}(0 ) ) & = & 0 , \nonumber \\ re(\delta\epsilon_{\infty}(\omega ) ) & = & 2.5 -\frac{2\omega}{\pi}\int_{0}^{\infty } \frac{dxim(\delta\epsilon_{\infty}(x))}{x^2-\omega^2 } .\end{aligned}\ ] ] in the mid infrared @xmath238 .
the model frequency dependencies of the optical masses and scattering rates @xmath239 are shown in panel [ inter]d for @xmath240 ( solid lines ) , @xmath241 ( dashed - dotted lines ) , and @xmath242 ( dashed lines ) .
the unperturbed @xmath81 and @xmath80 ( model 2 `` spectral function , @xmath109=2 , t=300 k ) are plotted by dotted lines for comparison .
as usually , the masses do not lie , and manifest that the ' ' unusual frequency dependence of the scattering rate is no more than the artefact of the chosen interpretation procedure .
the early band structure calculations @xcite predicted , that the interband contribution into the in - plane htsc permeability is a real constant between 10 and 15 at @xmath136 , and becomes complex for @xmath243 . in the vicinity of the plasma edge
its value is 4 + 2i .
this information was used by us for the successful description of the _ normal state _ optical properties of htsc materials both in fir and mir spectral regions @xcite . in conclusion ,
the overlooked interband transitions produces the unusual frequency dependence of the optical relaxation , which can be nevertheless analysed properly within the isb model .
let us rewrite the valid at t=0 eqs.([goptt0],[gameft0 ] ) in the differential form @xmath244 at a glance both eqs .
( [ 1prg ] ) and ( [ 2prg ] ) look nice and simple .
but if one attempts to apply them to an experimental data , the distinction becomes evident .
the point is that the experimental data are known only approximately . a finite value of the signal to noise ratio ( s / n ) is the reason why the eq .
( [ 2prg ] ) can not be applied to the fir data .
typical value of the reflectivity s / n do not exceed 50 , the multi - reflection techniques @xcite allow to reach s / n=2000 .
let us adopt for simplicity , that @xmath245 is sampled in equidistant points @xmath246 ,
@xmath247 and all values @xmath248 have the same errors @xmath249 note , that in the normal state at @xmath104 ( when the application of ( [ 1prg ] ) and ( [ 2prg ] ) is correct ) both @xmath245 and its first derivative @xmath250 are monotonically increasing functions , if a metal under consideration is the isb one .
the difference @xmath251 has the order of @xmath252 .
its error is the @xmath253 . as a result
the first derivative @xmath250 has the signal to noise ratio @xmath254 reduced by the factor @xmath255 in comparison with the original ( s / n ) in @xmath256 .
similar the second derivative @xmath257 has the signal to noise ratio @xmath258 reduced by the factor @xmath259 . in subsection [ favor ]
the four parameters formula ( [ padx ] ) was promoted as the universal fitting tool .
@xmath260 means @xmath261 . in another words ,
the normal state @xmath119 has to have the original signal to noise ratio ( s / n)=100 , in order to define _ four _ points in its second derivative @xmath262 with accuracy of 30% . as for k@xmath263
, this quantity grows reasonably .
of course , one can not measure the effective scattering rate @xmath114 and use the formula ( [ 1prg ] ) , nevertheless the title question is worth - while .
the point is that the first derivatives from the superconducting state optical scattering rate @xmath264 and the reflectivity @xmath265 reproduce the input spectral function @xmath0@xcite .
the well structured spectral function of lead obtained from the analysis of the first derivative of @xmath266 looks convincing ( at least for me ) due to the reported value of s / n=6000 .
analysing eq .
( [ 2prg ] ) we arrive at the same conclusions as discussed above .
the normal state optical conductivity provides us with restricted information similarly as other physical methods .
the only way to extract the all required material parameters values is the analysis of different measurements in frame of the _ same _ model .
let us consider from a general point of view the problems , which one has to take in mind , when one try to reconstruct a spectral function @xmath0 from experimental reflectivity data .
this analysis should be performed in frame of a model which is based on an appropriate formalism .
thus , it means , that the results obtained are model dependent .
the ideal , but rare situation occur , when the exact analytical solution of the inverse problem is available @xmath267 where @xmath268 is a known function , which could depend on @xmath0 . in the last case
the numerical solution can be found by iterations .
but what should we do , if such a function @xmath268 is not available by reason of the complexity of the direct formalism . in this case
we shell use the numerical _ adaptive strategy _
, based on the substitution of equations of the direct formalism by the _ adaptive _
formula(s ) , suitable for the numerical solution of ill - posed problems . since it is a strategy , let us consider how does the adaptivity work
. generally speaking , one can use _ any _ adaptive formula .
for example , one can consider the @xmath0 as an input function and the reflectivity as an output function for some formalism
. then one could write @xmath269 or in the fourier space @xmath270 where @xmath271 is the fourier image of the function @xmath268 and k(x ) is an for the time being unknown kernel .
k(x ) could be determined as follows .
* at first , one has to substitute a _ trial _
spectral function @xmath272 into the equations of the main formalism ( eliashberg equations + nam - rainer formula ) and obtains a _ trial _
reflectivity @xmath273 .
* next , one should substitute @xmath272 and the _ calculated _ @xmath273 into the adaptive formula ( [ furad ] ) and solve eq .
( [ furad ] ) for k@xmath274 .
the subscript 1 means the first iteration . *
finally , one has to substitute the _ now known _
kernel k@xmath274 and the _ experimental _
r@xmath275 into the same adaptive formula ( [ furad ] ) , and solve it for @xmath276 .
so , on each iteration we use the model formalism ones and the adaptive formula twice . at first , one have to calculate @xmath277 using @xmath278 , as input function . at second
, one should find the adaptive function(s ) @xmath279 and finally calculate the spectral function @xmath280 the presented trick is quite general .
one could replace analogously , for example , @xmath281 in ( [ ad0]-[ad2 ] ) by the tunnelling density of states @xmath24 or the arpes spectral function @xmath282 , and obtain adaptive methods for the tunnelling and electronic spectroscopies respectively , which is expected can work , with any formalism . as mentioned above ,
the experimental data are known only approximately , therefore the inverse problem is ill - posed . up to now the solution of ill - conditioned problems is the art more than the routine procedure . since one can use any adaptive formula(s ) , it will be pragmatic for him / her / ai to choose it from the short ( alas ! ) list of well investigated by numerical mathematics equations ( see , for example , sec .
18.4 - 18.7 in @xcite and references within ) . on this
understanding the discussed above convolution adaptive formula is one of the best possible selections , since the fourier transformation is linear and unitary .
the inverse problem for the superconducting state optical absorptivity of pb was successfully solved by b. farnworth and t. timusk @xcite within the weak coupling isb model , based on the theory proposed by p.b .
allen @xcite .
the difference between weak coupling theoretical calculations and experimental data @xcite was analysed in detail in @xcite .
it was pointed , that : * the first derivative from the superconducting state optical scattering rate contains the component , proportional to the input spectral function @xmath0 due to the influence of coherent factors ( [ cohfac ] ) . *
the weight @xmath283 of this component is strongly underestimated . *
if one substitutes the solutions of ( [ imel ] ) into the dominators of ( [ sigsup ] ) and keeps bsc values @xmath284 and @xmath285 for the coherent factors ( [ cohfac ] ) , one obtains the optical relaxation , close to the one , given by the weak coupling allen theory . below we will see that isb model based on the nam formalism have qualitatively the same properties .
there are several reasons , why the following convolution adaptive formula @xmath286 is suitable . at first
, we segregate the ill - posed part of the inverse _
s-_wave problem ( taking of the first derivative ) into the separate step . at second ,
the low and high frequency values of @xmath287 coincide ( = 0 ) , it is important for the fft algorithm . at third , the obtained kernel @xmath288 reminds us the instrumental resolution function of a badly adjusted spectrometer , allowing to profit by the applied spectroscopy experience .
if one thinks in terms of the transfer functions , laplace transformation adaptive formula could be more suitable .
the equation ( [ swaf ] ) was solved with respect to @xmath288 using optical scattering rates _ calculated _ in strong ( solid line , fig .
[ svertka]b ) and weak ( dashed line , fig .
[ svertka]b ) coupling approximations and input model 2 @xmath0 .
the results are plotted by the same type lines in fig .
[ svertka]a .
no doubt , all features of the kernel are important for the recovery of the precise peak amplitudes and energy positions .
nevertheless , the general shape and main physics of @xmath289 are defined by the _ master _
peak , located just above @xmath35 .
the arrow , pointing in fig .
[ svertka]a the frequency position of @xmath290 , is located within this peak .
the master peak income is illustrated by dotted line in fig .
[ svertka]b .
it was calculated as follows . *
all collateral maxima and minima of the shown by solid line @xmath288 were rejected and artificial kernel @xmath291 was set to zero anywhere besides the master peak frequency region , where it coincides with @xmath288 .
* @xmath291 was convoluted with the input spectral function @xmath0 and integrated over @xmath99 * since the discussed phenomenon is governed by the coherence factors , the resulting scattering rate
( dotted line , fig .
[ svertka]b ) is labelled as _
coherent relaxation_. * the simple frequency dependence of the coherent relaxation , reminding us the properties of the quasiparticle effective scattering rate @xmath114 , allow to enunciate the visual accessibility procedure , described in forthcoming section .
the normal state kernel ( multiplied on 3 ) is shown by the dotted line on fig .
[ svertka]a for comparison .
the pessimistic conclusion , made in the previous chapter , can be couched in current terms : the ( induced by the frequency and temperature averaging ) width of @xmath288 is too large to resolve the fine structure of the spectral function . the domination of the master peak makes the choice of the trial @xmath0 on the first iteration easy
. it can be the normalised @xmath292 or @xmath293 .
note , that the master peak is asymmetric .
the final frequency resolution is defined by the small value of its left half - width .
the self - consistent determination of the value of the superconducting gap is theoretically possible in frame of adaptive approach , but strongly not recommended . following this way one transforms the weakly nonlinear task to the strongly nonlinear problem . for the case of ill - conditioned equations
it is the wrong way .
the value of @xmath28 has to be known _
a priori_. the related issue is the difference between @xmath0 and @xmath9 .
substituted to the eliashberg equations transport spectral function @xmath9 lands us in difficulty , since the obtained value of the gap can be lower than the known @xmath28 . in my opinion , for the sake of simplicity in this case one might sacrifice the unknown ( from the first principle calculations ) coulomb pseudopotential and use @xmath294 as a simple fitting parameter . in conclusion ,
the _ adaptive _ method of the _ exact numerical inversion _ of the s - wave superconducting state nam equations is presented . in my opinion ,
one day the fir spectroscopy will provide not an alternative to the tunnelling , but a powerful independent tool for the determination of @xmath0 in _ s-_wave superconductors .
the frequency dependence of the superconducting state isb reflectivity @xmath295 in the local limit has the distinctive forms .
this features are determined by the physical properties of the optical scattering rate @xmath296 . in this section
we consider our simplified approach ( visual accessibility ) which allow us to analyse by eye the experimental reflectivity curve and get qualitatively the same results as the sophisticated method , described in the previous section , able to obtain quantitatively .
as far as the physics is determined by the @xmath119 , but the measured quantity is @xmath281 , let us derive a simple relation between them . in view of a high value of the permeability @xmath297 one can expand the formula @xmath298 in powers of @xmath299 . in view of eq .
( [ eqw ] ) , substituting @xmath300 instead of @xmath301 , one arrives the following approximate relation @xmath302 valid at @xmath303 both in the normal ( @xmath304=0 ) and superconducting ( @xmath304=@xmath28 ) states , except for small frequencies @xmath305 , where hagen - rubens approximation is suitable . fig .
[ facefig]a shows the frequency dependence of the superconducting state reflectivity @xmath295 ( t=10 k , solid line ) in comparison with the normal state reflectivity @xmath306 ( t=100k , dashed line ) .
both curves have been calculated using the model 2 spectral function .
the frequency dependencies of the optical scattering rates @xmath296 and @xmath307 are shown in the inset .
one observes that @xmath281 looks like the mirror image of the appropriate @xmath119 , reproducing all important features .
let us summarise our knowledge about the _
s-_wave superconducting state optical relaxation rate and the local ( london ) limit reflectivity ( see fig .
[ facefig ] ) .
the following features are visually accessible .
* the doubled gap value 2@xmath28 coincides with the absorption edge . if the @xmath308 , the reflectivity r(@xmath99)=1 and @xmath296=0 . *
the impurity scattering gives a sharp decrease of the reflectivity just above 2@xmath28 .
the magnitude of the `` jump '' is proportional to @xmath18 .
@xmath296 jumps approaching @xmath18 from above . * in superconducting state the coherent relaxation dominates .
hence , the first derivatives from @xmath309 and from r(@xmath99 ) reproduce the shape of the spectral function ( see fig . [
facefig]b ) .
* since the frequency dependence of the coherent relaxation rate grows faster , than the frequency averaged normal state one , and is fulfilled its maximum value at the frequency @xmath310 , at this point the ratio of the @xmath311 has a maximum , while the reflectivity ratio @xmath312 ) exhibits a minimum . in this section we consider briefly , how the procedure described in section [ vac ] should be corrected in order to make the main features of the _ d-_wave isb reflectivity visually accessible .
formally speaking , the frequency dependencies of the isotropic and the anisotropic parts of the spectral function are not obliged to coincide . at the same time , if the shapes differ substantially , the density of states ( [ nscm1 ] ) looks strange .
so , for the sake of tunnelling , the model spectral function was chosen in the simple form @xmath314 if one substitutes the function ( [ alfd ] ) into the @xmath315 isb eliashberg equations , one finds that at @xmath316 the _ d-_wave component entirely dominates .
hence , instead of the sophisticated @xmath315 approach , we can use for our analysis the pure _ d-_wave model described in detail in @xcite . for the sake of simplicity
we restrict ourselves here to the consideration of the clean limit @xmath177 and set the coulomb pseudopotential @xmath317 to zero .
the value of the parameter g was chosen little a bit larger than unity , since the normal state properties ( defined in frame of _ this _ model by @xmath318 ) manifest the strong renormalization due to the phonon - electron interaction .
it is interesting , that it is possible after minor rehash to use the available _ s-_wave computational programs for the strong coupling _ d-_wave calculations .
the clean limit version of the separable _ d-_wave model keeps the convolution structure of the set ( [ imel ] ) .
one has to substitute the angle averaged density of states @xmath319 instead of @xmath24 ( [ notom ] ) in ( [ imel ] ) , and the angle averaged density of pairs @xmath320 instead of @xmath25 ( [ dotom ] ) in ( [ imel ] ) .
similarly , the conductivity can be calculated using the nam formula ( [ sigsup ] ) . for this purpose one should substitute @xmath321 instead of the @xmath322 into the element of integration in ( [ sigsup ] ) and averages the obtained @xmath323 over the angle @xmath324 .
[ fig3]a shows the model reflectivity r@xmath325 , calculated in the frame of the separable _
d-_wave model ( dashed line ) in comparison with the experimental data @xcite ( solid line ) .
the spectral function @xmath326 is shown in the inset in fig .
[ fig3](a ) , its coupling constant @xmath109 was chosen to reproduce the value of the _ tunnelling _ gap @xmath327=200 @xmath7 derived from the tunnelling spectra @xcite .
note , that it was the same model 2 @xcite curve , used above , but without any high energy peak .
the value of the plasma frequency @xmath69 and low frequency value of the non - metallic part of dielectric permeability @xmath221 were obtained by the least - square fit .
[ fig3]b shows the first derivative from the _ d-_wave optical scattering rate @xmath328 ( dotted line ) in comparison with the shifted on 2@xmath329 input spectral function @xmath330(solid line ) .
here we define the _ optical _ gap @xmath331 as the physical quantity , with defines the shift of the spectral function @xmath332 in fir and tunnelling spectroscopy .
the distinction between @xmath329 and @xmath327 arises due to the angle averaging ( @xmath333 ) both in the eliashberg equations and in the conductivity calculation .
the tunnelling gap @xmath327 ( see fig . [
fig3]c ) is the maximum value of the _ d-_wave gap @xmath334 and the optical one is its mean square angle average .
note , that the dip in the density of states n(@xmath99 ) ( see fig .
[ fig3]c ) arises at @xmath335 and the electron - boson region starts in @xmath24 at @xmath336 .
it means , that the low - energy excitations ( if they contribute to the spectral function ) will degrade the sharp peak in the tunnelling density of states .
if one substitutes @xmath328 in eq .
[ swaf ] , one finds that coherent relaxation still dominates , but the main peak is substantially wider and can be approximated by a gaussian function @xmath337 ( ( fig .
[ fig3]b , inset ) . the function @xmath338 is plotted by the dashed line in the fig .
[ fig3]b and agrees well with the @xmath339 . here
@xmath340 means a convolution .
note , that the coincidence of the prefactor q with the @xmath341 is come - by - chance . in my opinion ,
the additional ill - posed step , the _ real _ deconvolution of the gaussian can not be performed using the data with reasonable value of ( s / n ) , since the spectral function in these compounds have half - width of the order of @xmath329 . on the other side , if one ( by chance ) will find that @xmath328 or @xmath265 are structured , it means , that the isb separable _ d_-wave model is not suitable for htsc and have to be replaced .
strictly speaking , the _ d-_wave reflectivity is less than unity at any frequency . at the same time , there is an onset of the absorption at some @xmath342 , where r(@xmath99 ) becomes visually different from unity .
since the convolution c(@xmath99 ) is wider than the input spectral function @xmath318 , the position of the onset @xmath343 . for the model spectral
function shown in the inset in fig .
[ fig3]a @xmath344 . generally speaking
, the selected boson modes could be responsible for the _ d-_wave pairing . in this case
the position of the absorption onset will be between @xmath345 and @xmath346 , where @xmath15 is the low frequency bound of the electron - boson interaction function .
similarly , the point where the coherent relaxation rate reaches its maximum , will be shifted by @xmath329 towards the high frequencies in comparison with its position in the _ s-_wave superconductors . as a result ,
the reflectivity ratio @xmath312 ) exhibits a minimum and the scattering rate ratio @xmath311 has a maximum approximately at @xmath347 . in isb _
s-_wave metals at the same frequency @xmath32 the density of states exhibits three important features . at first , in this point @xmath24 has a maximum . at second , since the averaged over * k * value of @xmath28 coincides in this case with itself , the density of states at @xmath32 changes most rapidly . at third , for @xmath348 , the value of @xmath24 is equal to zero . in _ d-_wave superconductors this structural feature has a different energy position
. one could denote * the position of the maximum in the density of states @xmath327 as the _ tunnelling _ gap ; * the value of the shift 2@xmath329 of the image of the spectral function in the first derivative from the optical relaxation rate @xmath328 as the _ optical _ gap , * the frequency @xmath349 where r@xmath350(@xmath99 ) becomes visually distinct from unity as the onset of the absorption or as the _ absorption _ gap .
the approximate relations between the gaps are @xmath351 .
similar to the description of the visually accessible features made in subsection [ vac ] for a _
s-_wave metals , we can write the same criteria for the separable _ d-_wave model . note , that in comparison with the _ s-_wave case , all presented below itemised statements are valid approximately only . *
the single gap value @xmath327 coincides with the absorption edge . if the @xmath352 , the reflectivity r(@xmath3531 and @xmath3540 .
* the impurity scattering does not give the sharp decreasing of the reflectivity just above the @xmath327 .
instead , it decreases the gap itself . *
the coherent electron - boson scattering dominates in the superconducting state .
the first derivatives from the @xmath309 and from the r(@xmath99 ) reproduce the spectral function shifted on 2@xmath329 and _ convoluted _ with the gauss distribution function @xmath355 ) ( see fig . [
* since the frequency dependence of the coherent relaxation rate grows faster than the normal state one , and reaches its maximum value when the frequency @xmath356 , at this point the ratio of the @xmath311 will exhibit a maximum and reflectivities ratio @xmath312 ) will exhibit a minimum .
finally , let us make the simple acceptance test for the presented above visual accessibility procedure .
the test spectral function has a long history . at that old time , when people did not distinguish between the _ effective _ scattering rate @xmath114 and the _ optical _ one @xmath119 , collins _
et al _ @xcite interpreted the high energy asymptotic behaviour of the @xmath119 ( see fig .
[ fig1 ] ) in terms of its step - like effective counterpart @xmath114 ( [ nel ] , [ gameft0 ] ) . as the result , they arrived at the spectral function having a long tail up to the very high frequency ( see fig .
[ fig3]d , inset ) . nowadays
, similar spectral functions occasionally arise in different models , since this shape suggests the spin fluctuations spectrum reported by neutron spectroscopy .
the calculations was performed in the same fashion as for model spectral function shown in fig .
the coupling constant was chosen to reproduce the value of the _ tunnelling _ gap @xmath327=200 @xmath7 .
the values of the plasma frequency @xmath69 and low - frequency value of the non - metallic part of the dielectric permeability @xmath221 were obtained by the least - square fit .
the peak position had the value 330 @xmath7 which was close to the one in the original paper @xcite .
i choose the power law @xmath357 for the tail and the high energy cut - off 1600 @xmath7 .
let us compare the calculated curve and the experimental one .
since the low frequency bound in the model spectrum @xmath358 @xmath7 , the absorption edge takes place at @xmath346 in accord with the results discussed in @xcite . at second , in the region between 400 @xmath7 and 800 @xmath7 , where the reflectivity integrates the spectral function , there is a systematic shift on 140 @xmath7 . for the compensation of this shift
one has to decrease the value of tunnelling gap by a. factor of two the most important discrepancy is connected with the upper frequency bound of the single - particle relaxation , that is , with the @xmath14 .
the reflectivity ratio @xmath359 has a minimum at 1400 @xmath7 in comparison with the 900 @xmath7 in the experimental one . in conclusion ,
the separable _
d_-wave model reproduce all important features of the traditional isb _
s_-wave one with the single exception : the resulting images of the spectral function are the convolution of the input electron - boson interaction function and the gauss distribution exp(@xmath360 ) .
i express my deep gratitude to s .- l .
drechsler , o. v. dolgov , e. g. maksimov , a. golubov , d. rainer , m. kulic , n. yu . boldyrev , v. m. burlakov , and r. s. gonnelli for numerous discussions , fruitful collaboration , and generous support of this work .
i am deeply grateful to k.d .
schotte , k.f .
renk , e.a .
vinogradov , v.n .
agranovich , j. fink , and v.l .
ginzburg for their stimulate decisive support during my work
. 99 nam , s.b .
( 1967 ) theory of electromagnetic properties of superconducting and normal systems , _ phys .
_ * 156 * , 470 - 493 allen , p.b .
( 1971 ) electron - phonon effects in the infrared properties of metals , _ phys . rev .
_ * b3 * , 305 - 320 farnworth , b. and timusk , t. ( 1976 ) phonon density of states of superconducting lead , _ phys .
_ * b14 * , 5119 - 5120 shulga , s.v . , drechsler , s .- l . , fuchs , g. , mller , k .- h . ,
winzer k. , heinecke , m. , and krug k. ( 1998 ) upper critical field peculiarities of superconducting yni@xmath4b@xmath4c and luni@xmath4b@xmath4c _ phys . rev . lett._,*80 * , 1730 - 1733 schutzmann , j. , gorshunov , b. , renk , k.f . ,
munzel , j. , zibold , a. , geserich , h.p . , erb , a. and muller - vogt g. ( 1992 ) far - infrared hopping conductivity in the cuo chains of a single - domain yba@xmath4cu@xmath5o@xmath13 crystal , _ phys .
rev . _ * b46 * , 512 - 515 sulewski , p.e . ,
sievers , a.j . , buhrman , r.a . ,
tarascon , j.m . , greene , l.h . and curtin , w.a .
( 1987 ) far - infrared measurement of @xmath361 in superconducting @xmath362 , _ phys .
rev . _ * b35 * , 5330 - 5333 shulga , s.v .
, dolgov , o.v . , and maksimov , e.g. ( 1991 ) electronic states and optical spectra of htsc with electron - phonon coupling , _ physica _ * c178 * , 266 - 274 carbotte , j.p .
( 1990 ) properties of boson - exchange superconductors , _ rev .
phys . _ * 62 * , 1027 - 1158 lee , w. , rainer , d. and zimmerman , w. ( 1989 ) holstein effect in the far - infrared conductivity of high t@xmath2 superconductors , _ physica _ * c159 * , 535 - 544 dolgov , o.v . , golubov , a.a . , and shulga , s.v .
( 1990 ) far infrared properties of high t@xmath2 superconductors , _ physics letters _ * a147 * , 317 - 322 allen , p.b . and mitrovic , b. ( 1982 ) theory of superconducting t@xmath2 , in _ solid state physics _
* 37 * , edited by ehrenreich h. , zeitz f. , and turnbull , d. , academic publ .
, 1 - 92 marsiglio , f. , startseva , t. and carbotte , j.p .
( 1998 ) inversion of k@xmath5c@xmath363 reflectance data , _ phys .
lett . _ * a245 * , 172 - 176 vidberg , h.j . and sirene , j. ( 1977 ) solving the eliashberg equations by means of n - point pade approximants , _ j. low temp
. phys . _ * 29 * , 179 - 192 degiorgi , l. , wachter , p. , gruner , g. , huang , s .- m . ,
wiley , j. and kaner , r.b .
( 1992 ) optical response of the superconducting state of k@xmath5c@xmath143 and rb@xmath5c@xmath143 , _ phys .
lett _ * 69 * , 2987 - 2990 degiorgi , l. , nicol , e.j . , klein , o. , gruner , g. , wachter , p. , huang , s .-
m . , wiley , j. and kaner , r.b .
( 1994 ) optical properties of the alkali - metal - doped superconducting fullerenes : k@xmath5c@xmath143 and rb@xmath5c@xmath143 , _ phys .
rev . _ * b49 * , 2787 - 2990 mazin , i.i .
, dolgov , o.v . ,
golubov , a. and shulga , s.v .
( 1993 ) strong - coupling effects in alkali - metal - doped c@xmath143 , _ phys .
rev . _ * b47 * , 538 - 541 joyce , r.r . and richards , p.l .
( 1970 ) phonon contribution to the far - infrared absorptivity of superconducting and normal lead , _ phys .
* 24 * , 1007 - 1011 farnworth , b. and timusk , t. ( 1974 ) far - infrared measurement of the phonon density of states of superconducting lead , _ phys .
rev . _ * b10 * , 2799 - 2802 collins , r.t . ,
schlesinger , z. , holtzberg , f. , chaudhary , p. and field , c. ( 1989 ) reflectivity and conductivity of yba@xmath4cu@xmath5o@xmath6 , _ phys .
* b39 * , 6571 - 6574 sulewski , p.e . ,
sievers , a.j . , maple , m.b . ,
torikachvili , m.s . , smith , j.l . and fisk , z. ( 1988 ) far - infrared absorptivity of upt@xmath5 , _ phys . rev .
_ * b38 * , 5338 - 5352 burlakov , v.m . ,
shulga , s.v . ,
keller , j. and renk , k.f .
( 1992 ) evidence for a fano - type interaction of infrared active phonons with electronic excitations in tl@xmath4ba@xmath4ca@xmath4cu@xmath5o@xmath364 , _ physica _ * c190 * , 304 - 308 basov , d.n , liang , r. , dabrowski , b. , bonn , d.a . ,
hardy , w.n . and timusk , t. ( 1996 ) pseudogap and charge dynamics in cuo@xmath365 planes in ybaco ,
lett . _ * 77 * , 4090 - 4093 kamaras , k. , herr , s.l . , porter , c.d . ,
tache , n. , tanner , d.b . ,
etemad , s. , venkatesan , t. , chase , e. , inam , a. , wu , x.d . ,
hegde , m.s . and dutta , b. ( 1991 ) in a clean high - t@xmath2 superconductor you do not see the gap , _ phys .
lett . _ * 64 * , 84 - 87 maksimov , e. g. , rashkeev , s. n. , savrasov , s. yu . ,
uspenskii , yu .
a. ( 1989 ) microscopic studies of the optical spectra of yba@xmath4cu@xmath5o@xmath6 _ phys .
l _ * 63 * , 1880 - 1883 dervenagas , p. , bullock , m. , zarestky , j. , canfield , p. , cho , b.k . , harmon , b. , goldman , a.i . and stassis , c. ( 1995 ) soft phonons in superconducting luni@xmath4b@xmath4c , _ phys .
rev . _ * b52 * , r9839-r9842 kawano , h. , yoshizawa , h. , takeya , h. , kadowaki k. ( 1997 ) new phonon peak in superconducting state of yni@xmath4b@xmath4c _ physica _ * c282 - 287 * 1055 - 1056 kawano , h. , yoshizawa , h. , takeya , h. , kadowaki k.(1996 ) anomalous photon scattering below tc in yni@xmath4b@xmath4c _ phys .
lett . _ * 77 * 4628 - 4631 schachinger , e. , carbotte , j.p . and marsiglio , f. effect of suppression of the inelastic scattering rate on the penetration depth and conductivity in a @xmath366 superconductor , _ phys . rev .
_ * b56 * , 2738 - 2750 zhao , g.l . ,
browne , d. a. and callaway , j. ( 1995 ) quasiparticle spectrum in superconducting yba@xmath4cu@xmath5o@xmath6 , _ phys .
rev . _ * b52 * , 16217 - 16222 | the interpretation of optical conductivity in the normal and superconducting states is considered in the frame of the standard isotropic single wide band ( isb ) model using the theory proposed by nam ( nam s.b .
, _ phys .
rev . _ * 156 * 470 - 486 ( 1967 ) ) .
the _ exact analytical _ inversion of the normal state nam equations is performed and applied for the recovery of the reliable information from the experimental data .
the allen formula is derived in the strong coupling approximation and used for the physically transparent interpretation of the fir absorption .
the phenomenological generalised `` drude formula is obtained from the nam theory in a high temperature approximation .
it is shown , that the reconstruction of the shape of the spectral function @xmath0 from the normal state optical data at t@xmath1t@xmath2 is not unique and the same data can be fitted by many spectral functions . this problem is considered in detail from different points of view . at the same time , using the exact analytical solution , one can get from the normal state data a useful piece of information , namely , the value of the coupling constant , the upper energy bound of the electron - boson interaction function , and the averaged boson frequency
. moreover , they are even visually accessible , if the optical data are presented in term of the frequency dependent optical mass and scattering rate .
the superconducting state nam formalism and related simplified theory are analysed from the user point of view .
a novel _ adaptive _ method of the _ exact numerical inversion _ of the superconducting state nam equations is presented . since this approach uses the _
first derivative _ from the experimental curve , the signal / noise ratio problem is discussed in detail .
it is shown that , the fine structure of the spectral function can be recovered from optical data in the case of _ s-_wave pairing symmetry . in contrast , in the _
d-_wave case the resulting image is approximately the convolution of the input electron - boson interaction function and the gauss distribution exp(@xmath3 ) . a simplified visual accessibility ( va ) procedure is proposed for ' ' by eye `` analyses of the superconducting state optical reflectivity of the isb metals with _ s _ and _ d _ pairing symmetry . the bosons , responsible for the superconductivity in yba@xmath4cu@xmath5o@xmath6 , exhibit a ' ' phonon - like spectral function with the upper frequency bound less than 500 @xmath7 and the averaged boson frequency near 300 @xmath7 . 15.85 cm 9.00 in |
a _ chain of linear mappings _ is a system of linear mappings @xmath14 of the form @xmath15 & u_2 \ar@{-}^{{a}_2}[r ] & u_3 \ar@{-}^{{a}_3}[r ] & \cdots \ar@{-}^{{a}_{t-1}}[r ] & u_{t } } \ ] ] in which each line is either the arrow @xmath3 or the arrow @xmath4 .
we assume that @xmath2 are unitary spaces ( or are euclidean spaces ) . without loss of generality
, the reader may think that all @xmath2 are @xmath16 ( or @xmath17 , respectively ) with a natural topology on them .
let @xmath18 & v_2 \ar@{-}^{{b}_2}[r ] & v_3 \ar@{-}^{{b}_3}[r ] & \cdots \ar@{-}^{{b}_{t-1}}[r ] & v_{t } } \ ] ] be a chain with the same orientation of arrows as in .
we write @xmath19 if @xmath20 is a system of bijections such that all squares in the diagram @xmath21\ar[d]_{\varphi_1 } & u_2 \ar@{-}^{{a}_2}[r]\ar[d]_{\varphi_2 } & u_3 \ar@{-}^{{a}_3}[r]\ar[d]_{\varphi_3 } & \cdots \ar@{-}^{{a}_{t-2}}[r ] & u_{t-1 } \ar@{-}^{{a}_{t-1}}[r ] \ar[d]_{\varphi_{t-1 } } & u_t\ar[d]_{\varphi_t } \\ v_1\ar@{-}^{{b}_1}[r ] & v_2 \ar@{-}^{{b}_2}[r ] & v_3 \ar@{-}^{{b}_3}[r ] & \cdots \ar@{-}^{{b}_{t-2}}[r ] & v_{t-1 } \ar@{-}^{{b}_{t-1}}[r ] & v_t } \ ] ] are commutative ; that is , @xmath22 for each @xmath23 .
[ def ] we say that @xmath19 is * an _ isometry _ if each @xmath24 is a linear bijection that preserves the scalar product ; that is , each @xmath10 is a unitary map ( or an orthogonal map if all spaces are euclidean ) ; * a _ linear isomorphism _ if each @xmath25 is a linear bijection ( in this definition , we forget that @xmath11 and @xmath12 are metric spaces and consider them as linear spaces ) ; * a _ topological isomorphism _ if each @xmath25 is a homeomorphism , which means that @xmath10 and @xmath13 are continuous and bijective ( we forget that @xmath11 and @xmath12 are linear spaces and consider them as metric spaces ) .
each linear bijection of unitary ( or euclidean ) spaces is a homeomorphism , hence @xmath26 the main result of this paper is the following theorem , which is proved in section [ s4 ] .
[ the ] two chains of linear mappings on unitary ( or euclidean ) spaces are topologically isomorphic if and only if they are linearly isomorphic .
note that the problem of topological classification was also studied for linear operators ( budnitska is the maiden name of the first author ) , affine operators @xcite , dynamical systems @xcite , and representations of lie groups @xcite .
the paper is organized as follows . in section [ s2 ]
we show that the problem of classifying chains up to isometry is hopeless for each @xmath27 . in section [ s3 ]
we recall a known classification of chains up to linear isomorphism ; we formulate it in terms of dimensions of some subspaces . in section [ s4 ]
we show that these dimensions are also topological invariants , which proves theorem [ the ] .
in this section , we consider chains of linear mappings on _ unitary _ spaces . it would be the most natural to classify them up to isometry .
if @xmath28 , then the classification of chains up to isometry is given by the singular value decomposition : there exist orthonormal bases in @xmath29 and @xmath30 in which the matrix of @xmath31 is @xmath32 , where @xmath33 are real numbers that are uniquely determined by @xmath31 .
unfortunately , _ the problem of classifying chains up to isometry must be considered as hopeless for @xmath34 ( and so for each @xmath27 ) _ since it contains the problem of classifying linear operators on unitary spaces up to unitary similarity , and hence all systems of linear mappings on unitary spaces ( see the end of this section ) .
this statement is proved sketchy in ( * ? ? ?
* section 2.3 ) ; for the reader convenience we prove in detail the following weaker assertion .
[ thee ] the problem of classifying chains @xmath35 up to isometry contains the problem of classifying linear operators on unitary spaces up to unitary similarity .
we say that matrices @xmath36 and @xmath37 are _ unitarily similar _ if there exists a unitary matrix @xmath38 such that @xmath39 .
let us consider chains of mappings @xmath40 that are given in some orthonormal bases by pairs of matrices @xmath41 and @xmath42 in which @xmath43 and all blocks are @xmath44 .
it suffices to prove that @xmath45 indeed , assume we know a set of canonical matrix pairs for . we take those of them that can be reduced to the form @xmath41 and reduce them to it .
due to , the obtained blocks @xmath36 form a set of canonical matrices for unitary similarity .
let us prove .
`` @xmath46 '' let the chains be isometric ; that is , there exist unitary matrices @xmath47 such that @xmath48 by the first equality in , @xmath49 since @xmath50 , we have @xmath51 for some @xmath44 matrices @xmath52 , and @xmath53 . by the second equality in , @xmath54 .
equating the corresponding horizontal strips , we obtain @xmath55 let @xmath56_{i , j=1}^2 $ ] . the first equality in implies that @xmath57 and @xmath58 .
since @xmath59 is unitary , @xmath60 and so @xmath61 . the second equality in implies that @xmath62 . the third equality in implies that @xmath63 and @xmath64
thus , @xmath36 and @xmath37 are unitarily similar .
`` @xmath65 '' conversely , if @xmath66 for some unitary @xmath67 , then holds for @xmath68 and @xmath69 , and so the chains are isometric .
recall that a _ quiver _ is a directed graph .
representation _ is given by assigning to each vertex a unitary space and to each arrow a linear mapping of the corresponding vector spaces .
a representation is _ unitary _ if all of its vector spaces are unitary .
it was shown in ( * ? ? ?
* section 2.3 ) that the problem of classifying linear operators on unitary spaces up to unitary similarity contains the problem of classifying unitary representations of an _ arbitrary _ quiver .
thus , we can not expect to find an observable system of invariants for linear operators on unitary spaces .
nevertheless , we can reduce the matrix of _ any given _ linear operator on a unitary space ( moreover , the matrices of any given unitary representation of a quiver ) to canonical form by using littlewood s algorithm ; see ( * ? ? ?
* section 3 ) . in the same way ,
the problem of classifying pairs of linear operators on a vector space is considered as hopeless ( and all classification problems that contain it are called _ wild _ ) since it contains the problem of classifying representations of each quiver .
nevertheless , we can reduce the matrices of any given representation of a quiver to canonical form by using belitskii s algorithm ; see @xcite .
in this section , we consider chains of linear mappings @xmath70 & u_2 \ar@{-}^{{a}_2}[r ] & u_3 \ar@{-}^{{a}_3}[r ] & \cdots \ar@{-}^{{a}_{t-1}}[r ] & u_{t } } \ ] ] on vector spaces without scalar product . without complicating the proofs ,
we consider them over _ any _ field @xmath71 . in theorem [ jtt ]
we recall the well - known classification of such chains up to linear isomorphisms ( see definition [ def](ii ) ) .
next we fix some subspaces of @xmath2 and prove in theorem [ ljtt ] that the set of their dimensions is a full system of invariants of chains with respect to linear isomorphisms . in section [ s4 ]
we establish that this set is also a full system of invariants of chains with respect to topological isomorphisms , which proves theorem [ the ] .
the directions ( @xmath72 or @xmath73 ) of all linear mappings @xmath74 in can be given by the directed graph @xmath75 & 2 \ar@{-}^{\alpha_2}[r ] & 3 \ar@{-}^{\alpha_3}[r ] & \cdots \ar@{-}^{\alpha_{t-1}}[r ] & { t } } \ ] ] in which each arrow @xmath76 is directed as @xmath74 .
( thus , each chain defines a representation of the quiver and a linear isomorphism of chains defines an isomorphism of the corresponding representations . )
write @xmath77 the _ direct sum _ of chains @xmath5 and @xmath6 & v_2 \ar@{-}^{{b}_2}[r ] & v_3 \ar@{-}^{{b}_3}[r ] & \cdots \ar@{-}^{{b}_{t-1}}[r ] & v_{t } } \ ] ] with the same directed graph is the chain @xmath78 for every pair of integers @xmath79 such that @xmath80 , we define the chain @xmath81 in which `` @xmath82 '' is the identity bijection and @xmath83 s are at the vertices @xmath84 of .
the following theorem is well known in the theory of quiver representations ; the representations of and the other quivers that have a finite number of nonisomorphic indecomposable representations were classified by gabriel @xcite .
[ jtt ] each chain @xmath85 is linearly isomorphic to a direct sum of chains of the form @xmath86 . this direct sum is uniquely determined by @xmath85 , up to permutation of summands .
an algorithm for constructing this canonical form of chains of linear mappings over @xmath87 is given in ( * ? ? ?
* section 4 ) ; it uses only transformations of unitary equivalence of matrices : @xmath88 in which @xmath89 and @xmath90 are unitary .
each chain @xmath85 of linear mappings on unitary ( or euclidean ) spaces is topologically isomorphic to a direct sum of chains of the form @xmath91 .
this direct sum is uniquely determined by @xmath85 , up to permutation of summands .
let @xmath85 be any chain of the form .
in each of its spaces @xmath11 , we define a series of subspaces @xmath92 by induction : @xmath93 and if is constructed for @xmath94 then @xmath95 ( here @xmath96 denotes the preimage of @xmath97 ) .
each chain of the form @xmath98 is given by the pair of matrices @xmath99 in some bases of @xmath100 .
changing the bases , we can reduce the pair by transformations @xmath101 it is convenient to give @xmath99 by the block matrix @xmath102 $ ] since the rows of @xmath103 and @xmath104 are transformed by the same matrix @xmath105 . due to , we can reduce it by elementary row transformations ( i.e. , by simultaneous elementary transformations with rows of @xmath103 and @xmath104 ) and by elementary column transformations within @xmath103 and @xmath104 . each @xmath102 $ ] can be reduced by these transformations to its canonical form @xmath106 as follows ( see ( * ? ? ?
* section 4 ) ) .
we first reduce @xmath103 to the form @xmath107 and denote the obtained block matrix by @xmath108 $ ] .
then we extend to @xmath109 the partition of @xmath110 into two horizontal strips and reduce the second horizontal strip of @xmath109 to the form : @xmath111 we make @xmath112 equal to zero by adding linear combinations of rows of @xmath113 . at last , we reduce @xmath114 to the form by elementary transformations ; these transformations may spoil @xmath115 , we restore it by column transformations . the obtained block matrix has the form .
for example , let the chain be given in some bases @xmath116 , @xmath117 , and @xmath118 of @xmath29 , @xmath30 , and @xmath119 by the following canonical block matrix of the form : @xmath120 then @xmath85 is the direct sum of 9 chains that are given by the action of the mappings on the basic vectors as follows : @xmath121 & 0&0 \ar@{|->}[l]\\ e_2\ar@{|->}[r ] & 0&0 \ar@{|->}[l]\\ 0\ar@{|->}[r ] & 0&g_1 \ar@{|->}[l]\\ e_3\ar@{|->}[r ] & f_1&g_2 \ar@{|->}[l]\\ e_4\ar@{|->}[r ] & f_2&g_3 \ar@{|->}[l]\\ e_5\ar@{|->}[r ] & f_3&0 \ar@{|->}[l]\\ 0\ar@{|->}[r ] & f_4&g_4 \ar@{|->}[l]\\ 0\ar@{|->}[r ] & f_5&g_5 \ar@{|->}[l]\\ 0\ar@{|->}[r ] & f_6&0 \ar@{|->}[l]\\ } \end{split}\ ] ] ( for simplicity of notation , we write @xmath122 instead of @xmath123 . )
thus , @xmath85 is linearly isomorphic to @xmath124 the subspaces @xmath97 defined in and are the following : @xmath125 here @xmath126 denotes the subspace spanned by @xmath127 .
note that @xmath128 in which @xmath129 and @xmath130 are the vector spaces of the chains given by @xmath131 & f_1&g_2 \ar@{|->}[l]\quad \text{and}\quad e_4\ar@{|->}[r ] & f_2&g_3 \ar@{|->}[l]\\ } \ ] ] [ ljtt ] each chain @xmath85 is fully determined , up to linear isomorphism , by the indexed set @xmath132 and @xmath97 are defined in . by theorem [ jtt ]
, @xmath85 possesses a canonical decomposition @xmath133 whose summands are determined up to renumbering and linear isomorphisms of summands .
@xmath8 is determined up to linear isomorphism by the family of pairs @xmath134 _ and this family is determined by @xmath5 up to renumbering ( i.e. , @xmath134 is an unordered set with repeating elements ) . for technical reason , it is better to prove the following statements that are stronger than the theorem : * @xmath135 uniquely determines @xmath136 , up to renumbering , * there are indices @xmath137 such that each of the spaces @xmath138 defined in is decomposed into the direct sum @xmath139 ( see ; we put @xmath140 if @xmath141 ) . , @xmath142 , @xmath143 , and @xmath144 . ] * all chains @xmath145 have the first nonzero space at the same position , i.e. @xmath146 * @xmath147 if @xmath148 .
we use induction on @xmath149 .
the induction base is trivial : the statements ( i)(iv ) hold for chains with @xmath150 vector spaces ; that is , for @xmath151 and @xmath152 .
suppose that ( i)(iv ) hold for chains with @xmath153 vector spaces , in particular , for the restriction @xmath154 & u_2 \ar@{-}^{{a}_2}[r ] & u_3 \ar@{-}^{{a}_3}[r ] & \cdots \ar@{-}^{{a}_{t-2}}[r ] & u_{t-1 } } \ ] ] of @xmath8 to the first @xmath153 spaces .
we can suppose that the summands in are numbered such that @xmath155 the canonical decomposition of @xmath156 can be obtained from as follows : @xmath157 in which @xmath158 is defined in and every @xmath159 is the restriction of @xmath160 to the first @xmath153 vector spaces . by induction hypothesis ,
* @xmath161 uniquely determines @xmath162 , up to renumbering , * there are indices @xmath163 such that each of the spaces @xmath164 is decomposed into the direct sum @xmath165 * @xmath166 * @xmath167 if @xmath168 .
we suppose that the summands in are numbered such that @xmath169 let us prove ( i)(iv ) .
consider two cases that are differ in the direction of the last arrow in .
_ case 1 : _ @xmath170 .
by , @xmath171 by , , and , we have @xmath172 which is the desired decomposition .
_ case 2 : _ @xmath173 . by , @xmath174 by , , and
, we have @xmath175 which is the desired decomposition . in both the cases , the family of pairs @xmath176 ( which is determined up to renumbering ) can be obtained from @xmath162 by replacing @xmath177 pairs @xmath178 with @xmath179 for each @xmath23 and by attaching @xmath180 pairs @xmath181 .
this proves the statement ( i ) since @xmath182 are expressed via @xmath183 : @xmath184 ( we set @xmath185 ) .
the statements ( ii)(iv ) follow from the induction hypothesis and cases 1 and 2 .
the goal of this section is to prove theorem [ the ] .
let @xmath19 be a topological isomorphism of chains of the form and .
due to theorem [ ljtt ] , it suffices to prove that their sets coincide ; that is , @xmath186 in which @xmath97 are the vector subspaces of @xmath11 that were constructed in , and @xmath187 are the vector subspaces of @xmath12 that are analogously constructed by the chain @xmath9 . due to definition [ def](iii ) , the topological isomorphism @xmath188 is formed by the homeomorphisms @xmath189 .
it suffices to show that each @xmath10 maps @xmath97 on @xmath187 since then each @xmath97 is homeomorphic to @xmath187 and by @xcite all homeomorphic vector spaces have the same dimension .
what is left is to prove the following lemma .
the assertion holds for @xmath191 since @xmath192 is a bijection .
suppose that holds for @xmath193 ( and all @xmath194 ) ; let us prove it for @xmath195 .
it suffices to prove that @xmath196 since then we can use for @xmath197 instead of @xmath198 and obtain @xmath199 which ensures @xmath200 . in the case
@xmath201 , the inclusion holds since if @xmath202 and @xmath203 , then @xmath204 \ar[d]_{\varphi_k}^{\text{biect } } & u_{k+1,j } \ar[d]_{\varphi_{k+1 } } & & x \ar@{|->}[r ] \ar@{|->}[d ] & y\ar@{|->}[d ] \\ v_{kj}\ar@{->}^{{b}_k}[r ] & v_{k+1 } & & \varphi_k x \ar@{|->}[r ] & { b}_k\varphi_k x } \ ] ] thus , @xmath205 . in the case
@xmath206 , the inclusion holds since if @xmath202 then @xmath207_{\varphi_k } & u_{k+1,j } \ar@{->}_{{a}_k}[l ] \ar[d]_{\varphi_{k+1 } } & & { a}_ky \ar@{|->}[d ] & y\ar@{|->}[d ] \ar@{|->}[l ] \\ v_{k } & v_{k+1}\ar@{->}_{{b}_k}[l ] & & \varphi_k{a}_ky & \varphi_{k+1}y \ar@{|->}[l ] } \ ] ] thus , @xmath208 and so @xmath209 . | we consider systems of linear mappings @xmath0 of the form @xmath1 & u_2 \ar@{-}^{{a}_2}[r ] & u_3 \ar@{-}^{{a}_3}[r ] & \cdots \ar@{-}^{{a}_{t-1}}[r ] & u_{t } } \ ] ] in which @xmath2 are unitary ( or euclidean ) spaces and each line is either the arrow @xmath3 or the arrow @xmath4 .
let @xmath5 be transformed to @xmath6 & v_2 \ar@{-}^{{b}_2}[r ] & v_3 \ar@{-}^{{b}_3}[r ] & \cdots \ar@{-}^{{b}_{t-1}}[r ] & v_{t } } \ ] ] by a system @xmath7 of bijections .
we say that @xmath8 and @xmath9 are linearly isomorphic if all @xmath10 are linear . considering all @xmath11 and @xmath12 as metric spaces
, we say that @xmath8 and @xmath9 are topologically isomorphic if all @xmath10 and @xmath13 are continuous .
we prove that @xmath8 and @xmath9 are topologically isomorphic if and only if they are linearly isomorphic .
chains of linear mappings , topological equivalence 5a21 ; 37c15 |
praesepe ( m44 = ngc 2632 = the beehive cluster ) and the hyades are often considered as twin open clusters because they appear to have essentially the same age ( @xmath4790 myr ) and metallicity ( [ fe / h]@xmath40.15 ) .
some authors attribute these similarities to the two clusters having been born within the same molecular cloud or cloud complex ( schwarzschild & hertzsprung 1913 , klein - wassink 1927 ) ; the space motions appear to support this picture ( see , _
e.g. _ , eggen 1992 ) . because the hyades is much closer than praesepe ( 44 pc vs. @xmath4180 pc ; perryman et al .
1998 , van leeuwen 2007 ) , its stars are much brighter ; the hyades has therefore received much more attention in the published literature .
however , proximity also has its disadvantages .
the hyades is spread out over a very large area on the sky , making it difficult to take advantage of multi - object spectrographs or wide - field cameras to efficiently survey many cluster members at once .
praesepe s four times greater distance can become an advantage in that sense , if a suitable wide - area facility becomes available .
we have taken advantage of one such circumstance , the unexpected availability of the 100 square degree fov kepler space telescope to obtain synoptic photometry of ecliptic plane fields for @xmath475 days .
there have been four prior large surveys to determine rotation periods for low mass stars in praesepe .
scholz et al .
( 2011 ) reported @xmath450 periods ; delorme et al .
( 2011 ) used data from the wasp telescope network to monitor @xmath470 members ; kovcs et al .
( 2014 ) monitored @xmath4400 members with the hat telescope network , and a pair of papers by agueros et al .
( 2011 ) and douglas et al .
( 2014 ) used data from the ptf telescope to monitor @xmath4500 members .
those groups reported periods for 49 , 52 , 180 , and 40 praesepe members , respectively ( and a total of 220 unique stars with periods ) , primarily covering stars with spectral types g , k and early m. the k2 data for praesepe were obtained in 2015 in campaign 5 .
k2 provides precision , sensitivity and long duration , continuous coverage that yields superb light curves ( lcs ) of much greater quality than can be obtained from the ground , and extending to much lower signal amplitudes and masses .
at least three groups have already obtained rotation periods for praesepe members based on these data .
libralato et al .
( 2016 ) derived rotation periods as a step in searching the light curves for exoplanet transits , but they did not discuss the periods nor the shapes of the phased light curves .
mann et al .
( 2016 ) similarly derive rotation periods while searching for exoplanet transits ; they illustrate the distribution of rotation period as a function of @xmath6 and color , but do not otherwise discuss the periods or light curve morphology .
finally , our group plotted an early version of the praesepe period distribution in stauffer et al .
( 2016 ) in the context of comparing the pleiades rotation periods to those in other clusters .
however , no paper has yet published a complete analysis of the k2 praesepe rotational data nor a discussion of the light curve morphologies shown by those stars .
this paper is devoted to those two topics .
much of our current analysis is very similar to that we conducted in the pleiades ( rebull et al .
2016a , b , stauffer et al .
2016 ) , and we often make comparisons to the pleiades analysis and results . in section [ sec : obs ] , we present the observations and data reduction , as well as the range of data from the literature that we assembled , and define the final sample of members that we analyze in the rest of the paper . the overall distribution of k2-derived rotation rates is discussed in section [ sec : rotationdistrib ] .
section [ sec : lcpcats ] places these lcs in the same categories as we defined in the pleiades ( rebull et al .
section [ sec : singleandmulti ] compares the locations of the single- and multi - period stars in a variety of ways , and section [ sec : deltap ] looks at the spacings between periods for those stars with clear multiple periods .
finally , we summarize our results in section [ sec : concl ] .
praesepe members and candidate members were observed in k2 campaign 5 , which lasted for 75 d between april and october 2015 .
[ fig : where ] shows the distribution of objects with k2 lcs on the sky ; note the gaps between detectors . all of the stars shown were observed in the long - cadence ( @xmath430 min exposure ) mode .
some of these stars were additionally observed in fast cadence ( @xmath41 min exposure ) , but those data are beyond the scope of the present paper .
there are 984 unique k2 long - cadence light curves corresponding to members or candidate members of praesepe ( see sec [ sec : membership ] below ) .
the tidal radius of praesepe is 12.1 pc ( holland et al .
2000 ) , which at a distance of 184 pc ( see sec .
[ sec : suppdata ] below ) is @xmath43.8@xmath7 across , which is easily covered by k2 .
( stars that are more than 5@xmath7 away from the cluster center are highlighted in app .
[ app : halo ] . )
kepler pixel sizes are relatively large , @xmath8 , and the 95% encircled energy diameter ranges from 3.1 to 7.5 pixels with a median value of 4.2 pixels . during the k2 portion of the mission , because only two reaction wheels can be used , the whole spacecraft slowly drifts and then repositions regularly every 0.245 d. this drift is @xmath40.1@xmath9 per hour ( cody et al .
2017 ) . since these data were reduced at the same time as we reduced our pleiades data ( rebull et al .
2016a ) , we have the same sets of lcs available to us .
( 1 ) the pre - search data conditioning ( pdc ) version generated by the kepler project and obtained from mast , the mikulski archive for space telescopes . ( 2 )
a version with moving apertures obtained following cody et al . in prep .
( 3 ) the version using a semiparametric gaussian process model used by aigrain et al .
( 2015 , 2016 ) .
( 4 ) the ` self - flat - fielding ' approach used by vanderburg & johnson ( 2014 ) as obtained from mast .
to this , late in the process , we added ( 5 ) the lcs from the everest pipeline ( luger et al .
2016 ) , which uses pixel level decorrelation .
we removed any data points corresponding to thruster firings and any others with bad data flags set in the corresponding data product .
again , following the same approach as in the pleiades , we inspected lcs from each reduction approach , and we selected the visually ` best ' lc from among the original four lc versions .
since the everest lcs became available late in our analysis process , for the most part , we used the everest lcs to break ties or clarify what the lc was doing .
about 7% had such severe artifacts that no ` best ' lc could be identified ; as for the pleiades , this is often a result of instrumental ( non - astrophysical ) artifacts ( because the star is too bright or too faint or adversely affected by nearby stars , etc . ) .
as for the pleiades , though , the periods we report here are not generally ambiguous , and are detected in all the lc versions ; they just look cosmetically better in one version or another .
our approach of comparing several different lc versions also minimizes the liklihood that any of these lc reductions removes stellar signal .
out of the 984 lcs , only one pair of stars for which a k2 lc was requested were within 4@xmath9 of each other ( within a kepler pixel ) .
these objects are discussed further in app .
[ app : pair ] . our approach for finding periods was identical to that which we used in the pleiades in rebull et al .
( 2016a ) . in summary , we used the lomb - scargle ( ls ; scargle 1982 ) approach as implemented by the nasa exoplanet archive periodogram service ( akeson et al .
we looked for periods between 0.05 and 35 d , with the upper limit being set by roughly half the campaign length .
[ fig : singlefreq ] shows lcs , periodograms , and phased lcs for some praesepe stars with single , unambiguous periods .
as we found in the pleiades , the overwhelming majority of them have zero , or effectively zero , false alarm probability ( fap ) ; a significant fraction have more than one peak with 0 fap ; see section [ sec : lcpcats ] below . for stars of the mass range considered here ,
the phased lcs are mostly sinsuoidal and therefore best attributed to star spot - modulated rotation .
we find periods for 828/984 k2 light curves , or 84% of all k2 light curves of candidate or confirmed praesepe members ( see sec .
[ sec : membership ] ) .
there is no significant trend with color ( as a proxy for mass ) for the fraction of periodic stars . as for the pleiades
, we removed from this distribution any periods that are likely to be eclipsing binaries ( see appendix [ app : binaries ] ) or those whose waveforms did not seem to be rotation periods ( see appendix [ app : timescales ] ) .
there are a few multi - periodic stars which we suspect to be pulsating variables ( see appendix [ app : pulsator ] ) ; many of them are also reported in the literature as @xmath10 scuti - type variables .
for some pulsators , the first period is still likely to be similar to the rotation period ( see discussion in rebull et al .
for the remaining multi - periodic stars ( @xmath425% of the sample ; see [ sec : lcpcats ] below ) , for the most part , we took the period corresponding to the strongest peak in the periodogram as the rotation period to be used for subsequent analysis .
in a few cases ( e.g. , double dip stars ; see sec . [ sec : singleandmulti ] ) , a secondary peak is the right @xmath11 to use .
we have chosen four of the most recent surveys looking for rotation rates in praesepe for a detailed comparison of our results ; they are summarized in figure [ fig : compareliterature ] .
there are 220 praesepe stars in the literature with at least one estimate of @xmath12 ; 60 of those do not have k2 lcs .
scholz et al .
( 2011 ) used the isaac newton telescope to monitor very low mass members , reporting 49 periods .
we have periods for 40 of the stars in common between the two studies ; the remaining scholz et al.stars were not monitored ( either because they are too faint for k2 , or because they are in the gaps between chips ) .
the periods derived for these 40 are shown in figure [ fig : compareliterature ] . there is generally very good agreement ; the median fractional difference ( @xmath13 ) is 2.4% .
there are 8 stars for which their period and ours do not match , and most of them are harmonics , because they are factors of two off ( too small or large ) . in four of these cases ,
the period we derive is very close to 1 day , which is hard to recover with a ground - based telescope .
we believe our periods to be correct in all cases for the epochs in which we observed this cluster .
two stars are of special note .
epic 211951438/hshj396 has a period of 0.593 d in scholz et al . ,
but we can not recover any period for this star , so we do not retain it as periodic .
epic 211984058/2massj08384128 + 1959471 was not a star for which we initially found a period , because all five light curve versions are different , and it was not clear which lc version is the ` best ' ( see discussion above in [ sec : obs ] ) .
however , upon examination of the k2 thumbnail and poss images of the region , several of the lc versions appear to have been dragged off to a nearby bright star .
the aigrain et al .
version is the only one that both stays on the target star and has a periodic signal .
the period we derive from this is identical to that from scholz et al .
( 2011 ) , so we retained this star as periodic with that period .
delorme et al .
( 2011 ) used superwasp to monitor 71 cluster members , looking for periods from 1.1 to 20 d , determining that 52 were periodic .
of these periodic stars , we have 39 in common ( the remaining 13 fall in the gaps between k2 chips , or are otherwise off the k2 fov ) . the periods derived for these
39 are shown in figure [ fig : compareliterature ] .
there is generally very good agreement ; the median fractional difference ( @xmath14 ) is 2.7% .
there are three stars for which our periods do not agree .
one is epic 212013132/js379=2massj08404426 + 2028187 , where it is likely that one of the surveys detected an harmonic .
superwasp obtains a period of 4.27 d ; we obtain three periods , including one that is close to 4.4 d , but the primary period we obtained for this star is 2.129 d. the second star is epic 211950227/2massj08402554 + 1928328 , where the k2 light curve is very , very messy .
superwasp obtains 13.15 d. the period we adopted for this one is 0.8984 d , and comes from only part of the lc , with most of the lc having been corrupted by instrumental effects .
the third star is epic 211995288/kw30=2massj08372222 + 2010373 , where we find a period of 7.8 d , a factor of 2 larger than the 3.9 d found in superwasp , so another harmonic .
having tested these periods by phasing our lcs at these alternate periods , we believe our periods to be correct in all cases for the epochs in which we observed these stars .
there are no stars in common where superwasp has a period and we do not have a period .
kovcs et al .
( 2014 ) used hatnet ( hungarian automated telescope networks ) to monitor 381 members , finding 180 rotation periods , all ranging between 2.5 and 15 d. they identified 10 more stars as having what we would call ` timescales ' ; that is , repeated patterns that may or may not correspond to periods .
there are 152 stars ( 150 out of their rotation periods , and 2 more out of their ` timescales ' ) in common between the surveys , shown in figure [ fig : compareliterature ] ; most of the remaining stars fall in the gaps between chips , while a few could have been but simply were not observed . for the stars that we have in common ,
the median fractional difference ( @xmath15 ) is 2.2% .
there are 13 stars for which the periods do not match , 7 of which are likely harmonics ; we believe our period to be correct .
one of the two from their ` timescales ' category is epic 211918335/hat-269 - 0000582/kw244=tx cnc , which kovcs et al .
identify as an eclipsing binary and whelan et al .
( 1973 ) identifies it as a w uma - type ; we drop this as @xmath16 .
the other is 211947631/hat-269 - 0000465/bd+19d2087 ( which kovcs et al . simply call ` miscellaneous ' ) ; we retain @xmath0=4.74 as a rotation period .
there are no stars for which we both have light curves but hatnet has a period and we do not .
ageros et al .
( 2011 ) and douglas et al .
( 2014 ) are a 2-part study on the palomar transient factory ( ptf ) observations of praesepe .
they used ptf to monitor 534 members , 40 of which were periodic between 0.52 and 35.85 d. there are 30 stars in common between the studies ( see figure [ fig : compareliterature ] ) , with the remaining 10 falling in the gaps between chips .
the derived periods are generally in very good agreement ; the median fractional difference ( @xmath17 ) is just 0.36% .
there is one star ( epic 211989299/ad2552=2massj08392244 + 2004548 ) for which one of us is likely to have found an harmonic , though the light curve is messy ( we obtain 12.791 and they obtain 25.36 ) .
we believe our periods to be correct for the epochs in which we observed these stars .
for another one ( epic 211886612/js525 ) , they found a period of 22.6 d ; we retrieved a ls peak at 17.2 d for this object , but relegated it to ` timescales ' ( in appendix [ app : timescales ] ; this is why only 29 stars appear in fig .
[ fig : compareliterature ] ) . for the light curve
we have , it just does not look as nice as the rest of the clear @xmath11 here .
we note here that , using k2 , we find periods for a far higher number and higher fraction of praesepe members than has been accomplished before .
moving to space enables higher precision and continuous photometry , resulting in a far higher fraction of detectable rotation periods .
we started with the list of praesepe members that we had proposed for this campaign , and added targets from other related programs . from this list
, we then amassed supporting data and assessed membership for each star .
for some objects , we obtained additional keck / hires spectroscopy ; see app .
[ app : keck ] .
table [ tab : bigperiods ] includes the supporting photometric data we discuss in this section , plus the periods we derive here ( in section [ sec : periods ] ) , for members of praesepe . for completeness , non - members ( nm )
appear in appendix [ app : nm ] .
cl epic & number in the ecliptic plane input catalog ( epic ) for k2 + coord & right ascension and declination ( j2000 ) for target + othername & alternate name for target + vmag & v magnitude ( in vega mags ) , if observed + kmag & @xmath18 magnitude ( in vega mags ) , if observed + vmk & @xmath1 , as directly observed ( if @xmath19 and @xmath18exist ) , or as inferred ( see text ) + p1 & primary period , in days ( taken to be rotation period ) + p2 & secondary period , in days + p3 & tertiary period , in days + p4 & quaternary period , in days + lc & lc used as ` best ' + single / multi - p & indicator of whether single or multi - period star ; if object has a ` timescale ' ( see app [ app : timescales ] ) , that is indicated + dd & indicator of whether or not it is a double - dip lc + ddmoving & indicator of whether or not it is a moving double - dip lc + shch & indicator of whether or not it is a shape changer + beat & indicator of whether or not the full lc has beating visible + cpeak & indicator of whether or not the power spectrum has a complex , structured peak and/or has a wide peak + resclose & indicator of whether or not there are resolved close periods in the power spectrum + resdist & indicator of whether or not there are resolved distant periods in the power spectrum + pulsator & indicator of whether or not the power spectrum and period suggest that this is a pulsator + despite praesepe s proximity and richness , its age and metallicity are a subject of considerable debate .
age estimates range from 600 to 900 myr : harris ( 1976 ) , 630 myr ; mermilliod ( 1981 ) , 660 myr ; vandenberg & bridges ( 1984 ) , 900 myr ; janes & phelps ( 1994 ) , 900 myr ; bonatto et al .
( 2004 ) , 729 myr ; salaris et al .
( 2004 ) , 700 myr ; brandt & huang ( 2015 ) , 790 myr . for this paper ,
we adopt the brandt & huang age , both because it is based on the most recent physics and because it is a reasonable average of the other estimates .
metallicity estimates for praesepe also span a somewhat surprisingly large range : friel & boesgaard ( 1992 ) , [ fe / h ] = + 0.04 ; an et al .
( 2007 ) , [ fe / h ] = + 0.19 ; pace , pasquini & francois ( 2008 ) , [ fe / h ] = + 0.27 ; boesgaard , roper & lum ( 2013 ) , [ fe / h ] = + 0.12 ; blanco - cuaresma et al .
( 2015 ) , [ fe / h ] = 0.20 ; and yang , chen & zhao ( 2015 ) , [ fe / h ] = + 0.16 ) .
we adopt [ fe / h ] = + 0.15 as a reasonable compromise value .
the difference in metallicity should not matter significantly in the context of comparisons to other kepler and k2 light curves , but we note it nonetheless , as it might affect stellar structure and angular momentum evolution . despite the lack of agreement on age and metallicity
, there is actually little dispute for the distance to the cluster , with modern values of @xmath20 = 6.32 ( salaris et al .
2004 ) , 6.33 ( an et al.2007 ) , and 6.30 ( van leeuwen 2009 ) .
we adopt @xmath20 = 6.32 , corresponding to a distance of 184 pc .
the reddening to the cluster is quite small , though possibly as large as @xmath21 = 0.027 ( taylor 2006 ) . for convenience ,
we adopt @xmath21 = 0.0 . to the accuracy that we have determined and are using @xmath22 ,
the error from reddening is negligible . in subsequent analysis here
, we often compare praesepe results to those from the pleiades ( rebull et al .
2016a , b ) .
we take the age of the pleiades to be 125 myr ( stauffer et al .
1998 ) , the distance to be 136 pc ( melis et al . 2014 ) , and the metallicity to be [ fe / h]=+0.04 ( soderblom et al . 2009 ) .
we started with a list of members based on an in - house / private open cluster database originally created by prosser & stauffer in the 1990s .
for praesepe , this was a merger of membership lists in klein - wassink ( 1927 ) , jones & cudworth ( 1983 ) , and jones & stauffer ( 1991 ) , with some candidate members deleted due to discrepant photometry or radial velocities .
this list was then merged with half a dozen recent proper motion membership studies ( adams et al .
2002 ; kraus & hillenbrand 2007 ; baker et al . 2010
; boudreault et al .
2012 ; khalaj & baumgardt 2013 ; wang et al .
2014 ) , retaining stars considered as likely members in those papers .
we then merged this praesepe membership catalog with the list of all stars observed in k2 campaign 5 within programs targeting praesepe .
about 600 did not have k2 lcs , sometimes due to the star falling in ccd gaps or just completely outside the k2 fov ; in other cases , the star may have been observable , but no light curve was obtained . at this point , then , we have a set of 984 praesepe members or candidate members with k2 light curves .
@xmath23 photometry on the kron system has been published by upgren , weis & deluca ( 1979 ) , weis ( 1981 ) , stauffer ( 1982 ) , and mermilliod et al .
( 1990 ) . for stars not in those papers
, we also queried the apass database ( henden et al .
2016 ) , particularly for @xmath19 magnitudes .
we similarly queried the gaia dr1 release ( gaia collaboration 2016 ) for their @xmath24 magnitudes ( the released proper motions do not go faint enough for this work ) .
finally , we queried the recently released pan - starrs1 database ( chambers et al .
2016 ) .
we added to this data from the two - micron all sky survey ( 2mass ; skrutskie et al .
2006 ) , from the spitzer space telescope ( werner et al .
2004 ) , including measurements from the spitzer enhanced imaging products , seip , and from the widefield infrared survey explorer ( wise ; wright et al.2010 ) . to refine the list of members , we queried the data from the united states naval observatory ( usno ) robotic astrometric telescope ( urat ; zacharias et al .
2015 ) astrometric catalog to extract the @xmath25 " magnitude ( a broad band optical / red magnitude ) and the urat proper motions for these stars .
we then plotted the candidate praesepe members in a vector - point diagram ( vpd ) and in two color - magnitude diagrams ( cmds ) , @xmath25 vs. @xmath26 and @xmath24 vs. @xmath27 . in the vpd
, we took the star as a member if the star s proper motion was within 15 mas yr@xmath28 of @xmath29 mas yr@xmath28 and @xmath30 mas yr@xmath28 ( which is a mean motion consistent with the urat proper motions for the cluster ) .
for the two cmds , we fit a polynomial curve to the single star locus and then took the star as a member if the star was displaced @xmath21.3 magnitude above the locus or 0.7 magnitudes below the locus . for the stars included in our member catalog , we required that two of three of these flags ( one flag for the vpd and one for each of the cmds ) be true " .
we removed stars flagged as non - members in the mermilliod et al.(1990 ) radial velocity and photometry survey for praesepe halo members .
for some objects identified as outliers in the analysis below , we obtained additional keck / hires spectroscopy ; see app .
[ app : keck ] . for those objects
whose radial velocities ( rv ) were inconsistent with membership , we identifed them as non - members ( see app .
[ app : nm ] for a list ) . from the initial sample of 984 lcs ,
then , 943 are members of praesepe by our criteria ( and 41 are non - members ; see appendix [ app : nm ] ) . since we want to compare to the results we obtained in the pleiades , we wanted to use @xmath1 as a mass proxy in the same fashion as we did in the pleiades .
we originally approached this the same way we had for the pleiades , and used data from urat .
however , gaia data are now available .
since @xmath22 is available ( with @xmath19 and @xmath31 directly observed ) for @xmath4250 praesepe stars with k2 lcs , we derived a formula to convert @xmath27 colors to @xmath22 colors by comparing the @xmath22 to the @xmath27 .
however , there are no observed @xmath19 photometry of known members for colors redder than about @xmath22=5.5 . to extend the calibration of @xmath27
, we have used faint pleiades and hyades stars ( where @xmath19 photometry was obtained for very low mass stellar and brown dwarf candidates in pleiades and hyades respectively by stauffer et al .
1989 , 1994 and bryja et al . 1992 , 1994 ) . to these , we added gj512b , a field star .
these stars are listed in table [ tab : redanchors ] .
the final relationship between @xmath22 and @xmath27 is : @xmath32 we used this to interpolate @xmath1 for those praesepe stars lacking @xmath1 , but having a gaia measurement . for
about a dozen more stars , there is no gaia measurement , but there is a urat measurement ; we used the relationship we derived for the pleiades to obtain a @xmath1estimate from the urat data for these remaining stars .
all of the stars with a measured period have a measurement or estimate of @xmath1 ; five of the non - periodic stars are missing a @xmath1 . for @[email protected] , median uncertainties ( based on the scatter in the gaia calibration above ) are @xmath40.017 mag ; for 2.5@xmath2@[email protected] , median uncertainties are @xmath40.085 mag .
lllll hhj3 & 22.14 & 18.78 & 14.13 & stauffer et al .
( 1998 ) + hhj5 & 20.71 & 18.45 & 13.93 & stauffer et al .
( 1998 ) + hhj8 & 20.60 & 18.38 & 14.23 & stauffer et al .
( 1998 ) + hhj10 & 20.74 & 18.39 & 13.77 & stauffer et al .
( 1998 ) + ppl1 & 22.08 & 19.49 & 14.31 & stauffer et al .
( 1989 ) + ppl2 & 22.12 & 19.32 & 14.37 & stauffer et al .
( 1989 ) + ppl14 & 21.56 & 19.12 & 14.48 & stauffer et al .
( 1994 ) + bry804 & 19.27 & 16.83 & 12.37 & bryja et al .
( 1994 ) + bry816 & 18.83 & 16.61 & 12.20 & bryja et al .
( 1994 ) + gj512b & 13.70 & 12.10 & 8.30 & field star + in the pleiades , we discarded stars with @xmath18@xmath336 and @[email protected] as being too bright and faint , respectively , for the k2 light curves to be reliable . here in praesepe ,
the appropriate limits are less obvious ; we have dropped the brightest ( @xmath18@xmath336 ) and retained the rest .
there are two stars with @xmath18@xmath26 , one of which we determined to be periodic , and both of which are listed in appendix [ app : bright ] .
both of them are discarded from our sample as too bright .
there are 21 stars with 6@xmath2@xmath18@xmath28 ; @xmath18=8 is roughly an f5 spectral type .
at least 11 of them are likely pulsators ( with 6 more likely pulsators that have fainter @xmath18 ) ; see appendix [ app : pulsator ] .
we have left these in the sample to allow for comparison to our pleiades work ( which also included likely pulsators ) , but have identified those pulsators where necessary in the remaining discussion . in the pleiades
, there were very few sources with k2 light curves in the optical cmd below @xmath18=14.5 ; as we will see below in fig .
[ fig : optcmd ] , such a cutoff is not as obvious here .
there are many targets with 14.5@xmath2@[email protected] that have clear periods , and we have left those in the sample .
final star counts are as follows . from an initial sample of 984 candidate members , we find 941 members that are also not too bright . limiting it further , there are 809 stars that we find to be periodic in these k2 data .
figure [ fig : optcmd ] shows color - magnitude diagrams for the stars detected as periodic and not detected as periodic .
the periodic stars for the most part follow the expected main sequence relation for praesepe .
the stars we do not detect as periodic appear to have a less - well - defined main sequence relation , which would be consistent with those stars more likely to be non - members , despite satisfying the membership criteria described in sec . [ sec : membership ] above . in the pleiades
, we determined 92% of the sample to be periodic ( rebull et al .
2016a ) ; here , we obtain 809/941=86% of the member sample to be periodic refers to the fraction of the entire initial sample that is periodic ; the 86% refers to the fraction of members that are periodic . ] .
if we have more non - members inadvertently included in the sample for praesepe than we did for the pleiades ( despite very similar selection methods ) , a lower fraction of periodic stars might be expected .
however , praesepe is considerably older than the pleiades , so the stars are expected to have fewer spots ( hence lower amplitude signals ) and rotate more slowly .
both of these factors would contribute to a lower fraction of detectably periodic members in praesepe .
although many of the praesepe stars , like many in the pleiades , are multi - periodic ( see sec . [
sec : lcpcats ] below ) , we have selected only one @xmath0 and color to be representative of the star in order to investigate the distribution of rotation rates .
now , we explore the overall distribution of @xmath0 , and the distribution of @xmath0 as a function of @xmath1 as a proxy for mass . the distribution of periods we found is shown in figure [ fig : phisto ] ; note that this excludes the stars with periods that were determined to be non - members ( see sec .
[ sec : membership ] and app .
[ app : nm ] ) .
the stars are rotating more slowly , on average , than the analogous figure from the pleiades ( see fig . 3 in rebull et al .
2016a or fig . 9 in rebull et al.2016b ) .
whereas the pleiades is strongly peaked at @xmath21 day , only @xmath420% of the praesepe stars with rotation periods rotate faster than a day , with @xmath442% rotating between 1 and 10 d , and @xmath437% rotating between 10 and 35d .
there seems to be a bimodal distribution of periods in praesepe , with one peak near @xmath41 d and another peak near @xmath410 d. assuming a skumanich law ( skumanich 1972 ) , @xmath35 , assuming that the pleiades is 125 myr and praesepe is 800 myr , we would expect the peak at @xmath40.3 d from the pleiades to become the peak at @xmath41 d in praesepe .
this is the case for the m stars , which compose most of the pleiades peak at @xmath40.3 d and most of the praesepe peak at @xmath41 d. there is a substantial number of praesepe stars , however , with periods near 10 - 15 d , about 35% of the distribution .
the stars contributing to this peak are for the most part more massive than those stars composing the other , faster rotating peak . using the skumanich law
, these stars should correspond to pleiades stars with periods of 4 - 6 d. indeed , about 20% of the pleiades stars with k2 rotation rates have periods greater than about 4 d ; presumably , these go on to populate the slower peak in older clusters .
however , there are fractionally more stars in the slower peak in praesepe , which suggests that this peak includes stars with a larger range of masses than we have assumed in the rough skumanich calculation .
note that the 17 stars we have identified as likely pulsators are still in this sample shown in fig .
[ fig : phisto ] ; their periods do not make a significant difference to the histograms of @xmath36 ( first period , e.g. , the @xmath11 ) or @xmath37 ( secondary period ) , just because so many stars are represented in these plots .
but , they do make a significant difference to the histograms of @xmath38 and @xmath39 ( tertiary and quaternary periods ) , where they are largely responsible for the periods in these histograms .
qualitatively , the morphology of the period vs. color diagram for praesepe , and the evolution of that distribution from young ages to praesepe age , has been well - documented in the literature .
the fact that by praesepe s age g and k dwarfs have a very narrow distribution in period at a given mass was first shown for hyades stars in the 1980s ( duncan et al .
1984 ; radick et al .
1987 ) . that f and early g stars arrive on the main sequence with higher rotational velocities and subsequently spin down on the main sequence was first shown by robert kraft via spectroscopic rotational velocities for stars in a number of open clusters ( kraft 1967 , and references therein ) .
later spectroscopic studies of those same clusters with more modern spectrographs and detectors showed that low mass stars of all masses arrive on the main sequence with a wide range in rotational velocities and that angular momentum loss on the main sequence causes those stars to converge over time to a much narrower range in rotation at a given mass , with the convergence time being longer for lower masses ( stauffer & hartmann 1986 ; stauffer et al .
1987 ; stauffer et al.1989 ) .
the subsequent development of wide - format ccds made it possible to obtain rotation periods for large samples of stars in open clusters , allowing the distribution of rotation rate as a function of mass to be determined for many of kraft s open clusters ( and other clusters ) , largely confirming the spectroscopic results but with better precision and larger samples of stars ; see gallet & bouvier ( 2015 ) and coker et al .
( 2016 ) for a review of the rotation period data and theoretical models of angular momentum loss which attempts to explain the data .
the light curves for open clusters and star - forming regions from k2 builds on this heritage , but also adds the benefits in photometric stability and signal - to - noise provided by space - based observations .
for the open clusters observed with k2 , these data for the first time allow the rotation periods of both stars in most binary systems to be determined and for all cluster members observed the light curves allow an assessment of the shape of the phased light curve , which provides insight into the size and location of star spots on their surfaces .
figure [ fig : pvmk ] shows the relationship between @xmath0 and @xmath1 for praesepe using k2 data . for comparison ,
figure [ fig : pvmk ] also shows the @xmath0 vs. @xmath1 plot for the pleiades k2 data
. there has clearly been significant changes in these distributions between the ages of the pleiades and praesepe .
as we noted above , praesepe stars are on the whole rotating more slowly . however , the m stars in both clusters are primarily rapid rotators . figure [ fig : pvmk2e ] has the @xmath0 vs. @xmath1 plots again for both praesepe and the pleiades , but with color coding to aid in this discussion . in praesepe , there are 3 relatively well - defined and apparently linear sequences . at the blue end
( omitting the pulsators ) , for @xmath40 ( @xmath33f8 ) , @xmath41 .
@xmath0 is changing rapidly over a very small range of @xmath1 .
there is an abrupt transition between the reddest end of this blue sequence and the bluest end of the next sequence ; this is the kraft break ( kraft 1967 ) , where magnetic braking becomes less efficient for bluer stars . in the middle ( yellow portion ) of fig .
[ fig : pvmk2e ] , @xmath42 ( @xmath4f8 to @xmath4m3 ) , @xmath43 . in this regime ,
the points are tightly clumped around this relationship . for @xmath44 , @xmath4m3 to @xmath4m6 , ( which encompasses both the green and red points in fig .
[ fig : pvmk2e ] ) , @xmath45 . here
again , @xmath0 is changing rapidly with @xmath1 .
this end of the distribution contributes substantially to the bimodal nature of the @xmath0 distribution ( fig .
[ fig : phisto ] ) ; there are many stars with @xmath44 and relatively few stars with @xmath46 .
the transition between the blue and yellow regions in fig .
[ fig : pvmk2e ] is abrupt .
the transition between the yellow and green regions is less obvious ; the green points , at least the ones with @xmath47 , could justifiably be included in the linear fit of the yellow points , and the bulk of these points are already consistent with that fit .
the green points , however , are also consistent with the relationship delineated by the red points ( or even a slightly steeper relation ) .
the end of the slow sequence ( yellow through the green points consistent with the relationship ) in praesepe is @[email protected] , or m4 .
figure [ fig : pvmk2e ] also includes the pleiades ( rebull et al.2016a ) for context . the color - coding in fig .
[ fig : pvmk2e ] follows stauffer et al .
( 2016 ) , fig . 24 , and is _ not _ meant to trace exactly the same populations as seen in praesepe , but simply illustrates different sections of the diagram that we call out in the text . in the pleiades
, we discussed the ` slowly rotating sequence ' for @xmath48@xmath1@xmath49 ( @xmath4f5 to @xmath4k9 colors and the corresponding spectral type .
this division might formally be at k7 , but we use ` k9 ' here to indicate succinctly ` just before g0 ' . ] , 2@xmath5011 d ) , which is the blue and yellow points together in fig .
[ fig : pvmk2e ] .
the transition of blue to yellow is placed at the location of the ` kink ' in the slowly rotating sequence ( @xmath4k3 ; see stauffer et al .
the green points delineate a region in which there seems to be a ` disorganized relationship ' between @xmath0 and @xmath1 between @xmath51@xmath1@xmath52 ( @xmath4k9 to @xmath4m3 , 0.2@xmath50 15 d ) .
finally , the red points represent the ` fast sequence , ' with @xmath1@xmath53 ( @xmath34m3 , 0.1 @xmath54 2 d ) .
now between the pleiades ( @xmath4120 myr ) and praesepe ( @xmath4800 myr ) , parts of this diagram see tremendous change ( earlier types ) , and parts show more subtle changes ( later types ) .
all of the g and k ( and even early m ) stars ( yellow in the praesepe plot ) have spun down into a well - defined , linear relationship from the curved relationship ( blue and yellow in the pleiades plot ) .
the ` disorganized region ' ( green in the pleiades plot , and even to some extent the yellow ) is no longer quite so disorganized by praesepe , where many of the earlier m stars have spun down into a relationship consistent with the g and k stars .
the praesepe early m stars are on average slow rotators , and the later m stars are on average more rapid rotators , but there is clear overlap between the rotation rates of early and late m stars ; one can not divide the stars by mass and have them also divided by period . in the pleiades ,
there is less obviously a bimodal distribution in m star periods , and ( in contrast to praesepe ) the division between fast and slow rotators is also roughly a division in mass . the overall m - star relationship ( the red points in both panels in fig .
[ fig : pvmk2e ] ) sees less obvious changes compared to the large changes for the more massive stars , but there are differences for the m stars too .
the slope of the m star relationship between period and color is overall much steeper in praesepe than in the pleiades ; this is most easily seen from the black curved line in fig .
[ fig : pvmk2e ] , which is the relationship for pleiades m stars derived in stauffer et al .
( 2016 ) . in the pleiades , some of the m stars
are still contracting ( still spinning up ) ; this is not the case in praesepe .
the m3 and m4 stars in praesepe have spun down considerably , and they have a shorter contraction time than the m5 and later stars .
the distribution of m5 stars has not changed much between the pleiades and praesepe . in the older cluster , angular momentum loss via wind braking
has had more time to counteract the contraction in the m3 - 4 stars , and the balance between contraction / spin up and angular momentum loss must be different in the m5 ( and later ) stars . the existence of a well - defined slow sequence of late f and even early g stars ( blue points in fig .
[ fig : pvmk2e ] ) in praesepe presumably points to their having at least some amount of angular momentum loss .
if they had no angular momentum loss at all , there would be a scatter in rotation reflecting a range in initial angular momentum and a range in disk lifetimes .
however , presumably their angular momentum loss rate is quite small since they have so little outer convective envelope . as a speculation , it is possible that the f dwarfs in praesepe still have rapidly rotating radiative cores , with their observed rotation periods representing a balance between the angular momentum feeding up from below with the angular momentum lost from their winds .
the g / k / m stars have much larger angular momentum loss rates , and have had time to spin down their cores .
so , the two sequences reflect that dichotomy : core / envelope still decoupled for the f dwarfs , core / envelope coupled for the g , k , and early ms .
the bimodal period distribution of m stars in praesepe ( but not the pleiades ) is interesting because field m stars are found to have a bimodal @xmath0 distribution in the kepler field ( _ e.g. _ , mcquillan et al.2013 , davenport 2016 ) and in nearby m field stars ( _ e.g. _ , newton et al.2016 , kado - fong et al . 2016 ) .
these field stars are older , on average , than praesepe , and the locations of the two peaks are slower ( @xmath419 and @xmath433 d ) compared to praesepe with @xmath41 and @xmath417 d for the @xmath4500 praesepe stars with @[email protected] , the color corresponding to m0 . assuming skumanich evolution , these two peaks can not both evolve together in lock - step .
however , it is interesting that both distributions are bimodal .
it may be that the mechanism which causes some m stars to rapidly spin down ( _ e.g. _ , newton et al.2016 , brown 2014 ) has started to operate in some of the praesepe stars .
we note , of course , that our data can not constrain the behavior of m stars later than about m5 or m6 .
while the relationships delineated by the majority of the praesepe stars in figures [ fig : pvmk ] and [ fig : pvmk2e ] are striking , it is worth looking at some of the outliers in this distribution .
notes on specific stars appear in appendix [ app : pvmkoutliers ] , as well as optical color - magnitude and @xmath0 vs. @xmath1 diagrams with these objects highlighted . the g and k stars in young open clusters like the pleiades have bimodal rotational velocity distributions , with a majority of stars on a slowly rotating branch and a minority in a rapidly rotating branch .
the latter stars are generally believed to descend from pre - main - sequence stars that lost their circumstellar disks ( and hence their ability to rapidly shed angular momentum ) very early .
our @xmath0 vs. color plot also shows several rapidly rotating g / k stars .
are these the descendents of the rapidly rotating g / k stars in the pleiades ? for most or all of these stars , we believe not . instead
, they are best explained as tidally locked short - period binaries . most of them are known short - period binaries ; many of the remaining ones have little to no spectroscopic information , but tidally locked binaries are a logical explanation for stars that are rotating too quickly in comparison to the other stars of similar colors
. the stars with periods much longer than average for their @xmath1color are also curious .
for a few of them , tidal synchronization ( or pseudo - synchronization ) in a short - period binary may be the explanation , as has been advocated for similar outliers in m35 ( meibom et al .
2006 ) and other clusters .
because of the steep dependence of the synchronization time on the binary - star separation , for the longer period outliers ( @xmath5620 days ) , this becomes a less viable possibility .
if these stars are all praesepe members , and if we are correct that all of them are rotation periods , then something about these stars made them spin down more than most other stars in the cluster .
for these longest - period stars , interpretation of their lcs as rotation periods is more fraught than other shorter - period stars , as there is more variation from cycle to cycle , and it becomes more of a judgement call as to whether or not the star has a @xmath11 , or just a repeated pattern that may or may not be tied to rotation .
slowly rotating stars may undergo more significant spot evolution , so this may be a real astrophysical effect .
having fewer complete cycles within the k2 campaign , coupled with significant changes every cycle , makes it hard to assess .
many of the long-@xmath0 g and k stars from the bulk of the distribution have more obviously sinusoidal lcs , and many of the longest @xmath0 m stars ( ones that are still with the bulk of the distribution ) share lc characteristics with these long-@xmath0 outlier g and k stars ; there is a continuum of lc properties such that it is not always easy to draw a line between @xmath11 and just a repeated pattern . specifically because of this ambiguity , we obtained follow - up keck / hires spectroscopy of many of these very slowly rotating stars ( see app . [ app : keck ] ) .
of the ones we observed , about half of them have radial velocities inconsistent with cluster membership .
we identified these as non - members ( sec .
[ sec : membership ] and app .
[ app : nm ] ) .
four of the longest @xmath0 objects ( for which we have yet to obtain spectra ) appear just below the single - star main sequence , suggesting that they may also be non - members ; given the uncertainties in @xmath1 , we have provisionally left them in the list of members ( see app . [ app : pvmkoutliers ] ) .
in rebull et al . ( 2016b ) , we presented a set of empirical structures in the k2 pleiades lcs and periodograms , finding via visual inspection 11 different categories .
these categories are discussed in detail in rebull et al .
( 2016b ) , and we do not repeat that discussion here .
all but one of the categories of light curves can be found in praesepe as well ; see table [ tab : countclasses ] for a list .
many of the categories have more than one significant period .
figure [ fig : multifreq ] shows some examples of these classes in praesepe , two of which have more than one measured @xmath0 ; these examples span a range of brightnesses , periods , and categories .
( for examples of each of the categories in the pleiades , see rebull et al .
2016a . )
before we can compare the total counts of objects in the various periodogram categories , we need to be sure that we are sampling the same range of masses in the two clusters .
figure [ fig : vkdistrib4 ] shows the sample fraction for the periodic member samples as a function of @xmath1 as a proxy for mass .
the two samples are comparable over most of the range of @xmath1 ; there are sample completeness effects at the reddest and bluest bins , which are the most poorly populated .
the most notable differences are at the bluest end ( @xmath1@xmath331 ) , which will affect primarily stars whose measured k2 periods are most likely to be pulsation rather than rotation . now with the knowledge that the samples are of comparable mass ranges , we can compare sample fractions of lc / periodogram categories .
table [ tab : countclasses ] summarizes the total counts of objects in each lc / periodogram category for praesepe and the pleiades ( in both cases , only cluster members are included ) .
there are some significant differences between the clusters .
overall , the fraction of stars that are singly periodic and multiply periodic are roughly comparable in the two clusters ( @xmath475 - 80% and @xmath420 - 25% , respectively ) . however , there is a slightly higher fraction of singly periodic sources ( and slightly lower fraction of multi - periodic sources ) in praesepe . in the context of this work
, we identified an inconsistency in the way that we were identifying close and distant resolved peaks in our earlier pleiades work . as stated there , we calculated @xmath57 ( see also sec . [
sec : deltap ] below ) , and used that value to identify close ( @xmath580.45 ) and distant ( @xmath590.45 ) peaks .
however , we went on to say ( without explanation ) that some objects could be identified as both close and distant peaks .
we now more consistently identify an object as both close and distant peaks if the difference in any two periods divided by the @xmath11 is @xmath20.45 for two of at least three periods , and @xmath30.45 for ( a different ) two of at least three periods ( unless the star is already identified as a pulsator , in which case it is not identified as either close or distant peaks ) . as a result
, three objects from the pleiades that were already identified as resolved close peaks should also have been tagged as resolved distant peaks ( epic 210877423 , 211112974 , and 211128979 ) .
this has been corrected in the statistics in table [ tab : countclasses ] .
lcccccc periodic & 809 & 0.86 & 1.00 & 759 & 0.92 & 1.00 + single period & 645 & 0.69 & 0.80 & 559 & 0.68 & 0.74 + multi - period & 164 & 0.17 & 0.20 & 200 & 0.24 & 0.26 + double - dip & 163 & 0.17 & 0.20 & 107 & 0.13 & 0.14 + moving double - dip & 121 & 0.13 & 0.15 & 31 & 0.04 & 0.04 + shape changer & 297 & 0.32 & 0.37 & 114 & 0.14 & 0.15 + orbiting clouds ?
& 0 & 0 & 0 & 5 & @xmath20.01 & @xmath20.01 + beater & 77 & 0.08 & 0.10 & 135 & 0.16 & 0.18 + complex peak & 68 & 0.07 & 0.08 & 89 & 0.11 & 0.12 + resolved , close peaks & 68 & 0.07 & 0.08 & 126 & 0.15 & 0.17 + resolved , distant peaks & 71 & 0.08 & 0.09 & 39 & 0.05 & 0.05 + pulsator & 17 & 0.02 & 0.02 & 8 & 0.01 & 0.01 + among the lc / periodogram categories , there is a higher fraction of moving double - dip and shape changers in praesepe than there are in the pleiades ; there are more than twice the fraction of shape changers as in the pleiades ( @xmath440% vs. @xmath415% ) . in rebull
et al.(2016b ) and stauffer et al .
( 2016 ) , we suggested that the shape changers and moving double dips are due to latitudinal differential rotation and/or spot / spot group evolution .
these lc types were found primarily in the slower rotators in the pleiades ; with more slower rotators in praesepe , it may not be surprising that we have more shape changers , because this may reflect a real difference in the incidence rate of differential rotation . because there is a slightly lower fraction of multi - period sources in praesepe
, there is a lower fraction of nearly all the multi - period subcategories in praesepe .
however , the most discrepant fractions are found in the resolved close peaks category , where the fraction is half what it was in the pleiades .
we postulated in rebull et al .
( 2016b ) and stauffer et al .
( 2016 ) that the resolved peaks categories could be a result of latitudinal differential rotation and/or spot / spot group evolution among the g and k stars .
the shape changers , moving double dips , and close resolved peaks ( in the g and k stars ) may all be subject to an observational bias in the following sense
. we noted in rebull et al .
( 2016b ) that , particularly for the shape changers / moving double dips , it was possible that if there had been much more data , @xmath6070 d , encompassing more cycles , then it might have been possible to differentiate the two ( or more ? ) periods contributing to the changing shape .
in praesepe , since the periods are on average longer than in the pleiades , consequently there are ( on average ) fewer complete cycles encompassed in the k2 campaign .
thus , perhaps the higher occurrence rate of shape changers / moving double dips and the lower occurrence rate of resolved close peaks ( for g and k stars ) in praesepe may be attributable to the number of complete cycles available for stars in praesepe ( as compared to the pleiades ) . for the m stars in the pleiades
, we postulated that the close resolved peaks were binaries , since most of them appeared above the single - star main sequence .
the distant resolved peaks we thought were most likely to be binaries .
the same basic result is true in praesepe as well the m stars with resolved peaks are above the single - star main sequence ( see sec .
[ sec : pvmkmultip ] below ) . from theory
, we expect stronger differential rotation in earlier stars ( _ e.g. _ , kitchatinov & olemskoy 2012 ) , and effectively solid body rotation in the m stars .
praesepe has a comparable if not slightly greater fraction of resolved distant peaks , though many more instances where the difference between the peaks is very large indeed ( @xmath36 d ; see [ sec : deltap ] below ) .
the phased light curves of most spotted stars are simple , showing a morphology that is more or less sinusoidal or one that has one or two broad humps " or dips .
that is as expected , because spots located at most positions on a stellar surface and when viewed from most vantage points will be visible for half or more of the rotation period .
that means that most spots will have a contribution to the phased light curve that spans 180@xmath7 or more in phase .
flux dips or other structures that cover less than 90@xmath7 in phase are very hard to produce with spots .
the only way they can be produced is by placing the spot very near the upper or lower limb of the star . because of limb darkening and geometric fore - shortening , light from such a location contributes little to the total integrated brightness of the star ; therefore , flux dips from such spots can not yield light curve features that are very deep . in the pleiades
( rebull et al.2016b ) , we nevertheless identified six stars with short - duration flux dips ( full width at zero intensity , fwzi , @xmath20.2 in phase ) ; one star had three such dips , another had two , and the remaining four had just one flux dip .
the dips were to first order constant in shape and depth over the duration of the k2 campaign .
all six stars were mid - m dwarfs with very short ( @xmath610.7 day ) periods .
we identified another 19 rapidly rotating , mid - to - late m dwarfs in the 8 myr old upper sco association whose k2 phased light curves seemed to show more structure than could be explained by spots ( stauffer et al . 2017 ) .
we have attributed the light curve structure to warm clouds of coronal gas orbiting the stars near their keplerian co - rotation radius , following models by jardine & van ballegooijen ( 2005 ) and townsend & owocki ( 2005 ) .
we have found no praesepe stars with the same properties ( short - duration flux dips or other structure , rapid rotation , mid - to - late m dwarf spectral type ) .
however , there are three slowly rotating praesepe m dwarfs ( @xmath1@xmath45 ) that do have more structured light curves than we expect ; these are shown in figure [ fig : batwingy ] : 212011416/2massj08330845 + 2026372 , 211915940/js208 , and 211931651/ad3196=cp cnc .
one ( epic 212011416/2massj08330845 + 2026372 ) has a nearby saturated column in the k2 data , but multiple lc versions obtain the same shape and ( large ! ) depth of the features .
note that in the first two cases , the highest peak in the periodogram results in a phased light curve that has more scatter than when phased at three times the peak value ; this is similar to the double - dips ( see discussion in rebull et al.2016b ) .
these stars have periods @xmath3410@xmath62 longer than the stars in pleiades or usco .
we do not believe the physical mechanism producing these light curve morphologies is the same as for the pleiades and upper sco . because the flux dips are broader than in the pleiades and are generally low amplitude , it is ( barely ) possible that spots could be responsible .
however , the full amplitude of one of them ( epic 212011416 ) is about 4% , which would be very difficult to produce using spots located near the limb of the visible hemisphere . in order to help constrain the nature of these stars
, we obtained single - epoch spectroscopy of these sources with keck / hires ( see app . [ app : keck ] ) .
all are consistent with being radial velocity members of praesepe , and all are narrow - lined ( and single - lined ) .
the phased light curve morphologies for these three stars are much more complex than most of the rest of the lcs in praesepe .
because they represent a tiny minority of the praesepe sample , we do not believe that our failure to understand their properties should affect any of the other conclusions in the paper .
we plan a future paper ( hebb et al . ) that will discuss modelling of these lcs .
kovcs et al .
( 2014 ) also identified lc / power spectrum classes in their study of praesepe .
they had ( a ) monoperiodic sinusoids ( 13% of their periodic sample ) ; ( b ) single but unstable sinusoid ( 39% ) ; ( c ) two peaks interpreted as latitudinal differential rotation ( 7% ) ; ( d ) power moved into first harmonic ( 41% ) .
\(a ) the monoperiodic sinusoids of their periodic sample can be matched to those in our sample with single periods but that are not double - dip nor shape - changer .
there are 366 of those , or @xmath445% of the periodic sources , in our member sample .
( b ) single but unstable sinusoids would be analogous to our shape changers , without the double - dips ; we find 164 of them , 20% of the periodic sources .
( c ) they rarely find two discrete peaks in the power spectrum ; this would be analogous to our resolved close and distant peaks combined , and we have 15% of our periodic sample falling into this category .
we agree that in some cases , at least , this could be attributable to latitudinal differential rotation , but we suspect ( also see discussion in rebull et al .
2016b ) that at least some ( most of the multi - peaked m stars ) are binaries .
lastly , their category where power is moved into the first harmonic is analogous to our double - dip category , which is 20% of our periodic sample . perhaps a more fair comparison is to work just with the 152 stars that are in common between the two studies . for that sample ,
we obtain ( a ) monoperiodic sinusoids : 6% ; ( b ) single unstable sinusoids : 23% ; ( c ) resolved peaks : 16% ; and ( d ) double dips : 47% . the closest match in terms of sample fraction
is this last category . assuming that we have correctly captured the relationship between our classes and those in kovcs et al .
( 2014 ) , we have roughly similar sample fractions for the same characteristics .
in the prior section , we identified the @xmath420% of stars with multiple periods . in this section ,
we focus on where the multi - periodic sources fall with respect to the single - period sources in a variety of parameter spaces .
we compare the single- and multi - period sources within praesepe , and also to our analysis of the same phenomenon in the pleiades ( from rebull et al .
2016b ) .
we calculated the amplitude of the periodic signal in the same fashion as we did in the pleiades ; we assembled the distribution of all points in the light curve , took the log of the 90th percentile flux , subtracted from that the log of the 10th percentile flux , and multiplied by 2.5 to convert to magnitudes .
figure [ fig : ampldistrib ] plots that amplitude against both @xmath0 and @xmath1 for the periodic light curves .
note that this is not necessarily the amplitude of a sinusoid overlaid or fit to the periodic signal , but the amplitude of the overall light curve , which necessarily includes long - term trends . in the pleiades
, there was no obvious trend of amplitude with color or period .
however , in praesepe , there is a trend in both of these panels . on the left of fig .
[ fig : ampldistrib ] , longer period stars have lower amplitudes .
this is consistent with expectations in that , certainly by this age , more slowly rotating stars should be less active and therefore have smaller spots .
note , however , that nearly all of the turn - down at the longest periods are higher - mass g and k stars ; likewise , the shorter @xmath0 stars are primarily mid to late - m stars .
the bimodal distribution of periods seen in fig .
[ fig : phisto ] is apparent here too .
there are more multi - period low - amplitude sources than single period low - amplitude sources ( though several of these are likely pulsators , which explains the low amplitude in those bluest and smallest @xmath0 cases ) . on the whole
, however , there is no clear distinction in the left panel of fig .
[ fig : ampldistrib ] between the distributions of singly and multi - periodic sources . on the right of fig .
[ fig : ampldistrib ] , it can be seen that bluer stars ( corresponding to @[email protected] , blue points in fig .
[ fig : pvmk2e ] ) are generally multiperiodic and have smaller amplitudes , not all of which are likely pulsators .
there are several lower amplitude lcs with colors redder than @[email protected] .
this , and the trend towards larger amplitudes at even redder colors ( @[email protected] ) , could be an observational bias in that stars need to be ` bright enough ' for a period to be derivable ; fainter stars need larger amplitudes to be seen as periodic , and brighter stars with smaller amplitudes are more easily detected .
note , however , that there is substructure within the right panel of fig .
[ fig : ampldistrib ] there is a ` clumping ' of the distribution for 1.3@xmath63@[email protected] ( corresponding to the yellow points in fig .
[ fig : pvmk2e ] ) that moves to lower amplitude as the color gets redder .
then , it turns around and moves to larger amplitude as the color gets redder for @[email protected] ( the green / red points in fig .
[ fig : pvmk2e ] ) . for _
both _ of these color regimes , the amplitude gets smaller as the period gets longer .
this makes sense in the standard rotation - activity sense , if the spot filling factor ( at least for the non - axisymmetrically distributed component ) decreases for longer periods . in the pleiades
, we found a strong correlation between multiple periodicities and @xmath1 most of the earlier stars were multiply periodic and nearly all the later stars were singly periodic .
the distribution in praesepe is different than in the pleiades ; see figure [ fig : vkdistrib ] . through most of the sample ,
the fraction that has multiple periods is roughly constant with color .
clearly the bins for @xmath1@xmath33 1 and @[email protected] are significantly affected by sample completeness .
there is a transition between where multiple periods dominate to where single periods dominate , and it is at @[email protected] ( @xmath4g0 ) .
the pleiades sample extends to bluer colors , and the transition between where multiple periods dominate to where single periods dominate is at @[email protected] ( early k ) .
inclusion of the likely pulsators in this analysis affects the results , particularly because of the sample differences for @xmath1@xmath331 ( see fig .
[ fig : vkdistrib4 ] ) . dropping the likely pulsators
causes the sample fraction of multi - period sources to plummet blueward of @xmath1@xmath41 .
however , the multi - period sources still dominate for 1@xmath33@[email protected] .
if latitudinal differential rotation dominates blueward of this @[email protected] transition point ( which was one of our hypotheses in the pleiades for the transition point there ) , then this transition has moved to more massive types by the age of praesepe .
this transition is also essentially where the bluest ` branch ' turns down in the @xmath0 vs.@xmath1 diagram in fig .
[ fig : pvmk ] above , again suggesting that something physically different happens blueward of this color .
stronger differential rotation is expected for hotter stars ( _ e.g. _ , kitchatinov & olemskoy 2012 ) , so this may be the dominant effect .
however , identification of a star as multi - period at all may be limited by the number of complete cycles in the k2 campaign , as discussed in sec .
[ sec : comparetypestopleiades ] . figure [ fig : pvmk3freq ] shows where the multi - period stars fall in the @xmath0 vs. @xmath1 and @xmath18 vs. @xmath1 parameter spaces .
this figure is in direct analogy to rebull et al .
( 2016b ) , figure 12 ; in the pleiades , most of the early - type stars were multi - periodic , and the later types were nearly all singly periodic , and the multi - periodic later type stars were nearly all photometric binaries . in praesepe ,
the single - period stars are distributed more uniformly throughout the diagram , consistent with fig .
[ fig : vkdistrib ] .
the earlier type stars are dominantly multiply periodic for @[email protected] - 2 , which is a smaller range than for the pleiades , again consistent with fig .
[ fig : vkdistrib ] .
it is still true that most of the multi - period stars with @[email protected] - 4 seem to be dominated by photometric binaries , given the position in the @xmath18vs .
@xmath1 diagram .
this is consistent with the m stars still rotating as solid bodies at praesepe age .
the panels in figure [ fig : optcmdclasses ] break down where the individual lc classes appear in the @xmath18 and @xmath0 vs. @xmath1 diagrams ; the analogous pleiades figures in rebull et al .
( 2016b ) are figs .
13 and 14 .
beaters and complex peaks were both , in the pleiades , found to dominate the stars with @[email protected] . in praesepe , the region where they dominate has shifted blueward , to @[email protected] - 3 , consistent with the discussion associated with fig .
[ fig : vkdistrib ] above . in both cases , beaters that are
also m stars are more likely to be binaries than single stars , based on position in the cmd .
shape changers dominated in the pleiades for 1.1@xmath2@[email protected] , with a significant fraction of the stars in the ` disorganized region ' ( with 3.7@xmath2@xmath1@xmath25 ) being in this category . in praesepe , there are a higher fraction of shape changers , and they dominate the ` middle branch ' ( yellow points in fig . [
fig : pvmk2e ] ) with 1.3@xmath63@[email protected] .
if the shape changers reflect rapid spot evolution and/or differential rotation , this happens more frequently in praesepe than it does in the pleiades .
note that the stars that exhibit shape changing behavior in praesepe are on average rotating much more slowly than in the pleiades .
the longer periods on the slowly rotating branch for 1.3 @xmath2@xmath1@xmath2 4.5 for praesepe relative to pleiades combined with the fixed campaign length for k2 makes it likely that stars will be moved from the resolved close peak category into the shape - changer or moving double - dip category , explaining at least some of the differences in light curve class distributions we see .
the lcs with two distinct periods can be found throughout the @xmath0 vs. @xmath1 diagram for both clusters , and , for m stars , tend to be on the brighter side of the cluster distribution in the cmd . for the m stars in particular , these are more likely to be binaries
. resolved distant peaks are distributed throughout the @xmath0 vs. @xmath1 diagram , but resolved close peaks tend to cluster in the bluest portion of the @xmath0 vs. @xmath1 diagram ( and to some extent in the m stars fast - rotating clump ) .
resolved close peaks and bluer colors may be differential rotation and/or spot evolution .
( see sec .
[ sec : deltap ] below . ) finally , moving double - dip stars dominate the ` middle branch ' ( yellow points in fig .
[ fig : pvmk2e ] ) of the praesepe @xmath0 vs. @xmath1 diagram to a much larger extent than in the pleiades
. there are fractionally many more moving double - dip stars in praesepe . there are fractionally far fewer m stars exhibiting any double - dip behavior in praesepe than in the pleiades .
finally , we calculated the @xmath66 metric for stars with resolved multi - period peaks in praesepe . in the pleiades work ( rebull et al .
2016b ) , we took the closest two periods out of those detected , subtracted the smaller from the larger , and divided by the period we take to be the rotation period of the primary star in the system . in this calculation
, it could be that the rotation period is not involved in the numerator .
we have refined our thinking on this calculation , and now take the closest peak to the @xmath11 , subtract the smaller from the larger , and divide by the @xmath11 .
this change ties the @xmath66 metric more closely to the rotation period of the star .
our previous calculation is effectively identical to this calculation for all but 5 stars in the pleiades multi - period sample , though it affects 15% of the praesepe stars with at least two periods ( which is half of those with at least three periods . )
the plot of this corrected @xmath66 vs. @xmath0 for praesepe and the pleiades is shown in figure [ fig : deltap ] .
( in both clusters , most of the likely pulsators are in a different region of the diagram than other stars . )
the morphology of this diagram in praesepe is missing two prominent features from the pleiades version of this figure . in the pleiades
, there was a very obvious linear feature in the lower right .
some of this was a selection effect in that points in the lower right are harder to observe ( harder to distinguish periods and thus harder to obtain two periods that can be used to place them in this plot ) . in praesepe ,
the feature is not as obvious .
we suspected that in the pleiades , this feature was a signature of differential rotation in stars primarily located on the slow sequence . as shown in fig .
[ fig : pvmk3freq ] , praesepe has very few multiperiod stars on the slow sequence , and for the k and m dwarfs on that sequence , nearly all the multiperiod stars appear to be binaries ( based on their location in the cmd ) . as discussed in sec .
[ sec : comparetypestopleiades ] , the relative paucity of resolved close peak and moving double dip stars on the slow sequence in praesepe ( the stars that primarily populate the linear sequence in the pleiades @xmath67 vs. @xmath0 plot ) is likely a result of their longer periods we simply do not have a long enough sequence of data to resolve two periods in the periodogram .
the second prominent feature in fig . [ fig : deltap ] that is present in the pleiades but absent in praesepe is a clump of stars near @xmath680.4 and @xmath690.3 .
those points are primarily binary m dwarfs .
we believe the lack of a well - defined peak of binary m dwarfs in praesepe is primarily a result of the dependence of angular momentum loss on mass and period .
( see figures [ fig : pvmk2e ] and [ fig : pvmk3freq ] . )
if the praesepe m dwarf binaries were drawn from the same parent distribution at pleiades age , they would have started with very few stars with periods more than 5 days .
figure [ fig : pvmk2e ] shows that many of those stars must spin down to periods @xmath310 days by praesepe age while others still maintain quite rapid rotation .
therefore , it is natural that the praesepe m dwarf binaries will have a much large range in period and a much larger range in @xmath70 than in the pleiades . indeed , praesepe has many more stars for which there are two very different periods .
those instances where @xmath65 is greater than 6 days are highlighted in the figure ; they clump in two different locations depending on whether the shorter or the longer period was taken to be the @xmath11 for the primary .
( there are only two such points in the pleiades with @xmath71 . )
none of these praesepe @xmath71 points are particularly outliers in the @xmath0 vs. @xmath1 diagram ; they are distributed between 1.8@xmath2@[email protected] within where the bulk of the distribution is located .
the five stars near @xmath681 , @xmath693 - 20 are in the m star fast rotating sequence , and the remainder are within the top , slowly rotating sequence .
they are not particularly clustered in the brighter portion of the cluster sequence in the cmd , so they are not necessarily photometric binaries .
however , it is hard to imagine a situation in which latitudinal differential rotation results in a @xmath65 of 6 days or more from pole to equator , and we suspect these are all binaries .
the angular momentum loss mechanism operating in these stars is probably a function of both period and stellar mass . for the stars with @xmath1@xmath344
, these could be examples of systems where one star has spun down and the other has not ; the difference in periods in these binaries has significantly increased since the age of the pleiades .
these pairs could perhaps shed light on the mechanism that causes some m stars to rapidly spin down ( see , _
e.g. _ , newton et al.2016 ) .
we have presented our analysis of nearly a thousand k2 lcs of praesepe members , increasing the known praesepe rotation periods by nearly a factor of four ; 809 of the 941 members ( 86% ) with k2 lcs have a measured period from the k2 data .
the distribution of @xmath0 vs. @xmath1 , a proxy for mass , has three different regimes : @[email protected] ( @xmath33f8 ) , where the rotation rate rapidly slows as the mass decreases ; 1.3@xmath2@[email protected] ( @xmath4f8 to @xmath4m3 ) , where the rotation rate slows more gradually as the mass decreases ; @[email protected] ( @xmath34m3 ) , where the rotation rate rapidly increases as the mass decreases . particularly in this last , primarily m star regime
, there is a bimodal distribution of periods , with few between @xmath42 and @xmath410 days .
we interpret this to mean that once m stars start to slow down , they do so rapidly ; this is likely a predecessor of the bimodal distribution of m star rotation rates found in much older field stars ( _ e.g. _ , newton et al .
2016 , kado - fong et al .
2016 ) .
the distribution of @xmath0 vs. @xmath1 exhibits significant changes between the pleiades ( @xmath4125 myr ) and praesepe ( @xmath4790 myr ) . for the late f , g , k , and early m stars ,
the overall distribution slows considerably compared to the pleiades , and the higher mass branch ( late f and earliest g ) steepens significantly .
the transition at @xmath1=1.3 is the kraft break .
the g and k and early m stars have a more shallow relationship , with the lower masses rotating more slowly . for the mid m stars , the period - color relationship changes relatively little between the pleiades and praesepe , though it is steeper in praesepe than the pleiades .
overall these late - type stars rotate significantly faster as mass decreases , but there is more than an order of magnitude spread in the rotation rates in praesepe at any given mass .
we found the same diversity of lc and periodogram classes in praesepe as we did in the pleiades . about three - quarters of the periodic member stars in both clusters have only one period ; the rest have multiple periods .
praesepe has a higher fraction of lc classes we have interpreted as latitudinal differential rotation and/or spot or spot group evolution , but this may be influenced by the observing window ; pleiades stars rotate faster , so there are more complete cycles observable in the @xmath470 d k2 campaign for the pleiades stars than for praesepe stars .
multi - periodic stars dominate the bluest end of the sample ; the transition between where multi - periodic stars dominate over single - period stars happens at a bluer color in praesepe ( @[email protected] ) than it does in the pleiades ( @[email protected] ) .
m stars in both clusters that have multiple periods are likely to be binaries . in praesepe , there are many more lcs that have two widely separated periods , @xmath726 days
. some of these could be examples of systems where one m star has spun down but the other has not .
k2 continues to revolutionize the study of rotation in young and intermediate age open clusters .
some of the data presented in this paper were obtained from the mikulski archive for space telescopes ( mast ) .
support for mast for non - hst data is provided by the nasa office of space science via grant nnx09af08 g and by other grants and contracts .
this paper includes data collected by the kepler mission .
funding for the kepler mission is provided by the nasa science mission directorate .
this research has made use of the nasa / ipac infrared science archive ( irsa ) , 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 ( ads ) abstract service , and of the simbad database , operated at cds , strasbourg , france .
this research has made use of data products from the two micron all - sky survey ( 2mass ) , 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 .
the 2mass data are served by the nasa / ipac infrared science archive , which is operated by the jet propulsion laboratory , california institute of technology , under contract with the national aeronautics and space administration .
this publication makes use of data products from the wide - field infrared survey explorer , which is a joint project of the university of california , los angeles , and the jet propulsion laboratory / california institute of technology , funded by the national aeronautics and space administration .
adams , j. , stauffer , j. , skrutskie , m. et al .
2002 , , 124 , 1570 aigrain , s. , hodgkin , s. , irwin , m. , lewis , j. , roberts , s. , 2015 , , 447 , 2880 aigrain , s. , parvianinen , h. , pope , b. , 2016 , , in press ( arxiv:1603.09167 ) ageros , m. , covey , k. , lemonias , j. , et al . , 2011 , , 740 , 110 akeson , r. , et al . , 2013 , , 125 , 989 an , d. , terndrup , d. , pinsonneault , m.,et al .
2007 , 655 , 233 baker , d. , jameson , r. , casewell , s. et al .
2010 , , 408 , 2457 barrado y navascues , d. , stauffer , j. , randich , s. , 1998 , , 506 , 347 blanco - cuaresma , s. , soubiran , c. , heiter , u. et al .
2015 , , 577 , 47 boesgaard , a. , roper , b. , & lum , m. 2013 , , 775 , 58 bonatto , ch . ,
bica , e. , girardi , l. et al .
2004 , , 415 , 571 boudreault , s. , lodieu , n. , deacon , n. & hambly , n. , 2012 , , 426 , 3419 brandt , t. , & huang , c. 2015 , , 807 , 24 breger , m. , 1970 , , 162 , 597 breger , m. , hareter , m. , endl , m. , et al . , 2012 , an , 333 , 131 brown , t. , 1991 , , 368 , 599 brown , t. , 2014 , , 789 , 101 bryja , c. , jones , t. , humphreys , r. , lawrence , g. , pennington , r. , zumach , w. , 1992 , , 388 , 23 bryja , c. , humphreys , r. , jones , t. , 1994 , , 107 , 246 chambers , k. , magnier , e. , metcalf , n. , et al . , 2016 , arxiv1612.05560 choi , h .- j .
, soon , w. , donahue , r. , baliunas , s. , henry , g. , 1995 , , 107 , 744 cody , a. m. , hillenbrand , l. , david , t. , carpenter , j. , everett , m. , & howell , s. , 2017 , in press , arxiv:1612.05599 coker , c. , pinsonneault , m. , & terndrup , d. 2016 , , 833 , 122 davenport , j. , 2016 , arxiv:1610.08563 delorme , p. , collier cameron , a , hebb , l. , et al . , 2011 , , 413 , 2218 douglas , s. , ageros , m. , covey , k. , et al .
, 2014 , , 795 , 161 duncan , d. , baliunas , s. , noyes , r. et al . 1984 , , 96,707 eggen , o. 1992 , , 104 , 1482 friel , e. & boesgaard , a. 1992 , , 387 , 1 , 70 gaia collaboration , prusti , t. , de bruijne , j.h.j .
, et al . , 2016 , arxiv:1609.04172 gallet , f. , & bouvier , j. 2015 , , 577 , a98 hauck , b. , 1971 , , 11 , 79 harris , g. 1976 , 30 , 451 henden , a. , templeton , m. , terrell , d. , smith , t. c. , levine , s. , welch , d. , 2016 , ycat , 2336 , 0 holland , k , jameson , r. , hodgkin , s. , davies , m. , & pinfield , d. , 2000 , , 319 , 956 howell , s , et al . , 2014 , , 126 , 398 janes , k. & phelps , r. 1994 , , 108 , 1773 jardine , m. & van ballegooijen , a. 2005 , , 361 , 1173 .
jones , b.f . & cudworth , k. 1983 , , 88 , 215 jones , b.f . &
stauffer , j. 1991 , , 102 , 1080 kado - fong , e. , williams , p. , mann , a. , et al .
, 2016 , , 833 , 281 khalaj , p. & baumgardt , h. , 2013 , , 434 , 3236 kitchatinov , l. , & olemskoy , s. , 2012 , , 423 , 3344 klein - wassink , w. 1927 , pgro , 41 , 1 kovcs , g. , hartman , j. , bakos , g. , et al . , 2014 , , 442 , 2081 kraft , r. , 1967 , , 150 , 551 kraus , a. & hillenbrand , l. 2007 , , 134 , 2340 libralato , m. , nardiello , d. , bedin , l. r. , et al . ,
2016 mnras , 463 , 1780 luger , r. , agol , e. , kruse , e. , et al .
2016 , , 152 , 100 ( arxiv:1607.00524 ) mann , a. , gaidos , e. , mace , g. , et al . , 2016a , , 818 , 46 mann , a. , gaidos , e. , vanderburg , a. , et al . , 2016b , , arxiv:1609.00726 mcquillan , a. , aigrain , s. , & mazeh , t. , 2013 , , 432 , 1203 melis , c. , reid , m. , mioduszewski , a. , stauffer , j. , bower , g. , 2014 , science , 345 , 1029 mermilliod , j .- c .
1981 , , 97 , 235 mermilliod , j .- c.,weis , e. , duquennoy , a. , mayor , m. , 1990 , , 235 , 114 mermilliod , j .- c . , & mayor , m. , 1999 , , 352 , 479 mermilliod , j .- c . , mayor , m. , udry , s. , 2009 , , 498 , 949 newton , e. , irwin , j. , charbonneau , d. , berta - thompson , z. , dittman , j. , west , a. , 2016 , , 821 , 93 obermeier , c. , henning , t. , schlieder , j. , et al . , 2016 , , in press arxiv 1608.04760 pace , g. , pasquini , l. , & francois , p. 2008
, , 489 , 403 pasquini , l. , de medeiros , j. , & girardi , l. 2000 , , 361 , 1011 paulson , d. , sneden , c. & cochran , w. 2003 , , 125 , 3185 perryman , m. , brown , a. , lebreton , y. et al . 1998 , , 331 , 81 pepper , j. , et al . ,
2017 , submitted pope , b. , parviainen , h. , aigrain , s. , 2016 , , 461 , 3399 radick , r. , thompson , d. , lockwood , g. et al .
1987 , , 321 , 459 rebull , l. , stauffer , j. , cody , a. , 2016a , , submitted rebull , l. , stauffer , j.,bouvier , j. , 2016b , , submitted salaris , m. , weiss , a. , & percival , s. 2004 , , 414 , 163 scholz , a. , irwin , j. , bouvier , j. , sipcz , b. , hodgkin , s. , eislffel , j. , 2011 , , 413 , 2595 schwarzschild , e. & hertzsprung , e. 1913 , astr .
196 , 9 skrutskie , m. , cutri , r. m. , stiening , r. , et al . , 2006 , , 131 , 1163 skumanich , a. , 1972 , , 171 , 565 soderblom , d. , laskar , t. , valenti , j. , stauffer , j. , rebull , l. , 2009 , , 138 , 1292 stauffer , j. 1982 , , 94 , 678 stauffer , j. , hamilton , d. , probst , r. , rieke , g. , mateo , m. , 1989 , , 344 , 21 stauffer , j. , hamilton , d. , probst , r. , , 108 , 155 stauffer , j. & hartmann , l. 1986 , , 98 , 1233 stauffer , j. , hartmann , l. , & latham , d. 1987 , , 320 , l51 stauffer , j. , hartmann , l. , & jones , b. 1989 , , 346 , 160 stauffer , j. , schultz , g. , & kirkpatrick , j. d. , 1998 , , 499 , 199 taylor , b. 2006 , , 132 , 2453 townsend , r. & owocki , s. 2005 , , 357 , 251 .
upgren , a. , weis , e. , & deluca , e. 1979 , , 84 , 1586 van leeuwen , f. , 2009 , , 497 , 209 vandenberg , d. & bridges , t. 1984 , , 278 , 679 vanderburg , a. , & johnson , j. , 2014 , , 126 , 948 vogt , s. , allen , s. , bigelow , b. , et al . , 1994 ,
spie , 2198 , 362 wang , p. , chen , w. , lin , c. et al .
2014 , , 784 , 57 weis , e. 1981 , , 93 , 437 werner , m. , roellig , t. , low , f. , et al . , 2004 ,
apjs , 154 , 1 whelan , j. , worden , p. , & mochnacki , s. , 2073 , , 183 , 133 wright , e. , eisenhardt , p. r. m. , mainzer , a. k. , et al . , 2010 , , 140 , 1868 yang , x. , chen , y. , & zhao , g. 2015 , , 150 , 158 zacharias , n. , finch , c. , subasavage , j. , et al . , 2015 , , 150 , 101
out of the 984 lcs , only one pair of stars for which a k2 lc was requested were within 4@xmath9 of each other ( within a kepler pixel ) .
epic 211934148 is at 08:43:07.40 + 19:14:15.4 , and as such , is matched to js519 ; epic 211934221 is at 08:43:07.44 + 19:14:19.2 , 3.88@xmath9 away , and as such is matched to ugcsj084307.42 + 191419.2 .
the two stars are of similar brightness in the poss images , but the latter is slightly fainter .
both of these stars are clearly m stars .
there are two obvious , effectively identical periods in each of the ( very similar ) light curves .
the one from the brighter star ( 211934148 ) is about double the counts than the one from the fainter star ( 211934221 ) .
we suspect that each star is contributing one periodic signal . however , it is less clear which period belongs to which star .
the two periods derived from the joint light curves are very different from each other . a period of 2.954 d
is recovered cleanly in both lcs .
the other period , at @xmath423 - 24 d , is less reliably recovered , presumably because there are fewer complete cycles in the k2 campaign .
the longer period derived from the light curves is 23.293 and 24.421 , respectively ( with the slightly longer period originating in the lc with fewer counts ) .
following the same approach as with the rest of the light curves , there is more power in the @xmath423 - 24 d peaks , so that period was assigned as @xmath36 , the likely rotation period for both epic numbers , with the other period assigned to @xmath37 .
however , since these are well - separated peaks , these two lcs were both identified as coming from likely binaries .
given our initial data amalgamations above ( sec .
[ sec : litphotom ] ) , we have @xmath31=12.222 and 12.618 mag , and no @xmath19 . our estimates of @xmath1 ( see sec .
[ sec : vmk ] ) are very different for our urat and gaia approaches .
gaia measures @xmath24=15.601 and 16.422 for the two stars , and
the @xmath1 we thus derive are 4.479 and 5.272 . via the urat approach ,
we obtain @xmath25 magnitudes of 15.05 and 15.53 ; this is at least consistent with the poss images that suggest the two stars are very close in brightness .
this results in @xmath1 of 4.017 and 4.158 .
we looked via vizier to find additional direct measurements of @xmath19 for these stars .
nomad reports @xmath19=16.12 and 17.160 , respectively , which would result in @xmath1=3.898 and 4.542 mag .
having no abundantly clear answer with respect to what the @xmath1should be for each of these stars , we sought input from the @xmath0 vs.@xmath1 plot ; see fig .
[ fig : pvmkpair ] .
if these stars are ` typical ' praesepe members , they should not be outliers .
following the same approach as with the rest of the sample , these stars would both appear within the ensemble , even if both use the @xmath424 d period .
taking the shorter period and tying it to the redder ( fainter ) source , it too is within the distribution at that location , though it is _ just _ above the denser population of m stars .
the shorter period does not appear to be appropriate for the brighter ( bluer ) star ; it is an outlier at that location at any @xmath1 . using the urat @xmath1 estimates results in the most discrepant points ; both stars would be outliers in those cases . using the nomad @xmath19 , the points move slightly right ( redder ) but are still not necessarily ` within the pack ' of the rest of the stars .
we suspect that the gaia - based @xmath1 estimates are the best we are going to have at this time .
we have retained those , and assigned the @xmath424 d period to the bluer star ( epic 211934148/js519 ) and the @xmath43 d period to the redder star ( epic 211934221=ugcsj084307.42 + 191419.2 ) .
there is limited information in the literature on binaries in praesepe . here
we collect what information we have .
figure [ fig : optcmdbin ] has a color - magnitude diagram and a @xmath0 vs. @xmath1 diagram of these binaries .
we now discuss eclipsing binaries ( ebs ) and binaries from the literature .
clcp9 cm 212009427 & ad1508 & 083129.87 + 202437.5 & sinsuoidal @xmath0 with dips superimposed ; to appear in gillen et al .
2017 in prep + 211916756 & js183 & 083727.05 + 185836.0 & also obermeier et al . 2016 ; mann et al . 2016b ; libralato et al .
2016 ; pepper et al .
2017 + 211969807 & jc126 & 083832.82 + 194625.7 & also mann et al . 2016b ; libralato et al .
2016 ; pope et al .
2016 + 212002525 & ugcsj083942.02 + 201745.0 & 083942.03 + 201745.0 & to appear in gillen et al .
2017 in prep + 211918335 & kw244=txcnc & 084001.70 + 185959.4 & w uma type ( whelan et al . 1973 ) + 211970147 & jc193 & 084013.45 + 194643.7 & also mann et al .
2016b + 211901114 & js441 & 084135.69 + 184435.0 & possible non - member + 211822797 & hshj385 & 084138.48 + 173824.0 & possible non - member , very shallow + 211946007 & hshj430 & 084239.43 + 192451.9 & to appear in gillen et al .
2017 in prep + 211922849 & js550 & 084344.73 + 190358.5 & grazing system + 211919680 &
hshj474 & 084403.90 + 190112.8 & transit signal may be from an unassociated projected background object + 211972086 & ad3814 & 085049.84 + 194836.4 & to appear in gillen et al .
2017 in prep + there are 12 objects that we identified in the process of light curve inspection as candidate eclipsing binaries ( or planet candidates ) in this dataset .
when we report a period here for these objects , we believe that the @xmath0 we report and use in our analysis is a rotation period .
all of the ebs are listed in table [ tab : ebs ] .
we note explicitly here that one object is a known w uma type eb according to whelan et al .
( 1973 ) : epic 211918335/kw244=tx cnc .
kovcs et al .
( 2014 ) also report this as an eb .
for this lc , we do not include our 0.191446 d period in our analysis of rotation periods .
mermilliod et al .
( 2009 , 1999 ) includes stars in praesepe that were monitored for spectral binarity ( radial velocity variations ) over @xmath420 years .
kovcs et al .
( 2014 ) also reports photometric periods they attribute to binary components .
barrado et al .
( 1998 ) identified spectral binaries from single - epoch observations .
table [ tab : sb ] includes a comparison of the periods derived here to those derived for the binaries in these papers .
many of the periods given in the literature are @xmath6035 d , so there is no way that we could have obtained that period from these data . for cases where the reported literature periods are both close ( and @xmath3335 d ) ,
more often than not , we do not recover both periods .
it could be that one of the two stars does not have spots , or organized enough spots , to create a detectable signal .
clcccp9 cm 211935509 & kw539=s4 & 083648.96 + 191526.4 & 7.475 & 3.72481,5551 & hat-269 - 0001402 ; our @xmath11 is twice theirs , and we believe ours to be correct + 211977390 & kw55=s27 & 083749.95 + 195328.8 & 6.789 & 7.13827,0.838434 & hat-269 - 0001352 ; our peak is broad and the waveform has changes over the campaign ; both 6.8 and 7.1 may be consistent ; no peak at 0.8d ; also fast outlier in @xmath0 vs. @xmath1 diagram + 211949097 & bd+19d2061 & 083924.97 + 192733.6 & 3.940 , 4.261 & 4.14405,1.41563 & hat-269 - 0000761 ; our peak is broad and both of our periods are probably consistent with theirs ; no peak at 1.4d + 211950081 & kw184 & 083928.58 + 192825.1 & 10.175 & 10.4866,4.52284 & hat-269 - 0001490 ; well - matched @xmath11 ; no peak at 4.5d + 211918335 & kw244=txcnc & 084001.70 + 185959.4 & ( 0.1914 ) & 0.1915 & hat-269 - 000058 ; wuma eb ( whelan et al.1973 ) ; also listed in ebs table + 211933215 & s137 & 084041.90 + 191325.4 & 6.112 & 3.11867,897 & hat-269 - 0000850 ; our peak is correct for our data + 211972627 & kw368=s155 & 084110.30 + 194907.0 & 9.054 & 9.00657,8.50340 & hat-269 - 0001570 ; our waveform undergoes many changes over the campaign , but we do not have two distinct peaks at 9.0 and 8.5d + 211935518 & kw434=s184 & 084154.36 + 191526.7 & 4.184 & 4.13291,0.951565 & hat-269 - 0001333 ; well - matched to @xmath11 ; we do not find a 0.95d peak ; also fast outlier in @xmath0 vs. @xmath1 diagram + 211947631 & bd+19d2087 & 084305.93 + 192615.2 & 4.736 & 4.65095,7.71605 & hat-269 - 0000465 ; well - matched @xmath11 ; no peak at 7.7d + 211969494 & s213 & 084320.19 + 194608.5 & 6.382 & 6.18544,648 & hat-269 - 0000913 ; well - matched @xmath11 + 211896596 & js655=fvcnc & 084801.74 + 184037.6 & 2.976 & 2.92535,1.01926 & hat-318 - 0000612 ; well - matched @xmath11 ; we do not find a 1.02d peak ; also appears mermilliod et al .
( 1990 ) with @xmath0=2.98d ; also fast outlier in @xmath0 vs. @xmath1 diagram + 211958646 & hd73081 & 083702.02 + 193617.2 & 4.413 , 2.093 & 45.97 & longer than we can detect + 211971690 & s12 & 083711.48 + 194813.2 & 8.242 & 5.86628 & our waveform undergoes many changes over the campaign , but no peak is apparent at @xmath46d + 211929178 & kw34=bd+19d2050 & 083728.19 + 190944.3 & 0.840 , 0.819 & 7383 & longer than we can detect + 211959522 & s16 & 083727.54 + 193703.1 & 8.873 & 144 & longer than we can detect + 211955820 & bd+20d2130 & 083727.94 + 193345.1 & 3.618 , 6.605 , 2.572 , 2.694 & 47.44 & longer than we can detect + 211947686 & s25 & 083746.61 + 192618.0 & 5.933 & 458 & longer than we can detect + 211936163 & hd73210 & 083746.76 + 191601.9 & 0.082 & 35.90 & we should have seen a @xmath436d period , but none are apparent + 211977390 & kw55=s27 & 083749.95 + 195328.8 & 6.789 & 1268 & longer than we can detect + 211983461 & kw58=s29 & 083752.09 + 195913.9 & 7.933 & 5567 & longer than we can detect + 211942703 & s38 & 083814.26 + 192155.4 & 7.225 , 9.554 , 6.369 & 117 & longer than we can detect + 211990866 & kw100=s42 & 083824.29 + 200621.8 & 4.255 , 4.604 , 3.971 & 4.92946 & comparable to measured periods , but it is unclear whether these are @xmath11 or @xmath16 .
+ clcccp9 cm 212034371 & js102 6 & 083556.94 + 204934.7 & 1.21051 & & listed as binary ; also fast outlier in @xmath0 vs. @xmath1 diagram + 212001830 & s11 & 083711.66 + 201704.9 & & 1149.5 & @xmath16 very much longer than we can detect .
noisy and not detected as periodic + 211927313 & kw47=bd+19d2052 & 083742.36 + 190801.5 & 3.077 , 3.303 , 3.626 & 34.619 & @xmath16 comparable to our max @xmath0 but this is not detected + 211988454 & kw127=s51 & 083850.00 + 200403.4 & 5.909 , 8.514&13.2803 & the waveform is very complicated , with more than one period and dense power spectra ; 13.3d is nt recovered , but could be embedded in the signal + 211898181 & kw547 & 084037.88 + 184200.4 & 2.524 & & listed as sb1 ; also fast outlier in @xmath0 vs. @xmath1 diagram + 211975006 & kw367=bd+19d2077 & 084109.60 + 195118.6 & 3.0675&1659 . & sinusoidal signal at 3.0675d very strong ; also fast outlier in @xmath0 vs. @xmath1 diagram + 211912407 & kw439=hd73994 & 084157.81 + 185442.2 & 2.0297 , 2.102&457.8 & longer than we can detect ; noisy + 211989010 & bd+20d2198 & 084407.33 + 200436.9 & 3.682,7.03816 & 40.6649 & longer than we can detect but no hint of structure on @xmath440d + 211958646 & hd73081 & 083702.02 + 193617.2 & 4.4134,2.093 & & listed as sb + 211991571 & kw146=hd73429 & 083905.23 + 200701.8 & 0.7856 & & listed as sb + 211956096 & kw236 & 083959.82 + 193400.2 & 11.509 & & listed as sb + 212003469 & kw297 & 084028.63 + 201844.8 & 8.1555 & & listed as sb + 211909748 & kw401 & 084130.70 + 185218.7 & 2.422 & & listed as sb ; also fast outlier in @xmath0 vs. @xmath1 diagram + 211956984 & bd+20d2193 & 084244.41 + 193447.8 & 0.4794,0.9488 , 0.9937 & & listed as sb + 211947631 & bd+19d2087 & 084305.93 + 192615.2 & 4.7357 & & listed as sb +
as for the pleiades , we identified some objects in praesepe that have a repeated pattern , but that we do not think are necessarily due to spot - modulated rotation periods ; see figure [ fig : timescales ] for four examples . we describe these as ` timescales ' ;
these objects are listed in table [ tab : nm ] .
they do not have a preferential color range ; see fig .
[ fig : optcmdts ] ( left ) .
they do , however , tend to have longer periods ; see fig .
[ fig : optcmdts ] ( right ) .
cccc 211985684 & 082409.33 + 200124.9 & 2massj08240933 + 2001249 & @xmath430 + 211995609 & 082635.20 + 201056.6 & 2massj08263520 + 2010567 & @xmath422 + 211935588 & 082820.20 + 191530.7 & 2massj08282020 + 1915307 & 25.6 + 212025207 & 082944.41 + 204023.1 & ad1268=tyc1391 - 40 - 1 & @xmath413 + 212173061 & 083008.81 + 233336.4 & 2massj08300882 + 2333365 & @xmath424 + 211787610 & 083055.77 + 170822.6 & 2massj08305580 + 1708223 & @xmath418 + 211852745 & 083147.19 + 180322.4 & 2massj08314719 + 1803225 & @xmath423 + 211981509 & 083239.47 + 195722.2 & jc10 & 28.8 + 211884968 & 083253.10 + 183029.3 & js16 & @xmath422 + 211993704 & 083550.36 + 200904.5 & js98 & 29.7 + 211975682 & 083627.12 + 195154.6 & js132=jc63 & @xmath420.1 + 211813885 & 083713.83 + 173048.7 & 2massj08371388 + 1730487 & @xmath420 + 211897713 & 083729.40 + 184135.5 & 2massj08372941 + 1841355 & 28.00 + 211795569 & 083813.64 + 171515.7 & 2massj08381365 + 1715158 & @xmath424.2 + 211776073 & 083824.86 + 165836.3 & ad2396=2massj08382489 + 1658360 & 22.8 + 211744621 & 083848.16 + 163155.9 & & 9.7 + 212030517 & 083903.86 + 204547.6 & js264 & @xmath430 + 212005623 & 083906.87 + 202054.2 & js270=2massj08390688 + 2020542 & @xmath420 + 212103507 & 083921.31 + 220520.7 & hshj259 & @xmath438 + 211935711 & 083936.43 + 191537.8 & js302=jc166 & @xmath418 + 211950716 & 083941.66 + 192900.4 & hshj283 & @xmath411 + 211914960 & 084002.22 + 185656.9 & jc183 & 22.4 + 211968228 & 084013.78 + 194455.9 & jc194 & 20.9 + 212112321 & 084111.62 + 221551.7 & hshj350 & @xmath435 + 211896120 & 084132.35 + 184010.7 & js439 & 30.3 + 211812292 & 084135.58 + 172927.1 & & 35.9 + 212005503 & 084151.91 + 202047.8 & js459=2massj08415192 + 2020479 & 21.1 + 212108286 & 084207.85 + 221105.1 & 2massj08420785 + 2211051 & @xmath428 + 211886612 & 084312.92 + 183150.9 & js525 & 17.2 + 211983811 & 084332.62 + 195933.0 & hshj458=2massj08433262 + 1959330 & @xmath430 + 211939484 & 084552.80 + 191900.6 & 2massj08455280 + 1919006 & 16.7 +
there are 18 stars that we suspect are pulsators ; that is , their power spectra have a ` forest ' of short - period peaks ( see sec .
[ sec : lcpcats ] and rebull et al .
one of these 18 is one of the stars that is too bright for our final best sample .
all of these pulsators are listed in table [ tab : pulsator ] .
most of them are bright ( @[email protected] ) , and blue ( @[email protected] ) .
there are some stars that are comparably bright and blue , but that do not have a comparable power spectrum ; if we derived single periods for them , we retained that period as the rotation period .
clcclp4 cm 211953002&br cnc = hd 73175 & 083740.70 + 193106.3 & 0.588 & 0.0872 , 0.1219 , 0.0965 & f0 ; @xmath10 scuti ( _ e.g. _ , breger 1970 ) + 211983602&cy cnc = hd 73345 & 083837.86 + 195923.1 & 0.530 & 0.051 , 0.058 , 0.053 , 0.067&f0v ; @xmath10 scuti ( _ e.g. _ , hauck 1971 ) + 211951863&hd 73397 & 083846.95 + 193003.3 & 0.779 & 0.5125 , 0.5950 , 0.6150 , 0.7568 & f4 + 211957791&bs cnc= hd 73450 & 083909.09 + 193532.7 & 0.625 & 0.0587 , 0.0644 & a9 ; @xmath10 scuti ( _ e.g. _ , breger 1970 , breger et al .
2012 ) + 211994121&hd 73616 & 083958.37 + 200929.6 & 0.884 & 0.8876 , 0.2713 , 0.5718 , 0.6137 & f2 + 211984704&39 cnc = hd 73665 & 084006.41 + 200028.0 & 2.219 & 1.2411 & g8iii ( dropped as too bright ) + 211941583&hd 73712 & 084020.13 + 192056.4 & 0.735 & 0.1413 , 0.4815 , 0.1179 , 0.1395 & spectroscopic binary ; a9v + 211931309&bv cnc = hd 73746 & 084032.96 + 191139.5 & 0.686 & 0.0608 , 0.0637 , 0.0494 , 0.0624 & f0 ; @xmath10 scuti ( _ e.g. _ , hauck 1971 ) + 211973314&hd 73854 & 084110.67 + 194946.5 & 0.850 & 0.7522 , 0.6970 , 0.7937 , 0.7409 & f5 + 211979345&hd 73872 & 084113.76 + 195519.1 & 0.556 & 0.0644 , 0.0437 , 0.0771 , 0.0649 & a5 ; @xmath10 scuti ( _ e.g. _ , breger et al .
2012 ) + 211935741&hi cnc = hd 73890 & 084118.40 + 191539.4 & 0.627 & 0.070 , 0.078 , 0.089 , 0.059&a7v ; @xmath10 scuti ( _ e.g. _ , hauck 1971 ) + 211945791&bx cnc = hd 74028 & 084206.49 + 192440.4 & 0.529 & 0.058 , 0.126 , 0.136 , 0.055&a7v ; @xmath10 scuti ( _ e.g. _ , breger 1970 ) + 211914004&by cnc = hd 74050 & 084210.80 + 185603.7 & 0.557 & 0.055 , 0.053 , 0.047 , 0.052&a7v ; @xmath10 scuti ( _ e.g. _ , breger 1970 ) + 211954593&bd+20d2192 & 084240.71 + 193235.4 & 1.081 & 0.4142 , 0.6202 , 1.2397 , 1.2719 & f2iii + 211956984&bd+20d2193 & 084244.41 + 193447.8 & 1.089 & 0.4794 , 0.9488 , 0.9937 & f6 + 212033939&hd 74135 & 084253.07 + 204909.1 & 0.777 & 0.4757 , 0.3128 , 0.4135 , 0.2734 & a9iii + 211909987&hd 74589 & 084520.53 + 185231.3 & 0.724 & 0.0434 , 0.0486 , 0.0552 , 0.0535 & f0 + 212008515&hd 74587 & 084528.25 + 202343.5 & 0.767 & 0.0587 , 0.0513 , 0.0519 , 0.0801 & a5 ; @xmath10 scuti in simbad and kovcs et al .
( 2014 ) but unclear if identified as @xmath10 scuti prior to kovcs et al .
we identified stars that we took to not be members of praesepe ( see sec .
[ sec : membership ] , with modifications as per the outliers discussion in sec .
[ sec : outliers ] and app .
[ app : pvmkoutliers ] ) .
table [ tab : nm ] lists these stars .
they appear in a color - magnitude diagram and a @xmath0 vs. @xmath1 diagram in fig .
[ fig : optcmdnm ] . in general
, they are not as well clustered along the main sequence in praesepe as the members . if they are periodic , these stars generally are outliers in the @xmath0 vs. @xmath1 diagram , consistent with being non - members .
cclccp4 cm 212158768&083012.13 + 231336.9&2massj08301213 + 2313370 & 2.158&31.220 & new spectrum has rv inconsistent with membership ; also too slow for this color + 211897926&083150.92 + 184147.1&2massj08315092 + 1841470 & 2.282&24.521 & new spectrum has rv inconsistent with membership ; also too slow for this color + 211981509&083239.47 + 195722.2&jc10 & 1.821 & &
+ 211803269&083300.85 + 172146.6 & & 1.709 & & + 211933061&083316.38 + 191317.3&js666 & 4.269 & & + 211885783&083334.80 + 183108.9&2massj08333480 + 1831089 & 3.279 & &
+ 211993704&083550.36 + 200904.5&js98 & 4.419 & &
+ 212033649&083608.34 + 204852.8 & & 4.663 & & + 212099156&083626.89 + 220016.8&2massj08362690 + 2200168 & 3.845 & &
+ 211955417&083649.56 + 193322.8&jc75 & 4.376&12.019 , 5.637 , 1.235 & + 211801256&083701.08 + 172005.1&2massj08370108 + 1720051 & 4.597 & &
+ 211930233&083703.45 + 191041.1&jc85 & 2.902&18.941 & new spectrum has rv inconsistent with membership ; also too slow for this color + 211833885&083707.22 + 174748.8&2massj08370722 + 1747487 & 3.676 & &
+ 212021664&083713.33 + 203653.6&js170 & 3.957&22.048 & also too slow for this color + 212072324&083816.83 + 212953.7&js223 & 4.069 & &
+ 211834700&083844.47 + 174829.4&anm690 & 1.436 & &
+ 211990313&084028.47 + 200551.2&js363 & 4.598 & &
+ 212066519&084030.72 + 212333.1&2massj08403072 + 2123331 & 4.624 & & + 212072257&084048.52 + 212949.5&2massj08404852 + 2129495 & 3.365 & &
+ 211984180&084053.83 + 195956.0&js392 & 4.662&23.994 , 1.863&also too slow for this color + 211988382&084058.92 + 200359.2&js399 & 4.000&25.393 & also too slow for this color + 211954582&084113.18 + 193234.9&jc243 & 3.823 & 3.191 & also too fast for this color + 212005583&084113.73 + 202051.7&js417 & 3.301&21.610 & also too slow for this color + 212094510&084120.89 + 215454.0&2massj08412090 + 2154540 & 3.821&30.708 & also too slow for this color + 212112522&084128.93 + 221605.3 & & 1.355&16.018 , 5.011 & also too slow for this color + 211812292&084135.58 + 172927.1 & & 4.218 & &
+ 212084898&084136.38 + 214353.8&js732 & 1.580 & &
+ 211784450&084155.14 + 170540.8&anm1065 & 2.676&10.992,13.714 &
+ 211995547&084221.63 + 201053.1 & & 0.994 & 0.591 & + 212137243&084221.95 + 224552.2&2massj08422195 + 2245521 & 3.657 & 8.950 & also too fast for this color + 211931928&084303.65 + 191215.9&js739 & 5.473 & &
+ 212112578&084338.80 + 221609.1&kw529 & 2.007 & 8.878 & + 212029850&084430.66 + 204505.0&sdssj084430.65 + 204504.7 & 5.162 & 0.213 , 0.227 & also too fast for this color + 212011328&084501.06 + 202631.9&js585 & 1.761 & &
+ 212069325&084811.53 + 212635.0 & & & & + 212120476&084850.34 + 222531.9&hshj506 & 3.171&17.097 & new spectrum has rv inconsistent with membership ; also too slow for this color + 211921444&084852.61 + 190247.0&2massj08485261 + 1902470 & 4.219 & & + 211905939&085001.55 + 184852.6&2massj08500156 + 1848526 & 4.955 & &
+ 211939409&085125.81 + 191856.3 & & & & + 211935447&085528.96 + 191523.3&2massj08552896 + 1915234 & 4.657 & & + 211875602&090222.36 +
182223.8&2massj09022236 + 1822238 & 2.371&15.490 & new spectrum has rv inconsistent with membership ; also too slow for this color +
the two stars with @xmath18@xmath36 are epic 211984704/39 cnc = hd 73665 and epic 211976270/hd 73974 .
these were discarded from our sample as too bright , but we note some characteristics of their lcs here . the star for which we do not determine a periodic signal using these reduced lcs is epic 211976270/hd 73974 ; it is a giant ( k0iii ; yang et al .
it is very bright , and the artifacts in the light curve reductions we have reflect that .
the dominant structure in all the lc versions seems to be the 0.245 day thruster firings .
we find a period in the other star , which is is epic 211984704/39 cnc = hd 73665 .
it is also a giant ( g8iii ; yang et al .
we determine only one period , 1.2411 d. while this object is also very bright , there seems to be a readily apparent @xmath41 d oscillation even in the raw lc versions , which is why we retained this period .
the phased lc has a lot of scatter in any of the lc versions , admittedly , but there is a peak near 1.24 d that persists across lc versions .
but , what does that periodicity represent ? there are 4 giants in praesepe , but there are only 2 k2 lcs
. the other two giants are hd73598 ( k0iii ) and hd73710 ( g9iii ) .
pasquini et al.(2000 ) provide @xmath73 values for two of them , and give an average @xmath73 for all four , and they are between 1 and 3 km s@xmath28 , comparable to field red giants .
choi et al .
( 1995 ) used long duration monitoring of calcium hk emission to find rotation periods for three of them .
( hd 73598 has @xmath68159 d ; hd73974 has @xmath68112 d ; hd73710 has @xmath68155 d. ) all of those are more or less consistent with the @xmath73 values .
so , the 1.2 d period we see in the k2 lc can not be the rotation period .
an estimate of the max power in the oscillations for stars is ( brown 1991 ) @xmath74 .
for the sun , @xmath75 is about 3100 @xmath76hz ( which converts to about 5 minutes for a period ) . assuming that the @xmath77for the giants in praesepe is about 4800 k , and taking their luminosity to be @xmath460 @xmath78 , and their masses as @xmath42.5 @xmath79 , the max oscillation power for a praesepe giant should be at about 4 hours .
so , the 1.2 d period from the k2 light curve is also not likely to be pulsation .
the star is saturated in the k2 thumbnail , but in general that should result in no derived period , or a period at 0.245 d ( or a multiple of that ) , reflecting the thruster firings ( as for the other giant with a k2 lc ) , not a 1.2 d period .
we are not sure if this 1.2 d period is real , nor how to interpret it .
section [ sec : outliers ] above highlights some of the outliers in the @xmath0 vs. @xmath1 plot , focusing on both the objects rotating too quickly for their color and those that are going too slowly .
here we focus on individual objects that were covered more broadly above .
[ fig : optcmdoutliers ] includes an optical cmd and the @xmath0 vs. @xmath1 plot with these objects highlighted ; they are listed in tables [ tab : fastoutliers ] and [ tab : slowoutliers ] . for the fast rotating stars
, we checked the veracity of their colors and periods , and all of them seemed consistent with the observations . among the fast rotating k stars ,
many of them are known short - period binaries , and we infer that the rest of them are probably binaries as well .
three of the ones with published orbital periods of less than 4 days are 211896596/js655=fv cnc , 211975006/kw 367 , and 211935518/kw 434=s184 ( mermilliod & mayor 1999 ; mermilliod et al . 1990 ) .
one more ( 211909748/kw401 ) is a known sb with a large radial velocity amplitude ( barrado et al .
1998 ) , and another is simply listed as an sb in mermilliod & mayor ( 1999 ) .
note that there is little literature binary monitoring work done for @xmath1@xmath33 . for the slowly rotating stars
, we also checked colors and periods .
the assessment of their periods is more difficult than other shorter - period stars , as described in sec .
[ sec : outliers ] above .
stars particularly subject to this are noted as such in table [ tab : slowoutliers ] .
most of the slow outliers with 1.7@xmath2@[email protected] and @xmath80 15 are relatively anonymous stars , with only a few references in simbad .
particularly for the few slowly rotating stars that are photometric binaries in fig .
[ fig : optcmdoutliers ] , tidal synchronization in a short - period binary system may be a good explanation for the slow rotation . in most cases ,
they are listed as candidate members in only one or two praesepe proper motion membership studies .
only one of them has a published radial velocity ; that star ( 212034762/js409 ) appears to be a member .
we have obtained keck hires spectra for 11 of these stars , which are noted in table [ tab : slowoutliers ] ( see app .
[ app : keck ] for details ) .
about half of the ones for which we have spectroscopy appear to have radial velocities inconsistent with membership .
many of these slow outliers have cmd locations and proper motions generally consistent with being praesepe members .
some have proper motions consistent with membership , but are slightly low in the cmd ; these are noted in table [ tab : slowoutliers ] .
the lack of detailed information for these stars makes it more likely that they might be non - members , but more information is needed .
cclccp7 cm 211898878&081903.68 + 184236.1&2massj08190368 + 1842361 & 5.100 & 0.075 & fastest , bluest m star ; it is one of the furthest west sources in fig .
[ fig : where ] ; it may be a cv + 212009427&083129.87 + 202437.5&ad1508=2massj08312987 + 2024374 & 4.021 & 1.557 & likely eb ( also listed in ebs table ) + 211971354&083140.45 + 194754.2&hshj15 & 4.153 & 8.297 , 9.374 , 0.798 & @xmath726 d , likely binary ; bluest end of sparsely populated @xmath0 range of m stars + 212034371&083556.94 + 204934.7&js102=2massj08355696 + 2049346 & 2.022 & 1.211 & binary ( also listed in binaries table ) + 211915860&083656.25 + 185747.9&js159=jc80 & 4.061 & 4.816&bluest end of sparsely populated @xmath0 range of m stars + 212021253&083728.46 + 203628.5&js181 & 3.909 & 7.238 & no rvs ; bluest end of sparsely populated @xmath0 range of m stars + 211977390&083749.95 + 195328.8&kw55=s27 & 2.054 & 6.789 & binary ( also listed in binaries table ) + 211988628&083915.79 + 200414.0&kw566=js280 & 4.123 & 7.964&bluest end of sparsely populated @xmath0 range of m stars + 211983725&083941.03 + 195928.8&kw570=js309 & 4.027 & 4.165&bluest end of sparsely populated @xmath0 range of m stars + 211918335&084001.70 + 185959.4&kw244=txcnc & 1.282&(0.1914 ) & already removed @xmath0 as wuma binary ( also listed in binaries table ) + 211920022&084005.71 + 190130.6&kw256=2massj08400571 + 1901307 & 2.611 & 4.675 & + 211984704&084006.41 + 200028.0&39cnc & 2.219 & 1.241 & already removed from sample as too bright ( g8iii giant ) + 211915202&084029.21 + 185709.4 & & 4.186 & 5.503 , 2.004&bluest end of sparsely populated @xmath0 range of m stars + 211898181&084037.88 + 184200.4&kw547 & 2.290 & 2.524 & binary ( also listed in binaries table ) + 212013132&084044.25 + 202818.6&js379=2massj08404426 + 2028187 & 3.528 & 2.129 , 4.367 , 4.053 & no rvs known
+ 211975006&084109.60 + 195118.6&kw367=bd+19d2077 & 1.761 & 3.068 & binary ( also listed in binaries table ) + 211954582&084113.18 + 193234.9&jc243 & 3.823 & 3.191 & already nm ( also listed in nm table ) + 211909748&084130.70 + 185218.7&kw401=2massj08413070 + 1852188 & 2.816 & 2.422 & binary ( also listed in binaries table ) + 211935518&084154.36 + 191526.7&kw434=s184 & 1.767 & 4.184 & binary ( also listed in binaries table ) + 212137243&084221.95 + 224552.2&2massj08422195 + 2245521 & 3.657 & 8.950&already nm ( also listed in nm table ) + 211890774&084232.06 + 183528.0&js488 & 4.170 & 9.428 & marginal outlier ; bluest end of sparsely populated @xmath0 range of m stars + 212066424&084300.54 + 212328.1&2massj08430054 + 2123281 & 6.217 & 0.112 & fastest m star in clump ; everything about it looks fine + 212029850&084430.66 + 204505.0&sdssj084430.65 + 204504.7 & 5.162 & 0.213 , 0.227 & already nm ( also listed in nm table ) + 211896596&084801.74 + 184037.6&js655=fvcnc & 1.664 & 2.975 & binary
( also listed in binaries table ) + 211773459&084832.70 + 165623.6&2massj08483271 + 1656236 & 4.353 & 5.919&bluest end of sparsely populated @xmath0 range of m stars + 211885995&084926.76 + 183119.5&2massj08492676 + 1831195 & 4.013 & 9.153 & marginal outlier ; bluest end of sparsely populated @xmath0 range of m stars + cclccp7 cm 212032123&082507.04 + 204725.1&2massj08250705 + 2047252 & 2.105&22.662 & just below ms , and @xmath33.6@xmath7 from cluster center ; judgement call as to whether @xmath0 is @xmath11 or timescale + 212158768&083012.13 + 231336.9&2massj08301213 + 2313370 & 2.158&31.220 & new spectrum has rv inconsistent with membership ( also listed in nm table ) + 212077235&083110.44 + 213522.4&2massj08311044 + 2135224 & 2.829&32.210 & slightly below ms ; judgement call as to whether @xmath0 is @xmath11 or timescale + 211897926&083150.92 + 184147.1&2massj08315092 + 1841470 & 2.282&24.521 & slightly below ms ; new spectrum has rv inconsistent with membership ( also listed in nm table ) ; judgement call as to whether @xmath0 is @xmath11 or timescale + 211898294&083249.71 + 184206.2&js14=jc12 & 2.978&17.808 & new spectrum is double lined and has rv consistent with membership + 211916015&083629.84 + 185757.0&kw536=bd+19d2045 & 1.121 & 9.175 & upper left of diagram ( near blue bend in distribution ) ; judgement call as to whether @xmath0 is @xmath11 or timescale + 211910450&083647.99 + 185258.0 & & 0.676&13.011&judgement call as to whether @xmath0 is @xmath11 or timescale ; far upper left of diagram + 211930233&083703.45 + 191041.1&jc85 & 2.902&18.941 & new spectrum has rv inconsistent with membership ( also listed in nm table ) + 212021664&083713.33 + 203653.6&js170 & 3.957&22.048 & already identified as nm ( also listed in nm table ) + 211946055&083751.85 + 192453.4 & & 2.299&17.459 & just below ms + 211892153&083821.66 + 183639.9&jc123 & 3.096&27.225 & judgement call as to whether @xmath0 is @xmath11 or timescale + 211931736&083958.06 + 191205.9&kw227=hd73641 & 1.008&10.285,14.992 , 1.536 & upper left of diagram ( near blue bend in distribution ) ; noisy lc + 211984180&084053.83 + 195956.0&js392 & 4.662&23.994 , 1.863 & already identified as nm ( also listed in nm table ) ; @xmath56d + 211988382&084058.92 + 200359.2&js399 & 4.000&25.393 & already identified as nm ( also listed in nm table ) + 212034762&084111.05 + 204957.8&js409 & 1.800&17.016&new spectrum has rv consistent with membership ; judgement call as to whether @xmath0 is @xmath11 or timescale + 212005583&084113.73 + 202051.7&js417 & 3.301&21.610 & already identified as nm ( also listed in nm table ) + 211990785&084119.43 + 200618.2 & & 2.582&22.292 & slightly below ms ; judgement call as to whether @xmath0 is @xmath11 or timescale + 212094510&084120.89 + 215454.0&2massj08412090 + 2154540 & 3.821&30.708 & already identified as nm ( also listed in nm table ) ; judgement call as to whether @xmath0 is @xmath11 or timescale + 212112522&084128.93 + 221605.3 & & 1.355&16.018 , 5.011 & already identified as nm ( also listed in nm table ) ; @xmath56d + 212098754&084212.66 + 215948.8 & & 2.055&23.601 & new spectrum has rv consistent with membership ; judgement call as to whether @xmath0 is @xmath11 or timescale + 211921647&084546.53 + 190258.1 & & 0.496 & 1.815 , 1.700 & above the bluest branch + 212120476&084850.34 + 222531.9&hshj506 & 3.171&17.097 & new spectrum has rv inconsistent with membership ( also listed in nm table ) ; judgement call as to whether @xmath0 is @xmath11 or timescale + 212027750&084914.76 + 204300.9&2massj08491476 + 2043009 & 2.907&20.837 & new spectrum has rv consistent with membership + 212008710&085502.23 + 202354.0&2massj08550224 + 2023540 & 3.025&23.573 & new spectrum has rv consistent with membership ; judgement call as to whether @xmath0 is @xmath11 or timescale + 211875602&090222.36 + 182223.8&2massj09022236 + 1822238 & 2.371&15.490 & new spectrum has rv inconsistent with membership ( also listed in nm table ) +
the tidal radius of praesepe is 12.1 pc ( holland et al . 2000 ) ,
which at a distance of 184 pc is @xmath43.8@xmath7 across .
six of the stars in the original sample of targets with k2 light curves are more than 5@xmath7 away from the cluster center ; see table [ tab : halo ] .
they appear in a cmd and the @xmath0 vs. color diagram in fig .
[ fig : halo ] .
one is dropped as a non - member ( sec .
[ sec : membership ] and app .
[ app : nm ] ) .
two more are listed as outliers in the @xmath0 vs. @xmath1 diagram ( app .
[ app : pvmkoutliers ] ) . in general , these stars are poorly studied , but we have retained those stars that met the rest of our criteria .
cccccp5 cm 211999629&081259.31 + 201453.6&2massj08125931 + 2014537& &6.4 & little additional relevant information found ; perhaps below single - star ms + 211756862&081551.30 + 164215.9 & & 1.455 , 2.627&6.4 & little additional relevant information found + 211984209&081615.53 + 195956.9&2massj08161554 + 1959570&13.754&5.6 & adams et al .
( 2002 ) has membership probability of 0.41 ; perhaps below single - star ms + 211898878&081903.68 + 184236.1&2massj08190368 + 1842361 & 0.075&5.0 & one of the fastest rotating m stars ( listed as fast outlier ) ; adams et al . ( 2002 ) has membership probability of 0.39 ; a cv rather than member ? + 212148559&082333.40 +
230007.9&2massj08233340 + 2300079& &5.1 & adams et al .
( 2002 ) has membership probability of 0.31 ; perhaps below single - star ms + 211875602&090222.36 + 182223.8&2massj09022236 + 1822238&15.490&5.4 & already dropped as a nm and long - p outlier ( listed in corresponding tables ) ; above single - star ms +
we obtained high resolution spectra for several of the anomalously slowly rotating stars ( app .
[ app : pvmkoutliers ] ) and all of the objects with odd lcs ( sec .
[ sec : batwingy ] ) using the keck hires spectrograph vogt et al .
the observations were taken on one of : 2016 october 14 , 2016 december 22 , 2016 december 26 , or 2017 january 13 , ut , and cover the wavelength range roughly 4800 to 9200 at a spectral resolution of @xmath81 ; the spectra were were reduced using the _ makee _ software written by tom barlow .
we measured radial velocities using the cross correlation techniques in the _ rv _ package in _ iraf _ , with absolute reference to between 3 and 5 ( depending on the night ) late type radial velocity standards . because these are slowly rotating stars , the errors are all under 0.5 km s@xmath28 , and are determined from the empirical scatter among orders and reference stars for each observation .
cclccp7 cm 212158768&083012.13 + 231336.9&2massj08301213 + 2313370 & [email protected] & non - member long - p outlier + 211897926&083150.92 + 184147.1&2massj08315092 + 1841470 & [email protected] & non - member long - p outlier + 211898294&083249.71 + 184206.2&js14=jc12 & [email protected] and [email protected] & double - lined , member long - p outlier + 212011416&083308.44 + 202637.3 & 2massj08330845 + 2026372 & [email protected] & unusual lc shape , member + 211930233&083703.45 + 191041.1&jc85&[email protected] & non - member long - p outlier + 211915940&083800.61 + 185752.9 & js208 & [email protected] & unusual lc shape , member + 212034762&084111.05 + 204957.8&js409 & 34.32 @xmath820.30 & non - member long - p outlier + 212098754&084212.66 + 215948.8 & hshj407 & [email protected] & member long - p outlier + 211931651&084322.40 + 191200.7 & ad3196=cp cnc & [email protected] & unusual lc shape , member + 212120476&084850.34 + 222531.9&hshj506 & [email protected] & non - member long - p outlier + 212027750&084914.76 + 204300.9&2massj08491476 + 2043009 & [email protected] & member long - p outlier + 212008710&085502.23 + 202354.0&2massj08550224 + 2023540 & [email protected] & member long - p outlier + 211875602&090222.36 + 182223.8&2massj09022236 + 1822238 & [email protected] & non - member long - p outlier + | we have fourier analyzed 941 k2 light curves of likely members of praesepe , measuring periods for 86% and increasing the number of rotation periods ( @xmath0 ) by nearly a factor of four .
the distribution of @xmath0 vs. @xmath1 , a mass proxy , has three different regimes : @[email protected] , where the rotation rate rapidly slows as mass decreases ; 1.3@xmath2 @[email protected] , where the rotation rate slows more gradually as mass decreases ; and @[email protected] , where the rotation rate rapidly increases as mass decreases . in this last regime
, there is a bimodal distribution of periods , with few between @xmath42 and @xmath410 days .
we interpret this to mean that once m stars start to slow down , they do so rapidly .
the k2 period - color distribution in praesepe ( @xmath4790 myr ) is much different than in the pleiades ( @xmath4125 myr ) for late f , g , k , and early - m stars ; the overall distribution moves to longer periods , and is better described by 2 line segments . for mid - m stars ,
the relationship has similarly broad scatter , and is steeper in praesepe .
the diversity of lightcurves and of periodogram types is similar in the two clusters ; about a quarter of the periodic stars in both clusters have multiple significant periods .
multi - periodic stars dominate among the higher masses , starting at a bluer color in praesepe ( @[email protected] ) than in the pleiades ( @[email protected] ) . in praesepe
, there are relatively more light curves that have two widely separated periods , @xmath56 days
. some of these could be examples of m star binaries where one star has spun down but the other has not . |
simulations of the quantum spin dynamics of coupled electron and nuclear spins are an essential component of many forms of electron paramagnetic resonance spectroscopy and are widespread in the field of spin chemistry . used to interpret experimental data , to guide the design and execution of experiments , and to explore ideas not immediately amenable to experiment ,
such calculations are usually restricted to a total of @xmath0 spins , because the size of the spin hilbert space increases exponentially with @xmath1 .
but many probably the majority of paramagnetic species whose spin dynamics can be probed experimentally contain more than 10 nuclei with hyperfine coupling to one or more electron spins . for example
, the radical pairs and triplet states responsible for magnetic field effects on the rates and yields of chemical reactions@xcite and for spin hyperpolarization@xcite often have several tens of magnetic nuclei ; the excitons , polaron pairs and bipolarons that appear to give rise to related magnetoresistance effects in polymeric organic semiconductors@xcite may have @xmath1 approaching 100 , and semiconductor quantum dots typically have @xmath2.@xcite in spin chemistry , the simulation problem becomes particularly acute in the context of magnetoreception the proposal that photochemically produced radical pairs are responsible for the magnetic compass sense of migratory birds and possibly other animals.@xcite the biological radicals thought to be involved here@xcite are multinuclear , long lived , slowly relaxing , and subject to an external magnetic field ( the earth s ) that is neither much stronger nor much weaker than the majority of the anisotropic hyperfine interactions all factors that exacerbate the difficulty of spin dynamics simulations .
the upper limit on the @xmath1 that it is practical to simulate can sometimes be relaxed by exploiting special properties of particular spin systems , such as coupling topology , symmetry , conservation laws , and the existence of non - interacting sets of spin states.@xcite experimental features can help too , for example by ensuring that a large fraction of the spin space is never populated and can therefore be removed from consideration.@xcite high - field , short - time , fast - relaxation and other approximations can also sometimes be useful .
however , none of these tricks is generally applicable in the spin chemical context .
a potential solution to the exponential scaling problem was proposed by schulten and wolynes in the 1970s,@xcite when spin chemistry was still in its infancy .
they described the electron spin motion induced by hyperfine coupling to many nuclear spins by means of a semiclassical approximation in which the electron spin precesses around a hyperfine - weighted sum of nuclear spin vectors , which is assumed to be fixed in space .
applied to a radical pair derived from pyrene and _ n , n_-dimethylaniline , the approach was shown to give excellent agreement with exact quantum simulations of the electron spin dynamics both with and without applied magnetic fields.@xcite in this paper , we shall examine a straightforward extension to the schulten - wolynes theory in which each individual nuclear spin is allowed to precess around the electron spin .
similar extensions have been examined previously in the condensed matter physics ( quantum dot ) literature,@xcite although in more sophisticated ways and with more sophisticated objectives than we shall pursue here ( such as obtaining an analytical understanding of the asymptotic long - time decay of the electron spin polarisation ) .
our goal is simply to find a simulation method that scales linearly with @xmath1 , approaches quantitative agreement with quantum mechanics as @xmath1 increases , and , unlike the schulten - wolynes theory , accurately captures the low field effect@xcite of a weak external magnetic field on a slow radical pair recombination reaction .
the motivation for this goal is that such a method would be very useful for simulating many of the current problems in spin chemistry .
to within a good approximation , the hamiltonian that governs the rate of a radical pair recombination reaction in solution in the presence of an applied magnetic field is@xcite @xmath3 where @xmath4 here @xmath5 , where @xmath6 is the gyromagnetic ratio of the electron in radical @xmath7 and @xmath8 is the applied magnetic field , @xmath9 is the number of nuclear spins in the radical , @xmath10 is an isotropic hyperfine coupling constant between the @xmath11-th nuclear spin and the electron spin , and @xmath12 and @xmath13 are the corresponding nuclear and electron spin angular momentum operators . note that this hamiltonian neglects the zeeman interactions between the nuclear spins and the magnetic field , which are typically far weaker than the terms that have been retained in eq .
we shall work throughout this paper in a unit system in which @xmath14 , in which the unit of time is the reciprocal of the unit of energy ( or frequency , or magnetic field strength ) that is used to specify @xmath15 and @xmath10 .
suppose that the radical pair is produced photochemically in its singlet state .
the initial density operator will then be @xmath16 where @xmath17 is the total number of nuclear spin states in radical @xmath7 and @xmath18 is the projection operator onto the singlet electronic subspace .
as time evolves , the singlet and triplet states will be mixed by the zeeman and hyperfine interactions in each radical , so the probability of finding the radical pair in the singlet state will decrease from one .
this singlet probability is given at time @xmath19 by @xmath20 , \eqno(5)\ ] ] where @xmath21 combining these equations with eq .
( 1 ) , and noting that @xmath22 and @xmath23 commute , one finds that @xmath24 can be written equivalently as@xcite @xmath25 where @xmath26 = { 1\over z_i}{\rm tr}\left[\hat{s}_{i\alpha}(0)\hat{s}_{i\beta}(t)\right ] \eqno(8)\ ] ] is an electron spin correlation tensor for the spin in radical @xmath7 .
the elements of this tensor can be shown to be real , and to satisfy @xmath27
@xmath28 @xmath29 the problem of calculating the time - dependent singlet probability @xmath24
is thus reduced to one of calculating the @xmath30 , @xmath31 , and @xmath32 components of the electron spin correlation tensor for each radical separately . in an experiment one
typically does not measure the singlet probability @xmath24 directly , but rather the overall quantum yield @xmath33 of the singlet product .
this is usually assumed to be given by a simple exponential model@xcite @xmath34 in which @xmath11 is a first order rate constant for radical pair recombination .
@xmath33 depends on the magnetic field by virtue of the @xmath35 term in eq .
( 2 ) , and there are two separate magnetic field effects that are of interest . in a high magnetic field
the electronic zeeman splitting is so large that the @xmath36 components of the triplet state are energetically inaccessible from the singlet , so @xmath37 can only mix with @xmath38 and the singlet yield is enhanced relative to its field - free value.@xcite conversely , a low but non - zero magnetic field breaks the symmetry of the zero - field problem and splits the degeneracy of the zero - field eigenstates , producing more pathways for ( energetically feasible ) singlet to triplet interconversion and lowering the singlet quantum yield from its field - free value.@xcite the schulten - wolynes theory is based on the assumption that , in a radical with a sufficiently large number of nuclear spins , the details of the nuclear spin dynamics will cease to matter : the electron spin in the radical will simply see a hyperfine - weighted sum of nuclear spin vectors , which can be regarded as being fixed in space .
the electron spin will then precesses around the resultant of this vector and the applied magnetic field.@xcite consistent with the assumption of a large number of nuclear spins , schulten and wolynes calculate the distribution of the hyperfine - weighted sum of nuclear spin vectors using the asymptotic ( @xmath39 ) result for the end - to - end distribution of a random flight polymer.@xcite if @xmath40 and each @xmath41 is interpreted as a classical vector of length @xmath42 , this is@xcite @xmath43 where @xmath44 and @xmath45 for each @xmath46 in this distribution , the electron spin in radical @xmath7 is assumed to precess around the vector @xmath47 where @xmath48 accounts for the applied magnetic field .
the contribution that this precession makes to the electron spin correlation tensor of the radical can be worked out either quantum mechanically or semiclassically .
schulten and wolynes chose to do the calculation quantum mechanically,@xcite but a semiclassical calculation gives the same result .
consider the contribution @xmath49\eqno(17)\ ] ] to the electron spin correlation tensor that comes from the precession of the electron spin around a particular vector @xmath50 in eq .
( 16 ) , where we have used @xmath51 to denote a trace over the electron spin states .
the simplest possible semiclassical ( classical vector model ) approximation to this is @xmath52 in which the quantum mechanical trace has been replaced by an integral @xmath53 over the orientation of a classical electron spin vector @xmath54 with @xmath55 .
the precession of this vector @xmath56 gives @xmath57 where @xmath58 and @xmath59 @xmath60{\bf s}_i(0),\ ] ] @xmath61 with @xmath62 .
substituting eq .
( 22 ) into eq .
( 18 ) , and doing the integrals over @xmath63 and @xmath64 in eq . ( 19 ) , one finds that @xmath65 , \eqno(24)\ ] ] where @xmath66 is the kronecker delta and @xmath67 is the alternating tensor .
this is the correct quantum mechanical result : the semiclassical approximation in eq .
( 18 ) is _ exact _ for the correlation tensor of an isolated spin in the presence of a fixed magnetic field .
the final stage of the calculation is to average @xmath68 in eq .
( 24 ) over the distribution of @xmath46 in eq .
it is most convenient to do this by re - writing the distribution as @xmath69 so that the schulten - wolynes approximation to @xmath70 becomes @xmath71 where @xmath63 and @xmath64 are now the polar and azimuthal angles of the vector @xmath50 .
rodgers@xcite has shown that when eqs .
( 24 ) and ( 25 ) are substituted into eq .
( 26 ) , all three integrals can be evaluated in closed form .
setting @xmath55 for the electron spin , the final results for the three independent components of the tensor are@xcite @xmath72\right)-4f(\omega_*,t_*)\right]/2\omega_*^3,\ ] ] @xmath73/2\omega_*^2,\ ] ] @xmath74/2\omega_*^3 , \eqno(27)\ ] ] where @xmath75 , @xmath76 , and the one remaining integral @xmath77 can be expressed in terms of complex error functions .
the assumptions of the schulten - wolynes theory are certainly likely to become valid at sufficiently short times as the number of nuclear spins in the radical tends to infinity .
however , it is not clear that they will be so well justified for a radical containing several tens of nuclear spins , which is typical for the radicals of interest in organic radical recombination reactions . in particular , while the assumption of an asymptotic ( gaussian ) distribution for the sum of hyperfine - weighted nuclear spin vectors might be reasonable for such a radical , the assumption that this vector remains fixed in space during the electron spin evolution may well not be .
( the argument that supports neglecting the nuclear spin precession is that in a radical with @xmath9 nuclear spins the precession frequency of the electron spin will be on the order of @xmath78 , whereas that of each nuclear spin will be on the order of @xmath79 , where @xmath79 is the average hyperfine coupling constant.@xcite for sufficiently large @xmath9 , the nuclear spin precession will therefore be much slower than the electron spin precession , and so from the point of view of the electron spin dynamics the nuclear spins can be regarded as constant . while this approximation may well be valid for moderate times for the electron spin in a quantum dot ( @xmath80 ) , it is not at all clear that the separation of time scales will be large enough to justify it when @xmath9 is only a small multiple of ten . ) fortunately , both assumptions can be removed by modifying the schulten - wolynes theory in a fashion suggested by the semiclassical approximation in eq .
suppose we make the same approximation to the traces over the nuclear spins in the radical as we have made for the electron spin , @xmath81 so that each nuclear spin is regarded as a classical vector of length @xmath42 with an initial orientation that is distributed uniformly on the surface of a sphere
. then it should be clear from eq .
( 18 ) that we can approximate the spin correlation tensor @xmath82 by @xmath83 where we have explicitly set @xmath55 and cancelled the factors of @xmath84 in eq .
( 29 ) with the factor of @xmath85 in eq .
the key difference between this and the schulten - wolynes theory is that each nuclear spin in the radical is now an independent vector which is free to precess around the electron spin at the same time as the electron spin precesses around it .
that is , we can use the dynamically coupled equations of motion @xmath86\times { \bf s}_i(t ) , \eqno(31)\ ] ] and @xmath87 to obtain the components @xmath88 of @xmath89 in eq .
( 30 ) , thereby eliminating the second of the assumptions made in the schulten - wolynes theory .
these modifications clearly preclude obtaining a simple closed form expression for @xmath82 : eqs .
( 30 ) to ( 32 ) have to be solved numerically.@xcite however , this is not especially difficult .
the integrals over @xmath90 and @xmath91 in eq .
( 30 ) can be done by monte carlo integration by sampling the initial electron and nuclear spin vectors at random from the surfaces of their respective spheres .
the coupled equations of motion in eqs .
( 31 ) and ( 32 ) could in principle be solved using any one of a number of ordinary differential equation integrators , although we prefer to use a specialised integrator for this purpose as described in appendix a. no matter what integrator one uses , the number of coupled equations that must be integrated increases only linearly with @xmath9 , so the calculation is entirely feasible for a radical with any number of nuclear spins .
as a first test of this modified semiclassical theory , we have used it to calculate the time - dependent triplet probability @xmath92 of the pyrene + _ n , n_-dimethylaniline ( @xmath93py@xmath94 + @xmath93dma@xmath95 ) radical pair that was considered by schulten and wolynes@xcite and subsequently by knapp and schulten.@xcite the spin dynamics of this radical pair had previously been solved quantum mechanically by werner _
et al._@xcite we have adopted the same ( simplified ) set of hyperfine coupling constants as in these earlier studies , namely @xmath96 for @xmath93py@xmath94 and @xmath97 for @xmath93dma@xmath95 . to facilitate a direct comparison with the results reported in refs .
@xcite , we have also converted the unit of time from mt@xmath98 to ns using the gyromagnetic ratio of a free electron ( @xmath99 ns@xmath98mt@xmath98 ) . fig . 1 shows the resulting triplet probabilities for a @xmath93py@xmath94 + @xmath93dma@xmath95 radical pair initially formed in the singlet state , for various magnetic field strengths .
the exact quantum mechanical ( qm ) results are compared with those of the schulten - wolynes ( sw ) theory and our modified semiclassical ( sc ) approximation .
one sees that both semiclassical theories do remarkably well in capturing the overall behaviour of the qm triplet probability , especially at the highest magnetic field strength considered in the figure ( 8 mt , almost an order of magnitude larger than the largest hyperfine coupling constant in the radical pair ) .
however , the present sc theory gives a somewhat more accurate result than the sw theory even at this field strength , and the difference between the two approximations becomes more pronounced as the field strength is decreased . for this problem , the sw approximation clearly works very well .
however the present sc approximation is noticeably more accurate when the field strength is comparable to the hyperfine coupling constants in @xmath93dma@xmath95 ( @xmath100 mt ) , and also in the limit of zero applied magnetic field . .hyperfine coupling constants used in the calculations reported in figs . 2 to 5 . [ cols="<,>",options="header " , ] considerably
more insight into the performance of the sw and sc approximations can be obtained by using them to compute the raw electron spin correlation tensors of a series of radicals of increasing size containing @xmath101 nuclei with hyperfine coupling constants distributed uniformly in the range @xmath102 mt .
the hyperfine constants that we actually used in these calculations were obtained from a random number generator and are listed for completeness in table i. fig .
2 shows the resulting electron spin correlation tensors for a magnetic field strength of 0.5 mt , which lies in the middle of the range of @xmath103 values and so probes the regime in which the zeeman and hyperfine interactions are of a comparable magnitude and contribute equally to the electron spin dynamics .
the qm , sw , and sc results are compared for radicals with 1 , 4 and 16 nuclear spins . in the smallest of these radicals , the components of the qm electron spin correlation tensor exhibit coherent oscillations which are not captured by either semiclassical theory .
however , the present sc method clearly does a better job of following the average of the quantum oscillations than the sw theory , especially at long times .
when there are 4 nuclear spins in the radical , the amplitude of the coherent quantum oscillations is significantly suppressed , and by the time there are 16 nuclear spins the oscillations are washed out almost completely : all that is left is the average classical precessional behaviour , which is captured almost perfectly by the present sc theory .
this is simply a consequence of the quantum decoherence that one expects to see in strongly - coupled multi - dimensional systems.@xcite we have also performed calculations for a variety of other magnetic field strengths and obtained results consistent with those shown in fig . 2 .
in the high - field limit , both the original sw theory and our modified sc theory become exact , for the reasons explained following eq .
( 24 ) the classical vector model gives the exact result for precession around a fixed field . that neither is exact in the low - field limit can be seen from fig . 3 , which shows how the electron spin correlation tensors behave in the absence of a magnetic field .
one again sees that , while neither semiclassical method captures the coherent oscillations in the qm electron spin correlation tensor of a radical with just one nuclear spin , the present sc theory becomes almost quantitatively accurate by the time the radical contains 16 nuclear spins .
this is not the case for the sw theory : the components of the electron spin correlation tensor in eq .
( 27 ) become constant beyond @xmath104 and fail to capture the long - time behaviour of the electron spin dynamics . the fact that the present sc theory begins to agree with quantum mechanics in a radical with 16 nuclear spins
is rather fortunate , because this is close to the maximum number of spins that can be included in an exact qm calculation . in a radical with 16 spin-1/2 nuclei ,
the hilbert space contains 2@xmath105 state vectors , and the qm calculation is already approaching the limit of practical feasibility .
this is illustrated in fig .
4 , which compares the relative computer times of the sc and qm calculations for radicals containing an increasing number of nuclear spins . for a radical with fewer than 10 nuclear spins ,
the sc calculation is significantly slower than the qm calculation , because of the need to run a large number of classical trajectories to converge the monte carlo integration over the initial spin orientations in eq .
( 30).@xcite however , since the computational effort in the sc case increases only linearly with the number of nuclear spins , whereas in the qm case it increases exponentially , the sc calculation must eventually become more efficient , and in our implementation the cross - over occurs at 12 nuclear spins . in view of the results in figs . 2 and 3 , it should be clear that one circumstance in which the present sc theory is likely to do significantly better than the sw theory is in capturing the low field effect in the singlet yield of a radical pair recombination reaction with a slow recombination rate . in order to investigate this ,
we have calculated the singlet yield in eq .
( 12 ) for a radical pair with no nuclear spins in one radical and 12 @xmath101 nuclear spins in the other , as a function of the magnetic field strength @xmath8 and the recombination rate constant @xmath11 .
the reason for taking one of the radicals in the pair to have no hyperfine interactions and the other to have many is that this has been established experimentally to be the situation in which the low field effect is most pronounced.@xcite the reason for stopping at 12 @xmath101 nuclear spins in the large radical is simply one of convenience : these calculations have to be done for a range of magnetic field strengths and for sufficiently long times to ensure the convergence of the integral in eq .
( 12 ) , and so are computationally more demanding than any of the calculations we have reported so far .
while this is not an issue for the semiclassical ( sw and sc ) methods , it is an issue for the qm method that is used to generate exact results for comparison .
5 compares the low field effects obtained from the qm , sc and sw theories for this idealised radical pair , the hyperfine coupling constants in the larger radical being distributed uniformly in the range @xmath102 mt .
the ordinate of the plot is the difference between the singlet yield at magnetic field @xmath8 and the singlet yield in the absence of a field , @xmath106 , which is accessible experimentally as the integrated signal obtained from a modulated mary ( magnetically affected reaction yield ) experiment.@xcite the three panels of the figure show the results for three different recombination rates , between 0.05 and 0.15 mt , all of which are smaller than the average hyperfine coupling in the larger radical . with these parameters , the low field effect in the exact qm calculation
is seen to be quite pronounced , but not unreasonable : low field effects of this magnitude have certainly been observed experimentally.@xcite as one would expect on the basis of the electron spin correlation tensors in figs . 2 and 3
, the present sc theory does a remarkably good job of capturing the correct low field effect , even in a radical pair with as few as 12 nuclear spins .
the sc and qm results are in almost quantitative agreement for all three radical pair recombination rates . however , while it is qualitatively correct in terms of its dependence on the magnetic field strength , the low field effect predicted by the sw theory for this radical pair is far too large by a factor of almost three at the lowest radical pair recombination rate we have considered ( @xmath107 mt ) .
clearly this is because the sw theory does not capture the correct long - time behaviour of the electron spin dynamics , which contributes more to the singlet yield as the recombination rate constant decreases .
this final numerical example therefore provides a clear illustration of the limitations of the sw approximation and the utility of our modification to it .
one important thing we have not been able to establish with the above numerical examples is whether the present sc theory will become even more accurate as the number of nuclear spins in the radical increases further .
however , we do believe that this will be the case on the basis of some recent results from the condensed matter physics literature.@xcite the electron spin dynamics generated by the hamiltonian @xmath108 in eq .
( 2 ) is known in that literature as the central spin problem. this problem has been studied extensively during the last decade because of its relevance to the electron spin dynamics in gaas quantum dots , which contain significantly more ( @xmath109 ) nuclear spins than the radicals of interest in radical pair recombination reactions . early theoretical work on the problem stressed the importance of quantum coherence in the electron spin dynamics and advocated a fully quantum mechanical treatment,@xcite which is clearly only possible in certain limiting cases ( such as the limit of a large applied magnetic field ) .
however , it has since been established quite generally , both on physical grounds and more rigorously , that the classical equations of motion will become exact for measurements of the electron spin dynamics in a radical with sufficiently many nuclear spins.@xcite we find it especially interesting that a fully quantum mechanical treatment was initially thought to be important for studying the electron spin dynamics in quantum dots.@xcite the same can be said for the low field effect in radical pair recombination reactions .
although simple classical vector arguments have provided useful insights into this effect,@xcite it is probably true to say that the majority of the existing literature attributes it to quantum coherences arising from the off - diagonal elements of the electron spin density matrix.@xcite but as we have shown in figs . 2 and 3 ,
these coherences are rapidly quenched in a radical with more than a handful of nuclear spins : all that survives is the average classical precessional behaviour , which accounts almost quantitatively for the low field effect in a radical with as few as 12 spins ( see fig . 5 ) .
from the perspective of simulating spin chemistry with which we began in the introduction , the results in figs . 1 - 5 are clearly very encouraging .
the present sc approximation misses the quantum coherence that is present in the electron spin dynamics of a radical with fewer than 10 or so nuclear spins , but we feel that this is largely irrelevant .
if one wants to simulate such a radical , one can always do so quantum mechanically ( see fig .
as soon as the number of nuclear spins increases to the point where the exact qm calculation becomes impractical , the accuracy of the present sc approximation becomes almost quantitative ( see figs . 2 , 3 , and 5 ) .
the present results therefore open up an exciting avenue for future research . because the sc theory is so simple ,
being based entirely on the primitive classical vector model , it is straightforward to see how to generalise it to include many of the effects that are not included in the zeroth - order spin hamiltonian in eq .
( 1 ) , such as anisotropic hyperfine interactions , exchange and dipolar interactions between the electrons in a radical pair , spin relaxation processes , and so on . to give just one example of the many possible applications of the theory
, it would be straightforward to use it to investigate potential low field effect - enhancing mutations of the carotenoid - porphyrin - fullerene triad that has recently been established as a proof - of - principle for the operation of a chemical compass.@xcite such an investigation would be totally impractical using exact quantum mechanical methodology , because the carotenoid radical in the photo - excited triad contains @xmath110 protons with hyperfine couplings larger than the earth s magnetic field , implying a hilbert space of @xmath111 states . with the present sc theory , this investigation would be entirely feasible , and the results we have presented here suggest that its conclusions would be perfectly reasonable .
we are indebted to david reichman for pointing us to work on the central spin problem in the condensed matter physics community , and to christopher rodgers for his unpublished work on the sw approximation ( ref .
dem acknowledges support from the wolfson foundation and the royal society and pjh acknowledges funding from darpa ( qube : n66001 - 10 - 1 - 4061 ) .
1 . let @xmath113 , and , regarding this as a constant vector , use eqs .
( 22 ) and ( 23 ) to evolve @xmath89 for time @xmath114 .
2 . for @xmath115 , let @xmath116 , and , regarding this as a constant vector , use eqs .
( 22 ) and ( 23 ) with @xmath117 to evolve @xmath118 for time @xmath112 .
3 . let @xmath113 , and , regarding this as a constant vector , use eqs .
( 22 ) and ( 23 ) to evolve @xmath89 for time @xmath114 .
this algorithm is only second order accurate , giving an error of @xmath119 per time step and an error of @xmath120 for propagation over a fixed time interval .
however it is easy to program , has minimal storage requirements , and exactly conserves both the lengths of all ( electron and nuclear ) spin vectors and the classical energy @xmath121 .
one constant of the motion that it does not exactly conserve is the projection of the total ( electronic and nuclear spin ) angular momentum along @xmath122 , which can therefore be used to check the accuracy of the integration .
the @xmath30 , @xmath31 and @xmath32 components of the quantum mechanical electron spin correlation tensor in eq .
( 8) can be computed as follows .
suppose that , for each composite nuclear spin state @xmath123 , we propagate the pair of state vectors @xmath124 with @xmath125 for @xmath126 and @xmath127 for @xmath128 forwards in time to obtain their time - dependent expansion coefficients @xmath129 then it follows from eq .
( 8) and the properties of the electron spin operators that @xmath130 , \eqno(b2)\ ] ] and @xmath131 .
\eqno(b3)\ ] ] the required components of @xmath70 can therefore be obtained by propagating @xmath132 independent pairs of state vectors forwards in time one pair at a time , and accumulating their expansion coefficients in the basis @xmath133 using these equations .
the time evolution of the pair of state vectors is conveniently performed with a symplectic integrator.@xcite this propagates each state vector in the pair @xmath134 through a time interval @xmath112 using the following @xmath135-step algorithm ( for @xmath136 ) , @xmath137 @xmath138 where @xmath139 and @xmath140 give @xmath141 and @xmath142 with an error of @xmath143 .
note that these equations only involve hamiltonian matrix - state vector multiplications , which can be facilitated by exploiting the sparsity of the matrix representation of @xmath144 : each of the @xmath145 basis states is coupled to at most @xmath146 others by the hamiltonian in eq .
( 2 ) . exploiting
this sparsity also leads to an algorithm with a minimal storage requirement , which is particularly useful for radicals with a large number of nuclear spins . in practice
, we have found that the four - stage , fourth - order ( @xmath147 ) symplectic integrator coefficients @xmath148 and @xmath149 given in ref .
@xcite provide an ideal balance between accuracy and efficiency for the present problem .
the classical equations of motion in eqs .
( 31 ) and ( 32 ) are in fact integrable : it is possible to obtain a closed - form solution of these equations subject to arbitrary initial conditions .
however , the solution is far from simple , involving klenian @xmath150 and @xmath151 functions of genus @xmath9 whose arguments require the evaluation of complicated integrals .
so it is not at all clear that integrating this analytical solution over the initial conditions of the spin vectors in eq .
( 30 ) will be any easier than solving the problem numerically .
see : e. a. yuzbashyan , b. l. altshuler , v. b. kuznetsov and v. z. enolskii , j. phys .
a * 38 * , 7831 ( 2005 ) . in all of the sc calculations reported in this paper
, we averaged over 10@xmath152 initial spin orientations , which we checked were enough to give well - converged results by averaging over ten times fewer . | we present a practical semiclassical method for computing the electron spin dynamics of a radical in which the electron spin is hyperfine coupled to a large number of nuclear spins .
this can be used to calculate the singlet and triplet survival probabilities and quantum yields of radical recombination reactions in the presence of magnetic fields .
our method differs from the early semiclassical theory of schulten and wolynes [ j. chem .
phys .
68 , 3292 ( 1978 ) ] in allowing each individual nuclear spin to precess around the electron spin , rather than assuming that the hyperfine coupling - weighted sum of nuclear spin vectors is fixed in space .
the downside of removing this assumption is that one can no longer obtain a simple closed - form expression for the electron spin correlation tensor : our method requires a numerical calculation .
however , the computational effort increases only linearly with the number of nuclear spins , rather than exponentially as in an exact quantum mechanical calculation .
the method is therefore applicable to arbitrarily large radicals .
moreover , it approaches quantitative agreement with quantum mechanics as the number of nuclear spins increases and the environment of the electron spin becomes more complex , owing to the rapid quantum decoherence in complex systems . unlike the schulten - wolynes theory , the present semiclassical theory predicts the correct long - time behaviour of the electron spin correlation tensor , and it therefore correctly captures the low magnetic field effect in the singlet yield of a radical recombination reaction with a slow recombination rate . |
we acknowledge useful discussions with b. altshuler , a. goldman , a. kamenev , v.e .
kravtsov , v. krasnov , and m. norman .
the work was supported by the u.s .
department of energy office of science under the contract no .
de - ac02 - 06ch11357 .
a.a.v . acknowledges support of the fp7-irses program , grant n 236947 `` simtech '' .
we study low - transparency junctions in the regime of weak fluctuations and find the tunneling current @xmath49 between a normal metal electrode and a disordered two - dimensional superconducting film placed in a perpendicular magnetic field throughout the whole phase diagram above the @xmath17 line .
this system can be described by the tunnel hamiltonian with interaction term@xmath50and the tunnel current can be identified with the time derivative of the particle number operator in one of the electrodes@xmath51averaged over the statistical ensemble:@xmath52 \right\rangle .\]]the procedure of such ensemble averages with the density matrix was performed in ref .
the tunnel current is then determined by the diagram presented in fig.[fig.all]*b * appearing in first orders of barrier transparency and strength of fluctuations @xmath15 .
solid lines correspond to the single - electron green s functions in the respective electrodes , the wavy line represents the fluctuation propagator , crossed circles stand for the matrix elements of the tunneling hamiltonian , and the solid triangles are the vertices accounting for impurity averaging . the quasi - particle current flowing through a tunnel junction
is expressed via the correlator @xmath53 of the electron green s functions of both electrodes .
being calculated as a series of imaginary matsubara frequencies@xcite @xmath54 @xmath55 the obtained expression has to be analytically continued into the whole upper half - plane of complex frequencies @xmath56 : @xmath57 .
finally , the real positive values of the latter are identified with the voltage at junction @xmath58 . as it is shown in the appendix , after summations over momenta and fermionic frequencies , the correlator @xmath59
, corresponding to the diagram from fig.[fig.all]*b * becomes @xmath60 \frac{\left [ \mathcal{e}_{m}^{\prime } \left ( k+2\nu \right ) -\mathcal{e}_{m}^{\prime } \left ( k\right ) \right ] } { \mathcal{e}_{m}\left ( |k|\right ) } \notag\end{aligned}\ ] ] the function @xmath61 -\psi \left ( \frac{1}{2}\right)\]]represents the denominator of the fluctuation propagator describing the fluctuation pairing of electrons in the normal phase of a superconductor over a wide range of temperatures and fields @xcite . here
@xmath62 and @xmath63 are dimensionless temperature and magnetic field normalized by the critical temperature and the value of second critical field respectively , @xmath64 is the exponential euler constant .
one can see that close to @xmath38 and for weak enough magnetic fields , @xmath65 is nothing else than the fundamental solution of the time - dependent ginzburg - landau equation lv05 .
the summation over the bosonic matsubara frequencies `` flowing '' through the fluctuation propagator done by an additional analytical continuation in upper half - plane results in the general expression for correlation function @xmath66 @xmath67 } { \mathcal{e}_{m}\left ( 0\right ) } \notag\ ] ] @xmath68 dz\!}{\sinh \left ( \pi z\right ) \sinh \pi \left ( z-\frac{ev}{2\pi t}\right ) \mathcal{e}_{m}\left ( iz\right ) } , \notag\end{aligned}\ ] ] which allows to obtain the fluctuation contribution to the tunnel current for arbitrary temperatures , magnetic fields and voltages .
the corresponding results for the conductivity are presented in figs .
[ fig.all]&[fig.plots ] .
_ the effect of sfs on the dos and corresponding pseudogap in tunnel conductivity_. according to the microscopic bcs theory @xcite , the superconducting state is characterized by a gap in the normal excitation spectrum , centered around the fermi level , @xmath69 , which vanishes along the transition line @xmath17 . however , it was predicted , as early as in 1970 @xcite , that even in the normal state of a superconductor , thermal fluctuations result in a noticeable suppression of the density of states ( dos ) in a narrow energy range around the fermi level ( see fig .
[ fig.ados]a ) .
more specifically , in the case of a disordered thin film @xcite the fluctuation correction to the dos takes form : @xmath70 , \label{deltanu2}\]]where @xmath71 is the electron density of the states per one spin of a normal metal at the fermi level , @xmath72 is the ginzburg - levanyuk number characterizing the strength of fluctuations in the film , @xmath73 is so - called ginzburg - landau time , characterizing the life - time of fluctuating cooper pair and , in accordance to the uncertainty principle , the inverse value of its characteristic energy scale @xmath74 .
one can see that eq .
( [ deltanu2 ] ) is a sign - changing function and its integral over the complete energy range must be equal zero:@xmath75the statement ( [ sumrule ] ) is nothing else as the sum rule : superconducting interactions can not create new states , it just redistributes existing ones to different energy levels .
namely , at the fermi level a sharp dip [ @xmath76 , the precursor of the superconducting gap is formed , while the released states are moved to higher energies , with maximum around @xmath77 the value corresponding to the characteristic energy of fluctuation cooper pairs ( see fig.1a of the main text ) . a major experimental tool for determining the density of states is by measurements of the differential tunnel conductivity . giaever and megerle @xcite ,
related the quasiparticle tunnel current to the densities of electron states of the left and right electrodes and to the difference of the equilibrium distribution functions in both of them ( see eq . ( 1 ) of the main text ) . assuming the left electrode being a normal metal with constant density of states @xmath78 and the right electrode being a thin superconducting film above its critical temperature one can write an explicit expression for the excess tunnel conductivity in terms of @xmath79 and the derivative of the fermi function . combining the latter with the sum rule ( [ sumrule ] )
one finds@xmath80and arrives at the disappointing conclusion that the predicted strong and narrow singularity in the density of states eq .
( [ deltanu2 ] ) manifests itself in the observable tunnel conductivity only as a wide pseudogap structure ( @xmath81 instead of @xmath82 ) and weak in the magnitude ( @xmath83 $ ] instead of @xmath84 ) , resembling that one in the superconducting phase @xcite ( see fig.1b of the main text ) .
indeed , due to the sum rule ( [ sumrule ] ) , almost the whole effect of fluctuations on the tunnel current is averaged out in the process of energy integration .
the strong divergence of eq .
( [ deltanu2 ] ) at zero energy is completely eliminated due to presence of @xmath85 in eq .
( [ tunfull ] ) and only a weak logarithmically singular behavior of the minimum and two bumps of @xmath86 are reminiscent of the closeness to the superconducting transition .
the commonly accepted giaver formula for the tunnel current does not allow to detect traces of the strong singularity of eq .
( [ deltanu2 ] ) , which should be manifested in the conductivity as a narrow zero bias anomaly in tunnel conductivity as we will see below .
one could be curious where does the difference between the microscopic approach and the giaver phenomenology lie ? in order to understand this let us follow the derivation of the latter from the former .
let us perform the summation of the green s functions of each electrode over the corresponding momenta in eq .
( 3 ) of the main text assuming the tunnel matrix elements to be momentum independent .
this makes the integrations of both matsubara green s functions independent and each of them can be presented in lehmann form agd @xmath87substituting eq .
( [ lehmann ] ) into eq .
( 3 ) of the main text , rewriting the product of the energy denominators in the form of simple fractions , and summation over fermionic frequencies , gives@xmath88 . \label{ktan}\end{aligned}\]]looking at this expression one might be tempted to perform an analytic continuation @xmath89 ( @xmath90 ) and apply the sokhotski
plemelj theorem to the integration over @xmath91 in eq .
( [ ktan]):@xmath92where the `` dashed '' integral symbol means that the integral is performed in the sense of a cauchy principal value .
this calculation of the imaginary part of eq .
( [ ktan ] ) with subsequent use of eq .
( 2 ) of the main text immediately reproduces giaever s and megerle s formula , i.e. , in accordance to the common believe , the microscopic approach confirms the phenomenological result .
nevertheless , one should remember , that the validity of eq .
( [ sokh ] ) requires the smoothness of the function @xmath93 .
however , this requirement is violated in the case under consideration : as we saw above , the fluctuation correction @xmath94 close to the transition temperature has a strong singularity at small energies .
hence , performing the integration of the exact expression eq .
( [ ktan ] ) using the rule eq .
( [ sokh ] ) , one looses the effect of the interplay between the parameters @xmath95 and @xmath96 ( or @xmath97 above the second critical field ) . the use of a finite - width @xmath98-function in the sokhotski
plemelj theorem washes out the result and makes the main difference .
we study the effect of sfs on the tunneling current @xmath99 between a normal metal electrode and a disordered two - dimensional superconducting film placed in a perpendicular magnetic field throughout the whole phase diagram above the @xmath17 line . describing this system by means of a tunnel hamiltonian
, the tunnel - current can be expressed in terms of the correlator @xmath100 of the electron green s functions of the corresponding electrodes , which is analytically continued from matsubara frequencies @xmath101 @xmath102 to the upper half - plane of complex frequencies @xmath103 , @xcite : @xmath104 being interested in low - transparency junctions and restricting our consideration to the first order in @xmath105 , one can see that in the case the second electrode is not subject to superconducting fluctuations
e.g. , is a normal stm tip the only diagram which contributes to the tunnel - current is that presented in fig .
2b of the main text .
this diagram describes the suppression of the tunnel - current due to the mechanism of fluctuation renormalization of the quasi - particle density of states , discussed above . in the absence of magnetic fields , the correlation function , eq .
( [ ikgen ] ) , was already studied in momentum representation @xcite .
the generalization to the case of a perpendicular magnetic field can be made by going over from the momentum to landau representation with an appropriate quantization of the cooper pair motion ( see , for example , refs . [ ] ) .
formally , this corresponds to a replacement of the energy associated with the motion of the center of mass of a free cooper pair with momentum @xmath106 by the eigen - energy of the landau state of level @xmath107 : @xmath108 @xmath109 . here
@xmath110 is the electron diffusion coefficient and @xmath111 is the cyclotron frequency corresponding to the rotation of the center of mass of a cooper pair in a magnetic field @xmath112 .
the integration over the two - dimensional momentum in correlator ( [ ikgen ] ) is replaced by a summation over landau levels according to the rule : @xmath113 = \frac{h}{2\pi ^{2}t}\sum_{m=0}^{m}f\left [ \omega _ { \mathrm{c}}(m+\frac{1}{2})\right ] , \]]where @xmath114 is a cut - off parameter related to the elastic electron scattering time @xmath115 ( see ref . [ ] for details ) .
this transformation is applied to the general expression for the correlation function @xmath116 , and one finds @xcite : @xmath117 \frac{\left [ \mathcal{e}_{m}^{\prime } \left ( k+2\nu \right ) -\mathcal{e}_{m}^{\prime } \left ( k\right ) \right ] } { \mathcal{e}_{m}\left ( |k|\right ) } \notag\end{aligned}\]]with @xmath118 @xmath4 being the tunneling resistance of the junction and @xmath35 is its surface area .
the function @xmath61 -\psi \left ( \frac{1}{2}\right ) \label{em}\]]represents the denominator of the fluctuation propagator ( wavy line in fig .
2b ) of the main text ) : @xmath119written in landau representation and describing the fluctuation pairing of electrons in the normal phase of a superconductor over a wide range of temperatures and fields @xcite . here
@xmath62 and @xmath120 are dimensionless temperature and magnetic field normalized by critical temperature and the value of second critical field respectively , @xmath64 is the exponential euler constant .
the cyclotron frequency of a cooper pair rotation in this parametrization is @xmath121 .
we clarify that @xmath122 denotes derivative of the function @xmath123 with respect to its argument @xmath124 , explicitly given by @xmath125 .
\label{deri}\ ] ] the two terms in eq .
( [ k1+k2 ] ) correspond to two fluctuation contributions to the tunnel - current with different analytical properties .
below we demonstrate how these contributions give rise to the pseudogap maxima and the low - bias anomaly ( lba ) in the tunneling conductivity in two - dimensional disordered superconductors .
we start our analysis with the first term of eq .
( [ k1+k2 ] ) .
since the external frequency @xmath26 enters the expression for @xmath126 only via the argument of the analytical function @xmath127 [ see eq .
( k1+k2 ) ] , one can easily perform its analytical continuation by just substituting @xmath128 . using eq .
( [ ikgen ] ) , one finds for the general expression of the corresponding current @xmath129 : @xmath130 the second contribution to the tunneling current is determined by @xmath131with@xmath132 } { \mathcal{e}_{m}\left ( k\right ) } .
\label{fm}\]]here the analytical contribution is more complex than in the case of @xmath133 since the frequency @xmath26 is not only present in the argument of function ( [ fm ] ) but also in the upper limit of the sum over @xmath134 in eq .
( [ i2 ] ) .
note , that this summation limit can be reduced from @xmath135 to @xmath136 since @xmath137 .
the analytical continuation of a function of the form @xmath138onto the upper half - plane of complex frequencies was performed in ref .
[ ] [ see also ref .
[ ] , equation ( 7.90 ) ] . by means of the eliashberg transformation @xcite
the corresponding sum can be presented as a counterclockwise integral over a closed contour @xmath139 consisting of two horizontal lines , two vertical lines , and two semicircles in the upper complex plane , where the latter exclude the points @xmath140 and @xmath141 ( see fig .
[ mtcontour ] ) : @xmath142the integrals over the vertical line segments become zero , the integral over the semi - circle at @xmath143 is zero since @xmath144 , the integral over the semi - circle at @xmath145 reduces to the residual of @xmath146 .
inverting the direction of integration over the line segment with @xmath147 and then shifting the integration variable as @xmath148 in the corresponding integral , one finds : @xmath149 dz .
\label{resi}\end{aligned}\ ] ] eq .
( [ resi ] ) is already an analytical function of @xmath26 and one can perform its continuation just by the standard substitution @xmath150 . shifting the variable in the second integral again as @xmath151 and using the identity @xmath152one finally finds@xmath153substituting the explicit expression for function @xmath154 from eq .
( [ fm ] ) into eq .
( [ thetar ] ) results in @xmath155 } { \mathcal{e}_{m}\left ( 0\right ) } \right .
\label{kr2 } \\ & \left .
+ i\sinh \left ( \frac{\omega } { 2t}\right ) { { \mathop{\,\,\vphantom{\intop}\!\!\ ! \mathpalette { \ooalign{$\relax\intop$\cr\hidewidth$\relax-$\hidewidth}}\nolimits}}\limits_{-\infty } ^{\infty } \frac{\left [ \mathcal{e}_{m}^{\prime } \left ( iz\!-\!\frac{i\omega } { \pi t}\right)\!-\ ! \mathcal{e}_{m}^{\prime } \left(\!-iz\right ) \right ] dz\!}{\mathcal{e}_{m}\left ( -iz\right ) \sinh \left ( \pi z\right ) \sinh \pi \left ( z+\frac{\omega } { 2\pi t}\right ) } \right\ }
. \notag\end{aligned}\ ] ] let us note that the first term of @xmath157 is nothing but half of the first summand ( with @xmath158 ) of the sum in eq .
( i1 ) with opposite sign .
technically it would be easy to incorporate the latter into @xmath159 .
however , such a procedure would be physically misleading : we will see below that this @xmath134 and @xmath160 independent term in eq .
( [ kr2 ] ) cancels the corresponding linear contribution stemming from the integral term at small voltages . as a result , the current @xmath161 , determined by the imaginary part of eqs .
( [ kr2 ] ) , does not contain a linear contribution if expanded in powers of voltage .
this means that it does not contribute to the magnitude of the differential tunnel conductivity at zero voltage @xmath162 , which is the easiest quantity to measure in experiments .
nevertheless , it contributes to the current - voltage characteristics at finite voltages and , as we will see below , can noticeably manifest itself even at very low voltages @xmath163 as a lba .
@xmath167 + \operatorname{im}{\mathcal e}_{m}\left ( iz\right ) \left [ \operatorname{im}{\mathcal e}_{m}^{\prime}\left ( iz - iev\right ) + \operatorname{im}{\mathcal e}_{m}^{\prime}\left ( iz\right ) \right ] } { \sinh\left ( \pi z\right ) \sinh\left [ \pi\left ( z - ev/2\pi t\right ) \right ] \left [ \operatorname{re}^{2}{\mathcal e}_{m}\left ( iz\right ) + \operatorname{im}^{2}{\mathcal e}_{m}\left ( iz\right ) \right ] } \right\ } .\nonumber\end{aligned}\ ] ] eq .
( [ itot ] ) is the main result of this work .
the first term @xmath168 has been studied in detail for different limiting cases using different approaches : close to @xmath38 , vd83 , ccrv90,v93,l10 : ( i ) in a wide temperature range in zero field vd83 , or ( ii ) close to @xmath38 in magnetic fields @xmath169 , r93 . the current contribution
@xmath170 has been omitted in all these works based on the `` standard '' argument that the zero frequency bosonic mode ( which traverses through the propagator ) is singular in the vicinity of the transition .
however , it is known that this argument sometimes works ( e.g. , in the case of the maki - thompson contribution to conductivity @xcite ) , but also sometimes fails ( e.g. , for the aslamazov - larkin contribution to conductivity al68 ) . in our case
this argument turns to out be correct only for very small voltages .
the reason being that voltage itself , together with temperature deviations from the transition point and finite magnetic fields , drives the system away from the immediate vicinity of the transition , which invalidates the argument regarding the dominance of the zero frequency bosonic mode .
we present several plots of the tunnel - current and the tunnel conductance .
since they depend on three parameters : @xmath45 , @xmath44 , and @xmath171 , only lines or planes in the full parameter space are presented as line or surface plots . in fig .
3 of the main text all parameter points and lines in the @xmath45-@xmath44 phase diagram for all following figures are shown .
the critical field line , @xmath172 , separating the superconducting ( sc ) and the normal fluctuation region ( sf ) is defined by @xmath173 .
each figure caption refers to these parameter locations .
[ fig.tunneliv ] shows the behavior of @xmath174 near @xmath38 and @xmath47 . in the following
we will carefully analyze the effect of superconducting fluctuations in the whole phase diagram .
we start our discussion with the regular contribution @xmath164 and then elucidate the important role of the anomalous contribution @xmath175 , which was neglected in literature so far .
[ [ analysis - of - the - asymptotic - behavior - of - ileft - mathrmregright - left - vright ] ] analysis of the asymptotic behavior of @xmath164 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ _ close to @xmath38 and for sufficiently weak magnetic fields @xmath176 _ , the most singular term in eq .
( [ i1 ] ) arises from the zero frequency bosonic mode @xmath158 , when the propagator has a pole at @xmath177 and @xmath178with @xmath179 as reduced temperature .
the summation over landau levels can be performed in terms of polygamma - functions , @xmath180 , and one finds an expression valid for any combination of @xmath181 and @xmath182 : @xmath183 \notag \\ & \cdot \operatorname{im}\psi ^{\prime } \left ( \frac{1}{2}-\frac{iev}{2\pi t}\right ) .
\label{eq.ireg}\end{aligned}\]]eq .
( [ eq.ireg ] ) reproduces the results of refs .
the corresponding contribution to the tunneling conductance is @xmath184 \notag \\ & \cdot \operatorname{re}\psi ^{\prime \prime } \left ( \frac{1}{2}-\frac{iev}{2\pi t}\right ) .\end{aligned}\ ] ] _ in the region of high temperatures @xmath185 and zero magnetic field _ we restrict our analytical consideration to the fluctuation contribution to the differential conductivity at zero voltage . performing an integration instead of a summation in eq .
( [ i1 ] ) one finds@xmath186which is again in complete agreement with ref .
[ ] . _ close to the line @xmath187 and for sufficiently low temperatures @xmath188 _ the lowest landau level approximation ( lll ) holds .
the corresponding propagator ( with quantum number @xmath189 has a pole structure and eq .
( [ em ] ) acquires the form : @xmath190with @xmath191 . keeping only the @xmath192 term in eq .
( [ i1 ] ) , one can write @xmath193\!=\!-\frac{2et_{c0}sh}{\pi ^{3}\sigma _ { n}r_{n}}\sum_{k=0}^{\infty } \frac{\operatorname{im}\mathcal{e}_{0}^{\prime } \left ( k-\frac{iev}{\pi t}\right ) } { \widetilde{h}+\frac{\pi ^{2}tk}{4h_{c2}(t ) } }
. \label{sum4}\]]the imaginary part @xmath194 can be explicitly written using eq .
( [ deri ] ) in the limit @xmath195 and the asymptotic behavior of @xmath196 : @xmath197 ^{2}+\left ( \frac{ev}{\pi t}\right ) ^{2}}.\]]the summation in eq .
( [ sum4 ] ) can then be performed exactly in terms of polygamma - functions , i.e. , using @xmath198which gives an expression for the regular part of the fluctuation current valid for low enough temperatures along the line @xmath172 : @xmath199 = -\frac{2est_{c0}h}{\pi ^{3}\sigma _ { n}r_{n}}\frac{v_{t}}{1+v_{t}^{2}}\left\ { \left [ \ln \left ( \frac{4h_{c2}\left ( t\right ) } { \pi ^{2}t}\right ) \sqrt{1+v_{t}^{2}}\allowbreak \allowbreak -\psi \left ( \frac{4h_{c2}}{\pi ^{2}t}\widetilde{h}\right ) \right ] -\frac{\arctan v_{t}}{v_{t}}\right\}. \label{curr1a}\ ] ] here , we introduced the dimensionless voltage @xmath200 which defines the characteristic scale of @xmath201 in the considered domain of the phase diagram .
we stress , that this scale depends on temperature via the parameter @xmath202 .
_ close to @xmath203 , in the region of very low temperatures @xmath204 _ , the argument of the @xmath205-function in eq . ( [ curr1a ] ) becomes large despite the smallness of @xmath206 , and the @xmath205-function can therefore be approximated by its asymptotic expression .
one gets @xmath207 \label{qf}\end{aligned}\]]with @xmath208 being the value of bcsgap .
the characteristic scale where the maximum of the tunnel conductance appears at these low temperatures is @xmath209 , i.e. @xmath210 _ in the region of high fields @xmath211 and low temperatures _ , the asymptotic behavior of the tunneling current can be studied in complete analogy to the case of high temperatures and weak fields .
the sums in eq .
( [ i1 ] ) can be approximated by integrals , which gives for the value of the differential conductivity at zero voltage : @xmath212one can see that this dependence is exactly the same as that one in the case of high temperatures with reversed roles of the reduced temperature and the reduced field .
( top ) and at low temperatures near @xmath213 ( bottom ) .
the regular part is presented by a solid line ( red ) , the anomalous by crossed line ( green ) and the their sum , i.e. , the total fluctuation contribution , is shown by a dashed line ( blue ) .
( top ) parameter point in fig .
[ fig.plots ] is @xmath214 [ see also fig .
[ fig.plots ] c ) ] , ( bottom ) @xmath215.,title="fig : " ] ( top ) and at low temperatures near @xmath213 ( bottom ) .
the regular part is presented by a solid line ( red ) , the anomalous by crossed line ( green ) and the their sum , i.e. , the total fluctuation contribution , is shown by a dashed line ( blue ) .
( top ) parameter point in fig .
[ fig.plots ] is @xmath214 [ see also fig . [ fig.plots ] c ) ] , ( bottom ) @xmath215.,title="fig : " ] in the low - voltage limit , @xmath216 , the general expression for @xmath217 , ( [ itot ] ) , can be expanded in small @xmath95 .
we start with the first order term of that expansion , where one can assume @xmath218 in the argument of integrand function and obtain in the region of temperatures close to the transition temperature @xmath38 and along the transition line for temperatures @xmath188 , the propagator has a simple pole structure [ see eqs . ( [ em0 ] ) and ( [ eline ] ) ] and the integral in eq .
( [ ian ] ) can be calculated analytically . performing this integration
one finds that the second term of eq .
( [ ian ] ) exactly annihilates the linear part of the first term .
this fact justifies the static approximation ( zero frequency bosonic mode ) made in refs .
yet , this static approximation turns out to be valid only for very low voltages . expanding the integrand in eq .
( [ itot ] ) to higher orders in voltage reveals an unexpected result .
one can see that the voltage @xmath16 enters the integrand of eq .
( [ itot ] ) in two different places : in the argument of @xmath220 in the numerator and in the argument of @xmath221 in the denominator .
the expansion of @xmath220 results in the appearance of a weakly voltage - dependent term of the order of @xmath222 in @xmath223 , while , as one can easily verify , the expansion of @xmath224 up to the third order in voltage after integration leads to a very singular correction@xmath225 } .
\notag\end{aligned}\ ] ] the strong divergency of this expression at small frequencies indicates that the process of generating current @xmath226([ancur ] ) should be limited in time .
indeed , from the physical picture described above , it is clear that the processes of anomalous cooper pairings of the injected electrons take place until the latter remain non - thermalized .
, i e. for times shorter than @xmath227 hence the frequency integral should be cut - off at @xmath228 what in dimensionless variables corresponds @xmath229 @xmath230 .
further integration is trivial and one finds for the non - linear current the expression@xmath231which valid for @xmath232 eq .
( [ difan ] ) describes two different regimes .
the first corresponds to the growth of the lba when temperature approaches @xmath38 but @xmath96 remains larger than the inverse energy relaxation time @xmath234@xmath235notoriously that the magnitude is directly proportional to the border area volue where the energy relaxation of the injected electrons takes place .
when @xmath96 reaches the value of @xmath236 the lba is saturated and does not grow more:@xmath237the complete expression for small voltages ( @xmath238 ) and in the case of low energy relaxation ( @xmath239 ) is@xmath240 .\]]from this expression one can estimate for the width of the peak : in the case of strong energy relaxation the anomalous contribution becomes of the order of the higher contributions of the regular part and it is not observable on the background of the pseudogap structure .
the effect of both fluctuation contributions , @xmath242 and @xmath217 , on the tunneling conductance is demonstrated in fig .
[ tuncond_reg_an ] .
similar behavior can be observed along the whole line @xmath243 the singularity in the low voltage behavior of tunneling conductance rapidly smears out when moving away from the transition line or increasing the temperature [ see fig .
[ fig.plots ] ] .
the temperature , magnetic field , and voltage dependencies of the tunneling conductance due to superconducting fluctuations , calculated numerically based on eq .
( [ itot ] ) , are presented in figs .
3a , f , h ) as surface plots . the numerical procedure to calculate the @xmath134-sum of the first term needs to take into account its relatively slow convergence .
therefore it is calculated explicitly up to a threshold at which the sum can be replaced by an integral and the polygamma functions by their asymptotic behavior .
( here we use as threshold-@xmath134 , the value @xmath244 at which the argument of the function @xmath245 reaches @xmath246 ) .
the `` rest''-integrals are calculated with inverse integration variable using a gauss - legendre method .
the second term requires a careful treatment of the two integrable poles , which is done by analytical calculation of the residuals in a small interval around them , where the denominator is linearized .
also the numerical integration outside the pole intervals is done by using adaptive integration point distances .
the overall behavior of both terms of the tunnel - current results in a pronounced pseudo - gap structure of the conductance near the superconducting region .
it is the non - linear anomalous term of the tunnel - current which is responsible for the fine structure ( `` local maximum '' ) at the center of the gap , the lba . at this point
it is worth mentioning that another sharp fine structure of tunnel conductance which should occur in the same scale @xmath247 was predicted in ref .
this structure appears due to _ interaction of fluctuations _ as the second order correction in ginzburg - levanyuk number @xmath248 ( but still in first order in the barrier transparency ) .
this contribution has an interference nature ( analogously to maki - thompson process ) and , in contrast to the discussed above nonlinear contribution @xmath249 ^ 2 $ ] , diverges at zero voltage as @xmath250 ^ 2\ln [ ( t - t_{c0})/ev]$ ] .
such divergency , in complete analogy to maki - thompson contribution , is cut off by any phase - breaking mechanism @xcite . analyzing the surface plot representation of the experimental results of ref .
[ ] , obtained at temperature close to @xmath38 , one notices their striking similarity to the theoretical surfaces presented in fig .
[ 3dt ] . indeed , the authors of ref .
[ ] mentioned the agreement of their results with the theoretical prediction of ref .
[ 3dh ] shows how the corresponding surface transforms at low temperatures and strong magnetic fields close to @xmath251 .
it is interesting to note that the behavior of the general expression ( itot ) clearly shows growth of the fluctuation effects in the domain of intermediate temperatures and magnetic fields , beyond the immediate vicinity of @xmath38 and @xmath47 , see plots of the zero - bias tunnel conductance @xmath252 in figs .
[ fig.plots]b,e ) . in fig .
[ fig.plots]f ) one can see the evolution of the pseudogap near the @xmath172 line ( slightly offset by a factor @xmath253 , see caption ) , exhibiting a deeper suppression for intermediate temperatures and fields .
this fact is in agreement with the general ideas of the theory of fluctuations establishing the growth of fluctuations strength ( characterized by the ginzburg - levanyuk number ) as one moves away from the extreme points [ @xmath38 and @xmath47 ] of the curve @xmath17 ( see chapter 2 of ref . [ ] ) . | electron tunneling spectroscopy pioneered by esaki@xcite and giaever@xcite offered a powerful tool for studying electronic spectra and density of states ( dos ) in superconductors .
this led to important discoveries that revealed , in particular , the pseudogap in the tunneling spectrum of superconductors above their critical temperatures@xcite .
however , the phenomenological approach of ref .
@xcite does not resolve the fine structure of low - bias behavior carrying significant information about electron scattering , interactions , and decoherence effects . here
we construct a complete microscopic theory of electron tunneling into a superconductor in the fluctuation regime .
we reveal a non - trivial low - energy anomaly in tunneling conductivity due to andreev - like reflection of injected electrons from superconducting fluctuations .
our findings enable real - time observation of fluctuating cooper pairs dynamics by time - resolved scanning tunneling microscopy measurements and open new horizons for quantitative analysis of the fluctuation electronic spectra of superconductors
. there have been rapid developments in scanning tunneling microscopy ( stm ) or scanning tunneling spectroscopy ( sts ) studies of superconductivity triggered by investigations of the pseudogap state and vortex state in high - temperature cuprates@xcite , observations of the pseudogap in 2d disordered films of conventional superconductors benjamen10 , investigations of the superconductor - insulator transition sacepe , measurements of the tunnel conductivity close to the superconducting transition in intrinsic josephson junctions ( see ref .
@xcite ) , and many others .
all this called for a quantitative theory capable to adequately describe high resolution stm / sts data uncovering subtle features of the tunneling spectra .
of special importance is the ability of analyzing data in the fluctuation regime as it is the domain that is key to reveal the microscopic mechanisms of high temperature superconductivity and the superconductor - insulator transition .
however , the restrictions of the phenomenological gm approach disguise the fine structure of the electronic spectrum . to see how this is happening ,
let us inspect the classical gm expression@xcite for the tunnel current @xmath0 \notag \\ & & \cdot \nu _ { l}\left ( e+ev\right ) \nu _ { r}\left ( e\right ) de,\ , \label{giaver}\end{aligned}\ ] ] , versus energy , @xmath1 , for 2d superconductors above the critical temperature for temperatures close to the critical one ( @xmath2 showing a pronounced divergence at zero energy . * b ) * the resulting pseudogap in the tunneling conductivity obtained by applying the gm eq .
( 1 ) to the fluctuation correction to the @xmath3
. * c ) * the low - voltage anomaly of the tunneling conductivity related to andreev - like reflection of injected electron from the fluctuating superconductive domain , which is beyond the possibilities of the gm approach . ]
( here @xmath4 is the tunnel junction resistance , @xmath5 is the fermi distribution function , and @xmath6 is the energy dependent density of states of the left ( right ) electrode , respectively ) and apply it to the calculation of the tunnel current of a n - i-(n+sf ) junction at temperatures above @xmath7 . using the explicit expression for the fluctuation correction to the electronic dos in a disordered superconducting film @xcite ( shown in the fig.[fig.ados]a ) , one sees with surprise that the sharp singularity in the dos at low energies gets smoothened out to a much wider pseudogap structure in the differential conductivity .
in particular , the latter has the width @xmath8 [ instead of @xmath9 and a small amplitude [ @xmath10 $ ] instead of @xmath11 in the dos @xcite ( see fig.fig.ados**b * * ) .
the reason for these dissimilarities is that the sign - change of the dos fluctuation correction ( see fig.fig.ados**a * * ) , almost averages out the whole effect of fluctuations on the tunnel current when integrated over energy . as a result
, quantum coherent effects like andreev reflections of injected electrons at domains of superconducting fluctuations in the biased electrode ( see fig.[fig.ados]*c * ) can not be described by the gm phenomenology . to construct a general approach to calculate the true tunnel conductivity taking into account the fine structure of the density of states , we employ the matsubara green functions technique .
the complete fluctuation contribution to the tunneling current in a typical stm / sts experiment sketched in fig.[fig.all]*a * is represented graphically by the matsubara diagram , shown in fig.[fig.all]*b*. this diagram describes both , regular and anomalous fluctuation tunneling processes depicted in fig.[fig.all]*a@xmath12 * and fig.[fig.all]*a@xmath13*. the former one , related to the depletion of the electron dos close to the fermi level , has already been discussed above in the framework of the phenomenological theory @xcite and , as we know , results in the appearance of the pseudogap - like feature in @xmath14 .
the latter process consists of andreev - like reflections of an injected , still energetically unrelaxed , electron from the fluctuation superconducting domain in the biased electrode , shown in fig.fig.all**a@xmath13**. in order to participate in fluctuation cooper pairing , the injected electron `` extracts '' an electron - hole pair from vacuum with momentum opposite to its own , forms a cooper pair with the electron , while the remaining hole returns along its previous trajectory ( see fig.[fig.all]*a@xmath13 * ) .
this quantum coherent contribution is missed by the phenomenological method , but is captured by the microscopic diagrammatic approach .
this anomalous tunneling process gives rise to an additional current , which , like the regular one , is proportional to the first power of the ginzburg number @xmath15 ( which characterizes the strength of fluctuations ) , but is cubic in voltage @xmath16 near zero bias and becomes relevant only close enough to the superconducting transition . as a result , a peculiar _ low - bias anomaly _
( lba ) appears near the superconducting transition line @xmath17 .
as the external parameter values move away from the transition line the amplitude of the lba rapidly decays .
the important feature of this novel andreev process is that it appears in lowest ( first ) order approximation with respect to the tunneling barrier transparency the same order as the usual tunneling current exhibiting the pseudogap .
this effect is stronger than the standard andreev conductance of a n - i - s junction which is proportional to the square of the transparency@xcite .
the reason is that the fluctuation - induced domain of superconducting phase in the biased electrode is not separated from the surrounding normal phase by any barrier and thus the process of andreev - like reflection does not involve an additional tunneling process .
remarkably , both complimentary physical processes shown in panels * a@xmath12 * and * a@xmath13 * of fig.[fig.all ] are straightforwardly expressed in terms of a graphic mathematical language : the calculation of the diagram of fig.[fig.all]*b * is reduced to the evaluation of the integrals of the electron green functions in the linked electrodes along two contours in the complex frequency plane shown in panels * b@xmath12 * and * b@xmath13 * of fig.[fig.all ] , respectively .
the upper contour corresponds to the conventional giaever - megerle ( gm ) tunneling , while the lower one describes the contribution due to andreev - like reflection from superconducting fluctuations .
accordingly , the fluctuation part of the tunneling conductance shown in fig.fig.all**c * * exhibits both , the pseudogap anomaly due to fluctuation depletion of the one - electron dos ( fig.[fig.all]*c@xmath18 * ) coming from the integration over the contour of fig.[fig.all]*b@xmath12 * , and andreev - like reflection induced lba ( fig.[fig.all]*c@xmath13 * ) , arising from the integration over the contour of the panel * b@xmath13*. important to remark is that the latter contribution is zero at zero bias voltage [ see fig.[fig.plots]*c * ] . in the framework of the diagrammatic matsubara formalism the tunneling current is presented as ( see methods ) : @xmath19where @xmath20here @xmath21 and @xmath22 are the exact matsubara green functions of the left and right electrodes respectively , the summations are performed over all fermionic frequencies @xmath23 and the electron states @xmath24 and @xmath25 * * * * in the corresponding electrodes . the external bosonic frequency @xmath26 accounts for the potential difference between the electrodes and the factor @xmath27 is due to the summation over the spin degrees of freedom .
the superscript r in eq .
( 2 ) means that the correlator @xmath28 is continued to the plane of complex voltages in such a way that it remains an analytic function through the complete upper complex half - plane .
the calculation of the sums in eq.([k ] ) is presented in the appendix .
it turns out that the discussed lba in the i - v characteristics appears only in the case where the energy ( or phase ) relaxation time @xmath29 of an electron injected into the explored electrode is long enough : @xmath30 the shape of the lba close to the critical temperature [ @xmath31 for low voltages @xmath32 , can be found analytically : @xmath33 , \label{final}\]]with @xmath34 as the electrode normal conductivity and @xmath35 as the junction surface area .
when @xmath36 decreases to the value @xmath37 the growth of the lba ceases .
one can show that close to the transition temperature @xmath38 the dip in the tunnel conductivity develops on the scale @xmath39 . at zero temperature , close to the second critical field @xmath40
the fluctuations acquire quantum nature and the corresponding voltage scale is @xmath41 / h_{c2}\left ( 0\right ) } \ll \delta _ { \mathrm{bcs}}$ ] . from the obtained eq .
( [ final ] ) one sees , that the intensity of the lba is directly proportional to the energy relaxation length @xmath42 which is in a complete agreement with the physical picture of this non - trivial quantum coherence effect presented above : anomalous cooper pairings take place only in a stripe of volume @xmath43 in the contact area , where the injected electrons still remain non - thermalized and differ from the local ones .
fig.[fig.plots ] shows the plots of fluctuation contributions to the tunneling conductivity for different parts of the temperature - magnetic field phase diagram of the superconducting film . the central panel
the @xmath44-@xmath45 phase diagram depicts the parameter combinations or ranges for the 2d graphs or 3d surface plots arranged around it in panels a ) - h ) . in accordance with the above theoretical speculations the strength of the singularity in the low - voltage behavior of the tunneling conductance smears out when moving away from the transition line ( panels a - d and g , h ) .
we point out that the lba is most pronounced roughly halfway between the endpoints of the transition line ( see panel f ) .
overall the panels clearly show that the lba is a pronounced important effect near the transition and is even noticeable at twice the transition temperature . in conclusion
, the lba provides an irreplaceable tool for determining microscopic material parameters including the energy relaxation time @xmath46 , the critical temperature @xmath38 , and the critical magnetic field @xmath47 by measuring the tunneling conductance and fitting the experimental data with the complete expression for the tunneling conductance .
remarkably , all the information about these parameters is encoded in merely the distance between the lba dips and the height of the central peak in the conductivity curve .
this introduces a new technique , a _ tunnel - fluctuoscope _ , in analogy to the recently developed conductivity fluctuoscopy@xcite .
the latter has already proven to be the first quantitative and precise method to determined material parameters of superconducting films in many instances .
an observation of the described lba in a d.c .
experiment is a fingerprint of the fact that at the point below the stm tip , fcps appear during the time of the experiment .
recent tunnel - current measurements of n - i - s junctions indeed indicate the presence of the lba@xcite . since the characteristic fcp lifetime is @xmath48 , a time - resolved stm measurement utilizing an a.c .
current with frequency on the scale of 1 - 10 ghz promises to make it possible , in principle , to visualize fcp directly in real time . |
a _ simplicial complex _ @xmath0 , on a finite set @xmath1 , is a set of subsets of @xmath1 closed under inclusion ; elements of @xmath0 are called _
faces_. the _ dimension _ of a face @xmath2 is @xmath3 and of a complex @xmath0 is the maximum of the dimensions of its faces .
a complex whose maximal faces , called _ facets _ , are equi - dimensional is called _ pure _ and a complex with a unique maximal face is called a _ simplex_. given two simplicial complexes @xmath0 and
@xmath4 with disjoint vertex sets , we define the _ join _ of @xmath0 and @xmath4 to be the simplicial complex @xmath5 if @xmath6 is a vertex , then @xmath7 is said to be the _ cone _ of @xmath4 with _ apex _ @xmath8 .
let @xmath2 be a face of @xmath0 , then the _ link _ and _ deletion _ of @xmath2 from @xmath0 are given by @xmath9 following ( * ? ? ?
* definition 2.1 ) , a pure complex @xmath0 is said to be to be _ vertex - decomposable _ if either @xmath0 is a simplex or there exists a vertex @xmath10 , called a _ shedding vertex _
, such that both @xmath11 and @xmath12 are vertex - decomposable .
checking if a particular simplicial complex is vertex - decomposable can be done using a computer program such as @xcite , in particular , the package described in @xcite provides the appropriate methods .
a _ poset _ @xmath13 is a set with a partial ordering , that is , a binary relation `` @xmath14 '' over the set which is reflexive , antisymmetric , and transitive .
given @xmath15 with @xmath16 , the _ interval _ @xmath17 $ ] is the sub - poset @xmath18 $ ] .
a _ chain _ is a sequence @xmath19 of elements of @xmath13 ; such a chain is said to have _ length _ @xmath20 .
the poset @xmath13 is called _ pure _ if every maximal chain has the same length .
the poset @xmath13 is _ bounded _ if there exists a unique minimal element , @xmath21 , and a unique maximal element , @xmath22 , of @xmath13 .
let @xmath23 be a subset of @xmath13 .
the _ meet _ of @xmath23 , if it exists , is the infimum of @xmath23 and is denoted @xmath24 .
the _ join _ of @xmath23 , if it exists , is the supremum of @xmath23 and is denoted @xmath25 .
if every distinct pair of elements in @xmath13 has a meet and a join , then @xmath13 is called a _ lattice _ ; if every subset of @xmath13 has a meet and a join , then @xmath13 is called a _
complete _ lattice .
let @xmath13 be a finite poset . for elements @xmath26
, we say that @xmath27 _ covers _
@xmath28 , denoted @xmath29 , if @xmath30 and @xmath31 implies @xmath32 ; in this case , we also say @xmath28 is _ covered by _
an _ atom _ of a bounded poset is an element which covers @xmath21 , the unique minimal element of @xmath13 . in @xcite , bjrner and wachs
define a bounded pure poset @xmath13 to _ admit a recursive atom ordering _ if either @xmath13 has maximal chains of length one or there is an ordering of the atoms , @xmath33 , of @xmath13 which satisfies the properties : 1 .
for @xmath34 , @xmath35 $ ] admits a recursive atom ordering on its atoms @xmath36 with the property that there exists a @xmath37 such that @xmath38 for some @xmath39 if and only if @xmath40 , and 2 . for @xmath41 ,
if @xmath42 , then there is a @xmath43 and @xmath44 such that @xmath45 .
associated to every finite poset @xmath13 is the _ order complex _
, denoted @xmath46 , which is a simplicial complex with faces given by chains in @xmath13 ; the facets of @xmath46 are exactly the maximal chains of @xmath13 , hence @xmath13 is pure if and only if @xmath46 is pure .
notice that the minimal non - faces of @xmath46 are exactly the pairs of incomparable elements of @xmath13 , so @xmath46 is a _
flag complex_. in @xcite , boij and sderberg made a pair of conjectures about the possible graded betti numbers of graded modules up to multiplication by positive rational numbers ; the conjecture was proven in @xcite and @xcite . in the course of their construction , they define a family of posets which we recall here .
[ def : boij - soederberg - posets ] let @xmath47 be strictly increasing sequences with @xmath48 for @xmath49 .
define the _ boij - sderberg poset _ of @xmath50 and @xmath51 to be the set @xmath52 of strictly increasing sequences @xmath53 such that @xmath54 for @xmath49 endowed with the partial ordering defined by @xmath55 whenever @xmath56 for @xmath49 .
the _ boij - sderberg order complex _ of @xmath50 and @xmath51 is the order complex of the boij - sderberg poset @xmath52 , that is , @xmath57 .
notice that in @xcite , the poset @xmath52 has the associated pure diagrams as vertices , but these are equivalent as pure diagrams are in bijection to strictly increasing sequences in @xmath58 .
we further note that @xmath57 is a pure complex by ( * ? ? ?
* proposition 2.7 ) .
[ exa : poset ] let @xmath59 and @xmath60 .
figure [ fig : poset ] gives both @xmath61 and its order complex .
note that we use concatenation of the sequence to label the vertices , e.g. we label @xmath62 as @xmath63 . ]
further , the family of boij - sderberg posets contains the family of posets of bounded root sequences described in ( * ? ? ?
* section 8) where it is shown that the supernatural cohomology tables of root sequences in a bounded range give a geometric realisation of the order complex of the associated bounded root sequences .
these are in turn used to prove the boij - sderberg conjectures .
let @xmath64 be strictly increasing sequences with @xmath48 for @xmath49 .
first we see that the order complex has very nice structure . [
pro : lattice ] the poset @xmath52 is a bounded complete lattice .
let @xmath65 .
define @xmath66 by @xmath67 and @xmath68 , for @xmath49 .
then @xmath69 for @xmath70 .
suppose @xmath71 for @xmath70 , then @xmath72 for @xmath49 and @xmath70 , so @xmath73 and @xmath74 is the meet of @xmath75 .
similarly , @xmath76 is the join of @xmath75 .
notice further that , by definition , @xmath77 for all @xmath78 .
thus , @xmath52 is a bounded complete lattice .
next we see that reversing the order on @xmath52 yields a ( possibly different ) boij - sderberg poset .
[ pro : reverse ] let @xmath79 be the set @xmath52 endowed with the partial ordering given by @xmath80 if @xmath81 for @xmath49 .
then @xmath79 is isomorphic to some boij - sderberg poset . without loss of generality ,
assume @xmath82 .
let @xmath83 and define the map @xmath84 from @xmath85 onto itself by @xmath86 and extend @xmath84 component - wise to @xmath87 .
further , define the map @xmath88 from @xmath87 onto itself by @xmath89 .
notice @xmath90 is a bijection .
then for @xmath91 , we have that @xmath92 that is , @xmath93 exactly reverses the order of elements in @xmath52 and hence @xmath94 is isomorphic to @xmath79 .
an immediate result of this is that particular posets are isomorphic to themselves after reversing the order .
[ cor : linear - symmetric ] let @xmath95 and suppose @xmath96 and @xmath97
. then @xmath52 is isomorphic to @xmath79 .
let @xmath84 and @xmath88 be as in the proof of proposition [ pro : reverse ] ; recall that @xmath90 is a bijection .
applying @xmath90 to @xmath50 , we get @xmath98 thus , @xmath90 is a poset isomorphism , that is @xmath99 .
this then allows us to see that some posets can be simplified .
[ pro : shearing ] suppose @xmath100 and @xmath101 for some @xmath102 . then @xmath103 by corollary [ cor : linear - symmetric ] , @xmath104 .
let @xmath105 be given by @xmath106 .
extending @xmath107 component - wise provides a bijective map from elements of @xmath79 to elements of @xmath108 .
further still , for @xmath109 , we have that @xmath80 if and only if the @xmath110 largest missing element of @xmath111 is at least the @xmath110 largest missing element of @xmath112 for all @xmath113 .
that is , @xmath80 if and only if @xmath114 .
hence @xmath107 is a poset isomorphism .
the binomial coefficients and the multi - dimensional catalan numbers give the number of vertices and the number of facets , respectively , for boij - sderberg posets with upper and lower sequences given by consecutive integers .
[ lem : vertex - facet ] let @xmath100 and @xmath101 for some positive integer @xmath115 .
then @xmath52 has @xmath116 vertices and @xmath117 facets .
every vertex of @xmath52 is a sequence of @xmath118 numbers which can be seen as a @xmath118 subset of @xmath119 .
similarly , every such @xmath118 subset can be seen as a strictly increasing sequence of @xmath118 numbers .
hence there are @xmath116 vertices .
consider the vertices of @xmath52 with @xmath50 subtracted from them , then we are considering weakly increasing sequences of length @xmath118 with entries from @xmath120 . thus under the aforementioned consideration of the vertices ,
the maximal chains in @xmath52 then correspond directly to the @xmath121 walk diagrams of @xmath122-steps as described in ( * ? ? ?
* section iv ) .
moreover , the number of such walk diagrams is given in ( * ? ? ? * equation ( 4.8 ) ) as @xmath123 .
as every boij - sderberg poset contains and is contained in such a poset , we can give a bound on the number of vertices and number of facets of an arbitrary boij - sderberg poset .
let @xmath64 be strictly increasing sequences with @xmath48 for @xmath49 .
if we set @xmath8 to be the number of vertices and @xmath124 to be the number of facets of @xmath52 , then @xmath125 and @xmath126 we further make the observation that vertex - decomposability of @xmath57 implies the vertex - decomposability of any @xmath127 where @xmath128 .
[ rem : vd - transfer ] let @xmath78 .
then @xmath129 where @xmath130 and @xmath131 , hence @xmath132 and @xmath133 . for @xmath128
, we then have that @xmath127 is obtained from @xmath57 by linking , taking half of a join , and coning .
thus , properties like vertex - decomposability and cohen - macaulayness , which respect linking , joining , and coning , are preserved .
boij - sderberg posets admit recursive atom orderings which are simple to describe . [
thm : atom - ordering ] let @xmath64 be strictly increasing sequences with @xmath48 for @xmath49 . then @xmath52 admits a recursive atom ordering when the atoms are ordered lexicographically from smallest to largest . if @xmath50 and @xmath51 differ in only one position , then @xmath135 and clearly @xmath52 admits a recursive atom ordering .
suppose then that @xmath50 and @xmath51 differ in more than one position , i.e. , @xmath136 .
define @xmath137 by @xmath138 if @xmath139 and @xmath140 .
the atoms of @xmath52 are exactly the elements @xmath141 where @xmath142 or @xmath143 ; let @xmath144 be the indices of the atoms . then under the lexicographic ordering , @xmath145 .
furthermore , by induction on the number of positions where @xmath50 and @xmath51 differ , we may assume that @xmath146 admits a recursive atom ordering when the atoms are ordered lexicographically from smallest to largest . for @xmath147 ,
the atom @xmath148 of @xmath149 covers @xmath150 for @xmath151 and these are the only atoms of @xmath149 which cover some @xmath150 for @xmath151 .
moreover , for @xmath152 not equal to some @xmath153 for @xmath151 , then @xmath154 so @xmath155 is not a member of @xmath149 .
hence , the atoms @xmath148 , for @xmath151 , of @xmath149 are those which come first in the ordering of @xmath149 .
as the poset admits a recursive atom ordering , then the associated order complex is vertex - decomposable , hence squarefree glicci ( so , in particular , in the * g*orenstein * li*aison * c*lass of a * c*omplete * i*ntersection , see @xcite ) and cohen - macaulay .
let @xmath64 be strictly increasing sequences with @xmath48 for @xmath49 .
then by theorem [ thm : atom - ordering ] , @xmath52 admits a recursive atom ordering .
thus by ( * ? ? ?
* theorem 3.2 ) and ( * ? ? ?
* theorem 11.6 ) , the order complex of @xmath52 is vertex - decomposable . removing either @xmath170 or @xmath171 creates a situation where a single change is followed by a triplet of changes or the opposite either case is impossible in our family .
last , removing the other four vertices creates a situation which is , after tedious calculation , demonstrable impossible .
in particular , the form implies @xmath172 but also that @xmath21 would have exactly one consecutive pair of entries with nonconsecutive values .
checking the nine possible situations yields a contradiction in each case .
the author would like to thank his advisor , uwe nagel , for reading drafts of this article and making comments thereover .
the author would also like to thank heide gluesing - luerssen for assistance in the aesthetics of the figures . | boij and sderberg made a pair of conjectures , which were subsequently proven by eisenbud and schreyer and then extended by boij and sderberg , about the structure of betti diagrams of graded modules .
in the theory , a particular family of posets , and their associated order complexes , play an integral role .
we explore the structure of this family .
in particular , we show the posets are bounded complete lattices and the order complexes are vertex - decomposable , hence cohen - macaulay and squarefree glicci .
boij and sderberg recently conjectured in @xcite a complete characterisation , up to multiplication by a positive rational , of the structure of betti diagrams of finitely generated graded modules .
their conjectures were proven for the cohen - macaulay case in @xcite .
these were further extended to arbitrary graded modules in @xcite .
this characterisation centered around constructing the convex hull of a particular class of betti diagrams and showing it is equal to a geometric realisation of a particular simplicial complex the order complex of a poset .
the family of posets described in ( * ? ? ?
* definition 2.3 ) , which we refer to as the _ boij - sderberg posets _ , is the focus of this paper . in particular , we study the structure of the posets and their associated order complexes . while no immediate applications are presented here , we hope that our results will help shed light on the mysterious relation between the decomposition of betti tables and the associated modules . in section [ sec : preliminaries ] we recall the relevant combinatorial definitions and explicitly define the boij - sderberg posets ( definition [ def : boij - soederberg - posets ] ) .
following this , in section [ sec : basic - structure ] we determine some basic structural results for the posets . in particular , we show that the boij - sderberg posets are indeed bounded complete lattices ( proposition [ pro : lattice ] ) . and
in section [ sec : recursive - structure ] we discuss the recursive structure of the posets by finding a recursive atom ordering ( see @xcite ) for each of the posets ( theorem [ thm : atom - ordering ] ) .
this allows us to conclude that the order complexes are vertex - decomposable , cohen - macaulay , and squarefree glicci ( corollary [ cor : vd ] ) . |
the interest in minimal energy problems in the presence of an external field , which goes back to the pioneering works by gauss @xcite and frostman @xcite , was initially motivated by their direct relations with the dirichlet and balayage problems . a new impulse to this part of potential theory ( which is often referred to as the gauss variational problem ) appeared in the 1980 s when gonchar and rakhmanov @xcite , mhaskar and saff @xcite efficiently applied logarithmic potentials with external fields in the investigation of orthogonal polynomials and rational approximations to analytic functions .
e.g. , the vector setting of the problem , earlier suggested by ohtsuka ( * ? ? ?
* section 2.9 ) , has nowadays become particularly interesting in connection with hermite pad rational approximations ( see @xcite ) . however , the potential - theoretical methods applied in these studies were mainly based on the vague ( = weak@xmath8 ) topology , which made it possible to establish the existence of a solution only for vector measures of finite dimensions and compact support , though with a rather general matrix of interaction between their components .
see , e.g. , ( * ? ? ?
* theorem 2.30 ) . in order to treat the gauss variational problem for vector measures @xmath9 of infinite dimensions and/or noncompact support , associated with a condenser , in @xcite
we have suggested an approach based on introducing a metric structure on the class of all @xmath10 with finite energy , which agrees properly with the vague topology , and also on establishing an infinite dimensional version of a completeness theorem ( see below for details ) .
this enabled us to obtain sufficient conditions for the solvability of the problem ( see ( * ? ? ? *
theorem 8.1 ) ) ; moreover , we have shown these sufficient conditions to be sharp by providing examples of the nonsolvability ( see ( * ? ? ?
* examples 8.1 , 8.2 ) ; see below for some details ) .
it is worth emphasizing that , as is seen from these examples , such a phenomenon of the nonsolvability occurs even for the coulomb kernel @xmath11 in @xmath12 and a standard condenser of two oppositely signed plates , one of them being noncompact , which at first glance looks to be rather surprising because of an electrostatic interpretation of the problem .
the purpose of this study is to establish conditions that are simultaneously necessary and sufficient for the solvability of the problem in the infinite dimensional vector setting . to formulate the problem and to outline some of the results obtained , we start by introducing relevant notions . in all that follows
, @xmath13 denotes a locally compact hausdorff space and @xmath14 the linear space of all real - valued scalar radon measures @xmath15 on @xmath13 equipped with the _ vague _ topology , i.e. the topology of pointwise convergence on the class @xmath16 of all continuous functions @xmath17 on @xmath13 with compact support . a _ kernel _
@xmath18 on @xmath13 is meant to be an element from @xmath19 , where @xmath20 consists of all lower semicontinuous functions @xmath21 $ ] such that @xmath22 unless @xmath23 is compact .
given @xmath24 , the _ mutual energy _ and the _ potential _ relative to the kernel @xmath18 are defined by @xmath25 respectively .
( when introducing a notation , we always tacitly assume the corresponding object on the right to be well defined as a finite number or @xmath26 . ) for @xmath27 , @xmath28 defines the _ energy _ of @xmath15 .
let @xmath29 consist of all @xmath30 with @xmath31 .
we shall mainly be concerned with a _ positive definite _ kernel @xmath18 , which by @xcite means that it is symmetric ( i.e. , @xmath32 for all @xmath33 ) and @xmath34 , @xmath30 , is nonnegative whenever defined .
then @xmath35 forms a pre - hilbert space with the scalar product @xmath28 and the seminorm @xmath36 .
the topology on @xmath35 defined by this seminorm is called _
strong_. a positive definite kernel @xmath18 is _ strictly positive definite _ if the seminorm @xmath37 is a norm .
recall that a measure @xmath38 is _ concentrated _ on a set @xmath39 if @xmath40 is locally @xmath15-negligible ; or equivalently , if @xmath41 is @xmath15-measurable and @xmath42 , where @xmath43 is the trace of @xmath15 on @xmath41 ( see , e.g. , @xcite ) .
let @xmath44 be the convex cone of all nonnegative measures concentrated on @xmath41 , and let @xmath45 .
we also write @xmath46 and @xmath47 . the _ interior capacity _ of a set @xmath41 relative to the kernel @xmath18 is given by the formula . ]
@xmath48 where the infimum is taken over all @xmath49 with @xmath50 .
we consider a countable , locally finite collection is a singleton , then we preserve the normal fonts instead of the bold ones . ]
@xmath51 of closed sets , called plates , @xmath52 with the sign @xmath53 or @xmath54 prescribed such that the oppositely signed sets are mutually disjoint .
let @xmath55 stand for the cartesian product @xmath56 ; then @xmath57 is a ( nonnegative ) _ vector measure _ @xmath58 with the components @xmath59 . the topology of the product space @xmath56 , where every @xmath60 is endowed with the vague topology , is likewise called _
vague_. if @xmath61 and a vector - valued function @xmath62 with the @xmath63-measurable components @xmath64 $ ] are given , then we write @xmath65 .
is meant to be well defined provided that so is every summand @xmath66 and the sum does not depend on the order of summation though might be @xmath26 .
then , by the riemann series theorem , the sum is finite if and only if the series converges absolutely . ] in accordance with an electrostatic interpretation of a condenser , we assume that the interaction between the charges lying on the conductors @xmath4 , @xmath5 , is characterized by the matrix @xmath67 , where @xmath68 . given @xmath69
, we define the _ mutual energy _ @xmath70 and the _ vector potential _
@xmath71 , @xmath72 , as a vector - valued function with the components @xmath73 for @xmath74 , the mutual energy @xmath75 defines the _ energy _ of @xmath10 .
let @xmath76 consist of all @xmath57 with @xmath77
. fix a vector - valued function @xmath78 , where each @xmath79 $ ] is treated as an external field acting on the charges on @xmath4 . then the @xmath80-_weighted vector potential _ and the @xmath80-_weighted energy _ of @xmath81 are defined by @xmath82 respectively . throughout this paper
, we assume that for all @xmath5 ( which is , in particular , the case if each @xmath83 is the potential of a _ dirac measure _ ) .
however , for a finer analysis to be carried out in the present study in order to establish necessary and sufficient conditions of the solvability , we need to assume the external field to be generated in the sense of ( [ f]) by a charge of _ finite _ energy . ]
@xmath84 where a ( signed ) measure @xmath85 is given , and we also write @xmath86 . also fix a numerical vector @xmath87 with @xmath88 for all @xmath5 such that @xmath89 and a continuous function @xmath90 on @xmath91 with @xmath92 we are interested in the problem of minimizing @xmath93 over the class of all @xmath94 normalized by @xmath95 for all @xmath5 , referred to as the _ gauss variational problem_. the main question is whether minimizers @xmath96 in the problem exist . if the condenser @xmath97 is finite and compact , the kernel @xmath18 is continuous on @xmath98 whenever @xmath99 , and @xmath100 for all @xmath5 , then the existence of those @xmath96 can easily be established by exploiting the vague topology only , since then the class of admissible vector measures is vaguely compact , while @xmath101 is vaguely lower semicontinuous .
see ( * ? ? ?
* theorem 2.30 ) ( cf .
also @xcite for the logarithmic kernel in the plane ) .
however , these arguments break down if any of the above - mentioned four assumptions is dropped , and then the problem on the existence of minimizers becomes rather nontrivial .
in particular , the class of admissible vector measures is no longer vaguely compact if any of the @xmath4 is noncompact .
another difficulty is that , under assumption ( [ f ] ) , the @xmath80-weighted energy @xmath101 might not be vaguely lower semicontinuous . for a positive definite kernel satisfying fuglede s condition of consistency between the strong and vague topology on @xmath102 ( see definition [ def : const ] below ) , these difficulties have been overcome in @xcite in the frame of an approach , @xmath103 , and @xmath104 ) and also the author s studies @xcite@xcite ( where @xmath105 is finite ) .
] based on the following crucial arguments .
the set @xmath76 has been shown to be a semimetric space with the semimetric @xmath106^{1/2},\ ] ] and one can define an inclusion @xmath107 of @xmath108 into @xmath35 such that @xmath108 becomes isometric to its @xmath107-image , the latter being regarded as a semimetric subspace of the pre - hilbert space @xmath35 . see ( * ? ? ?
* theorem 3.1 ) ; cf .
also theorem [ lemma : semimetric ] below .
we therefore call the topology of the semimetric space @xmath76 _ strong_. another crucial fact is that if , in addition , the kernel @xmath18 is bounded from above on the product of the oppositely signed plates , then the topological subspace of @xmath76 consisting of all @xmath10 with @xmath109 ( where @xmath110 is given ) is strongly complete . moreover , the strong topology on this subspace is in agreement with the vague one in the sense that any strong cauchy net converges strongly to each of its vague cluster points .
see ( * ? ? ?
* theorem 9.1 ) ; cf .
also theorem [ th : complete ] below .
all these enabled us to prove in @xcite that if , moreover , @xmath90 is bounded from above , then the gauss variational problem is solvable provided that each @xmath4 either is compact or has finite interior capacity .
see ( * ? ? ?
* theorem 8.1 ) ; cf .
also theorem [ th : suff ] below .
however , if some of the plates , say @xmath1 for @xmath2 , are noncompact and have infinite interior capacity then , in general , there exists a vector @xmath111 such that the problem admits _ no _ solution
. see examples 8.1 and 8.2 in @xcite , illustrating such a phenomenon of the nonsolvability ( cf .
also example [ ex : sharp ] below ; see also @xcite for some related numerical experiments ) .
then , for given @xmath18 , @xmath112 , @xmath90 , and @xmath113 , what is a description of the set @xmath114 of all vectors @xmath111 for which the gauss variational problem is nevertheless solvable ? in this paper ,
such a characterization is established ; it is given in terms of a solution to an auxiliary extremal problem intimately related to the operator of orthogonal projection onto the cone of all positive scalar measures supported by @xmath115 .
in the case where @xmath116 is a singleton , the description obtained is _ complete_. to give a hint how it looks like , we start with the following example , where @xmath117 while @xmath18 is the riesz kernel @xmath118 of order @xmath119 $ ] , @xmath120 , in @xmath121 , @xmath122 . for this kernel ,
the operator of orthogonal projection in @xmath123 onto @xmath124 is , in fact , the operator of riesz balayage @xmath125 onto @xmath1 , related to the notion of @xmath126-green function @xmath127 of @xmath128 by the formula @xmath129 @xmath130 being the unit dirac measure at @xmath131 ( see , e.g. , ( * ? ? ?
* chapter 4 , section 5 ) ) .
[ ex:11 ] under these hypotheses , let the euclidean distance between the oppositely signed plates be nonzero , @xmath132 for all @xmath133 , @xmath134 for all @xmath5 , and let @xmath135 be compactly supported in @xmath136 .
[ pr : ex:11 ] then the set @xmath137 consists of all vectors @xmath138 such that is a scalar radon measure due to the local finiteness of @xmath112 , so that formula ( [ bal ] ) makes sense . ]
@xmath139 where @xmath140 is a solution ( it exists ) to the problem of minimizing the @xmath126-green energy @xmath141 over all @xmath142 with the properties that @xmath63 , @xmath133 , is supported by @xmath4 and @xmath143 .
the rest of the paper is organized as follows . in sections
[ sec : cond ] and [ sec : gausspr ] we summarize without proof some results from @xcite , necessary for the understanding of the subject matter .
a description of the set @xmath114 is provided by theorems [ th : solvable1 ] and [ th : solvable2 ] , the main results of the study ; it is given in terms of a solution @xmath144 to an auxiliary extremal problem ( see section [ sec : gausspr ] for its formulation ) , while the existence of this @xmath144 is guaranteed by theorem [ lemma : exist ] .
the proofs of theorems [ lemma : exist ] , [ th : solvable1 ] and [ th : solvable2 ] are provided in sections [ proof : lemma : exist ] , [ proof : th : solvable1 ] and [ proof : th : solvable2 ] , respectively ; see also sections [ sec : prel ] , [ sec : j ] , [ sec : extr ] and [ sec : descr ] for some crucial auxiliary results .
finally , section [ sec : ex ] contains some further examples with the riesz kernels , illustrating the results obtained .
we write @xmath145 or @xmath146 for the support of @xmath30 .
a measure @xmath15 is called _ finite _ if @xmath146 is compact , and _ bounded _ if @xmath147 . here
@xmath148 , where @xmath149 and @xmath150 are respectively the positive and negative parts in the hahn jordan decomposition of @xmath15 . in this section
we assume the kernel @xmath18 to be positive definite
. then @xmath35 forms a pre - hilbert space with the scalar product @xmath28 and the seminorm @xmath151 , while the potential @xmath152 of any @xmath153 is well defined and finite _ nearly everywhere _
( n.e . ) in @xmath13 , i.e. except for a subset with interior capacity zero ; see @xcite . in minimal energy problems , the following lemma from @xcite is often useful .
[ lemma : convex]consider a convex set @xmath154 and assume there exists @xmath155 such that @xmath156 then @xmath157 in addition to the _ strong _ topology on @xmath35 , determined by the seminorm @xmath158 , sometimes we shall consider the _ weak _ topology on @xmath35 , defined by means of the seminorms @xmath159 , @xmath160 ( see , e.g. , @xcite ) .
the cauchy
schwarz inequality @xmath161 implies immediately that the strong topology on @xmath35 is finer than the weak one . in @xcite
, fuglede introduced the following two _ equivalent _ properties of consistency between the induced strong , weak , and vague topologies on @xmath102 : * _ every strong cauchy net in @xmath102 converges strongly to any of its vague cluster points ; _ * _ every strongly bounded and vaguely convergent net in @xmath102 converges weakly to the vague limit . _
[ def : const ] following fuglede @xcite , we call a kernel @xmath18 _ consistent _ if it satisfies either of the properties ( c@xmath162 ) and ( c@xmath163 ) , and _ perfect _ if , in addition , it is strictly positive definite .
[ metriz ] one has to consider _ nets _ or _ filters _ in @xmath164 instead of sequences , since the vague topology in general does not satisfy the first axiom of countability .
we follow moore s and smith s theory of convergence , based on the concept of nets ( see @xcite ; cf .
also ( * ? ? ?
* chapter 0 ) and ( * ? ? ?
* chapter 2 ) ) .
however , if @xmath13 is metrizable and can be written as a countable union of compact sets , then @xmath165 satisfies the first axiom of countability ( see ( * ? ? ?
* lemma 1.2.1 ) ) and the use of nets may be avoided .
( fuglede @xcite)[th:1 ] a kernel @xmath18 is perfect if and only if @xmath102 is strongly complete and the strong topology on @xmath102 is finer than the vague one . in @xmath121 , @xmath166 ,
the newtonian kernel @xmath167 is perfect @xcite .
so are the riesz kernels @xmath168 , @xmath169 , in @xmath121 , @xmath122 @xcite , and the restriction of the logarithmic kernel @xmath170 in @xmath171 to the open unit disk @xcite . furthermore
, if @xmath172 is an open set in @xmath121 , @xmath173 , and its generalized green function @xmath174 exists ( see , e.g. , ( * ? ? ?
* theorem 5.24 ) ) , then the kernel @xmath174 is perfect as well @xcite . it is seen from definition [ def : const ] and theorem [ th:1 ] that the concept of consistent or perfect kernels is an efficient tool in minimal energy problems over classes of _ nonnegative scalar _ radon measures with finite energy .
indeed , fuglede s theory of capacities of _ sets _ has been developed in @xcite for exactly those kernels .
the following fundamental result of this theory will often be used below . to the interior capacities of condensers with finitely many plates . ] [ thf:1 ] assume that the kernel @xmath18 is consistent .
then for any set @xmath39 with @xmath175 there exists a measure @xmath176 with the properties @xmath177 this @xmath178 , called an interior equilibrium measure associated with @xmath41 , is a solution to the problem of minimizing @xmath34 over the class @xmath179 of all @xmath153 such that @xmath180 n.e . in @xmath41 , and it is determined uniquely up to a summand with seminorm zero . in @xcite we have shown that the concept of consistent or perfect kernels is efficient , as well , in minimal energy problems over classes of _ vector measures _ of finite or infinite dimensions , associated with a condenser @xmath112 .
this is guaranteed by a theorem on the completeness of certain topological subspaces of the semimetric space @xmath76 ( see ( * ? ? ?
* theorem 13.1 ) and ( * ? ? ? * theorem 9.1 ) , cf .
also theorem [ th : complete ] below ; compare with theorem [ th:1 ] ) .
in the following lemmas [ lemma2.3.1][lemma : vague.cont ] , the kernel is arbitrary ( not necessarily positive definite ) .
( fuglede @xcite)[lemma2.3.1 ] for any given @xmath39 , it holds that @xmath181 .
[ a.e.]if @xmath182 is bounded , then any proposition holds @xmath15-almost everywhere ( @xmath15-a.e . ) in @xmath13 , provided that it holds n.e . in @xmath41 . indeed ,
then @xmath41 is @xmath15-integrable and hence , by ( * ? ? ?
* proposition 4.14.1 ) , any locally @xmath15-negligible subset of @xmath41 is @xmath15-negligible . since , according to lemma [ lemma2.3.1 ] , a set of interior capacity zero is locally @xmath183-negligible for any @xmath184 , the lemma follows .
we shall need the following lemmas on continuity , the first being well known ( see , e.g. , @xcite ) .
[ lemma : lower ] for any @xmath185 the map @xmath186 is vaguely lower semicontinuous on @xmath164 . in particular
, this implies that the potential @xmath152 of any @xmath187 belongs to @xmath188 .
[ lemma : vague.cont2 ] assume that @xmath187 is bounded , @xmath189 is continuous for @xmath190 and @xmath191 then the potential @xmath152 is continuous at every @xmath192 .
having fixed a point @xmath192 and its compact neighborhood @xmath193 so that @xmath194 , we consider the function @xmath195 on @xmath196 defined by @xmath197 under the hypotheses of the lemma , @xmath198 is nonnegative and continuous ; hence , @xmath199 is lower semicontinuous as the potential of @xmath187 relative to the kernel @xmath198 . on the other hand , integrating ( [ kappastar ] ) with respect to the ( bounded ) measure @xmath15 , we conclude from assumption ( [ kern.bound ] ) that @xmath200 , @xmath201 , coincides up to a finite summand with the restriction of @xmath202 to @xmath193 . what has been shown just above
therefore implies that @xmath203 is lower semicontinuous at @xmath204 , and the lemma follows .
[ inf]a kernel @xmath18 is said to _ possess the property _
@xmath205 if @xmath206 as @xmath207 uniformly on compact sets ; then for every @xmath208 and every compact @xmath209 there exists a compact set @xmath210 such that @xmath211 for all @xmath212 and @xmath213 . for any @xmath214 , write @xmath215 .
[ lemma : vague.cont ] fix a closed set @xmath216 , a closed subset @xmath217 of @xmath218 , and @xmath214 .
if a kernel @xmath189 is continuous for @xmath190 and possesses the property @xmath205 , then the mapping @xmath219 is continuous in the product topology , where @xmath217 and @xmath220 are considered to be topological subspaces of @xmath13 and @xmath221 , respectively .
fix @xmath222 and @xmath223 , @xmath224 , such that @xmath225 ( as @xmath226 ranges along @xmath227 ) in the topology of the product space @xmath228 .
we need to show that @xmath229 due to the property @xmath205 , for any @xmath208 one can choose a compact neighborhood @xmath230 of the point @xmath204 and a compact neighborhood @xmath231 of the set @xmath230 so that @xmath232 and @xmath233 certainly , there is no loss of generality in assuming @xmath234 .
let @xmath235 and @xmath236 denote respectively the complement and the boundary of @xmath41 relative to @xmath237 , where @xmath237 is treated as a topological subspace of @xmath13 .
having observed that @xmath238 is continuous , we proceed by constructing a function @xmath239 such that @xmath240 @xmath241 to this end , consider a compact neighborhood @xmath242 of @xmath243 in @xmath237 and write @xmath244 note that @xmath245 is a compact subset of the hausdorff and compact ( hence , normal ) space @xmath246 , while the function @xmath90 is continuous on @xmath247 . by using the tietze
urysohn extension theorem ( see , e.g. , ( * ? ? ?
* chapter 0 ) ) , we deduce from relation ( [ 115 ] ) that there exists a continuous function @xmath248 $ ] such that @xmath249 .
hence , the function in question can be defined by means of the formula @xmath250 furthermore , since such a function @xmath17 is continuous on @xmath251 and has compact support , one can choose a compact neighborhood @xmath252 of @xmath204 in @xmath230 so that @xmath253 therefore , for any @xmath254 and @xmath255 we get , due to relations , @xmath256 @xmath257 @xmath258 @xmath259\,d\nu(y)\bigr|\leqslant \varepsilon.\label{124}\ ] ] choose @xmath260 so that for all @xmath224 that follow @xmath261 there hold @xmath262 and @xmath263 combined with relations ( [ 121])([124 ] ) , this shows that for all @xmath264 , @xmath265\,d\nu_s(y)\bigr|+\bigl|\int \varphi(x_0,y)\,d(\nu_s-\nu_0)(y)\bigr|+4\varepsilon\\ { } & \leqslant\varepsilon+\varepsilon+4\varepsilon=6\varepsilon , \end{split}\ ] ] which in view of the arbitrary choice of @xmath266 establishes ( [ cont1 ] ) . throughout this section
we require the kernel @xmath18 to be perfect ( see definition [ def : const ] ) .
fix @xmath153 and a closed set @xmath267 with @xmath268 , and let @xmath269 be the operator of _ orthogonal projection _ in the pre - hilbert space @xmath35 onto the convex cone @xmath270 ( see ( * ? ? ?
* section 1.12.3 ) ; note that @xmath271 due to lemma [ lemma2.3.1 ] ) .
then @xmath272 is a measure in @xmath270 such that @xmath273 observe that @xmath270 , treated as a metric subspace of @xmath35 , is complete in consequence of theorem [ th:1 ] ; hence , according to ( * ? ? ?
* theorem 1.12.3 ) , @xmath272 _ exists and is determined uniquely_. [ lemma : hryu ] if @xmath247 ranges through the increasing filtering family @xmath274 of all compact subsets of @xmath237 , then @xmath275 for each @xmath276 one can certainly assume @xmath277 , which causes no loss of generality because of @xmath278 ( see @xcite ) ; hence , the projection @xmath279 exists ( and is unique ) .
we next observe that @xmath280 decreases as @xmath281 and @xmath282 on the other hand , applying ( * ? ? ?
* lemma 1.2.2 ) to the measure @xmath283 and the function @xmath18 , as well as to @xmath272 and each of @xmath284 and @xmath285 , we obtain @xmath286 therefore @xmath287 when combined with relation ( [ preg ] ) , this establishes the equality @xmath288 fix @xmath289 such that @xmath290 .
applying lemma [ lemma : convex ] to the ( convex ) set @xmath291 in view of @xmath292 we get @xmath293 as @xmath280 , @xmath276 , is fundamental in @xmath294 because of ( [ preg1 ] ) , the last relation shows that @xmath295 is a strong cauchy net in @xmath296 .
since @xmath270 is strongly complete , this net converges to some @xmath297 strongly and , hence , weakly .
repeated application of ( [ preg1 ] ) then yields @xmath298 which due to the uniqueness statement gives @xmath299 , and the lemma follows . as an immediate consequence of lemmas [ lemma : lower ] and [ lemma : hryu ] , we get @xmath300 [ lemma : proj.desc ] it holds that @xmath301 if , moreover , @xmath189 is continuous for @xmath190 , @xmath149 is bounded , @xmath302 and @xmath303 then @xmath304 and therefore @xmath305 according to ( * ? ? ?
* proposition 1.12.4 ) , @xmath272 is uniquely characterized by the relations @xmath306 we observe that assertion ( [ orth1 ] ) can be obtained from inequality ( [ orth4 ] ) with the help of arguments similar to those in ( * ? ? ?
* proof of theorem 4.16 ) . indeed , for each @xmath307 the set @xmath308 is @xmath309-measurable and , hence , one can consider @xmath310 , the trace of @xmath309 on @xmath41 . since @xmath310 is an element of @xmath270 as well , relation ( [ orth4 ] )
gives @xmath311 which , however , is possible only provided that @xmath41 is locally @xmath309-negligible . in view of the arbitrary choice of @xmath307 , this together with lemma [ lemma2.3.1 ] yields @xmath312 as desired .
let now all the hypotheses of the second part of the lemma be satisfied .
it follows from inequality ( [ proj : pot : lower ] ) that , while proving assertion ( [ orth2 ] ) , one can assume @xmath237 to be compact .
then , by lemma [ a.e .
] , relation ( [ orth1 ] ) holds @xmath272-a.e . in @xmath13
this implies @xmath313 for if not , we would arrive at a contradiction with ( [ orth5 ] ) when integrating inequality ( [ orth1 ] ) with respect to @xmath272 . in turn
, the last relation yields that for every @xmath314 one can choose a net @xmath315 with the properties that @xmath316 and @xmath317 hence , assertion ( [ orth2 ] ) will be proved once we show that @xmath318 is upper semicontinuous on @xmath237 .
as @xmath319 is lower semicontinuous on @xmath13 , it is enough to establish the continuity of @xmath320 on @xmath237 , but this is a direct consequence of lemma [ lemma : vague.cont2 ] .
( see , e.g. , @xcite)[def : h ] a kernel @xmath18 satisfies the _ generalized maximum principle _ with a constant @xmath321 if for every finite @xmath187 with the property @xmath322 one has @xmath323 for all @xmath72 . [ lemma : bound]suppose that the kernel @xmath18 is @xmath324 and satisfies the generalized maximum principle with a constant @xmath321 .. ] if , moreover , all the assumptions of lemma [ lemma : proj.desc ] hold true , then @xmath325 it follows from inequality ( [ proj : masses : lower ] ) that , while proving the lemma , one can assume @xmath237 to be compact ; then @xmath326 ( @xmath18 being strictly positive definite ) .
hence , according to theorem [ thf:1 ] , an equilibrium measure @xmath327 of @xmath328 exists and satisfies the relations @xmath329 since then , due to the boundedness of both @xmath272 and @xmath330 , relations ( [ orth3 ] ) and ( [ equ:1 ] ) hold true @xmath330-a.e . and @xmath272-a.e .
, respectively , on account of @xmath331 we get @xmath332 on the other hand , inequality ( [ equ:2 ] ) yields , by definition [ def : h ] , @xmath333 when substituted into the preceding relation , this yields the lemma .
let @xmath334 and @xmath335 be fixed countable , disjoint sets of indices @xmath336 , where the latter is allowed to be empty , and let @xmath337 .
assume that to every @xmath5 there corresponds a nonempty , closed set @xmath52 .
a collection @xmath51 is called an @xmath338-_condenser _ ( or simply a _ condenser _ ) in @xmath13 if every compact subset of @xmath13 intersects with at most finitely many @xmath4 and @xmath339 we call @xmath97 _ compact _ if so are all @xmath4 , @xmath340 , and _ finite _ if @xmath105 is finite . the sets @xmath4 , @xmath341 , and @xmath342 , @xmath343 , are called the _ positive _ and _ negative plates _ , respectively .
( note that any two equally signed plates can intersect each other or even coincide . ) for any @xmath344 , write @xmath345 then @xmath346 is closed ( @xmath97 being locally finite ) , and it is compact if @xmath4 , @xmath347 , are compact while @xmath348 is finite . given @xmath97 , let @xmath55 consist of all ( nonnegative ) _ vector measures _
@xmath9 , where @xmath349 for all @xmath5 .
the product topology on @xmath350 , where every @xmath351 is equipped with the vague topology , is likewise called _
vague_. in accordance with an electrostatic interpretation of a condenser , we assume that the law of interaction between the charges lying on the plates @xmath4 , @xmath5 , is determined by the matrix @xmath67 , where @xmath352 equals @xmath53 if @xmath341 and @xmath54 otherwise .
given @xmath353 , we then define the _ mutual energy _ @xmath75 and the _ vector potential _ @xmath71 , @xmath72 , by relations ( [ vectoren ] ) and ( [ vectorpot ] ) , respectively . for @xmath74 ,
the mutual energy @xmath75 defines the _ energy _ of @xmath10 .
let @xmath76 consist of all @xmath57 with @xmath77 . from now on we shall always require the kernel @xmath18 to be positive definite . then , according to ( * ? ? ? * lemma 3.3 ) , for @xmath57 to have finite energy , it is sufficient that @xmath354 . in this paper
we are concerned with minimal energy problems over subsets of @xmath76 , to be treated in the frame of the approach developed in @xcite . for the convenience of the reader , in this and the next sections
we summarize without proof some results from @xcite , necessary for the understanding of the subject matter . due to the local finiteness of @xmath112 , one can define the mapping @xmath355 , @xmath356 such a mapping is , in general , non - injective ; it is injective if and only if @xmath4 , @xmath340 , are mutually disjoint . also observe that @xmath357 is a _ signed _ ( scalar ) measure ; because of assumption ( [ non ] ) , the positive and negative parts in the hahn
jordan decomposition of @xmath357 are given by @xmath358 [ lemma3.5]for any @xmath185 and @xmath57 , the value @xmath359 is finite if and only if @xmath360 , and then @xmath361 [ lemma3.3 ] @xmath57 has finite energy if and only if so does its scalar @xmath107-image , @xmath357 .
convex cone_. ] [ relation]for any @xmath362 , it holds that @xmath363 furthermore , @xmath364 is well defined and finite n.e .
, @xmath5 , subsists n.e . in @xmath13 if this is the case for each @xmath365 . ] in @xmath13 and has the components @xmath366 [ lemma : semimetric ] @xmath76 forms a semimetric space with the semimetric @xmath367 , defined by ( [ vseminorm ] ) , and such a space is isometric to its @xmath107-image , @xmath368 . that is , @xmath369 the semimetric @xmath367 is a metric if and only if the kernel @xmath18 is strictly positive definite while all @xmath4 , @xmath5 , are mutually disjoint .
[ cor : ea]given @xmath112 , define @xmath370 as the set of all @xmath15 such that @xmath371 and @xmath372 and equip it with the semimetric structure induced from @xmath35 .
then @xmath373 so that , by theorem [ lemma : semimetric ] , the semimetric spaces @xmath370 and @xmath374 are isometric . indeed , @xmath375 can be obtained directly from relation ( [ hj ] ) and corollary [ lemma3.3 ] . to prove the converse inclusion , fix @xmath376 and define @xmath377 and @xmath378\biggl|_{a_{i+1 } } \quad\text{for all \ } i=1,2,\ldots.\
] ] then @xmath349 and @xmath379 , where @xmath380 .
since @xmath34 is finite , repeated application of corollary [ lemma3.3 ] shows that @xmath381 , and identity ( [ ea ] ) follows .
because of ( [ twonorms ] ) , the topology on @xmath108 defined by the semimetric @xmath382 is called _ strong_. two elements of @xmath76 , @xmath383 and @xmath384 , are _ equivalent _ if @xmath385 .
the equivalence in @xmath108 implies @xmath386 provided that the kernel @xmath18 is strictly positive definite , and it implies @xmath387 if , moreover , all the @xmath4 are mutually disjoint .
[ lemma : potequiv]if @xmath383 and @xmath384 are equivalent in @xmath76 , then @xmath388 n.e . in @xmath13 .
for any @xmath214 , let @xmath389 consist of all @xmath61 with @xmath390 this class is vaguely bounded and closed ; hence , by ( * ? ? ?
* lemma 3.1 ) , it is vaguely compact .
the following theorem on the strong completeness of @xmath391 , treated as a topological subspace of @xmath76 , is crucial in our approach ( compare with theorem [ th:1 ] ) .
is isometric to its @xmath107-image , theorem [ th : complete ] has singled out a _ strongly complete _ topological subspace of the pre - hilbert space @xmath35 , whose elements are _ signed _ radon measures .
this is of independent interest since , according to a well - known counterexample by cartan @xcite , the whole space @xmath35 is strongly incomplete even for the newtonian kernel @xmath167 in @xmath121 , @xmath392 . ]
[ th : complete]assume that the kernel @xmath18 is consistent and bounded on @xmath393 . of order @xmath394 in @xmath121 ,
@xmath122 , theorem [ th : complete ] remains valid without assuming @xmath395 to be bounded on @xmath393 ; cf .
* theorem 1 ) and ( * ? ? ?
* remark 9.2 ) .
the proof is based on deny s theorem @xcite stating that the pre - hilbert space @xmath123 can be completed by making use of distributions of finite energy . ] then the following two assertions hold : * the semimetric space @xmath396 is complete . in more detail ,
if @xmath397 is a strong cauchy net and @xmath10 is one of its vague cluster points , then @xmath398 and @xmath399 strongly , i.e. @xmath400 * if the kernel @xmath18 is strictly positive definite ( hence , perfect ) , while all @xmath4 , @xmath340 , are mutually disjoint , then the strong topology on @xmath401 is finer than the vague one . in more detail ,
if @xmath402 converges strongly to @xmath403 , then actually @xmath404 and @xmath405 vaguely .
fix @xmath85 and define @xmath78 where @xmath406 for all @xmath5 ; such an @xmath407 is well defined and finite n.e . in @xmath13 , while all its components are _ universally measurable _ , i.e. @xmath15-measurable for every @xmath187 .
we treat each @xmath83 as an external field acting on the charges on @xmath4 and then we define the @xmath407-weighted vector potential @xmath408 and the @xmath80-_weighted energy _
@xmath409 of @xmath94 by relations ( [ def : wpot ] ) and ( [ def : wener ] ) , respectively . given @xmath94 , application of lemma [ lemma3.5 ] to each of @xmath410 and @xmath411
gives @xmath412 combined with ( [ twonorms ] ) , this implies that @xmath413 is finite and can be written in the form @xmath414 fix a vector @xmath87 with @xmath88 for all @xmath5 and a continuous function @xmath90 on @xmath415 satisfying conditions ( [ ag ] ) and ( [ ginf ] ) . also having fixed @xmath416 such that @xmath417 , we write @xmath418 and further we consider @xmath419 then , because of ( [ gmu ] ) , @xmath420 in the case @xmath421 , the symbol @xmath416 in these and other notations will often be dropped .
that is , we write @xmath422 , @xmath423 , and so on .
below we use the following terminology . given @xmath97 , @xmath424 , @xmath90 , @xmath113 , and @xmath416 , the @xmath425-_problem _ is that on the existence of @xmath426 with @xmath427 the @xmath425-problem is said to be _
solvable _ if the class @xmath428 of all its minimizers is nonempty .
the @xmath429-problem ( or shortly the @xmath430-problem ) , main in the present study , is referred to as _ the gauss variational problem_. a minimizer in the @xmath430-problem will often be denoted simply by @xmath96 ( with the tilde dropped ) . to make these problems well defined , we shall always suppose that @xmath431 then for every @xmath416 one has @xmath432 where the inequality on the right is an obvious consequence of assumption ( [ g : finite ] ) , while that on the left follows from ( [ ga ] ) due to the positive definiteness of the kernel .
in addition to all the assumptions stated above , in the rest of the paper we shall always require that the kernel @xmath18 is consistent and bounded on @xmath393 , i.e. @xmath433 while for every @xmath5 either @xmath434 is bounded or there exist
@xmath435 and @xmath436 such that @xmath437 then sufficient conditions for the solvability of the gauss variational problem are given by the following theorem ( see ( * ? ? ?
* theorem 8.1 ) ) .
[ th : suff]under the above - mentioned hypotheses , the @xmath430-problem is solvable if each @xmath4 , @xmath5 , either is compact or has finite interior capacity .
moreover , this theorem is sharp ; that is , if at least one of the plates is noncompact and has infinite interior capacity , then the @xmath430-problem , in general , admits _ no _ solution .
this can be illustrated by the following example ( cf .
* proposition 8.1 ) ) . in @xmath12 , and also @xcite for some related numerical experiments . ]
[ ex : sharp]in @xmath121 , @xmath122 , consider the riesz kernel @xmath118 of order @xmath126 , where @xmath119 $ ] and @xmath120 .
assume @xmath438 , @xmath439 , @xmath440 and @xmath441 are closed disjoint sets with @xmath442 , @xmath443 , and let @xmath440 be compact while @xmath444 connected .
if , moreover , @xmath445 is compactly supported in @xmath444 and @xmath446 , then the corresponding @xmath430-problem has _ no _ solution if and only if @xmath447 though @xmath441 is @xmath126-thin at @xmath448 . is _
@xmath126-thin at _
@xmath448 if @xmath449 , the inverse of @xmath237 relative to the unit sphere , is @xmath126-thin at @xmath450 , or equivalently ( * ? ? ?
* theorem 5.10 ) , if either @xmath237 is bounded or @xmath450 is an @xmath126-irregular point for @xmath449 .
see @xcite ; for @xmath451 , such a definition is due to m. brelot ( see @xcite ; cf .
also @xcite ) .
we refer to @xcite for an example of a set of infinite newtonian capacity , though @xmath452-thin at @xmath448 ( cf .
also example [ ex:13 ] in section [ sec : ex ] below ) . ]
[ pr]given @xmath18 , @xmath453 , @xmath90 , and @xmath113 , what is a description of the set @xmath114 of all vectors @xmath138 for which the @xmath430-problem is nevertheless solvable ?
in addition to the standing assumptions stated in section [ sec : gausspr ] , in the rest of the paper we assume that @xmath13 is noncompact ( hence , @xmath331 ) . this involves no loss of generality , since for a compact @xmath13 the gauss variational problem is always solvable due to theorem [ th : suff ] .
we also require that the kernel @xmath18 is strictly positive definite ( hence , perfect ) , continuous for @xmath190 , possesses the property @xmath205 and satisfies the generalized maximum principle with a constant @xmath454 ( see definitions [ inf ] and [ def : h ] ) .
furthermore , assume that the measure @xmath85 , generating the external field by means of ( [ f ] ) , is bounded and satisfies the assumptions @xmath455 [ rem : lower]then for each @xmath5 the @xmath456-component of the external field , @xmath457 , is lower semicontinuous on @xmath4 , which is seen from lemma [ lemma : vague.cont2 ] .
[ rem : fin]suppose for a moment that @xmath85 is compactly supported in @xmath136 .
it is seen from ( * ? ? ?
* lemma 6.3 ) that , under the above hypotheses on the kernel @xmath18 , assumption ( [ g : finite ] ) would hold automatically if one would require @xmath458 for all @xmath5 .
note that the standing assumptions on a kernel are not too restrictive .
in particular , in @xmath121 , @xmath392 , they all are satisfied by the newtonian kernel @xmath167 , riesz kernels @xmath168 with @xmath394 , or green kernels @xmath174 , provided that the euclidean distance between @xmath459 and @xmath460 is nonzero . here
@xmath461 is a regular domain ( in the sense of the solvability of the classical dirichlet problem ) and @xmath174 its green function .
recall that we are working under the standing assumptions stated in sections [ sec : gausspr ] and [ sec : standing ] . before having formulated an answer to problem [ pr ] , we observe that in the case @xmath462 the auxiliary @xmath463-problem can equivalently be formulated in terms of orthogonal projections @xmath464 onto the convex cone @xmath465 in the sense of the following lemma . due to the fact that @xmath466 for all @xmath5 ( see the footnote to formula ( [ g : finite ] ) ) , so that for any @xmath153 an orthogonal projection @xmath467 exists and is determined uniquely ( see section [ sec : proj ] ) . ]
given @xmath112 and @xmath462 , define @xmath468 and @xmath469 , the latter being thought of as an @xmath470-condenser .
for any @xmath471 , also write @xmath472 . with these assumptions and notations ,
consider the minimum energy problem @xmath473 [ aux1]for @xmath144 to solve the @xmath474-problem with @xmath462 , it is necessary and sufficient that there exist a solution @xmath475 to the problem ( [ pr : pr ] ) such that @xmath476 fix @xmath477 and let @xmath478 be one of the @xmath107-preimages of @xmath479 .
consider @xmath480 such that @xmath481 and @xmath482 .
then , by corollaries [ lemma3.3 ] and [ cor : ea ] , @xmath483 ; actually , @xmath484 . therefore , using representation ( [ gmu ] ) , we get @xmath485 where the first inequality is an equality if and only if @xmath486 . in view of the arbitrary choice of @xmath477 , this yields the lemma .
let @xmath116 consist of all @xmath487 such that @xmath488 .
assume @xmath489 , for otherwise problem [ pr ] has already been solved by theorem [ th : suff ] . in all that follows , we also suppose that @xmath490 .
a description of the set @xmath114 , required in problem [ pr ] , is given by theorems [ th : solvable1 ] and [ th : solvable2 ] , the main results of this study .
it is formulated in terms of a solution to the auxiliary @xmath491-problem , while the solvability of the latter is guarantied by the following theorem with @xmath492 .
[ lemma : exist ] given @xmath416 , where @xmath417 , assume that @xmath493 then the @xmath494-problem is solvable , and the class @xmath495 of all its solutions is vaguely compact . fix a solution @xmath144 to the auxiliary @xmath491-problem .
then , equivalently , @xmath496 gives a solution to the minimum energy problem ( [ pr : pr ] ) with @xmath492 , while @xmath497 is one of the @xmath107-preimages of @xmath498 ; see lemma [ aux1 ] .
[ th : solvable1 ] consider @xmath138 such that , @xmath2 , can be shown to be finite ( see section [ proof : th : solvable1 ] ) , so that inequalities ( [ br ] ) make sense .
] @xmath499 then @xmath500 while the main @xmath501-problem is solvable if and only if all the inequalities ( [ br ] ) are equalities . in the case where @xmath116 is a singleton , theorem [ th : solvable1 ]
can be refined as follows to establish a _
complete description _ of the set @xmath114 .
[ th : solvable2 ] let @xmath117 and assume , moreover , that @xmath502 then the set @xmath114 consists of all vectors @xmath138 with the property @xmath503 or , equivalently , @xmath504 theorem [ th : solvable2 ] makes sense , since , under its hypotheses , the value @xmath505 does not depend on the choice of @xmath506 ( see section [ proof : th : solvable2 ] ) .
see sections [ proof : lemma : exist ] , [ proof : th : solvable1 ] , and [ proof : th : solvable2 ] for the proofs of theorems [ lemma : exist][th : solvable2 ]
. combining theorem [ th : solvable1 ] with lemma [ lemma : h ] for @xmath492 ( see section [ sec : j ] below ) , we arrive at the following corollary .
[ cor : solvable1 ] assume @xmath116 is finite and @xmath507 .
then the @xmath501-problem is nonsolvable for every @xmath111 with @xmath508\quad\text{for all \ } \ell\in l,\ ] ] @xmath321 being the constant in the generalized maximum principle . in this section ,
we withdraw all the standing assumptions stated in sections [ sec : gausspr ] and [ sec : standing ] .
to illustrate the results obtained , we consider the riesz kernel @xmath118 of order @xmath119 $ ] , @xmath120 , in @xmath121 , @xmath122 , and we assume that @xmath1 , where @xmath509 , is the only plate with infinite interior capacity .
for this kernel , the operator of orthogonal projection @xmath510 is , in fact , the operator of riesz balayage @xmath125 onto @xmath1 , while ( see , e.g. , @xcite ) @xmath511 where @xmath127 is the @xmath126-green function of the open set @xmath128 .
hence , the auxiliary problem ( [ pr : pr ] ) with @xmath512 can be rewritten as @xmath513 the infimum being taken over all @xmath514 such that @xmath515 for all @xmath133 , while @xmath516 from formula ( [ aell ] ) now takes the form @xmath517 where @xmath475 is a solution to the minimum @xmath126-green energy problem ( [ pr : pr ] ) . combined with theorem [ th : solvable2 ] , this yields proposition [ pr : ex:11 ] ( see example [ ex:11 ] ) , providing a complete description of the set @xmath518 .
see also examples [ ex:12][ex:14 ] below , where such a description takes a much simpler form , given in terms of the geometry of the plate @xmath1 .
in each of the following examples , let @xmath134 for all @xmath5 and let @xmath135 be compactly supported in @xmath136 . )
holds automatically ( cf .
remark [ rem : fin ] ) .
] we also require that @xmath519 , @xmath520 , @xmath521 , @xmath128 is connected , and @xmath103 .
[ ex:12 ] under the hypotheses stated above , assume moreover that @xmath522 [ pr : ex:12 ] then the gauss variational problem ( relative to @xmath395 , @xmath112 , @xmath111 , @xmath519 , and @xmath103 ) is solvable if and only if @xmath523 is not @xmath126-thin at @xmath448 , while @xmath524 it is known from the riesz balayage theory ( see , e.g. , @xcite )
that , for any bounded positive measure @xmath525 , it holds @xmath526 , while according to ( * ? ? ?
* theorem 4 ) the inequality here is an equality if and only if @xmath1 is not @xmath126-thin at @xmath448 . combining this for @xmath527 with theorem [ th : solvable2 ] , where @xmath516 is now given by @xmath528 and taking assumption ( [ init ] ) into account
, we obtain the proposition .
[ ex:13]consider the coulomb kernel @xmath529 in @xmath12 and let @xmath1 be a rotational body consisting of all @xmath530 such that @xmath531 and @xmath532 , where @xmath533 and @xmath534 approaches @xmath535 as @xmath536 .
we focus with the following three cases : @xmath537,\label{ex2}\\ \rho(r)&=\exp(-r^s),\quad\text{where \ } s>1.\label{ex3}\end{aligned}\ ] ] then @xmath1 is not @xmath452-thin at @xmath538 in case ( [ ex1 ] ) , has finite ( newtonian ) capacity in case ( [ ex3 ] ) , and it is @xmath452-thin at @xmath538 though @xmath539 in case ( [ ex2 ] ) ; see @xcite . hence ,
if @xmath111 satisfies condition ( [ init ] ) , then , by proposition [ pr : ex:12 ] , the gauss variational problem is nonsolvable in case ( [ ex2 ] ) , while in case ( [ ex1 ] ) it admits a solution if and only if ( [ init ] ) additionally holds . also observe that , by theorem [ th : suff ] , in case ( [ ex3 ] ) the problem is solvable for _ any
_ @xmath111 .
[ ex:14]consider @xmath540 with @xmath541 and @xmath542 , and let @xmath543 .
then , in consequence of theorem [ th : solvable2 ] , the set @xmath518 forms the cone in @xmath544 defined by @xmath545 where @xmath546 is obtained from the equilibrium measure of @xmath440 relative to the kernel @xmath547 when divided by @xmath548 . note that @xmath549 ; hence , by ( * ? ? ?
* theorem 4 ) , the value on the right in relation ( [ aa ] ) is equal to @xmath550 if @xmath441 is not @xmath126-thin at @xmath448 , while otherwise it is @xmath551 .
for any @xmath214 and @xmath416 , @xmath417 , write @xmath553 , where @xmath554 has been defined in section [ sec : cond ] .
[ lemma : h ] the value @xmath555 remains unchanged if the class @xmath556 in its definition is replaced by @xmath557 , where @xmath558.\ ] ] observe that @xmath559 is finite , which is clear from assumptions ( [ ag ] ) , ( [ ginf ] ) and the boundedness of @xmath113 .
also note that , for any @xmath560 , @xmath561 hence , if @xmath421 , then @xmath562 and the lemma is obvious . therefore , assume @xmath462 . then , as is clear from the proof of lemma [ aux1 ] , @xmath555 remains unchanged if the class @xmath556 in its definition is replaced by its subclass consisting of all @xmath563 with the additional property that @xmath564 .
thus , the lemma will follow once we prove @xmath565 but this is a direct consequence of relation ( [ upper ] ) and lemma [ lemma : bound ] .
[ min : proj]if @xmath144 is a minimizer in the @xmath425-problem , then @xmath566 . for @xmath462
this follows from relations ( [ relrel ] ) and ( [ est1 ] ) , while otherwise it is obvious .
[ def : minnet]we call a net @xmath567 _ minimizing _ in the @xmath425-problem if @xmath568 @xmath559 being defined by ( [ h ] ) , and furthermore @xmath569 let @xmath570 consist of all those nets ; this class is nonempty in consequence of assumption ( [ g : finite ] ) and lemma [ lemma : h ] .
we denote by @xmath571 the union of the vague cluster sets of @xmath567 , where @xmath567 ranges over @xmath572 . taking a subnet if necessary
, we shall always assume @xmath573 to be strongly bounded .
then so are @xmath574 , @xmath575 and @xmath576 for all @xmath5 ; that is , @xmath577 indeed , this can be obtained from @xmath331 and @xmath578 for all @xmath224 , the latter being a consequence of @xmath579 and assumption ( [ kern.bound ] ) .
[ def : extr ] we call @xmath580 _ extremal _ in the @xmath425-problem if there exists @xmath573 that converges to @xmath581 both strongly and vaguely ; such a net @xmath567 is said to _ generate _
. the class of all those @xmath581 will be denoted by @xmath582 . , the tilde in the notation of an extremal measure will often be dropped . ]
it follows from lemma [ lemma : lower ] with @xmath583 that , for any @xmath584 , @xmath585 also observe that @xmath586 since , because of corollary [ min : proj ] , each @xmath587 ( provided exists ) can be treated as an extremal measure generated by the constant sequence @xmath588 . [
lemma : wm ] furthermore , the following assertions hold true : * from every minimizing net one can select a subnet generating an extremal measure ; hence , @xmath589 is nonempty .
moreover , @xmath590 * for any @xmath591 , it holds that @xmath592 .
hence , @xmath593 , while @xmath594 if and only if all @xmath4 , @xmath5 , are mutually disjoint .
* @xmath589 is vaguely compact .
choose arbitrary @xmath567 and @xmath595 in @xmath596 .
then @xmath597 where @xmath598 is the directed product of the directed sets @xmath227 and @xmath599 ( see , e.g. , ( * ? ? ?
* chapter 2 , section 3 ) ) . since @xmath600 is convex ( cf .
the footnote to corollary [ lemma3.3 ] ) , we get @xmath601 on the other hand , applying the parallelogram identity in the pre - hilbert space @xmath35 to @xmath602 and @xmath603 and then adding and subtracting @xmath604 , we obtain @xmath605 when combined with the preceding relation , this gives @xmath606 which yields ( [ fund ] ) when combined with ( [ min ] ) .
relation ( [ fund ] ) implies that @xmath567 is strongly fundamental . by theorem
[ th : complete ] with @xmath607 , for every vague cluster point @xmath10 of @xmath567 ( it exists ) we therefore get @xmath399 strongly .
this shows that @xmath10 is an extremal measure ; actually , @xmath608 . since the converse inclusion is obvious , relation ( [ wm ] ) follows .
having thus proved assertion ( i ) , we proceed by verifying ( ii )
. choose @xmath567 and @xmath609 generating @xmath610 and @xmath611 , respectively .
then application of ( [ fund ] ) shows that @xmath567 converges to both @xmath610 and @xmath611 strongly , hence @xmath592 , and consequently @xmath612 by ( [ twonorms ] ) . in view of the strict positive definiteness of the kernel ,
assertion ( ii ) follows . to establish assertion ( iii ) , fix @xmath613 ; then @xmath614 .
since the class @xmath615 is vaguely compact , there exists a vague cluster point @xmath616 of @xmath617 .
let @xmath618 be a subnet of @xmath617 that converges vaguely to @xmath616 ; then for every @xmath619 one can choose a net @xmath620 converging vaguely to @xmath621 . consider the cartesian product @xmath622 that is , the collection of all functions @xmath623 on @xmath599 with @xmath624 , and let @xmath172 denote the directed product @xmath625 . for every @xmath626 , write @xmath627 .
then the theorem on iterated limits from ( * ? ? ?
* chapter 2 , section 4 ) implies that the net @xmath628 , @xmath626 , belongs to @xmath629 and converges vaguely to @xmath616 ; thus , @xmath630 .
combined with ( [ wm ] ) , this yields assertion ( iii ) .
[ cor : uniqu]any two minimizers @xmath631 ( provided exist ) are equivalent in @xmath374 . hence , @xmath632 , while @xmath633 if and only if all @xmath4 , @xmath5 , are mutually disjoint .
this is a direct consequence of inclusion ( [ incl ] ) and assertion ( ii ) of lemma [ lemma : wm ] .
[ iorii ] for every extremal measure @xmath634 it holds that @xmath635 applying ( [ gmu ] ) to @xmath636 and each of @xmath637 , @xmath224 , where @xmath638 is a net generating @xmath639 , and using the fact that @xmath640 strongly , we get @xmath641=\lim_{s\in s}\,g_{\chi}(\boldsymbol\mu_s),\ ] ] which yields ( [ eqiorii ] ) when combined with ( [ min ] ) .
[ extr.min]let @xmath421 .
if the @xmath430-problem is solvable , then the class @xmath642 of all its solutions is compact in the vague topology ; moreover , @xmath643 due to assertion ( iii ) of lemma [ lemma : wm ] , the corollary will follow once we prove ( [ hrryu ] ) . as is seen from relations ( [ incl ] ) and ( [ eqiorii ] ) , to this end
it is enough to show that , for any given @xmath644 , @xmath645 because of assertion ( ii ) of lemma [ lemma : wm ] , for any @xmath646 one has @xmath647 , so @xmath648 , hence @xmath649 and , as an application of lemma [ lemma3.5 ] with @xmath583 , @xmath650 @xmath651 being finite by assumption ( [ ag ] ) . in view of relation ( [ inequal ] ) , this yields ( [ ex.min ] ) .
given @xmath416 , where @xmath417 , fix an extremal measure @xmath652 ; it exists according to assertion ( i ) of lemma [ lemma : wm ] .
a part of the proof of theorem [ lemma : exist ] is given as the following lemma , to be needed also in the sequel .
[ lemma : lemma]if @xmath653 for some @xmath654 , then @xmath655 we treat @xmath342 as a locally compact space with the topology induced from @xmath13 . given a set @xmath656 , let @xmath657 denote its characteristic function and @xmath658 the interior equilibrium measure associated with @xmath41 ( as the kernel is perfect , the existence of @xmath178 is guaranteed by theorem [ thf:1 ] ) .
also write @xmath659 , and let @xmath660 be the increasing filtering family of all compact subsets @xmath661 of @xmath342 .
we then observe that there is no loss of generality in assuming @xmath662 to satisfy condition ( [ growth ] ) with some @xmath663 and @xmath664 . indeed , otherwise @xmath662 has to be bounded from above ( say by @xmath665 ) , which combined with relation ( [ equ : f2 ] ) for @xmath666 results in inequality ( [ growth ] ) with @xmath667 , @xmath663 being arbitrary . to establish ( [ 24 ] ) , choose a ( strongly bounded ) net @xmath668 generating @xmath581 .
since @xmath669 is upper semicontinuous on @xmath342 while @xmath670 vaguely , we get @xmath671 on the other hand , application of lemma 1.2.2 from @xcite yields @xmath672 which together with the preceding inequality and relation ( [ inequal ] ) gives @xmath673 @xmath674 being the directed product of the directed sets @xmath227 and @xmath660 . hence , if we prove @xmath675 the lemma follows . for any @xmath676 such that @xmath677 , consider @xmath678 and @xmath679 , the interior equilibrium measures associated with @xmath680 and @xmath681 , respectively .
then @xmath679 minimizes the energy in the class @xmath682 , while @xmath678 belongs to @xmath682 ; see theorem [ thf:1 ] . by lemma [ lemma : convex ] with @xmath683 and @xmath684 , this yields @xmath685 but , as is seen from ( [ equf:1 ] ) , the net @xmath686 , @xmath687 , is bounded and nonincreasing , and so it is fundamental in @xmath294 .
the preceding inequality then yields that the net @xmath688 is strongly fundamental in @xmath102 .
since , clearly , it converges vaguely to zero , the property ( c@xmath162 ) ( see section [ sec:2 ] ) implies that zero is also its strong limit ; hence , @xmath689 write @xmath690 , where @xmath663 is the number involved in condition ( [ growth ] ) . combining relations ( [ growth ] ) and ( [ equ : f2 ] ) , the latter with @xmath691 , shows that the inequality @xmath692 subsists n.e . in @xmath342 , and hence @xmath693-a.e . in @xmath13 by lemma [ a.e
having integrated this relation with respect to @xmath693 , we then apply the hlder and , subsequently , the cauchy schwarz inequalities to the integrals on the right .
this gives @xmath694^{1/r_j}\ , \biggl[\int\kappa(x,\theta_{k^c_j})\,d\mu_s^j(x)\biggr]^{1/q_j}\leqslant \|\nu_j\|^{1/r_j}\|\theta_{k^c_j}\|^{1/q_j}\|\mu_s^j\|.\ ] ] taking limits here along @xmath674 and using relations ( [ twostars ] ) and ( [ 27 ] ) , we get ( [ 25 ] ) as desired .
let now assumption ( [ capfinite ] ) in theorem [ lemma : exist ] hold .
then , in consequence of lemma [ lemma : lemma ] , the extremal measure @xmath581 belongs to @xmath695 . because of
( [ eqiorii ] ) , this yields that , actually , @xmath581 is a solution to the @xmath696-problem , and the statement on the solvability follows .
it has thus been proved that @xmath697 . since the converse also holds according to inclusion ( [ incl ] )
, these two classes have to be equal . on account of assertion ( iii ) of lemma [ lemma : wm ] , this implies the vague compactness of @xmath698.@xmath699
recall that @xmath490 consists of @xmath487 with @xmath488 .
fix @xmath700 ( it exists due to theorem [ lemma : exist ] with @xmath492 ) and assume @xmath701 , @xmath702 , to satisfy relation ( [ br ] ) .
observe that @xmath703 is finite for each @xmath5 , so that inequalities ( [ br ] ) make sense . indeed , since @xmath704 according to corollary [ min : proj ] , one can assume @xmath90 to satisfy condition ( [ growth ] ) , for if not , then @xmath90 has to be bounded from above and the finiteness of @xmath705 is obvious . by the hlder and cauchy schwarz inequalities ,
we then obtain @xmath706^{1/r_i}\ , \biggl[\int1\,d\tilde{\lambda}^i\biggr]^{1/q_i}\leqslant \|\nu_i\|^{1/r_i}\|\tilde{\lambda}^i\|^{1/r_i } h^{1/q_i}<\infty\ ] ] as desired . here ,
@xmath707 and @xmath708 are the same as in condition ( [ growth ] ) and @xmath709 . to establish the theorem , for each @xmath2 we choose probability measures @xmath710 , @xmath711 , such that @xmath712 vaguely as @xmath713 and @xmath714 such @xmath715 exist due to the fact that @xmath716 for any compact @xmath247 , which in turn follows from @xmath488 because of the strict positive definiteness of the kernel .
further , for each @xmath711 define @xmath717 , where @xmath718 for all @xmath719 and @xmath720{\tau}^i_n}{\langle g,\tau^i_n\rangle}\quad\text{otherwise}.\ ] ] since @xmath721 for all @xmath702 because of assumption ( [ ginf ] ) , we have @xmath722 and also , due to relations ( [ ag ] ) and ( [ tau ] ) , @xmath723 estimate ( [ on ] ) yields that @xmath724 for all @xmath711 ( see section [ sec : cond ] ) ; hence , in view of the convexity of @xmath374 , @xmath725 and consequently , by ( [ gmu ] ) , @xmath726 where @xmath727 . but
@xmath728 while @xmath729 as @xmath713 , the latter being a consequence of estimate ( [ on ] ) .
combining relations ( [ raz1 ] ) and ( [ raz2 ] ) and then letting @xmath713 , we thus get @xmath730 this establishes assertion ( [ lll ] ) and , on account of inclusion ( [ raz3 ] ) , also the fact that @xmath731 as @xmath732 vaguely because of relation ( [ tauvague ] ) , the last inclusion yields @xmath733 and hence , by ( [ wm ] ) with @xmath421 , @xmath734 using corollary [ extr.min ] , we then see that the main @xmath501-problem is solvable if and only if @xmath144 serves as one of its minimizers , which in turn , due to ( [ eqiorii ] ) with @xmath421 , holds if and only if @xmath735 since the latter is equivalent to the requirement that all the inequalities ( [ br ] ) are equalities , the proof is complete.@xmath699
it is seen from ( [ rpot ] ) that , for any @xmath736 , the @xmath737-weighted vector potential @xmath738 is finite n.e . in @xmath13 and its components can be written in the form @xmath739 therefore , for any @xmath740 , @xmath741 which follows from lemma [ lemma3.5 ] when applied to @xmath742 and each of the functions @xmath743 and @xmath744 .
let @xmath745 be extremal in the gauss variational problem .
taking ( [ gmu ] ) and ( [ eqiorii ] ) into account , we then conclude from ( [ xxx ] ) with @xmath746 that @xmath747,\ ] ] where the expression on the right ( hence , also that on the left ) does not depend on the choice of @xmath745 in consequence of assertion ( ii ) of lemma [ lemma : wm ] . below . ]
the proof of theorem [ th : solvable2 ] , to be given in section [ proof : th : solvable2 ] , is based on a description of the @xmath407-weighted extremal potential @xmath748 provided as follows .
[ th : descpot1 ] given @xmath644 , it holds that for all @xmath5 ( see the footnote to assumption ( [ g : finite ] ) ) . ]
@xmath749 relations ( [ wsum ] ) and ( [ descpot1 ] ) characterize uniquely the vector @xmath750 , @xmath5 , in the sense that , if these two relations are satisfied by some @xmath751 in place of @xmath750 , then @xmath752 having fixed @xmath5 , we start with the observation that , for any @xmath753 , @xmath754 is a strongly continuous function of @xmath755 . indeed , this is clear from representation ( [ wpopot ] ) in view of the isometry between @xmath374 and @xmath756 .
choose a net @xmath757 generating the extremal measure @xmath745 .
then the above observation can slightly be generalized in the following way : @xmath758 indeed , according to relation ( [ twostars ] ) , the net @xmath759 is strongly bounded ( say by @xmath665 ) . since it converges to @xmath760 vaguely , the property @xmath761 ( see section [ sec:2 ] ) yields that @xmath762 weakly ; hence , for every @xmath208 , @xmath763 whenever @xmath224 is large enough . furthermore , by the cauchy
schwarz inequality , @xmath764 as @xmath765 strongly , the last two relations combined give @xmath766 which is equivalent to ( [ threestars ] ) because of representation ( [ wpopot ] ) . in order to establish assertion ( [ descpot1 ] ) , we now assume , on the contrary , that there is a set @xmath767 of interior capacity nonzero with the property @xmath768 having chosen @xmath769 so that @xmath770 ( such a measure exists ) , we then obtain @xmath771 to get a contradiction , for all @xmath772 and @xmath224 we define @xmath773 , where @xmath774 since then @xmath775 corollary [ lemma3.3 ] implies @xmath776 . actually , @xmath777 for all @xmath224 , which in view of relations ( [ gmu ] ) , ( [ twostars ] ) , and ( [ wpopot ] ) yields @xmath778 @xmath779 being positive .
passing here to the limit as @xmath226 ranges along @xmath227 , on account of ( [ min ] ) , ( [ threestars ] ) and the continuity of @xmath780 relative to @xmath781 we get @xmath782 which leads to @xmath783 by letting @xmath784 .
this contradicts to inequality ( [ threeheart ] ) . to complete the proof ,
assume now relations ( [ wsum ] ) and ( [ descpot1 ] ) to hold also with some @xmath751 , @xmath5 , in place of @xmath750 .
using the fact that the union of two universally measurable sets with interior capacity zero has interior capacity zero as well ( see @xcite ) , for each @xmath5 we then have @xmath785 which according to lemma [ a.e . ] also holds @xmath786-a.e . in @xmath13 for each @xmath224 . having integrated this relation with respect to @xmath786 and then summing up the inequalities obtained over all @xmath5 , on account of ( [ xxx ] ) and ( [ wsum ] ) we get @xmath787=\sum_{i\in i}\,\bigl\langle w_{\boldsymbol\gamma}^i,\gamma^i\bigr\rangle=\kappa(\chi+r\boldsymbol{\gamma},r\boldsymbol{\gamma}),\end{aligned}\ ] ] which in view of the strong ( hence , weak ) convergence of @xmath788 to @xmath789 proves ( [ etaeta ] ) .
[ eta3]for any @xmath790 , it holds that @xmath791 indeed , then @xmath792 by assertion ( ii ) of lemma [ lemma : wm ] , which in view of corollary [ lemma : potequiv ] yields @xmath793 n.e . in @xmath13 .
combining this with relation ( [ descpot1 ] ) , we get @xmath794 which establishes the corollary due to the uniqueness statement in theorem [ th : descpot1 ] .
fix @xmath795 ; it exists according to theorem [ lemma : exist ] with @xmath796 . in consequence of theorem
[ th : solvable1 ] , theorem [ th : solvable2 ] will be proved once we establish the existence of a solution to the ( main ) @xmath797-problem under the hypothesis @xmath798 ( observe that the value on the right does not depend on the choice of @xmath144 , which is clear from assumption ( [ nonint ] ) in view of @xmath799 for all @xmath800 ; see corollary [ cor : uniqu ] . )
fix @xmath807 . for every @xmath803 ,
consider @xmath808 , the trace of @xmath809 on @xmath802 . applying ( * ? ? ?
* lemma 1.2.2 ) , we then obtain for the notation @xmath810 . ]
@xmath811 choose a compact set @xmath812 so that @xmath813 , which is possible due to ( [ w ] ) , and for all @xmath803 that follow @xmath814 define @xmath815 , where @xmath816 since then @xmath817 from corollary [ lemma3.3 ] we get @xmath818 , and hence , by ( [ gmu ] ) and ( [ w])([hatmu ] ) , @xmath819 which in view of the arbitrary choice of @xmath820 proves @xmath821 since the converse inequality is obvious , the lemma is established .
in the rest of the proof we assume @xmath822 for all @xmath803 , which involves no loss of generality because of condition ( [ g : finite ] ) and lemma [ lemma : cont ] .
then , according to theorem [ th : suff ] , for every @xmath823 there exists @xmath824 , while in consequence of lemma [ lemma : cont ] , these minimizers form a net minimizing in the @xmath501-problem , i.e. @xmath825 note that , under the standing assumptions ( cf . also remark [ rem : lower ] ) , ( * ? ? ?
* theorem 7.2 ) is applicable to each of @xmath833 , @xmath803 .
this gives @xmath834 another observation is that , according to ( [ threestars ] ) with @xmath826 instead of @xmath567 , @xmath835 having substituted ( [ wpopot ] ) with @xmath836 into the left - hand side of inequality ( [ nuuu ] ) , we pass to the limit as @xmath832 . in view of ( [ nu ] ) and the lower semicontinuity of @xmath837 on @xmath4 ( see remark [ rem : lower ] ) , we then see that assertion ( [ descpot2 ] ) will be proved once we establish @xmath838 but these two follow directly from lemma [ lemma : vague.cont ] and the lower semicontinuity of the mapping @xmath839 on @xmath840 ( see ( * ? ? ?
* lemma 2.2.1 ) ) , respectively .
assume , on the contrary , that @xmath845 then , according to ( [ xxx ] ) with @xmath746 , @xmath846 as all the coordinates of the given @xmath795 are bounded by corollary [ min : proj ] , for each @xmath5 relation ( [ descpot1 ] ) holds @xmath847-a.e . in @xmath13 ; that is , @xmath848 having integrated this relation , divided by @xmath849 , with respect to @xmath847 and then summing up the inequalities obtained over all @xmath5 , on account of equalities ( [ wjzero ] ) , ( [ sumwj ] ) and @xmath850 for all @xmath133 we get @xmath851 which because of ( [ xxx ] ) with @xmath852 and @xmath853 can be rewritten in the form @xmath854 or , equivalently , @xmath855 applying the cauchy schwarz inequality and , subsequently , identity ( [ gmu ] ) , we then obtain @xmath856 which in view of inclusion ( [ g000 ] ) yields @xmath857 .
now , having integrated ( [ descpot3 ] ) for @xmath859 with respect to the ( bounded ) measure @xmath860 , on account of lemma [ 0000 ] we obtain ( [ g0000 ] ) , which together with inclusion ( [ g000 ] ) shows that , actually , @xmath861 .
combined with corollary [ iorii ] , this proves that @xmath745 is , in fact , a solution to the @xmath501-problem , and theorem [ th : solvable2 ] follows.@xmath699 * acknowledgments * the author is very grateful to professor w.l .
wendland for the careful reading of the manuscript and his suggestions to improve it .
a part of this research was done during the author s visit to the dfg cluster of excellence simulation technologies at the university of stuttgart during february of 2012 , and the author acknowledges this institution for the support and the excellent working conditions . | we continue our investigation of the gauss variational problem for infinite dimensional vector measures associated with a condenser @xmath0 .
it has been shown in potential anal .
, doi:10.1007/s11118 - 012 - 9279 - 8 that , if some of the plates ( say @xmath1 for @xmath2 ) are noncompact then , in general , there exists a vector @xmath3 , prescribing the total charges on @xmath4 , @xmath5 , such that the problem admits _ no _ solution .
then , what is a description of all the vectors @xmath6 for which the gauss variational problem is nevertheless solvable ?
such a characterization is obtained for a positive definite kernel satisfying fuglede s condition of perfectness ; it is given in terms of a solution to an auxiliary extremal problem intimately related to the operator of orthogonal projection onto the cone of all positive scalar measures supported by @xmath7 .
the results are illustrated by examples pertaining to the riesz kernels .
.2 cm _ subject classification _ : 31c15 .
.1 cm _
key words : _ infinite dimensional vector measure , external field , minimal energy problem * necessary and sufficient conditions for the solvability of the gauss variational problem for infinite dimensional vector measures * natalia zorii |
type i agns are often classified into two subclasses according to the full width at half maximum ( fwhm ) of their broad h@xmath0 lines . those agns with fwhm greater than approximately 2000 km@xmath1s@xmath2
are called broad line seyfert 1 ( bls1 ) , and those with fwhm less than approximately 2000 km@xmath1s@xmath2 are called narrow line seyfert 1 ( nls1 ) .
there is a tight correlation between the black hole mass and the stellar velocity dispersion of the bulge of normal galaxies , the so called m - sigma relation ( e.g. magorrian et al .
1998 ; gebhardt , et al . 2000 ; tremaine et al . 2002 ; marconi & hunt 2003 ; ferrarese & ford 2005 ) .
bls1s are found also to follow this relation well ( greene & ho 2006 ) .
however , nls1s seem to deviate from this relation ; they seem to have much smaller masses or much higher stellar velocity dispersions ( e.g. , wang & lu 2001 ; bian & zhao 2004 ; zhou et al .
2006 ; komossa & xu 2007 ) .
_ we call this under - massive black hole problem of nls1s . _ since there are a lot of uncertainties on measuring the black hole mass and sigma ( decarli et al . 2008 ; komossa & xu 2007 ; komossa 2008 ; komossa et al .
2008 ) , it is difficult to determine whether the deviation reflects an intrinsic mass or / and sigma difference between the nls1s and bls1s , or it is only caused by the measurements . nls1s also show softer x - ray spectra , strong fe ii emission , rapid continuum variation , and high accretion rate ; the last one may partly be caused by the incorrect measurement of the black hole mass ( e.g. , grupe & mathur 2004 ; komossa 2008 ) .
however , williams et al .
( 2004 ) found some optically defined nls1s are different from those defined by x - ray selected nls1s , suggesting that strong , ultrasoft x - ray emission is not a universal characteristic of nls1s and challenging the current paradigm that nls1s accrete near the eddington limit . in the simple unified model for agn , type
i and type ii objects differ only in terms of the angle between the observer s line of sight and the normal axis of a dusty torus ( urry & padovani 1995 ) .
seds of qsos also imply that the location of the inner wall of torus is determined by the sublimation of dust by the central radiation ( e.g. kobayashi et al .
1993 ; elvis et al .
it is widely accepted that @xmath3 ( e.g. , elitzur 2006 ; peterson 2007 ; nenkova 2008a ) , consistent with that predicted by the toy model of krolik & begelman ( 1988 ) .
reverberation mapping ( rm ) based on infrared emission also supports this relation ( suganuma et al .
2006 ) .
many observations show that the fraction of type i agns increases with luminosity ( ueda et al .
2003 ; hasinger 2004 , 2005 ; steffen et al .
2003 , 2004 ; barger et al . 2005
; wang et al . 2005 ) , i.e. , brighter agns are mostly type i agns .
this is also equivalent observationally to the situation that it is less probable to find very luminous type ii agns ( mostly seyfert ii galaxies ) in a random sample of agns .
_ we call this under - populated luminous type ii problem of seyfert galaxies .
_ however , the standard receding torus model with constant height first suggested by lawrence ( 1991 ) failed to describe this luminosity fraction , and thus a modified receding torus whose height slightly increases with luminosity with the relation @xmath4 is needed ( simpson 2005 ) .
based on the luminosity fraction for type i agn , wang et al .
( 2005 ) obtained a relation between covering factor of the torus and luminosity , @xmath5 .
the variation of covering factor with luminosity may be totally caused by the receding of the inner torus ( wang et al . 2005 ) .
clearly both the broad line regions and the dust tori of agns show strong dependence of luminosity , yet the physical connections between these two key ingredients of agn s unification scheme are not yet fully understood .
it is also not clear if the luminosity dependence of dust torus and broad line region is simultaneously related to both problems of under - massive black holes in nls1 galaxies and under - populated luminous type ii seyfert galaxies .
progress towards solving these problems may shed new lights to the understanding the hierarchical evolution of agn and its host galaxy , as well as the feeding , fueling and growth of supermassive black holes . in this paper
we attempt to address both problems in a coherent way by studying a sample of sdss agns with strong broad emission lines .
the geometry and kinematics of broad line region in agn have been studied for about three decades but far from been fully understood .
there are mainly two kinds of views to interpret the structure of blr based on the profiles of balmer lines .
one interprets the profile as a gaussian / lorentz profile and blr as a extended region has different projected velocity distributions ( e.g. , goncalves et al .
1999 ; collin et al .
the other model is that the profile is a superposition of double gaussian profiles ( intermediate component+very broad component in hu et al .
2008 ; broad component+very broad component in sulentic et al , 2000 and marziani et al .
these two components will be re - named as intermediate gaussian component ( imgc)+very broad gaussian component(vbgc ) in this paper . ) , which come from two physically distinct emission regions due to their substantial differences in the line widths ( e.g. , hu et al . 2008 ; sulentic et al . 2000 ; brotherton et al .
1996 ; mason et al .
1996 ) ; we call these two regions very broad line region ( vblr ) and intermediate line region ( imlr ) . for the two - components model ,
the results from these papers are not consistent with each other and the physical interpretations to these two emission regions are also different .
sulentic et al ( 2000 ) suggested that vbgc is likely to arise in an optically thin region close to the central source which is slightly red - shifted , whereas hu et al .
( 2008 ) concluded that imgc is systematically red - shifted and may come from the inflow .
in addition , the study of partially obscured quasars suggests that there is an inner narrow line region covered by dust which may be consistent with our imlr ( zhang et al .
the existence of these two emission regions need to be confirmed further and their dynamical and physical properties also need to be further studied . for this purpose , we select a sdss sample with strong h@xmath6 , h@xmath0 and h@xmath7 lines , and decompose these balmer lines based on the two - components model . our goal is to provide further evidence for this model , carry out more detailed analysis about the dynamics and evolution of these two emission regions , and ultimately to understand the above two problems ( the under - massive black hole problem of nls1s and the under - populated luminous type ii problem of agns ) in a coherent way .
this paper is organized in the following way . detailed decomposition and fwhm measurement
are described in section 2 . in section 3
we present h@xmath6 , h@xmath0 , and h@xmath7 decomposition and statistical analysis . in section 4
we confirm the two - components model based on the line decomposition and give detailed analysis of these two emission regions .
section 5 presents other supporting evidence for this model .
section 6 shows that this model provides a coherent interpretation to the two problems .
section 7 is conclusion and discussion .
the continuum luminosity and some other parameters of agns are directly taken from table 2 in la mura et al .
fwhm is defined as fwhm of the whole line , fwhmi means fwhm of the imgc and fwhmb represents fwhm of the vbgc when they are not clear in the context .
the unit is km s@xmath2 for the width " of all lines and components , as well as the inferred velocity , throughout this manuscript .
the sample contains 90 objects with clear h@xmath6 , h@xmath0 , and h@xmath7 line profiles that can be decomposed .
they are selected by la mura et al .
( 2007 ) from sdss 3 based on their balmer line intensities .
narrow emission lines have already been removed by la mura et al .
( 2007 ) from the spectra we adopted , with templates extracted from oiii and with compatible width ; the detailed description on how they have produced the broad line spectral catalog we used here is presented in la mura et al .
therefore all analysis and discussions in this paper do not concern with those narrow lines .
21 of them are classified as nls1s .
a two - component model ( a disk line plus a gaussian component ) to fit the profile of the broad lines has been presented by popovi et al .
( 2008 ) and bon et al .
we first tried this model .
however , the fitting is not acceptable statistically for most of the objects in this sample . instead
, a double gaussian component model can fit the profile very well .
first , we fit h@xmath6 and h@xmath0 lines freely with two gaussian components , i.e. , a vbgc and an imgc ; a wavelength deviation of 10 @xmath8 from their rest frame wavelength is allowed in the fitting for the central value of the two components .
when the imgc is statistically insignificant , we refit the spectrum with just one gaussian component .
our results are generally consistent with the results reported by mullaney & ward ( 2008 ) ; they decomposed nine h@xmath6 lines , three of which only need one gaussian component .
the decomposition of h@xmath6 is thus quite straight forward .
however , it is more complicated to decompose the h@xmath0 lines , because of the contamination of a broad line he i at 4922 @xmath8 and fe ii on the red wing of h@xmath0 ( vron et al .
2002 ; mullaney & ward 2008 ) .
we thus add a gaussian component with centers ranging from 4922 @xmath8 to 4940 @xmath8 when necessary .
the profile of the feii is not important here . due to the above complication and the relative weakness of the h@xmath0 lines ,
sometimes there are relatively large uncertainties in the h@xmath0 line decomposition .
second , with the result of the first step , to every h@xmath6 and h@xmath0 line , we get the width of their vbgcs , @xmath9 and @xmath10 , respectively . if @xmath11 , we refit the h@xmath6 by requiring @xmath12 and fit the corresponding h@xmath7 by also requiring @xmath13 .
conversely , if @xmath14 , we refit the h@xmath0 by requiring @xmath15 and fit the corresponding h@xmath7 by also requiring @xmath16 .
the motivation of the above requirement is to ensure the fittings are physical , i.e. , the vbgcs of these three lines are produced from regions with radius not very different , since @xmath17 , if the same gravitationally - bound gas dynamics applies to all three lines .
we do not impose the same radius for the emission line regions for the three vbgcs , because rm measurements have found different delay times for different emission lines ( e.g. , peterson & wandel 1999 ; kaspi et al .
the difficulty of fitting h@xmath7 comes from its weak flux , roughly about two times weaker than h@xmath0 .
since h@xmath6 has a clear red wing , it also helps to reduce the influence of fe ii contamination in h@xmath0 fitting .
a single gaussian component model is not acceptable for the majority of the sources in this sample .
this is consistent with the results of collin et al .
( 2006 ) , who found that most broad line profiles of agns are non - gaussian and deviations from a gaussian profile are correlated with the properties of agns .
almost all of the spectra ( 88/90 for h@xmath6 and h@xmath0 , 74/90 for h@xmath7 ) can be fitted very well with this double gaussian components model .
for this reason , model lines based on two gaussian components are used to measure the fwhm of the whole line ( e.g. , greene & ho 2005 ) .
since the error estimate is very complex , we just use the typical error of 10% of fwhm , following greene & ho ( 2005 ) and vestergaard & peterson ( 2006 ) .
those that do not fit well have been excluded from our analysis .
several typical decomposition examples are presented in fig.[fig : hlines ] and fig.[fig : habk ] .
the decomposition parameters of these objects are presented in the table . ,
h@xmath6 and h@xmath7 lines . from ( a ) to ( c ) , fwhm of the line increases , and the imgc becomes weaker and may disappear sometimes , e.g. , in subpanels c1 & c3 .
note that for in subpanel c2 , the imgc of h@xmath6 line is still detectable , in contrast to h@xmath0 and h@xmath7 lines , because the imgc of h@xmath6 is normally much stronger than h@xmath0 and h@xmath7 lines ( see fig.[fig : wr ] for line intensity ratios between these three lines ) . ) ] since these parameters are somewhat degenerated in the fitting , monte carlo simulations are carried out to examine if this degeneration affects the reliability of our spectral decompositions . as shown in fig.[fig : mont ] , 80 simulated spectra are produced , with each one composed of an imgc and a vbgc ; each parameter of these gaussian components is a random choice from a corresponding parameter group which covers the similar parameter range in our sample .
proper noise is also added to each simulated spectrum .
we then decompose these simulated spectra and obtain the fitted parameters .
fig.[fig : mont ] shows that the fitted parameters do not deviate from the initially set parameters significantly .
first , we focus on h@xmath0 lines ; several examples are shown in fig.[fig : hlines ] ( a1 , b1 & c1 ) and fig.[fig : habk ] ( b ) .
the statistical analysis is presented in fig.[fig : hstatistics ] ( a1 , b1 & c1 ) .
the vbgc has a much larger fwhm than the whole line when fwhm of the whole line is small .
the fwhm ratios range from 1.5 to 4 in the nls1 group which lie on the left of fig.[fig : hstatistics](a1 ) .
the intensity ratio of the vbgc to the whole line is about 0.6 in these objects .
this means that for nls1s the imgc has an intensity comparable to the vbgc , and consequently the fwhm of the whole line of nls1s is dominated by the imgc . with fwhm increasing ,
the imgc becomes weaker and finally disappeared , which can be seen from fig.[fig : hstatistics](b1 ) , in which the intensity ratio of the vbgc to the whole line reaches unity when fwhm of the whole line reaches about 5000 km@xmath1s@xmath2 .
naturally , the fwhm ratio of the vbgc to the whole line also reaches unity .
the whole line can thus be simply described by a single vbgc and the vbgc becomes totally dominant .
this trend agrees with that obtained by hu et al .
( 2008 ) , who have shown that most of the agns with very large black hole masses ( normally with largest fwhm ) do not need two components for their broad h@xmath0 lines .
our result also agrees with collin et al .
( 2006 ) , who found that nls1s seem to have a prominent narrower component ( the imgc identified in our work ) on top of a broad gaussian profile ( the vbgc identified in our work ) . in fig.[fig :
hstatistics](c1 ) , fwhm ratio of the imgc to the vbgc becomes larger with increasing fwhm .
it ranges from @xmath18 to @xmath19 , which is another reason for the fwhm of the whole line reaching the fwhm of the vbgc .
it indicates that the imgc also becomes broader when it becomes weaker .
the points lie on @xmath20 of the bottom picture represent the objects which have only a single gaussian component .
the three points with maximum fwhm ( but still with two gaussian components ) fall off the trend slightly , due to perhaps the large uncertainties in fitting when the intensity of the imgc becomes very low . compared to fig.2 ,
the correlations in fig.3 are much tighter , suggesting that the two - components decomposition is physically meaningful , not just a mathematical convenience . moreover ,
the systematical evolution show in fig.3 suggests that the two components are physically distinct , and both components responses to a common source .
fig.[fig : hlines ] ( a2 , b2 & c2 ) and fig.[fig : habk ] ( a ) are several decomposition examples for h@xmath6 lines .
they behave similarly when fwhm is small .
however when fwhm is large , the imgc is very broad and strong .
statistical analysis is shown in fig.[fig : hstatistics ] ( a2 , b2 & c2 ) in a similar manner to h@xmath0 lines .
the obvious difference is that the number of points near unity in panel ( b2 ) is less than that in panel ( b1 ) for h@xmath0 lines ; only a weakly increasing trend can be seen in panel ( b2 ) .
this means that the imgc is still strong even when fwhm reaches about 5000 km@xmath1s@xmath2 .
h@xmath7 lines behave similarly to h@xmath0 lines as shown fig.[fig : hlines ] ( a3 , b3 & c3 ) and fig.[fig : habk ] ( c ) ; when fwhm is very large , only a single gaussian component ( the vbgc ) is required to fit the profile . however , when fwhm is very small , a large number of h@xmath7 lines loose the vbgc , i.e , only the imgc is required to fit the line , because most of their fwhm are very close to the imgcs of their h@xmath0 lines .
the points lie on the x - axis in fig.[fig : hstatistics ] ( b3 ) are the cases when the h@xmath7 lines are described by a single imgc ; an example is presented in fig.[fig : habk ] ( c ) .
however some of the single gaussian component h@xmath7 lines have fwhm neither close to the vbgc nor close to the imgc of the corresponding h@xmath0 lines , therefore we exclude these points in our analysis .
the loss of the vbgcs may be caused by the weak intensity of h@xmath7 , since its vbgc is particularly weak when fwhm is small , which can cause confusion in the continuum subtraction .
the gap pointed by an arrow in fig.[fig : hstatistics ] ( b3 ) also reveals a sudden decrease of the intensity of the broad gaussian component when it is very weak .
previous works suggested that a highly significant correlation exists between fwhm(h@xmath0 ) and source luminosity ( e.g. , joly et al .
1985 ; corbett et al . 2003 ) .
more recent work with a larger sample which spans a fwhm range from 1000 to 16000 km s@xmath2 showed that the correlation is not so strong but still statistically significant ; the correlation in this large sample is likely driven by the minimum fwhm trend ( marziani et al 2009 ) .
fig.[fig : l - fwhm]a presents this correlation in our sample ; the correlation is more obvious in this smaller sample than in marziani et al ( 2009 ) , and it is very similar to the correlation in joly et al ( 1985 ) .
the extremely broad h@xmath0 ( fwhm @xmath21 about 7000 km s@xmath2 ) may have a rather different line profile and may originate from a different physical region from the less broad h@xmath0 ( e.g. , strateva et al .
2003 ; eracleous et al . 2003 ) , therefore the analysis with only mediate broad ( fwhm @xmath22 about 7000 km s@xmath2 ) h@xmath0 with the same line profile seems to be more meaningful .
fig.[fig : l - fwhm]b shows that the logarithm of fwhm of the vbgc or imgc increases with the logarithm of luminosity in a more linear way than fwhm of the whole line .
this may indicate that fwhm of vbgc or imgc is more physical than fwhm of the whole line . on the other hand , fig.[fig
: l - fwhm]b also shows that fwhm of imgc increases with luminosity much faster than vbgc ; the two components has a trend to merge into one
. this trend will be confirmed in another way in section 4.1 .
therefore fwhm evolution also supports the picture that the broad line region is composed of two physically distinct regions .
we assume that the two gaussian components come from two distinct regions , and both of the these two regions ( vblr and imlr ) are bounded by the central black hole s gravity .
we therefore have + @xmath23 and + @xmath24 equation ( 1 ) and ( 2 ) lead to + @xmath25 where @xmath26 we assume that @xmath27 is a constant for all sources , because currently we can not determine the exact geometry and velocity distribution of gases in the imlr .
the validity of this assumption and the exact value of @xmath27 needs to be determined with further studies of more observation data . for convenience
we simply redefine @xmath28 for the rest of this paper , i.e. , we only consider the special case when @xmath29 ( see section 5.2 for a tentative support to this rather arbitrary choice ) .
rm is used to measure radius of the blr , which is related to an agn s continuum luminosity by the following empirical radius - luminosity relation ( kaspi et al .
2005 ) : + @xmath30 or the starlight - corrected r - l relation ( bentz et al .
2006 ) : + @xmath31 we use equation ( 6 ) to calculate @xmath32 throughout this work , unless indicated otherwise . equation ( 5 ) is only used in section 4.4 where a comparison is made . because rm is used to measure radius of blr
, one might expect that it would normally measure the radius of the innermost emission - line region , i.e. , the vblr in our model ( see discussion on this issue in section 5.2 ) .
therefore , the radius calculated from @xmath33l5100 relation is taken as @xmath34 .
consequently , we can calculate the radius of imlr from equation ( 3 ) . both h@xmath6 and
h@xmath0 lines can be used to do this calculation .
because they seem to behave slightly differently in the line evolution , @xmath35 is obtained from h@xmath6 and h@xmath0 lines separately .
fig.[fig : rrf ] is the evolution of @xmath34 and @xmath35 with fwhm .
imlr radii derived from h@xmath6 and h@xmath0 lines have only subtle differences .
the imlr radius just varies around a constant with increasing fwhm , i.e. does not have a systematic increase or decrease . on the other hand , vblr radius becomes larger with increasing fwhm .
vblr and imlr are clearly separated with fwhm@xmath36 km@xmath1s@xmath2 as we can see in fig.[fig : rrf ] .
the circles filled with dots represent the objects whose lines are fitted well with a single ( very ) broad gaussian component ; they appear in the fig.[fig : rrf ] with largest fwhm .
the evolution of @xmath34 and @xmath37 with luminosity and black hole mass are shown in fig.[fig : rr ] .
the radius of the imlr increases more slowly than vblr with black hole mass and luminosity .
the two emission regions have a trend to merge into one with higher luminosity or larger black hole mass ; it is consistent with the emission lines behavior as shown in section 3.4 .
the points which represent objects with zero imgcs also mainly lie on the large mass and large luminosity side .
the slopes of these relations will be discussed in section 4.4 .
although most of the h@xmath6 lines still need two gaussian components to fit , there are three of them which only need one gaussian component in our sample .
the two gaussian components in h@xmath6 lines also show a trend to merge into one single gaussian component with fwhm becoming larger , though the trend is not as obvious as for h@xmath0 lines .
once again the systematically different trends of evolution of @xmath34 and @xmath37 are consistent with our assumption that they are physically two distinct emission regions .
it is possible that density , temperature and geometric shape of the two regions ( vblr and imlr ) are different , then the collisional effect and different ionization energies of the three balmer lines may cause different balmer decrement in the two regions . here
we show the balmer decrement of the whole line , vbgc and imgc in fig.[fig : wr ] .
there are two sources whose @xmath38 or @xmath39 or both do not fit well with the double gaussian model ; these two points are excluded in fig.[fig : wr]a when calculating the intensity ratio of @xmath38 to @xmath39 .
similarly , there are sixteen sources whose @xmath39 or @xmath40 or both do not fit well ; these points are excluded when calculating the intensity ratio of @xmath39 to @xmath40 .
the points excluded in fig.[fig : wr]a are also excluded in fig.[fig : wr]b and fig.[fig : wr]c .
in addition , fifteen sources @xmath40 do not have broad gaussian ; these points do not appear in intensity ratio of h@xmath0 to @xmath40 in fig.[fig : wr]b .
fourteen sources @xmath38 or @xmath39 or both only need a vbgc to fit ( i.e. , they do not have imgc ) , so these sources are not included in fig.[fig : wr]c when calculating the @xmath38 to @xmath39 ratio .
similarly 22 sources @xmath39 or @xmath40 or both only have vbgc , and are thus excluded in fig.[fig : wr]c when calculating the @xmath39 to @xmath40 ratio . in summary
, there are 88 , 88 and 76 points used to calculating h@xmath6 to h@xmath0 ratio in fig.[fig : wr]a , b and c , respectively ; correspondingly , there are 74 , 59 and 52 points in h@xmath7 to h@xmath0 ratio . in vblr
, h@xmath6/h@xmath0 is 2.54 and h@xmath7/h@xmath0 is 0.37 , which is consistent with a pure h model with , ne=@xmath41 @xmath42 ( krolik et al .
1978 ) . in imlr ,
h@xmath6/h@xmath0 is 4.78 , which is much higher than that in vblr , consistent with previous results ( netzer & laor 1993 ; mullaney & ward 2008 ) .
these can be caused by collisional effect which can lead to an effective emission increase for h@xmath6 but not for h@xmath0 or h@xmath7 ( e.g. osterbrock 1989 ) .
this result suggests a higher gas density in imlr .
it has also been argued that dust in the emission region can cause higher balmer decrement ( binette et al .
therefore , it also suggests that imlr may be contaminated by dust . on the other hand , the h@xmath7 to h@xmath0 ratio is also slightly higher in imlr ( 0.41 ) than that in vblr ( 0.37 ) . as h@xmath0 and
h@xmath7 are produced from similar process , the slightly higher h@xmath7 to h@xmath0 ratio is not well understood . in fig.[cor_lines1 ] , we show the correlations between the first three balmer lines for the imgc , i.e. , between the h@xmath6 and h@xmath0 lines , and between the h@xmath7 and h@xmath0 lines .
the linear correlations between them indicates that the imgcs for all these three lines originate from physically connected regions , even if not exactly from the same region . in fig.[cor_lines2 ] , we show the distributions of fwhm differences between the three lines for the imgc .
the fwhm of h@xmath6 and h@xmath7 is offset systematically by around -200 km s@xmath2 and + 200 km s@xmath2 around that of h@xmath0 , respectively .
this suggests a stratified geometry for the imlr , where h@xmath7 , h@xmath0 and h@xmath6 lines are produced at increasing radii respectively .
fwhm of the whole line , h@xmath6 is also systematically larger than h@xmath0 ( greene & ho 2005 , shen et al 2008 ) .
it is consistent with the rm result ( kaspi et al .
2000 ) which showed that radius of blr obtained from h@xmath6 is larger than that of h@xmath0 in most of the sources .
although there is no difference in the ionization degree of h@xmath6 , h@xmath0 and h@xmath7 , the above result can be explained that the inner skin of torus is the imlr , therefore , it has more dust in the region with larger radius .
there are mainly two processes .
first , dust reddening will make the h@xmath7 produced in the region much closer to the torus be extincted most seriously , the h@xmath0 will suffer less extinction and h@xmath6 will be least extincted .
therefore the average radius of the h@xmath6 region will be larger than h@xmath0 and h@xmath7 .
second , collisional excitation contributes a lot to h@xmath6 line but not significantly to h@xmath0 and h@xmath7 line , so the region much closer to the torus with a higher density will produce relatively more h@xmath6 .
this also causes the fwhm of h@xmath6 to be smaller than h@xmath0 and h@xmath7 .
the relation between the irradiation region and the central continuum is important , since the emission lines are thought to be mostly caused by photoionization .
fig.[fig : cw ] is plotted based on the analysis of h@xmath0 lines .
baldwin effect can be seen in the bottom plot .
the pearson correlation coefficient ( pcc ) is -0.39 .
we bootstrap it 100000 times and obtain 10 pcc larger than 0 , so pcc is smaller than 0 at 99.99% confidence level .
pearson rank correlation coefficient ( prcc ) is -0.48 , at 99.9% confidence level smaller than 0 using the same method as the confidence level calculation of pcc .
however no baldwin effect is seen in the top ( pearson correlation coefficient is -0.1 ) or in the middle plot ( pearson correlation coefficient is 0.1 ) .
it means that the equivalent width ( ew ) of the imgc becomes smaller when the agn becomes brighter , whereas the ew of the vbgc does not change with the continuum luminosity .
this phenomenon has also been seen by marziani et al .
since ew reflects the covering factor of the emission region , the above result supports the scenario that the vblr is nearly spherical , but the imlr has a flattened geometry .
therefore in the absence of other knowledge of other parameters of the gas , we can use the higher ew as a supportive ( albeit not conclusive ) evidence for a higher covering factor .
there is also evidence that the dust torus has a smaller covering factor with higher luminosity ( wang et al . 2005 ) .
we therefore suggest that the cause of the slight baldwin effect of imlr can be the same , i.e. , the innermost region of the dusty torus can be sublimated away by the strong irradiation of the agn
. therefore we can infer that imlr and vblr have different geometries .
based on the above analysis , a picture of imlr has emerged . in comparison with vblr ( the traditionally called
broad line region " with a near spherical structure in virialization equilibrium ) , it has a larger radius and higher density , contains more dust , is more flattened , and is thus consistent as being the inner boundary region of the dust torus of agn . as shown in fig.[fig : rr ] , the radii of both vblr and imlr are primarily determined by the continuum luminosity of the agn . @xmath32
obtained with equation ( 6 ) is used as the radius of vblr there .
then @xmath43 is derived from equation ( 3 ) as we have shown in fig.[fig : rr](a ) .
the stratified geometry of imlr , as inferred from fig.[cor_lines2 ] in section 4.2 , is also consistent with the above picture
. the receding velocity of torus with luminosity can be also obtained from the analysis of type i agn fraction . a relation between the covering factor @xmath44 and @xmath45
can be obtained from maiolino et al .
( 2007 ) : @xmath46 , consistent with wang et al ( 2005 ) .
if @xmath37 is used as the radius of the inner torus , then @xmath47 , and the height of the inner torus @xmath48 can be calculated from this relation , as shown in fig.[fig : rhh ] . clearly the relationship between @xmath48 and @xmath37 can be fitted with a power - law form .
since @xmath49 affects the index very weakly in this fitting , we just let @xmath50 in fig.[fig : rhh ] . the height of the inner torus increases when its radius increases with the relation @xmath51 , which controls the geometry of torus .
rm based on infrared emission is consistent with @xmath52 ( suganuma et al .
2006 ) , which agrees with the prediction of a toy model for the sublimation process ( krolik & begelman 1988 ) . however , suganuma et al . ( 2006 ) did not provide the index error of @xmath53 relation .
we refit the data with error and the best fitting is @xmath54 , statistically consistent with our relation between imlr radius and luminosity ( @xmath43 ) . however , the radius of imlr that we calculated here is strongly dependent on the radius of vblr , i.e. , dependent on the @xmath55 relation .
if equation ( 5 ) is used to calculate the radius of blr which is used as the radius of vblr here , we would obtain @xmath56 , which is also consistent with the infrared rm result of @xmath57 .
however , we prefer equation ( 6 ) because it has taken into account the correction of starlight ( bentz et al .
equation ( 6 ) is also consistent with a simple radiation pressured dominated vblr where @xmath58constant .
all of these analysis suggest that imlr has roughly the same receding velocity as torus .
combined with the analysis in the sections above , we conclude that imlr is the inner part of a dust torus .
based on the analysis above , a simple scenario of the blr ( blr = vblr+imlr ) evolution can be constructed as shown in fig.[fig : car ] .
the inner spherical region is the vblr .
it expands to a larger radius with luminosity increase and perhaps also black hole mass increase .
we suggest that the imlr is the inner part of the torus , which can be sublimated by the central radiation and thus its radius also increases with luminosity increase .
naturally the imlr will be photoionized by the irradiation of the central agn and the material gravitationally bound by the central black hole will be varialized with the gravitational potential of the central black hole , as determined by equations ( 1 ) and ( 2 ) . since the height of torus increases slower than the inner radius of the torus when luminosity increases , the covering factor of the imlr decreases with luminosity ( and larger radius ) as shown in the cartoon .
when the luminosity is high enough , the two regions may eventually merge into one .
the observed broad emission line is thus the superposition of the emission from the two regions .
when luminosity is large , the broad line region and torus become one single entity , but with different physical conditions .
this scenario is consistent with the dust bound blr hypothesis ( laor 2004 ; elitzur 2006 ) .
suganuma et al .
( 2006 ) also showed that the delay time of infrared emission is always longer than the delay time of emission lines of the corresponding agn , but sometimes they are very close to each other and become almost the same .
this is also consistent with the inner torus region as the origin of the imgc .
a piece of supportive evidence for the existence of imlr comes from the study of the micro - lensing of the broad line region in the lensed quasar j1131 - 1231 ( sluse et al .
2007 , 2008 ) . in this work they found evidence that the h@xmath0 emission line ( as well as h@xmath6 ) is differentially microlensed , with the broadest component ( fwhm @xmath59 4000 km / s ) being
much more micro - lensed than the narrower component ( fwhm @xmath59 2000 km / s ) .
the emission line can be well decomposed into two components and the emission line profile has been significantly changed after microlensed compared to the original line . because the amplitude of micro - lensing depends on the size of the emitting region , it can be naturally explained as the broadest component of the emission line comes from a more compact region than that of the narrower component .
although it can also be explained as a single region with a range of gas velocities at different radii , the explanation comes out from our model is natural .
then , we turn to the rm experiment .
if the imlr is really a physically distinct emission region apart from the vblr , another peak corresponding to the radius of this region might appear in the cross correlation function between the lightcurves of the emission line and the continuum .
in fact , a possible example exists in the database of rm observations ( peterson et al . 1998 ; 2004 ) .
mrk 79 shows obviously double peaks in the cross correlation centroid distribution map as shown in fig.[fig : mrk2 ] , which has not been explained well so far .
there are four subsets of rm data for mrk 79 which were taken in different time periods , and the double - peaked cross - correlation centroid distribution of mrk 79 is from the fourth period .
the measured time delays are about @xmath60 days in the subsets @xmath61 ( peterson et al .
2004 ) , while in the 4th subset there are two typical time delays , one is about @xmath62 days which is consistent with subsets @xmath61 , and the other is about @xmath63 days , as shown in fig.[fig : mrk2 ] . here
we suggest that the shorter time delay , i.e. @xmath62 days , is the delay time of vblr , which was also observed in subsets @xmath61
. the longer time delay , i.e. @xmath64 days , is possibly the delay time of imlr , which was only observed in the fourth subset .
the mean h@xmath0 spectrum of mrk 79 can be well described in a three gaussian component model ( peterson et al .
1999 ) , i.e. , a normal narrow line , an imgc and a vbgc , as shown in fig.[fig : mrk1 ] .
based on the simple relation in section 4.1 ( equation 3 ) , we get @xmath65 .
therefore the delay time for imlr should be 32 - 54 days , fully agrees with the second peak of cccd .
this agreement also suggests that our rather arbitrary choice of @xmath66 does not deviate from its true value significantly , at least not for mrk 79 .
however , it should be noted that in the mean h@xmath0 the vbgc dominates over the imgc , but in the cccd the second peak appears to be much stronger . probably , in the fourth subset , the imlr responses to the continuum more than the vblr , i.e. , the imgc is more variable . on the other hand , for the majority of other sources with rm observations ,
the imgc is not important .
we re - examined the ten rm sources with public data on peterson website .
six of them only need one gaussian component to describe the broad line of h@xmath0 , one of them has double peak profile and one has irregular profile , only two of them need vbgc+imgc to fit . in addition , 90% of the sources that plot on the r - l relation ( kaspi et al 2005 ) has luminosity larger than @xmath67 erg s@xmath2 and more than 60% larger than @xmath68 erg s@xmath2 , which prefers the emission line to be one single gaussian profile ( in our sample , there are 14 sources s h@xmath0 only need one gaussian component , 11 of them has luminosity larger than @xmath67 erg s@xmath2 and 6 of them larger than @xmath68 erg s@xmath2 ) .
therefore the radius - luminosity relationship established in emission line rm measurements may be valid only for the vblr , i.e. , the r - l relation only gives the radius of vblr , as assumed in section 4.1 .
this is probably why we did not see double peaks in most of the cccd maps ( as shown above , even for mrk 79 , the two peaks were obtained in only one observation period out of four periods in total . ) .
two of the sources with double gaussian line profiles are ngc 4051 and mrk 509 ; the lag times for the two components of both sources satisfy this simple relationship @xmath69 , consistent with our two - components blr assumption .
interestingly the lag time for the vbgc of ngc 4051 is between 1 to 2 days , significantly different from the lag time of about 4 days for the whole line .
this revised lag time makes ngc 4051 consistent with the radius - luminosity relationship in equation ( 5 ) or ( 6 ) , resolving the outstanding problem on the significant deviation of ngc 4051 from the radius - luminosity relationship ( kaspi et al .
2005 ) . the detailed results on our re - analysis of the rm data with this two - components model will be presented elsewhere ( zhu & zhang 2009 ) .
we propose that the vblr and imlr are identified with the variable regions that scales with luminosity .
if this is the case , one might expect that the vbgc would vary more than the mean spectrum as it has smaller radius .
however , that is not observed ; the rms spectrum of the emission line is normally ( but not always ) narrower systematically than the mean spectrum ( collin et al .
2006 ) . in our model ,
imlr has a flattened geometry , and our simulations show that it can create a narrower response function to a delta impulse even with radius much larger than the vblr .
we carried out a straightforward test by calculating the responses of the two components to the continuum . in this simple geometrical model ,
the vblr is a spherical shell and the imlr is a cylindrical shell ( representing a flattened disk - like geometry with inner and outer boundary ) .
the gas density inside each region is independent of radius , i.e. , the gas distribution could be clustered , but the distribution of gas clusters is independent of radius .
the thickness of the vblr shell is chosen to be three times that of the imlr , because the vblr is thought to have much lower density .
the emissivity law of the two emission line regions is chosen in such a simple way that the emission line flux is proportional to the density times the continuum flux received at any point in the two regions .
this means that the broad emission line intensity is proportional to the covering factor of each region .
the line profile at any radius is assumed to have a gaussian profile , with velocity determined from equation ( 3 ) .
the overall broad emission line from each region is thus the superposition of all broad lines from all radii .
the response functions with different combinations of parameters for the two shells and the inclination angle of imlr are shown in fig.[fig : p ] .
it can be seen that the imlr normally can produce a narrower response function when the inclination angle is not very large .
a narrower rms spectrum can be easily calculated based on the narrower response function of imlr . however , the fwhm difference in modeled rms and mean spectra is not as significant as the observed .
the simulation is carried out with noiseless , evenly sampled data , so further work is needed to see if this model can explain the narrower rms spectrum quite well . in any case , the narrower rms is not in conflict to our model of two distinct broad line regions , with the outer region has a flattened geometry .
the requirement for small inclination angles simply suggests that most of the reverberation mapped objects have small inclination angles , a generic property of agns with broad emission lines .
further more extensive calculations on the detailed responses of the two broad line regions to characteristic continuum light curves of agns and comparisons with data will be presented in a future work ( zhu & zhang 2009 ) .
rm - based black hole mass is calculated by the virial equation + @xmath70 where fwhm@xmath71 is the fwhm of the whole h@xmath72 , which represents the virial velocity of the blr . for an isotropic velocity distribution , as generally assumed , @xmath73 ( onken et al .
@xmath32 is the blr radius that can be calculated from equation ( 5 ) or ( 6 ) .
as assumed in section 4.1 and further discussed in section 5.2 , the radius obtained from rm may represent the radius of the vblr actually .
if we take this assumption , fwhm of the vbgc , instead of fwhm of the whole line , should be used to calculate the black hole mass .
we therefore correct the black hole mass in this way , + @xmath74 where fwhmb is the fwhm of the vbgc , and @xmath34 is taken as the @xmath32 in equation ( 5 ) or ( 6 ) .
it gives a more significant mass correction for nls1s than for bls1s as shown in fig.[fig : mbh](a ) .
the correction factor is near unity when fwhm reaches about 5000 km@xmath1s@xmath2 .
we use l5100 as an indicator for continuum luminosity , and a correction for accretion rate has also been shown in fig.[fig : mbh](c ) .
after such correction , nls1s still have smaller black hole masses but normal accretion rate in units of the eddington rate .
fifteen objects in our samples have velocity dispersion measurement data ( sigma ) in shen et al .
( 2008 ) , as shown in fig.[fig : sm ] , where comparisons are made between black hole masses measured here and that predicted by the currently used m - sigma relation @xmath75 ( tremaine et al .
it is obvious that the masses of all nsl1 ( filled symbols in the upper panel of fig.[fig : sm ] ) are well below , but become more very close to , the predictions of the m - sigma relation before and after the correction , respectively . for other agns
no significant changes to their masses are introduced by the correction process . as shown in the lower panel of fig.[fig : sm ] , after the correction , the median value of @xmath76 is much closer to zero , and the dispersion is reduced from 0.76 to 0.50 dex , which is consistent with the black hole mass uncertainty of 0.5 dex in rm ( peterson 2006 ) . the effect of the correction is obvious , albeit small number statistics due to the limited sample . it has been shown that the receding torus model with constant torus height fails to provide a good fit to the data of type ii agn fraction as a function luminosity , and a good fitting can be given when @xmath48 increases slowly with luminosity with the relation @xmath4 ( simpson 2005 ) , in excellent agreement with that of imlr as we have shown in section 4.4 .
note that here we consider the change of covering factor is totally because of the change of the opening angle during the calculation , following wang et al .
this agreement suggests that the receding of torus is sufficient to explain the decrease of covering factor with decreasing luminosity .
our model is consistent with the torus that simpson ( 2005 ) needed to explain under - populated luminous type ii agns .
we conclude that the decomposition of broad h@xmath0 , h@xmath6 , and h@xmath7 line of the agns confirms the two component model of blr which has been suggested by several previous studies ( e.g. , brotherton 1996 ; sulentic et al .
2000 ; hu et al . 2008 . ) .
we have made detailed analysis about the two emission regions ( vblr and imlr ) based on the balmer line decomposition , and find other supportive evidence for this model .
our main conclusions are : 1 .
the two gaussian components exhibit evolutions with increasing fwhm ( we note in passing that because the two components show quite different dependence with fwhm and luminosity as to be shown in the following , we rule out the possibility that the dependence is caused by systematic biases in the decomposition process ) .
the evolution is much stronger for the h@xmath0 and h@xmath7 lines .
our results offer strong evidence of the evolution of the broad line region ( consists of a imlr and a vblr ) evolution from nls1s to bls1s .
we obtain the luminosity dependence for the radius of imlr , @xmath77 , if the luminosity dependence for the vblr is taken as @xmath78 ( kaspi et al . 2005 ) .
the two emission regions have a trend to merge into one region with luminosity increasing .
balmer decrement and the baldwin effect in imlr indicate that it has a flattened geometry , higher density and contains more dust , compared to the vblr .
the receding velocity of imlr is consistent with dust torus .
therefore , we suggest the imlr is the hot inner skin of torus .
a cartoon of the evolution of blr emerge from these analysis .
there are other evidence in support of this two component model .
the study of micro - lensing provides possible evidence for the existence of imlr .
the double - peaked cccd from the rm data of mrk 79 also provides possible evidence .
simulations suggest that the narrower rms spectra of broad emission lines from many agns may be consistent with our model , although more work needs to be done to establish this as the case .
the existence of a weak imgc in the broad emission lines of many sources with rm measurements may cause systematic biases for the measurements of the radius of the vblr , thus biasing the radius - luminosity relation for the vblr .
it will be helpful to decompose each broad emission line into two gaussian components as we have done here , and then do cross - correlation analysis between the continuum and each of the two components , in order to measure the radius - luminosity relations for the two components independently .
3 . in our model ,
only the vbgc should be used to estimate the black hole mass , and the radius measured by reverberation mapping based on emission lines normally represents the radius of this region .
after correction for black hole masses , nls1s still have smaller black hole masses ( compared to bls1s ) but normal accretion rate in units of eddington rate .
therefore , the black hole mass increases from nls1s to bls1s by following the m - sigma relation established for normal galaxies .
+ we obtain the luminosity dependence for the height of imlr as @xmath79 .
it can well explain the luminosity function of agn , if the decreasing fraction of type ii agns for higher luminosity is due to completely the decreasing covering angle of the dusty torus to the central irradiation source ( simpson .
+ therefore both the problem of under - massive black hole in nls1s and the problem of under - populated luminous type ii seyfert galaxies can be understood properly if our model is true , still more concrete evidences for this model are needed .
hu et al . (
2008 ) found evidence that imlr is related to inflow towards the vblr . combining with this conclusion , we suggest that the inflow from the inner boundary region of agn s dusty torus may provide the supply to the accretion disk surrounding the central black hole
the strong and positive luminosity dependence of the geometry of imlr suggests that the dust sublimation by the central accreting black hole s radiation dominates the structure and evolution of imlr .
if imlr is related to inflow ( hu et al . 2008 ) , we may be able to further suggest that the inflow is caused by the dust sublimation , i.e. , a consequence of the feed - back of the black hole s accretion and radiation . because imlr is also ionized similarly to vblr , the imlr induced viscosity allows efficient angular momentum transfer to drive gas inflow from the dust torus and consequently fuel the accretion flow onto the central black hole . in this scenario ,
the accretion flow is self - regulated by the radiation from the accretion disk through irradiation to the dust torus .
therefore the growth of the supermassive black hole is at the expense of consuming the material in the dust torus during the agn phase ; this is consistent with the observation that for very low luminosity agns , the luminosity decreases with decreasing absorption column , i.e. , the agns in their last stages are running out accretion material supplied by the torus ( zhang & soria et al .
2009 ) . of course
not all material in the accretion flow falls into the black hole horizon to increase the black holea mass , since accretion winds and outflows are common in agns .
after the material in the dust torus is completely consumed , the agn phase will be turned off and the galaxy becomes a normal and inactive galaxy .
indeed many agns in low luminosity ( because of low accretion rate and low radiation efficiency ) show very little , or even no signs of torus and/or broad line region ; there is also no evidence for dust torus or broad line region in the centers of normal and inactive galaxies , including the milky way .
therefore the agn s dust torus is the missing link or bridge between the coeval growth of a black hole and its host galaxy .
this would require that during the merging of two galaxies , a dust torus is first formed , perhaps due to the residual orbital angular momentum of the two galaxies .
the dust torus then fuels the accretion and growth of the supermassive black hole through the self - regulation of irradiation to the dust torus by the accretion disk .
the initial trigger to this self - regulation process may be bondi accretion of gas with negligible angular momentum , or the low level agn activities of the two black holes in the two parent galaxies before the merger .
this evolutionary scenario is illustrated in fig.[fig : evolution ] .
the above scenario is generally consistent with that proposed by wang & zhang ( 2007 ) , but also with some important difference .
we suggest that the torus is formed by the merging of two galaxies and disappears after each agn cycle ; the different appearance ( mainly geometry ) of torus in different types of agns are mostly due to the self - regulation of the accretion and irradiation of the agn . therefore in our scenario ,
torus evolution is fast ( only lasting for one episode of agn activity ) and synchronized with that of the broad line region , whose evolution is also dominated by the luminosity of the agn . _
our torus evolution is mostly hierarchical .
_ for example , although nls1s also have generally smaller black hole masses , we do not find nls1s deviate from the m - sigma relation more than the bls1s after black hole mass correction made here .
therefore in our scenario nls1s are produced by mergers of smaller galaxies compared to bls1s ; nls1 may or may not show up as bls1s in the future , depending upon if more galaxy mergers grow them up in the future . in the scenario of wang & zhang ( 2005 ) , nls1s are in their early growth stage and will grow to become bls1 during this particular agn cycle . _
therefore their torus evolution is mostly secular . _ in practice , both hierarchical and secular evolutions should be needed for the black hole , torus and host galaxy .
it is natural that hierarchical evolution dominates at high redshifts where merger rate is very high , and secular evolution dominates at low redshifts .
we are extremely grateful to g. la mura and l.c .
popovic for sending us the processed spectra with both the narrow lines and continuum already removed , for the sample we used in this work .
snz thanks jianmin wang for many discussions , as well as sharing the early results of their work on the similar subject prior to its publication ( hu et al .
2008 ) , which motivated us to pursue the work presented here .
xuebing wu , chen hu and yuan liu are thanked for their comments and suggestions , when the initial results of this work was presented in a black hole workshop organized by feng yuan , xinwu cao and wenfei yu during april 26 - 28 , 2008 , in shanghai astronomical observatory .
yuan liu , jianmin wang , chen hu and xinlin zhou are also appreciated for proof reading and suggestions on the draft of this manuscript .
zhixing ling is acknowledged for helping us in calculating the responses functions presented in section 5.6 ( fig .
we appreciate very much the insightful comments and help suggestions by anonymous referee .
snz acknowledges partial funding support by the yangtze endowment from the ministry of education at tsinghua university , directional research project of the chinese academy of sciences under project no .
kjcx2-yw - t03 and by the national natural science foundation of china under grant nos .
10521001 , 10733010,10725313 , and by 973 program of china under grant 2009cb824800 .
barger , a. j. , cowie , l. l. , mushotzky , r. f. , yang , y. , wang , w .- h . , steffen , a. t. , & capak , p. 2005
, , 129 , 578 collin , s. , et al .
2006 , a&a , 456 , 75c greene , j. e. , & ho , l. c. 2006 , , 641 , l21 bian , w. , & zhao , y. 2004 , , 352 , 823 bentz , m.c .
2006 , apj , 644 , 133 binette , l. , et al .
1993 , apj , 414 , 535 bon , e. , popovi , l.c , et al .
2006 , newar , 50 , 716 brotherton , m. s. 1996 , , 102 , 1 collin , s. , kawaguchi , t. , peterson , b.m . , & vestergaard , m. 2006 , a&a , 456 , 75 corbett , e.a . , et al . 2003 , mnras , 343 , 705 decarli , r. , et al .
2008 , mnras , 386 , 15 elitzur , m. 2006 , in the central engine of active galactic nuclei , ed . l. c. ho & j .-
wang ( san francisco : asp ) , 3 elvis , m. , et al . , 1994 , apjs , 95 , 1 eracleous , m. & halpern , jules p. 2003
, apj , 599 , 886 ferrarese , l. , & ford , h. 2005 , space science reviews , 116 , 523 gebhardt , k. , et al.2000 , , 539 , l13 goncalves , a.c .
, et al . 1999 , a&a , 341 , 662 greene , j.e .
& ho , l.c .
2005 , apj , 630 , 122 grupe , d. , & mathur , s. 2004 , , 606 , l41 hasinger , g. 2004 , nuclear physics b proceedings supplements , 132 , 86 hasinger , g. 2005 , growing black holes : accretion in a cosmological context , 418 hu , c. et al .
2008 , apj , 683 , l115 joly , m. et al . 1985 ,
a&a , 152 , 282 kaspi , s. , et al .
2000 , apj , 533 , 631 kaspi , s. , et al .
2005 , apj , 629 , 61k kobayashi , y. , sato , s. , yamashita , t. , shiba , h. , takami , h. , 1993 , apj , 404 , 94 komossa , s. , & xu , d. 2007 , , 667 , l33 komossa , s. , et al .
2008 , apj , 680 , 926 komossa , s. 2008 , revista mexicana de astronomia y astrofisica conference series , 32 , 86 krolik , j.h .
, et al . 1978 , apjs , 37 , 459 krolik , j. h. & begelman , m. c. 1988 , apj , 329 , 702 la mura , g. , et al .
2007 , apj , 671 , 104 laor , a. 2004 , agn physics with the sloan digital sky survey , asp conference series , 311 , 169 lawrence , a. 1991 , mnras , 252 , 586 magorrian , j. , et al .
1998 , , 115 , 2285 maiolino , r. , et al .
2007 , a&a , 468 , 979 mason , k.o .
1996 , mnras , 283 , l26 marziani , p. , et al .
2009 , a&a , 495 , 83 m marconi , a. , & hunt , l. k.2003 , , 589 , l21 mullaney , j.r . & ward , m.j .
2008 , mnras , 385 , 53 nenkova , m. et al . 2008a , apj , 685 , 147 ( 2008arxiv:0806.0511 ) netzer , h. & laor , a. 1993 , apj , 404 , l51 onken , c.a . ,
2004 , apj , 615 , 645 osterbrock , d.e .
astrophysics of gaseous nebulae and active nuclei , 1989 university science books press , p 337 - 338 .
peterson , b.m . , et al .
1998 , apj , 501 , 82 peterson , b.m . , et al . 1999 , aspc , 175 , 41 peterson , b. m. , & wandel , a. 1999 , , 521 , l95 peterson , b.m . , et al .
2004 , apj , 613 , 682 peterson , b.m .
2006 , memorie della societa astronomica italiana , 77 , 581 peterson , b. m. 2007 , in the central engine of active galactic nuclei , ed .
l. c. ho & j .-
wang ( san francisco : asp ) , 3 popovi , l.c , et al .
2008 , rmxac , 32 , 99 simpson , c. 2005 , mnras , 360 , 565 sluse , d. , et al .
2007 , a&a , 468 , 885 sluse , d. , et al .
2008 , rmxac , 32 , 83 strateva , iskra v. et al .
2003 , aj , 126 , 1720 steffen , a. t. , barger , a. j. , cowie , l. l. , mushotzky , r. f. , & yang , y. 2003 , , 596 , l23 steffen , a. t. , barger , a. j. , capak , p. , cowie , l. l. , mushotzky , r. f. , & yang , y. 2004 , , 128 , 1483 suganuma , m. , et al .
2006 , apj , 639 , 46 sulentic , j. w. , marziani , p. , zwitter , t. , dultzin - hacyan , d. , & calvani , m. 2000 , , 545 , l15 shen , j.j .
, et al . 2008 , aj , 135 , 928 tremaine , s. , et al .
2002 , apj , 574 , 740 ueda , y. , akiyama , m. , ohta , k. , & miyaji , t. 2003 , apj , 598 , 886 urry , c.m .
, & padovani , p. 1995 .
pasp , 107 , 803 vestergaard , m. & peterson , b.m .
2006 , apj , 641 , 689 vron , p . ,
2002 , a&a , 384 , 826 wang , j - m . , et al .
2005 , apj , 627 , 5 wang , j .- m .
, & zhang , e .- p . 2007 , , 660 , 1072 wang , t. , & lu , y. 2001 , , 377 , 52 williams , r. j. , mathur , s. , & pogge , r. w. 2004 , apj , 610 , 737 zhou , h. , wang , t. , yuan , w. , lu , h. , dong , x. , wang , j. , & lu , y. 2006 , , 166 , 128 zhang , kai . ,
wang , tinggui .
, et al . 2009 , arxiv:0902.4390 . 1 & sdssj1152 - 0005 & 0.275 & 77.88 & 208.85 & 18.9 & 0 & 3610 & 3579 & 0 & 13.6 & 0 & 3640 & 3601 & 0 & 4.26 & 2.15 & 2529 & 3913 & 1493 + 2 & sdssj1157 - 0022 & 0.178 & 140.86 & 1241.78 & 75.2 & 58.5 & 4844 & 6926 & 3241 & 48.56 & 0 & 6602 & 6598 & 0 & 15.7 & 5.77 & 4689 & 6627 & 2301 + 3 & sdssj1307 - 0036 & 0.188 & 48.84 & 149.05 & 26.9 & 36.9 & 1873 & 3770 & 1285 & 11.3 & 8.45 & 2838 & 3793 & 1857 & 7.24 & 2.76 & 3023 & 4143 & 1597 + 4 & sdssj1059 - 0005 & 0.283 & 92.33 & 224.16 & 16.2 & 24.4 & 1965 & 3839 & 1368 & 8.17 & 5.66 & 3208 & 4109 & 2224 & 5.8 & 0 & 4010 & 4362 & 0 + 5 & sdssj1342 - 0053 & 0.129 & 23.61 & 293.32 & 14.3 & 27.8 & 2925 & 5767 & 2272 & 7.43 & 6 & 4195 & 6041 & 2716 & 3.43 & 3.39 & 3763 & 6445 & 2840 + 6 & sdssj1307 + 0107 & 0.26 & 183.28 & 1056.59 & 40.6 & 28.1 & 4433 & 6240 & 2864 & 16.4 & 8.38 & 4874 & 5784 & 3311 & 11.4 & 0 & 4380 & 4851 & 0 + 7 & sdssj1341 - 0053 & 0.17 & 32.02 & 97.72 & 22.9 & 57.5 & 2285 & 4490 & 1835 & 12.4 & 14.1 & 2653 & 4173 & 1869 & 6.15 & 8.04 & 2406 & 4791 & 1853 + 8 & sdssj1344 + 0005 & 0.276 & 113 & 1133.83 & 31.1 & 24.9 & 5027 & 6849 & 3541 & 18.8 & 0 & 6478 & 6396 & 0 & 7.38 & 0 & 4751 & 5269 & 0 + 9 & sdssj1013 - 0052 & 0.327 & 324.69 & 2732.11 & 38.1 & 0 & 4342 & 4257 & 0 & 23.1 & 0 & 4627 & 4623 & 0 & 11.1 & 0 & 4504 & 4937 & 0 + 10 & sdssj1010 + 0043 & 0.237 & ... & ... & ... & ... & ... & ... & ... & ... & ... & ... & ... & ... &
... & ... & ... & ... & ... + 11 & sdssj1057 - 0041 & 0.087 & 5.28 & 42.13 & 7.78 & 51.8 & 2148 & 4946 & 1921 & 8.43 & 17.9 & 2468 & 5342 & 1894 & 4.84 & 10.3 & 2406 & 4028 & 2211 + 12 & sdssj0117 + 0000 & 0.245 & 33.58 & 124.96 & 27.5 & 34.6 & 1828 & 3547 & 1185 & 13.1 & 8.99 & 2653 & 3801 & 1533 & 5.59 & 4.3 & 2591 & 4248 & 1853 + 13 & sdssj0112 + 0003 & 0.074 & 4.16 & 32.75 & 15.4 & 41 & 1873 & 4566 & 1487 & 7 & 10.2 & 2221 & 4318 & 1527 & 0 & 9.48 & 1789 & 0 & 1923 + 14 & sdssj1344 - 0015 & 0.141 & 30.36 & 78.27 & 24.5 & 68.4 & 1919 & 3881 & 1550 & 12.9 & 23.6 & 2097 & 3643 & 1598 & 7.85 & 10 & 2344 & 4208 & 1826 + 15 & sdssj1343 + 0004 & 0.114 & 25.02 & 57.24 & 27.5 & 71 & 1736 & 3639 & 1391 & 10.9 & 25.9 & 1974 & 3997 & 1559 & 7.58 & 10.8 & 1974 & 3059 & 1633 + 16 & sdssj1519 + 0016 & 0.233 & 100.02 & 628.19 & 35.6 & 16.8 & 3884 & 5251 & 2109 & 16.1 & 0 & 4874 & 4793 & 0 & 7.32 & 0 & 4010 & 4481 & 0 + 17 & sdssj1437 + 0007 & 0.179 & 74.76 & 146.48 & 21 & 63.6 & 1965 & 4012 & 1631 & 11.7 & 22.7 & 2406 & 4315 & 1877 & 8.27 & 2.21 & 3578 & 4377 & 2278 + 18 & sdssj1659 + 6202 & 0.31 & 210.04 & 752.96 & 16.9 & 36.1 & 3564 & 5479 & 2955 & 12.9 & 8.59 & 4319 & 5154 & 3303 & 4.88 & 10.9 & 2776 & 3989 & 2655 + 19 & sdssj0121 - 0102 & 0.36 & 439.86 & 724.5 & 18.5 & 50.4 & 2833 & 4770 & 2377 & 14.6 & 15.4 & 3393 & 4688 & 2599 & 26.4 & 0 & 4134 & 4506 & 0 + 20 & sdssj1719 + 5937 & 0.174 & 137.49 & 946.25 & 117 & 60.5 & 4936 & 6088 & 3232 & 56.2 & 0 & 5615 & 5550 & 0 & 12.2 & 0 & 3640 & 4077 & 0 + 21 & sdssj1717 + 5815 & 0.279 & 208.76 & 511.13 & 38.6 & 25.6 & 2925 & 3814 & 1902 & 14.4 & 14.5 & 3578 & 4109 & 3076 & 7.64 & 0 & 3023 & 3381 & 0 + 22 & sdssj0037 + 0008 & 0.362 & 277.83 & 369.99 & 33.2 & 55.2 & 1416 & 3424 & 948 & 14.7 & 6.71 & 2961 & 3801 & 1724 & 3.2 & 1.48 & 3332 & 4224 & 1708 + 23 & sdssj2351 - 0109 & 0.252 & 64.75 & 246.21 & 10 & 11.8 & 3107 & 4733 & 2283 & 5.98 & 3.64 & 3640 & 5137 & 1907 & 0 & 18.6 & 2159 & 0 & 2382 + 24 & sdssj2349 - 0036 & 0.046 & 2.13 & 35.98 & 27.2 & 72 & 2468 & 4222 & 2053 & 16.7 & 20.6 & 3023 & 4623 & 2236 & 4.84 & 5.69 & 3702 & 6560 & 2929 + 25 & sdssj0013 + 0052 & 0.239 & 128.04 & 639.96 & 16.4 & 34.8 & 3427 & 5892 & 2795 & 12.3 & 7.01 & 4195 & 5423 & 2505 & 13.7 & 17 & 2838 & 4258 & 2418 + 26 & sdssj1720 + 5540 & 0.055 & 19.96 & 175.4 & 64.5 & 78.7 & 4021 & 5098 & 3296 & 39.8 & 22.5 & 4134 & 5085 & 2790 & 7.96 & 3.14 & 2838 & 3607 & 1697 + 27 & sdssj0256 + 0113 & 0.081 & 6.76 & 104.17 & 26.2 & 22 & 3336 & 4251 & 2474 & 13.9 & 4.93 & 3887 & 4623 & 2188 & 23.2 & 0 & 5121 & 5676 & 0 + 28 & sdssj0135 - 0044 & 0.335 & 845.06 & 2960.01 & 101 & 19 & 4981 & 5451 & 2588 & 46.6 & 0 & 5861 & 5850 & 0 & 5.63 & 5.91 & 2714 & 4023 & 1963 + 29 & sdssj0140 - 0050 & 0.146 & 22.95 & 107.38 & 23.9 & 38.7 & 2148 & 4241 & 1577 & 13.5 & 13.6 & 2776 & 4623 & 1499 & 9.19 & 0 & 4380 & 4792 & 0 + 30 & sdssj0310 - 0049 & 0.206 & 57.01 & 471.05 & 24.9 & 45.7 & 3245 & 5174 & 2617 & 23.7 & 0 & 4997 & 4900 & 0 & 3.73 & 3.04 & 2776 & 3913 & 2123 + 31 & sdssj0304 + 0028 & 0.368 & 120.96 & 306.39 & 19.6 & 0 & 3153 & 3066 & 0 & 8.99 & 1.83 & 3455 & 3595 & 1980 & 4.37 & 4.78 & 3023 & 5507 & 2346 + 32 & sdssj0159 + 0105 & 0.198 & 37.36 & 188.85 & 12.4 & 41.2 & 2605 & 6198 & 2179 & 7.88 & 11.4 & 3270 & 6288 & 2305 & 7.99 & 0 & 3270 & 3568 & 0 + 33 & sdssj0233 - 0107 & 0.177 & 26.13 & 184.98 & 22.5 & 40.8 & 2605 & 4514 & 2046 & 11.3 & 7.13 & 3578 & 4472 & 2335 & ... & ... & ... & ... & ... + 34 & sdssj0250 + 0025 & 0.045 & 2.07 & 9.93 & 18.6 & 94.1 & 1188 & 3439 & 981 & 9.84 & 28.2 & 1542 & 3801 & 1214 & 0 & 16.6 & 1542 & 0 & 1634 + 35 & sdssj0409 - 0429 & 0.081 & 41.92 & 213.82 & 54.9 & 203 & 2650 & 5935 & 2254 & 28.9 & 58.7 & 3270 & 5576 & 2588 & 17.3 & 17.8 & 3270 & 5870 & 2463 + 36 & sdssj0937 + 0105 & 0.108 & 14.45 & 63.25 & 20.7 & 98.4 & 2102 & 4946 & 1813 & 11.7 & 32.8 & 2468 & 5342 & 1969 & 5.64 & 17.8 & 2406 & 4028 & 2345 + 37 & sdssj0323 + 0035 & 0.186 & 318.83 & 323.42 & 109 & 216 & 2239 & 4416 & 1710 & 54.9 & 84.1 & 2406 & 4841 & 1725 & 37.1 & 21.5 & 2714 & 4016 & 1617 + 38 & sdssj0107 + 1408 & 0.216 & 56.76 & 120.52 & 28.6 & 54.7 & 1416 & 3194 & 1043 & 13.9 & 9.74 & 2591 & 3493 & 1699 & 4.76 & 6.17 & 2468 & 4678 & 1944 + 39 & sdssj0142 + 0005 & 0.077 & 1.73 & 5.59 & 6.31 & 34.7 & 1051 & 3424 & 882 & 4.19 & 13.2 & 1234 & 3698 & 962 & 2.41 & 5.99 & 1419 & 3683 & 1251 + 40 & sdssj0306 + 0003 & 0.095 & 3.14 & 12.56 & 14.1 & 39.2 & 1096 & 3262 & 844 & 5.99 & 8.83 & 1666 & 3595 & 1097 & 1.79 & 3.88 & 2159 & 4028 & 1897 + 41 & sdssj0322 + 0055 & 0.09 & 6.44 & 33.31 & 18.9 & 56 & 1142 & 3442 & 917 & 4.86 & 10.9 & 2097 & 3801 & 1699 & ... & ... & ... & ... & ... + 42 & sdssj0150 + 1323 & 0.037 & 8.78 & 227.13 & 127 & 76.3 & 3473 & 5450 & 1703 & 45.07 & 0 & 5985 & 5949 & 0 & 19.7 & 0 & 4010 & 4488 & 0 + 43 & sdssj0855 + 5252 & 0.069 & 5.81 & 23.72 & 31.7 & 109 & 1051 & 2739 & 834 & 12.5 & 20.7 & 1789 & 3082 & 1363 & 0 & 15.1 & 1542 & 0 & 1591 + 44 & sdssj0904 + 5536 & 0.039 & 1.09 & 20.98 & 20.9 & 33.8 & 1828 & 3354 & 1389 & 9.34 & 7.07 & 2653 & 3801 & 1646 & ... & ... & ... & ... & ... + 45 & sdssj1355 + 6440 & 0.051 & 9.14 & 23 & 110 & 225 & 1325 & 2843 & 1010 & 54.6 & 87.1 & 1604 & 3063 & 1157 & 25.2 & 36 & 1666 & 2720 & 1405 + 46 & sdssj0351 - 0526 & 0.075 & 31.62 & 120.58 & 83.7 & 255 & 2422 & 4566 & 2022 & 54.3 & 113 & 2776 & 4620 & 2183 & 0 & 76.9 & 2529 & 0 & 2831 + 47 & sdssj1505 + 0342 & 0.058 & 7.98 & 18.08 & 23.2 & 98.9 & 1416 & 3887 & 1233 & 12.4 & 32.6 & 1604 & 4077 & 1254 & 8.47 & 14.3 & 2036 & 3309 & 1727 + 48 & sdssj1203 + 0229 & 0.093 & 35.99 & 171.89 & 56 & 124 & 2559 & 4718 & 2051 & 41.9 & 27.2 & 3332 & 4392 & 2216 & 10.6 & 21.6 & 2961 & 5409 & 2704 + 49 & sdssj1246 + 0222 & 0.078 & 45.44 & 170.3 & 41.3 & 148 & 2650 & 4870 & 2330 & 30.2 & 46.4 & 3023 & 4520 & 2348 & 16.4 & 27.3 & 2653 & 3875 & 2468 + 50 & sdssj0839 + 4847 & 0.024 & 7.86 & 279.34 & 289 & 113 & 3976 & 5228 & 1568 & 130 & 0 & 5491 & 5428 & 0 & 63.9 & 0 & 4874 & 5327 & 0 + 51 & sdssj0925 + 5335 & 0.087 & 5.44 & 17.86 & 7.9 & 36.8 & 1234 & 3757 & 1032 & 4.04 & 10.1 & 1666 & 4006 & 1272 & ... & ... & ... & ... & ... + 52 & sdssj1331 + 0131 & 0.048 & 11.05 & 26.69 & 41.9 & 172 & 1188 & 3294 & 987 & 25.6 & 61.7 & 1727 & 3698 & 1338 & 9.19 & 34.3 & 1480 & 4143 & 1312 + 53 & sdssj1042 + 0414 & 0.08 & 7.41 & 38.49 & 7.82 & 51.3 & 2193 & 4946 & 2013 & 6.93 & 13.3 & 2283 & 4520 & 1734 & 0 & 11.8 & 2283 & 0 & 2428 + 54 & sdssj1349 + 0204 & 0.033 & 2.48 & 45.98 & 111 & 220 & 2468 & 4908 & 1920 & 42.9 & 27 & 3455 & 5239 & 1558 & ... & ... & ... & ... & ... + 55 & sdssj1223 + 0240 & 0.072 & 8.15 & 59.21 & 35.8 & 27.9 & 1965 & 3044 & 1155 & 15 & 6.49 & 2776 & 3287 & 1777 & 11.5 & 0 & 2653 & 2959 & 0 + 56 & sdssj0755 + 3911 & 0.034 & 7.88 & 26.39 & 58.2 & 164 & 1462 & 3321 & 1172 & 34.3 & 51.2 & 1912 & 3647 & 1344 & ... & ... & ... & ... & ... + 57 & sdssj1141 + 0241 & 0.047 & 3.37 & 26.76 & 27.4 & 53 & 1599 & 3576 & 1159 & 8.6 & 13.1 & 1789 & 3801 & 1221 & ... & ... & ... & ... & ... + 58 & sdssj1122 + 0117 & 0.04 & 4.94 & 17.08 & 45.6 & 144 & 1371 & 3298 & 1079 & 17 & 40.1 & 1604 & 3698 & 1235 & ... & ... & ... & ... & ... + 59 & sdssj1243 + 0252 & 0.077 & 4.31 & 9.24 & 13.8 & 84.3 & 1096 & 3292 & 961 & 6.56 & 29.6 & 1295 & 3287 & 1035 & 0 & 15.2 & 1295 & 0 & 1383 + 60 & sdssj0832 + 4614 & 0.061 & 7.87 & 44.3 & 47.2 & 152 & 1965 & 5297 & 1624 & 24.3 & 44.9 & 2283 & 5342 & 1657 & 6.99 & 24.1 & 2776 & 6312 & 2588 + 61 & sdssj0840 + 0333 & 0.053 & 4.57 & 124.7 & 40.9 & 0 & 3884 & 3805 & 0 & 15.5 & 0 & 4627 & 4551 & 0 & 10.2 & 0 & 2653 & 2920 & 0 + 62 & sdssj1510 + 0058 & 0.036 & 21.63 & 365.27 & 212 & 276 & 3702 & 5935 & 2735 & 127 & 25.1 & 4874 & 5548 & 1735 & 31 & 48.1 & 3085 & 6675 & 2506 + 63 & sdssj0110 - 1008 & 0.078 & 53.19 & 187.64 & 234 & 164 & 2102 & 3372 & 997 & 99.1 & 38.8 & 3085 & 3698 & 1376 & 59 & 0 & 3023 & 3288 & 0 + 64 & sdssj0142 - 1008 & 0.031 & 20.11 & 266.31 & 416 & 256 & 3884 & 4946 & 2631 & 238 & 0 & 4380 & 4303 & 0 & 103 & 31.3 & 4134 & 5524 & 2405 + 65 & sdssj1519 + 5908 & 0.069 & 7.95 & 15.03 & 25.8 & 122 & 1279 & 3337 & 1086 & 14 & 51.4 & 1419 & 3698 & 1157 & 0 & 26.1 & 1542 & 0 & 1681 + 66 & sdssj0013 - 0951 & 0.074 & 3.86 & 8.08 & 9.59 & 53.5 & 1051 & 3216 & 899 & 4.61 & 18 & 1234 & 3493 & 1027 & ... & ... & ... & ... & ... + 67 & sdssj1535 + 5754 & 0.062 & 12.3 & 168.21 & 18.9 & 60.5 & 2970 & 5479 & 2569 & 14.3 & 17.4 & 3887 & 5959 & 2823 & 12.7 & 1.59 & 3085 & 3470 & 1732 + 68 & sdssj1654 + 3925 & 0.042 & 3.6 & 35.71 & 36 & 155 & 2148 & 4490 & 1834 & 19.7 & 33.2 & 2591 & 4109 & 2054 & 0 & 23.2 & 2406 & 0 & 2562 + 69 & sdssj0042 - 1049 & 0.058 & 3.89 & 33.23 & 20.4 & 69.7 & 1599 & 3757 & 1318 & 6.31 & 11.9 & 2283 & 4109 & 1729 & 6.51 & 2.04 & 2406 & 3377 & 650 + 70 & sdssj2058 - 0650 & 0.09 & 4.15 & 28.15 & 13.6 & 51.4 & 1736 & 4598 & 1486 & 8.96 & 14 & 2097 & 5034 & 1420 & 0 & 23.7 & 3393 & 0 & 3752 + 71 & sdssj1300 + 6139 & 0.052 & 9.25 & 214.11 & 67 & 49.3 & 3656 & 4948 & 2398 & 31.7 & 9.2 & 4504 & 5021 & 2657 & 1.43 & 4.69 & 1727 & 4834 & 1626 + 72 & sdssj0752 + 2617 & 0.095 & 6.39 & 33.3 & 7.76 & 24.8 & 1599 & 4367 & 1283 & 4.35 & 7.55 & 2159 & 4726 & 1549 & 5.18 & 28 & 1480 & 4143 & 1430 + 73 & sdssj1157 + 0412 & 0.082 & 14.61 & 28.9 & 53.9 & 163 & 1234 & 3325 & 982 & 25 & 55.9 & 1604 & 3647 & 1246 & 6.82 & 15.3 & 2221 & 3913 & 2064 + 74 & sdssj1139 + 5911 & 0.085 & 15.6 & 77.92 & 17.9 & 58.8 & 2330 & 4718 & 1969 & 9.91 & 26.1 & 2591 & 5137 & 2110 & ... & ... & ... & ... & ... + 75 & sdssj1345 - 0259 & 0.028 & 4.39 & 65.57 & 18.4 & 144 & 2742 & 5859 & 2515 & 26.8 & 44.6 & 3332 & 6061 & 2521 & 0 & 38.6 & 2653 & 0 & 2969 + 76 & sdssj1118 + 5803 & 0.061 & 25 & 275.2 & 334 & 215 & 3016 & 4362 & 1722 & 151 & 28.4 & 3887 & 4234 & 1706 & 88.4 & 0 & 3640 & 4036 & 0 + 77 & sdssj1105 + 0745 & 0.074 & 7.83 & 253.4 & 24.6 & 46.3 & 3702 & 7001 & 2847 & 13.5 & 9.4 & 5121 & 7397 & 3150 & 2.57 & 8.4 & 2838 & 5754 & 2710 + 78 & sdssj1623 + 4104 & 0.045 & 15.86 & 32.67 & 78.4 & 272 & 1645 & 3515 & 1388 & 56.9 & 126 & 1789 & 3801 & 1337 & 27.9 & 45.4 & 1851 & 3454 & 1571 + 79 & sdssj0830 + 3405 & 0.07 & 21.87 & 454.94 & 75.2 & 84.6 & 4570 & 5859 & 3671 & 53.2 & 0 & 5491 & 5424 & 0 & 30.8 & 0 & 4874 & 5439 & 0 + 80 & sdssj1619 + 4058 & 0.034 & ... & ... & ... & ... & ... & ... & ... & ... & ... & ... & ... & ... & ... & ... & ... & ... & ... + 81 & sdssj0857 + 0528 & 0.038 & 2.64 & 29.51 & 27.2 & 55.5 & 1828 & 3531 & 1394 & 8.26 & 14.7 & 2344 & 3801 & 1788 & 0 & 17.2 & 2036 & 0 & 2151 + 82 & sdssj1613 + 3717 & 0.059 & 9.27 & 227.9 & 40.1 & 55.1 & 3336 & 5935 & 2399 & 21.6 & 8.94 & 4997 & 6370 & 2468 & 4.97 & 12.7 & 2529 & 6330 & 2301 + 83 & sdssj1025 + 5140 & 0.062 & 9.42 & 34.11 & 47.3 & 96.9 & 1873 & 4617 & 1405 & 22.6 & 27.6 & 1912 & 4575 & 1194 & 5.56 & 16.2 & 2406 & 4258 & 2339 + 84 & sdssj1016 + 4210 & 0.055 & 14.11 & 41.73 & 94.5 & 267 & 1645 & 3576 & 1318 & 44.6 & 112 & 1851 & 3904 & 1463 & 24 & 47.6 & 1789 & 3452 & 1572 + 85 & sdssj1128 + 1023 & 0.051 & 10.94 & 16.71 & 99.5 & 248 & 1142 & 2601 & 880 & 39.7 & 64.7 & 1357 & 2783 & 974 & ... & ... & ... & ... & ... + 86 & sdssj1300 + 5641 & 0.072 & 13.98 & 54.01 & 54.5 & 83.1 & 1645 & 3151 & 1171 & 23.3 & 20.3 & 2221 & 3227 & 1494 & ... & ... & ... & ... & ... + 87 & sdssj1538 + 4440 & 0.041 & 3.79 & 66.77 & 15 & 41.8 & 2787 & 5859 & 2273 & 9.7 & 15 & 3763 & 5548 & 2938 & 0 & 11.8 & 2900 & 0 & 3165 + 88 & sdssj1342 + 5642 & 0.073 & 4.99 & 52.18 & 34.7 & 35.6 & 1736 & 3120 & 1074 & 13.9 & 8.59 & 2776 & 3390 & 2034 & 0 & 10.4 & 2036 & 0 & 2276 + 89 & sdssj1344 + 4416 & 0.055 & 8.3 & 17.68 & 41.7 & 113 & 1188 & 2961 & 964 & 26.1 & 49.7 & 1480 & 3185 & 1070 & 7.61 & 26.2 & 1419 & 3537 & 1300 + 90 & sdssj1554 + 3238 & 0.049 & 17.46 & 384.87 & 60.4 & 82.8 & 4387 & 5859 & 3594 & 38 & 0 & 5491 & 5397 & 0 & 0 & 19.3 & 2653 & 0 & 2836 + , h@xmath6 and h@xmath7 lines . with the fwhm increasing ,
fwhm ratio of vbgc to the whole line becomes smaller and finally reaches unity , the intensity ratio of the vbgc to the whole line becomes larger and finally also reaches unity , and the imgc becomes broader and weaker .
the uncertainty is roughly as large as the dispersion . ] , h@xmath0 , h@xmath7 of sdssj0250 + 0025 .
its h@xmath0 and h@xmath6 lines behave similarly as sdssj1344 + 4416 shown in fig.1(a1 ) and fig.1(a2 ) , respectively . the h@xmath7 line lost its vbgc , which can be caused by confusion in the continuum subtraction , when it is very weak and broad . ]
line increase with luminosity of the source .
lower panel is the correlation between fwhm of a single gaussian component ( vbgc / imgc ) and luminosity .
the sum squared error ( sse ) of the l5100-fwhm fitting in ( a ) is 2.34 ( 88 points ) , sse of the l5100-fwhmb fitting in ( b ) is 0.66 ( 88 points ) and sse of the l5100-fwhmi fitting is 1.07 ( 74 points ) .
the correlation of a single gaussian component fwhm and luminosity is tighter than that of the whole line .
the typical uncertainty is plotted in the corner ] line .
radii are plotted in logarithm scale .
circles are the vblr radius , `` @xmath80 '' represent imlr radii derived from h@xmath0 lines and `` @xmath81 '' are obtained from h@xmath6 lines .
circles filled with dots represent the objects missing the imgc .
least - absolute - residuals fitting results are also plotted : for imlr @xmath82 , and for vblr @xmath83 .
it is interesting to note that the radius of imlr is not correlated with the fwhm of the whole h@xmath0 line .
the typical uncertainty is plotted in the corner.,title="fig : " ] + represent imlr radius derived from h@xmath0 lines and @xmath81 from h@xmath6 lines .
circles filled with dots represent the objects missing the imgc .
the radius of the vblr is obtained from the @xmath84l5100 ( equation 3 ) relationship on the top figure .
these correlations supports a scenario of hierarchical evolution of agns .
the typical uncertainty is plotted in the corner.,title="fig : " ] + represent imlr radius derived from h@xmath0 lines and @xmath81 from h@xmath6 lines .
circles filled with dots represent the objects missing the imgc .
the radius of the vblr is obtained from the @xmath84l5100 ( equation 3 ) relationship on the top figure .
these correlations supports a scenario of hierarchical evolution of agns .
the typical uncertainty is plotted in the corner.,title="fig : " ] + and h@xmath0 ; _ lower panel _ : between the h@xmath7 and h@xmath0 .
the linear correlations between them indicates that the imgcs for all these three lines originate from physically connected regions , even if not exactly from the same region .
typical uncertainty is plotted as a cross.,title="fig : " ] + and h@xmath0 ; _ lower panel _ : between the h@xmath7 and h@xmath0 .
the linear correlations between them indicates that the imgcs for all these three lines originate from physically connected regions , even if not exactly from the same region .
typical uncertainty is plotted as a cross.,title="fig : " ] and h@xmath0 ; _ lower panel _ : between the h@xmath7 and h@xmath0 .
the fwhm of h@xmath6 and h@xmath7 is offset systematically by around -200 km / s and + 200 km / s around that of h@xmath0 , respectively .
this suggests a stratified geometry for the imgc , where h@xmath6 , h@xmath0 and h@xmath7 lines are produced at increasing radii respectively.,title="fig : " ] + and h@xmath0 ; _ lower panel _ : between the h@xmath7 and h@xmath0 .
the fwhm of h@xmath6 and h@xmath7 is offset systematically by around -200 km / s and + 200 km / s around that of h@xmath0 , respectively .
this suggests a stratified geometry for the imgc , where h@xmath6 , h@xmath0 and h@xmath7 lines are produced at increasing radii respectively.,title="fig : " ] obtained from @xmath37 based on h@xmath85 and h@xmath86 separately ; they have negligible differences .
the height of the inner torus ( imlr ) increases when its radius increases with an index smaller than unity.,title="fig : " ] + line .
the green line is the normal narrow line which we do not discuss here .
the yellow line represents the imgc with fwhm of 2522 km@xmath87 , and the blue line represents the vbgc with fwhm of 5856 km@xmath87.,title="fig : " ] + , y - axes are black hole mass , black hole mass correction factor and relative luminosity to @xmath88 ( eddington luminosity ) from top to bottom .
black hole masses ( in units of solar mass @xmath89 ) are plotted in logarithm scale .
circles are black hole mass and relative luminosity calculated with equation ( 1 ) .
dots are that after correction using equation ( 4 ) . clearly the correction is more effective for agns with smaller fwhm , i.e. , nls1s . after the correction , their luminosity ( in units of eddington ) appear to be in the same range , i.e. , nls1s do not show exceptionally higher luminosity .
typical uncertainty is plotted as a cross.,title="fig : " ] + ( tremaine et al . 2002 ) , for fifteen sources in our sample with dispersion measurements ( shen et al .
_ upper panel _
: circles represent those with black hole masses obtained with fwhm of the whole line used .
pentagrams represent those with corrected black hole mass .
filled symbols are nls1s . _
lower panel _ : histogram of the mass ratios before and after the correction . after the correction ,
the median value of @xmath76 is much closer to zero , and the dispersion is reduced from 0.76 to 0.50 dex.,title="fig : " ] + ( tremaine et al . 2002 ) , for fifteen sources in our sample with dispersion measurements ( shen et al .
_ upper panel _
: circles represent those with black hole masses obtained with fwhm of the whole line used .
pentagrams represent those with corrected black hole mass .
filled symbols are nls1s . _
lower panel _ : histogram of the mass ratios before and after the correction . after the correction ,
the median value of @xmath76 is much closer to zero , and the dispersion is reduced from 0.76 to 0.50 dex.,title="fig : " ] + | a two - components model for broad line region ( blr ) of active galactic nuclei ( agn ) has been suggested for many years but not widely accepted ( e.g. , hu et al . 2008 ; sulentic et al . 2000 ; brotherton et al .
1996 ; mason et al .
1996 ) .
this model indicates that the broad line can be described with superposition of two gaussian components ( very broad gaussian component ( vbgc ) and intermediate gaussian component ( imgc ) ) which are from two physically distinct regions ; i.e. , very broad line region ( vblr ) and intermediate line region ( imlr ) .
we select a sdss sample to further confirm this model and give detailed analysis to the geometry , density and evolution of these two regions .
micro - lensing result of blr in j1131 - 1231 and some unexplained phenomena in reverberation mapping ( rm ) experiment provide supportive evidence for this model .
our results indicate that the radius obtained from the emission line rm normally corresponds to the radius of the vblr , and the existence of the imgc may affect the measurement of the black hole masses in agns .
the deviation of nls1s from the m - sigma relation and the type ii agn fraction as a function of luminosity can be explained in this model in a coherent way .
the evolution of the two emission regions may be related to the evolutionary stages of the broad line regions of agns from nls1s to bls1s .
based on the results presented here , a unified picture of hierarchical evolution of black hole , dust torus and galaxy is proposed . |
optimally doped films of the high-@xmath0 superconductor yba@xmath3cu@xmath4o@xmath5 ( ybco ) grown by pulsed laser deposition ( pld ) show high critical currents that reach a substantial fraction of the theoretical maximum depairing current in applied magnetic field oriented in parallel to the @xmath6 axis@xcite .
evidence exists that this is primarily due to edge dislocations that thread the film along the @xmath6 axis in pld - ybco@xcite .
the critical current of ybco films in @xmath6-axis aligned applied magnetic field can also be enhanced substantially by the introduction of bazro@xmath4 ( bzo ) or of basno@xmath4 ( bso ) material that naturally form nano - rods along the @xmath6 axis during the film growth process@xcite@xcite@xcite .
such increases in the in - field performance of the critical current in ybco films by columnar microstructures are very important to the development of second - generation high-@xmath0 superconducting wire technology .
achieving a fundamental understanding of how they occur is therefore desirable . in this paper
, we shall compute the in - field critical due to natural or to artificial columnar microstructures in ybco thin films by employing two - dimensional ( 2d ) collective pinning theory@xcite@xcite and by performing direct computer simulations of the corresponding 2d vortex dynamics@xcite@xcite@xcite@xcite@xcite@xcite@xcite .
pinning of vortex lines will be assumed to be due to variations in @xmath0 about columnar defects that in turn are a result of the strain field about them@xcite .
heterogeneous columnar defects are thereby predicted that are composed of component lines of pins and of antipins of equal strength@xcite@xcite . these form a dipolar `` clover - leaf '' pattern in the case of an edge dislocation line defect ( see fig .
[ dislocation ] ) , while they form a quadropolar pattern in the case of a nano - column inclusion ( see fig .
[ nano_column ] ) . both 2d collective pinning theory@xcite@xcite and direct computer simulation of 2d vortex dynamics
obtain the following central result : that the critical current due to an arrangnement of weak heterogeneous columnar pins is equal to that expected from an arrangement of conventional columnar pins@xcite obtained by unbinding the component pin / antipin lines from the heterogeneous cluster , and by then converting the isolated antipin lines that result into pin lines of equal strength .
figure [ pin_landscapes ] depicts the two pinning landscapes in question .
the antipin components of heterogeneous columnar defects therefore increase the critical current by a factor of two compared to the contribution due to the true pin components alone in the collective pinning regime .
it is well known that the critical temperature in optimally doped ybco couples most strongly to pure shear strain in the @xmath1-@xmath2 plane@xcite . in particular ,
the variation in the critical temperature depends linearly on the symmetric strain tensor @xmath7 following @xmath8 with strain derivatives @xmath9 and @xmath10 .
below , we shall show how the shear strain about a line of edge dislocations or about a nano - column inclusion in a ybco thin film combined with this experimental fact results in a heterogeneous columnar pinning line for a vortex line in the mixed state .
consider a line of edge dislocations that threads a ybco film in parallel to the @xmath6 axis .
gurevich and pashitskii calculated the variation in the critical temperature expected from a linear dependence on strain ( [ tc ] ) and obtained the result@xcite @xmath11 with constants@xcite @xmath12 above , @xmath13 denotes the poisson ratio and @xmath2 denotes the magnitude of the burgers vector .
it can be aligned along either the @xmath1 ( @xmath14 ) or the @xmath2 ( @xmath15 ) axes .
we now note that the contribution of the vortex core to the vortex line tension is approximated by the fundamental energy scale per unit length @xmath16 , where @xmath17 denotes the london penetration depth , and where @xmath18 denotes the flux quantum in a type - ii superconductor .
the potential - energy landscape experienced by a vortex line then has a contribution @xmath19 due to the variation in the critical temperature about the dislocation line ( [ disc ] ) .
it is dominated by the shear component ( @xmath20 ) in the case of optimally doped ybco , which is shown in fig .
[ dislocation ] .
the potential energy landscape is essentially dipolar in such case , with two anti - pinning lobes ( + ) and two pinning lobes ( - ) .
an analogous heterogeneous pinning potential for a vortex line about a nano - column inclusion has been obtained more recently by one of the authors in collaboration with barnes and varanasi@xcite . the lattice mismatch between the inclusion and the ybco matrix results in pure shear strain about the inclusion .
these authors then ultimately find a variation of the critical temperature in the ybco matrix about the nano - column given by @xmath21 here @xmath22 is the mismatch between the radius of the inclusion , @xmath23 , and the radius of the corresponding void in the ybco matrix , @xmath24 , while @xmath25 and @xmath26 are the 2d compression and shear moduli , respectively .
the potential - energy landscape experienced by a vortex line again then has a contribution @xmath27 due to such a variation in the critical temperature about the nano - column inclusion , but with d - wave symmetry .
figure [ nano_column ] depicts the constant - energy contours .
notice that the potential energy is characterized by diametrically opposite lobes of pins and of antipin oriented with respect to each other by 90 degrees .
last , the force @xmath28 that a vortex core experiences due to changes in @xmath0 about both a line of edge dislocations ( [ disc ] ) and about a nanocolumn inclusion ( [ nano_rod ] ) is long - range .
suppose now that a random field of parallel lines of heterogeneous pins thread a ybco film along @xmath6 axis at transverse locations @xmath29 .
statistical cancellations in the net force experienced by a vortex core , @xmath30 , cut off the long - range force .
the effective finite range @xmath31 of each heterogeneous pin is determined by matching the fluctuation in the net force over the system with the long - range force due to a single material line defect : @xmath32 .
it is given by the expression @xmath33 in the case of strain - induced pinning of vortex lines about nano - column inclusions@xcite , where @xmath34 denotes their density .
above , we have shown how the strain field about edge dislocations and nano - column inclusions that thread a ybco film along the @xmath6 axis result in heterogeneous pinning lines for vortex lines that are present in the mixed phase .
as shown by figs .
( [ dislocation ] ) and ( [ nano_column ] ) , the heterogeneous columnar pins are clusters with equal numbers of pins and antipin , each of equal strength .
[ see also eqs .
( [ disc ] ) and ( [ nano_rod ] ) . ]
consider now a ybco film threaded by a field of such heterogeneous pinning lines in applied magnetic field aligned along the @xmath6 axis as well .
next , suppose that a rigid vortex lattice appears as a result . among the fraction of vortex lines that experience forces from the heterogeneous columnar pins , half will lie in valleys of the vortex - core potential while the other half will lie on the hills of the vortex - core potential
. because the magnitude of the force experienced by the vortex lines that lie in the valleys and on the hills of the vortex - core potential are equal , the critical current should be the same as that due to a field of columnar pins which results from ( _ i _ ) unbinding the cluster of pins and antipins from the heterogeneous pins , and ( _ ii _ ) converting each of the antipins that are now isolated into pins of equal strength and range .
figure [ pin_landscapes ] depicts the two columnar pinning arrangments in question for the case of nano - column inclusions .
below , we shall see that this prediction is indeed borne out by 2d collective pinning theory .
previous work by one of the authors and maley@xcite implies that many heterogeneous columnar defects collectively pin a rigid abrikosov vortex lattice .
theoretical arguments and numerical monte carlo simulations indicate , in particular , that a hexatic bose glass state can exist at low temperature in such a case@xcite@xcite . here
the vortex lattice is threaded by isolated lines of edge dislocations parallel to the lines of relatively weak pins / anti - pins .
( see fig .
[ critical_state_config ] . ) the critical current @xmath35 is then limited by plastic creep of the vortex lattice associated with glide by such edge dislocations@xcite .
application of 2d collective pinning theory@xcite yields a transverse larkin scale @xmath36 that is set by the density of larkin domains @xmath37 here @xmath38 denotes the density of vortex lines pinned by the material line defects and @xmath39 denotes the maximum force per unit length at a pin / antipin that makes up the heterogeneous material line defect .
also , @xmath40 denotes the shear modulus of the vortex lattice , while @xmath41 is the magnitude of burgers vectors that glide .
it can be shown that @xmath36 is of order the separation between isolated lines of dislocations in the hexatic bose glass@xcite . a statistical summation of the pinning forces inside of a corresponding larkin volume , in which the vortex lattice is perfectly rigid
, then yields a net pinning force density @xmath42 the above is valid if many vortex lines experience pinning forces inside of a larkin volume : @xmath43 .
last , study of equations ( [ disc ] ) and ( [ nano_rod ] ) shows that the potential energy about a heterogeneous pinning line experienced by a vortex line has zero angle average .
this implies that the occupation of a heterogeneous pin by a vortex line is purely random .
the density of vortex lines that they collectively pin is then equal to @xmath44 , where @xmath34 denotes the density of heterogeous columnar defects , where @xmath45 is the effective crossectional area of a heterogeneous pinning / antipinning line , and where @xmath46 denotes the density of vortex lines .
observe now that both @xmath39 and the product @xmath47 remain constant after unbinding the pins and anti - pins that make up heterogeneous columnar defects , and after then converting the isolated antipins that result into isolated true pins ! we conclude that 2d collective pinning theory@xcite@xcite implies that the critical current due to comparable arrangements of conventional columnar pins and of heterogeneous columnar pins like those shown in fig . [ pin_landscapes ] are the same .
we shall now compute the in - field critical current due to columnar pins instead by simulating over - damped vortex dynamics with the langevin equation .
our aim is to once again compare the critical - state performance of type - ii superconductors threaded by conventional columnar pins@xcite with that of type - ii superconductors threaded by heterogeneous ( d - wave ) pinning lines that can result from the strain field about nano - column inclusions in ybco@xcite .
the abrikosov vortex lattice in a field of parallel columnar pins is in a bose glass state characterized by an infinite macroscopic tilt modulus@xcite .
the longitudinal larkin scale @xmath48 is then infinite@xcite , which means that vortex lines may be considered as rigid rods at long longituginal length scales . in a film geometry that is perpendicular the applied magnetic field , their dynamics can therefore be described by the over - damped 2d langevin equations for the transverse motion of a given rigid vortex line at location @xmath49 @xcite@xcite@xcite@xcite@xcite@xcite@xcite : @xmath50 it is governed by the bardeen - stephen viscosity @xmath51 for a vortex line .
each rigid vortex line experiences forces @xmath52 and @xmath53 due to interactions with other vortices at locations @xmath54 and due to the field of material line defects , respectively .
the former is taken to be long range : @xmath55\over{4 - 2\ , { \rm cos}(q_x a ) - 2\ , { \rm cos}(q_y a)}}. \label{interaction}\ ] ] note that periodic boundary conditions are imposed over a rectangle of dimensions @xmath56 . the lattice constant @xmath1 of the grid is understood to be of order the diameter of a vortex core . above , @xmath57 denotes the thickness of the film .
values for both potential energies @xmath58 and @xmath59 that span a @xmath60 grid are stored in look - up tables .
the corresponding forces @xmath61 and @xmath62 are computed from these potentials by taking finite differences over the grid .
they are also stored in look - up tables .
last , a force at an arbitrary location is determined by linear interpolation from those determined off the grid .
in the special case of a featureless ( s - wave ) line pin that lies at a grid point @xmath63 , the above linear interpolation scheme yields a pinning force of the form @xmath64 where @xmath65 is the pinning force along a principal axis of the grid that contains the pinning center .
both it and its corresponding 1d potential energy @xmath66 are shown in fig .
[ conservative ] .
the scale @xmath67 that appears there denotes the magnitude of the maximum pinning force , which is related to the depth @xmath68 of the line pin potential by @xmath69 .
the transverse interpolation of forces is set by the function @xmath70 , which is also shown in fig .
[ conservative ] .
it is given explicitly by @xmath71 if @xmath72 , and by @xmath73 otherwise .
notice now that @xmath74 , which means that the above force field for a line pin is not conservative . on the other hand , leaving the 1d function @xmath70 free while imposing a conservative force implies that it must be given by @xmath75 instead . as shown by fig .
[ conservative ] , both functions have qualitatively similar shapes .
they both also have the same area , @xmath76 .
we believe , therefore , that the model for a linear pin displayed by the equations above is qualitatively correct .
the rigid vortex lines also experience a uniform lorentz force @xmath77 that push them across the width of the wire , in addition to a random brownian force @xmath78 that brings the vortex lattice into thermal equilibrium .
we numerically evolved the above langevin equation ( [ langevin ] ) in order to describe the over - damped dynamics of @xmath79 rigid vortex lines .
the second - order runge - kutta method was employed .
values for the interaction forces and for the pinning forces to be stored in look - up tables were determined by taking finite differences over a @xmath80 square grid that spans the system , and which has a lattice constant @xmath1 that is understood to be of order the superconducting coherence length .
the computer simulations were conducted in two stages .
we first annealed the vortex lattice down to low temperature in the presence of a landscape of weak columnar pins .
figure [ pin_landscapes ] depicts the two pinning ( anti - pinning ) landscapes that were studied .
all ( component ) line pins had equal strengths , each with a maximum pinning / antipinning energy @xmath81 equal to 1% of the dominant interaction scale @xmath82 .
the initial temperature was taken to be very many times the halperin - nelson - young melting temperature of the 2d vortex lattice , @xmath83 , and it was successively halved until a very small fraction of @xmath83 was reached .
two million time steps were taken at each temperature , with a time interval @xmath84 equal to a few percent of the natural time scale @xmath85 that is set by the dominant interaction energy among vortex lines , eq .
( [ interaction ] ) . here
@xmath86 is the average separation between vortex lines .
the annealed vortex lattice was found to be in a hexatic vortex glass state characterized by long - range orientational order and unbound dislocations@xcite@xcite@xcite . only a fraction of the vortices in the lattice experience pinning forces . in the second stage
, we drove the annealed vortex lattice at zero temperature with a uniform lorentz force in the presence of the same landscape of columnar pins .
the driving force increased in 300 steps from zero to about twice the critical force ( current ) .
each step of constant lorentz force lasted @xmath87 time iterations , @xmath84 , and the lorentz force was increased at each step by a part in @xmath88 of the interaction scale @xmath89 .
figure [ jc ] shows the current - voltage properties that result from driving the vortex lattice at zero - temperature through the pinning landscapes that are shown in fig .
[ pin_landscapes ] .
the landscape in the left panel is the field of conventional ( s - wave ) columnar pins that is shown in the left panel of fig .
[ pin_landscapes ] .
all of the pins in this case are featureless and of equal strength , with a number that matches the number of vortex lines ( 896 ) . as depicted by fig .
[ conservative ] , each pin lies at a grid point and inherits the minimum range of the grid .
notice the critical current that separates a pinned vortex lattice at low currents from a vortex lattice that experiences flux creep and flux flow at high currents .
it is interesting to note that although the vortex lattice was threaded by isolated dislocation defects throughout the simulation , these became well seperated precisely at the critical state .
( see figs .
[ critical_state_config ] and [ jc ] . )
the dislocation defects then gradually began to pair up and annihilate with increasing driving force in the flux - creep / flux - flow regime .
the right panel of fig .
[ jc ] displays the current - voltage property for a driven vortex lattice that experiences the arrangement of quadrapolar columnar pins that is shown in the right panel of fig .
[ pin_landscapes ] .
the latter aims to model the strain - induced pinning potential shown in fig .
[ nano_column ] that is predicted to exists about nano - column inclusions in ybco films@xcite .
this pinning landscape contains half the number ( @xmath90 ) of true line pins , by comparison with the previous case , with an equal number ( @xmath90 ) of antipin lines .
each component line of pins / antipins has the same range and strength as in the previous case , on the other hand .
figure [ jc ] indicates that the critical current in the present case of heterogeneous columnar pins is comparable to that obtained in the previous case of featureless columnar pins , which had _ twice _ the number of true pins !
this result is nevertheless anticipated by 2d collective pinning theory , eq .
( [ critical_state ] ) .
it is due to the fact that the pin and the antipin components of the heterogeneous pinning line contribute equally to the critical current of a rigid vortex lattice .
last , two well - spaced dislocation defects with equal and opposite burgers vectors threaded the vortex lattice throughout the second simulation .
in summary , strain about columnar microstructure in ybco films generates heterogeneous pin / antipin lines that can multiply the critical current expected from the true pin line components alone by a factor of two .
this result is obtained for weak correlated pinning inside of the collective pinning regime , however .
it is unlikely to persist in the limit of strong correlated heterogeous pins , where all vortex lines that experience pinning forces then lie along a true pin component of the composite line defect .
antipin line components of the heterogenous pin have no effect on the critical current in that limit .
a key assumption of the collective pinning analysis is that the critical state be found in a hexatic bose glass state characterized by isolated dislocation defects that thread the vortex lattice@xcite .
figure [ critical_state_config ] demonstrates that this assumption is indeed confirmed by the molecular dynamics simulations that we conducted across the critical state for 896 rigid vortex lines .
previous 2d simulations find evidence for the break - up of the vortex lattice into grains at much higher numbers of rigid vortex lines when the latter experience short - range interactions@xcite@xcite .
thermodynamic monte carlo simulations on over 2000 vortices that experience the same long - range logarithmic interaction studied here , eq .
( [ interaction ] ) , find evidence for an equilibrium hexatic vortex glass phase@xcite , on the other hand .
it therefore remains to be seen if the hexatic glass phase obtained in the present molecular - dynamics simulations persists in the thermodynamic limit .
a. goyal , s. kang , k.j .
leonard , p.m. martin , a.a .
gapud , m. varela , m. paranthaman , a.o .
ijaduola , e.d .
specht , j.r .
thompson , d.k .
christen , s.j .
pennycook and f.a list , supercond .
* 18 * , 1533 ( 2005 ) . ) .
all forces and velocities are given in units of @xmath89 and of @xmath96 , respectively .
the right panel above depicts the current voltage curve obtained from driving 896 rigid vortex lines in the presence of 224 quadrapolar columnar pins .
each component pin / antipin line has a strength and range equal to that of the conventional s - wave columnar pins .
( see fig .
[ pin_landscapes ] . ) ] | theoretical work predicts that the strong dependence of @xmath0 on pure shear strain within the @xmath1-@xmath2 plane of optimally doped yba@xmath3cu@xmath4o@xmath5 results in heterogenous columnar pins of vortex lines about dislocation lines and about nano - columns inclusions aligned in parallel to the @xmath6 axis .
the critical current of a rigid vortex lattice driven by the lorentz force in the presence of such clusters of pin / antipin lines is computed using two - dimensional ( 2d ) collective pinning theory and by numerical simulation of the corresponding 2d vortex dynamics . both theory and computer calculation
find that the antipin component of the heterogenous columnar pins contributes substantially to the net in - field critical current . in - field
critical current , nanorod inclusions , strain field |
it is well known that tilting theory is a theoretical basis in the representation theory of finite - dimensional algebras , in which the notion of tilting modules is fundamental . moreover , in the representation theory of algebras the notion of mutation " often plays an important role .
mutation is an operation for a certain class of objects in a fixed category to construct a new object from a given one by replacing a summand . in [ hu2 ] ,
happel and unger gave some necessary and sufficient conditions under which mutation of tilting modules is possible ; however , mutation of tilting modules is not always possible . in [ air ] , adachi , iyama and reiten introduced the notion of support @xmath0-tilting modules which generalizes that of tilting modules , and showed that mutation of support @xmath0-tilting modules is always possible .
this is a big advantage of support @xmath0-tilting mutation " which tilting mutation " does not have . note that the @xmath0-tilting theory developed in [ air ] has stimulated several investigations ; in particular , there is a close relation between support @xmath0-tilting modules and some other important notions in the representation theory of algebras , such as torsion classes , silting complexes , cluster - tilting objects , grothendieck groups and @xmath7-modules , see [ air ] , [ aii ] , [ bdp ] , [ ijy ] , [ ir ] , [ j ] , [ w ] , and so on .
moreover , adachi gave in [ a ] a classification of @xmath0-tilting modules over nakayama algebras and an algorithm to construct the exchange quiver of support @xmath0-tilting modules .
zhang studied in [ z ] @xmath0-rigid modules which are direct summands of support @xmath0-tilting modules over algebras with radical square zero . on the other hand , the notion of skew group algebras was introduced in [ rr ] .
let @xmath1 be a finite - dimensional algebra and @xmath2 a finite group such that its order @xmath8 is invertible in @xmath1 acting on @xmath1 .
the algebra @xmath1 and the skew group algebra @xmath6 have a lot of properties in common .
the aim of this paper is to introduce and study @xmath2-stable support @xmath0-tilting modules and moreover to establish bijections among them and @xmath2-stable two - term silting complexes , @xmath2-stable functorially finite torsion classes , and @xmath2-stable cluster - tilting objects .
moreover , we investigate the relationship between stable support @xmath0-tilting modules over @xmath1 and the skew group algebra @xmath6 .
this paper is organized as follows . in section 2 , we give some terminology and some known results . in section 3 , we prove the following theorem .
( [ theorems 3.4 , 3.7 , 3.13 and 3.15])[t : integer2 ] let @xmath1 be a finite - dimensional algebra and @xmath2 a finite group acting on @xmath1
. then there exist bijections among * the set @xmath2-@xmath9-@xmath10 of isomorphism classes of basic @xmath2-stable support @xmath0-tilting modules in @xmath11 ; * the set @xmath2-@xmath12-@xmath13 of isomorphism classes of basic @xmath2-stable two - term silting complexes in @xmath14 ; * the set @xmath2-@xmath15-@xmath16 of @xmath2-stable functorially finite torsion classes in @xmath11 .
furthermore , if @xmath17 where @xmath5 is a @xmath4-finite 2-calabi - yau triangulated category with a @xmath2-action and @xmath3 is a @xmath2-stable cluster - tilting object , then there exists also a bijection between the following set and any one of the above sets . * the set @xmath2-@xmath18-@xmath19 of isomorphism classes of basic @xmath2-stable cluster - tilting objects in @xmath5
. let @xmath1 be a finite - dimensional self - injective algebra and @xmath20 the subgroup of the automorphism group of @xmath1 generated by the nakayama automorphism . then @xmath21-stable support @xmath0-tilting modules introduced by mizuno in [ m ] are exactly @xmath2-stable support @xmath0-tilting modules in our sense .
so theorem 1.1 is a generalization of [ m , theorem 1.1 ] . in section 4
, we investigate the relationship between @xmath2-stable support @xmath0-tilting @xmath1-modules and @xmath22-stable support @xmath0-tilting @xmath6-modules , where @xmath22 , the group of characters of @xmath2 , naturally acts on @xmath6 .
we have the following ( [ theorems 4.2(3 ) and 4.6])[t : integer2 ] let @xmath1 be a finite - dimensional algebra and @xmath2 a finite group acting on @xmath1 such that @xmath8 is invertible in @xmath1 . then the functor @xmath23 preserves stability and induces the following injection : @xmath2-@xmath9-@xmath24-@xmath9-@xmath25 via @xmath26 .
moreover , if @xmath2 is solvable , then this map is a bijection .
finally we give an example to illustrate this theorem .
in this section , we give some terminology and some known results .
let @xmath27 be an algebraically closed field and we denote by @xmath28 . by an algebra @xmath1 , we mean a finite - dimensional algebra over @xmath27 .
we denote by @xmath11 the category of finitely generated left @xmath1-modules , by @xmath29 and @xmath30 the subcategories of @xmath11 consisting of projective modules and injective modules respectively , and by @xmath0 the auslander - reiten translation of @xmath1 .
we denote by @xmath31 the homotopy category of bounded complexes of @xmath29 . for @xmath32 ,
we denote by @xmath33 the subcategory of @xmath11 consisting of all direct summands of finite direct sums of copies of @xmath34 , and by @xmath35 the subcategory of @xmath11 consisting of all factor modules of finite direct sums of copies of @xmath34 . *
2.1 . @xmath0-tilting theory
* first we recall the definition of support @xmath0-tilting modules from [ air ] .
let @xmath36 be a pair with @xmath37 and @xmath38 .
* we call @xmath34 in @xmath11 _ @xmath0-rigid _ if @xmath39 .
we call @xmath36 a _
@xmath0-rigid pair _
if @xmath34 is @xmath0-rigid and @xmath40 .
* we call @xmath34 in @xmath11 _ @xmath0-tilting _ ( respectively , _
almost complete @xmath0-tilting _ ) if @xmath34 is @xmath0-rigid and @xmath41 ( respectively , @xmath42 ) , where @xmath43 denotes the number of non - isomorphic indecomposable direct summands of @xmath34 .
* we call @xmath34 in @xmath11 _ support @xmath0-tilting _ if there exists an idempotent @xmath44 of @xmath1 such that @xmath34 is a @xmath0-tilting ( @xmath45)-module .
we call @xmath36 a _ support @xmath0-tilting pair _
( respectively , _ almost complete support @xmath0-tilting pair _ ) if @xmath36 is @xmath0-rigid and @xmath46 ( respectively , @xmath47 ) .
we say that @xmath36 is _ basic _ if @xmath34 and @xmath48 are basic .
moreover , @xmath34 determines @xmath48 uniquely up to isomorphism .
we denote by @xmath49@xmath0-@xmath10 the set of isomorphism classes of basic support @xmath0-tilting @xmath1-modules .
( [ air , proposition 2.3 ] ) let @xmath37 and @xmath50 , and let @xmath44 be an idempotent of @xmath1 such that @xmath51 .
* @xmath36 is a @xmath0-rigid pair for @xmath1 if and only if @xmath34 is a @xmath0-rigid @xmath52-module . * if both @xmath36 and @xmath53 are support @xmath0-tilting pairs for @xmath1 , then @xmath54 .
in other words , @xmath34 determines @xmath48 and e uniquely up to equivalence .
let @xmath55 be a full subcategory of @xmath11 .
assume that @xmath56 and @xmath57 .
the morphism @xmath58 is called a _ left @xmath55-approximation _ of @xmath59 if @xmath60 is exact for any @xmath61 . the subcategory @xmath55 is called _ covariantly finite _ in @xmath11 if every module in @xmath11 has a left @xmath55-approximation . the notions of _
right @xmath55-approximations _ and _ contravariantly finite subcategories _ of @xmath11 are defined dually .
the subcategory @xmath55 is called _ functorially finite _ in @xmath11 if it is both covariantly finite and contravariantly finite in @xmath11 ( [ ar ] ) .
recall that @xmath62 is called _ partial tilting _ if the projective dimension of @xmath3 is at most one and @xmath63 .
a partial tilting module is called _ tilting _ if there exists an exact sequence @xmath64 in @xmath11 with @xmath65 , @xmath66 ( see [ hu1 ] and [ b ] ) .
we have that @xmath67 for any tilting module @xmath3 by [ b , theorem 2.1 ] .
the following result gives a similar criterion for a @xmath0-rigid @xmath1-module to be support @xmath0-tilting .
( [ j , proposition 2.14 ] ) let @xmath68 be a @xmath0-rigid @xmath1-module .
then @xmath68 is a support @xmath0-tilting @xmath1-module if and only if there exists an exact sequence @xmath69 in @xmath11 with @xmath70 and @xmath15 a left @xmath71-approximation of @xmath1 .
functorially finite torsion classes * let @xmath55 be a full subcategory of @xmath11 . recall that @xmath55 is called a _ torsion class
_ if it is closed under factor modules and extensions .
we denote by @xmath15-@xmath16 the set of functorially finite torsion classes in @xmath11 .
we say that @xmath72 is @xmath73-_projective _ if @xmath74=0 .
we denote by @xmath75 ) the direct sum of one copy of each of the indecomposable @xmath73-projective objects in @xmath55 up to isomorphism .
we have that @xmath76 if @xmath77-@xmath16 ( [ as , corollary 4.4 ] ) .
the following result establishes a relation between @xmath9-@xmath10 and @xmath15-@xmath16 .
( [ air , theorem 2.7 ] ) there exists a bijection : @xmath9-@xmath78-@xmath16 given by @xmath9-@xmath79-@xmath16 and @xmath15-@xmath80-@xmath10 . *
silting complexes * recall from [ aii ] that @xmath81 is called _ silting _
if @xmath82)=0 $ ] for any @xmath83 and @xmath14 is the smallest full subcategory of @xmath14 containing @xmath48 and is closed under cones , @xmath84 $ ] and direct summands ; and a complex @xmath85 in @xmath14 is called _ two - term _ if @xmath86 for all @xmath87 .
we denote by @xmath12-@xmath13 the set of isomorphism classes of basic two - term silting complexes in @xmath14 .
the following result establishes a relation between @xmath12-@xmath13 and @xmath9-@xmath10 .
( [ air , theorem 3.2 ] ) there exists a bijection : @xmath12-@xmath88-@xmath10 given by @xmath12-@xmath89-@xmath10 and @xmath9-@xmath90-@xmath13 , where @xmath91 is a minimal projective presentation of @xmath68 . *
cluster tilting objects * let @xmath5 be a @xmath27-linear @xmath4-finite krull schmidt triangulated category
. assume that @xmath5 is a 2-calabi - yau triangulated category , that is , there exists a functorial isomorphism : @xmath92 an important class of objects in such categories is that of cluster - tilting objects .
following [ bmrrt ] , an object @xmath93 is called _ cluster - tilting _ if @xmath94)=0\}$ ] .
we denote by @xmath18-@xmath19 the set of isomorphism classes of basic cluster - tilting objects in @xmath5 .
assume that @xmath5 has a cluster - tilting object @xmath3 and @xmath95 .
for @xmath96 , we have a triangle @xmath97 , \eqno{(\ast)}\ ] ] where @xmath98 and @xmath15 is a minimal right @xmath99-approximation .
we have the following results , which will be used frequently in this paper .
( [ bmr , theorem 2.2 ] and [ kr , p.126 ] ) there exists an equivalence of categories @xmath100\rightarrow { { \operatorname{\mathsf{mod}}}\nolimits}\lambda$ ] , where @xmath101 $ ] is the ideal of @xmath5 consisting of morphisms which factor through @xmath102 $ ] .
( [ air , theorem 4.1 ] ) there exists a bijection : @xmath18-@xmath103-@xmath10 given by @xmath18-@xmath104)})\in s\tau$]-@xmath10 , where @xmath105 is a maximal direct summand of @xmath34 belonging to @xmath102 $ ] .
( [ air , theorem 4.7 ] ) there exists a bijection : @xmath18-@xmath1062@xmath107-@xmath13 given by @xmath18-@xmath1082@xmath107-@xmath13 , where g is the morphism in @xmath109 . *
skew group algebras * in this subsection , we recall the definition of skew group algebras and some useful results from [ rr ] .
let @xmath1 be an algebra and @xmath2 be a group with identity 1 .
consider an action of @xmath2 on @xmath1 , that is a map @xmath110 via @xmath111 such that * for any @xmath112 in @xmath2 , the map @xmath113 is an algebra automorphism .
* @xmath114 for any @xmath115 and @xmath116 .
* 1@xmath117 for any @xmath118 .
let @xmath2 be a finite group .
for any @xmath119 and @xmath120 , let @xmath121 be a @xmath1-module as follows : as a @xmath27-vector space @xmath122 , the action on @xmath121 is given by @xmath123 for any @xmath116 and @xmath124 . given a morphism of @xmath1-modules @xmath125 ,
define @xmath126 @xmath127 @xmath128 by @xmath129 for any @xmath130 @xmath121 .
then @xmath131 is also a @xmath1-homomorphism .
indeed , for any @xmath124 and @xmath116 , we have @xmath132 for any @xmath133 and @xmath134 , we define a functor @xmath135 by @xmath136=@xmath121 and @xmath137
. one can check that @xmath138 is an automorphism and the inverse is @xmath139 .
so we have that @xmath119 is indecomposable ( respectively , projective , injective , simple ) if and only if so is @xmath121 in @xmath11 . the _ skew group algebra _ @xmath6 of @xmath2 over @xmath1 is given by the following data : * as an abelian group
@xmath6 is a free left @xmath1-module with the elements of @xmath2 as a basis . *
the multiplication in @xmath6 is defined by the rule @xmath140 for any @xmath141 and @xmath142 .
when @xmath2 is a finite group such that @xmath8 is invertible in @xmath1 , the natural inclusion @xmath143 induces the induction functor @xmath144 and the restriction functor @xmath145 .
( [ rr , p.227 and p.235 ] ) let @xmath2 be a finite group such that @xmath8 is invertible in @xmath1 .
then we have * @xmath146 and @xmath147 are adjoint pairs of functors .
consequently , @xmath148 and @xmath149 are both exact , and hence both preserve projective modules and injective modules .
* let @xmath150 and @xmath120 .
then the subset @xmath151 of @xmath152 has a structure of @xmath1-module given by @xmath153 for any @xmath116 , so that @xmath154 and @xmath155 are isomorphic as @xmath1-modules .
therefore , as @xmath1-modules , we have @xmath156 and then @xmath157
in this section we investigate the relationship among @xmath2-stable support @xmath0-tilting modules , @xmath2-stable two - term silting complexes , @xmath2-stable functorially finite torsion classes and @xmath2-stable cluster - tilting objects . from now on
, @xmath1 is an algebra with action by a finite group @xmath2 . *
3.1 . some definitions
* in this subsection , we introduce the notions of @xmath2-stable support @xmath0-tilting modules , @xmath2-stable torsion classes and @xmath2-stable two - term silting complexes .
recall from [ dls ] that a tilting @xmath1-module @xmath3 is called _
@xmath2-stable _ if @xmath158 for any @xmath159 .
motivated by this , we introduce the following * * we say that a support @xmath0-tilting module @xmath34 in @xmath11 is _
@xmath2-stable _ if @xmath160 for any @xmath159 .
* we say that a support @xmath0-tilting pair ( or a @xmath0-rigid pair ) @xmath161 for @xmath1 is _
@xmath2-stable _ if @xmath160 and @xmath162 for any @xmath159 .
* we say that a torsion class @xmath55 is _
@xmath2-stable _ if @xmath163 for any @xmath159 .
we denote by @xmath2-@xmath9-@xmath10 the set of isomorphism classes of basic @xmath2-stable support @xmath0-tilting @xmath1-modules and @xmath2-@xmath15-@xmath16 the set of @xmath2-stable functorially finite torsion classes in @xmath11 .
the following result shows that in a support @xmath0-tilting pair @xmath164 the @xmath2-stability of @xmath3 implies the @xmath2-stability of the pair .
let @xmath3 be a @xmath1-module and @xmath48 a projective @xmath1-module .
then @xmath165-@xmath10 if and only if @xmath166-@xmath10 . moreover , if @xmath165-@xmath10 and @xmath3 is a @xmath2-stable support @xmath0-tilting module , then @xmath48 is @xmath2-stable .
since @xmath135 is an automorphism commuting with @xmath0 ( see the proof of [ rr , lemma 4.1 ] ) , we have that @xmath167 if and only if @xmath168 , and that @xmath169 if and only if @xmath170 , and if and only if @xmath171 .
so @xmath3 is @xmath0-rigid if and only if @xmath172 is @xmath0-rigid .
thus the former assertion follows . if @xmath3 is a @xmath2-stable support @xmath0-tilting module , then by proposition 2.1 we have that @xmath48 is also @xmath2-stable . for any complex @xmath173=@xmath174@xmath175 over @xmath11 and @xmath120 , let @xmath176 be the complex @xmath177 , where @xmath178 is the ring of integers .
moreover , given another complex @xmath179 over @xmath11 and a morphism of complexes @xmath180 , let @xmath181@xmath175 .
clearly , @xmath131 is a morphism of complexes .
since @xmath182 is an automorphism , this construction is compatible with the homotopy relation and preserves projective modules .
this allows defining an automorphism @xmath183 for any @xmath120 . in this way
, we obtain an action by @xmath2 on @xmath14 .
we call a basic two - term silting complex @xmath184 _ @xmath2-stable _ if @xmath185 for any @xmath120 .
we denote by @xmath2-@xmath12-@xmath13 the set of isomorphism classes of basic @xmath2-stable two - term silting complexes for @xmath1 . *
connection of @xmath2-@xmath9-@xmath10 with @xmath2-@xmath15-@xmath16 and @xmath2-@xmath12-@xmath13 * in this subsection , we show that @xmath2-stable support @xmath0-tilting modules correspond bijectively to @xmath2-stable functorially finite torsion classes as well as @xmath2-stable two - term silting complexes .
the following result establishes a one - to - one correspondence between @xmath2-stable support @xmath0-tilting @xmath1-modules and @xmath2-stable functorially finite torsion classes in @xmath11 .
the bijection of theorem 2.3 restricts to a bijection : @xmath2-@xmath49@xmath0-@xmath10 @xmath186-@xmath15-@xmath16 .
assume that @xmath3 is @xmath2-stable .
for any @xmath187 @xmath11 and @xmath120 , we have that for any @xmath188 , @xmath189 is surjective if and only if so is @xmath190 .
so we have @xmath191 for any @xmath120 , that is , @xmath192 is @xmath2-stable .
conversely , if @xmath77-@xmath16 is @xmath2-stable , then @xmath193 for any @xmath120 .
since @xmath138 is an automorphism , we have @xmath194 so @xmath195 for any @xmath120 , that is , @xmath196 is @xmath2-stable .
recall that @xmath197 is _ sincere _ if every simple @xmath1-module appears as a composition factor in @xmath68 .
this is equivalent to the condition that @xmath198 for any indecomposable projective module @xmath48 .
also recall that @xmath150 is _ faithful _ if its left annihilator @xmath199 a class of left @xmath1-modules @xmath55 is _ sincere _ if for any indecomposable projective module @xmath48 , we have @xmath200 a class of left @xmath1-modules @xmath55 is _ faithful _ if @xmath201 the following result shows that support @xmath0-tilting modules can be regarded as a common generalization of @xmath0-tilting modules and tilting modules .
( [ air , proposition 2.2 ] ) * @xmath0-tilting modules are precisely sincere support @xmath0-tilting modules . *
tilting modules are precisely faithful support @xmath0-tilting modules .
we denote by @xmath2-@xmath202-@xmath16 ( respectively , @xmath2-@xmath203-@xmath16 ) the set of @xmath2-stable sincere ( respectively , faithful ) functorially finite torsion classes in @xmath11 . using proposition 3.5
, we get the following the bijection in theorem 3.4 restricts to bijections @xmath2-@xmath0-@xmath204-sf-@xmath16 and @xmath2-@xmath204-ff-@xmath16 .
let @xmath3 be a @xmath2-stable support @xmath0-tilting @xmath1-module .
it follows proposition 3.5 that @xmath3 is a @xmath0-tilting @xmath1-module ( respectively , tilting @xmath1-module ) if and only if @xmath3 is sincere ( respectively , faithful ) .
claim 1 : @xmath3 is sincere if and only if @xmath192 is sincere . if @xmath3 is sincere , it is obvious that @xmath192 is sincere by definition .
conversely , if @xmath192 is sincere , then for any indecomposable projective module @xmath48 we have @xmath205 , that is , there exists @xmath206 such that @xmath207 . since @xmath206 ,
there exist exact sequences : @xmath208 @xmath209 where @xmath188 .
so we have @xmath210 for any indecomposable projective module @xmath48 .
therefore @xmath3 is sincere .
claim 2 : @xmath3 is faithful if and only if @xmath192 is faithful .
it suffices to show that @xmath211 .
it is obvious that @xmath212 @xmath213 by definition .
conversely , for any @xmath214 and @xmath215 , there exists an exact sequence @xmath216 with @xmath188 and @xmath217 .
then we have @xmath218 , that is , @xmath219 @xmath220 .
we end this subsection with the following result .
the bijection of theorem 2.4 restricts to a bijection : @xmath2-@xmath12-@xmath221-@xmath9-@xmath10 . if @xmath222-@xmath49@xmath0-@xmath10 , then @xmath223 and @xmath224 for any @xmath120 .
let @xmath225 be a minimal projective presentation of @xmath3 .
then @xmath226 @xmath227 is a minimal projective presentation of @xmath172 since @xmath228 is an automorphism .
since @xmath223 for any @xmath120 , it follows that @xmath185 and @xmath229 ) \cong p^{\bullet}\oplus p[1]$ ] .
conversely , if @xmath230-@xmath12-@xmath13 , then @xmath231 because @xmath112 commutes with taking cokernel .
it follows that @xmath232 is @xmath2-stable . *
connection of @xmath2-@xmath49@xmath0-@xmath10 with @xmath2-@xmath18-tilt@xmath5 * let @xmath5 be a @xmath27-linear hom - finite krull - schimidt 2-calabi - yau triangulated category . an action of @xmath2 on @xmath5 is a group homomorphism @xmath233 from @xmath2 to the group of triangulated automorphisms of @xmath5 , that is , @xmath234 is a triangle automorphism . for any @xmath96 and @xmath120 , @xmath121 denotes the image of @xmath34 under @xmath135 .
an object @xmath34 in @xmath5 is called _
@xmath2-stable _ if @xmath235 for any @xmath120 .
we denote by @xmath2-@xmath18-@xmath19 the set of isomorphism classes of basic @xmath2-stable cluster - tilting objects . throughout this subsection ,
let @xmath3 be a @xmath2-stable cluster - tilting object in @xmath5 with a fixed isomorphism @xmath236 such that @xmath237 and @xmath238 .
then @xmath95 admits a @xmath2-action via @xmath239 for any @xmath120 and @xmath116 .
the following proposition plays an important role in this subsection .
the action of @xmath2 on @xmath5 induces that on @xmath240 $ ] .
it suffices to show that @xmath102 $ ] is closed under the @xmath2-action . by the definition of triangle functors we have that the shift functor [ 1 ] and the functor @xmath135 commute on objects .
so @xmath241)={{\operatorname{\mathsf{add}}}\nolimits}^{\sigma}\hspace{-2pt}(t[1 ] ) = { { \operatorname{\mathsf{add}}}\nolimits}(^{\sigma}\hspace{-2pt}t)[1 ] = { { \operatorname{\mathsf{add}}}\nolimits}t[1]$ ] , and the assertion follows . by theorem 2.5
, there exists an equivalence of categories between @xmath240 $ ] and @xmath11 .
so the action of @xmath2 on @xmath240 $ ] induces an action of @xmath2 on @xmath11 . on the other hand
, the action of @xmath2 on @xmath1 also induces that on @xmath11 .
the following result shows that these two actions coincide .
the functor @xmath242 commutes with @xmath2-action .
it suffices to prove that for any @xmath243 , there exists a @xmath1-module isomorphism : @xmath244 take @xmath245 for any @xmath246 .
then @xmath247 is clearly an isomorphism as @xmath27-vector spaces . because @xmath248 we have that @xmath247 is a @xmath1-module isomorphism . if both @xmath249 and @xmath250 are basic cluster - tilting objects with @xmath251 ( respectively , @xmath252 ) a maximal direct summand of @xmath253 ( respectively , @xmath254 ) which belongs to @xmath102 $ ] , then @xmath255 .
let @xmath253 , @xmath254 and @xmath3 be cluster - tilting and @xmath256 $ ] .
then we have @xmath257)={{\operatorname{\mathsf{hom}}}}_{\mathcal{c}}(x , y'[1])\oplus{{\operatorname{\mathsf{hom}}}}_{\mathcal{c}}(y '' , y'[1])=0,\ ] ] @xmath258)={{\operatorname{\mathsf{hom}}}}_{\mathcal{c}}(x , y''[1])\oplus{{\operatorname{\mathsf{hom}}}}_{\mathcal{c}}(y ' , y''[1])=0.\ ] ] it is easy to get that @xmath259 and @xmath260 from the definition of cluster - tilting objects .
since @xmath251 ( respectively , @xmath252 ) is a maximal direct summand of @xmath253 ( respectively , @xmath254 ) which belongs to @xmath102 $ ] by assumption , we have @xmath255 . for @xmath120 , we have @xmath261-@xmath19 if and only if @xmath262-@xmath19 .
let @xmath261-@xmath19 .
then @xmath263)=0\}$ ] .
so we have @xmath264)=0\}\\ & = \{{^{\sigma}c}\in\mathcal{c}\mid { { \operatorname{\mathsf{hom}}}}_{\mathcal{c}}({^{\sigma}x},{^{\sigma}c}[1])=0\}\\ & = \{c\in\mathcal{c}\mid { { \operatorname{\mathsf{hom}}}}_{\mathcal{c}}({^{\sigma}x},c[1])=0\},\end{aligned}\ ] ] and hence @xmath262-@xmath19 .
dually , we get that @xmath262-@xmath19 implies @xmath261-@xmath19 .
the following observation is useful . if @xmath34=@xmath265 @xmath18-@xmath19 with @xmath105 a maximal direct summand of @xmath34 which belongs to @xmath102 $ ] , then @xmath34 is @xmath2-stable if and only if @xmath266 is @xmath2-stable . by proposition 3.8
, we have @xmath267 = { { \operatorname{\mathsf{add}}}\nolimits}{^{\sigma}\hspace{-2pt}(t[1])}={{\operatorname{\mathsf{add}}}\nolimits}(^{\sigma}t)[1 ] = { { \operatorname{\mathsf{add}}}\nolimits}t[1].\ ] ] so @xmath268 $ ] if and only if @xmath269 $ ] . since @xmath105 is a maximal direct summand of @xmath34 which belongs to @xmath102 $ ] by assumption
, @xmath270 is also a maximal direct summand of @xmath271 $ ] .
if @xmath34 is @xmath2-stable , then @xmath272 for any @xmath120 .
since @xmath270 ( respectively , @xmath105 ) is a maximal direct summand of @xmath121 ( respectively , @xmath34 ) which belongs to @xmath102 $ ] , we have @xmath273 and @xmath274 for any @xmath120 . by lemma 3.11
we have @xmath275 @xmath276 c - tilt@xmath5 . if @xmath266 is @xmath2-stable , then @xmath277 @xmath270 and @xmath278 are basic cluster - tilting objects . by lemma 3.10
we have @xmath279 .
thus @xmath34 is @xmath2-stable .
now we are in a position to prove the following the bijection of theorem 2.6 restricts to a bijection : @xmath2-@xmath18-@xmath280-@xmath49@xmath0-@xmath10 . by lemma 3.9
, we have @xmath281(@xmath3 , @xmath155 ) for any @xmath282 $ ] . if @xmath283-c-@xmath19 , then @xmath266 is @xmath2-stable by proposition 3.12 . for any @xmath120 , we have @xmath284 thus @xmath285 is @xmath2-stable . moreover , @xmath286-s@xmath0-@xmath10 is @xmath2-stable .
conversely , if @xmath285 is @xmath2-stable , then we have @xmath287(@xmath3 , @xmath288 ) @xmath289 as before .
then by theorem 2.5 , we have @xmath274 for any @xmath120 , that is , @xmath266 is @xmath2-stable
. thus @xmath283-@xmath18-@xmath19 by proposition 3.12 . in the following we
establish a bijection between @xmath2-stable cluster - tilting objects in @xmath5 and @xmath2-stable two - term silting complexes in @xmath14 .
let @xmath34 be a basic object of @xmath5 and take a triangle @xmath97\ ] ] with @xmath98 and @xmath15 a minimal right @xmath99-approximation .
then the following statements are equivalent .
* @xmath34 is @xmath2-stable in @xmath5 .
* @xmath290 is @xmath2-stable in @xmath14 . since @xmath291 is a triangulated equivalence .
so we have the following diagram : @xmath292^{g}\ar[d ] & t_{0}\ar[r]^{f}\ar[d ] & x\ar[r]\ar[d ] & t_{1}[1]\ar[d]&\\ & ^{\sigma}\hspace{-2pt}t_{1}\ar[r]^{^{\sigma}\hspace{-2pt}g } & ^{\sigma}\hspace{-2pt}t_{0}\ar[r]^{^{\sigma}\hspace{-2pt}f } & ^{\sigma}\hspace{-2pt}x\ar[r ] & ( ^{\sigma}\hspace{-2pt}t_{1})[1].&}\ ] ] by the proof of theorem 3.13 , we have that @xmath293 is @xmath2-stable if and only if @xmath290 is @xmath2-stable .
@xmath294 we already have @xmath293 is @xmath2-stable , it follows that @xmath34 is @xmath2-stable .
@xmath295 if @xmath34 is @xmath2-stable , then @xmath235 . since @xmath15 is a minimal right @xmath99-approximation by assumption
, @xmath131 is a minimal right @xmath296-approximation and hence a minimal right @xmath99-approximation .
thus @xmath297 and @xmath298 .
therefore @xmath293 is @xmath2-stable and the assertion follows .
immediately , we get the following the bijection of theorem 2.7 restricts to a bijection : @xmath2-@xmath18-@xmath280-@xmath12-@xmath13 .
it follows from theorem 2.7 and lemma 3.14 .
throughout this section , @xmath1 is an algebra and @xmath2 is a finite group acting on @xmath1 such that @xmath8 is invertible in @xmath1 .
we denote by @xmath22 the group of characters on @xmath2 , that is , the group homomorphisms @xmath299 .
then @xmath34 acts on @xmath6 via @xmath300 .
we prove that there exists an injection from @xmath2-stable support @xmath0-tilting @xmath1-modules to @xmath22-stable support @xmath0-tilting @xmath6-modules . in the case
when @xmath2 is a solvable group , the injection turns out to be a bijection .
we only need to prove @xmath303 for any @xmath304 .
note that @xmath305 is a @xmath6-module whose underlying set and the additive structure is the same as @xmath302 , in which @xmath306 is defined to be @xmath307 .
define @xmath308 @xmath309 via @xmath310 .
clearly it is a bijection . because @xmath311 we have that @xmath312 is a @xmath6-homomorphism , and hence an isomorphism .
* from the set of isomorphism classes of @xmath2-stable @xmath0-rigid @xmath1-modules to the set of isomorphism classes of @xmath22-stable @xmath0-rigid @xmath6-modules . * from the set of isomorphism classes of @xmath2-stable @xmath0-rigid pair in @xmath11 to the set of isomorphism classes of @xmath22-stable @xmath0-rigid pair in @xmath315 .
* @xmath2-s@xmath0-@xmath316-@xmath9-@xmath25 .
we claim that the functor @xmath148 restricting to the set of isomorphism classes of basic @xmath2-stable @xmath1-modules is an injection . if both @xmath317 and @xmath318 are @xmath2-stable @xmath1-modules and @xmath319 , then @xmath320 , that is , @xmath321 .
since @xmath317 and @xmath318 are @xmath2-stable , we have @xmath322 with @xmath323 .
thus @xmath324 , and the claim follows .
\(1 ) by definition , @xmath3 is @xmath0-rigid if and only if @xmath169 . by the proof of [ rr ,
lemma 4.2 ] , we have that @xmath148 commutes with @xmath0 .
so we have @xmath325 since @xmath3 is @xmath2-stable @xmath0-rigid in @xmath11 , we have @xmath302 is @xmath0-rigid in @xmath315 .
now the assertion follows from lemma 4.1 .
\(2 ) note that @xmath326 is a @xmath2-stable @xmath0-rigid pair if and only if @xmath3 is @xmath2-stable @xmath0-rigid , @xmath48 is @xmath2-stable projective and @xmath327 .
it follows from the injection in ( 1 ) that @xmath302 is @xmath0-rigid in @xmath315 . by lemma 2.8 ,
@xmath148 preserves projective modules and @xmath328 is a projective module in @xmath315 .
we have @xmath329 thus @xmath330=0 , and therefore @xmath331 is a @xmath22-stable @xmath0-rigid pair in @xmath315 by lemma 4.1 .
\(3 ) by propositions 2.1 and 2.2 , @xmath332-@xmath10 if and only if @xmath3 is @xmath0-rigid and there exists an exact sequence @xmath333 in @xmath11 with @xmath334 and @xmath15 a left @xmath99-approximation of @xmath1 .
it follows from the injection in ( 1 ) that @xmath302 is a @xmath0-rigid @xmath6-module and there exists an exact sequence @xmath335 in @xmath315 with @xmath336 . then by proposition 2.2
, we only have to prove that @xmath337 is a left @xmath338-approximation of @xmath6 , that is , @xmath339 is surjective for any @xmath340 .
first we prove that @xmath341 is surjective .
consider the following commutative diagram : @xmath342^{{{\operatorname{\mathsf{hom}}}}_{\lambda g}(ff , ft)}\ar[d]^{\cong } & { { \operatorname{\mathsf{hom}}}}_{\lambda g}(\lambda g , ft)\ar[d]^{\cong } & \\ & { { \operatorname{\mathsf{hom}}}}_{\lambda}(t ' , hft)\ar[r]^{{{\operatorname{\mathsf{hom}}}}_{\lambda}(f , hft)}\ar[d]^{\cong } & { { \operatorname{\mathsf{hom}}}}_{\lambda}(\lambda , hft)\ar[d]^{\cong } & \\ & { { \operatorname{\mathsf{hom}}}}_{\lambda}(t ' , \bigoplus_{\sigma\in g}{^{\sigma}t})\ar[r]^{{{\operatorname{\mathsf{hom}}}}_{\lambda}(f,\bigoplus_{\sigma\in g}{^{\sigma}t } ) } \ar[d]^{\cong } & { { \operatorname{\mathsf{hom}}}}_{\lambda}(\lambda , \bigoplus_{\sigma\in g}{^{\sigma}t})\ar[d]^{\cong } & \\ & { { \operatorname{\mathsf{hom}}}}_{\lambda}(t ' , t^n)\ar[r]^{{{\operatorname{\mathsf{hom}}}}_{\lambda}(f , t^n ) } & { { \operatorname{\mathsf{hom}}}}_{\lambda}(\lambda , t^n ) , & } \ ] ] where @xmath323 .
the last row is surjective since @xmath15 is left @xmath99-approximation of @xmath1 .
so the first row is also surjective .
now let @xmath340 and @xmath343 .
then there exist @xmath344 and @xmath345 such that @xmath346 .
so we have a split exact sequence @xmath347 & m\ar[r]^{i } & ( ft)^{m}\ar[r]\ar@<1ex>[l]^{p } & n\ar[r ] & 0 & } \ ] ] in @xmath315 with @xmath348 . by the above argument
, there exists @xmath349 such that @xmath350 .
so we have @xmath351 and hence @xmath339 is surjective .
let @xmath3 be a @xmath2-stable ( basic ) tilting @xmath1-module .
then by proposition 3.5(2 ) , @xmath3 is a @xmath2-stable faithful support @xmath0-tilting module .
so @xmath1 is cogenerated by @xmath3 and there exists an injection @xmath352 in @xmath11 .
since @xmath148 is exact , we get an injection @xmath353 in @xmath315 . so @xmath302 is a @xmath22-stable faithful support @xmath0-tilting @xmath6-module by theorem 4.2(3 ) , and hence it is a tilting @xmath6-module by proposition 3.5(2 ) again .
let @xmath354 be a @xmath6-module .
for any @xmath355 and @xmath356 , we define a map @xmath357 by @xmath358 .
then for any @xmath359 , we have @xmath360 so @xmath361 is a @xmath1-module homomorphism .
we also have that @xmath361 is an isomorphism with the inverse @xmath362 @xmath363 such that @xmath364 for any @xmath356 . *
if @xmath3 is @xmath0-rigid in @xmath11 , then @xmath365 is @xmath2-stable @xmath0-rigid in @xmath11 . *
if @xmath3 is support @xmath0-tilting in @xmath11 , then @xmath365 is @xmath2-stable support @xmath0-tilting in @xmath11 .
it follows from [ rr , corollary 5.2 ] that @xmath370 is morita equivalent to @xmath371 , where @xmath372 is the commutator subgroup of @xmath2 .
let @xmath2 be solvable , and let @xmath373 be its derived series , that is , every subgroup is the commutator subgroup of the preceding one .
denote by @xmath374 the character group of @xmath375 . by [ rr , proposition 5.4 ]
, we can get from @xmath6 to @xmath1 by using a finite number of skew group algebra constructions , combined with morita equivalences . to be more precise
, there exists a chain of skew group algebras @xmath376 where each algebra @xmath377 is the skew group algebra of the preceding algebra @xmath378 under the action of the group @xmath379 .
then we have the induced functors @xmath380 under the above assumption , we give the following let @xmath2 be an abelian group .
it is well known that @xmath22 is isomorphic to @xmath2 .
so @xmath6 admits an action by @xmath2 ; and moreover , by [ rr ] the skew group algebra @xmath394 is morita equivalent to @xmath1 .
now the following is an immediate consequence of theorem 4.6 .
let @xmath1 be the path algebra of the quiver @xmath396 ( see below ) , and let @xmath397 act on @xmath1 by switching @xmath12 and @xmath398 , @xmath399 and @xmath400 and fixing the vertex @xmath401 .
then the following @xmath402 is the quiver of @xmath6 . @xmath403^{\gamma } & & & \\ & q=\hspace{-15pt}&1\ar[ur]^{\alpha}\ar[dr]_{\beta } & & & & & q'=\hspace{-15pt}&&2&\\ & & & 2'&&&&&1'\ar[ur]_{\delta } & & & } \ ] ] the auslander - reiten quivers of @xmath11 and @xmath315 are the following , where each module is represented by its radical filtration .
@xmath404 & & \scriptsize{\begin{matrix } 1\\2'\\ \end{matrix}}\ar[dr ] & & & & \scriptsize{\begin{matrix } 1\\2\\ \end{matrix}}\ar[dr ] & & 1'&\\ & & \scriptsize{\begin{matrix } 1\\2\ \ 2'\\ \end{matrix}}\ar[ur]\ar[dr]&&1&&2\ar[ur]\ar[dr ] & & \scriptsize{\begin{matrix } 1\ \ 1'\\2\\ \end{matrix}}\ar[ur]\ar[dr ] & & \\ & 2'\ar[ur ] & & \scriptsize{\begin{matrix } 1\\2\\ \end{matrix}}\ar[ur ] & & & & \scriptsize{\begin{matrix } 1'\\2\\ \end{matrix}}\ar[ur ] & & 1&\\ } \ ] ] @xmath405 we denote by @xmath406 the set of isomorphism classes of indecomposable @xmath1-modules and by @xmath407 the set of isomorphism classes of indecomposable @xmath6-modules .
one can calculate the left mutation of support @xmath0-tilting @xmath1-modules by exchanging sequences that are constructed from left approximations .
therefore we can draw the support @xmath0-tilting quiver of an algebra by its auslander - reiten quiver .
now we draw @xmath419-@xmath420 and @xmath419-@xmath421 as follows .
@xmath419-@xmath420 : @xmath422 & & & \\ & & \color{red}{2\oplus 2'}\ar[ur]\ar[r ] & 2'\ar[rr ] & & \color{blue}{0 } & \\ & \color{green}{\scriptsize{\begin{matrix } 1\\2\ \
2'\\ \end{matrix}}\oplus 2'\oplus 2}\ar[ur]\ar[r]\ar[dr ] & \scriptsize{\begin{matrix } 1\\2\ \
2'\\ \end{matrix}}\oplus\scriptsize{\begin{matrix } 1\\2'\\ \end{matrix}}\oplus 2'\ar[r]\ar[dr ] & \scriptsize{\begin{matrix } 1\\2'\\ \end{matrix}}\oplus 2'\ar[u]\ar[r ] & 1\oplus\scriptsize{\begin{matrix } 1\\2'\\ \end{matrix}}\ar[dr ] & & \\ & & \scriptsize{\begin{matrix } 1\\2\ \ 2'\\ \end{matrix}}\oplus2\oplus\scriptsize{\begin{matrix } 1\\2\\ \end{matrix}}\ar[r]\ar[dr ] & \color{orange}{\scriptsize{\begin{matrix } 1\\2\ \
2'\\ \end{matrix}}\oplus\scriptsize{\begin{matrix } 1\\2'\\ \end{matrix}}\oplus\scriptsize{\begin{matrix } 1\\2\\ \end{matrix}}}\ar[r ] & \color{brown}{1\oplus\scriptsize{\begin{matrix } 1\\2'\\ \end{matrix}}\oplus\scriptsize{\begin{matrix } 1\\2\\ \end{matrix}}}\ar[u]\ar[d ] & \color{purple}{1}\ar[uu ] & \\ & & & 2\bigoplus\scriptsize{\begin{matrix } 1\\2\\ \end{matrix}}\ar[r]\ar@/^3pc/[uuuu ] & 1\bigoplus\scriptsize{\begin{matrix } 1\\2\\ \end{matrix}}\ar[ur ] & & \\ } \ ] ] @xmath419-@xmath421 : @xmath423\ar[dr ] & \color{red}{2}\ar[drrr ] & & & & \\ & \color{green}{\scriptsize{\begin{matrix } 1\\2\\ \end{matrix}}\oplus\scriptsize{\begin{matrix } 1'\\2\\ \end{matrix}}\oplus2}\ar[ur]\ar[r]\ar[dr ] & \scriptsize{\begin{matrix } 1\\2\\ \end{matrix}}\oplus2\ar[ur]\ar@/_1.5pc/[ddrr ] & 1'\oplus \scriptsize{\begin{matrix } 1'\\2\\ \end{matrix}}\ar[r ] & 1'\ar[rr ] & & \color{blue}{0}&\\ & & \color{orange}{\scriptsize{\begin{matrix } 1\\2\\ \end{matrix}}\oplus\scriptsize{\begin{matrix } 1'\\2\\ \end{matrix}}\oplus\scriptsize{\begin{matrix } 1\ \ 1'\\2\\ \end{matrix}}}\ar[r]\ar[dr ] & 1'\oplus\scriptsize{\begin{matrix } 1'\\2\\ \end{matrix}}\oplus\scriptsize{\begin{matrix } 1\ \ 1'\\2\\ \end{matrix}}\ar[u]\ar[r ] & \color{brown}{1'\oplus1\oplus\scriptsize{\begin{matrix } 1\ \ 1'\\2\\ \end{matrix}}}\ar[r ] & \color{purple}{1'}\oplus1\ar[d]\ar[ul]&&\\ & & & \scriptsize{\begin{matrix } 1\\2\\ \end{matrix}}\oplus1\oplus\scriptsize{\begin{matrix } 1\ \
1'\\2\\ \end{matrix}}\ar[ur]\ar[r ] & \scriptsize{\begin{matrix } 1\\2\\ \end{matrix}}\oplus1\ar[r ] & 1\ar[uur ] & & \\ } \ ] ] the colored support @xmath0-tilting modules in the graph are all the basic @xmath2-stable support @xmath0-tilting modules in @xmath11 and @xmath315 respectively
. moreover , the bijection in theorem 4.6 takes a @xmath2-stable support @xmath0-tilting module in @xmath419-@xmath420 to that in @xmath419-@xmath25 ) in the same color . the @xmath2-stable support @xmath0-tilting modules in green ,
orange and brown are @xmath2-stable tilting .
* acknowledgement*. this research was partially supported by nsfc ( grant no .
11571164 ) and a project funded by the priority academic program development of jiangsu higher education institutions .
the authors would like to thank dong yang and yuefei zheng for their helpful discussions .
m. auslander , i. reiten and s. o. smal@xmath424 , representation theory of artin algebras , corrected reprint of the 1995 original , cambridge studies in adv . math . *
36 * , cambridge univ . press , cambridge , 1997 .
k. bongartz , _ tilted algebras _ , representations of algebras , proceedings of the third international conference held in puebla , august 48 , 1980 , edited by m. auslander and e. lluis , lecture notes in math .
* 903 * , springer , berlin - new york , 1981 , pp.2638 .
t. brstle and d. yang , _ ordered exchange graphs _ , in : advances in representation theory of algebras ,
the 15th workshop and international conference on representations on algebras ( icra 2012 ) held in bielefeld , august 817 , 2012 , edited by d. j. benson , h. krause and a. skowroski , ems ser .
soc . , zrich , 2013 , pp.135193 . | motivated by @xmath0-tilting theory developed by adachi , iyama and reiten , for a finite - dimensional algebra @xmath1 with action by a finite group @xmath2 , we introduce the notion of @xmath2-stable support @xmath0-tilting modules
. then we establish bijections among @xmath2-stable support @xmath0-tilting modules over @xmath1 , @xmath2-stable two - term silting complexes in the homotopy category of bounded complexes of finitely generated projective @xmath1-modules , and @xmath2-stable functorially finite torsion classes in the category of finitely generated left @xmath1-modules . in the case when @xmath1 is the endomorphism of a @xmath2-stable cluster - tilting object @xmath3 over a @xmath4-finite 2-calabi - yau triangulated category @xmath5 with a @xmath2-action , these are also in bijection with @xmath2-stable cluster - tilting objects in @xmath5 .
moreover , we investigate the relationship between stable support @xmath0-tilitng modules over @xmath1 and the skew group algebra @xmath6 .
-stable support @xmath0-tilting modules , @xmath2-stable two - term silting complexes , @xmath2-stable functorially finite torsion classes , @xmath2-stable cluster - tilting objects , bijection , skew group algebras . ] |
the irregularities in the cosmic ray ( cr ) energy spectrum of `` knee '' type at @xmath2 ev and `` ankle '' type at @xmath3 ev found in @xcite are yet of special interest from the point of view of interpretation of these phenomena from the position of astrophysics . in recent years
a few papers @xcite have been published which try to explain such a behavior of cr spectrum with the help of new models of generation and propagation of crs .
there exists also another namely a nuclear - physical point of view for the formation of irregularity at @xmath4 ev @xcite . in some sense , the answer to the problem on reasons of the formation of breaks is in the detailed study of different eas characteristics in the region of the first and second irregularities in the spectrum .
we have made such a work in @xcite .
the cascade curves of eas development in fig.[fig2 ] were reconstructed according to the method suggested in @xcite .
it is seen from fig.[fig2 ] that the maximum depth of cascade curves depends on the primary particle mass composition as well as the hadron interaction model .
it is seen from fig.[fig2 ] that to describe the experimental cascade curve ( @xmath5 , @xmath6 ) the qgsjet model is better - suited @xcite .
so we use this model for the estimation of mass composition of primary particles .
fig.[fig3 ] presents the calculations of the maximum depth @xmath5 using the qgsjet model for the primary proton and iron nucleus , and experimental data obtained at the yakutsk eas array .
it is seen that the velocity of shift of @xmath5 to sea level depends on the energy range . in the framework of the qgsjet model
the experimental data are indicative of the change of mass composition of primary particles in the energy range @xmath7 ev and at @xmath8 ev .
ev ) with different models of hadron interactions.,scaledwidth=42.0% ] fig.[fig3 ] and fig.[fig4 ] present experimental data : a ) the density of muons with @xmath9 gev at a distance of 1000 m from the shower core @xcite , b ) the root - mean - square radius @xmath10 of charged particle ( ldf ) @xcite .
these data are compared with calculations using the qgsjet model .
protons , iron nuclei and @xmath11-quanta are considered as primary particles . a confidence interval taken in calculations is @xmath12 . from fig.3
it follows that in the energy range @xmath13 ev within the boundaries of confidence there are @xmath14% of showers for the iron nucleus and @xmath15% for the proton . at @xmath16
ev the portion of iron nuclei decreases , the per cent of protons and @xmath11-quanta increases .
it is seen from fig.[fig4 ] that the `` heaviest '' mass composition is observed at @xmath17 ev . vs @xmath18 relations for observed events @xmath19 ev ( squares ) , @xmath20 ev ( points ) and @xmath21 ev ( triangles ) . expected @xmath12 bounds for the distributions are indicated for p , fe and @xmath11 primary by different curve as in the legend.,scaledwidth=42.0% ] the most sensitive instrument for the model of hadron interactions and mass composition of primary particles is the eas muon component . at the yakutsk complex eas array
the muons with @xmath0 gev are measured by a shower registration in 70% cases .
fig.[fig5 ] presents the correlation of @xmath22 parameters and fig.[fig6 ] gives the portion of muons ( ratio of muons to all charged particles ) at a distance of 300 m and 600 m from the shower core . in the same place
the calculations by the qgsjet model are given .
the tendency for an increase of light nuclei in the cosmic ray primary flux at @xmath23 ev is marked by these data . and
muons @xmath24 at sea level .
the curves are a calculation by the qgsjet model for the primary proton and iron nuclei.,scaledwidth=42.0% ] gev ( % ) . @xmath25 and @xmath26.,scaledwidth=42.0% ] at the yakutsk complex eas array
the shower energy is determined by measurements of the total flux of eas cherenkov light @xmath27 , the total number of charged particles , @xmath28 , and muons , @xmath24 , at sea level @xcite .
fig.[fig7 ] presents the energy - dependence of @xmath29 ratio .
it is seen from calculations that the ratio strongly depends on a mass composition . from comparison of the calculations by the qgsjet model for the proton , iron nucleus and experimental data
it follows that the mass composition changes just after the first knee in the spectrum , i.e. in the interval @xmath30 ev .
it is evident from fig.[fig8 ] where the portion of energy transmitted in to the electromagnetic eas cascade is shown . on a primary energy.,scaledwidth=42.0% ] in this section we consider the fluctuations of @xmath5 and @xmath10 obtained by the measurement of cherenkov eas light and density of charged particle flux ( see fig.[fig9 ] and fig.[fig10 ] ) . in both cases at @xmath31
ev fluctuations are considerable and correspond to the mixed mass composition of primary particles . from comparison with calculations by the qgsjet model we have the following relationship
: the light nuclei are @xmath32% and heavy nuclei are @xmath33% ( see the dotted line in fig.[fig9 ] ) . at @xmath34
ev.,scaledwidth=42.0% ] at @xmath31 ev .
curves are the qgsjet model calculations for the primary proton and iron nucleus.,scaledwidth=42.0% ]
at the yakutsk complex array for almost the 35-year period of continuous observations the unique experimental data on electron , muon and cherenkov eas components in the region of superhigh and ultrahigh energies have been accumulated .
results on longitudinal and radial development of eas are presented in figures . from the total combination of data
one can select two energy regions where , as is seen from figures , the characteristics of eas have the complex dependence on the energy .
this is the energy ranges @xmath35 ev and @xmath19 ev . as is known , in these energy ranges the irregularities of `` knee '' and `` ankle '' in the energy spectrum of eas are observed . from the comparison of all experimental data with calculations by the qgsjet model in @xcite
the results of the cosmic ray mass composition in the energy range @xmath36 ev have been obtained .
it follows from the analysis that after the `` knee '' the mass composition becomes heavier and in the region of `` ankle '' , on the contrary , becomes lighter .
such a conclusion does nt contradict the hypothesis on a cosmic ray generation up to the energy @xmath37 ev in our galaxy and their propagation according to the model of anomalous diffusion in the fractal interstellar medium . beginning with @xmath38 ev
the mass composition becomes lighter and it does nt contradict the presence of cosmic rays metagalactic origin in the total flux of cosmic rays . | the energy dependence of such characteristics as a ratio of the total number of charged particles to the total flux of eas cherenkov radiation , a ratio of @xmath0 gev muon flux density at the distance of @xmath1 m from a shower core to charged particle flux density , a ratio of the energy transferred to the electromagnetic component of eas to the primary particle energy is presented .
their comparison with two - component mass composition of cosmic rays ( p - fe ) in the framework of calculations by a qgsjet model is given . |
shabat polynomial of a plane bipartite tree is not unique , but we can made it unique , if we demand that : a ) critical values are @xmath0 and @xmath1 ; b ) sum of coordinates of white vertices ( i.e. inverse images of @xmath2 ) is @xmath2 ; c ) sum of coordinates of black vertices ( i.e. inverse images of @xmath1 ) is @xmath1 .
shabat polynomial with these properties will be called shabat polynomial in zapponi form , or shabat - zapponi polynomial , or sz - polynomial @xcite .
thus , we can correspond to a tree the julia set , i.e. the julia set of its sz - polynomial .
we want to understand is there a correspondence between geometry of a plane tree and such properties of its julia set as form , connectedness and hausdorff dimension ? at first it was expected that julia set of a shabat polynomial is something simple with hausdorff dimension approximately @xmath2 ( because shabat polynomial is a generalized chebyshev polynomial ) .
this assumption turned out to be wrong .
so , we decided to study the zapponi form of shabat polynomial , because if there exists a sz - polynomial for a given bipartite tree , then such polynomial is unique .
in the course of this experimental work we found that : a ) there is some similarity between the form of a given tree and the form of its julia set ; b ) the connectedness of julia set is probably the main characteristic of an embedded tree .
we consider plane bipatite trees , i.e. trees embedded into plane , with vertices properly colored in black and white . a polynomial @xmath3 with exactly two finite critical values one and minus one will be called _ shabat polynomial _ @xcite .
the inverse image @xmath4 $ ] of segment @xmath5 $ ] is a plane bipartite tree , where white vertices are images of @xmath2 and black of @xmath1 . for each plane bipartite tree @xmath6
there exists a shabat polynomial @xmath3 such that trees @xmath6 and @xmath7 are isotopic .
such polynomial will be called a shabat polynomial of the tree @xmath6 . if polynomials @xmath3 and @xmath8 are shabat polynomials of the same tree @xmath6 , then @xmath9 for some constants @xmath10 and @xmath11 .
a shabat polynomial is in zapponi form @xcite , if the sum of coordinates of white vertices is @xmath2 and black vertices @xmath1 .
let @xmath6 be a bipartite tree and @xmath12 its sz - polynomial .
then @xmath13 .
let @xmath14 be roots of polynomial @xmath15 with multiplicities @xmath16 , respectively , and @xmath17 , @xmath18 be roots of @xmath19 and their multiplicities . then @xmath20 also we have , that @xmath21 hence , @xmath22 from ( 1 ) and ( 2 ) we have that @xmath13
. if p is a shabat polynomial and @xmath13 , then @xmath23 let @xmath6 be a bipartite tree .
if there exist its sz - polynomial @xmath3 , then @xmath3 is unique and its field of definition coincides with the field of definition of the tree @xmath6 @xcite .
if @xmath24 is a shabat polynomial of a tree @xmath6 and @xmath13 , then @xmath3 is unique up to variable change @xmath25 and the unique choice of @xmath26 in this variable change gives us sz - polynomial .
let now @xmath27 be the field of definition of a tree @xmath6 and @xmath28 $ ] be its shabat polynomial . the variable change @xmath29 preserves the field of definition , but turns coefficient at @xmath30 to zero .
if @xmath31 , then @xmath32 . if @xmath33 , then the the variable change @xmath34 also preserves the field of definition , but turns ( in new coordinates ) @xmath35 to one . then @xmath36 .
if @xmath37 , then shabat polynomial in zapponi form does not exist for the tree @xmath6 . in @xcite
it was proved that sz - polynomial always exists for trees with prime number of edges .
sz - polynomial obviously does not exist , if the tree is symmetric , i.e. if it has a nontrivial rotation automorphism with the center in one of vertices .
* conjecture . *
_ sz - polynomial exists for non - symmetric trees .
_ in what follows sz - polynomial for a tree @xmath6 will be denoted @xmath38 .
definitions of fatou and julia sets see , for example , in the book @xcite . for us
the following properties of julia sets will be important .
* julia set of a polynomial @xmath3 is the boundary of the basin of infinity , i.e. the boundary of open set of those points , whose iterations converge to infinity .
* let @xmath39 be the set of stationary repelling points of @xmath3 and let @xmath40 , @xmath41 .
then julia set of @xmath3 is the closure of @xmath42 . *
julia set of @xmath3 is connected if and only if iterations of critical points of @xmath3 constitute a bounded set . in the case of a sz - polynomial
it means that connectedness of julia set is equivalent to the boundedness of iterations of @xmath2 and @xmath1 .
if iterations of @xmath2 and @xmath1 are both unbounded , then julia set is totally disconnected ( two dimensional cantor compact ) . if iterations of @xmath2 , for example , are bounded , and iterations of @xmath1
not , then julia set is a union of infinite number of connected components @xcite .
definition of hausdorff dimension see in @xcite .
there are several methods of its computation .
description of `` box counting '' and `` packing dimension '' methods see in @xcite .
description of jenkinson - pollicott algorithm ( jp - algorithm ) see in @xcite .
it must be noted that performance of these algorithms differs from case to case .
box counting method does not work , if julia set is totally disconnected .
it also demonstrate bad performance , if hausdorff dimension of julia set is @xmath43 .
if there is a stationary point and derivative in this point is close to @xmath2 , then jp - algorithm demonstrates a bad convergence .
let t be a bipartite tree and let @xmath44 be the same tree , but with inverse colors ( i.e. white vertices in @xmath6 are black in @xmath44 and black vertices in @xmath6 are white in @xmath44 ) .
then @xmath45 . let @xmath46 be an arbitrary point and @xmath47 , @xmath48 , @xmath49 and so on .
then @xmath50 , @xmath51 , @xmath52 , and so on .
it means that julia sets of polynomials @xmath53 and @xmath38 are the same up to rotation on @xmath54 around the origin , i.e. characteristics of julia set depends only on tree and not on its coloring .
in what follows we will study one tree from the pair @xmath55 .
let @xmath6 be a bipartite tree .
by fixing some white vertex of degree @xmath56 at @xmath2 and some black vertex of degree @xmath56 at @xmath1 we uniquely define shabat polynomial @xmath3 of @xmath6 . in this case
@xmath3 will be a postcritically finite polynomial ( a pcf - polynomial ) , i.e. a polynomial with finite orbit of set of critical points ( see @xcite ) .
it must be noted that shabat pcf - polynomial of a tree @xmath6 is not unique .
let @xmath6 be a tree with four edges : @xmath57 then @xmath58 is its pcf - polynomial and @xmath59 is its sz - polynomial .
julia sets of @xmath3 and @xmath38 are quite different : @xmath60{z4pcf}\hspace{2 cm } \includegraphics[width=4cm , height=3cm]{z4sz}\ ] ] figure 1 .
figure 2 .
julia set of pcf - polynomial @xmath3 .
julia set of sz - polynomial @xmath38 . in the left figure iterations of yellow points
converge to infinity , of green points to @xmath1 , of red points to @xmath2 .
julia set is connected .
its hausdorff dimension approximately equals @xmath61 ( box counting method ) or @xmath62 ( packing dimension method ) .
sz - polynomial @xmath38 has a weakly attracting 10-cycle .
let @xmath63 be a union of the domain @xmath64 and @xmath65-neighborhood of the attracting cycle .
in the right figure points that get into @xmath63 in 5 steps or less are white , in 6 or 7 steps green , in 8 , 9 or 10 steps red .
all other points ( including points of julia set ) are blue .
julia set is connected . for its hausdorff dimension box counting method
gives estimation @xmath66 , packing dimension method @xmath67 , jp - algorithm @xmath68 .
let @xmath6 be a tree and @xmath38 its sz - polynomial .
characteristics of julia set @xmath69 depend on behavior of iterations of @xmath70 .
there are several types of this behavior .
g1 : : iterations on @xmath70 converge to an attracting point @xmath3 . here
julia set is a common border of two basins : the basin @xmath71 of infinity and the basin @xmath72 of attracting point @xmath3 . as all vertices of @xmath6 belong to @xmath72 , then the form of julia set resembles the form of the tree ( in some general manner ) .
hausdorff dimension here is close to 1 .
g2 : : iterations on @xmath70 converge to an attracting 2-cycle .
julia set approximates the tree better , than in the previous case .
also the `` fractality '' of set is greater , hence the hausdorff dimension is greater ( in average ) .
julia sets in this case are more `` interesting '' , than in previous .
good example see in figure 4 .
g3 : : iterations on @xmath70 converge to an attracting @xmath73-cycle , where @xmath74 .
julia sets can be very `` interesting '' .
the form of julia set even more closely resembles the form of the tree .
`` fractality '' is great and hausdorff dimension is greater , than @xmath75 .
good examples see in figures 2 , 3 and 6 .
g4 : : iterations on @xmath70 converge to infinity .
julia set here is totally disconnected , but some small similarity to the form of the tree remains .
hausdorff dimension can be smaller , than 1 .
julia sets are rather `` uninteresting '' .
see example of s3 case in figure 5 .
s1 : : iterations of @xmath2 ( for example ) converge to an attracting point and iterations of @xmath1 converge to an attracting @xmath73-cycle , @xmath76 .
s2 : : iterations of @xmath2 converge to one attracting @xmath73-cycle , @xmath76 , and iterations of @xmath1 converge to another attracting @xmath77-cycle , @xmath78 .
s3 : : iterations of @xmath2 ( for example ) converge to an attracting point and iterations of @xmath1 converge to infinity .
julia set here is a union of infinite number of connected components .
good example see in figure 5 . in what follows
we will give several most interesting examples of julia sets of sz - polynomials .
in this section for each 5-edge tree @xmath6 we will compute its sz - polynomial @xmath38 and find characteristics of @xmath69 .
the passport of a tree @xmath6 is the list of degrees of white vertices ( in non increasing order ) and the list of degrees of black vertices ( also in non increasing order ) .
we will always assume that `` white '' list is lexicographically not less , than `` black '' list .
estimations of hausdorff dimension we will write in order : the box counting estimation , the packing estimation and the jp - algorithm estimation .
if some method is inapplicable , then we will put `` ? '' in the corresponding position .
1 . passport @xmath79 .
+ @xmath80 + polynomial @xmath38 has an attracting 24-cycle .
iterations of @xmath70 converge to this cycle , i.e. @xmath38 is of g3-type .
the set @xmath69 is very similar to the set in figure 2 .
hausdorff dimension : @xmath81 , @xmath82 , ? .
2 . passport @xmath83 .
+ @xmath84 polynomial @xmath38 has an attracting 2-cycle : @xmath85 iterations @xmath0 and @xmath1 converge to this cycle , i.e. @xmath38 is of g2-type .
hausdorff dimension : @xmath86 , @xmath87 , ? .
3 . passport @xmath88 .
+ @xmath89 + iterations of @xmath0 and @xmath1 converge to infinity , i.e. @xmath38 is of g4-type .
julia set is totally disconnected .
hausdorff dimension : ?
, @xmath90 , @xmath91 .
passport @xmath92 .
+ @xmath93 iterations of @xmath0 and @xmath1 converge to infinity , i.e. @xmath38 is of g4 .
hausdorff dimension : ?
, @xmath94 , @xmath95 . 5 .
passport @xmath96 .
+ @xmath97 the polynomial @xmath38 has two attracting 4-cycles : @xmath98 and @xmath99 iterations of @xmath2 converge to the first cycle and iterations of @xmath1 to the second , i.e. @xmath38 is of s2-type .
hausdorff dimension : @xmath100 , @xmath101 , ? .
only one non - symmetric 6 edge tree generates a connected julia set : @xmath102 the polynomial @xmath38 has a superattracting 2-cycle : @xmath103 , @xmath104 , i.e. @xmath38 is of g2-type .
hausdorff dimension : @xmath105 , @xmath95 , ? .
julia set is similar to julia set in figure 4 .
here we have many trees that generate connected julia sets . for such tree @xmath6 we will present behavior of iterations of @xmath106
, characteristics of julia set @xmath69 and the picture of this set in interesting cases . in the picture of julia set points that quite fast come into attracting domain of infinity ( or into attracting domain of attracting point or a cycle ) are white , points that come there more slowly are yellow , even more slowly are green , then light red , then deep red .
we will use the following notations : * @xmath107 means that iterations of @xmath2 and @xmath1 converge to infinity , so julia set is totally disconnected ; * `` p '' means that sz - polynomial has an attracting point ; * `` @xmath108 '' means that sz - polynomial has an attracting @xmath73-cycle ; * `` @xmath109 '' means that iterations of @xmath2 converge to attracting 2-cycle and iterations of @xmath1 to attracting @xmath110-cycle .
\1 ) @xmath111 . @xmath112
the polynomial @xmath38 is of g3-type .
julia set here is similar to julia set in figure 2 .
\2 ) @xmath113 . @xmath114
the polynomial @xmath38 is of g2-type .
julia set here is similar to julia set in figure 4 .
\3 ) @xmath115 .
@xmath116 polynomial @xmath38 is of g1-type .
convergence rate is quite good : @xmath117 .
\4 ) @xmath118 .
@xmath119 polynomial @xmath38 is of g1-type .
convergence rate is weak : @xmath120 .
\5 ) @xmath121 .
@xmath122 polynomial @xmath38 is of g1-type .
\6 ) @xmath123 .
@xmath124 polynomial @xmath38 is of g3-type .
here we have a high rate of convergence to the attracting cycle : the product of derivatives in cycle points is around @xmath125 .
@xmath126{z7_4_32}}\ ] ] figure 3 \7 ) @xmath127 . @xmath128
the polynomial @xmath38 is of g1-type .
\8 ) @xmath129 . @xmath130
the polynomial @xmath38 is of g2-type .
here we have a medium rate of convergence to the attracting cycle : the product of derivatives in cycle points is around @xmath131 .
@xmath126{z7_322_3}}\ ] ] figure 4 \9 ) @xmath132 . @xmath133
the polynomial @xmath38 is of g4-type .
\10 ) @xmath132 . @xmath134
the polynomial @xmath38 is of s2-type .
\11 ) @xmath135 . @xmath136
the polynomial @xmath38 is of g1-type .
\12 ) @xmath137 . @xmath138
the polynomial @xmath38 is of s3-type .
@xmath139{z7_32_222}}\ ] ] figure 5
there are five trees whose sz - polynomials have an attracting point , i.e. are of g1-type . @xmath140
@xmath141 there are three trees whose sz - polynomials has an attracting 2-cycle , i.e. are of g2-type .
@xmath142 next three cases are more interesting . 1 .
polynomial @xmath38 is of s1-type . 2 .
polynomial @xmath38 is of g3-type and its julia set is visually interesting .
@xmath139{z8_422_22 } } \quad\ ] ] figure 6 3 .
it is an interesting example of g3-type polynomial @xmath38 that is not defined over @xmath143 , i.e. where @xmath6 is not mirror symmetric .
if the passport is relatively simple , then sz - polynomials can be computed for trees with big number of edges . here
are some examples . passport @xmath144 . if @xmath145 , then julia set is totally disconnected .
otherwise : @xmath146 passport @xmath147 . if @xmath148 , then julia set is totally disconnected .
otherwise : @xmath149 passport @xmath150 . if @xmath151 then @xmath152 . if @xmath153 , then julia set is totally disconnected .
otherwise : @xmath154 passport @xmath155 .
@xmath156 @xmath157 @xmath158 this example demonstrates that when we consider a set of trees with the same passport , then almost all trees in this set have totally disconnected julia sets , but for `` nearly symmetric '' trees this set is connected .
further work in this field is related to the following problems
. 99 douady a. , hubbard j. , _ exploring the mandelbrot set .
the orsay notes .
_ eremenko a. , lyubich m. , _ the dynamics of analytic transformations _ , leningrad math
. j. , 1990 , 1(3 ) , 563 - 634 .
falconer k. , fractal geometry , wiley , 2003 .
jenkinson o. , pollicott m. , _ calculating hausdorff dimension of julia sets and kleinian limit sets _ ,
, 2002 , 124(3 ) , 495 - 545 .
lando s. , zvonkin a. , graphs on surfaces and their applications , springer - verlag , 2004 .
peitgen h .- o . ,
richter p.h . ,
the beauty of fractals , springer - verlag , 1986 .
zapponi l. , _ the arithmetic of prime degree trees _ ,
notices , 2002(4 ) , 211 - 219 . | a tree , embedded into plane , is a dessin denfant and its belyi function is a polynomial shabat polynomial .
zapponi form of this polynomial is unique , so we can correspond to an embedded tree the julia set of its shabat - zapponi polynomial . in this purely experimental work
we study relations between the form of a tree and properties ( form , connectedness , hausdorff dimension ) of its julia set . |
we study theoretically light propagations at the zigzag edge of a honeycomb photonic crystal consisting of dielectric rods in air , analogous to graphene . within the photonic band gap of the honeycomb photonic crystal ,
a unimodal edge state may exist with a sharp confinement of optical fields .
its dispersion can be tuned simply by adjusting the radius of the edge rods . for the edge rods with a graded variation in radius along the edge direction ,
we show numerically that light beams of different frequencies can be trapped sharply in different spatial locations , rendering wideband trapping of light .
graphene , a flat monolayer of graphite with carbon atoms arranged in a honeycomb lattice , has many interesting physical properties with great promise in nanoelectronics owing to its unique electronic properties , e.g. , conical dispersion ( dirac cones ) in the electronic band structure @xcite .
inspired by graphene , its photonic analog , namely , two - dimensional photonic crystals @xcite that possess dirac cones in photonic band structures , have drawn considerable attention in recent years @xcite . the existence of dirac cones may lead to many interesting optical phenomena such as conical diffraction and gap solitons @xcite , pseudodiffusive transport of light @xcite , zitterbewegung of photons @xcite and the realization of zero - refractive - index materials @xcite .
wideband trapping of light , also called `` rainbow '' trapping , is an optical phenomenon that light of different frequencies can be slowed down and stopped at different spatial positions in specially designed waveguides @xcite .
it has many potential applications ranging from optical information storage to enhanced light - matter interactions @xcite .
several structures for realizing wideband trapping have been proposed , including metamaterial waveguides @xcite , plasmonic waveguides @xcite , metal - coated optical nano waveguides @xcite and plasmonic gratings @xcite .
however , these structures may suffer inevitably a serious problem of optical loss due to constituting metals , which are rather lossy especially in the visible regime .
in contrast , dielectric photonic crystal waveguides can also slow light with a very small loss @xcite , which could be exploited in light trapping @xcite .
but they are usually multimodal @xcite , which may lead to undesirable mode coupling . by breaking the inversion symmetry of photonic crystals that possess dirac cones , a photonic band gap ( pbg ) around dirac points
may open up @xcite . with
the presence of edges , edge states within the pbg may appear with a sharp confinement of optical fields at the edges @xcite . with further breaking the time - reversal symmetry , confined edge states become even unidirectional , showing robustness against scattering or disorder @xcite . in this paper , we show theoretically that edge states in honeycomb photonic crystals ( hpcs ) can be exploited in realizing wideband trapping of light .
since no metal is introduced , the optical loss should be much lower than those in metamaterial and plasmonic waveguides .
compared with photonic crystal waveguides , our system is unimodal and more compact as well .
although exemplified by hpcs with rods being infinitely long , our idea can be applied to two - dimensional photonic crystals with finite rod lengths or membrane photonic crystals with drilled holes @xcite . , @xmath0 and @xmath1 representing points of high symmetry .
( c ) photonic band structures of hpcs .
red lines are those for @xmath2 , while blue lines for @xmath3 and @xmath4 .
the colored area stands for the pbg of the latter hpc .
( d ) pbg ( colored area ) map as a function of @xmath5 for fixed @xmath3 .
frequency is in units of @xmath6 , where @xmath7 is the lattice constant and @xmath8 is the wavelength in vacuum.,width=377 ] hpcs under study consist of dielectric rods which are arranged in a honeycomb lattice in air , analogous to graphene , as schematically shown in figure [ fig1](a ) .
the inter - distance between adjacent rods is denoted by @xmath9 . in an infinite hpc
, there are two different sets of rods whose radii are denoted respectively by @xmath10 and @xmath5 .
the radius of edge rods , which are changeable , is denoted by @xmath11 .
the lattice constant of an hpc is thus @xmath12 . without loss of generality ,
the dielectric constant of all rods is taken to be 11.4 , a typical value for silicon at telecommunication wavelengths . in our following discussions , we only consider the transverse magnetic polarization , i.e. , the case that the electric field is parallel to the rods . in figure [ fig1](c ) , photonic band structures of infinite hpcs , calculated by a plane - wave expansion method @xcite , are shown .
for an hpc possessing the inversion symmetry , i.e. , @xmath13 , dirac points exist where the first and second photonic bands cross each other at the @xmath1 point with a reduced frequency of @xmath14 . with broken inversion symmetry , i.e. , @xmath15
, the degeneracy at the @xmath1 point will be lifted and a pbg will open up @xcite .
the lower ( upper ) edge of the pbg is defined by the first ( second ) photonic band at the @xmath1 ( @xmath0 ) point . to attain an hpc with an optimal or desired pbg
, one can tune the radius of the rods or adjust the dielectric constant of the rods . in figure [ fig1](d ) , the pbg map as a function of @xmath5 for fixed @xmath3 is shown . for @xmath2 , there is no pbg due to the existence of the dirac points .
for @xmath15 , a pbg always exists . for @xmath3 and @xmath4 ,
the pbg ranges in reduced frequency @xmath6 from 0.26 to 0.343 . for @xmath3 and @xmath16 , the relative pbg width , @xmath17 , can be as high as @xmath18 , where @xmath19 is the pbg width and
@xmath20 is the mid - gap frequency .
( a ) , where @xmath21 is the wavevector along the edge direction in units of @xmath22 . the radius parameters for the hpc are @xmath3 and @xmath4 .
gray areas represent projected photonic bands .
gray , blue , and dark gray lines represent the edge states for the edge rods with different radius .
dotted straight lines stand for the light line in vacuum .
inset shows the cutoff frequency ( at @xmath23 ) as a function of @xmath11 .
two doted lines defines the range of the pbg .
( b)(d ) distributions of the electric field intensity @xmath24 for the edge state with @xmath25 at ( b ) @xmath26 , ( c ) @xmath27 , and ( d ) @xmath22 , marked by dots in ( a )
. doted circles stand for the rods.,width=377 ] figure [ fig2](a ) shows the dispersion of edge states of an hpc with a zigzag edge , calculated by the plane - wave expansion method @xcite . in our calculations ,
a supercell with a size of @xmath7 along the edge direction and @xmath28 along the direction perpendicular to the edge direction is used .
periodical boundary conditions are imposed in both directions .
the supercell contains a finite hpc with a thickness of @xmath29 along the direction perpendicular to the edge direction , corresponding to 48 rows of rods including the edge rods .
since the separation of the two zigzag edges are large enough , there is no coupling between the two edges .
the resulting dispersion can thus be considered as that for a single zigzag edge of a semi - infinite hpc .
projected photonic bands are obtained by projecting the bulk photonic band structure over the wavevector along the edge direction , @xmath21 .
in addition to photonic bands , for a given radius of the edge rods @xmath11 , an edge state may appear between the first and second photonic bands .
interestingly , this edge state is unimodal , whose dispersion depends on @xmath11
. even for different values of @xmath11 , the dispersion of the edge state has a similar behaviour : its frequency increases with increasing @xmath21 up to the brillouin zone boundary @xmath23 . at @xmath23 , it takes a zero slope . in other words
, the group velocity decreases with increasing @xmath21 and approaches zero at the brillouin zone boundary .
this interesting feature could be exploited to slow down and even stop light @xcite .
the frequency of an edge state at @xmath23 is dubbed cutoff frequency since the frequencies of the edge state at other wavevectors are always smaller than the cutoff frequency . for different values of @xmath11 ,
the corresponding cutoff frequency is different , showing a monotonic decrease with increasing @xmath11 , as shown in the inset of figure [ fig2](a ) .
note that edge states lie below the light line , implying that they are evanescent waves . in figs .
[ fig2](b)(d ) , the field distributions for the edge state with @xmath25 at different @xmath21 are shown .
obviously , optical fields are confined to the edge rods . for the edge mode at @xmath26 , it is near the first photonic band , leading to some bulk excitations in the vicinity of the edge , which can be clearly seen from figure [ fig2](b ) .
in contrast , for the edge modes away from the photonic bands there are nearly no bulk excitations .
as shown by us @xcite , these edge modes are of subwavelength , similar to surface plasmons bounded at metallic surfaces @xcite .
as aforementioned , an edge state at its cutoff frequency has a zero group velocity and the cutoff frequency decreases monotonically with increasing @xmath11 .
these features can render wideband trapping of light by taking advantage of edge states , provided that the edge rods are designed to have a spatially graded variation in radius .
if the grade is small enough , the graded edge can be considered approximately as a series of small edges , in each of which the edge rods possess a constant radius . as a result , each location seems to have a different cutoff frequency . for a wave propagating along such a graded edge
, its group velocity will be gradually reduced along the edge direction and eventually approach zero in a location where the corresponding cutoff frequency coincide with the frequency of the wave . in other words , the wave will stop around this location . for an hpc with a graded zigzag edge at different reduced frequencies ( a ) 0.276 , ( b ) 0.286 , and ( c ) 0.297 .
the radius parameters for the hpc are @xmath3 and @xmath4 .
a right - propagating gaussian beam of different frequencies is launched at @xmath30 .
circles denote dielectric rods .
the bottom ( top ) horizontal axis indicates the position ( radius of edge rods ) along the edge direction.,width=510 ] figure [ fig3 ] shows light trapping for different frequencies at the graded zigzag edge of an hpc , simulated by a finite - difference time - domain ( fdtd ) method @xcite .
the radii of the edge rods are assumed to vary linearly with the position of the edge rods , namely , @xmath31 , where @xmath32 is the position along the edge direction . in our fdtd simulations ,
a continuous gaussian beam with an intensity of 1 and a width of @xmath33 is launched at @xmath30 .
perfectly matched layer absorbing boundaries @xcite are used .
obviously , from figure [ fig3 ] light beams of different frequency can be trapped spatially in different locations .
the scenario is as follows . for a right - propagating gaussian beam
, it will propagate rightwards after launching .
the beam will be slowed down towards right since its group velocity decreases rightwards due to the reduction of the radius of the edge rods .
it will eventually be stopped around the location where the corresponding cutoff frequency of the location coincides with the beam frequency . by inspecting the beam with a reduced frequency of 0.286 , for example
, the maximum intensity of the standstill beam occurs around the position @xmath34 .
the radius of the edge rods around this location is about @xmath35 , very close to but a bit smaller than the radius of @xmath36 that offers a cutoff frequency of 0.286 [ see the inset of figure [ fig2](a ) ] .
for the standstill beam , there are some small additional peaks on the left side of the main peak .
this is because the right - propagating wave may get reflected more or less since the grade of the edge rods is not small enough .
the reflected and the right - propagating waves may interfere with each other , leading to these small additional peaks , which could be suppressed in principle by using a finer grade . in summary , we studied theoretically the edge states at the zigzag edge of hpcs consisting of dielectric rods in air for the transverse magnetic polarization .
with broken inversion symmetry , a pbg exists between the first and second photonic bands .
a unimodal edge state may appear within the pbg , whose dispersion and cutoff frequency can be tuned by changing the radius of the edge rods . for a zigzag edge with a gradually graded radius in edge rods , light beams of different frequencies
can be slowed down and trapped spatially in different locations .
our results show that edge states in hpcs could be exploited in optical storage and enhanced light - matter interactions . + this work was supported by the 973 program ( grant nos .
2011cb922004 , 2012cb921604 and 2013cb632701 ) and the nsfc . |
|
angular momentum boundary conditions boulder non - linear ( name of a company ) beam splitter charge - coupled device circular dichroism computed generated hologram electromagnetic focused ion beam generalized lorenz - mie theory half - wave plate first kerker condition second kerker condition left circularly polarised laguerre - gaussian linear polariser lookup table morphological - dependent resonance microscope objective numerical aperture quarter - wave plate right circularly polarised red green blue scanning electron microscope spatial light modulator stimulated emission depletion transverse electric transverse magnetic working distance whispering gallery mode `` growing up in the place i did i never was aware of any other option but to question everything . ''
[ intro ] invented by the sumerians in ancient mesopotamia around 3000 bc , @xcite , writing may be considered one of the greatest human inventions of all - times . since then , it has been used to store , transmit and manipulate information .
concurrently , the technology employed in writing has changed over the years .
the development of computers , perhaps one of mankind s most significant technological revolutions in the last century , was intrinsically linked to writing . despite the principles of modern computers having been described by alan turing in 1936 @xcite , the first programmable * electronic * computer was not built until 1946 , at pennsylvania university @xcite . in the following 6 years , john von neumann and co - workers laid the ground work upon which modern computers rest to this day @xcite .
the first generation of commercial electronic computers was made available in the us in 1951 , but the giant leap towards our current technologies occurred in 1955 , when the so - called ` second generation computers ' reached the market @xcite .
since that time , the evolution of computers has been linked to the advent of semiconductor electronics .
simply put , this is the technology associated with the control of how the electrons move in semiconductor materials
. the majority of our current information devices function because of our ability to control electron currents from one spot to another one , and re - interpret this current as chains of 0 s and 1 s , which are used to encode information . since the 1970 s , our means of re - interpreting electron currents as chains of 0 s and 1 s have remained fundamentally unchanged .
however , the techniques and devices to move electrons around in a controlled manner have changed drastically .
of particular note , the size of the electronic devices has decreased by many orders of magnitude .
this can be observed in figure [ fig11 ] . ] ` the past ' includes the electronic circuits ( using semiconductor electronics , or vacuum tubes ) used up until the 1980 s . from then onwards ,
the semiconductor manufacturing processes achieved sizes below 1@xmath1 m currently known as nano - electronics .
in fact , nano - electronics has advanced at a very fast pace , consistently following moore s law @xcite . at present , is fabricating transistors at 22 nm , and current projections are that the 11 nm technology will be obtained by 2015 .
+ + the fact that electronics has been so successful over the past 60 years does not imply that it is the ultimate technology to carry out digital writing or information processing .
* photonics * has been following a delayed but similar development .
the term photonics began to be used in the late 1960 s , once it became evident that photons could be used as bits of information in the same way as electrons were being used .
however , it did not reach the common use until the 1980 s , when the ieee lasers and electro - optics society established an archival journal named photonics technology letters .
some of the early crucial developments in the field were the invention of the laser in 1958 @xcite ; the diode laser in the 1962 @xcite ; the first demonstration of optical communication using optical fibres in 1966 , done at the university of ulm , germany @xcite ; and the invention of the optical fibre amplifier in the early 1980 s @xcite .
photonics presents some significant advantages over electronics .
the most important one is speed . due to its much larger carrier frequencies , photonics has a much larger bandwidth ( or data transmission rate ) than electronics .
this is depicted in figure [ fig12 ] , which shows that the operating speed of photonic devices can be up to five orders of magnitude higher than electronics . ]
another important feature of photons is their weak interaction with the environment .
photons can travel long distances almost without changing their state .
this implies much lower losses in transmission .
actually , this feature is also exploited in quantum information , in particular in the field of quantum cryptography @xcite to create secure protocols to data exchange . as a consequence ,
photonic integrated circuits have been gaining a lot of attention in recent times @xcite .
the most ambitious goal is to eventually usurp the predominant role of electronic integrated circuits in information devices .
however , the goal seems to be unrealistic at the moment .
photonic circuits suffer from a lack of storage capacity , _
i.e. _ it is difficult to stop and store light .
hence , the storage needs to be done in an electronic medium . moreover , integrated electronics is non - linear , whereas non - linear integrated photonics still needs to evolve to be competitive in the market .
last but not least , the typical size of photonic components is limited by the wavelength of light . due to the diffraction limit of light , dielectric photonic circuit can not resolve or retrieve information smaller than a fraction of the wavelength ( 200 nm approximately ) @xcite , as opposed to nano - electronics where 22 nm transistors are already a reality .
+ + most of the efforts to overcome the diffraction limitations in photonics have been happening in the field of plasmonics .
in fact , lots of authors consider that nanophotonics should be based on metallic nano - plasmonics ( see figure [ fig13 ] ) @xcite . ]
plasmonics is the science that studies the interaction between light and free electrons on a metal .
the first theoretical studies in plasmonics were done in the 1950 s @xcite , and the first experiments with metallic surfaces in the 1970 s @xcite . in general , the strategy followed in plasmonics to bring photonics down to the nano - scale has a sample - based perspective .
that is , given a fixed external excitation beam ( see inset in figure [ fig14](a ) for a typical example ) , samples are engineered ( shape and materials ) so that they produce certain effects .
, width=548 ] for example , in figure [ fig14](a ) , an array of metallic nano - rods is fabricated to guide light from one location to another .
similarly , in figure [ fig14](b ) a circular nano - hole drilled in a gold film deposited on top of a glass layer is inserted into a liquid chamber .
the excitation of the structure enabled the authors of @xcite to trap 50 nm particles that had been dissolved in the water solution .
there is no doubt that fabrication and characterization of samples play a huge role in the current plasmonic technologies .
interestingly , the spatial profile of the incident beam also plays a very important role .
however , most of the work in the field is done with gaussian beams ( the beam on the inset of figure [ fig14 ] ) . in this thesis
, it will be shown that more elaborate beams of light can be used to retrieve plenty of additional information from nano - structures .
an illustrative example is the stimulated emission depletion ( sted ) microscopy .
sted microscopy was invented by stefan hell and co - workers in 1994 @xcite .
it is one of the so - called super - resolution microscopy methods , as it can resolve defects as tiny as 30 nm .
a very complete schematic of its working principle is shown in figure [ fig15 ] , and its working principle can be found in @xcite . ] even though figure [ fig15 ] is complicated , the message should be clear . probing a sub - wavelength specimen with a gaussian beam ( green beam in figure [ fig15 ] ) results in a blurry image ( figure [ fig15](c ) ) . nevertheless , the combined use of a gaussian and a doughnut - shaped beam ( red ) results in a much neater image ( see figure [ fig15](d ) ) .
there are different kinds of doughnut - shaped beams , but the ones used in sted are called vortex beams @xcite .
+ + vortex beams are defined by their topological charge @xmath2 , which is an integer number .
the topological charge @xmath2 accounts for the number of times that the phase of the beam wraps around its centre in a @xmath3 circle .
examples of four different vortex beams are shown in figure [ fig16 ] . .
for the intensity plots , red indicates maximum and dark blue minimum . for the phase ,
white means @xmath4 phase , and black means @xmath3 .
[ fig16 ] ] it can be seen that when @xmath5 , the phase goes from @xmath4 to @xmath3 one time in counter - clockwise direction .
in contrast , when @xmath6 , the phase goes three times from @xmath4 to @xmath3 in a clockwise direction . in chapter [ ch5 ] , sections [ ch5_cgh ] and [ ch5_slm ]
have been dedicated to explain how to characterize and generate vortex beams in the laboratory .
note that the definition of vortex beam and topological charge is independent of the polarisation .
however , for many applications in nano - photonics ( sted microscopy is one of them ) , vortex beams need to be tightly focused .
when that is the case , vortex beams become much more complex .
in fact , in that regime , the definition of the phase singularity is polarisation - dependent .
this is what many authors have described as one of the manifestations of the spin - to - orbit conversion @xcite .
then , characterising vortex beams in terms of their symmetries becomes enormously useful .
it allows for a systematic study of the beam without the need of considering different focusing regimes .
+ + sections [ ch1_basis ] and [ ch1_multipoles ] are dedicated to a systematic characterisation of beams in terms of their symmetries .
thanks to the understanding and analysis of the different transformations under which light beams are symmetric , different results have been obtained , both theoretically and experimentally . in the succeeding chapters , three symmetries
have been specially studied : mirror symmetry , rotational symmetry , and duality @xcite . both samples and beams of light which are ( or are not )
symmetric under these transformations have been used and as a consequence new fundamental phenomena have been found . in chapter
[ ch3 ] , the interaction between single spheres and cylindrically symmetric beams has been theoretically studied . among others ,
a technique to excite whispering gallery modes ( wgms ) has been developed . in chapter
[ ch4 ] , the cylindrical symmetry of a single sphere and the duality symmetry of a light beam have been used to cancel the forward scattering of a sample .
furthermore , it has been proven that a sample can effectively be dual in a controlled manner if its dimensions are changed in a certain way with respect to an incident cylindrically symmetric beam . in chapter
[ ch6 ] , a mirror and cylindrically symmetric sample has been used to induce a giant circular dichroism ( cd ) . again , the use of dual , cylindrically and mirror symmetric modes has been a key point to be able to achieve it . finally , chapter [ ch7 ]
merges all the knowledge accumulated in the previous chapters , both theoretical and experimental . using various dual cylindrically symmetric beams , the resonant scattering behaviour of a dielectric sphere is unveiled .
this resonant behaviour is hidden under the gaussian beam ( or plane wave ) excitation , and the use of vortex beams is crucial to find it .
due to time constraints , this last study is not as thorough as the others presented in the previous chapters . nevertheless , it allowed for an experimental observation of some of the theoretical predictions of the thesis .
+ + now , even though the details of what has been done in each chapter may be too technical for a non - specialised reader , these should only be considered to emphasise the point made before : nanophotonics can also be done from another perspective different from the prevailing sample - based view . in particular , the consideration of the sample and the exciting light as a whole system , along with symmetry considerations , allows for the discovery of new phenomena at the nano - scale .
the future in this unexplored direction looks as exciting as promising . in the next chapters ,
the reader will find a taste of it .
`` if i could explain it to the average person , i would nt have been worth the nobel prize . ''
[ ch1 ] the material presented in this thesis can be described using the tools of em field theory @xcite . in particular , the subdivisions of em field theory that this thesis especially deals with are electromagnetic optics and vectorial scattering theory @xcite . in this chapter , i will introduce all the necessary concepts and tools to understand the theoretical developments scattered across the next chapters . for completeness
, the starting point will be the macroscopic maxwell equations for a source - free medium @xcite : @xmath7 where @xmath8 and @xmath9 are the electric and magnetic fields , @xmath10 is the electric displacement , and @xmath11 is the magnetic induction .
the relations between @xmath10 and @xmath8 , as well as between @xmath11 and @xmath9 are given by the properties of the medium : @xmath12 with @xmath13 being the electric permittivity and magnetic permeability of vacuum .
@xmath14 is defined as the polarization density , which is the density of molecular dipole moments induced by the electric field @xcite .
similarly , @xmath15 , defined as the magnetization density , is the density of molecular magnetic dipole moments induced by the magnetic field .
i will be particularly interested in linear , non - dispersive , homogeneous and isotropic media . in these cases
, a linear relation can be established between @xmath14 and @xmath8 , and therefore the following relation holds @xcite : @xmath16 where @xmath17 is a scalar quantity defining the electrical properties of the medium called electric permittivity . in the same way , a linear relation can be cast between @xmath15 and @xmath9 , yielding the following linear relation between @xmath11 and @xmath9 @xmath18 with @xmath1 being the so - called magnetic permeability constant of the medium
hence , the macroscopic maxwell equations for a linear , non - dispersive , homogeneous , isotropic and source - free medium can be written only as functions of @xmath8 and @xmath9 @xcite : @xmath19 here , i will be especially interested in the monochromatic solutions of these equations .
that is , i will suppose that all the fields have a @xmath20 time dependence , with @xmath21 being the frequency of light .
then , any field can be expressed as the real part of a product of a purely temporal part and a purely spatial part : @xmath22 $ ] , where @xmath23 represents a complex - amplitude vector .
then , equations ( [ e_max_eh ] ) can be re - written just in terms of their spatially dependent parts : @xmath24 now , applying the curl operation to equations ( [ e_max_eh_mono ] ) and making use of the vector identity @xmath25 , one obtains that both @xmath26 and @xmath27 fulfil the vectorial helmholtz equation @xcite : @xmath28 where @xmath29 is known as the wave - number .
nonetheless , equations ( [ e_max_eh_mono ] , [ e_max_helm ] ) can not be uniquely solved unless some boundary conditions ( bc ) are applied .
firstly , the em fields must have a proper decay at infinity and be finite everywhere else .
secondly , when equations ( [ e_max_eh_mono ] ) are applied between two different media , the following bc need to hold : @xmath30 where @xmath31 is the surface charge density at the surface ; @xmath32 is the surface current density ; @xmath33 is the unitary normal vector to the interface between 1 and 2 ; and @xmath34 and @xmath35 account for the electric and magnetic fields at two different media .
it is observed that the tangential components of @xmath26 and @xmath27 must be continuous , while the normal components of @xmath36 and @xmath37 can be discontinuous depending on the properties of the surface .
equations ( [ e_max_eh_mono ] ) plus the bc listed above will be the framework in which this thesis will be grounded on .
in the previous section , i have introduced the monochromatic maxwell equations for linear , non - dispersive , homogeneous , isotropic and source - free media ( see equation ( [ e_max_eh_mono ] ) ) .
i will impose that all the theoretical developments done in this thesis fulfil equations ( [ e_max_eh_mono ] ) .
that is , i will require that any meaningful * em field * fulfils equations ( [ e_max_eh_mono ] ) , bcs ( [ e_bc ] ) , it decays properly at infinity and it is finite everywhere else .
+ + another framework to do optical physics is the paraxial equation @xcite , which is an approximation of equations ( [ e_max_eh_mono ] ) but it is much simpler to use . in paraxial optics , a complex - amplitude electric field @xmath38 can be expressed as @xmath39 , with @xmath40 being a complex - wave amplitude scalar function that must fulfil the paraxial equation @xmath41 and @xmath42 being an arbitrary polarization vector .
then , it is clear that in paraxial optics the vectorial character of the field ( given by @xmath42 ) is decoupled from its spatial properties ( given by @xmath40 ) .
the importance of the paraxial equation is that collimated beams , which are very frequently used in the laboratory , can be accurately described within this framework . in those beams , if the propagation direction of the beam is the @xmath43 axis , a plane wave decomposition of the wave yields partial waves whose @xmath44 , and @xmath45 ( see [ ch1_planewaves ] to check the definitions of @xmath46 and @xmath47 ) .
this implies that the complex amplitude function @xmath40 can be expressed as @xmath48 , where @xmath49 is a magnitude called the envelope of the wave which varies slowly with @xmath43 : @xmath50 however , the paraxial approximation does not properly work in nano - optics . in nano - optics ,
light beams are either tightly focused or interact with sub - wavelength objects . in those cases ,
the polarization degrees of freedom of the beams can not be decoupled from their spatial properties @xcite .
then , imposing that all em fields must be solutions of the full maxwell equations rules out all those beams that fulfil the paraxial equation , but are not solutions of the full maxwell equations given by equation ( [ e_max_eh_mono ] ) @xcite . + +
now , i will define an * operator * as an entity that picks up any em field fulfilling equations [ e_max_eh_mono ] and transforms it into a different one that still fulfils equations [ e_max_eh_mono ] .
finally , i will define a * symmetry group * as a set of operators with group structure .
a group @xmath51 is defined as a set of elements @xmath52 along with an inner product .
the inner product ( @xmath53 ) is such that the product of two arbitrary elements of the group yields an element of the group @xcite .
furthermore , the product must satisfy the following conditions @xcite : * it must be associative , _
i.e. _ @xmath54 , @xmath55 .
* there must be an element @xmath56 , called the identity , which is such that @xmath57 , @xmath58 .
* for each @xmath59 , there must exist an element @xmath60 , called the inverse of @xmath61 , such that @xmath62 .
groups can be subdivided into two categories - discrete , and continuous . in physics ,
most of continuous groups of interest are part of a subdivision called ` linear lie groups ' @xcite . linear lie groups can be labelled by one or more continuous variables .
in fact , each label is related to a generator .
the generators of the group define the local behaviour of the group near the identity , and any element of the group can be obtained as a smooth function of the generator . for example , for the @xmath63 group of 2-dimensional rotations , any rotation @xmath64 can be obtained as @xmath65 , with @xmath66 the generator of the group .
note that this implies that if the element of the group is an operator , so must be the generator .
in fact , group generators are very good candidates to compose a complete set of commuting operators .
a complete set of commuting operators of a vector space is a set of operators that commute with each other and whose eigenvalues completely specify the state of a system @xcite .
the idea is that if an em field is an eigenvector of a generator of a symmetry , then it is also invariant under the symmetry transformation generated by that generator , and vice versa @xcite . then , using generators to make a set of commuting operators will be useful as the fields will be characterized in terms of their symmetries .
now , i will differentiate between em fields and * em modes*. em fields are solutions of maxwell equations ( [ e_max_eh_mono ] ) , whereas em modes are elements of a basis of solutions of maxwell equations ( [ e_max_eh_mono ] ) .
thus , in general , any em field can be decomposed into em modes .
in addition , em modes will be chosen so that they are eigenstates of some generators , _
i.e. _ they will be invariant under certain symmetries .
next , i will list the most common operators that i will use to describe em modes and their differential representation will be given .
some of them are generators of symmetries , so in those cases i will also comment on the symmetries that they give rise to . * , @xmath67 .
the hamiltonian is the generator of temporal translations ( or time evolution ) @xcite .
that is , the transformation @xmath68 applied to an em field @xmath69 the state evolve from @xmath70 to @xmath71 : @xmath72 = \psi(\mathbf{r},t+\delta t)\ ] ] * @xmath73 .
this is a vectorial operator , _
i.e. _ it is composed of three operators in three different directions : @xmath74 .
its three components commute with each other @xmath75 = 0 $ ] .
each of the components generates a translation in its corresponding axis .
hence , the vectorial addition of all of them is the generator of linear translations in any direction of space : @xmath76 @xcite .
that is , @xmath77 = \psi(\mathbf{r } + \delta \mathbf{r } , t)\ ] ] * , @xmath78 , with @xmath79 and @xmath80 respectively , where @xmath81 is the total anti - symmetric tensor and @xmath82 .
it is also composed of three operators in three components , @xmath83 , but they do not commute with each other : @xmath84 = i \epsilon_{ijk } j_k$ ] .
the reason why the three components of the angular momentum ( am ) @xmath85 do not commute is that they are the generators of rotations on three different axis , and rotations do not commute ( whereas spatial translations do , see above ) @xcite .
a general rotation of an angle @xmath86 along an axis on the direction @xmath33 is given by : @xmath87 .
then , the application of a rotation onto a general em field yields : @xmath88 = \mathbf{m}_{{\mathbf{\hat{n}}}}(\varphi ) \cdot \psi(\mathbf{m}_{{\mathbf{\hat{n}}}}^{-1}(\varphi ) \cdot \mathbf{r},t ) \label{e_rotpsi}\ ] ] where @xmath89 is a rotation matrix @xcite ( see appendix [ appendix ] ) . finally , note that even though the @xmath90 and @xmath91 operators used to define @xmath85 are usually called orbital and spin am respectively , they are not proper am momentum operator on their own @xcite .
none of them generates proper rotations , only their addition does .
furthermore , they are not proper operators , as their application to an em field fulfilling maxwell equations yields a field that does not fulfil maxwell equations @xcite . * , @xmath92 , where @xmath93 have been defined in iii ) .
@xmath94 is called a casimir operator , and commutes with all the operators . even though it does not generate any meaningful transformation , it is very useful for problems with spherical symmetry . * , @xmath95 .
the helicity operator is defined as the projection of the am onto the direction of the linear momentum .
it was proven in @xcite that @xmath96 generates generalized duality transformations @xcite : @xmath97 .
generalized duality transformations mix the electric and magnetic parts of an em field . that is , if we express @xmath98 as @xmath99 , then @xmath100 = d_{\varphi } \left [ \left ( \begin{array}{l } { \mathbf{e}}(\mathbf{r},t ) \\ { \mathbf{h}}(\mathbf{r},t )
\end{array } \right ) \right ] = \left ( \begin{array}{l } { \mathbf{e}}(\mathbf{r},t ) \cos \varphi - { \mathbf{h}}(\mathbf{r},t ) \sin \varphi \\ { \mathbf{e}}(\mathbf{r},t )
\sin \varphi + { \mathbf{h}}(\mathbf{r},t ) \cos \varphi \end{array } \right)\ ] ] its differential representation for monochromatic fields is @xmath101 .
unlike all the rest of symmetries listed hereby , duality symmetry is a non - geometrical symmetry .
it can be proven that its preservation only depends on the material properties of the medium in consideration @xcite
. a very thorough study of duality symmetry and its generator ( helicity ) can be found in @xcite .
* , @xmath102 .
parity transformations @xmath103 are discrete , therefore they are not generated by any other operator .
the application of @xmath102 onto a maxwell field yields @xcite : @xmath104 = \psi(-\mathbf{r},t)\ ] ] * , @xmath105 .
a mirror transformation of the kind @xmath105 indicates that the spatial coordinates are reflected off a plane given by its normal vector @xmath33 .
the difference is that in @xmath105 the reflection is off a plane whose normal vector is @xmath33 , whereas @xmath106 indicates that the reflection is done upon a plane that contains the axis @xmath33 . ] .
actually , it can be re - written as @xmath107 , _
i.e. _ a mirror symmetry operator can be obtained by multiplying a parity and rotation operators . like parity , it is a discrete transformation , so it is not generated by any other operator .
its application onto @xmath98 gives @xcite : @xmath108 = \psi(\mathbf{r } - 2 { \mathbf{\hat{n}}}({\mathbf{\hat{n}}}\cdot \mathbf{r}),t)\ ] ] last but not least , i will give the most useful commutation rules between the operators listed above @xcite .
these commutation rules are crucial in order to define a set of commuting operators . in fact
, it is known that only four commuting operators are needed to describe em fields , as they are elements of the poincar group @xcite : * @xmath109 commutes with all the other operators of the list above .
* @xmath110 commutes with @xmath109 , @xmath94 , @xmath96 , and @xmath105 .
it does not commute with @xmath85 , their commutator is @xmath111 = i \epsilon_{ijk}p_k$ ] .
it anti - commutes with parity @xmath112 . * @xmath85 commutes with @xmath109 , @xmath94 , @xmath96 , @xmath102 and @xmath105 ( when @xmath33 is parallel to @xmath85 )
. it does not commute with @xmath113 , in fact it anti - commutes : @xmath114 . * @xmath94 commutes with all the rest of operators , as it is a casimir operator . *
@xmath96 commutes with @xmath109 , @xmath110 , @xmath85 and @xmath94 .
it anti - commutes both with @xmath102 and @xmath105 .
* @xmath102 commutes with @xmath109 , @xmath85 , @xmath94 , and @xmath105 .
it anti - commutes with @xmath96 and @xmath110 .
the aim of this chapter is to establish the basis on which the works presented on this thesis are grounded . in section [ ch1_hilmax
] , i have introduced maxwell equations .
then , in section [ ch1_symm ] , i have discussed that i will require all the em fields to fulfil maxwell equations in its form given by equations ( [ e_max_eh_mono ] ) . furthermore , some operators and generators have been introduced . in this section and the next one , six different sets of em modes that stem from six different sets of complete commuting operators will be described .
each of these sets of em modes will be especially suited to describe some particular em interactions .
in fact , since generators of symmetries will be used as the operators in a set , a set of modes will be chosen over another depending on the symmetries of the interaction .
there are different ways of constructing these sets of em modes . from a symmetries point of view , probably the clearest technique is ` the induced representation method ' .
wu - ki tung describes it in chapter 9 of his book @xcite . as in most of the cases ,
the use of symmetries simplifies the amount of operations tremendously .
unfortunately , this method is not very common in optics .
hence , another equivalent method much better - known in the optics community will be used here .
this method is based on some mathematical theorems about the solutions of the vectorial helmholtz equation ( [ e_max_helm ] ) .
the method is described in @xcite among others .
it is based on the following property .
given a general solution of the scalar helmholtz equation in a system of coordinates where the solution is in separated variables @xmath115 , then it can be proven that if @xmath116 is a unitary vector in a privileged direction is valid .
some conditions apply .
these conditions can be found in @xcite .
] , @xmath117 \label{e_mn } \end{array}\ ] ] are general solutions of the vectorial helmholtz equation and fulfil the divergence equations @xmath118 @xcite .
that is , any @xmath26 field can be obtained as a superposition of the modes @xmath119 and @xmath120 defined in equation ( [ e_mn ] ) .
then , the magnetic field @xmath27 can be derived from maxwell equations ( [ e_max_eh_mono ] ) can always be derived from @xmath26 , i will only focus on the properties of @xmath26 , and derive @xmath27 using equations ( [ e_max_eh_mono ] ) . besides , any symmetry considerations regarding @xmath26 will also apply to @xmath27 , except for parity and mirror inversion .
it can be shown that when @xmath26 is odd , @xmath27 is even and viceversa @xcite ] .
note that i have differentiated the behaviour of the two general solutions by its te or tm character .
the notation of te / tm behaviour is a bit arbitrary , as both the electric and magnetic fields are ` transverse ' in the maxwell sense , _
i.e. _ @xmath118 . in fact , this notation gets even more cumbersome when spherical coordinates are used , as te modes are usually referred to as ` magnetic ' and tm as electric. here , i will stick to the most common notation and denote the first solution in equation ( [ e_mn ] ) as te and the second one as tm for cartesian and cylindrical coordinates @xcite .
then , in spherical coordinates i will refer to the first solution as magnetic and the second one as electric @xcite , as that is the dominant notation .
it can be shown that @xmath121 is a general solution of the scalar helmholtz equation for cartesian coordinates @xcite , with @xmath122 being the position vector and @xmath123 the wave - vector , whose modulus @xmath124 is the wave - number . then , using @xmath125 as the symmetry direction , the two following vectorial solutions arise : @xmath126 { \exp(i \textbf{k } \cdot \textbf{r})}\end{aligned}\ ] ] where @xmath127 are the polarization vectors in cartesian directions .
these te and tm modes are very well - known in optics and are referred to as @xmath128 and @xmath129 plane waves .
in fact , they define a trihedron such that @xmath130 .
as previously mentioned , all the fields considered here are monochromatic , therefore both @xmath128 and @xmath129 waves will be eigenvectors of the @xmath131 operator .
in addition , it can be checked that they are eigenvectors of the following operators : @xmath132 & = & k_x \ { \widehat{\mathbf{s } } } { \exp(i \textbf{k } \cdot \textbf{r } ) } & \qquad p_x \left [ { \widehat{\mathbf{p } } } { \exp(i \textbf{k } \cdot \textbf{r})}\right ] & = & k_x \ { \widehat{\mathbf{p } } } { \exp(i \textbf{k } \cdot \textbf{r})}\\ p_y \left [ { \widehat{\mathbf{s } } } { \exp(i \textbf{k } \cdot \textbf{r})}\right ] & = & k_y \ { \widehat{\mathbf{s } } } { \exp(i \textbf{k } \cdot \textbf{r } ) } & \qquad p_y \left [ { \widehat{\mathbf{p } } } { \exp(i \textbf{k } \cdot \textbf{r})}\right ] & = & k_y \ { \widehat{\mathbf{p } } } { \exp(i \textbf{k } \cdot \textbf{r})}\\ m_{{\widehat{\mathbf{s } } } } \left [ { \widehat{\mathbf{s } } } { \exp(i \textbf{k } \cdot \textbf{r})}\right ] & = & - \ { \widehat{\mathbf{s } } } { \exp(i \textbf{k } \cdot \textbf{r } ) } & \qquad m_{{\widehat{\mathbf{s } } } } \left [ { \widehat{\mathbf{p } } } { \exp(i \textbf{k } \cdot \textbf{r})}\right ] & = & + \ { \widehat{\mathbf{p } } } { \exp(i \textbf{k } \cdot \textbf{r})}\label{e_sp_eigen } \end{array}\ ] ] then , @xmath128 and @xmath129 are em modes invariant under temporal translations , spatial translations along @xmath133 and @xmath134 axis , and mirror symmetry transformations .
hence , they are eigenvectors of @xmath131 , @xmath135 , @xmath136 , and @xmath137 . now
, if we look at equations ( [ e_mn ] ) , we can see that the @xmath120 mode ( @xmath129 wave ) is obtained by applying @xmath138 to the @xmath119 mode ( @xmath128 wave ) . in section [ ch1_symm
] , i have shown that the operation @xmath138 is actually the differential form of the helicity operator for monochromatic fields : @xmath139 then , it can immediately be seen that : @xmath140 & = & { \widehat{\mathbf{p } } } { \exp(i \textbf{k } \cdot \textbf{r})}\\ { \lambda } \left [ { \widehat{\mathbf{p } } } { \exp(i \textbf{k } \cdot \textbf{r})}\right ] & = & { \widehat{\mathbf{s } } } { \exp(i \textbf{k } \cdot \textbf{r})}\end{aligned}\ ] ] thus , additions and subtractions of @xmath128 and @xmath129 waves yield states with a well - defined helicity : @xmath141 & = & + \left [ { \dfrac{{\widehat{\mathbf{s } } } + { \widehat{\mathbf{p } } } } { \sqrt{2 } } { \exp(i \textbf{k } \cdot \textbf{r})}}\right ] \\ { \lambda } \left [ { \dfrac{{\widehat{\mathbf{s } } } - { \widehat{\mathbf{p } } } } { \sqrt{2 } } { \exp(i \textbf{k } \cdot \textbf{r})}}\right ] & = & - \left [ { \dfrac{{\widehat{\mathbf{s } } } - { \widehat{\mathbf{p } } } } { \sqrt{2 } } { \exp(i \textbf{k } \cdot \textbf{r})}}\right ] \end{aligned}\ ] ] i will denote these combinations as @xmath142 , with @xmath143 depending on the value of helicity : @xmath144 \\ { \mathbf{\hat{e}}_-}= \dfrac{{\widehat{\mathbf{s } } } - { \widehat{\mathbf{p } } } } { \sqrt{2 } } = \dfrac{1}{2 } \left [ \left(1 + \dfrac{k_z}{k } \right ) e^{i \phi_k } \ { \mathbf{\hat{\sigma}_-}}+ \left(\dfrac{k_z}{k } -1 \right ) e^{-i \phi_k } \
{ \mathbf{\hat{\sigma}_+}}- \dfrac{\sqrt{2}{k_{\rho}}}{k } \ { \mathbf{\hat{z}}}\right]\end{aligned}\ ] ] where @xmath46 and @xmath145 are the wave - vector components in cylindrical coordinates , _
i.e. _ @xmath146 , @xmath147 and @xmath148 are the circular polarizations vectors : @xmath149 is the left circular polarization and @xmath150 is the right one .
then , a new basis of em modes can be created with @xmath151 and @xmath152
. this will be symmetric under temporal translations , spatial translations along @xmath133 and @xmath134 axis , and generalized duality transformations : @xmath153 & = & k_x \
{ \mathbf{\hat{e}}_+}{\exp(i \textbf{k } \cdot \textbf{r } ) } & \qquad p_x \left [ { \mathbf{\hat{e}}_-}{\exp(i \textbf{k } \cdot \textbf{r})}\right ] & = & k_x \ { \mathbf{\hat{e}}_-}{\exp(i \textbf{k }
\cdot \textbf{r})}\\ p_y \left [ { \mathbf{\hat{e}}_+}{\exp(i \textbf{k } \cdot \textbf{r})}\right ] & = & k_y \ { \mathbf{\hat{e}}_+}{\exp(i \textbf{k } \cdot \textbf{r } ) } & \qquad p_y \left [ { \mathbf{\hat{e}}_-}{\exp(i \textbf{k } \cdot \textbf{r})}\right ] & = & k_y \ { \mathbf{\hat{e}}_-}{\exp(i \textbf{k } \cdot \textbf{r})}\\ \lambda \left [ { \mathbf{\hat{e}}_+}{\exp(i \textbf{k } \cdot \textbf{r})}\right ] & = & + \ { \mathbf{\hat{e}}_+}{\exp(i \textbf{k } \cdot \textbf{r } ) } & \qquad \lambda \left [ { \mathbf{\hat{e}}_-}{\exp(i \textbf{k } \cdot \textbf{r})}\right ] & = & - \ { \mathbf{\hat{e}}_-}{\exp(i \textbf{k } \cdot \textbf{r})}\label{e_ep_eigen } \end{array}\ ] ] finally , it can be checked that the two sets of em modes given here are normalized in the following way : @xmath154 where @xmath155 is a dirac delta , and @xmath156 is a kronecker delta @xcite .
cylindrically symmetric interactions will play a very significant role in this thesis .
therefore , a basis that exploits this symmetry needs to be found . a way of attacking the problem is to use equations ( [ e_mn ] ) in cylindrical coordinates with @xmath125 as well .
this method , used by @xcite and several others , naturally gives rise to the so - called te / tm bessel beams .
nevertheless , the symmetries of the problem are a bit blurred by the methodology
. then , in order to highlight the symmetries of the bessel beams , i will use an alternative method to derive them .
this method is based on the fact that any field can be decomposed into plane waves .
then , the superpositions are chosen so that the symmetry requirements are fulfilled .
a bessel beam can be defined as an em mode symmetric under temporal translations , spatial translations and rotations along a symmetry axis ( usually , the @xmath43 is chosen ) .
this implies that bessel beams need to be eigenstates of @xmath131 , @xmath157 , and @xmath158 .
as mentioned earlier , since i am only considering monochromatic fields , they are immediately eigenstates of @xmath131 . then , having a set @xmath21 fixes a relation between @xmath47 and @xmath46 : @xmath159 .
therefore , the @xmath43 component and the transverse component of the linear momentum are linked via the frequency .
thus , having a well - defined @xmath157 is analogous of having a well - defined @xmath160 .
consequently , adding plane waves on a circular cone symmetric under the @xmath43 axis will yield a field with a well - defined @xmath160 .
then , i will be able to find the conditions on the superposition coefficients by imposing that the beam is an eigenvector of @xmath158 . to start ,
let me re - write the four plane waves derived in the previous sub - section in terms of rotations ( see appendix [ appendix ] ) : @xmath161 where @xmath162 is the operator of rotations defined in appendix [ appendix ] .
then , due to the symmetry considerations given by equations ( [ e_sp_eigen ] , [ e_ep_eigen ] ) , a superposition of @xmath128 waves will yield a mode with a well - defined frequency and a defined te / tm character . in general , it will no longer be an eigenstate of @xmath135 and @xmath136 , but if we only add @xmath128 waves on a cone , as mentioned in the paragraph above , we can retrieve a mode with a well defined @xmath157 ( or @xmath160 ) .
this can be given by the general superposition : @xmath163 where @xmath164 . by construction , the field @xmath165 is an eigenstate of @xmath131 and @xmath160 . in order to make it a bessel beam , a right function
@xmath166 needs to be chosen so that @xmath165 is an eigenstate of @xmath158 .
it can be proven that @xmath167 does exactly this , and it also assures the unitarity of the transformation : @xmath168 = & \dfrac{\text{id}-\mathbf{r}({\text{d}}\phi,0)}{i{\text{d}}\phi } [ \mathbf{b}_{{k_{\rho}}}^{\text{te } } ] \\ = & - \int_0^{\pi } \sin { \theta_k}{\text{d}}{\theta_k}\int_0^{2\pi } { \text{d}}{\phi_k}f({\phi_k } ) \delta ( { \theta_k}- { \theta_k } ' ) \dfrac{{\mathbf{r } ( { \phi_k } , { \theta_k})}- \mathbf{r}({\phi_k}+ { \text{d}}\phi , { \theta_k } ) } { i{\text{d}}\phi } i { \mathbf{\hat{y}}}\exp(ikz ) \\ = &
\int_0^{\pi } \sin { \theta_k}{\text{d}}{\theta_k}\int_0^{2\pi } { \text{d}}{\phi_k}\dfrac{\partial f({\phi_k})}{\partial { \phi_k } } \delta ( { \theta_k}- { \theta_k } ' ) { \mathbf{r } ( { \phi_k } , { \theta_k})}{\mathbf{\hat{y}}}\exp(ikz ) = m \mathbf{b}_{{k_{\rho}}}^{\text{te } } \label{e_jzb } \end{split}\ ] ] where an integration by parts has been carried out on the last step .
it is straightforward to see that similar bessel beams could have been obtained if instead of rotating @xmath169 , the other three plane waves listed on equations ( [ e_p - x]-[e_ep - r ] ) had been used .
this fact allows for the definition of two sets of em modes .
i will denote the first set as @xmath170 , with @xmath171 , or bessel beams . besides @xmath131 , they are eigenstates of the following operators : @xmath172 & = & { k_{\rho}}\ { \mathbf{b}_{m{k_{\rho}}}^{(y)}}\\ j_z \left [ { \mathbf{b}_{m{k_{\rho}}}^{(y)}}\right ] & = & m \ { \mathbf{b}_{m{k_{\rho}}}^{(y)}}\label{e_bx_eigen } \end{array}\ ] ]
their expressions as functions of @xmath128 and @xmath129 waves are : @xmath173 whilst their expressions in the real space are : @xmath174 e^{ik_z z}\\ { \mathbf{b}_{m{k_{\rho}}}^{\text{tm } } } & = & i^m \sqrt{\dfrac{{k_{\rho}}}{4\pi } } \left [ - \dfrac{k_z}{k } \left ( i j_{m+1}({k_{\rho}}\rho ) e^{i ( m+1 ) \phi } \ { \mathbf{\hat{\sigma}_-}}- i j_{m-1}({k_{\rho}}\rho ) e^{i ( m-1 ) \phi } \ { \mathbf{\hat{\sigma}_+}}\right ) \right .
\\ & & \left .
+ \sqrt{2 } \dfrac{{k_{\rho}}}{k } j_m({k_{\rho}}\rho ) e^{im\phi } \ { \mathbf{\hat{z}}}\right ] e^{ik_z z } \\
\label{e_bbtem } \end{array}\ ] ] where @xmath175 and the following equality has been used @xcite : @xmath176 with @xmath177 being a bessel function of the first kind @xcite .
an analogous procedure can be followed to derive the bessel beams with a well - defined helicity , @xmath178 .
their expressions as functions of the @xmath179 waves can be compacted into : @xmath180 which yields the following expression for the real space fields : @xmath181 \\ \end{array } \label{e_bmp}\ ] ] as for the symmetries of @xmath178 , they are eigenstates of @xmath131 , and the following operators : @xmath182 & = & { k_{\rho}}\ { \mathbf{b}_{m{k_{\rho}}}^{p}}\\ j_z \left [ { \mathbf{b}_{m{k_{\rho}}}^{p}}\right ] & = & m \ { \mathbf{b}_{m{k_{\rho}}}^{p}}\\ \lambda \left [ { \mathbf{b}_{m{k_{\rho}}}^{p}}\right ] & = & p \ { \mathbf{b}_{m{k_{\rho}}}^{p}}\label{e_bp_eigen } \end{array}\ ] ]
it is clear , then , that the relation between the parity and helicity modes is the following one : @xmath183 as with the plane waves , the two sets of bessel beams are also normalized : @xmath184 finally , for completeness , the decomposition of plane waves into bessel beams is also given : @xmath185 the multipolar fields are em modes specially useful to describe em interactions with spherical symmetry .
they are widely used in antenna theory @xcite , mie theory ( see sections ( [ ch2]-[ch4 ] ) ) , nuclear @xcite , atomic and molecular physics @xcite , astrophysics @xcite , and many more fields . the fact that they are used in such vast variety of fields makes a unique notation almost impossible .
sometimes , even within the same field , different notations are used , making it hard to reconcile calculations from different sources @xcite .
my notation is very similar to rose s @xcite , and it is based on the symmetries of the modes .
i will denote a multipolar field as @xmath186 , where @xmath187 accounts for the parity of the field .
the notation becomes clear when we look at the symmetries of the modes . besides the hamiltonian @xmath131 ,
these modes are also symmetric under : @xmath188 & = & j(j+1 ) \ { \mathbf{a}_{jm_z}^{(m ) } } & \qquad j^2 \left [ { \mathbf{a}_{jm_z}^{(e)}}\right ] & = & j(j+1 ) \ { \mathbf{a}_{jm_z}^{(e)}}\\ j_z \left [ { \mathbf{a}_{jm_z}^{(m)}}\right ] & = & m_z \ { \mathbf{a}_{jm_z}^{(m ) } } & \qquad j_z \left [ { \mathbf{a}_{jm_z}^{(e)}}\right ] & = & m_z \ { \mathbf{a}_{jm_z}^{(e)}}\\ \pi \left [ { \mathbf{a}_{jm_z}^{(m)}}\right ] & = & ( -1)^j \ { \mathbf{a}_{jm_z}^{(m ) } } & \qquad \pi \left [ { \mathbf{a}_{jm_z}^{(e)}}\right ] & = & ( -1)^{j+1 } \ { \mathbf{a}_{jm_z}^{(e)}}\label{e_ax_eigen } \end{array}\ ] ] like bessel beams , they can also be expressed as general superpositions of plane waves .
their expressions can be found in @xcite : @xmath189 where @xmath190 is the wigner rotation matrix @xcite .
note that the symmetry requirements defined by equations ( [ e_ax_eigen ] ) do not imply anything about the radial dependence of the multipoles .
nonetheless , in the plane - wave decomposition given by equations ( [ e_ame_sp ] ) , the radial function has been chosen so that these multipoles are regular at the origin .
thus , they have the following expressions in the real space @xcite : @xmath191\\ { \mathbf{a}_{jm_z}^{(e ) } } & = & c_{jm_z } \left [ j(j+1)p_j^{m_z}(\eta ) \dfrac{j_j(kr)}{kr } e^{im_z\phi } \ { \mathbf{\hat{r}}}+ \dfrac{1}{kr } \dfrac{{\text{d}}\left [ kr j_j(kr ) \right]}{{\text{d}}(kr ) } \dfrac{{\text{d}}p_j^{m_z}(\eta ) } { { \text{d}}\theta } e^{im_z\phi } \
\mathbf{\hat{\theta } } \right .
\\ & & \left .
+ ie^{im_z\phi } \dfrac{m_z } { \sin \theta } p_j^{m_z}(\eta ) \dfrac{1}{kr } \dfrac{{\text{d}}\left [ kr j_j(kr ) \right]}{{\text{d}}(kr ) } \ \mathbf{\hat{\phi } } \right ] \label{e_ame_real } \end{array}\ ] ] where @xmath192 is a spherical bessel function @xcite , @xmath193 is a generalized legendre polynomial @xcite , @xmath194 , @xmath195 are the polarization vectors in spherical coordinates , and @xmath196 is a constant so that the modes are normalized .
its expression is given by : @xmath197 indeed , with @xmath196 the multipolar fields are an orthonormal basis : @xmath198 similarly , to the plane waves and bessel beams , a new set of multipolar fields can be defined by adding and subtracting the electric and magnetic modes : @xmath199 this implies that their plane wave decomposition is @xcite : @xmath200 which yields the following real space expression : @xmath201}{{\text{d}}(kr ) } \dfrac{{\text{d}}p_j^{m_z}(\eta ) } { { \text{d}}\theta } \right ) \mathbf{\hat{\theta } } \\ & & \left . - e^{im_z\phi } \left ( \dfrac{{\text{d}}p_j^{m_z}(\eta ) } { { \text{d}}\theta } j_j(kr ) e^{im_z\phi } + p \dfrac{m_z } { \sin \theta } p_j^{m_z}(\eta ) \dfrac{1}{kr } \dfrac{{\text{d}}\left [ kr j_j(kr ) \right]}{{\text{d}}(kr ) } \right ) \mathbf{\hat{\phi } } \right ] \label{e_ajmp_real } \end{array}\ ] ] besides of the @xmath131 operator , the multipolar fields with well - defined helicity @xmath202 are eigenstate of the following three operators : @xmath203 & = & j(j+1 ) \
{ \mathbf{a}_{jm_z}^{p}}\\ j_z \left [ { \mathbf{a}_{jm_z}^{p}}\right ] & = & m_z \ { \mathbf{a}_{jm_z}^{p}}\\ \lambda \left [ { \mathbf{a}_{jm_z}^{p}}\right ] & = & p \ { \mathbf{a}_{jm_z}^{p}}\end{array}\ ] ] finally , by construction , @xmath202 are also normalized : @xmath204 now , as it will be clear in chapter [ ch2 ] , there is a need to define other multipolar fields .
indeed , the two sets of multipolar fields defined by equations ( [ e_ame_real ] , [ e_ajmp_real ] ) have a far - field ( @xmath205 ) behaviour which is not desired for scattering problems @xcite .
this behaviour stems from the spherical bessel function : @xmath206 that is , the multipolar fields given by equations ( [ e_ame_real ] , [ e_ajmp_real ] ) behave as stationary spherical waves in the far - field .
nevertheless , lots of em problems with spherical symmetry have an emitting nature , _
e.g. _ radiation emitted from antennas @xcite , quantum dots @xcite , nitrogen - vacancies @xcite , electrons @xcite , among others .
one of these problems , which lays in the core of this thesis , is the mie scattering problem , where a sphere embedded in a homogeneous medium scatters light that is illuminated on it @xcite . in all these problems ,
the spherical bessel function of first kind must be replaced by a hankel function of first kind @xmath207 instead of @xmath20 , the hankel function of second kind @xmath208 should have been chosen @xcite . ]
because @xmath208 will not be used throughout this thesis , i denote the hankel function of the first kind simply as @xmath209 . in the far - field , the hankel function produces a desired response , _
i.e. _ it behaves like an outgoing spherical wave : @xmath210 the real expressions of these new multipolar fields is analogous to the ones presented by equations ( [ e_ame_real ] , [ e_ajmp_real ] ) , where @xmath209 replaces @xmath192 : @xmath211\\ \mathbf{a}_{jm_z}^{(e),h } & = & c_{jm_z } \left [ j(j+1)p_j^{m_z}(\eta ) \dfrac{h_j(kr)}{kr } e^{im_z\phi } \ { \mathbf{\hat{r}}}+ \dfrac{1}{kr } \dfrac{{\text{d}}\left [ kr h_j(kr ) \right]}{{\text{d}}(kr ) } \dfrac{{\text{d}}p_j^{m_z}(\eta ) } { { \text{d}}\theta } e^{im_z\phi } \
\mathbf{\hat{\theta } } \right .
\\ & & \left .
+ ie^{im_z\phi } \dfrac{m_z } { \sin \theta } p_j^{m_z}(\eta ) \dfrac{1}{kr } \dfrac{{\text{d}}\left [ kr h_j(kr ) \right]}{{\text{d}}(kr ) } \ \mathbf{\hat{\phi } } \right ] \\
\mathbf{a}_{jm_z}^{p , h } & = & \dfrac{c_{jm_z}}{\sqrt{2 } } \left [ ip j(j+1)p_j^{m_z}(\eta ) \dfrac{h_j(kr)}{kr } e^{im_z\phi } \ { \mathbf{\hat{r}}}\right . \\ & & + i e^{im_z\phi } \left ( \dfrac { m_z}{\sin \theta } p_j^{m_z}(\eta ) h_j(kr ) + p \dfrac{1}{kr } \dfrac{{\text{d}}\left [ kr h_j(kr ) \right]}{{\text{d}}(kr ) } \dfrac{{\text{d}}p_j^{m_z}(\eta ) } { { \text{d}}\theta } \right ) \mathbf{\hat{\theta } } \\ & & \left . - e^{im_z\phi } \left ( \dfrac{{\text{d}}p_j^{m_z}(\eta ) } { { \text{d}}\theta } h_j(kr ) e^{im_z\phi } + p \dfrac{m_z } { \sin \theta } p_j^{m_z}(\eta ) \dfrac{1}{kr } \dfrac{{\text{d}}\left [ kr h_j(kr ) \right]}{{\text{d}}(kr ) } \right ) \mathbf{\hat{\phi } } \right ] \end{array}\ ] ] however , the plane wave decomposition of @xmath212 changes dramatically .
the hankel function is singular in the origin , @xmath213 , and consequently it can not be expanded only as a superposition of propagating plane waves .
evanescent waves must appear in the superposition .
moreover , it is necessary to split the domain into two subspaces , @xmath214 .
the decomposition is given by the following equation @xcite : @xmath215 where @xmath216 is given by : @xmath217 and the spherical momentum variables @xmath218 and @xmath219 must be computed as functions of @xmath220 : @xmath221 the expression for the semi - plane @xmath222 can be obtained substituting the factor @xmath223 by @xmath224 @xcite .
equations ( [ e_ajm_hank ] ) have large implications in nano - optics .
evanescent waves play a key role in obtaining information beyond the diffraction limit of light @xcite . here
, it is shown that evanescent waves are an integral part of hankel multipolar fields .
furthermore , it is seen that , like evanescent waves , hankel multipolar fields are intrinsically linked to interfaces , as they are not well - defined on the whole space .
this fact is largely exploited in the excitation of wgms ( see section [ ch3_wgm ] ) . as described by oraevsky in @xcite ,
given a sphere of radius @xmath225 , hankel multipolar fields have an evanescent dependence for a short range of @xmath226 s outside the spherical surface : @xmath227 , \qquad r > r\ ] ] last but not least , some authors have managed to link the evanescent behaviour of plane waves to complex rotations @xcite .
in fact , this is what is seen in the expressions ( [ e_ajm_hank ] ) , where complex angles would be needed if the double integral @xmath228 was to be expressed in spherical coordinates .
the work presented in this thesis belongs to the field of nano - optics .
that is , most of the work deals with light - matter interactions where the light is at optical wavelengths ( @xmath229 $ ] nm @xcite ) and the matter is a nano - structure .
i will be specially interested in single nano - structures , therefore it will be almost mandatory to tightly focus the beam onto the structure , so that the light - matter interaction is stronger .
however , this is not the typical approach in nanophotonics or metamaterials , where structures are mostly designed to interact with paraxial gaussian beams or plane waves @xcite . in general
, the approach followed in the design of nano - circuits or nano - materials is the following one .
the nano - structure is characterized by its scattering matrix @xmath230 .
this scattering matrix is a function of many geometrical and material properties of the system , @xmath231 , where @xmath232 and @xmath233 are general sets of variables describing the geometrical and material properties of the structure .
however , the properties of the structure do not depend on the incident field .
then , the response of the structure to an incoming field @xmath234 can be cast as a convolution of @xmath231 with @xmath234 : @xmath235 since the properties of the structure do not depend on @xmath234 and the incoming field is well - known ( a plane wave ) , the light - matter interaction is reduced to a complete characterization of the scattering matrix @xmath236 . then , adjusting the geometry or material of the structure , a controlled interaction can be carried out . in this thesis
, i will look at these light - matter interactions from a different perspective @xcite .
i will study single highly symmetric nano - structures whose scattering matrix are analytical or easy to compute .
then , i will use different plane waves decompositions to control the scattering .
that is , instead of controlling @xmath237 with the geometry and materials of the structure , the interaction will be controlled with the incoming field @xmath234 : @xmath238 where @xmath239 is each of the different plane waves that take part in the superposition , and @xmath240 is the fourier transform of the incident field , which modulates the plane wave decomposition .
thus , it will be crucial to know the analytical expression and the symmetries of the incoming beam .
in addition , as previously mentioned , due to the fact that single structures will be used , tightly focused light beams will be used to strengthen the light - matter interaction .
tightly focused beams can be analytically described with the aplanatic model of a lens .
this is explained in the following section . in this section ,
i study one of the most frequently used models to describe the focusing of light beams , _
i.e. _ the aplanatic lens model @xcite . in particular
, i will be especially interested in the symmetries of this model .
i will show that the aplanatic lens model does not change the helicity nor the am content of the beam as long as some conditions are met .
when those are met , the focusing of a bessel beam with a well - defined helicity such as @xmath178 can be easily modelled as @xmath241 .
thus , @xmath242 will still be an eigenvector of @xmath96 and @xmath158 .
but before getting there , the aplanatic model of a lens is explained .
the variables playing a role in the model are the collimated incident beam @xmath234 , the focal distance of the lens @xmath243 , and the refractive indexes of the media on the two sides of the lens @xmath244 , @xmath245 .
the geometrical representation of the aplanatic model is given by figure [ fig21 ] .
is the incident collimated electric field ; @xmath246 is the focused field ; @xmath244 , @xmath245 are the refractive indexes of the media embedding the lens before and after the field is focused ; @xmath247 and @xmath248 are polarization vectors in cylindrical and spherical coordinates .
[ fig21],width=491 ] as it can be seen there , the collimated beam @xmath234 travels in a medium whose refractive index is @xmath244 and hits the back - aperture of a lens .
the analytical description of the collimated beam at the a @xmath43 plane of the back - aperture of the lens is done in the paraxial approximation , _
i.e. _ @xmath249 with @xmath42 being the polarisation vector . the incident beam gets refracted and its expression @xmath250 is obtained as : @xmath251 \sqrt{\frac{n}{n_1}}(\cos { \theta_k})^{1/2 } \label{e_einf}\ ] ] where @xmath252 are the effective fresnel coefficients of the lens for @xmath128 and @xmath129 waves respectively @xcite ; @xmath253 and @xmath254 are the polarization vectors in cylindrical and spherical coordinates ( see figure [ fig21 ] ) .
note that three assumptions have been made here .
first , the sine condition of geometrical optics has been considered . the sine condition states that any ray emerging or converging to the focus of an aplanatic system meets its conjugate ray at the surface of a sphere of radius @xmath243 , at a distance of the optical axis @xmath255 , where @xmath256 is the divergence angle @xcite .
second , the intensity law has also been assumed .
the intensity law implies that the energy incident on the aplanatic lens equals the energy leaving it .
it can be expressed as @xmath257 @xcite .
finally , it is imposed that the real space - expression of @xmath234 on that @xmath43 plane gives rise to the angular spectrum of the focused beam .
that is , when introducing the expression of @xmath234 in equation ( [ e_einf ] ) , the following change of variables needs to be done : @xmath258 where no @xmath43 dependence is supposed as the expression is taken on a plane with a constant @xmath43 . with all this , the electric field at the focus of the lens is obtained as : @xmath259 with @xmath243 being the focal distance of the lens , and @xmath260 with na the numerical aperture of the lens . the fact that the aplanatic model preserves the helicity of light is not trivial nor a casualty .
in fact , simple lenses made of glass do not preserve helicity .
however , it is interesting for different applications that the polarisation of light is preserved under a microscope set - up . in order to achieve
that ( which implies helicity preservation ) , microscope objective manufacturers need to apply a special coating to microscope objectives @xcite
. this fact is translated into a condition relating the effective fresnel coefficients @xmath261 . in order to show this condition , a general plane wave with a well - defined helicity @xmath262
will be focused down using the aplanatic lens model .
it will be seen that the helicity of the beam is preserved provided a certain condition is met .
consider a general incident paraxial beam with a well - defined helicity @xmath262 within the paraxial approximation and a @xmath263 dependence : @xmath264 where @xmath265 is the amplitude of the electric field on the plane of the back - aperture of the lens .
then , the computation of @xmath250 yields : @xmath266 \sqrt{\frac{n}{n_1}}(\cos { \theta_k})^{1/2 } \label{e_einf_hel}\ ] ] where the following equalities have been used : @xmath267 equation ( [ e_einf_hel ] ) can be expressed in terms of @xmath128 and @xmath129 : @xmath268 \sqrt{\frac{n}{n_1}}(\cos { \theta_k})^{1/2 } \label{e_einf_sp}\ ] ] where the two following relations have been used : @xmath269 hence , examining equation ( [ e_einf_sp ] ) , it is straightforward to conclude that the aplanatic lens model preserves helicity if and only if @xmath270 i will consider that equation ( [ e_tstp ] ) is valid for the rest of my theoretical calculations .
looking at figure [ fig21 ] , one notices that the geometrical representation of the aplanatic model is cylindrically symmetric around the @xmath43 axis .
thanks to noether s theorem , in principle it follows that the model preserves @xmath158 . nevertheless , in the same way as it happened in the previous subsection , the fresnel coefficients play an important role .
indeed , next it will be proven that either the fresnel coefficients fulfil a certain condition , or the model is not cylindrically symmetric and therefore does not preserves @xmath158 . using equation ( [ e_apl_model ] ) ,
the expression for the electric field at the focus of the aplanatic lens when an incoming beam given by equation ( [ e_ein_ep ] ) can be computed : @xmath271 it is interesting to see that equation ( [ e_efoc_ep ] ) can be expressed as a superposition of rotated plane waves with a well - defined helicity .
the way to construct such a superposition is intuitive and reflects very well the definition of the helicity of a beam .
we take a circularly polarized plane wave propagating along the z axis with helicity @xmath262 .
we rotate it in all directions and we sum all the different contributions with a certain weight @xmath272 .
this rotation does not change the polarization in the system of reference of the plane wave ( it maintains transversality ) , therefore the helicity has still a well - defined value @xmath262 .
@xmath273 where @xmath274 is a general field with a well - defined helicity . then , using equations ( [ e_ep - l ] , [ e_ep - r ] ) , @xmath274 can be re - written as : @xmath275 at this point , a direct comparison between equations ( [ e_ap ] ) and ( [ e_efoc_ep ] ) yields a function @xmath276 that makes both equations equal : @xmath277 where @xmath278 is the heaviside step function . with @xmath276
given by equation ( [ e_gthphi ] ) , the field at the focus of a lens can be given by : @xmath279 now , i will show that @xmath246 has a well - defined @xmath158 if the incoming field does , too .
that is , i will show that the aplanatic model of a lens preserves @xmath158 by verifying the following condition : @xmath280 = m { \mathbf{e}^\mathrm{in } } & \longrightarrow & j_z [ \mathbf{e}^{\mathrm{foc } } ] = m \mathbf{e}^{\mathrm{foc } } \end{array}\ ] ] a paraxial beam with a well - defined helicity @xmath262 and an value of @xmath281 with eigenvalue @xmath282 . ] is given by any expression of the kind : @xmath283 if this expression is substituted in equation ( [ e_gthphi ] ) , the following expression for @xmath276 follows : @xmath284 now , in order to prove that the aplanatic lens model does not change the @xmath158 content of @xmath285 , i apply @xmath158 to @xmath246 when @xmath276 is given by equation ( [ e_g_par ] ) .
the fact that @xmath246 can be expressed as a rotation helps to compute the result . as seen by equation ( [ e_jzb ] ) , @xmath158 can be applied as a partial derivative with respect to @xmath218 under the integral operation .
then , it can be seen that : @xmath286 = \int_0^{\pi } \sin ( { \theta_k } ) { \text{d}}{\theta_k}\int_0^{2\pi } { \text{d}}{\phi_k}\ \dfrac{\partial g({\phi_k},{\theta_k})}{\partial { \phi_k } } { \mathbf{r } ( { \phi_k } , { \theta_k})}i { \mathbf{\hat{\sigma}_p}}\exp\left ( ikz \right)\ ] ] with @xmath276 only depending on @xmath218 via two terms , @xmath287 and @xmath288 . as one could expect
, it is easy to see that the aplanatic model preserves @xmath158 of @xmath234 if and only if @xmath289 . throughout this thesis
, i will consider that all lenses ( as long as they are centred with the beam axis ) preserve @xmath158 and therefore @xmath290 . in this chapter , different theoretical methods to describe light - matter interactions have been explained .
they will be used in the remaining of this thesis , both to understand the experimental results , as well as to carry out different theoretical developments .
next , i present a summary of the methods that have been described above as well as their applications in the remaining of the thesis .
+ + in section [ ch1_hilmax ] , maxwell equations for a linear , non - dispersive , homogeneous , isotropic and source - free medium have been introduced .
these equations will be used in chapters [ ch2 ] , [ ch3 ] and [ ch4 ] to deepen in understanding of the interaction of vortex beams with spherical particles . + + in section [ ch1_symm ] , the paraxial approximation , which is a very relevant limit of maxwell equations , has been explained .
this approximation is going to be used in all the experimental chapters ( [ ch5 ] , [ ch6 ] , and [ ch7 ] ) .
then , the concepts of em field , operator , and symmetry have been given .
of particular interest is the concept of symmetry , which is used throughout this thesis to predict theoretical results or interpret experimental results .
a complete list of symmetries with their respective generators can also be found in section [ ch1_symm ] .
many of these generators will be used to characterize the light beams as well as the samples that will be used in this thesis . in particular , chapters
[ ch3 ] , [ ch4 ] , [ ch6 ] , [ ch7 ] describe different phenomena that arise when beams with duality and cylindrical symmetry interact with samples with mirror and cylindrical symmetry .
+ + in section [ ch1_basis ] , a general method to construct solutions of the vectorial helmholtz equation is sketched .
then , plane waves are constructed using this method .
plane waves are characterised in terms of their symmetries , and they are used to construct bessel beams , which are symmetric under some different symmetries .
both plane waves and bessel beams are later used in section [ ch5_cgh ] .
+ + section [ ch1_multipoles ] introduces two different basis of multipolar fields with two different radial functions each .
the understanding of the symmetries of these beams will be crucial in chapters [ ch2 ] , [ ch3 ] and [ ch4 ] , where they are used to solve maxwell equations in a spherical domain .
+ + section [ ch1_scatt ] presents in a formal way the key idea behind the work done in this thesis , which was introduced in [ intro ] . that is , controlling and/or characterising the scattering of nano - structures using symmetric light .
this method is exploited both theoretically and experimentally in chapters [ ch3 ] , [ ch4 ] , [ ch6 ] and [ ch7 ] .
it is seen that symmetric beams can be used to effectively turn a non - dual particle into dual , or inducing circular dichroism in a non - chiral smaple .
+ + finally , in section [ ch1_apla ] , the description of the aplanatic model of a lens is given .
the aplanatic model is used in chapter [ ch3 ] to theoretically control the multipolar content of a beam .
furthermore , the symmetry conditions proven in [ ch1_apla_hel ] and [ ch1_apla_jz ] are then assumed in chapters [ ch6 ] and [ ch7 ] to understand the experimental results .
`` i learned very early the difference between knowing the name of something and knowing something . ''
[ ch2 ] in this chapter , i will use the tools described in the preceding chapter to solve the electromagnetic problem described in figure [ f_glmt_schematics ] .
an arbitrary electromagnetic field @xmath291 excites a single homogeneous , isotropic sphere embedded in a homogeneous , isotropic and lossless medium is an abuse of language .
the incident electromagnetic field has also a magnetic component @xmath292 , and it is indeed used to solve the maxwell s equations . ] .
( red ) propagates through a lossless , isotropic and homogeneous medium with optical constants @xmath293 .
the field interacts with an isotropic , homogeneous sphere with optical constants @xmath294 . as a result
, the incident field is scattered in all directions giving rise to @xmath295 ( blue).,width=529 ] as mentioned in the introduction , the scattering of a sphere illuminated by a plane wave was independently solved by ludvig lorenz in 1890 and gustav mie in 1908 @xcite .
unfortunately for lorenz , mie s name has stuck in the literature and nowadays the problem is mostly known as the mie scattering problem @xcite .
a very complete review about the history of this problem can be found in @xcite . before the invention of lasers in 1958 @xcite ,
mie theory had already been successfully applied to a very wide range of fields , _
e.g. _ chemistry , material science or atmospheric physics @xcite
. however , once mie theory started being tested with lasers , it became clear that a more general formulation of the problem had to be established .
although in many cases the light beam produced by a laser could be modelled as a plane wave with a good approximation , a more general theory that took into account the spatial non - uniformity of light beams was needed . in this context , the generalized lorenz - mie theory ( glmt ) was created in the 1980 s @xcite .
the glmt development was especially led by grard gouesbet and grard grhan , both professors at the university of rouen , france . in its first ten years
, the progress of glmt was mainly focused on the theoretical part . in 1985 ,
gouesbet _ et al . _ managed to solve the scattering problem of a free propagating gaussian beam interacting with an on - axis arbitrary sphere using bromwich potentials @xcite .
and then , in 1988 , a seminal paper was also published by gouesbet and co - workers where the scattering of a sphere located in an arbitrary position with respect to the axis of propagation of a gaussian beam was solved @xcite .
this paper laid the foundations of the glmt , as it solved the scattering problem of a single sphere excited with an arbitrary beam of light @xcite . since then , as the theoretical framework had already been laid , the development of the theory happened in the applications and computation side .
due to computation limitations at that time , a lot of effort was put into finding ways of efficiently computing the so - called beam shape coefficients of an arbitrary beam @xcite .
in addition to that , lots of applications were found , including phase - doppler instruments , imaging , optical characterization and optical manipulation , among others @xcite . in this chapter
, i will summarize the main theoretical results of glmt and interpret them from a symmetry - based point of view .
the wide variety of applications of glmt is a proof of the success of the theory .
however , due to the typical method of solving the problem via vector potentials ( bromwich , debye or hertz ) @xcite , the symmetries of the system are usually blurred . in order to gain a bit of insight about the symmetries in the problem ,
let s take a look at figure [ f_mie_symmetries ] , where the different symmetries of the mie problem are presented .
i have ruled out the incident beam of any consideration , as i consider it as an excitation to the system , but not part of the system itself .
the system is formed by the sphere and the surrounding medium .
the origin of the frame of reference in which the problem is described is placed at the centre of the sphere .
this is necessary for the three last symmetry considerations to hold : for the blue wave , @xmath296 for the red one ) .
the interaction is the same one for both .
b ) rotations around the @xmath43 axis .
the system remains invariant when it is rotated around the @xmath43 axis .
c ) independence of orientation of reference frame .
the problem is cylindrically symmetric around the @xmath43 axis independently of the orientation of the reference frame .
. a parity @xmath297 change of coordinates leaves the problem invariant.,width=529 ] * _ temporal translations , @xmath109_. because both the sphere and the medium surrounding it are assumed to be linear media , the excitation of the sphere is independent of the instant when the sphere is excited . *
_ rotations around the @xmath43 axis , @xmath298_. a rotation around the @xmath43 axis leaves the system invariant .
this will also be referred to as cylindrical symmetry . * _
independence of orientation of the reference frame_. as long as the origin is the center of the sphere , the orientation of the axis will be irrelevant . that is
, any axis is an axis of cylindrical symmetry . * _ parity , @xmath102_. the system is invariant under parity transformations , _
i.e. _ changes of @xmath299 .
these four transformations commute with each other .
thus , the natural modes of the system will be eigenmodes of these four transformations .
now , due to the fact that continuous symmetries can be expressed as a function of their generators , i will use the generators of the three first symmetries to classify the natural modes of the system ( see section [ ch1_symm ] ) : * hamiltonian , @xmath131 .
temporal translations are generated by the hamiltonian @xmath300 .
* projection of the am in the @xmath43 axis , @xmath158 .
the @xmath43 component of the am generates @xmath301 .
* the am squared , @xmath94 .
the transformation @xmath302 is meaningless .
nevertheless , the independence of orientation of axis implies that the eigenmodes of the system need to be eigenstates of @xmath94 .
parity ( @xmath102 ) is not a continuous transformation , therefore i will use the transformation itself to classify the eigenmodes of the system .
now , in [ ch1_multipoles ] we have seen that one of the complete basis of solutions of source - free maxwell equations are the multipolar modes @xmath303 . as shown in there ,
these modes are eigenmodes of @xmath131 , @xmath158 , @xmath94 , and @xmath102 respectively .
thus , the natural modes of the system depicted in figures [ f_glmt_schematics ] , [ f_mie_symmetries ] will be the multipolar modes , as they fulfil all the symmetry requirements . a harmonic time dependence @xmath20 will be assumed throughout the rest of the chapter , as non - linear phenomena are not going to be studied . in this section
i will solve the problem sketched in figure [ f_glmt_schematics ] .
that is , i will find the em field ( @xmath26 , @xmath27 ) in all space for a given arbitrary incident field @xmath291 . in order to achieve
that , i will describe all the em fields as a superposition of multipolar modes .
this will simplify all the calculations , as they are the natural modes of the system .
the procedure will be the following one .
first , i will break down the domain into two sub - domains which will be separated by the surface of the sphere .
the em field inside the sphere will be given by @xmath304 and @xmath305 , the interior electric and magnetic fields .
the expression of the em field outside the sphere will be given by the total field , which can be computed as the addition of the scattered field ( @xmath295 , @xmath306 ) and the incident field ( @xmath291 , @xmath292 ) , _ i.e. _ @xmath307 .
then , i will impose that the tangential components of the fields inside and outside the sphere are continuous @xcite . to make the problem as general as possible
, i will suppose that the incident electric field is the most general superposition of multipolar modes : @xmath308 where @xmath309 and @xmath310 are some general coefficients modulating the multipolar superposition .
then , the magnetic field can be obtained using the maxwell equations ( [ e_max_eh_mono ] ) and the relations between multipolar modes given in @xcite : @xmath311\ ] ] the scattered and interior fields are also supposed to be a general superposition of multipolar fields , whose expression are given by : @xmath312 \\
{ \mathbf{e}^\mathrm{int } } & = & \sum_{j=1}^{\infty } \ \sum_{m_z =- j}^{j } { c_{j , m_z}}{\mathbf{a}_{jm_z}^{(m)}}+ { d_{j , m_z}}{\mathbf{a}_{jm_z}^{(e)}}\\ { \mathbf{h}^\mathrm{int } } & = & \dfrac{-1}{\omega \mu_1 } \left [ \sum_{j=1}^{\infty }
\ \sum_{m_z =- j}^{j } { c_{j , m_z}}{\mathbf{a}_{jm_z}^{(e)}}- { d_{j , m_z}}{\mathbf{a}_{jm_z}^{(m)}}\right ] \label{e_all_fields4}\end{aligned}\ ] ] where @xmath313 and @xmath1 are the magnetic permeabilities of the sphere and the surrounding medium respectively .
however , as mentioned in [ ch1_multipoles ] , even though we use the same notation for the multipolar modes in scattering and free propagation , their radial functions are different . for scattering ,
the radial function is a hankel function of first kind , @xmath314 . whereas for the interior and incident fields , the radial function is a bessel function , @xmath315 @xcite .
once the general expression of the fields has been found , we apply the tangential maxwell boundary conditions , which yield the following four equalities : @xmath316 equations ( [ tangential ] ) can be solved with cramer s rule , yielding the following results : @xmath317 where @xmath318 are the so - called mie coefficients , whose expressions are given by : @xmath319'-\mu_1 j_j(x)[n_r x j_j(n_rx)]'}{\mu n_r^2 j_j(n_r x)[x h_j^{(1)}(x ) ] ' - \mu_1 h_j^{(1)}(x ) [ n_r x j_j(n_rx ) ] ' } \\ \displaystyle b_j = \frac{\mu_1 j_j(n_r x ) [ xj_j(x)]'- \mu j_j(x ) [ n_r x j_j(n_r x)]'}{\mu_1 j_j(n_r
x)[x h_j^{(1)}(x ) ] ' - \mu h_j^{(1)}(x ) [ n_r x j_j(n_rx ) ] ' } \\ \displaystyle c_j=\frac{\mu_1
j_j(x ) [ xh_j^{(1)}(x)]'-\mu_1 h_j^{(1)}(x ) [ xj_j(x)]'}{\mu_1 j_j(n_r x)[x h_j^{(1)}(x ) ] ' - \mu
h_j^{(1)}(x ) [ n_r x j_j(n_rx ) ] ' } \\
\displaystyle d_j=\frac{\mu_1 n_r j_j(x ) [ xh_j^{(1)}(x)]'-\mu_1 n_r h_j^{(1)}(x ) [ xj_j(x)]'}{\mu n_r^2 j_j(n_r x)[x h_j^{(1)}(x ) ] ' - \mu_1 h_j^{(1)}(x ) [ n_r x j_j(n_rx ) ] ' } \label{mie_coeffs } \end{array}\ ] ] where @xmath320 is the so - called size parameter , with @xmath225 being the radius of the sphere in consideration and @xmath321 the excitation wavelength , inversely proportional to the frequency @xmath21 ; @xmath322 is the relative refractive index of the sphere with respect to the surrounding medium , @xmath323 .
it can be observed from equations ( [ general ] ) that once the coefficients @xmath324 have been found for the incident field , the solution of the scattering problem is straightforward : the expression for the scattered and interior fields are a mere formal copy of the incident field , where each multipole is modulated by a mie coefficient .
hence , the key point to solve the generalized lorenz - mie problem is finding the coefficients @xmath324 . both @xmath309 and @xmath310
are usually known as beam shape coefficients , except for a subtlety . in glmt ,
the beam shape coefficients are defined as @xmath325 and @xmath326 , and they are equal to @xmath310 and @xmath309 divided by a factor @xcite : @xmath327 where @xmath328 are the plane - wave coefficients : @xmath329 now , given @xmath291 , we can find them applying the techniques shown in ( [ ch1_basis ] ) . that is , due to the orthonormality relations between @xmath330 and @xmath331 , we can find @xmath309 and @xmath310 projecting @xmath291 onto the multipolar modes , yielding a unique result .
thus , from equation ( [ ei ] ) , we obtain : @xmath332 equations ( [ int_gjm ] , [ int_gje ] ) give us the formal solution of the problem .
hence , solving the glmt problem basically consists in computing two sets of two - dimensional angular integrals .
with our current computational power , @xmath309 and @xmath310 can be computed using standard numerical integration techniques , without too much of a struggle .
nevertheless , as mentioned in [ ch2_intro ] , that was not the case when glmt was firstly formulated , and so a lot of effort was put into computing the beam shape coefficients .
a very complete review of all the different available techniques to compute the beam shape coefficients can be found in @xcite .
i will not explicitly comment further on those techniques , as in the proceeding sections i will simplify ( [ int_gjm ] , [ int_gje ] ) thanks to some symmetry considerations . in the previous section
i have given the expressions of the em field in all points of space . nonetheless , it is difficult to measure the em field in experiments at optical frequencies . therefore , we need to derive some quantities that we can experimentally determine , even though they might provide a less complete description of the scattering process .
one of these quantities is the em power carried by the radiation , since many light detectors are sensitive to it .
thus , we need to mathematically characterize how much power is carried by the three fields in the problem . in order to do that , we first define the time - averaged poynting vector dependence is supposed ] @xcite : @xmath333 \label{poyting}\ ] ] now , the total fields are obtained adding the scattered and incident fields , _
i.e. _ @xmath307 .
hence , we can split the time - averaged poynting vector in three parts : @xmath334 \\ \langle { \mathbf{s}^\mathrm{s}}\rangle & = & \dfrac{1}{2 } \text{re } \left [ { \mathbf{e}^\mathrm{s}}\times ( { \mathbf{h}^\mathrm{s}})^ * \right ] \\
\langle { \mathbf{s}^\mathrm{ext}}\rangle & = & \dfrac{1}{2 } \text{re } \left [ { \mathbf{e}^\mathrm{i}}\times ( { \mathbf{h}^\mathrm{s}})^ * + { \mathbf{e}^\mathrm{s}}\times ( { \mathbf{h}^\mathrm{i}})^ * \right]\end{aligned}\ ] ] the flux of the poynting vector across a surface gives us the energy flux per unit of area and unit of time , which accounts for the loss of energy inside the surface .
we will denote this flux as minus the absorption cross - section : @xmath335 where @xmath336 is a differential surface element , with @xmath33 being the normal vector to this element at every point .
due to the distributive property of integrals , this can be also split in three terms , to which we will refer as incident , scattering and extinction cross - sections : @xmath337 clearly , if we consider the flux of the poynting vector across a spherical surface , @xmath338 , as the surrounding medium is non - absorbing .
the choice of a spherical surface simplifies the calculations without any loss of generality , therefore i will use it for the rest of the calculations .
this will directly imply that the surface element @xmath339 , and the normal vector @xmath33 will point in the radial direction @xmath340 .
i will start by computing @xmath341 . in order to do this calculation
, the following orthonormality relations of the multipolar fields will be considered @xcite : @xmath342 where @xmath343 is @xmath344 and @xmath345 is a bessel function for the incident and interior field and a hankel function of the first kind for the scattered field .
then , it can be proven that @xmath346 where the two following relations have been used @xcite : @xmath347 note that equation ( [ e_cs ] ) differs from the expression given in @xcite in some factors due to the fact that the normalisation of the multipolar modes given by equation [ e_mul_norm ] is not considered in @xcite using the bromwich potentials . very similarly , the following expression can be obtained for @xmath348 @xcite : @xmath349 \label{e_cext}\ ] ] consequently , @xmath350 can be obtained subtracting expressions ( [ e_cext ] ) and ( [ e_cs ] ) : @xmath351 .
+ as i have shown above , the expressions for the interior and scattered field depend on the incident field .
in particular , their expressions depend on the intensity carried by @xmath291 .
this fact naturally translates into the cross sections , _
i.e. _ the cross sections depend on the intensity of the incident beam . in order to remove the intensity dependence from the cross - section expression ,
we normalize the cross sections over the intensity of the beam across a transverse section @xcite : @xmath352 where @xmath353 can take three values @xmath354 .
i will not give general expressions for the efficiencies .
instead , i will either explicitly compute them ( see chapters [ ch3 ] and [ ch4 ] ) , or measure them ( see chapter [ ch7 ] ) . a subtlety about the measurements in chapter [ ch7 ]
is in order at this point . despite normalising the scattering intensity over the power of the incident beam ( @xmath355 ) ,
the result is modified by the limited na of the detector .
that is , the definitions of cross section and efficiency factors given by equations ( [ c1]-[c5 ] ) and ( [ e_efficiency ] ) suppose that the scattering in all angles @xmath356 is collected . however , this is not the case in the experiments done in chapter [ ch7 ] .
there , only scattered light in the backward semi - space is collected .
now , even though backward cross sections do not generally have the same @xmath321 dependence as proper cross sections @xcite , the features that i will be especially interested in will still remain .
namely , as it will be shown in section [ ch3_wgm ] , the excitation of spheres with plane waves hides some resonant behaviour that can be unveiled using vortex beams .
`` i think it only makes sense to seek out and identify structures of authority , hierarchy , and domination in every aspect of life , and to challenge them ; unless a justification for them can be given , they are illegitimate , and should be dismantled , to increase the scope of human freedom . ''
[ ch3 ] in the previous chapter , i have solved the generalized lorenz - mie problem .
that is , i have found the em fields in the whole space when a homogeneous and isotropic sphere is excited by an arbitrary monochromatic light beam @xmath291 . in the present chapter , i will particularize the solution given by equations ( [ e_all_fields1]-[int_gje ] ) for an incident monochromatic cylindrically symmetric beam .
the definition of a general cylindrically symmetric beam stems from the definition of cylindrical symmetry . as it has been described in chapters [ ch1]-[ch2 ] , a light beam is symmetric under a transformation @xmath357 when its em field is an eigenstate of the transformation generated by @xmath357 : @xmath358 .
i define a cylindrically symmetric beam as an eigenstate of @xmath359 , the operator of rotations around the @xmath43 axis . now , because @xmath359 can be expressed as @xmath301 , a cylindrically symmetric beam can be alternatively defined as an eigenstate of @xmath158 . as we have seen in chapter [ ch1 ] , there are different families of monochromatic modes that have a well - defined @xmath158 .
both bessel beams and multipolar fields are basis of modes that are eigenstates of @xmath158 with eigenvalues @xmath282 and @xmath360 respectively ( see equations ( [ e_bx_eigen ] , [ e_ax_eigen ] ) ) : @xmath361 an arbitrary superposition of them ( keeping @xmath282 or @xmath360 constant ) will also be an eigenstate of @xmath158 .
now , both basis are valid to describe the problem , but i will use the multipolar fields as they are the normal modes of a sphere ( see section [ ch2_symmetries ] ) . in this basis ,
the expression for the electric field of a general cylindrically symmetric beam is : @xmath362 where @xmath363 and @xmath364 are proportional to the beam shape coefficients , ( see section [ ch2_scatt ] ) .
note that a general superposition of multipoles is given by equation ( [ ei ] ) . that decomposition contains two summations , on @xmath365 and @xmath366 respectively .
in contrast , expression ( [ e_ei_c ] ) only involves a summation in @xmath367 .
first , @xmath360 needs to be fixed to @xmath368 so that @xmath291 is still an eigenstate of @xmath158 . and
secondly , due to the fact that @xmath369 , @xmath368 bounds @xmath370 to values @xmath371 .
+ last but not least , i will also impose that the cylindrically symmetric beams are eigenstates of the helicity operator , @xmath96 . in this way , they will be symmetric under duality transformations and it will be easier to study the duality properties of the system ( see chapter [ ch4 ] ) . imposing that the incident field @xmath291 has a well - defined helicity has further implications on expression ( [ e_ei_c ] ) . as it has been demonstrated in chapter [ ch1 ] , a mode of light with a well - defined helicity
can be expressed as a superposition of electric and magnetic modes .
consequently , the two families of beam shape coefficients can be reduced to one : @xmath372 where @xmath373 is the only family of beam shape coefficients remaining and @xmath374 are the helicity values for the beam .
thus , the multipolar decomposition of a cylindrically symmetric beam with a well - defined helicity is : @xmath375 equation ( [ e_ei_cp ] ) will be the starting point for this chapter .
i will use it to solve the glmt problem with paraxial cylindrically symmetric beams in [ ch3_paraxial ] and with general non - paraxial cylindrically symmetric beams in [ ch3_non ] . then , in [ ch3_wgm
] , i will show how the use of these beams of light along with the proper choice of ratio between the particle size and the excitation wavelength is enough to excite wgms @xcite . in this section
i solve the glmt when the incident field @xmath291 is a laguerre - gaussian beam , lg@xmath376 $ ] .
this phase dependence generally gives rise to an intensity profile that resembles that of a doughnut .
nevertheless , a beam can have a doughnut - like intensity profile without having a @xmath377 $ ] phase dependence .
the lg modes are a well - defined set of modes ( see equation ( [ e_lgbeam ] ) ) .
their phase dependence is that of a vortex beam , but their intensity profile is not always doughnut - like .
in fact , their intensity profile is only doughnut - like when the radial parameter @xmath378 . ] .
the lg modes are the normal modes of a cylindrical cavity and can also be obtained as superpositions of hermite - gaussian modes , which are the normal modes of spherical - mirror resonator @xcite .
they are a basis of solutions of the paraxial equation in cylindrical coordinates @xcite , _
i.e. _ any paraxial beam can be described as a superposition of lg modes @xcite .
the lg modes had been known by the optics community for a long time , but their use highly escalated after a seminal paper by allen and co - workers in 1992 @xcite . in that paper , it was established that there was a relation between the phase and the am content of these modes .
this laid out the grounds of the field of the angular momentum of light .
since then , the am of light has been used in many diverse fields such as quantum optics @xcite , optical manipulation @xcite , optical communications @xcite or astrophysics @xcite . the general expression for a lg modes
is given by : @xmath379 \rho^l l^l_q(\dfrac{2\rho^2}{w^2(z)})\left ( \dfrac{\sqrt{2}}{w(z ) } \right)^{l+1 } e^ { i \left ( \frac{-k\rho^2 z}{2(z^2+z_0 ^ 2 ) } + ( 2q+l+1 ) \tan^{-1}(z / z_0 ) + l\phi \right ) } \label{e_lgbeam}\ ] ] where @xmath380 is the radial index , and @xmath2 the azimuthal one .
the azimuthal number @xmath2 describes how the phase winds around the center of the beam , whereas @xmath381 has to do with the number of intensity nodes in the radial direction .
@xmath382 are the cylindrical coordinates ; @xmath383 are the generalized - laguerre functions ; @xmath384 is the waist of the beam at a plane of @xmath43 constant ; @xmath385^{1/2}$ ] is a normalization constant ; and @xmath386 is the wave - vector .
note that due to the fact that a lg mode is a paraxial beam , the polarization is added to the beam independently of its spatial shape ( see section [ ch1_symm])@xcite .
next , i will decompose a circularly polarized lg beam into multipoles .
in fact , a circularly polarized paraxial beam has a well - defined helicity within the paraxial approximation @xcite .
that is , if we apply the helicity operator @xmath96 to a circularly polarized lg mode , and we only consider paraxial terms , we obtain the same mode times the helicity value of the beam : @xmath387 \approx p \left [ \text{lg}_{l , q } { \mathbf{\hat{\sigma}_p}}\right]\ ] ] with @xmath388 being a general circularly polarized vector defined as @xmath389 , and @xmath390 and @xmath391 the horizontal and vertical polarization vectors . the differential form of @xmath96 is given by the expression @xmath392 @xcite .
then , the multipolar decomposition of lg@xmath393 is given by equation ( [ e_ei_cp ] ) and only one set of beam shape coefficients is needed , @xmath373 . before computing them ,
one comment needs to be made .
a beam such as lg@xmath393 does not fulfil maxwell equations .
however , its decomposition into multipoles using equation ( [ e_ei_cp ] ) does . this is a known paradox in optics that was solved by lax in 1975 @xcite
the idea is that the maxwell field given by the multipolar decomposition is equal to the paraxial beam only in the paraxial limit .
furthermore , lax also showed that a paraxial field gives rise to many different fields fulfilling the maxwell equations . the general formalism to find one of the possible multipolar decompositions of a circularly polarized lg mode was given by my supervisor in @xcite .
then , in @xcite , we applied that formalism to give an analytical formula for the beam shaping coefficients @xmath373 .
our findings are shown next .
if @xmath394 , then we can re - express it as : @xmath395 \\
{ c_{jm_z^*p}}&= & \displaystyle\int_0^{\pi}d_{m_z^ * p}^j({\theta_k})f_{l , q}(k\sin{\theta_k } ) \
\sin{\theta_k}{\text{d}}{\theta_k}\end{array } \renewcommand{{2.5}}{1 } \label{e_general}\ ] ] where @xmath396 stands for value of @xmath158 , the am content of a paraxial beam can be obtained by adding the topological charge of the beam ( @xmath2 for a lg@xmath376 beam ) and the value of the helicity @xmath262 .
this is what is usually described as am @xmath397 oam @xmath398 sam , _ i.e. _ the total angular momentum can be obtained adding its orbital and spin parts ] .
the reduced rotation matrix @xmath399 can be found in @xcite , and the function @xmath400 is related to the fourier transform of the incident field , with @xmath46 being the transverse momentum , _ i.e. _ @xmath401 .
note that the function @xmath402 is still defined as an integral , although it is no longer a double one , unlike ( [ int_gjm ] , [ int_gje ] ) .
this is a consequence of the fact that @xmath291 and the multipolar fields are eigenstates of @xmath158 with value @xmath368 , therefore projections are only needed on @xmath94 values .
also , note that the definition of @xmath402 in equation ( [ e_general ] ) is slightly different from its definition in ( [ e_ei_cp ] ) : a factor @xmath403 has been made explicit
. the definition of @xmath402 given by equation ( [ e_general ] ) will be the one used for the rest of the thesis .
+ + actually , equation ( [ e_general ] ) is valid for any beam whose @xmath158 is well - defined . for the specific case of a lg@xmath376 beam , the integral in equation ( [ e_general ] )
can be solved analytically thanks to the expression of @xmath404 and provided the integral [ e_general ] is extended to infinity .
the following expression is obtained : @xmath405^{1/2 } \\
\sum_s \frac{(-1)^s}{(j - l - p - s)!(j+p - s)!(s+l)!s ! } \frac{m_{l+2s}(f_{l , q})}{2^{l+2s } } \end{split } \label{cjlp}\ ] ] where the number @xmath406 is the momentum of order @xmath61 of the function @xmath404 .
the expression for @xmath407 when the incident beam is a lg@xmath376 is : @xmath408 with @xmath409 being the beam width of the lg mode in real space .
finally , i give the expression of @xmath410 when the function @xmath404 is given by the expression ( [ f(lg ) ] ) .
as it is shown , in this particular case , the integral is analytical . also , i particularize the result for @xmath378 : @xmath411 @xmath412 where @xmath413 is the gamma function and @xmath414 is the hypergeometric function .
the problem , then , is solved .
using equations ( [ e_general ] , [ cjlp ] ) and ( [ ma ] ) the decomposition of the incident beam is found .
then , using the glmt formulation introduced in the preceding chapter , the scattered and the interior fields are obtained ( see section [ ch2_scatt ] ) : @xmath415 \\ { \mathbf{e}^\mathrm{int}}=&&\displaystyle\sum_{j=\vert m_z^ * \vert}^{\infty } i^j ( 2j+1)^{1/2 } { c_{jm_z^*p}}\left [ c_j { \mathbf{a}_{jm_z^*}^{(m)}}+ip d_j { \mathbf{a}_{jm_z^*}^{(e)}}\right ] \\ \end{array } \label{e_fields}\ ] ] as already mentioned in chapter [ ch2 ] , it is clear that the solution of the scattering problem only depends on the beam shape coefficients @xmath402 .
the beam shape coefficients depend on the parameters of the incident lg modes , _
i.e. _ the azimuthal and radial numbers @xmath416 , the helicity @xmath262 , and the waist of the beam @xmath417 .
that means that once the lg mode and its polarization have been chosen , the only way of tailoring the multipolar content of the beam is with the width of the beam , @xmath417 .
however , @xmath417 can not get arbitrary values , as the beam still needs to fulfil the paraxial equation @xcite . as a consequence , the multipolar content of the beam
is almost fixed for every single mode .
+ + in figure [ f_multipolar ] , the multipolar content for four different lg modes is depicted .
it can be observed that over 40 multipolar modes of each parity are needed to describe a lg mode of an arbitrary order gives rise to the two families of beam shape coefficients ( see equation ( [ cjmgjm ] ) ) . ] .
this fact will be further explored in the next section .
i advance that using high na lenses will enable us to describe a cylindrically symmetric beam with a handful of multipoles
. this fact will be a key point to excite single multipolar resonances in section [ ch3_wgm ] . for different lg@xmath376 beams .
a ) lg@xmath418 b ) lg@xmath419 c ) lg@xmath420 d ) lg@xmath421 .
the relation between the width ( @xmath417 ) of the beam and the wavelength is @xmath422 .
the helicity is @xmath423 for the four different cases .
the insets are intensity plots of the multipolar decomposition for each of the four cases .
the intensity of each of the lg modes is then retrieved . ] in this section , i will extend the results of the previous section to the non - paraxial case .
that is , in this case the incident field @xmath291 will fulfil maxwell equations .
mathematically , the only difference is the way the beam shape coefficients @xmath373 are computed .
however , physically there is a very big difference : the use of an aplanatic lens enables us to control the multipolar decomposition at the focal plane with a much higher precision .
this fact is of great interest to control multipolar resonances both in nanostructures and atoms @xcite .
i will further develop this idea on section [ ch3_wgm ] .
in addition , i will show that by adjusting the focal distance of a lens , the magnitude of the scattering efficiency can be increased by orders of magnitude .
+ + to start , i will show how the aplanatic model of a lens tailors the multipolar content of a beam . in section [ ch1_apla ]
, i have demonstrated the conditions under which the aplanatic lens model does not change the helicity and the angular momentum content ( @xmath158 ) of the incident fields . in order to do that ,
i have expressed the aplanatic lens transformation @xcite as a rotation in the momentum space .
next , i will use the same mathematical technique to retrieve the beam shape coefficients @xmath373 as a function of the focal distance of the lens , the na , and the media surrounding the lens .
on one hand , i will consider a general superposition of circularly polarized propagating plane waves with the same helicity : @xmath424 on the other hand , the aplanatic model can be modelled as the integral ( [ e_apl_model ] ) of a collimated ( or paraxial ) beam @xmath425 over the surface of a sphere @xcite : @xmath426 in chapter [ ch1 ] , i proved that the polarization transformation produced by equations ( [ e_apl_model2 ] ) and ( [ e_eint ] ) is the same one provided the two transmissivities of the lens are equal , @xmath427 ( see section [ ch1_apla ] ) .
i also gave the value of the function @xmath276 that makes @xmath428 , which is ( see equation [ e_gthphi ] ) : @xmath429 where @xmath243 is the focal of the lens ; @xmath430 , with na the numerical aperture of the lens ; @xmath245 and @xmath244 are the index of refraction of the lens and the medium surrounding it respectively ; @xmath278 is the heaviside step function ; and @xmath431 is the value of the paraxial incident field at the back - aperture of the lens . note @xmath432 .
that is , there is a match between the cylindrical coordinates in the real space of the paraxial incident beam , and the spherical coordinates of its angular spectrum in the momentum space . as mentioned in section [ ch1_apla ] , this fact is a key factor to prove that the aplanatic model does not change the helicity nor the am of the incident fiend .
now , using the properties of the multipolar fields under rotations @xcite , the expression ( [ e_eint ] ) can be re - written as : @xmath433 \\ = & \int_0^{\pi } \sin{\theta_k}{\text{d}}{\theta_k}\int_0^{2\pi } { \text{d}}{\phi_k}\ g({\theta_k},{\phi_k } ) \sum_{j , m_z } i^j ( 2j+1)^{1/2 } { d^j_{m_zp}({\theta_k})}e^{-im{\phi_k } } \left [ { \mathbf{a}_{jm_z}^{(m)}}+ip { \mathbf{a}_{jm_z}^{(e)}}\right ] \\ = & \sum_{j , m_z } i^j ( 2j+1)^{1/2 } { c_{jm_zp}}\left [ { \mathbf{a}_{jm_z}^{(m)}}+ip { \mathbf{a}_{jm_z}^{(e)}}\right ] \end{split } \label{e_emulti}\ ] ] with @xmath434 if the expression for @xmath435 found in ( [ e_gthetaphi ] ) is used in the previous expression ( [ e_cjmp ] ) , the value for the @xmath373 is the following one : @xmath436 as i mentioned in the beginning of the chapter , i will be interested in cylindrically symmetric beams . the helicity has been set to @xmath262 already , so the only other requirement on the incident paraxial field @xmath437 is having an azimuthal dependence of the kind @xmath438 .
in general , this can be achieved by a superposition of lg beams with a fixed azimuthal number @xmath2 : @xmath439 however , for simplicity , the calculations will be done for a single lg@xmath376 mode .
now , using the following orthogonality property of the exponentials , @xmath440 it can be seen that the second integral in ( [ e_cjmp_non ] ) can be analytically solved , setting @xmath441 .
then , @xmath402 can be expressed as a 1-dimensional integral .
its expression , as well as the one for the incident field @xmath291 , is given by the following equation : @xmath442 \\ { c_{jm_z^*p } } & = & \displaystyle \int_0^{{\theta_k}^{m } } \sin { \theta_k}{\text{d}}{\theta_k}\dfrac{fe^{-ikf}}{2\pi } \sqrt{\dfrac{n_1}{n_2 } } \cos^{1/2}({\theta_k } ) { d^j_{m_z^*p}({\theta_k})}n_{l , q } \\ & & \exp\left [ { \dfrac{-(f\sin{\theta_k})^2}{w(z)^2}}\right](f\sin{\theta_k})^l l^l_q(\dfrac{2(f\sin{\theta_k})^2}{w^2(z)})\left ( \dfrac{\sqrt{2}}{w(z ) } \right)^{l+1 } \\ & & \exp \left [ i \left ( \dfrac{-k(f\sin{\theta_k})^2 z}{2(z^2+z_0 ^ 2 ) } + ( 2q+l+1 ) \tan^{-1}(z / z_0 ) \right ) \right ] \end{array } \label{e_eifinal } \renewcommand{{2.5}}{1}\ ] ] ) for different focusing scenarios . the insets represent the intensity plots of the paraxial lg modes at entrance pupil of the lens .
these paraxial modes are focused with two lenses of na@xmath443 and na@xmath444 , and their multipolar decomposition given by @xmath402 is found at the focal plane .
the red coloured bars indicate na@xmath443 , and the yellow ones na@xmath444 .
the paraxial beams used are a ) lg@xmath418 , b ) lg@xmath419 , c ) lg@xmath420 , and d ) lg@xmath421 .
note that @xmath402 can be described with very few multipoles when a high na lens is used . ]
next , i show the influence of the na of a lens in the beam shape coefficients . in figure
[ f_cjmp ] , i have plotted the multipolar decomposition of the same four lg beams used in figure [ f_multipolar ] .
the four lg modes are plotted for two different focusing scenarios , one of them being focused with na@xmath444 ( yellow ) , and the other one with na@xmath443 ( red ) . in all cases
i have used lg modes with a well - defined helicity @xmath423 .
the width of all beams in the paraxial approximation is chosen so that the entrance pupil of the lens is filled .
the results presented in figure [ f_cjmp ] show that the higher the na , the narrower the distribution of multipolar modes is . for some cases such as figure [ f_cjmp](a ) , the beam shape coefficients @xmath402
can be described with only a handful of multipoles . in those cases ,
the value of @xmath445 increases considerably , as the incident field needs to be normalised on a transverse plane @xcite . as a consequence
, the following condition needs to be fulfilled : @xmath446 the other parameter to control the distribution of multipoles is the @xmath158 of the incident beam . as mentioned earlier , the distribution of multipolar amplitudes is zero for values of @xmath447 .
this second degree of freedom can be also used in the paraxial case , as displayed in figure [ f_multipolar ] , but its influence is less important .
also , comparing figures [ f_multipolar ] and [ f_cjmp ] , one can see that in the paraxial case more multipoles are needed to describe the same lg modes , even if a lens with small na@xmath444 has been used .
now , similarly to the previous section , once @xmath402 has been calculated , the fields in the whole space are obtained applying the glmt , and expressions ( [ e_fields ] ) are re - obtained : @xmath448 \\ { \mathbf{e}^\mathrm{int}}&=&\displaystyle\sum_{j=\vert m_z^ * \vert}^{\infty } i^j ( 2j+1)^{1/2 } { c_{jm_z^*p}}\left [ c_j { \mathbf{a}_{jm_z^*}^{(m)}}+ip d_j { \mathbf{a}_{jm_z^*}^{(e)}}\right ] \end{array}\ ] ] where @xmath402 is given by expression ( [ e_eifinal ] ) .
finally , i want to acknowledge that before i published these results in @xcite , based on the work done by my supervisor in @xcite , n.m .
et al . _ had observed the influence of the na of a lens in focusing a plane wave onto plasmonic particles @xcite .
however , their study only dealt with a single focused @xmath128 plane waves , and therefore it could not be applied to any paraxial field .
also , they never applied their formalism to more elaborated beams such as vortex beams . in this section
, i will look into one of the applications of glmt - the excitation of wgms , or morphological - dependent resonances ( mdrs ) @xcite .
wgm and mdr are interchangeable expressions , however the term mdr has been typically only used in the context of glmt , whereas wgm has been used in almost any other branch of physics .
hence , i will use the term wgm throughout this thesis to refer to mdr . in the next pages , i will use paraxial and non - paraxial cylindrically symmetric beams to excite single wgms in dielectric spheres . i will show that provided an adequate ratio of @xmath449 is obtained the excitation beam , a single wgm can be excited with direct light , without the need of going to evanescent coupling . in order to do that
, i will need to compute the scattering efficiency factors defined in section [ ch2_eff ] for an incident field given by expression ( [ e_general ] ) and a scattered and interior fields given by equations ( [ e_fields ] ) .
as described in chapter [ ch2 ] , the first step is computing the temporal average of the poynting vector , @xmath450 .
then , the flux of it across a spherical surface is computed to obtain the scattering cross section @xmath341 .
the general result was given in chapter [ ch2 ] by equation ( [ e_cs ] ) : @xmath451 then , if the result is particularized for @xmath452 and @xmath453 ( see equations ( [ cjmgjm ] , [ e_general ] ) ) , the scattering efficiency factor @xmath454 can be computed @xcite : @xmath455 where @xmath456 , @xmath262 being the helicity of the incident beam , and @xmath2 the topological charge of the lg mode used to produce the focused beam ; and @xmath457 is the size parameter of the problem .
it is insightful to compare equation ( [ e_qs ] ) with the expression for @xmath454 in mie theory .
this expression is @xcite : @xmath458 expressions ( [ e_qs ] ) and ( [ e_qsmie ] ) give rise to very different scattering efficiency as a function of the wavelength .
i showed these differences in @xcite using paraxial lg beams as @xmath291 .
that is , i used the equation ( [ cjlp ] ) to compute @xmath402 .
the differences between the scattering efficiency using equations ( [ e_qs ] ) and ( [ e_qsmie ] ) are summarised in figures [ f_qsmie ] , [ f_qswavelength ] , [ f_enhancement ] .
firstly , i use equation ( [ e_qsmie ] ) to compute the scattering efficiency for mie theory .
the result is shown in figure [ f_qsmie ] .
it can be observed that high order multipoles only give rise to ripples in ths scattering structure .
that implies that the scattering is dominated by the low order modes .
( blue curve , on top ) for a single sphere with @xmath459 and @xmath460 $ ] . at the bottom , four mie coefficients ( @xmath461 , @xmath462 , @xmath463 and @xmath464 ) are plotted .
it is seen that the ripples in the scattering efficiency are caused by high order mie resonances.,width=453 ] secondly , i use equation ( [ e_qs ] ) to plot the scattering efficiency for the same incident four lg modes whose multipolar decomposition has been given in figure [ f_multipolar ] , _ i.e. _ lg@xmath418 , lg@xmath419 , lg@xmath420 , lg@xmath421 .
as seen in figure [ f_qswavelength ] , an increase in the am of the incident beam is translated into an enhancement of the ripple structure with respect to the background produced by the low order modes .
this phenomenon is clear when we compare figure [ f_qswavelength](a ) with ( [ f_qswavelength](d ) ) . in figure [ f_qswavelength](a ) , it is observed that the scattering is dominated by the low multipolar orders , and the high ones only contribute to the scattering as ripples , very similarly to figure [ f_qsmie ] .
furthemore , both of them have the same am content , @xmath465 . ] . nevertheless , in figure [ f_qswavelength](d ) the ripple structure is largely enhanced and the scattering is dominated by the ripple structure itself . as a function of wavelength , @xmath466 $ ] nm , for different incident lg beams .
a ) lg@xmath418 b ) lg@xmath419 c ) lg@xmath420 d ) lg@xmath421 .
all the rest of parameters do not vary from a ) to d ) : @xmath467 , @xmath423 , @xmath468 m .
the multipolar decomposition of each of the lg beams is given in figure [ f_multipolar ] . ] in order to highlight this effect and quantify it , the influence of the same four excitation beams ( lg@xmath418 , lg@xmath419 , lg@xmath420 , lg@xmath421 ) on a single ripple is plotted in figure [ f_enhancement ] .
as a function of wavelength , @xmath469 $ ] nm , and am of the incident beam . the excitation beam is a ) lg@xmath418 b ) lg@xmath419 c ) lg@xmath420 d ) lg@xmath421 .
all the rest of parameters do not vary : @xmath467 , @xmath423 , @xmath468 m .
the multipolar decomposition of each of the lg beams is given in figure [ f_multipolar ] .
the chosen resonance corresponds to @xmath470 , that is , an electric multipole with @xmath471 . ]
the maximum value of the scattering efficiency , @xmath472 , which peaks at @xmath473 nm , and the minimum value are computed for each plot .
the minimum value is computed as the average of two values @xmath474 nm apart in wavelength from the position of the maximum .
then , the ratio @xmath475 can be defined as the relative quotient between the subtraction of the maximum and the minimum , and the minimum .
@xmath476 where @xmath477 and @xmath478 .
the results are presented in table ( [ t_gamma ] ) .
it can be seen that the background is highly reduced . when a lg@xmath418 is used to excite the sphere
, the peak only represents 8@xmath479 of the background
. nevertheless , when the incident beam is a lg@xmath421 , the peak contribution to @xmath454 is two times larger than the addition of all the other modes .
this is easily understood looking at equation ( [ e_qs ] ) . due to the rotational symmetry of the system ,
the am needs to be preserved .
that is , if @xmath291 is an eigenstate of has @xmath158 with value @xmath368 , its multipolar decomposition is such that the first @xmath480 multipolar modes are not included in the scattering .
.enhancement of the ripple structure using different lg beams . [ cols="^,^,^,^,^",options="header " , ] [ t_cgh_charact ]
here , the experimental basics to separate and isolate the 1st diffraction order of a hologram are explained . in section [ ch5_imod ] , a method to separate the different contributions from a hologram has been explained .
indeed , if an oblique phase @xmath481 is added to the hologram expression , the different diffraction orders of a hologram are separated by an angle @xmath482 , where @xmath483 .
now , this implies that at a distance @xmath43 from the hologram the different diffraction orders are separated a transverse distance : @xmath484 nevertheless , the width of the diffraction orders also grows following gaussian optics .
that is , if the beam waist at the hologram plane is @xmath417 , then the waist of every single diffraction order at a distance @xmath43 from the hologram is @xmath485 where @xmath2 is the order of the phase singularity given by the hologram .
it is clear that the condition that assures that two diffraction orders are separated is @xmath486 note that equations ( [ e_x])-([e_x0-w ] ) are valid at distances such that a free - space fourier transform of the field at the plane of the hologram is carried out . in this case , the complex amplitude of the plane wave components at the plane of the hologram are related to different positions in the fourier plane .
experimentally , a free - space fourier transform is carried out when the distance to the hologram @xmath43 is @xmath487 due to lack of space or need to make set - ups compact , it is not always possible to propagate beams .
however , there is another way of doing a fourier transform , and that is using a lens .
a widely used set - up to separate the diffraction orders of a hologram is a 4-f filtering system .
a sketch of a 4-f filtering system is shown in figure [ f_b1 ] . ] on one hand , the diffraction orders are separated a distance @xmath488 at the fourier plane : @xmath489 where @xmath243 is the focal distance of the lens . on the other hand ,
the waist of a diffraction order at the focus of a lens is given by gaussian optics : @xmath490 similarly to the free - space case , the condition that needs to be fulfilled is @xmath491 , which yields the exact same condition given by equation ( [ e_orders ] ) .
a rule of thumb to separate the diffraction orders is : @xmath492 here , the efficiency of the slm is characterized as a function of some controllable parameters in the laboratory , such as the polarization of the incoming beam , the angle formed between the incident beam and the normal to the slm , and the number of @xmath3 phase jumps displayed at the hologram ( see equation ( [ e_kx ] ) ) .
contrary to cghs made of photographic films , the performance of the slm is very dependent on all these parameters .
, width=491 ] , width=491 ] 14 degrees .
the red markers display the efficiency in the y axis , whereas the blue markers display the efficiency in the x axis.[fig616],width=491 ] figures [ fig614 ] and [ fig615 ] show the influence of the polarization and the angle of incidence on the efficiency phase jumps . ] .
in fact , in each of the figures , two lines of data points are plotted .
the red markers correspond the efficiency measured when the phase hologram is @xmath493 , while the blue markers are obtained with @xmath494 .
that is , the first ( red ) one measures the efficiency steering the beam in the @xmath133 axis , and the second one in the @xmath134 axis ( the @xmath134 axis is perpendicular to the plane of the optical table ) . for all the cases
, the vertical axis displays the efficiency of the slm measured in the way it has been described above ; whilst the horizontal axis displays the polarization angle .
this polarization angle is a direct readout of the angular position of a rotation - mount where a half - wave plate is hold .
hence , the value is arbitrary , as a rotation of the slm would make it vary .
however , the number is given so that it can be checked how the maximum is displaced when the slm is tilted .
the first conclusion that can be drown from figures [ fig614 ] and [ fig615 ] is that the efficiency significantly depends on the polarization of the incident beam .
second , it is observed that when the angle between the normal to the slm and the beam axis increases , the efficiency decreases more rapidly for a same range of variation of polarization degrees .
that is , figure [ fig614 ] , which corresponds to almost normal incidence , shows that a 10 degree variation in polarization is matched with a 5@xmath479 variation in efficiency .
figure [ fig615 ] , which depicts the efficiency data obtained when the angle between the incident beam and the normal to the slm is 5 degrees , shows that a 10 degree variation in polarization is matched with a 12@xmath479 variation in efficiency .
furthermore the maximum value for the efficiency drops a little and changes its position .
the position of the maximum varies 5 degrees approximately , which is approximately the angle that the slm has been rotated . in figure
[ fig616 ] , where the results obtained when the angle between the normal to the slm and the incident beam is @xmath49514 degrees are displayed , the same trends commented previously are maintained .
that is , the efficiency of the maximum decreases ; the angle for which the maximum of efficiency is moved as much as the angle between the normal of the slm and the incident beam ; and the gap between the efficiency in the @xmath133 and @xmath134 axis increases .
the pixels are squares , so one would expect that the efficiencies in both axis were the same . however , a tilt of the slm breaks the symmetry between the two axis and accounts for the measured difference .
+ + next , the importance of the hologram displayed on the slm is characterized . in figures
[ fig614]-[fig616 ] the same hologram was used .
the hologram is the oblique phase in figure [ fig65 ] for the efficiency in the @xmath134 axis , and a @xmath496 rotation of it for the @xmath133 axis .
this hologram contains twenty - four @xmath3 phase jumps . in figure [ fig617 ]
the efficiency is plotted as a function of the number of @xmath3 phase jumps given by the hologram .
the angle of incidence is set at 8 ( same as figure [ fig614 ] ) and the optimum position of the half - wave plate angle is set to maximize the efficiency .
obviously , as given by equation ( [ e_kx ] ) , the larger number of @xmath3 phase jumps the hologram gives , the larger is the angle between the 0th and the 1st diffraction order .
then , it can be observed that there is an almost linear relation between the efficiency of the slm and the number of @xmath3 phase jumps in for this range of values .
the larger the number of @xmath3 phase jumps given by the hologram , the lesser pixels per ramp are used , and consequently the efficiency is reduced .
phase jumps displayed on the slm .
the angle formed by the normal to the slm and the beam axis is 8. the polarization is set to optimize the efficiency .
the red markers display the efficiency in the @xmath134 axis , whereas the blue markers display the efficiency in the @xmath133 axis.[fig617],width=491 ] to finalize , the lookup table ( lut ) provided by the producer of the slm ( bns ) was tested .
all the data points collected to plot figures [ fig614]-[fig617 ] were done at @xmath497 nm , using a he - ne laser ( see figure [ fig613 ] ) .
the phase shift given by an slm highly depends on @xmath321 .
the transmittance function created by the liquid crystals depends on @xmath321 as shown in equation ( [ e_txy_general ] ) .
this means that the efficiency of the slm can vary a lot depending on @xmath321 . in order to account for this dependence and some other imperfections of the slm such as non - linear behaviour of the pixels
, luts change the theoretical value of pixels to give them an effective one which increases the performance of the hologram as a whole . in @xcite ,
a linear method to correct the imperfections of the slm was developed .
the idea is to multiply the phase of the phase - hologram @xmath498 by a contrast constant @xmath499 .
then , when the new @xmath500 is such that @xmath501 or @xmath502 , those phases are re - set to @xmath503 or @xmath504 .
using this method , bowman _
_ managed to greatly increase the efficiency of an slm . using a hologram with ten @xmath3 phase ramps and maximizing the polarization angle at an angle of incidence of 8
, i tried to use the same method to improve the efficiency of the slm .
the results are shown in figure [ fig618 ] .
the @xmath134 axis shows the efficiency and @xmath133 axis the contrast parameter @xmath499 .
leaving @xmath505 means not doing any changes to the lut .
it can be observed that the highest efficiency is achieved for @xmath506 .
hence , the lut had been properly optimized at the frequency of the hene laser .
i believe that whenever a new slm or wavelength are used , this is a sanity check that should be done .
in fact , some producers do not send luts along with the slm . in those cases ,
it is highly recommended to apply the method described in @xcite to improve the performance of the slm .
phase ramps .
the angle of incidence is 8 , and the polarization is optimized to maximize the efficiency .
[ fig618],width=491 ]
[ appendix3 ] in this appendix , the equation ( [ iq ] ) is demonstrated . in order to do that ,
the following relations given in chapter [ ch6 ] are used : @xmath508 using all of them , it can be obtained that @xmath509 can be expressed as : @xmath510 following an identical procedure , the following equation yields for @xmath511 : @xmath512 choosing the mirror symmetric operator to be @xmath513 , then equation ( [ -epq ] ) can be used to get a relation between the coefficients in equations ( [ iq],[i - q ] ) : @xmath514 which implies that @xmath515 and @xmath516 , due to the orthogonality of @xmath388 and @xmath517 .
consequently , it follows that : @xmath518 as the integrations limits remain the same under the change @xmath519 .
[ appendix4 ] in the current appendix , intensity distributions of multipolar fields are displayed .
bearing their spatial shape in mind is of special importance to understand the excitation of wgms ( section [ ch3_wgm ] ) , the suppression of backscattering ( [ ch4_kerker ] ) , and the intensity profiles recorded by the ccd camera in the experiments done in chapter [ ch7 ] .
furthermore , it is fairly common to forget about the fact that there are four different kinds of multipolar fields depending on the radial function used : @xmath192 , @xmath520 , @xmath207 , and @xmath208 . also , as it has been shown across this whole thesis , the multipolar fields can be classified either in terms of parity or helicity . hereafter , some frequently used multipolar fields are displayed .
they are displayed for both parities , helicities , and typically used radial functions ( @xmath192 and @xmath207 ) .
the plots are done in the @xmath521 plane , @xmath43 being the horizontal axis and @xmath133 the vertical one . then , because they are cylindrically symmetric around @xmath43 , the expression in the other planes can be deduced by applying a @xmath3 revolution symmetry .
the axis are normalised to the wavelength , so they are dimensionless .
all the units are normalized to @xmath321 .
also , due to the properties of the wigner d - matrices @xmath190 @xcite , the plots have only been done for @xmath522 , as it can be observed that the following properties hold : @xmath523 first , the dipolar orders are displayed .
it is apparent from figure [ d1 ] ( and it can be re - checked with all the rest of figures displaying multipoles with well - defined helicity ) that equation ( [ e_ajmsemi ] ) holds .
in fact , equation ( [ e_ajmsemi ] ) is a consequence of the fact that @xmath524 second , it can also be observed that @xmath202 has the same behaviour for the two @xmath222 and @xmath525 semi - spaces , whilst @xmath526 does not .
this property is the basis of the measurements carried out in chapter [ ch7 ] .
third , the stationary wave behaviour of the bessel function is only observed for @xmath527 , but not for @xmath202 .
remember that a spherical bessel function @xmath192 is a superposition of two spherical hankel functions @xmath528 , where the spherical hankel functions have a respective outwards and inwards radiative behaviour ( see equations ( [ e_jj ] , [ e_hj ] ) in section [ ch1_multipoles ] ) .
in addition , looking at all the captions of the different figures , it is observed that the plot area does not remain constant .
this has been done to visualize the spatial shape of the multipolar fields .
indeed , high order hankel multipolar fields are highly evanescent , therefore the plot area has been reduced .
in contrast , high order bessel multipoles require a larger plot area . also note that a central region surrounding the origin of coordinates has been removed from the plots of hankel multipoles .
the reason for this is that hankel functions are singular at the origin .
finally , figures [ d11]-[d16 ] depict multipolar beams with @xmath529 .
note the similarities between figure [ d11 ] and figure [ f_intensity ] in section [ ch3_wgm ] , where almost a single mode was excited . and last but not least , figure [ d16 ] , which depicts @xmath530 , shows an intensity plot which highly resembles the typical plots for wgms field distribution .
it is interesting to see that those plots are achieved when @xmath531 and @xmath370 is large , in contrast to oraevsky s definition of wgm , which states that a wgm is a multipolar field with very large @xmath370 and @xmath532 @xcite . that , of course , does not affect the q factor of the sphere , as the q factor is only given by the linewidth of a mie resonance , usually the first one for wgms @xcite . and in mie theory
the @xmath360 value is degenerated , therefore a multipolar field @xmath530 has the same q - factor as @xmath533 . and @xmath534 for @xmath143 .
the plot area is a square of side @xmath535 .
a contour of @xmath536 around the origin has been removed to plot @xmath534 .
[ d1],width=396 ] and @xmath537 for @xmath538 .
the plot area is a square of side @xmath535 .
a contour of @xmath536 around the origin has been removed to plot @xmath537 .
[ d2],width=396 ] and @xmath539 for @xmath143 .
the plot area is a square of side @xmath535 .
a contour of @xmath536 around the origin has been removed to plot @xmath539 .
[ d3],width=396 ] and @xmath540 for @xmath538 .
the plot area is a square of side @xmath535 .
a contour of @xmath536 around the origin has been removed to plot @xmath540 .
[ d4],width=396 ] and @xmath541 for @xmath143 .
the plot area is a square of side @xmath542 .
a contour of @xmath536 around the origin has been removed to plot @xmath541.[d5],width=396 ] and @xmath543 for @xmath538.the plot area is a square of side @xmath542 .
a contour of @xmath536 around the origin has been removed to plot @xmath543 .
[ d6],width=396 ] and @xmath544 for @xmath143 .
the plot area is a square of side @xmath542 .
a contour of @xmath545 around the origin has been removed to plot @xmath544 .
[ d7],width=396 ] and @xmath546 for @xmath538 .
the plot area is a square of side @xmath542 .
a contour of @xmath545 around the origin has been removed to plot @xmath546 .
[ d8],width=396 ] and @xmath547 for @xmath143 .
the plot area is a square of side @xmath542 .
a contour of @xmath545 around the origin has been removed to plot @xmath547 .
[ d9],width=396 ] and @xmath548 for @xmath538 .
the plot area is a square of side @xmath542 .
a contour of @xmath545 around the origin has been removed to plot @xmath548 .
[ d10],width=396 ] and @xmath549 for @xmath143 . for @xmath550 ,
the plot area is a square of side @xmath551 . for @xmath549 ,
the plot area is a square of side @xmath552 , where a contour of @xmath553 around the origin has been removed .
[ d11],width=389 ] and @xmath554 for @xmath538 . for @xmath533 ,
the plot area is a square of side @xmath551 . for @xmath554 ,
the plot area is a square of side @xmath552 , where a contour of @xmath545 around the origin has been removed .
[ d12],width=389 ] and @xmath555 for @xmath143 . for @xmath556 ,
the plot area is a square of side @xmath551 . for @xmath555 ,
the plot area is a square of side @xmath552 , where a contour of @xmath553 around the origin has been removed .
[ d13],width=389 ] and @xmath557 for @xmath538 . for @xmath558 ,
the plot area is a square of side @xmath551 .
for @xmath557 , the plot area is a square of side @xmath552 , where a contour of @xmath545 around the origin has been removed .
[ d14],width=389 ] and @xmath559 for @xmath143 . for @xmath560 ,
the plot area is a square of side @xmath551 . for @xmath559 ,
the plot area is a square of side @xmath561 , where a contour of @xmath553 around the origin has been removed.[d15],width=389 ] and @xmath562 for @xmath538 . for @xmath530 ,
the plot area is a square of side @xmath551 . for @xmath562 ,
the plot area is a square of side @xmath561 , where a contour of @xmath545 around the origin has been removed .
[ d16],width=389 ]
general em field c. h. bennett , g. brassard , _
et al . _ _ quantum cryptography : public key distribution and coin tossing_. in _ proceedings of ieee international conference on computers , systems and signal processing _ , vol .
175 , p. 8
( new york , 1984 ) .
k. y. bliokh , e. a. ostrovskaya , m. a. alonso , o. g. rodrguez - herrera , d. lara , and c. dainty . _ spin - to - orbital angular momentum conversion in focusing , scattering , and imaging systems_. opt .
express * 19*(27 ) , 26132 ( 2011 ) .
i. fernandez - corbaton , x. zambrana - puyalto , n. tischler , x. vidal , m. l. juan , and g. molina - terriza .
_ electromagnetic duality symmetry and helicity conservation for the macroscopic maxwell s equations_. phys .
lett . * 111 * , 060401 ( 2013 ) . i. fernandez - corbaton , x. zambrana - puyalto , and g. molina - terriza . _ helicity and angular momentum : a symmetry - based framework for the study of light - matter interactions_. phys . rev .
a * 86 * , 042103 ( 2012 ) .
i. fernandez - corbaton , x. zambrana - puyalto , and g. molina - terriza .
_ on the transformations generated by the electromagnetic spin and orbital angular momentum operators_. j. opt .
b. * 31 * , 2136 - 2141 ( 2014 ) .
a. jorio , a. souza filho , v. brar , a. swan , m. nl , b. goldberg , a. righi , j. hafner , c. lieber , r. saito , _
et al . _
_ polarized resonant raman study of isolated single - wall carbon nanotubes : symmetry selection rules , dipolar and multipolar antenna effects_. phys .
b * 65*(12 ) , 121402 ( 2002 ) . s. carretero - palacios , o. mahboub , f. j. garcia - vidal , l. martin - moreno , s. g. rodrigo , c. genet , and t. w. ebbesen .
_ mechanisms for extraordinary optical transmission through bull s eye structures_. opt .
express * 19*(11 ) , 10429 ( 2011 ) .
l. lorenz .
_ sur la lumire rflchie et rfracte par une sphre transparente_. oeuvres scientifiques de l. lorenz .
revues et annotes par h. valentiner .
tome premier , libraire lehmann & stage , copenhague pp . 403529 ( 1898 ) .
b. maheu , g. gouesbet , and g. grehan .
_ a concise presentation of the generalized lorenz - mie theory for arbitrary location of the scatterer in an arbitrary incident profile_. j. opt . * 19*(2 ) , 59 ( 1988 ) .
g. gouesbet , j. lock , and g. grhan .
_ generalized lorenz - mie theories and description of electromagnetic arbitrary shaped beams : localized approximations and localized beam models , a review_. j. quant .
transfer * 112*(1 ) , 1 ( 2011 ) .
g. molina - terriza , j. p. torres , and l. torner .
_ management of the angular momentum of light : preparation of photons in multidimensional vector states of angular momentum_. phys .
* 88 * , 013601 ( 2001 ) .
h. he , m. friese , n. heckenberg , and h. rubinsztein - dunlop . _
direct observation of transfer of angular momentum to absorptive particles from a laser beam with a phase singularity_. phys .
lett . * 75*(5 ) , 826 ( 1995 ) .
b. thid , h. then , j. sjholm , k. palmer , j. bergman , t. carozzi , y. n. istomin , n. ibragimov , and r. khamitova .
_ utilization of photon orbital angular momentum in the low - frequency radio domain_. phys .
lett . * 99*(8 ) , 087701 ( 2007 ) .
f. tamburini , e. mari , a. sponselli , b. thid , a. bianchini , and f. romanato .
_ encoding many channels on the same frequency through radio vorticity : first experimental test_. new j. phys . * 14*(3 ) , 033001 ( 2012 ) .
x. zambrana - puyalto and g. molina - terriza .
_ the role of the angular momentum of light in mie scattering .
excitation of dielectric spheres with laguerre - gaussian modes_. j. quant .
126*(0 ) , 50 ( 2013 ) .
z. chen , a. taflove , and v. backman .
_ photonic nanojet enhancement of backscattering of light by nanoparticles : a potential novel visible - light ultramicroscopy technique_. opt .
express * 12*(7 ) , 1214 ( 2004 ) .
d. grard , j. wenger , a. devilez , d. gachet , b. stout , n. bonod , e. popov , h. rigneault , _
et al . _
_ strong electromagnetic confinement near dielectric microspheres to enhance single - molecule fluorescence_. opt .
express * 16*(19 ) , 15297 ( 2008 ) .
a. garca - etxarri , r. gmez - medina , l. s. froufe - prez , c. lpez , l. chantada , f. scheffold , j. aizpurua , m. nieto - vesperinas , and j. j. senz .
_ strong magnetic response of submicron silicon particles in the infrared_. opt .
express * 19*(6 ) , 4815 ( 2011 ) .
d. s. filonov , a. e. krasnok , a. p. slobozhanyuk , p. v. kapitanova , e. a. nenasheva , y. s. kivshar , and p. a. belov .
_ experimental verification of the concept of all - dielectric nanoantennas_. appl .
100*(20 ) , 201113 ( pages 4 ) ( 2012 ) .
k. bendtz , d. milstead , h .-
hchler , a. m. hirt , p. mermod , p. michael , t. sloan , c. tegner , and s. b. thorarinsson .
_ search for magnetic monopoles in polar volcanic rocks_. phys .
lett . * 110 * , 121803 ( 2013 ) .
b. garca - cmara , r. a. de la osa , j. m. saiz , f. gonzlez , and f. moreno .
_ directionality in scattering by nanoparticles : kerker s null - scattering conditions revisited_. opt .
lett . * 36*(5 ) , 728 ( 2011 ) .
j. geffrin , b. garca - cmara , r. gmez - medina , p. albella , l. froufe - prez , c. eyraud , a. litman , r. vaillon , f. gonzlez , m. nieto - vesperinas , j. senz , and f. moreno . _ magnetic and electric coherence in forward- and back - scattered electromagnetic waves by a single dielectric subwavelength sphere_. nat commun * 3 * , 1171 ( 2012 ) .
r. gmez - medina , b. garca - cmara , i. surez - lacalle , f. gonzlez , f. moreno , m. nieto - vesperinas , and j. j. senz . _ electric and magnetic dipolar response of germanium nanospheres : interference effects , scattering anisotropy , and optical forces_. j. nanophotonics * 5*(1 ) , 053512 ( 2011 ) .
n. tischler , i. fernandez - corbaton , x. zambrana - puyalto , a. minovich , x. vidal , m. l. juan , and g. molina - terriza .
_ experimental control of optical helicity in nanophotonics_. light sci appl * 3 * , e183 ( 2014 ) .
x. vidal , x. zambrana - puyalto , a. f. barbara , i. fernandez - corbaton , and g. molina - terriza .
_ chiral elements and dual systems : towards the design of an omnidirectional optical active media_. in _
spie micro+ nano materials , devices , and applications _ , pp .
89233j89233j ( international society for optics and photonics , 2013 ) .
h. he , n. heckenberg , and h. rubinsztein - dunlop .
_ optical particle trapping with higher - order doughnut beams produced using high efficiency computer generated holograms_. j. mod .
* 42*(1 ) , 217 ( 1995 ) .
r. w. bowman , g. m. gibson , a. linnenberger , d. b. phillips , j. a. grieve , d. m. carberry , s. serati , m. j. miles , and m. j. padgett .
_ `` red tweezers '' : fast , customisable hologram generation for optical tweezers_. comput .
commun . * 185*(1 ) , 268 ( 2014 ) .
a. f. demirors , a. jannasch , p. d. j. van oostrum , e. schaffer , a. imhof , and a. van blaaderen .
_ seeded growth of titania colloids with refractive index tunability and fluorophore - free luminescence_. langmuir * 27*(5 ) , 1626 ( 2011 ) .
a. j. haes and r. p. van duyne . _ a nanoscale optical biosensor : sensitivity and selectivity of an approach based on the localized surface plasmon resonance spectroscopy of triangular silver nanoparticles_. j. am .
soc . * 124*(35 ) , 10596 ( 2002 ) .
s. carretero - palacios , f. j. garca - vidal , l. martn - moreno , and s. g. rodrigo .
_ effect of film thickness and dielectric environment on optical transmission through subwavelength holes_. phys .
b * 85 * , 035417 ( 2012 ) . j .- m .
yi , a. cuche , f. de len - prez , a. degiron , e. laux , e. devaux , c. genet , j. alegret , l. martn - moreno , and t. w. ebbesen .
_ diffraction regimes of single holes_. phys .
* 109 * , 023901 ( 2012 ) .
a. guerrero - martnez , b. augui , j. l. alonso - gmez , z. doli , s. gmez - graa , m. ini , m. m. cid , and l. m. liz - marzn .
_ intense optical activity from three - dimensional chiral ordering of plasmonic nanoantennas_. angew .
int . edit . *
50*(24 ) , 5499 ( 2011 ) .
a. kuzyk , r. schreiber , z. fan , g. pardatscher , e .-
roller , a. hgele , f. c. simmel , a. o. govorov , and t. liedl . _ dna - based self - assembly of chiral plasmonic nanostructures with tailored optical response_. nature * 483*(7389 ) , 311 ( 2012 ) .
n. a. abdulrahman , z. fan , t. tonooka , s. m. kelly , n. gadegaard , e. hendry , a. o. govorov , and m. kadodwala .
_ induced chirality through electromagnetic coupling between chiral molecular layers and plasmonic nanostructures_. nano lett . *
12*(2 ) , 977 ( 2012 ) .
x. vidal , w. j. kim , a. baev , v. tokar , h. jee , m. t. swihart , and p. n. prasad .
_ coupled plasmons induce broadband circular dichroism in patternable films of silver nanoparticles with chiral ligands_. nanoscale * 5 * , 10550 ( 2013 ) .
r. bowman , v. dambrosio , e. rubino , o. jedrkiewicz , p. trapani , and m. padgett .
_ optimisation of a low cost slm for diffraction efficiency and ghost order suppression_. eur .
j .- spec .
top . * 199*(1 ) , 149 ( 2011 ) .
s. arnold , t. okeeffe , k. leung , l. folan , t. scalese , and a. pluchino .
_ optical bistability of an aqueous aerosol particle detected through light scattering : theory and experiment_. appl . opt . * 29*(24 ) , 3473 ( 1990 ) .
v. garbin , g. volpe , e. ferrari , m. versluis , d. cojoc , and d. petrov .
_ mie scattering distinguishes the topological charge of an optical vortex : a homage to gustav mie_. new j. phys . *
11*(1 ) , 013046 ( 2009 ) .
e. karimi , b. piccirillo , e. nagali , l. marrucci , and e. santamato .
_ efficient generation and sorting of orbital angular momentum eigenmodes of light by thermally tuned q - plates_. appl .
. lett . * 94*(23 ) , 231124 ( 2009 ) . | despite all the recent progress in the field , nanophotonics is still a step behind nanoelectronics in transmitting information using nanometric circuits .
a lot of effort is being put into making very elaborate structures that can guide light and control light - matter interactions at the nano - scale .
the field of plasmonics has been especially successful in this . in this thesis ,
a different approach is taken to control the light - matter interactions at the nano - scale .
the approach is based on considering light and sample as a whole system and exploiting its symmetries .
thanks to this new perspective , new phenomena have been unveiled .
these new phenomena have been developed theoretically and/or experimentally and are scattered across this thesis .
+ in chapter [ ch1 ] , the theoretical grounds of this thesis are settled . even though every physicist is familiar with the concept of symmetry , a formalism to systematically describe the symmetries of electromagnetic fields is explained . with this formalism , some well - known symmetry considerations can be as easily retrieved as some much less intuitive .
for example , it can be demonstrated that a linearly polarised bessel beam is not cylindrically symmetric ; whilst a circularly polarised bessel beam is both cylindrically and dual symmetric .
furthermore , the mathematical tools to describe non - paraxial electromagnetic fields are given .
due to the fact that this work deals with sub - wavelength scatterers , the light - matter interaction can not usually be described within the paraxial approximation . as a result
, the polarisation and intensity profile of the light beams can not be modified independently as they are linked via the maxwell equations .
+ chapters [ ch2 ] , [ ch3 ] and [ ch4 ] deepen in the study of generalized lorenz - mie theory . using the formalism developed in chapter [ ch1 ] , various new effects are discovered . in chapter
[ ch3 ] , the excitation of wgm modes on micron - sized spheres is described . indeed , using cylindrically symmetric beams
, light can be coupled into spherical resonators without the use of evanescent coupling .
furthermore , it is shown that the use of cylindrically symmetric modes also allows for the enhancement of the ripple structure in scattering .
finally , chapter [ ch4 ] generalizes the kerker conditions and uses cylindrically and dual symmetric beams to control the helicity content in scattering .
it is shown that non - dual materials such as tio@xmath0 spheres can behave as dual if the correct excitation beam and wavelength is used to illuminate them .
+ chapters [ ch5 ] , [ ch6 ] and [ ch7 ] are devoted to experiments . in chapter [ ch5 ] , a description of the experimental techniques used in chapters [ ch6 ] and [ ch7 ] is carried out .
in particular , the basics of spatial light modulators and computer generated holograms are given .
spatial light modulators are used in chapters [ ch6 ] and [ ch7 ] to create vortex beams . in chapter
[ ch6 ] , the symmetries of these vortex beams turn out to be crucial to induce a giant circular dichroism in a non - chiral sample .
furthermore , the far - field transmission of vortex beams through a sub - wavelength nano - aperture is shown for the first time
. finally , chapter [ ch7 ] presents the dependence of scattering measurements on the wavelength and the topological charge of the incident vortex beam .
as predicted in chapter [ ch3 ] , it is seen that some scattering resonances are hidden under a gaussian beam excitation .
these resonances can be unveiled when the illumination is a vortex beam .
+ overall , this work shows a number of new effects ( theoretical and/or experimental ) produced by the excitation of symmetric structures with symmetric light .
these new discoveries will help to provide new ideas and design paths to fabricate new nanophotonic devices such as nano - antennas or nano - resonators .
a study of the symmetries of the system should always be kept in mind for any photonic device where the spatial degrees of freedom and the polarisation can not be decoupled . |
many issues related to the formation and evolution of galactic bars remain unsettled ( e.g. , villa - vargas , shlosman & heller 2009 , hereafter paper i ) . in paper
i , we have revisited the properties of barred disks embedded in the dark matter ( dm ) halos , with an emphasis on the angular momentum redistribution in _ collisionless _ systems . here
we attempt to understand some effects of the gas component on the bar evolution . in a follow - up work ,
we investigate the effect of varying the halo parameters in the gaseous / stellar disk models . due to its viscous and dissipative nature , the gas can influence the stellar component in the disk , well beyond its observed mass fraction ( e.g. , shlosman & noguchi 1993 ; friedli & benz 1993 ; heller & shlosman 1994 ; berentzen et al .
1998 ; bournaud & combes 2002 ; berentzen et al .
2007 ; curir et al . 2007 ) . the gas dissipation , especially in the bar s presence ,
triggers the mass redistribution in the disk .
a substantial fraction of barred orbits are self - intersecting , which while irrelevant for stars has a strong effect on the gas .
the gas populating such orbits is short - lived , resulting in shocks and loss of rotational support .
different viscosities in gas and stellar ` fluid ' lead to a delayed response to the gravitational torques and to an exchange of angular momentum between these components .
the loss of angular momentum in the gas induces a flow down to the inner disk , forming a central mass concentration ( cmc ) , and possibly forming and fueling the central supermassive black hole ( sbh ) via nested bars mechanism ( e.g. , shlosman et al .
1989 , 1990 ; pfenniger & norman 1990 ; friedli & martinet 1993 ; knapen et al .
1995 ; begelman & shlosman 2009 ; hopkins & quataert 2010 ) , and additional processes closer to the sbh ( e.g. , shlosman 1999 ; hopkins et al .
the gas is capable of modifying the disk orbital structure ( e.g. , berentzen et al .
moreover , the cold gas is inherently clumpy which effectively heats up the stellar ` fluid ' and populates the disk with an increased fraction of chaotic orbits , resulting in a decay of the bar strength ( e.g. , shlosman & noguchi 1993 ; berentzen et al . 2007 ) .
the vertical buckling instability in the bar is progressively damped by the two - fluid gas - star interaction ( e.g. , berentzen et al .
2007 ) .
formation and evolution of stellar bars is intricately related to the redistribution of angular momentum , @xmath3 , in the disk - halo system .
this means the presence of sources and sinks of angular momentum .
sellwood ( 1981 ) has shown that the former reside in the disk and the latter in the dm halo ( see also debattista & sellwood 1998 ; athanassoula 2002 ) . interactions of sources and sinks of @xmath3 are dominated by resonances ( lynden - bell & kalnajs 1972 ; athanassoula 2002 ; martinez - valpuesta , shlosman & heller 2006 ; paper i ) .
details of these resonance interactions are not fully understood and quantified .
this is especially true in the presence of the gas component , as the above cited works are based on a purely collisionless modeling . in this paper
we analyze the effect of the gaseous component on the evolution of galactic bars and take a numerical approach . in particular , we vary the gas content in the disk and advance models with various numerical resolution , but keep the stellar disk and dm halo paramaters unchanged , using the standard model of paper i. we are especially interested in how the gas presence and the associated numerical resolution affect the basic parameters of a stellar bar , as well as the angular momentum transfer process in a barred disk and in a disk - halo system . for this purpose
, we perform a detailed comparison of models with gas to the standard collisionless model published in paper i.
we use the hybrid @xmath4-body and smooth particle hydrodynamics ( sph ) ftm-4.4 code ( e.g. , heller & shlosman 1994 ; heller , shlosman & athanassoula 2007 ; romano - diaz et al .
2009 ) to evolve the stellar and gaseous disks embedded in the dm halos .
the gravitational forces are calculated using the falcon routine ( dehnen 2002 ) which scales as @xmath5 .
the adopted units are the same as in paper i : the units of mass and distance are taken as @xmath6 and 10 kpc respectively .
this makes the unit of time equal to @xmath7 yr , when @xmath8 , and the velocity unit @xmath9 .
the gravitational softening is @xmath10 for stars and dm particles . for the gas particles we use a dynamical softening .
the gravitational softening is set to the smoothing length unless the smoothing length falls below the fixed limiting value @xmath11 .
the models consist of a stellar disk with @xmath12 , a gas disk with @xmath13 and dm halo with @xmath14 collisionless particles .
models were evolved for about a hubble time , @xmath15 in the adopted units , which translates to 12.7 gyr . during the model evolution ,
we have routinely observed the formation of very compact accumulations of gas particles in the central disk region . in the models with a substantial gas component
we have observed the formation of secondary gaseous bars which decoupled from the large - scale bars and contracted subsequently to spatial scales where insufficient resolution resulted in flattened ` blobs ' a few softening lengths in radius , in agreement with simulations of englmaier & shlosman ( 2004 ) . besides the gradual capture of additional gas particles ,
the morphological structure of the blob has evolved very little and was beyond the resolution of our models . for the sake of shortening the computational time , we have replaced the gas particles trapped in the center by stellar ones .
this operation was repeated whenever the central gas accumulation was substantially slowing down the overall evolution .
we kept the softening of the individual particles when they were converted from one type to the other .
we have run a large number of tests to verify that this action did not affect the evolution of the bar .
this was achieved by running parallel models with and without the gas particle replacement .
the initial conditions of the stellar and dm particles were created with the procedures described in paper i , using the density profiles from hernquist ( 1993 ) .
the mass volume density distribution in the disk is given in cylindrical coordinates by @xmath16 where @xmath17 is the disk mass , @xmath18 is a radial scale length and @xmath19 is a vertical scaleheight .
the density of the spherical halo is given by @xmath20 where @xmath21 is the mass of the halo , @xmath22 is a gaussian cutoff radius and @xmath23 is the core radius .
@xmath24 is the normalization constant defined by @xmath25 \ } ^{-1}\ ] ] with @xmath26 .
the particle velocities , dispersion velocities and asymmetric drift corrections were calculated using moments of the collisionless boltzmann equation . since models thus constructed are not in exact virial equilibrium , the halo component was relaxed for @xmath27 in the frozen disk potential .
because we are interested to quantify the effect of the gas fraction and _ gas _ spatial resolution on the bar evolution , we use the pure stellar model sd from paper i as our benchmark model ( table 1 ) .
a fixed fraction @xmath0 of stellar disk particles at @xmath28 were converted to identical mass gas particles and re - balanced using the central attraction forces from the total mass distribution .
the gas is considered to be isothermal with @xmath29 k , and initially moves on circular orbits .
we have created a set of models covering a two dimensional parameter space : by varying the gas mass fraction , @xmath0 , in the disk , and by changing the limiting value of the gravitational softening , @xmath11 , in the gas .
the sum of stellar@xmath30gas mass was kept constant in the models .
we used the values @xmath31 , 2% , 4% , 8% , 15% , 30% and 50% , and @xmath32@xmath33 , 0.05 and 0.1 . the softening used for the stellar and dm components
is fixed at 0.016 .
table [ param ] shows the combination of @xmath0 and @xmath32 values used in each model .
.parameters of the standard model [ cols= " < , > , < , > " , ] [ sd_param ] l r r + model & @xmath0(% ) & @xmath32 + sd & 0 &
+ + sd_g2s1 & 2 & 0.016 + sd_g4s1 & 4 & 0.016 + sd_g8s1 & 8 & 0.016 + sd_g15s1 & 15 & 0.016 + sd_g30s1 & 30 & 0.016 + sd_g50s1 & 50 & 0.016 + + sd_g2s2 & 2 & 0.050 + sd_g4s2 & 4 & 0.050 + sd_g8s2 & 8 & 0.050 + sd_g15s2 & 15 & 0.050 + sd_g30s2 & 30 & 0.050 + sd_g50s2 & 50 & 0.050 + + sd_g2s3 & 2 & 0.10 + sd_g4s3 & 4 & 0.10 + sd_g8s3 & 8 & 0.10 + sd_g15s3 & 15 & 0.10 + sd_g30s3 & 30 & 0.10 + sd_g50s3 & 50 & 0.10 + [ param ]
disks with a high content of gas tend to form dense clumps of gas which interact with the stellar component and raise the velocity dispersion in the disk . if this rise in the disk ` temperature ' is substantial , the bar instability may be lessened or completely suppressed ( shlosman & noguchi , 1993 ) .
we have encountered this problem when evolving models with @xmath0=100% which prevented us from completing these runs .
the _ stellar _ bar strength has been quantified using the fourier amplitude @xmath34 of the @xmath35 mode normalized by the @xmath36 mode ( paper i ) .
it is obtained by integration over restricted cylindrical volumes where the bar is a dominant morphological feature , namely , over @xmath37@xmath38 range , where @xmath38 is the bar size defined in section 3.3 ( see also paper i ) . in figure
[ a2b ] , we plot @xmath34 as a function of time @xmath39 for various gas fractions , @xmath0 , and gravitational softening in the gas , @xmath32 .
the initial stages in the evolution of @xmath34 are very similar in all models : an initial stage of an accelerated growth , and a peak followed by a sudden drop .
the duration of this dynamical stage varies from model to model , but is completed by @xmath40 .
the first peak in @xmath34 is followed by the vertical buckling instability in the bar leading to an abrupt weakening in the bar but not a complete dissolution ( e.g. , martinez - valpuesta & shlosman 2004 ) .
we observe that the bar weakening is quite independent of @xmath0 and @xmath32 , as expected . for gas - poor models ,
the first peak of @xmath34 is independent of @xmath32 and @xmath0 . for gas - rich models , with @xmath0@xmath41 , the peak is lowered gradually for @xmath32@xmath42 , up to a factor of 0.4 .
models with @xmath32@xmath43 appear not to be affected at all by this trend .
the post - buckling evolution of the bars is much more affected by @xmath0 and @xmath32 , and shows a bimodal behavior .
namely , in gas - poor models , the bar resumes its growth but at a more gradual pace as compared to the dynamical growth . in gas - rich models ,
the bar strength declines over the simulation ( i.e. , hubble ) time .
this bimodal behavior was noted by berentzen et al .
( 2007 ) , but the current models show it more explicitly
. we shall refer to the first bar evolution phase , including the buckling , as the _ dynamical _ phase , and to the subsequent evolution as the _ secular _ phase .
the secular stage allows the separation of the secularly growing models from the secularly declining ones . in no models
have the bars disappeared completely at least a substantial oval distortion remained .
the borderline @xmath0 between these two trends , growth and decline , depends on @xmath32 . for @xmath32@xmath44
, it lies in the range of @xmath0@xmath45 .
larger @xmath32 moves the borderline @xmath0 toward more gas - rich models .
models sd_g8s2 and sd_g15s3 lie close to this borderline , @xmath0@xmath46 for @xmath32@xmath47 and @xmath0@xmath48 for @xmath32@xmath43 , and show a very singular evolution after the sudden drop @xmath34@xmath49const . for about @xmath50 , and a second period of bar growth
this is a surprisingly mixed behavior showing the prolonged constancy in @xmath34 of the gas - rich models and the secular bar growth of the gas - poor ones .
model sd_g50s2 has a peculiar evolution in the dynamical stage that deserves special attention .
even though this is a gas - rich model , the bar resumes is secular growth after the buckling .
the rate of growth is much more moderate than that in the other models but is clearly noticeable .
the evolution of the pattern speed ps .
is shown in fig .
[ figps ] .
it remains about constant during the dynamical phase of the bar evolution , slightly rising in the gas rich models .
the bimodality of gas - poor and rich models shows up in the secular evolution where ps .
drops in the former and stays constant or rises in the latter disks . in the secular phase , ps .
nearly always anticorrelates with @xmath34 a sustained drop in ps
. corresponds to a rise in @xmath34 .
thus , in the gas - poor models , the bar tumbling slows down , while in the gas - rich models it stays nearly constant .
2 provides a hint to what one can expect for the evolution of the angular momentum in the disk - halo system , although in no way does this figure account for _ all _ the angular momentum in the disk , and even in the bar region it represents only the tumbling of the bar .
the gas - poor disks lose their tumbling angular momentum efficiently , while gas - rich ones nearly conserve it or even speed up their tumbling during the secular phase .
the length of the bar semi - major axis , @xmath38 , has been taken as the radius where the bar equatorial ellipticity drops by 15% off its peak ( paper i ) .
this method has been tested in comparison with alternative method based on the last stable orbit supporting the bar ( martinez - valpuesta et al . 2006 ) .
it is the most reliable when applied after the first maximum of the bar strength .
the ellipticity of the bar at different radii is obtained by fitting ellipses to the isodensity contours in the face - on disk . as in the collisionless models
, the bar length exhibits the initial period of an accelerated growth and reaches a maximum which always coincides ( in time ) with the maximum in @xmath34 ( fig .
[ figrbar ] ) .
this symbolizes the period of a dynamic growth of the bar , or growth related to the bar instability itself .
similarly to the pure stellar model , there is a subsequent drop in @xmath38 as a result of the vertical buckling ( section 3.4 ) . during the secular phase ,
the bar length exhibits a more complicated behavior than in the collisionless models .
most importantly , it grows in gas poor models while it stagnates or even shortens in the gas rich disks . in the late stages of secular evolution ,
the gas poor models show a sudden decrease in the bar size which has no immediate correspondence to @xmath34 evolution .
rather generally , the evolution of @xmath38 does not go in tandem with changes in @xmath34 . as an example , in model sd_g4s1 after @xmath51 , @xmath38 continues its growth while @xmath34 growth saturates .
this behavior is caused by the formation , growth and detachment of the _ ansae _ , as discussed in paper i. in the following , we shall show that @xmath38 correlates with other properties of the bar - disk system . the bar corotation ( cr ) radius @xmath52 has been computed using linear approximation .
@xmath52 grows or shrinks as a consequence of the variation of the bar tumbling speed , ps . .
it grows substantially in the sd and gas poor models , while it stays constant or even drops slightly in the gas rich models .
its initial value among all models is the same . as in paper
i , we follow the ratios of @xmath52 to that of the disk @xmath53 ( fig
. [ figrcrrd ] ) and bar @xmath38 sizes .
both @xmath53 and @xmath38 have been defined in paper i. clearly , when @xmath52/@xmath53@xmath54 , its growth experiences a temporary slowdown and then resumes , rising to larger values .
this effect is the result of the combination of a temporary slowdown of the outwards radial movement of @xmath52 and a temporary expansion of the disk radial border which results from the drag on some of the disk material trapped at the cr resonance . as in the pure stellar models ,
the growth of the bar strength is sensitive to the moment at which @xmath52 crosses the edge of the disk , seen as a temporary period of slower growth of @xmath34 .
note , that @xmath52/@xmath53 is always less than unity in the gas rich models and it seems that for a fixed @xmath0 , the cr - to - disk size ratio is damped stronger for small @xmath32 , i.e. , for higher numerical resolution . the ratio @xmath52/@xmath38 shows a pronounced dip during the initial growth of the bar .
this is followed by an equally abrupt rise due to the bar shortening that results from the vertical buckling instability .
after these variations , both @xmath38 and @xmath52 grow in tandem , and @xmath52/@xmath38@xmath55 in most of the models .
exceptions are models sd_g4s2 which has a rise and drop resulting from a temporary stall of @xmath38 , and models sd_g30s3 and sd_g50s3 where @xmath52@xmath49 const . while the bar shrinks .
finally , a pronounced rise above @xmath52/@xmath38@xmath56 occurs in models due to detachment of the ansae ( paper i ) .
this rise is absent in models , in which no ansae are formed or which form too late in the run and do not have time to detach from the bar , as in sd_g15s3 .
we find that the ansae do not form in models with more than @xmath57 of the gas .
we note that the bar size evolution given in fig .
3 , includes the ansae
. however , in section 4 , we find it advantageous to exclude the ansae when discussing some correlations .
the vertical buckling observed in many numerical bars is an event that in some cases can reshape the phase - space density of the disk . as we found in paper i , when the disk is heated vigorously by the buckling instability , the secular growth of the bar can be seriously diminished or completely halted .
we have measured the vertical asymmetry of the stellar disk with the index @xmath58 , computed as described in paper i. the evolution of @xmath58 is shown in fig .
[ figa1z ] .
most of the models have at least one clear peak between @xmath59 and 100 , and this peak coincides in time with the drop in @xmath34 .
exceptions are models sd_g15s1 , and sd_g50s2 , which have no clear peak above the noise level .
some models have secondary and even higher order peaks or prolonged periods where @xmath58 is clearly above the noise level , as found by martinez - valpuesta et al .
( 2006 ) .
clearly , the height of the first peak varies with the gas content of the disk . in the sequences with @xmath32@xmath44 and 0.05 ,
the peak becomes gradually lower in models with higher @xmath0 .
this result is in good agreement with berentzen et al .
the big surprise is in the sequence with @xmath32@xmath43 , where the value of the peak drops as @xmath0 increases from 0% to 8% , and then rises in models with @xmath0@xmath60 and 50% .
this is discussed further in section [ s_disc ] .
the signature of the vertical buckling can be observed in the behavior of ps . .
as the bar strengthens , it brakes against the outer disk and against the dm halo , at later time and ps .
declines . at the time of buckling ,
the bar weakens abruptly and ps .
experiences a break its slope changes and becomes much smaller .
one of the differences that arise between pure stellar models of galactic bars and models including a gas component is the formation of a central mass concentration .
the central densities , both stellar and dm , have notably increased even in pure stellar models as a response to the asymmetric potential of the bar and the vertical buckling ( paper i ; dubinski , berentzen & shlosman 2009 ) .
however , this effect is by far more intense in models with gas .
the growth of central mass concentration ( cmc ) is boosted by the bar which channels the gas toward the central kpc .
this leads to a cmc composed mainly of gas , and , in smaller proportion , of stellar and dm particles . in all our models
, the cmc has resided inside a radius smaller than 0.1 . to follow the growth of the cmc
, we have measured the mass @xmath61 contained within a sphere of radius 0.1 placed at the center of mass of the disk .
contributions from all three components have been included ( fig .
6 ) .
model sd_g2s1 serves as a benchmark of evolution in the gas poor models . with the formation of the stellar bar , a massive accumulation of gas appears at the center , with an elongated shape of @xmath62 , in the disk plane
this cmc is approximately aligned with the major axis of the ( gas ) bar .
it rapidly captures the gas fed by the bar and simultaneously contracts into a small , more axisymmetric and somewhat flattened circular ` blob ' with @xmath63 and @xmath64 .
these characteristic sizes are determined largely by @xmath32 .
the accelerated @xmath61 rise seen in fig .
6 encompasses the exponential bar growth phase , up to the time of buckling ( or the time when @xmath34 drops in models with no buckling ) .
typically , about 50% of the gas mass is captured by the cmc during this stage . during the secular phase of the bar evolution , the cmc captures the gas at a much slower pace .
this is mostly the gas which remains outside @xmath38 and even @xmath52 at the onset of the secular phase .
the slowing down of the cmc growth results from a combination of two factors .
first , the availability of the disk gas has been severely reduced by the violent initial inflow , and second , the bar strength has been diminished by the buckling . as a consequence
, the gas - poor models have been left with very little gas in the disk after the initial inflow , although they show a healthy secular bar growth . whereas gas - rich models have gas left , but their bars are weak and do not grow , so no fresh gas crosses @xmath52 which stagnates as well .
three models exhibit a somewhat different evolution for @xmath61 . instead of a short period of a very accelerated growth
, the cmc grows at a somewhat slower rate but over an extended period of time .
these are models sd_g15s3 , sd_g30s3 and sd_g50s3 , which are all gas - rich and have the lowest resolution . in these models ,
the gas content of the bar has formed without a visibly prominent cmc . instead of a rapid influx to the center , the gas in the bar contracts rather slowly , gradually increasing its density at the center .
the process saturates when the gas is completely contained within the central @xmath65 .
models sd_g8s2 and sd_g15s3 show a rate of @xmath61 growth that increases with time at the end of the run .
this behavior is related to the late period of accelerated bar growth .
angular momentum evolution in the disk has been followed within a number of characteristic radii , namely , the cr ( @xmath66 ) and the disk radius , @xmath67 ( which contains 98% of the disk mass by definition ) . the latter exhibits a clear evolutionary sequence as a function of @xmath0 ( e.g. , fig .
both the amount of @xmath3 lost by the disk over dynamical and secular evolution decreases monotonically from pure stellar disks to progressively more gas - rich ones . in a way
this is a reflection of weaker bars along @xmath0 .
even more interesting is the evolution of @xmath66 ( fig .
after the initial adjustment , the disk within the cr is losing its angular momentum .
the loss is not dramatic , @xmath68 , and somewhat increases with @xmath0 .
this decline in @xmath66 continues until the vertical buckling sets in . for low @xmath0 disks ,
the buckling is associated with a nearly sudden increase in @xmath66 which restores the pre - buckling value followed by a subsequent gradual decline .
gas - rich disk show no sudden increase in @xmath66 but rather a slow decline .
@xmath66 stays quite flat until @xmath69 , a long time after the buckling , as can bee seen from fig .
8 ( with the exception of a small initial decline and increase , as mentioned above ) .
interestingly , this time corresponds to @xmath52/@xmath53@xmath70 in the gas - poor models ( fig .
crossing @xmath52/@xmath53@xmath71 border affects the bar growth ( fig .
3 ) , which saturates immediately .
it also is reflected in the evolution of @xmath52/@xmath38 which increases abruptly thereafter .
clearly , the bar growth ceases at some point when the @xmath52 exceeds @xmath53 and the bar can not capture additional orbits and be fed by the angular momentum from these orbits .
this situation is similar to that analyzed in paper i ( section 3.6 ) , where the angular momentum of the disk inside @xmath52 stays about constant as long as @xmath52 remains within the disk .
our present models can be directly compared to the standard model of paper i. we return to this point in section 4 .
the loss of the angular momentum , @xmath72 , with @xmath0 by the outer disk appears to be much more severe than @xmath66 .
but one should remember that the outer disk , i.e. , outside the cr , accounts for less mass . before the buckling
, @xmath72 behaves similarly to @xmath66 , but differently at the later times .
the trend here is that @xmath72 declines steeply with time for gas poor disks after the buckling period .
it stays nearly constant with time for the gas rich models .
models with different @xmath32 show a behavior consistent with our understanding .
there is a slight increase in the @xmath3 transfer during dynamical and secular stages of the bar evolution .
much less difference is observed in the evolution of @xmath66 and @xmath72 in this case .
we have studied the effects of gas fraction , @xmath0 , and spatial resolution , @xmath32 , on some aspects of bar evolution in galactic disks embedded in dm halos . specifically , we aimed at understanding their effect on the basic parameters characterizing stellar bars , e.g. , bar size , corotation radius , strength , etc .
we have also followed the angular momentum redistribution in the disk - halo systems as a function of @xmath0 and @xmath32 . the sd pure stellar model ( paper i ) has been used as a template the disk and halo parameters in current models have been fixed at those of the sd .
as before , the bar evolution has been separated into two main phases the dynamical phase , which represents the bar instability itself de facto , and the secular phase .
these phases are separated by the vertical buckling instability in the bar .
we investigate a number of new correlations which are corollaries to evolution discussed in section 3 . in the follow - up work ( villa - vargas et al .
, in preparation ) , we analyze the angular momentum transfer and bar evolution as a function of disk gas fraction and by varying the basic parameters of the dm halo .
overall , there appears to be a substantial difference in the evolution of the gas - poor and rich models , as has been shown in the previous section .
this is discussed below in more detail .
the boundary between the gas - poor and gas - rich models depends on the spatial resolution for the gas component it lies around 5%7% for @xmath32=0.016 , and shifts to @xmath48 for @xmath32=0.1 .
the loss of spatial resolution , in a way , changes the nature of the gas component it becomes less dissipative .
we define the bar strengthening during its dynamical phase as @xmath73@xmath34 @xmath74@xmath34(@xmath75)@xmath34(@xmath28 ) , where @xmath76 is the time when the bar has reached its maximal amplitude prior to the onset of buckling ; the second term @xmath34(@xmath28)@xmath77 .
similarly , the bar growth during the secular phase is defined as @xmath73@xmath34 @xmath74@xmath34(@xmath78 )
@xmath34(@xmath79 ) , where @xmath80 is the time of the minimal bar amplitude immediately following the buckling . in the _ dynamical _ stage of the bar evolution
, the largest differences have been observed in the gas - rich models , where @xmath73@xmath34 drops dramatically with @xmath0 for @xmath32@xmath44 and 0.05 , while it exhibits no dependence on @xmath0 for @xmath32@xmath43 ( figs . 1 , 9a ) .
in the gas - poor models , the maximal bar amplitude achieved in the dynamical phase is nearly independent of the gas fraction , nor does it depends on the spatial resolution used in the modeling of the gas component ( figs . 1 , 9a ) .
9a supports the berentzen et al .
( 1998 , 2007 ) analysis that the increasing presence of the gas component makes the bar instability milder . in the _ secular _ stage ,
this sharp decrease in @xmath73@xmath34 with @xmath0 persists , including the lowest resolution sequence as well .
the bar growth is severely restrained in the gas - rich models in this phase .
it either had vanished or even becomes negative .
the underlying physical process has been quantified by berentzen et al .
( 1998 ) : the _ gas affects the orbital structure _ of the collisionless components , which in turn modifies the bar properties
. however detailed understanding of how this influences particular parameters of the bars requires more work .
we have attempted to clarify this dependence of the bar strength on @xmath0 and @xmath32 . as a working hypothesis
, we have tested whether the final mass of the cmc , @xmath61 , in each model can serve as an underlying hidden parameter replacing @xmath0 and @xmath32 .
the dependency of the final bar strength on the cmc mass in the pure stellar disks has been analyzed by athanassoula , lambert & dehnen ( 2005 ) .
extensive study of the bar evolution in the presence of an analytical cmc has been performed by shen & sellwood ( 2004 ) .
bornaud et al .
( 2005 ) have attributed the bar weakening to both the cmc and the angular momentum transfer from the gas to the stellar bar , using analytical cmc and dm halo .
however , the latter process has been found not important by berentzen et al .
( 2007 ) following a detailed analysis of the angular momentum transfer in live potentials of the cmc and dm halo . moreover , berentzen et al .
( 1998 , 2007 ) have used self - consistently growing cmcs in the gaseous / stellar disks to demonstrate that the secular bar growth is strongly affected .
the set of numerical models analyzed here is in fact the most controlled experiment performed so far to test the influence of gas fraction and its resolution on the bar evolution .
10a shows the dependence of the bar strengthening after the first buckling , @xmath81@xmath34 on final @xmath61 , with all other parameters characterizing the stellar disk and the dm halo being fixed .
one model , sd_g8s1 , has been omitted , as its cmc secular evolution is unusually flat and falls out of the sequence with @xmath32@xmath44 . if @xmath61 is the single defining variable which controls @xmath81@xmath34 , all three @xmath32-sequences are expected to merge into a single sequence in fig .
what is actually observed is that the three sequences exhibit a very similar behavior each curve is flat for the gas - poor models , experiences an abrupt drop , and flattens out .
however , the curves do not coincide completely as would be expected if @xmath61 were the _
unique _ underlying parameter .
in fact , in some cases the points corresponding to the same @xmath61 but different @xmath32 have a substantial vertical dispersion in @xmath81@xmath34 .
hence , the issue remains inconclusive .
in contrast , we show the final @xmath61 value normalized by the disk mass at @xmath28 as a function of @xmath0 ( fig .
this fractional @xmath61 value is indeed unique for the three sequences , as all three curves have nearly merged .
we now turn to a plausible correlation between the buckling amplitude and the bar strengthening .
the vertical buckling ( e.g. , toomre 1966 ; combes et al .
1990 ; raha et al .
1991 ) is a recurrent instability ( martinez - valpuesta et al .
2006 ) that was recently analyzed by berentzen et al .
( 2007 ) in the presence of gas ( see also berentzen et al .
the gas component , it has been concluded , leads to a milder instability . here
we have attempted to relate the buckling amplitude , @xmath58 , to the change in the bar amplitude , @xmath73@xmath34 , in the dynamical and secular phases of evolution ( fig .
11 ) . in both phases
we observe a clear correlation between @xmath58 and @xmath73@xmath34 .
stronger bar instability leads to a stronger buckling , and indeed , increasing @xmath0 makes the bar and buckling instabilities milder ( figs .
11a ; see also fig .
9a ) . on the other hand ,
stronger buckling goes in tandem with the bar secular growth ( fig .
this trend shows saturation for the strongest bars .
higher resolution sequences with @xmath32@xmath82 and 0.5 behave in a very similar fashion , while the lowest resolution sequence stands out of this correlation .
we have also checked the value of the drop in @xmath34 during the buckling and find a clear match between the higher resolution sequences , while the lower resolution models behave differently .
the physical extent of the bar depends on its ability to capture additional orbits . while in principle
this capture can proceed at all radii , the fertile region lies between the bar s end and its cr radius , where various families of orbits can be easily destabilized .
therefere , the bar growth due to the orbit capture should go in tandem with the angular momentum _ influx _ because the near - cr orbits will have a larger momentum - to - energy , j / e , ratio than the bar orbits . in paper
i , we have shown , and this is confirmed here , that the influx of angular momentum across the cr and into the bar is able to maintain @xmath66@xmath49const . in time , as long the cr radius lies within the disk , but in the presence of the gas _ this statement is limited to the gas - poor models only_. this happens despite that the cr remains within the disk at all times and for all gas - rich models ( fig . 4 ) .
in other words , in gas - rich models the @xmath3 influx across the cr can not compensate for its loss within the cr . the amplitude @xmath34 is expected to be related to the ability of the bar to capture additional orbits in the bar cr region , leading to its geometrical growth . in both the dynamical and secular phases of the bar evolution we find that the final bar amplitude , @xmath34 , correlates with the final bar size ( fig .
so stronger bars appear to be longer as well .
taken at face value , this property of bar evolution appears to be supported by recent observations in that the @xmath83-band images of barred galaxies exhibit a correlation between the bar amplitude , measured by the @xmath84 fourier component , and its size ( elmegreen et al .
. however , there are caveats .
we find that the bar sizes are substantially smaller in the gas - rich models compared to the gas - poor ones ( figs . 3 , 13 ) .
while the median bar size at the end of the simulation in the latter models appears to be @xmath85 , it is around 0.5 in the former ones , i.e. , 12 kpc vs 5 kpc , embedded in the identical disks and halos .
this difference arises almost entirely during the secular phase of the bar evolution .
more precisely , by the end of the dynamical stage , one can notice that the bar size slightly anti - correlates with @xmath0 for the gas - rich models and only for @xmath32@xmath44 and 0.05 sequences .
this trend is sufficiently weak and bars differ by not more than 10%15% of their length .
so essentially , the bar growth in the dynamical phase is independent of @xmath0 .
what is rather striking , is the abrupt change in the bar s growth ` habit ' in the secular phase bars grow in the gas - poor models and stagnate in the gas - rich ones .
this leads to a _
evolution of the final bar sizes with respect to the gas mass fraction ( fig .
we note that this diverging evolution should be observable only in the long - lived bars which naturally reside in the older disks .
the relevant evolutionary timescale is that of a few gyrs .
the milky way disk is a good candidate , as it is probably about 10 gyr old and was not affected by major mergers over this time period ( e.g. , gilmore , wyse & norris 2002 ) . the anti - correlation between the final bar sizes and disk gas fractions , found here , complicates the simplistic picture of bar evolution .
the reason for this is that our gas - rich models lead to more massive cmcs which increase bulge - to - disk mass ratios and , therefore , should be associated with earlier type disks .
this means that smaller bars in our models lie in early - type disks ( fig .
13 ) , while they are expected to be found in late - type disks , as noted by a number of surveys ( e.g. , erwin 2005 ; see also laine et al .
the simplest resolution of this discrepancy can lie in the omission of star formation processes and especially the feedback from stellar evolution and from the central supermassive black holes in this work .
this leads to a gross over - estimate of the amount of gas which reaches the central kpc and hence contributes to the growth of the cmc by dragging stars and even dm inwards .
we note , however , that the purpose of this numerical exercise was to understand the effect of gas fraction and gas spatial resolution on the disk - bar - halo interactions in the system .
therefore , we have simplified the long list of processes known to affect the galaxy evolution at large . in summary , we find that the spatial resolution in the gas component becomes increasingly important for the bar evolution in the gas - rich disks .
this is true for the dynamical but especially for the secular phase of evolution . in most cases ,
model sequences with @xmath32@xmath82 and 0.05 show a similar behavior , while differing substantially from the sequence with @xmath32@xmath43 .
a bimodal behavior has been found for models based on their gas fractions .
the border line between the gas - poor and gas - rich systems appears to lie around @xmath86 for higher resolution models .
it shifts to @xmath48 for the lowest resolution models .
the switch from a gas - poor to a gas - rich behavior appears to be sufficiently abrupt .
it is clearly visible in all basic characteristics of bar evolution , such as the bar strength , the cmc mass , the bar buckling amplitude , the bar size , etc .
the largest differences in the evolution have been found in the secular phase .
we find that the presence of the gas component severely limits the bar growth and affects its pattern speed evolution .
while pure stellar models ( paper i ) exhibit a rapid slowdown of the bar tumbling , as known for a long time ( e.g. , debattista & sellwood 1998 ; athanassoula 2003 ) , the addition of a substantial amount of gas reverses this trend completely .
furthermore , the cr - to - disk size ratio , @xmath52/@xmath53 , was determined to be an important dynamic discriminator between various phases of barred disk evolution .
here we find that the gas - rich models are characterized by @xmath52/@xmath53 @xmath70 and by @xmath87 const . in these models ,
@xmath88 can even slightly increase with time .
in addition , the gas - rich models maintain @xmath52/@xmath38 @xmath89 which is much more in agreement with observations of stellar bars being fast rotators .
the angular momentum evolution displays the same degree of bimodality with the gas fraction . for low @xmath0 ,
the total disk @xmath3 drops steeply and monotonically with time after the buckling , while it decreases weakly for the gas - rich models .
the reason for this behavior is of course that the bar amplitude is substantially lower in the gas - rich models .
our attempt to explain this difference between the gas - poor and rich models in terms of the more massive cmcs in the latter ones has been rather inconclusive .
we shall return to this issue in the forthcoming work .
next , we have confirmed our previous claim ( paper i ) that the angular momentum , @xmath66 , within the cr radius is maintained at a constant level due to the influx of angular momentum across the cr as it expands .
we have extended this statement to the gas - poor models . for the gas - rich models ,
@xmath66 drops abruptly to a lower level and stays constant thereafter .
a number of corollaries follow from the above results .
we find that the bar strength inversely correlates with the gas fraction , both in the dynamical and secular phases of bar evolution .
the only exception seems to be the lowest resolution sequence in the dynamical phase which fails to capture this trend .
we also find that the buckling amplitude becomes larger for stronger bars prior to the onset of buckling . on the other hand ,
the secular growth is most prominent in bars which show a large buckling amplitude .
finally , we show that stronger bars are also the longest ones throughout both evolutionary phases , and that bar sizes anti - correlate with the gas fraction . | we conduct numerical experiments by evolving gaseous / stellar disks embedded in live dark matter halos aiming at quantifying the effect of gas spatial resolution and gas content on the bar evolution .
three model sequences have been constructed using different resolution , and gas fraction has been varied along each sequence within the range of @xmath0@xmath1 , but keeping the disk and halo properties unchanged .
we find that the spatial resolution becomes important with an increase in the gas content .
for the higher resolution model sequences , we observe a bimodal behavior in the bar evolution with respect to the gas fraction , especially during the secular phase of this evolution .
the switch from the gas - poor to gas - rich behavior is abrupt and depends on the resolution used , being reasonably confined to @xmath0@xmath2 .
the diverging evolution has been observed in nearly all basic parameters characterizing bars , such as the bar strength , central mass concentration , bar vertical buckling amplitude , bar size , etc .
we find that the presence of the gas component severely limits the bar growth and affects its pattern speed evolution .
gas - poor models display rapidly decelerating bars , while gas - rich models exhibit bars with constant or even slowly accelerating tumbling .
we also find that the gas - rich models have bar corotation ( cr ) radii within the disk at all times , in contrast with gas - poor and purely stellar disks .
in addition , the cr - to - bar size ratio is less than 2 for gas rich - models .
next , we have confirmed that the disk angular momentum within the cr remains unchanged in the gas - poor models , as long as the cr stays within the disk , but experiences a sharp drop before leveling off in the gas - rich models .
finally , we discuss a number of observed correlations between various parameters of simulated bars , such as between the bar sizes and the gas fractions , between the bar strength and the buckling amplitude , the bar strength and its size , etc . |
dirichlet s theorem , or the dirichlet prime number theorem , states that for any @xmath3 and @xmath4 such that @xmath5 , there are infinitely many primes in the progression @xmath6 , or equivalently there are infinitely many @xmath7 .
euler was the first to make progress on this theorem , proving it in the case @xmath8 , and it was shown in full generality by dirichlet in 1837 @xcite .
the prime number theorem gives an overall order of growth of the number of primes less than @xmath1 .
letting @xmath0 denote the number of primes in the interval @xmath9 $ ] ( where @xmath1 is not necessarily an integer ) , the prime number theorem asserts that @xmath10 .
this was first conjectured by legendre in 1797 , and was first proven using complex analysis and the zeta function by jacques hadamard and charles jean de la valle - poussin in 1896 .
two `` elementary '' proofs were later discovered by selberg and erds in 1949 @xcite .
the first formal proof of the prime number theorem was written by jeremy avigad et .
al . in 2004
@xcite , in the isabelle proof language , following selberg s proof , a landmark result for mathematics formalization .
hadamard and valle - poussin s proof was formalized in hol light by john harrison in 2009 @xcite .
john harrison also later formalized dirichlet s theorem in hol light , in 2010 @xcite .
metamath is a formal proof language and computer verification software developed for the purpose of formalizing mathematics in a minimalistic foundational theory @xcite . on may 12 , 2016 and june 1 , 2016 respectively , the author formally verified the following theorems in the metamath formal system : [ , dirichlet s theorem][thm : dirith ] @xmath11 [ , the prime number theorem][thm : pnt ] @xmath12 the latter expression is metamath s notation for @xmath13 , restricted to @xmath14 , which is the domain of definition of the function .
these two theorems are interesting formalization targets as they both have simple statements and `` deep '' proofs , and they are also both members of the `` formalizing 100 theorems '' list maintained by freek wiedijk @xcite , which tracks formalizations of 100 of the most famous theorems in mathematics .
both proofs were written concurrently , over the course of about seven weeks between april 7 and june 1 , 2016 .
this was done mostly because both theorems are in the same general subject ( elementary number theory ) and required similar techniques ( mostly asymptotic approximation of finite sums of reals ) . the primary informal text used for the proof was shapiro @xcite , which devotes a section to dirichlet s theorem and the whole final chapter to selberg and erds s proof of the prime number theorem .
the present work is only a broad overview of the problem and proof method .
interested readers are invited to consult the main theorems http://us.metamath.org/mpegif/pnt.html [ ] and http://us.metamath.org/mpegif/dirith.html [ ] at @xcite , where the exact proof is discussed in detail .
the main arithmetic functions used in the formalization are : @xmath15 additionally , the mbius function @xmath16 is a very useful tool in sum manipulations .
it is the unique multiplicative function such that @xmath17 and @xmath18 for @xmath19 .
this yields the mbius inversion formula : if @xmath20 , then @xmath21 . since @xmath22 ,
this is a very powerful technique for estimating sums `` by inversion '' .
the proof of hadamard and valle - poussin relies on some deep theorems in complex analysis , such as cauchy s theorem , which were not available at the time of this formalization , so instead we targeted the `` elementary '' proof discovered half a century later semi - independently by erds and selberg .
the key step in both proofs is the selberg symmetry formula : [ , selberg symmetry formula][thm : selberg ] @xmath23 in selberg s proof , we leverage this theorem to produce a bound on the residual @xmath24 : [ ] [ thm : pntrlog2bnd ] @xmath25 the goal is to show @xmath10 , but it is easily shown that @xmath26 , so it is equivalent to show that @xmath27 , or @xmath28 , to establish the pnt . given an eventual bound @xmath29 and the estimation @xmath30 , an application of reproduces the original estimate @xmath31 , but using improved bounds on @xmath32 on small intervals we can improve the estimate to @xmath33 for a fixed constant @xmath34 , which produces a sequence of eventual bounds approaching zero , which proves @xmath28 as desired . in dirichlet
s theorem , the focal point is instead the dirichlet characters @xmath35 , which are group homomorphisms from @xmath36 to @xmath37 , extended to @xmath38 with value @xmath39 at non - units , but the general theme of estimation of sums involving @xmath40 and the characters @xmath41 is the same .
in keeping with metamath s tradition of minimal complexity , we used a minimum of definition .
asymptotic estimations are reduced to the class @xmath42 of eventually bounded functions , partial functions @xmath43 such that for some @xmath44 , @xmath45 implies @xmath46 .
an equation such as @xmath47 is rewritten as @xmath48 ( which is correct as long as @xmath49 is eventually nonzero , which is always true in cases of interest ) , and similarly @xmath50 is rewritten as @xmath51 .
a few finite summation theorems take us a long way ; two number - theory specific summation theorems are the following divisor sum commutations : @xmath52 @xmath53 a small amount of calculus was used in the proof , mostly through the following sum estimation theorem , which for example evaluates @xmath30 : [ ] [ thm : dvfsumrlim ] if @xmath54 is a differentiable function with @xmath55 , and @xmath56 is a positive decreasing function that converges to zero , then @xmath57 converges to some @xmath58 and @xmath59 .
.comparison of the present proof with @xcite .
`` ? '' marks an estimated or unknown value . [ cols=">,^,^,^,^ " , ] the comparison of parallel proof attempts in different systems is usually confounded by the many other factors , so these statistics should not be given undue credence . according to @xcite ,
avigad s pnt project was a year - long project by four people , with the majority of the work happening during one summer , while this was a solo project over about seven work weeks .
dirichlet s theorem is 10 pages of informal text of @xcite , and the pnt is 37 pages . although the number of lines in the current proofs seem competitive , this is lost in the gzipped version , because the stored metamath proof is already largely compressed , while the isabelle and hol scripts are plain text .
the de bruijn factors for this work had to be estimated because the tex source for the informal text was not available , but indications suggest that it fares poorly with comparatively large factors 19.9 and 7.67 , respectively . however , when reading these statistics it is important to realize that metamath stores _ proofs _ , not _ proof scripts _ like isabelle and hol .
every inference in the proof is an axiom or theorem of the system , and no proof searches are conducted by the verifier .
this is reflected in the incredibly small verification time , which is normal for metamath proofs .
we do not have exact data on verification time for hol light , but it is believed to be on the order of minutes to hours .
these proofs are important milestones for the metamath project .
they demonstrate that even the largest of formalization projects in high level languages can also be conducted in a `` full transparency''-style system like metamath , with entirely worked - out proofs and with all automation offloaded from the verifier to the proof generation .
abc99 dirichlet , p. g. l. : beweis des satzes , dass jede unbegrenzte arithmetische progression , deren erstes glied und differenz ganze zahlen ohne gemeinschaftlichen factor sind , unendlich viele primzahlen enthlt .
berlin * 48 * , 313342 ( 1837 ) selberg , a. : an elementary proof of the prime - number theorem . ann . of math .
( 2 ) , vol .
50 , pp . 305313 ; reprinted in atle selberg collected papers , springer
verlag , berlin heidelberg new york , 1989 * 1 * , 379387 ( 1949 ) | we present the formalization of dirichlet s theorem on the infinitude of primes in arithmetic progressions , and selberg s elementary proof of the prime number theorem , which asserts that the number @xmath0 of primes less than @xmath1 is asymptotic to @xmath2 , within the proof system metamath . |
pentagon chains as main representatives of conjugated polymers which have conducting properties have broad application possibilities covering several fields . starting from the nobel prize awarded to the subject in 2000 , in 15 years
several technological applications have emerged , such as energy storage @xcite , solar cells @xcite , or sustained drug release @xcite , which are now considered classical .
but besides these , several other application possibilities are present as rechargeable batteries , electrochromic displays , information memory , anti - static materials , anti - corrosives , electrocatalysis , sensors , electromechanical devices , infra - red polarizers @xcite , bioanalytical sensors @xcite , or biocontacts @xcite .
in addition to these , there are several other applications related explicitly to the action of external ( magnetic and/or electric ) fields on the conducting polymer of pentagon chain type . in order to exemplify , we mention the applications related to pseudocapacitor electrodes @xcite , fabrication of cell layers creating in vivo electric fields @xcite , radar or microwave absorbing materials which converts the electromagnetic energy into heat @xcite , electrods for battery applications @xcite , artificial muscles based on actuators with large strain and stress induced electrically @xcite , photothermal agents @xcite , covering layers for magnetic nanoparticles @xcite , or transport mediators in liquid electrolites @xcite .
contrary to this technological interest , the action of external electric and/or magnetic fields on pentagon chains has not been analyzed et all on microscopic level .
the present paper tries to fill up at least partially this gap by reporting a detailed study of the action of external magnetic and electric fields on pentagon chains at microscopic level .
we mention that pentagon chains in the absence of external fields have been intensively analyzed in the last period . using technical steps presented in details in ref .
@xcite , the high concentration limit has been described @xcite , and the low concentration limit has been investigated @xcite .
inspite of the fact that these systems are non - integrable , and strongly correlated @xcite where the on - site coulomb repulsion may even be as large as 10 ev @xcite , the method used is special and provides exact results @xcite .
the strategy is to transcribe the hamiltonian in exact terms into a positive semidefinite form @xmath0 where @xmath1 is a positive semidefinite operator while @xmath2 is a scalar , and in a second step to find the exact ground state by constructing the most general solution of the equation @xmath3 . by deducing total particle number ( @xmath4 ) dependent ground states in this manner ,
we can find exact result relating the low lying part of the excitation spectrum as well , e.g. via the n - dependent chemical potential .
the technique itself has been successfully used previously in characterizing different non - integrable quantum many - body systems as periodic anderson model ( pam ) in one @xcite , two @xcite , and three @xcite dimensions , non - integrable chain structures @xcite , two dimensional effects as stripe or droplet formation @xcite , delocalization effect of the interaction @xcite , or even study of the effect of randomness @xcite . using the above presented procedure , in the present paper
we show that placing the pentagon chain in external fields , main physical properties of the system can be substantially changed .
the remaining part of the paper is structured as follows : section ii .
describes the studied systems in external fields , section iii . presents the system hamiltonian , sections iv - v .
present the deduced ground states in two different regions of the parameter space , section vi . containing
the summary closes the presentation , while at the end of the paper two appendices containing mathematical details close the presentation .
the system under consideration is a pentagon chain taken with periodic boundary conditions and containing @xmath5 unit cells as presented in fig.1 .
we consider the chain in an external magnetic field perpendicular to the plain containing the cells . in these conditions
each hopping matrix element @xmath6 will gain a peierls phase factor @xmath7 .
this means that along the bond @xmath8 described at @xmath9 by @xmath10 , the hopping matrix element for @xmath11 case becomes @xmath12 .
the peierls phase factor is given by an integral along the bond @xmath8 of the form [ h ] characterize the geometrical extension of the pentagon cell .
the hopping matrix elements along different bonds are given by @xmath13 , while the arrows on the bonds show the hopping direction for which the peierls phase factor @xmath14 has been calculated .
one notes that @xmath15 is the lattice site where the presented pentagon cell is placed , @xmath16 is the unique bravais vector of the system , and different sites of the unit cell placed at the site @xmath15 are denoted by @xmath17 .
the sublattice index is @xmath18 , while for mathematical convenience @xmath19 .
b ) the on - site one - particle potentials @xmath20 at different sites of the pentagon , @xmath18 .
c ) the system of coordinates xoy used during the calculation of the peierls phase factors @xmath21 .
the direction of the external magnetic field is presented by @xmath22 , @xmath23 shows the direction of the corresponding vector potential @xmath23 , the unique bravais vector direction is presented by @xmath16 , while the direction of the wave vector is shown by @xmath24 .
, width=566,height=377 ] @xmath25 where @xmath26 represents the flux quantum .
we must underline that if one moves in positive trigonometric direction an electron starting from the site @xmath27 , around a closed path c containing the sites @xmath28 , hence turning back to the starting point @xmath27 , one obtains @xmath29 = \frac{2\pi}{\phi_0 } \oint_c { \bf a } d { \bf l } = \frac{2\pi}{\phi_0 } \int_s ( \nabla \times { \bf a } ) d { \bf s } \nonumber\\ & = & \frac{2\pi}{\phi_0 } \int_s { \bf b } d{\bf s } = \frac{2\pi}{\phi_0 } \phi , \label{ue2}\end{aligned}\ ] ] where @xmath30 is the magnetic flux through the surface s enclosed by the path c. in eq.([ue2 ] ) , at the third step the stoke s theorem , at the fourth step the relation between the magnetic induction @xmath22 and the vector potential @xmath23 , namely @xmath31 , and at the fifth step the definition of the magnetic flux @xmath32 has been used . as a consequence of eq.([ue2 ] ) , if one adds the peierls phase factors @xmath33 corresponding to the bond @xmath34 along bonds enclosing an arbitrary surface s with a closed path c , one must reobtain the flux @xmath30 threading the surface s , multiplied by @xmath35 , namely ( see also fig.2 ) [ h ] ) .
, width=151,height=151 ] @xmath36 the eq.([ue3 ] ) can be used for checking the calculations providing the peierls phase factors @xmath33 .
the calculation of the peierls phase factors is performed in details in appendix a. the obtained ( and see ( [ ue11 ] ) , checked ) result is @xmath37 the external electric potential is taken into consideration by producing modifications in the on - site one particle potentials ( @xmath38 ) in a site selective manner .
this means that the same @xmath39 variation appears in the on - site one particle potential in each cell at the same site , i.e. for an arbitrary site @xmath40 , given by the field , @xmath41 appears for each integer @xmath42 , where @xmath16 is the unique bravais vector of the system .
we mention that for the site selective modification of the local on - site one - particle potentials different techniques can be used in practice . in order to exemplify ,
we mention : one dimensional fets ( field effect transistors ) disposed in an array geometry @xcite , directed electric fields applied on short distances @xcite , charge transfer through site - selective interaction @xcite , site - selective conjugation into the polymer @xcite , site selective binding to the polymer @xcite , site - selective application of potential during bipolar electrolysis @xcite , or even site selective introduction of functional groups into the polymer @xcite . at the level of the description , we simply use the notation @xmath38 for the on - site potentials in the hamiltonian , keeping in mind that this value , if needed , can be modified in a site selective manner as mentioned above .
taking into account the peierls phase factors from eq.([ue10 ] ) , and the presence of the site selective on - site one particle potentials , one has @xmath43 , where @xmath44 , \nonumber\\ \hat h_{kin,2}&=&\sum_{\sigma } \sum_{i=1}^{n_c } \big[\epsilon_1 \hat n_{{\bf i}+{\bf r}_1,\sigma } + \epsilon_2 \hat
n_{{\bf i}+{\bf r}_2,\sigma } + \epsilon_3 ( \hat n_{{\bf i}+{\bf r}_3,\sigma } + \hat n_{{\bf i}+{\bf r}_6,\sigma } ) + \epsilon_4 ( \hat n_{{\bf i}+{\bf r}_4,\sigma } + \hat n_{{\bf i}+{\bf r}_5,\sigma } ) \big ] , \nonumber\\ \hat h_{int } & = & \sum_{i=1}^{n_c } \big[u_1 \hat n_{{\bf i}+{\bf r}_1,\uparrow } \hat n_{{\bf i}+{\bf r}_1,\downarrow}+u_2 \hat n_{{\bf i}+{\bf r}_2,\uparrow } \hat n_{{\bf i}+{\bf r}_2,\downarrow}+u_3 ( \hat n_{{\bf i}+{\bf r}_3,\uparrow } \hat n_{{\bf i}+{\bf r}_3,\downarrow}+\hat n_{{\bf i}+{\bf r}_6,\uparrow } \hat n_{{\bf i}+{\bf r}_6,\downarrow } ) \nonumber\\ & + & u_4 ( \hat n_{{\bf i}+{\bf r}_4,\uparrow } \hat n_{{\bf i}+{\bf r}_4,\downarrow}+\hat n_{{\bf i}+{\bf r}_5,\uparrow } \hat n_{{\bf i}+{\bf r}_5,\downarrow } ) \big ] .
\label{ue12}\end{aligned}\ ] ] we append the following observations relating the used hamiltonian presented above .
a ) the studied system is in fact a conducting polymer free of magnetic atoms , hence ( [ ue12 ] ) describes itinerant carriers holding the same charge .
hence the inter - carrier interaction is of coulomb type .
b ) the studied chain is in fact a many - body itinerant system , hence a strong screening of the coulomb interaction is present . because of this reason the longer tails of the coulomb repulsion are neglected , and the inter - carrier interaction in ( [ ue12 ] ) appears as on - site coulomb repulsion , whose strength at the site @xmath40 is given by @xmath45 for an arbitrary @xmath46 , where @xmath42 is an arbitrary integer ( i.e. @xmath45 is fixed for a given type of site in each cell ) .
c ) the carriers created by the @xmath47 canonical fermi operators can be considered in fact fermionic quasiparticles , so their mobility and effective mass must not coincide with the bare electron mobility and effective mass . in this view , ( [ ue12 ] ) can be considered as an effective hamiltonian in which even polaronic concepts are in a given extent incorporated .
further electron - phonon interactions are neglected in the starting hamiltonian .
these neglected interactions become important at system half filling , but away from this carrier concentration ( i.e. electron or hole doped polymer ) their importance is small @xcite . by transforming the @xmath48 expression in @xmath24-space we use :
i ) @xmath49 , where @xmath50 is the sublattice index , ii ) @xmath51 , one obtains @xmath52 . \label{ue13}\end{aligned}\ ] ] now taking into consideration that @xmath24 is directed along the line of the chain ( the line connecting the sites @xmath15 and @xmath53 , hence @xmath24 and @xmath16 are parallel ) , and @xmath54 where @xmath55 , one has ( see fig.1a ) : @xmath56 consequently the @xmath57 from ( [ ue13 ] ) becomes @xmath58 .
\label{ue15}\end{aligned}\ ] ] introducing the @xmath59 column vector @xmath60 composed in order from the components @xmath61 , whose transposed adjoint provides the @xmath62 row vector @xmath63 , the @xmath57 operator can be written as @xmath64 where the @xmath65 matrix @xmath66 has the form @xmath67 the band structure is obtained from the secular equation of @xmath66 , hence @xmath68=0 $ ] , where @xmath69 is the @xmath70 unity matrix , and @xmath71 provides the eigenvalues . introducing the notations @xmath72 the secular equation for @xmath66 gives as eigenvalues the @xmath71 solutions of the following equation : @xmath73 \big]+ \big[e_1e'_4(t_1 ^ 2t_c^2+t^2{t'}^2)+e_1e_4(t_1 ^ 2t_c^2+t^2{t"}^2 ) -e_1e'_3t_1 ^ 2(e_3e'_4-{t"}^2 ) \nonumber\\ & & -e_1e'_3t^2(e_4e'_4-t_n^2 ) -e_1e_3t_1 ^
2(e_4e'_3-{t'}^2)-e_1e_3t^2(e_4e'_4-t_n^2 ) \big ] \nonumber\\ & & + 2e_1t_ct^2(e_4e'_4-t_n^2)\cos(k-2\delta)+2e_1t^2t't"t_n \cos(2\delta+2\delta'+\delta_n)+2e_1t_1 ^ 2t_ct't"\cos[k+(2\delta'+2\delta_1 ) ] \nonumber\\ & & -2e_1t_1t_ct(e'_4t'+e_4t")\cos[k+(\delta_1+\delta'-\delta)]+ + 2e_1t_1t_ct_nt(t'+t")\cos[k+(\delta'+\delta_n-\delta-\delta_1 ) ] \nonumber\\ & & -2e_1t_1t_nt(e_3t'+e'_3t")\cos(\delta+\delta'+\delta_n-\delta_1)+ 2e_1tt_1[t'(e_3e'_4-{t"}^2)+t"(e_4e'_3-{t'}^2)]\cos(\delta+\delta'+\delta_1 ) \nonumber\\ & & + 2e_1t_1 ^ 2t_n(e_3e'_3-t_c^2)\cos(\delta_n-2\delta_1 ) = 0 .
\label{ue19}\end{aligned}\ ] ] from this equation results that at @xmath74 the band structure is obtained from the equation
@xmath73 \big]+ t^2
e_1 \big[e'_4{t'}^2+e_4{t"}^2 -e'_3(e_4e'_4 -t_n^2)-e_3(e_4e'_4-t_n^2 ) \big ] \nonumber\\ & & + 2e_1t^2 \big [ t_c(e_4e'_4-t_n^2)\cos(k-2\delta)+t't"t_n \cos(2\delta+2\delta'+\delta_n)\big]= 0 . \label{ue20}\end{aligned}\ ] ] now if @xmath75 and because of this reason @xmath76 , furthermore @xmath77 and @xmath78 , equation ( [ ue20 ] ) becomes @xmath79 \big ] \nonumber\\ & & + 2t^2 e_1 \big[e_4{t'}^2-e_3(e_4 ^ 2-t_n^2)+ t_c(e^2_4-t_n^2)\cos(k-2\delta)+{t'}^2t_n \cos(2\delta+2\delta'+\delta_n)\big]= 0 .
\label{ue21}\end{aligned}\ ] ] without external magnetic field , eq.([ue21 ] ) reads @xmath80\cos k + ( e_1e_2-t_f^2)[(e_3e_4-{t'}^2)^2-t_c^2(e_4-t_n^2)-t_n^2e_3 ^ 2]+ \nonumber\\ & & 2t^2e_1[{t'}^2(e_4+t_n)-e_3(e_4 ^ 2-t_n^2)]=0 .
\label{ue22}\end{aligned}\ ] ] this last equation , at @xmath81 and @xmath82 gives back the old result known for the pentagon chains @xcite @xmath83 + \nonumber\\ & & 2t^2e_1[e_3(t_n^2-e_4 ^ 2)+t^2(e_4+t_n)]=0 .
\label{ue23}\end{aligned}\ ] ] the study of ( [ ue19 ] ) , or the comparison between e.g. ( [ ue22 ] ) and ( [ ue21 ] ) shows that the external fields have a substantial effect even at the level of the bare band structure modifying the physical properties of the system .
these modifications , besides the effective mass and carrier average velocity changes that they automatically produce , can lead to band flattening effects or modifications of proximities to flat bands ( see for importance e.g. refs .
@xcite ) , or level crossing effects ( qualitative changes produced by this effects are exemplified e.g. by refs .
the modifications in the bare band structure presented in the previous section lead also to modifications of the phase diagram of the system described by the whole hamiltonian .
this means e.g. that in several region of the parameter space new phases appear .
we exemplify this fact in the following two sections for the case of magnetic phases .
the deduction of the many - body ground states will be made in a non - approximated manner following the technique of the hamiltonian transformation in positive semidefinite form presented in the introductory part ( sect.i . ) . for simplicity ,
only the case of the symmetric cell with nearest neighbor hoppings ( @xmath84 , @xmath85 , @xmath86 , @xmath87 ) will be analyzed in details in sections iv - v .
we use 5 block operators for decomposition ( see fig.4 ) for each cell .
one has @xmath88 [ h ] the matching equations providing @xmath89 in the form @xmath90 -p\hat n , \nonumber\\ & & \hat o = \sum_{{\bf i},\sigma } [ \hat a^{\dagger}_{{\bf i},\sigma } \hat a_{{\bf i},\sigma } + \hat b^{\dagger}_{{\bf i},\sigma } \hat b_{{\bf i},\sigma } + \hat d^{\dagger}_{{\bf i},\sigma } \hat d_{{\bf i},\sigma } + \hat f^{\dagger}_{{\bf i},\sigma } \hat f_{{\bf i},\sigma } + \hat g^{\dagger}_{{\bf i},\sigma } \hat g_{{\bf i},\sigma } ] + \hat h_{int } , \ : c= -p\hat n , \label{e2.2}\end{aligned}\ ] ] becomes @xmath91 solving the matching equations , 8 parameters can be directly expressed @xmath92 and the remaining 7 matching equations with unknowns @xmath93 become @xmath94 now the first relation from ( [ e2.5 ] ) gives @xmath95 for the case @xmath74 the equalities ( [ e2.6 ] ) provide @xmath96 } } , \nonumber\\ & & |b_4|^2=\frac{|d_6|^2}{|a_3|^2}t_ne^{-i[\delta_n+2(\delta+\delta ' ) ] } , \label{e2.7}\end{aligned}\ ] ] from where is seen that @xmath97 } \geq 0 $ ] must hold .
this means that at @xmath74 , with @xmath98 as an arbitrary integer , one has @xmath99 based on ( [ e2.5],[e2.7 ] ) a simple solution of the matching equations is obtained .
the starting point in deducing it is the @xmath100 relation which is provided by eqs .
( [ e2.3 ] ) , see appendix b. based on this equality one directly finds @xmath101 . in this case
the last line of ( [ e2.5 ] ) is satisfied and provides @xmath102 .
the @xmath103 equation gives a condition , namely @xmath104 , and the @xmath105 value as @xmath106 .
for @xmath107 one finds from ( [ e2.7 ] ) that @xmath108 .
the equation of @xmath109 determines the @xmath110 value . summarizing the results one has (
note that @xmath74 ) : @xmath111 the solutions ( [ e2.9 ] ) are valid with the conditions @xmath112 by solving the matching equations in the previous subsection , the transformation in ( [ e2.2 ] ) now becomes explicitly known , so we can start the deduction of the ground state .
the deduction procedure is based on constructing the most general @xmath113 which satisfies @xmath114 .
the technical steps followed in the presented case are described in extreme details e.g. in refs .
note that @xmath115 is considered , where @xmath116 is the bare fock vacuum state with no carriers present , and @xmath117 is a condensed index . in this subsection
we are looking for the possibility of @xmath118 solutions in @xmath113 that do not touch each other , i.e. are defined on disconnected geometrical blocks .
this means that @xmath118 and @xmath119 for @xmath120 do not have common operators , hence @xmath121 automatically holds . for a disconnected solution , the operators from the wave vector have to have the form ( see fig.4a ) @xmath122 the condition which determines the unknown coefficients
@xmath123 is @xmath124 which has to be satisfied for all values of all indices .
here @xmath125 , for @xmath126 represents all block operators from ( [ e2.1 ] ) i.e. @xmath1 from ( [ e2.2 ] ) becomes of the form @xmath127 .
we underline that if ( [ e2.11a ] ) is satisfied , it automatically implies @xmath128 based on ( [ e2.9 ] ) , from here one obtains @xmath129x_2=0 , \label{e2.13}\end{aligned}\ ] ] which , taking into account the definition of @xmath130 leads to @xmath131 .
this equality can not be satisfied since @xmath132 , and @xmath133 hold .
so disconnected ground state wave vector for this solution does not exist .
[ h ] together with the coefficients of the sites.,width=529,height=188 ] contrary to the disconnected solution , now we are looking for @xmath118 operators for @xmath113 with the property that the @xmath134 manifold has @xmath120 components which touch each other at least on one site , i.e. two different @xmath118 and @xmath119 for some different @xmath135 and @xmath136 have common creation operators .
the operators of the wave vector in the case of the connected solution ( see fig.4b ) have the expression @xmath137 and the equations for the coefficients , based on ( [ e2.11a ] ) become @xmath138 the system ( [ e2.15 ] ) allows for arbitrary @xmath139 the solution
@xmath140}{[a_2b_4d_5+a_4(d_2b_5-b_2d_5)]}= -\frac{\bar t_ne^{2i\delta'}}{t_ct_ne^{-i\delta_n } } \frac { [ d_2a_4b_5+d_5(a_2b_4-b_2a_4)]}{[a_2b_4d_5+a_4(d_2b_5-b_2d_5)]}. \nonumber\end{aligned}\ ] ] based on the considerations from the previous subsection , the unnormalized ground state wave vector has the expression @xmath141 where the @xmath142 operators are given in ( [ e2.14],[e2.16 ] ) , @xmath5 represents the number of cells in the system taken with periodic boundary conditions , and @xmath4 is the total number of carriers . for @xmath143 , one has @xmath144 , hence the spin projection being fixed , the system is ferromagnetic .
this is because neighboring operators are `` connected '' , hence the connectivity condition is satisfied , and the unique way to obtain @xmath145 is to drop the double occupancy , hence to have @xmath144 .
the obtained ferromagnetic ground state is localized in the thermodynamic limit .
for @xmath146 the ground state is constructed from ferromagnetic clusters whose cluster moment has arbitrary orientation , hence the ground state is paramagnetic . in the thermodynamic limit
this state is also localized .
the presented ground states emerge when the conditions ( [ e2.10 ] ) are satisfied @xcite . as ( [ e2.8 ] ) shows
, these conditions are dependent on the external magnetic field , and through @xmath38 , also dependent on the site selective electric potential .
in particular , for zero external fields , at @xmath147 , ferromagnetic solution does not exist @xcite .
consequently , as ( [ e2.10 ] ) shows , in the presented case the external fields are able to double the parameter space region where the ordered phase emerges .
as explained in details in ref .
@xcite , in a non - integrable case , usually is not possible to obtain the desired solution for the ground state in a single mathematical expression valid for the whole parameter space .
ground state solutions can be written only in a restricted parameter domain which is fixed by the obtained solution of the matching equations . on its turn
, the matching equations are connected to a given decomposition of the hamiltonian in positive semidefinite form .
consequently , other decompositions provide ground state solutions in other regions of the parameter space .
we explicitly use this observation below by presenting a new decomposition of the same starting hamiltonian in positive semidefinite form .
[ h ] we use again 5 block operators for decomposition at @xmath75 for each cell , but now the blocks are constructed as shown in fig.5 .
one has in this case @xmath148 the transformed hamiltonian again becomes of the form @xmath149 -p\hat n , \label{e2.2x}\end{aligned}\ ] ] but now the matching equations are different ( again @xmath150 , @xmath75 hold ) , namely @xmath151 solving the matching equations , 11 parameters can be directly and explicitly expressed @xmath152 and one remains with the following four equations written for the remaining three ( @xmath153 ) unknown variables @xmath154 the first equality of the second line from ( [ e2.5x ] ) gives @xmath155 which , because it is satisfied only at @xmath156 , provides back the conditions relating the external magnetic field from ( [ e2.8 ] ) .
furthermore , ( [ e2.6x ] ) determines the @xmath110 value as well .
the second equality from ( [ e2.5x ] ) provides @xmath157 which becomes a conditions for the solution of the matching equation .
together with ( [ e2.6x],[e2.7x ] ) , the following inequalities must hold in order to have the presented solution of the matching equations @xmath158 the inequalities ( [ e2.8x ] ) define the parameter space region where the second deduced ground state solution is valid .
the used procedure is the same as applied in section iv . for a possible disconnected solution the @xmath159 operator must have again the expression presented in eq.([e2.11 ] ) , hence in the presented case , instead of ( [ e2.12 ] ) one finds @xmath160 where now the block operator parameters are taken from ( [ e2.4x ] ) .
but ( [ e2.9x ] ) allows only the solution @xmath161 , hence disconnected solution does not exist . in this case
the @xmath159 operator has again the form presented in ( [ e2.14 ] ) , while now , because the block operators constructing @xmath162 are different , instead of ( [ e2.15 ] ) one obtains @xmath163 where again the block operator parameters have to be taken from ( [ e2.4x ] ) and the first line of ( [ e2.5x ] ) .
the system of equations ( [ e2.10x ] ) , for an arbitrary @xmath164 presents the non - trivial solution @xmath165 consequently , in the here presented case one has again a connected solution of the form ( [ e2.14 ] ) whose coefficients are provided by ( [ e2.11x ] ) .
the deduced ground state in this section is again of the form ( [ e2.17 ] ) but is present on a different region of the parameter space specified by ( [ e2.6x],[e2.7x],[e2.8x ] ) .
note that now , as explained above , the expression of @xmath166 operators present in ( [ e2.17 ] ) is different , only the qualitative form of the ground state remains the same .
the physical properties of @xmath113 are similar to the physical properties of the ground state deduced in the previous section .
in particular , the ordered phase present in the absence of external fields at @xmath167 , is extended also in the @xmath147 region when external magnetic field is present , which practically double the parameter space domain of the ferromagnetic phase .
this fact has been demonstrated in two different phase diagram regions in the last two sections , hence the extension of the ordered phase created by the presence of external fields is considerably higher .
we investigate pentagon chains described by hubbard type of hamiltonians in external magnetic and electric fields .
the system is in fact a conducting and organic polymer with pentagon cell not containing magnetic atoms at all .
the external magnetic field perpendicular to the plane of the cell is taken into account via peierls phase factors multiplying the hopping terms of the hamiltonian , while the external electric fields by site selective electric potentials modifying the on - site one - particle potentials . in the presented conditions
one shows that the external fields have a substantial effect on the physical properties of the chains .
first , this modifies considerably the bare band structure of the system provided by the one - particle part of the hamiltonian , and second , the emergence domains of condensed phases are also redrawn .
we exemplify this by deducing ferromagnetic ground states in two different regions of the parameter space and show that the parameter space domain where the ordered phase occurs is doubled .
since the studied system is non - integrable and strongly correlated , the deduction process of the multielectronic ground states is non - approximated , and is based on a technique which transforms the hamiltonian in a positive semidefinite form .
z. gulcsi kindly acknowledges financial support provided by the alexander von humboldt foundation , otka - k-100288 ( hungarian research funds for basic research ) and tamop 4.2.2/a-11/1/konv-2012 - 0036 ( co - financed by eu and european social fund ) .
the calculation of the peierls phase factors is performed by using eq.([ue1 ] ) . for the vector potential one uses the landau gauge expression , namely @xmath168 , where @xmath169 , and the used system of coordinates @xmath170
is presented in fig.1c .
note that @xmath171 holds , and @xmath22 is directed along the z - axis perpendicular to the ( x0y ) plane in fig.1c .
the calculations provide the following peierls phase factors : \1 ) along the bond @xmath172 , because the vectors @xmath23 and @xmath173 are orthogonal , one has @xmath174 , hence @xmath175 holds .
\2 ) along the bonds @xmath176 , and @xmath177 , because of @xmath178 , one has @xmath179 , hence @xmath180 .
\3 ) along the bond @xmath181 , @xmath182=constant , consequently @xmath183 please note that @xmath184 , see fig.1a , consequently @xmath185 .
\4 ) along the bond @xmath186 , denoting by @xmath135 the angle between the bond @xmath186 and @xmath187 axis [ @xmath188 , where @xmath189 is the lenght of the bond @xmath186 ] , taking @xmath191 , @xmath192 , as the projection of the vector potential along the bond @xmath186 , and denoting by @xmath193 the @xmath194 variable along the bond @xmath186 , one obtains @xmath195 where one had @xmath196 , @xmath197 , and @xmath198 , see fig.1a .
consequently , one obtains @xmath199 .
\5 ) the same result is obtained along the bond @xmath200 . indeed
, if @xmath201 , the @xmath194 variable along the bond @xmath202 , is measured from the site 6 , taking into account that now @xmath203 , one has @xmath204 this means @xmath205 , consequently @xmath206 .
\6 ) along the bond @xmath207 , denoting by @xmath208 the angle between the bond @xmath207 and 0x axis @xmath209 $ ] , taking the @xmath194 variable along the bond @xmath207 to be @xmath210 as measured from the origin ( site 3 ) , observing that the @xmath23 projection to the bond @xmath207 is @xmath211 , but now @xmath212 because we are placed on the negative region of the 0x axis , hence @xmath213 , one has @xmath214 consequently , taking into account that @xmath215 where @xmath216 , one obtains @xmath217 .
but since @xmath218 , we have as result @xmath219 .
\7 ) along the bond @xmath220 we obtain the result given in @xmath221 .
this is because now the projection of @xmath23 on the bond with the @xmath222 orientation is negative , i.e. @xmath223 , but @xmath224 remains @xmath225 .
consequently @xmath226 this means that @xmath227 .
\8 ) along the bond @xmath228 the peierls factor is obtained as follows .
let introduce the @xmath229 axis connecting the sites 2 and 5 .
this axis intersects the translated x axis as shown in fig .
6 . let us denote by @xmath230 the angle between the x axis and @xmath229 axis .
[ h ] axis along the line connecting the sites 2 and 5 .
this axis is rotated by the angle @xmath230 relative to the x axis.,width=151,height=151 ] one has @xmath231 ( see fig.1a ) , where the distance between the sites 2 and 5 is @xmath232 .
the connection between the coordinate x and the coordinate @xmath229 is @xmath233 , hence @xmath234 holds .
the projection of the vector potential on the @xmath229 axis is @xmath235 , and because @xmath236 , one has @xmath237 .
consequently @xmath238 consequently @xmath239 .
\9 ) along the bond @xmath240 one obtains the same result .
indeed , now the integrals must be taken from @xmath241 to @xmath242 , but @xmath243 since the @xmath229 axis is directed from site 4 to the site 2 , hence the two negative signs compensate each other . consequently @xmath244 . summarizing the obtained results one has @xmath245 now one check the calculations based on ( [ ue3 ] ) .
the total surface of the pentagon @xmath246 is composed from the surface of the upper triangle @xmath247 , and the surface of the lower trapezoid @xmath248 , consequently , one finds @xmath249 $ ] .
furthermore the middle triangle @xmath250 enclosed into the pentagon has the surface @xmath251 , hence the two lateral triangles with the same area @xmath252 and @xmath253 have the surface @xmath254/4 $ ] .
now one can check that as required by eq.([ue3 ] ) one has @xmath255 consequently , the calculated peierls phase factors are correct .
from the subsection b.1 it seems at first view that the condition @xmath100 is artificially imposed , so provides only a particular solution of ( [ e2.5 ] ) .
this impression is not true .
this fact will be demonstrated below .
for this reason one concentrates below on the last equation from ( [ e2.5 ] ) , namely that written for @xmath256 .
indeed , without presumptions , from ( [ e2.6 ] ) one has @xmath257 this introduced in the first equality of @xmath256 gives @xmath258 .
but the same ( [ e2.18 ] ) introduced in the second equality of @xmath256 gives @xmath259 now taking @xmath260 from ( [ e2.19 ] ) and ( [ e2.20 ] ) one obtains ( note that @xmath261 holds ) : @xmath262 this relation introduced in ( [ e2.19 ] ) provides @xmath263 as seem , @xmath264 is not a simplification of the solution , is the only existing solution of ( [ e2.5 ] ) .
a. nan , i. craciunescu and r. turcu , in _ aspects on fundaments and applications of conducting polymers _ ,
159 - 182 , chapt . 8 : conducting polypyrrole shell as a promis ing covering for magnetic nanoparticles , pg .
159 - 182 , chapt . 8 : intech publication , rijeka- shanghai , 2010 .
yoshiki chujo in _ macromolecular design of polymeric materials _ , eds . k. hatada , t. kitayama and o. vogl , marcel dekker inc .
usa 1997 , chapt .
18 : pg.323 - 337 : design of reactive polymers and their applications , pg . 333 | we consider a pentagon chain described by a hubbard type of model considered under periodic boundary conditions .
the system i ) is placed in an external magnetic field perpendicular to the plane of the cells , and ii ) is in a site selective manner under the action of an external electric potential . in these conditions
we show in an exact manner that the physical properties of the system can be qualitatively changed .
the changes cause first strong modifications of the band structure of the system created by the one - particle part of the hamiltonian , and second , produce marked changes of the phase diagram .
we exemplify this by deducing ferromagnetic ground states in the presence of external fields in two different domains of the parameter space . |
this work was funded by nsf dmr 99 - 81869 . | barkhausen noise , including both periodic and aperiodic components , is found in and near the relaxor regime of a familiar relaxor ferroelectric , pbmg@xmath0nb@xmath1o@xmath2 , driven by a periodic electric field .
the temperature dependences of both the amplitude and spectral form show that the size of the coherent dipole moment changes shrink as the relaxor regime is entered , contrary to expectations based on some simple models .
although relaxor ferroelectrics form locally ferroelectric polar nanodomains , underlying chemical disorder somehow prevents the formation of long - range ferroelectric or antiferroelectric order @xcite .
instead , there is a crossover , characterized by faster - than - arrhenius temperature dependence of relaxation rates ( approximately of the vogel - fulcher form , e.g. @xcite ) , to a glassy `` relaxor '' regime .
the key ingredients of a successful model may include static random polar fields ( e.g. @xcite ) , random anisotropy ( e.g. @xcite ) , random interactions among the polar nanodomains ( e.g.@xcite ) , and interactions with soft to phonons @xcite and slow charge - transfer modes @xcite .
nonequilibrium noise and mesoscopic noise provide model - sensitive probes of glassy freezing @xcite , unlike macroscopic linear response and the corresponding equilibrium fluctuations , which show rather generic properties .
the barkhausen effect ( see , e.g. , @xcite ) arises because the polarization changes unevenly as an applied field changes , with the random part of the response only statistically similar between nominally identical samples . for two materials with similar average response functions
, the barkhausen noise will be larger for the material whose polarization makes larger coherent changes .
although simple pictures of interacting polar nanodomains would predict that the dynamically coherent units grow as the frozen regime is entered , a previous attempt to find barkhausen noise in the relaxor regime found none , setting an upper limit on the cooperative changes of the dipole moment @xcite .
large discrete steps in polar order have been found in the best - studied relaxor , pbmg@xmath0nb@xmath1o@xmath2 ( pmn ) , but only after the applied dc electric field was large enough to drive the material into an ordinary ferroelectric regime with macroscopic domains @xcite .
here we report barkhausen noise in the _ nonferroelectric _ ( paraelectric and relaxor ) regimes of pmn , where conventional barkhausen noise from macroscopic domains is absent . the simplest barkhausen model @xcite consists of a set of polarization steps at fixed fields distributed with uniform probability over the field sweep range . for periodically driven fields ,
the noise consists of a periodically repeated random walk , so @xmath3 , the fourier power spectral density of the voltage fluctuations , will consist only of harmonics of the driving frequency @xmath4 , and its spectral envelope will decay as @xmath5 .
if the step pattern changes between cycles ( e.g. , due to thermal jitter in the step times ) then there will be aperiodic noise too .
for fast - relaxing domains which remain in quasi - equilibrium as the field changes , the polarization steps follow the smooth change in boltzmann factor vs. time ( on a coarse time scale ) , causing the @xmath5 envelope to cut off rapidly ( @xmath6 ) above a characteristic frequency @xmath7 @xcite : @xmath8 ( here , @xmath9 is the maximum amplitude of the sinusoidal driving field , @xmath10 is the typical change in electric dipole moment in a single step , and @xmath11 is boltzmann s constant . ) for domains with large barriers to switching , the switches occur as single abrupt steps .
however , thermal jitter in the timing of otherwise reproducible steps also reduces the periodic component above @xmath7 , with a ( numerically computed ) envelope very similar to that for the fast switchers . the missing harmonic power should appear as an aperiodic component of roughly lorentzian shape and corner frequency comparable to @xmath7 .
when this simple independent - step picture applies , the voltage variance ( the integral of @xmath3 ) will grow linearly with @xmath9 ( so long as @xmath12 ) @xcite , being @xcite @xmath13 here @xmath14 is the geometrical capacitance , and we assume the dimensionless dielectric constant @xmath15 . when there is a range of @xmath10 s , @xmath10 in eq .
[ eq : totbkhnoise ] is replaced by @xmath16 . generically , if the non - arrhenius slowing down of relaxation rates arises because interacting nanodomains form clusters which reorient coherently but with barriers which increase with cluster size ( e.g. as in the random - field ising model @xcite ) , then the dipole moments of the coherent clusters ( even if the constituents are randomly aligned ) would grow as @xmath17 was reduced , causing @xmath18 and @xmath7 to rise .
if the activation barrier heights for some clusters become too large for them to respond at or near @xmath4 for a given @xmath9 , then the typical @xmath10 would saturate near the moment of the largest responding clusters .
the size of the largest responding cluster would shrink slightly as @xmath17 is reduced further , but only as @xmath17 to a power less than 1 .
we found neither continued growth nor approximate saturation of the noise magnitude and frequency range , but instead abrupt shrinkage .
our sample consisted of two pieces , each about 0.75 mm thick , from a single crystal of pmn , grown by the chokhralsky method at the rostov - on - don institute of physics .
the sputtered gold contact pads on the main ( 001 ) faces had about 1 mm@xmath19 surface area each . and
@xmath20 at 50 hz , measured at 13 v / cm .
the inset shows the nonlinear out - of - phase response , most apparent at high-@xmath17 where the linear out - of - phase response is tiny.,scaledwidth=40.0% ] standard ac susceptibility measurements at 50 hz showed the usual pmn relaxor behavior , as shown in fig .
[ fig1 ] .
the response was close enough to linear over the field range employed in the noise experiments for the linear susceptibility to suffice for approximate calculations .
previous experiments @xcite have shown that the apparent vogel - fulcher freezing temperature of pmn ( @xmath21 k ) determined by ac response is essentially unchanged over the field range used in these experiments , although it drops sharply at higher fields .
g is the gain of the driving op - amps .
data were taken on the 14-bit converter in sets of 30,000 points with a 10 khz sampling rate , giving 1/3 hz frequency bins in the fourier spectrum .
usually 16 such spectra were averaged .
the anti - alias filter corner frequency was 5 khz , although it should have been 4 khz.,scaledwidth=40.0% ] for the noise measurements , the two halves of the sample were incorporated in a balanced - bridge circuit , as shown in fig .
[ fig2 ] .
this setup allowed us to apply an ac bias up to 10 volts on each half of the sample without overloading the low - noise amplifiers , despite the generation of harmonics by the sample nonlinearity , slow capacitance changes due to systematic aging effects @xcite , and imperfect common - mode rejection .
however , any residual systematic differences in the nonlinearity of the two parts would generate harmonics .
an experiment in which one sample arm was replaced with a polystyrene capacitor indicated significant systematic harmonics out to the fifth harmonic .
we avoided using harmonics lower than the tenth in the data analysis , since systematic nonlinearities are distinct from barkhausen noise .
the ac source has very low levels of noise and distortion , and the analog - to - digital sampling is synchronized well with the ac drive . with @xmath22
v / cm at @xmath23 hz at 244 k ( spectrum 2 ) is shown above a background ( spectrum 1 ) consisting of equilibrium noise from the dissipative sample plus amplifier noise , mainly from current noise from the stanford 550 preamp .
60 hz and its harmonics have been edited out.,scaledwidth=40.0% ] fig .
[ fig3 ] shows a typical @xmath3 with , roughly as expected , a mixture of periodic and broad aperiodic noise .
the wings around @xmath4 ( and to a lesser extent around low harmonics ) were not anticipated , but these can arise from small drifts in the bridge cancellation as the sample ages .
the size of the polarization events , calculated below from @xmath3 , was far too small for them to be discerned individually . for ( a ) aperiodic and ( b ) periodic components.,scaledwidth=40.0% ] since the spectral form is close to the independent - step expectation , we can extract a typical step size @xmath10 , although we doubt that the polarization changes literally consist of simple discrete steps . as shown in fig .
[ fig4 ] , both periodic and aperiodic components of the noise grow nonlinearly with @xmath9 over the range explored .
thus the @xmath10 calculated at some @xmath9 gives only a typical dipole step size under the particular drive conditions .
the form of the nonlinearity suggests a broad distribution of @xmath10 s , with many small @xmath10 s showing up in the noise only as @xmath9 is increased enough to make @xmath24 for that @xmath10 .
the form of the aperiodic spectrum indicates a similar distribution . although we are not yet confident in extracting absolute step sizes from this new technique ( mainly due to the difficulty of removing all artifacts from the lower harmonics , which contain the most power ) , we can use eq .
[ eq : totbkhnoise ] to calculate a lower - bound estimate for a typical @xmath10 by integrating the broad aperiodic spectrum only .
( we assume this @xmath10 reflects about how big the cooperative changes are regardless of whether they are simple discrete steps . ) at @xmath25 k , just above the relaxor regime , at the largest @xmath9 used , the result is about @xmath26 c cm , which apparently would grow at higher @xmath9 . under the same conditions ,
the envelope of the harmonic noise has a @xmath5 tail extending from about 1 khz to 4 khz , the top of our measured range , indicating that there are some discrete dipole switches as large as @xmath27 c cm .
an order - of - magnitude estimate of the moment of a polar nanodomain at 250 k , based on the range of the static polar correlations @xcite and the saturation polarization @xcite , would be about @xmath28 c cm .
thus it seems that even in the _ paraelectric _ regime there are some dynamically coherent units larger than single polar nanodomains . despite the uncertainties in the absolute calibration , we can obtain a good indication of how typical @xmath10 s depend on @xmath17 , a key issue for relaxor models . for independent steps , the quantity @xmath29 is proportional to the actual @xmath10 s under particular experimental conditions .
fig .
[ fig5 ] shows how some periodic and aperiodic components of @xmath30 depend on @xmath17 at fixed @xmath9 .
we pick spectral components in an octave around @xmath31 , because in this range the signal - to - background is still good , we expect few artifacts in the harmonics , and the @xmath17-dependence of @xmath10 shows up in @xmath3 both via the overall magnitude and via the @xmath17-dependence of @xmath7 .
( other frequency components show the same sort of behavior , just less clearly . ) , are shown vs. @xmath17 at two different @xmath9.,scaledwidth=40.0% ] as seen in fig .
[ fig5 ] , as the relaxor regime is approached from high @xmath17 , the spectral components of @xmath32 either rise slowly , as one would guess from the growth of medium - range static correlations @xcite , or show little change .
however , below 235 k to 250 k , both periodic and aperiodic components drop sharply , in sharp contrast to the continued growth of static correlations .
the factor @xmath33 accounts for less than half of this drop .
the width of the envelope of the harmonic components is also @xmath17-dependent , being largest ( for @xmath34 v / cm ) at 250 k. in contrast to the long @xmath5 tail at 250 k , below 250 k the harmonics fell sharply into the anharmonic background , e.g. becoming undetectable above about 900 hz at 210 k , confirming that the largest dipole moment steps occur near 250 k. in summary , our noise results indicate abrupt shrinkage of the typical net dipole moment of the dynamically coherent changes ( at fixed driving field ) as the relaxor regime is entered .
we do not want to burden this experimental result with much theory , but some initial comments should be helpful .
we know of no proposed model in which the coherent dipole moment differences between the _ equilibrium _ states at different fields would shrink sharply in the relaxor regime ( i.e. , relaxors are not believed to be frustrated antiferroelectrics ) .
rather , the shrinkage of the steps found on a fixed frequency scale should be related to the vogel - fulcher kinetics @xcite , i.e. , to the rapid growth of barrier heights as the relaxor regime is approached .
although our results do not dictate a model for the relaxor transition in pmn , we can rule out some models . a model of the dynamics with a fixed set of relaxation modes with a fixed broad unimodal distribution of activation barriers ( presumably monotonically increasing in @xmath10 ) @xcite , would predict that the characteristic @xmath10 is a weakly increasing function of @xmath17 at all @xmath17 , and hence is inconsistent with our results . for
simple growth of independent polar nanodomains , the dipoles grow along with the anisotropy barriers , so the typical dipole moment for which the field ( plus thermal activation ) allows switches at rates of order @xmath7 would continue to grow .
for formation of coherently flipping clusters , the typical moment also grows with the number of nanodomains ( probably as a square root ) . although the moment to barrier ratio would be lower for larger clusters , the moment of the largest clusters with rapid enough driven kinetics would approximately saturate . unless there were a strange distribution of cluster sizes , so that as the typical size grew , the mean of those remaining below some fixed cutoff would shrink , such models would not reproduce our results . thus the barkhausen results strongly suggest that the rapid growth of barriers is not just the result of nanodomain growth or of collective nanodomain clusters forming . instead , it seems that each nanodomain s contribution to the barriers grows much more rapidly than its contribution to the dipole moment , as in a recent phenomenological model @xcite , suggesting that less - polar degrees of freedom are freezing along with the polar nanodomain orientations .
another indication of some such effects is that the low - temperature linear heat capacity specifically associated with the relaxor @xcite gives a dimensionless entropy of about @xmath35 per nanodomain .
models of relaxor freezing invoking strong coupling of nanodomain polarization to multiple slow local degrees of freedom ( particularly charge transfers ) and to soft to phonon modes @xcite have been proposed to account for clusters with multiple metastable polarization states having a roughly uniform distribution in solid angle @xcite .
such models would allow complex aging @xcite even in small clusters of nanodomains .
whether or not such models prove successful , the barkhausen data provide a new qualitative constraint on models for the relaxor transition , which has thus far provided a target for somewhat under - constrained theory . |
in 1984 cummings and stell @xcite proposed a simplified hamiltonian model for a reactive system of two types of homogeneus fluids with the same density and diameter . in this model , type @xmath6 and @xmath7 molecules , can be associated by means of a selective square interaction that , in the appropiate limits , can be reduced to a sticky baxter s potential @xcite located inside the hard core at a distance from the center of the particle @xmath2 .
the cases of @xmath8 for the baxter s function have been solved in @xcite using the same formalism and , at @xcite , they apply the results for a pair of reactive fluids and a solvent .
lee and rasaiah solved the @xmath9 , @xmath10 and proposed a solution for @xmath11 for chemical association and dipolar dumbbells @xcite .
the model of lee and rasaiah of association in electrolytes @xmath12 is studied in @xcite . in this case
the authors add a selective coulombian part , @xmath13 , out of core , to the associating original csm .
the sticky site inside the hard core incorporates geometrical conditions of steric saturation in the molecule and this idea is shown , using computational simulations in different ensembles , in ref .
for different bonding length parameters the system allows formation of dimers for small @xmath5 , chains for @xmath5 slightly larger , and vulcanization of species for bonding length values close to the diameter @xmath14 of particles .
huerta and naumis studied the connectivity of a binary mixture using a selectively hard sphere potential and a superposition as @xcite : @xmath15 where @xmath16 @xmath17 and @xmath18 where @xmath5 is the bonding distance , @xmath0 the intracore square well width and @xmath19 , @xmath20 represents the species in the mixture .
the final potential ( see figure [ potential ] ) is equivalent to the original cummings - stell in the adecuate limits for sticky approximation . the same idea is implemented by pizio and blum @xcite for a hard - sphere fluid with dimerization @xmath21 . in the development
most of the models maintain @xmath5 as a parameter ( bonding distance ) and finally take the case @xmath22 , however some other possibilities are presents having the analytical solution for arbitrary @xmath5 .
kalyuzhnyi and stell @xcite present a recount of cases for different ranges of the location @xmath5 .
as we show here , the sticky potential into the hard sphere shell produces a discontinuity step in the auxiliary functions of baxter .
this fact allows us to think of systems with more than one sticky site inside the shell or , even , a distribution of sticky wells .
the vertical dashed arrows show the sense of the baxter s sticky limit @xcite .
the molecular diameter is @xmath14 and the dashed curves ( with labels `` aa '' and `` ab '' ) corresponds to the colulombian interaction for electrolyte @xcite . in the following sections , we first develop the matrix of differential and shift operators mdso , and its inverse , for the simplest case @xmath23 , followed for the inversion of the mdso s for a very general case that corresponds to the set of @xmath4 dde s , to @xmath4 subintervals of [ 0 , @xmath14 ] and to a sticky location @xmath25 .
the statistical mechanical model of chemical reactions of cummings and stell @xcite represents the association of two species @xmath26 , with the same density and diameter , which simplifies the mathematical problem .
the potential proposed in the csm consists of a hard sphere repulsion between like species ( @xmath6-@xmath6 or @xmath7-@xmath7 ) and a mound of width @xmath14 with a deep , narrow , and attractive square well , with width @xmath0 centred on @xmath5 . here
@xmath27 and @xmath28 for @xmath29 interactions : @xmath30 the geometric consideration of this model for the @xmath29 interactions ensures steric saturation in the system ( there is no formation of @xmath4-mers for @xmath31 ) due to overlapping .
in addition , this model has a solution in the py approximation , mapping the square well onto an infinitely deep and stretch well like the sticky potential of baxter @xcite .
the connection between the baxter s original and this model , is obtained equating the second virial coefficients : first it is considered the limit @xmath32 , @xmath33 , @xmath34 .
the limits of @xmath33 and @xmath34 are taken to maintain tractable the problem in the py approximation @xcite .
the limit @xmath35 ( in the repulsive part : see figure [ potential ] ) does nt change very much the results , but simplifies the solution @xcite .
the total and direct correlation functions are related by the ornstein - zernike ( oz ) equation that , for this binary mixture , can be written as @xcite @xmath36 where the integral is evaluated in the whole space , @xmath37 , and @xmath38 is the number density of species @xmath39 particles . in the csm
the densities are considered equal . briefly reviewing the formulation of the cs model :
we need the factorized form of the oz @xcite equations ( [ ozij ] ) with the py closures @xmath40 and , in the considered limits @xmath41 given the conditions of the problem , the factorized oz equations are written @xcite @xmath42\\ & & \\
r h_{12 } ( r ) & = & - q_{12 } ' ( r ) + 2 \pi \rho \int^{\sigma}_0 d t ( r - t ) \left [ q_{11 } ( t ) h_{12 } ( |r - t| ) + q_{12 } ( t ) h_{11 } ( |r - t| ) \right ] .
\end{array } \label{ozfac}\ ] ] where we changed the index of @xmath43 or @xmath44 to @xmath45 or @xmath46 respectively .
the same for all functions . substituting the closure relations ( [ py1 ] ) and ( [ py2 ] ) in the set of oz equations ( [ ozfac ] ) ,
the following system of difference - differential equations ( dde ) for the auxiliary baxter s functions @xmath47 is obtained : @xmath48 & = & ( a_{11 } + d a_{12 } ) r + b_{11 } + d b_{12}\\ & & \label{qijs}\\ q_{12 } ' ( r ) + p [ q_{11 } ( r + l ) - q_{11 } ( r - l ) ] & = & ( d a_{11 } + a_{12 } ) r + d b_{11 } + b_{12 } - \frac{\lambda l^2}{12 } \delta ( r - l ) , \end{array}\ ] ] where @xmath49 and @xmath50 and satisfies the boundary conditions @xmath51 after integration , a step appears in @xmath52 for the auxiliary function @xmath53 due to the delta term associated with the well . in the first work of cummings and stell @xcite a new pair of functions
were used and defined as the sum and difference of the originals @xmath54 and @xmath53 .
the advantage of this trick is to obtain two uncoupled equations , one for @xmath55 and another for @xmath56 which can be solved in a separate way . if the functions @xmath57 and @xmath58 are defined then , adding and substracting eqs .
( [ qijs ] ) , @xmath59 = a_+ r + b_+ - \frac{\lambda l^2}{12 } \delta ( r - l ) \label{qmas}\ ] ] and @xmath60 = a_- r + b_- + \frac{\lambda l^2}{12 } \delta ( r - l ) , \label{qmenos}\ ] ] with the obvious definitions , so that @xmath61
. this not implies changes in the results .
we asume this fact in the rest of paper . ]
@xmath62\ ] ] and @xmath63 the last term in ( [ qmas ] ) and ( [ qmenos ] ) can be omitted since it is equal to zero in all subintervals except where @xmath64 .
this condition is fixed in the boundary conditions ( [ condiciones ] ) . in the rest of subintervals , for the general case , must be true that @xcite @xmath65 we use here @xmath66 .
the aim of this proposal is to find an analytical form for the function @xmath67 assuming that : i ) the solution must be made in subintervals @xcite , ii ) this implies that @xmath67 will be defined also in subintervals , and iii ) the original functions @xmath47 can be recovered : @xmath68 and @xmath69 .
identical procedure shows that is sufficient to replace @xmath70 to obtain @xmath56 .
the cases shown in @xcite are solved here using mdso . for convenience
we show the case @xmath22 of the cs model @xcite in detail , and the cases @xmath71 and @xmath9 summarized .
the first case yields the system of coupled differential equations @xmath72 and @xmath73 where , evidently , @xmath74 corresponds to the first half of the interval and @xmath75 to the second .
we define here the differential operator @xmath76 as @xmath77 and the shift operator @xmath78 by @xmath79 . with this
operators defined , the set of ( [ mdsoa1 ] ) and ( [ mdsob1 ] ) can be rewritten as @xmath80 and @xmath81 or , in matricial form , as @xmath82 these equations can be reduced to a symbolic form @xmath83 where @xmath84 is the matrix of differential and shift operators , or mdso , that appears in ( [ mdsocero ] ) , applied to the vector @xmath85 of functions @xmath86 . the right side is the vector @xmath87 of functions @xmath88 that , in this case , are linear functions of @xmath89 .
the index in @xmath90 , corresponds to the number of equations ( or partitions in the interval of solution ) .
the main idea of this paper is to find a solution for the system represented in ( [ formalset ] ) as @xmath91 this implies the knowledge of an explicit analytical form of the inverse of @xmath84 , and how it operates on @xmath92 .
one way of defining the inverse of the differential operator @xmath76 is by using the equation @xmath93 or @xmath94 whose solution leads us to define the inverse operator @xmath95 as @xmath96 in the previous expression , the case @xmath97 implies that the inverse mdso is reduced to the trivial definition of inverse differential operator as an integral operator .
the case where @xmath98 is a complex number ( or a pure imaginary one ) implies harmonic solutions @xcite and fourier transform of the right hand side of differential equation .
continuing with the case @xmath22 , the inverse of @xmath84 is @xmath99 where the commutation properties of the operators @xmath76 and @xmath100 were used .
direct calculation gives the determinant - operator of @xmath101 as . ]
@xmath102 so that ( [ m2inverse ] ) and ( [ detinverse ] ) define completely the inverse determinant - operator of @xmath84 as the product of two inverse operators of the form of ( [ defineop ] ) : @xmath103 this is the formal inverse determinant of the mdso , however we still need to find the appropiate coefficients to satisfy the boundary conditions .
so that , the direct application of the inverse mdso , ( [ m2inverse ] ) , on ( [ mdsoa1 ] ) and ( [ mdsob1 ] ) gives @xcite @xmath104 with @xmath105 .
now , considering ( [ condiciones ] ) and the fact of ( [ mdsob1 ] ) must be satisfied we obtain , explicitly , @xmath106 which agree exactly with the results in @xcite . the second equation , now has the same set of constants that the first .
the harmonic functions have been interchanged and their arguments are shifted by @xmath107 . in the case @xmath71
there are three equations out of @xmath109 $ ] . ] : @xmath110 @xmath111 @xmath112 with the mdso given as @xmath113 and the inverse operator of @xmath114 @xmath115 where @xmath116 and . ]
@xmath117 obviously @xmath118 is a product of inverse operators in the form of ( [ defineop ] ) .
applying these to the right hand side of ( [ case3 ] ) we obtain , directly @xmath119 @xmath120 @xmath121 and , imposing bound conditions in the respective subintervals and the fact of ( [ case3 ] ) must be satisfied ( as in ( [ q1q2ab ] ) for @xmath23 ) , we obtain @xmath122 and @xmath123 in terms of @xmath124 and @xmath125 @xmath126
@xmath127 @xmath128 the case @xmath9 has a tridiagonal matrix @xmath129 , whose determinant @xmath130 has roots @xmath131 and @xmath132 so that @xmath133 \left [ \mathcal{d}- i \frac{\sqrt{5 } - 1}{2 } p \right ] \left [ \mathcal{d}+ i \frac{\sqrt{5 } + 1}{2 } p \right ] \left
[ \mathcal{d}- i \frac{\sqrt{5 } + 1}{2 } p \right ] .\ ] ] lee and rasaiah , in @xcite , called these roots @xmath134 and @xmath135 , and correspond to the @xmath136 s defined below in this paper . the case @xmath137 or @xmath10 has a determinant @xmath138 whose roots are @xmath139 , and the solutions have the same structure as ( [ q1q2ab ] ) and ( [ sol3 ] ) , and that showed in @xcite .
making the same construction for @xmath4 divisions in the solution interval , one obtains @xmath4 functions @xmath140 that represent a continuous solution . each of them must be such that @xmath141 = a r + b , { \ensuremath{\operatorname{for } } } \frac{i - 1}{n } \sigma < r < \frac{i}{n } \sigma\ ] ] where @xmath142 and @xmath143 , in which @xmath144 the generalized mdso for arbitrary @xmath4 has the form tridiagonal @xmath145 or , in matrix form @xmath146 due to symmetry of @xmath147 the shift operators are mutually canceled in the inverse of @xmath148 .
this fact enables us to put all solutions @xmath149 in terms of inverse differential operators of the form ( [ defineop ] ) and ( [ deltainverse ] ) . from the tridiagonal matrix
obtained , ( [ thematrix ] ) , the determinants for different values of @xmath4 can be evaluated : @xmath150 it is easy to prove that the recurrence relation between determinants of mdso s of different order is , for @xmath151 , @xmath152 where @xmath153 and @xmath154 .
the index of @xmath148 corresponds to the order of the mdso . from this recurrence relation
we obtain the general expression for @xmath148 @xmath155 which , finally , can be reduced to @xmath156 where , additionally @xmath157 in these expressions @xmath158 is the integral part of @xmath159 . in ( [ deltafac ] )
we have written @xmath148 in factorized polynomial form . here
@xmath160 are all the @xmath4 roots of @xmath148 which are all pure imaginary and proportional to @xmath161 . when @xmath4 is even there are exactly @xmath162 pairs of complex conjugated roots . ]
@xmath163 and , if @xmath4 is odd , there is a further null root of @xmath148 , which requires an additional integration to obtain the @xmath140 functions .
we can write @xmath148 , with @xmath164 as defined above , as @xmath165 where we define @xmath166 and @xmath167 .
also , @xmath168 is the complex conjugate of @xmath160 , @xmath136 is a real number , and @xmath169 is the operator associated with the root @xmath170 for @xmath4 odd
. direct application of individual inverse @xmath171 on a linear function gives @xcite @xmath172 same as before , @xmath173 is defined by mean of @xmath174 and @xmath175 . in the general case , a partition @xmath4 even of the interval [ @xmath176 results in @xmath177 it is easy to see that if we define the vectorial funcion @xmath178 , using the definition in ( [ m2inverse ] ) or ( [ mgorro ] ) , as the application of @xmath179 to the vector @xmath92 , @xmath180 we obtain a linear function @xmath178 .
the general solution of the system of equations become @xmath181 where @xmath182 and @xmath86 is as defined in ( [ qider ] ) .
the constants @xmath183 and @xmath184 that appears in ( [ d2minverse ] ) can be stablished from bound conditions at the frontiers of the subintervals given by ( [ condiciones ] ) and ( [ diez ] ) and the recurrence stablished in the original set of equations .
a general solution of the set of dde s for the original cs model has been discussed here .
it requires the roots of the polinomial expresion for the inverse determinant @xmath185 which always are of the form @xmath186 with @xmath173 a real number .
one of the advantages of this method is the fact that it allows to choose the site of the sticky potential , not only at @xmath187 for a few values of @xmath4 : the solution is valid for any @xmath4 .
it allows to change the sticky potential site to positions more and more close to ( or away of ) the center of the particle .
for example in the case @xmath188 the step of discontinuity @xmath189 can be defined at @xmath190 or at @xmath191 .
this fact bring us the possibility to set the well at @xmath192 to obtain different molecular structures @xcite : by locating the step @xmath189 in the site of sticky and represent a step discontinuity in the solution at the position @xmath25 . with this in mind
one can think , also , the posibility of two or more sticky square wells into the hard shell . + * acknowledgements * p.t . cummings and g. stell
statistical mechanical models of chemical reactions ii .
analitic solution of the percus - yevick approximation for a model of homogeneus association .
_ , 55(1):3348 , 1985 .
a. huerta . a two - dimensional model associating fluid with spherically integral equations and monte carlo simulation study symmetric intracore square well shell . _ mol .
_ , 96(5):795804 , march 1999 .
orest pizio and lesser blum .
analitic solution of the mean spherical approximation for a dipolar hard - sphere fluid with intracore anisotropic sticky interactions .
52(1):572579 , july 1995 .
guadalupe m. mungu@xmath194a gmez and mart@xmath194n g. garc@xmath194a alvarado .
ecuaciones diferenciales con retardo .
_ memorias de la xvii semana regional de nivel superior investigacin y docencia en matemticas .
mosaicos matemticos .
_ , ( 20 ) , august 2007 . | in a series of publications the cummings - stell model ( csm ) , for a binary mixture of associative fluids with steric effects , has been solved analytically using the percus - yevick approximation ( pya ) .
the solution consists in a square well potential of width @xmath0 , whose center is placed into the hard sphere shell ( @xmath1 ) : at @xmath2 ( @xmath3 ) .
this paper presents a general solution , for any @xmath4 , of the first order difference differential equation ( dde ) , for the auxiliary baxter s function that appears in the csm , using recursive properties of these auxiliary functions and a matrix composed by differential and shift operators ( mdso ) .
this problem is common in some other models of associative fluids such as the csm for homogeneus and inhomogeneus mixtures of sticky shielded hard spheres including solvent effects under pya , and in that of mean - spherical approximation ( msa ) , for chemical ion association and dipolar dumbbells and polymers .
the sticky potential implies a discontinuity step at @xmath5 in the solution of auxiliary baxter s functions so that , one side , @xmath5 now is arbitrary and , for some additional effects , it can be placed one or more sticky potentials at different positions into the hard shell .
pacs : 02.30.ks , 61.20.qg |
supersymmetry is at present one of the most favored ideas for physics beyond the standard model ( sm ) @xcite .
a particularly attractive implementation of the idea of supersymmetry is the minimal supersymmetric standard model ( mssm ) obtained by introducing the supersymmetric partners of the of the sm states , and introducing an additional higgs doublet , with opposite hypercharge to that of the sm higgs doublet , in order to cancel the gauge anomalies and generate masses for all the fermions of the standard model @xcite .
supersymmetry must obviously be a broken symmetry . in order for broken supersymmetry to be effective in protecting the weak scale against large radiative corrections ,
the supersymmetric partners of the sm particles should have masses of the order of a few hundred gev .
their discovery is one of the main goals of present and future accelerators . in particular , a @xmath4 linear collider with a high luminosity @xmath5 fb@xmath6 , and a center - of - mass energy of @xmath7 gev in the first stage ,
will be an important tool in determining the parameters of the low energy supersymmetric model with a high precision @xcite .
furthermore , polarization of the electron ( and positron ) beam can enhance the capability of such a linear collider @xcite in unravelling the structure of the underlying supersymmetric model . in the minimal supersymmetric standard model
the fermionic partners of the two higgs doublets ( @xmath8 ) mix with the fermionic partners of the gauge bosons to produce four neutralino states @xmath9 , @xmath10 , and two chargino states @xmath11 , @xmath12 in the mssm the lightest neutralino is favored to be the lightest supersymmetric particle , and assuming , @xmath13-parity ( @xmath14 ) conservation , is absolutely stable . the neutralino states of the minimal supersymmetric standard model with @xmath14 conservation have been studied in great detail , because the lightest neutralino , being the lightest supersymmetric particle ( lsp ) , is the end product of any process involving supersymmetric particle in the final state . there are alternatives to the mssm , an elegant one being the model with an additional chiral electroweak gauge singlet higgs superfield @xmath15 which couples to the two higgs doublet superfields @xmath16 and @xmath17 via a dimensionless trilinear term @xmath18 in the superpotential .
this model can solve the the @xmath19 problem of the mssm in a natural manner . when the scalar component of the singlet superfield @xmath15 obtains a vacuum expectation value , a bilinear term @xmath20 involving the two higgs doublets is naturally generated .
furthermore , when this scalar component of the chiral singlet superfield @xmath15 acquires a vacuum expectation value of the order of the @xmath21 breaking scale , it gives rise to an effective value of @xmath19 ( @xmath22 ) of the order of the electroweak scale .
however , the inclusion of the singlet superfield leads to additional trilinear superpotential coupling @xmath23 in the model , the so called nonminimal , or next - to - minimal @xcite , supersymmetric standard model ( nmssm ) .
the absence of @xmath24 term , and the absence of tadpole and mass couplings , @xmath15 and @xmath25 in the nmssm is made natural by postulating a suitable discrete symmetry @xcite .
the nmssm is attractive on account of the simple resolution it offers to the @xmath19 problem , and of the scale invariance of its classical action in the supersymmetric limit @xcite . since no dimensional supersymmetric parameters are present in the superpotential of nmssm , it is the simplest supersymmetric extension of the standard model in which the electroweak scale originates from the supersymmetry breaking scale only .
its enlarged higgs sector may help in relaxing the fine - tuning and little hierarchy problems of the mssm , thereby opening new perspectives for the higgs boson searches at high energy colliders @xcite , and for dark matter searches @xcite .
in the nonminimal supersymmetric standard model the mixing of fermionic partners of higgs and gauge bosons @xcite produces five neutralino states @xmath9 , @xmath26 , and two chargino states @xmath11 , @xmath12 furthermore , because of the presence of the fermionic partner of the singlet higgs boson , the neutralino states can have an admixture of this @xmath21 singlet fermion , thereby affecting the phenomenology of the neutralinos in the nonminimal supersymmetric standard model .
the lightest neutralino state ( @xmath27 ) of mssm or nmssm , being typically the lsp , is stable and therefore , a possible dark matter candidate @xcite .
since the neutralinos are among the lightest particles in low energy supersymmetric models , they are expected to be the first states to be produced at the colliding beam experiments . at an electron - positron collider , such as the international linear collider ( ilc )
, the lightest neutralino can be produced in pairs @xmath28 this process proceeds via @xmath29 boson and selectron exchange @xcite . in collider experiments the lightest neutralino escapes detection .
in such a situation the production of the lightest neutralino pair ( [ pairneuts ] ) is invisible .
therefore , we must look for the signature of neutralinos in the radiative process @xmath30 despite this process being suppressed by the square of the electromagnetic coupling , it might be the first process where the lightest supersymmetric states could be observed at the @xmath31 colliders .
the signal of the radiative process ( [ radiative ] ) is a single high energy photon with the missing energy carried away by the neutralinos .
the process ( [ radiative ] ) has been studied in detail in the minimal supersymmetric standard model @xcite .
some of these studies underline the importance of longitudinal @xcite , and even transverse beam polarizations .
on the other hand , the signature `` photon plus missing energy , '' that arises in the process ( [ radiative ] ) has been studied in detail by different lep collaborations @xcite .
furthermore , the radiative neutrino process @xmath2 in the sm is the leading process with this signature , for which the cross section depends on the number @xmath32 of light neutrino species @xcite .
the lep collaborations have found no deviations from the sm prediction , and , therefore , only bounds on the masses of supersymmetric particles have been set @xcite . for a review of the experimental situation , see ref .
@xcite .
most of the theoretical studies on radiative neutralino production in the literature have been carried out in the framework of the minimal supersymmetric standard model .
this includes calculations relevant to ilc with a high center - of - mass energy , high luminosity and longitudinally polarized beams , as well as study of the sm background from the radiative neutrino production @xmath33 and the supersymmetric background from radiative sneutrino production @xmath34 the discovery potential of ilc may be significantly enhanced @xcite if both beams are polarized , particularly if other susy states like heavier neutralino , chargino or even slepton pairs are too heavy to be produced at the first stage of the ilc at @xmath7 gev . in a previous paper
( referred to as paper i ) we have carried out a detailed study of the radiative process ( [ radiative ] ) in the nonminimal supersymmetric model and compared the predictions with those of the minimal supersymmetric standard model @xcite .
this study was carried out for unpolarized electron and positron beams . in this paper
we continue this study and consider the radiative process ( [ radiative ] ) in the nonminimal supersymmetric standard model with polarized electron and positron beams to understand in detail if the signal can be enhanced by the use of polarized beams .
furthermore , the sm background photons from radiative neutrino production process ( [ radiativenu ] ) with beam polarizations will have to be taken into account for a proper analysis of the radiative neutralino production process ( [ radiative ] ) .
beam polarizations could enhance the signal photons for the process ( [ radiative ] ) for nmssm and reduce those from the sm background at the same time , which could lead to the enhancement of the statistics .
we will also consider supersymmetric background photons from radiative sneutrino production process ( [ radiativesnu ] ) with polarized beams .
this is important if sneutrino production is kinematically accessible and if the sneutrino decay is invisible
. we will compare and contrast the results obtained for nmssm with those for the minimal supersymmetric standard model with polarized beams .
this will include the signal for the radiative neutralino process , and the dependence of the cross sections on the parameters of the neutralino sector .
this comparison will allow us to assess the feasibility of observing the radiative neutralino process for the most popular low energy supersymmetric models at a @xmath31 collider .
the plan of the paper is as follows . in sec .
[ sec : signal ] , we calculate the cross section for the signal process ( [ radiative ] ) in the nonminimal supersymmetric standard model , and compare it with the corresponding cross section in the minimal supersymmtric standard model , for unpolarized and polarized electron and positron beams . in order to calculate the cross section in the nmssm , we fix the parameter space that we use in our calculations .
this is done by using various theoretical and experimental constraints on the parameter space of nmssm . in particular , we constrain the values of the trilinear superpotential parameters @xmath35 and @xmath36 which enter the neutralino mass matrix of the nmssm .
we also describe the phase space for the signal process as well as the cuts on outgoing photon angle and energy that we use to regularize the infrared and collinear divergences in the tree level cross section .
we then analyze numerically the dependence of the cross section on the parameters of the neutralino sector , and on the selectron masses , for unpolarized and polarized beams . here
we also calculate the photon energy distribution for the radiative production of the second lightest neutralino in the nmssm and compare it with the corresponding distribution for the lightest neutralino for unpolarized and polarized beams , respectively . in sec .
[ sec : backgrounds ] we analyze the backgrounds to the radiative neutralino process ( [ radiative ] ) with polarized beams .
this includes the background from sm process ( [ radiativenu ] ) , as well as from the supersymmetric process ( [ radiativesnu ] ) for the case of polarized beams . in sec .
[ sec : pol ] we study in detail the beam polarization dependence of the cross section for the radiative neutralino production process as well as for the backgrounds processes .
here we also consider the statistical significance for measuring the excess of photons from radiative neutralino production over the backgrounds , and calculate this quantity for nmssm , and compare it with the corresponding results in mssm .
we summarize our results and conclusions in sec .
[ sec : conclusions ] .
our notations and results on the neutralino mass matrices and couplings are summarized in appendix [ appendix : superpotential and couplings ] .
the feynman diagrams for the radiative process @xmath37 are shown in fig .
[ fig : radneutralino ] , where the symbols in the brackets denote the four momenta of the respective particles . in nmssm , and in mssm , this process proceeds at the tree level via @xmath38- and @xmath39-channel exchange of right and left selectrons @xmath40 , and via @xmath29 boson exchange in the @xmath41 channel . in order to calculate the cross section for the radiative production of neutralinos
we need to compute the couplings of the neutralinos to electrons , to the scalar partners of electrons , the selectrons , and to @xmath42 bosons .
we summarize these couplings for mssm and nmssm @xcite in appendix [ appendix : superpotential and couplings ] .
= 1.0pt .,title="fig : " ] as can be seen in appendix [ appendix : superpotential and couplings ] , the couplings of the lightest neutralino are determined by the corresponding elements of the neutralino mixing matrix ( @xmath43 or @xmath44 ) . for numerical calculation of the radiative neutralino cross section in the mssm
, we have chosen to work with the parameters in the snowmass points and slopes ( sps 1a ) scenario @xcite .
the parameters of the sps 1a scenario are summarized in table [ parmssm ] . however , since in the sps 1a scenario the value of the parameters @xmath19 and @xmath45 are fixed , we shall use a different set of parameters to study the dependence of the neutralino mass and the radiative neutralino production cross section on @xmath19 and @xmath46 and on the selectron masses .
this set of parameters is shown in table [ parmssmewsb ] .
we shall call this set of parameters as the mssm electroweak symmetry breaking scenario ( ewsb ) @xcite . as in paper
i , for the nmssm we use a set of parameters that is obtained by imposing theoretical and experimental constraints on the parameter space of the nmssm .
the parameters that enter the neutralino mass matrix of the nmssm are , apart from @xmath47 and @xmath45 , @xmath48 and @xmath36 . for @xmath49 and @xmath50 we use the values which are consistent with the usual gut relation @xmath51 we note that for the mssm in sps 1a scenario , the value of the parameter @xmath52 . in order to remain close to the sps 1a scenario of mssm ,
we have chosen for our numerical calculations in nmssm values of @xmath53 whereas the rest of the parameters are chosen in such a way that the lightest higgs boson mass , the lightest neutralino mass and the lightest chargino mass satisfy the present experimental lower limits .
we have also imposed on the parameter space of nmssm , the theoretical constraint that there is no charge and color breaking global minimum of the scalar potential , and that a landau pole does not develop below the grand unified scale ( @xmath54 gev ) . the consequence of imposing these constraints on the parameter space of nmssm , and the resulting masses for various particles for a particular choice of input parameters is summarized in table [ parnmssm ] .
.input parameters and resulting masses for various states in the mssm sps 1a scenario . [ cols="^,^,^,^,^ " , ]
the nonminimal supersymmetric standard model ( nmssm ) solves @xmath19 problem of mssm in an elegant manner and is , thus , an attractive alternative to the mssm .
we have carried out a detailed study of the radiative neutralino production process @xmath55 in nmssm for ilc energies , and compared the results with the corresponding results in the mssm for both unpolarized and polarized @xmath56 and @xmath57 beams .
this process has a signature of a high energy photon and missing energy .
we have obtained a typical set of parameter values for the nmssm by imposing theoretical and experimental constraints on the parameter space of nmssm .
for the set of parameter values that we obtain in this manner , the lightest neutralino in nmssm has a significant admixture of the fermionic component of the singlet chiral superfield @xmath15 . using this parameter set ,
we have studied in detail the radiative neutralino production cross section in nmssm for the ilc energies for both unpolarized and polarized @xmath56 and @xmath57 beams . for comparison with mssm
, we have used the mssm sps 1a and mssm ewsb models . the background to
this process comes from the sm process @xmath2 , as well as the background from the supersymmetric process @xmath3 .
all these processes have a signature of a highly energetic photon with missing energy .
the photon energy distribution @xmath58 , and the total cross section as a function of the total energy have been calculated for the nmssm and for mssm sps 1a scenario at @xmath59 gev using calchep package . because of the admixture of a singlet in the lightest neutralino , the cross section as a function of energy for the radiative neutralino production is lower in nmssm than in mssm .
we have also studied the dependence of the cross section for radiative neutralino production on the @xmath60 gaugino mass parameter @xmath45 and the higgs(ino ) mass parameter @xmath19 , as well as its dependence on the selectron ( @xmath61 ) masses in nmssm , and compared it with the corresponding results in mssm . in order to quantify whether an excess of signal photons , @xmath62 , can be measured over the background photons , @xmath63 , from radiative neutrino production ,
we have analyzed the theoretical statistical significance @xmath64 . at the ilc , electron and positron beam polarizations can be used to significantly enhance the signal and suppress the background simultaneously .
we have shown that the significance can then be increased almost by an order of magnitude , e.g. with @xmath65 compared to @xmath66 .
a possible feed - back of ilc@xmath67 results could motivate the immediate use of the low - luminosity option of the ilc at @xmath68 gev in order to resolve model ambiguities between nmssm and mssm even at an early stage of the experiment and outline future search strategies at the upgraded ilc at 1 tev . in our scenarios ,
the signal cross section for @xmath69 is larger than @xmath70 , the significance @xmath71 , and the signal to background ratio is about @xmath72 .
the background cross section can be reduced to @xmath73 . increasing the positron beam polarization to @xmath74 , both the signal cross section and the significance increase significantly .
thus the electron and positron beam polarization at the ilc are essential tools to observe radiative neutralino production .
p. n. p. would like to thank the institute of mathematical sciences , chennai for hospitality where was this work was initiated .
the work of p. n. p. is supported by the council of scientific and industrial research , india , and by the j. c. bose national fellowship of the department of science and technology , india .
p. n. p. would like to thank the inter university centre for astronomy and astrophysics for hospitality while this work was completed .
for completeness we summarize here the couplings of the neutralinos to electrons and the scalar partners of electrons , the selectrons , in mssm and in nmssm .
these couplings can be obtained from the neutralino mixing matrix . to obtain the neutralino mixing matrix for the mssm
, we recall that the neutralino mass matrix obtains contributions from part of the mssm superpotential @xmath75 where @xmath16 and @xmath17 are the two higgs doublet chiral superfields , and @xmath19 is the supersymmetric higgs(ino ) parameter . in addition to the contribution from the superpotential , the neutralino mass matrix receives contributions from the interactions between gauge and matter multiplets , as well as contributions from the soft supersymmetry breaking masses for the gauginos . including all these contributions , the neutralino mass matrix , in the bino , wino
, higgsino basis @xmath76 can be written as @xcite @xmath77 where @xmath47 and @xmath45 are the @xmath78 and the @xmath60 soft supersymmetry breaking gaugino mass parameters , respectively , and @xmath79 is the ratio of the vacuum expectation values of the neutral components of the two higgs doublet fields @xmath16 and @xmath17 , respectively .
furthermore , @xmath80 is the @xmath29 boson mass , and @xmath81 is the weak mixing angle .
we shall assume that all the parameters in the matrix @xmath82 are real , in which case @xmath82 is a real symmetric matrix and can be diagonalised by an orthogonal matrix .
if one of the eigenvalues of @xmath82 is negative , one can diagonalize this matrix using a unitary matrix @xmath83 , the neutralino mixing matrix , to get a positive diagonal matrix @xcite : @xmath84 where @xmath85 are neutralino masses arranged in order of increasing value . for the nmssm ,
the relevant part of the superpotential is @xmath86 where @xmath15 is the higgs singlet chiral superfield . in the basis @xmath87 , the neutralino mass matrix for the nmssm
can then be written as @xcite @xmath88 where @xmath89 is the vacuum expectation value of the singlet higgs field .
as in the case of mssm , we can use a unitary matrix @xmath90 to get a positive semidefinite diagonal matrix with the neutralino masses @xmath91 @xcite : @xmath92 the lagrangian for the interaction of neutralinos , electrons , selectrons and @xmath29 bosons for mssm is given by @xcite @xmath93 { f}_e + \mathrm{h . c. } , \label{mssmlagrangian}\end{aligned}\ ] ] with the electron , selectron , neutralino and @xmath29 boson fields denoted by @xmath94 , @xmath95 , @xmath96 , and @xmath97 , respectively , and @xmath98 .
the corresponding interaction lagrangian for nmssm is obtained from ( [ mssmlagrangian ] ) by replacing @xmath43 with @xmath99 the different vertices following from ( [ mssmlagrangian ] ) are shown in table [ feynmandiag ] .
the couplings of the lightest neutralino to electrons , selectrons and @xmath29 boson are determined by the corresponding elements of the neutralino mixing matrix ( @xmath43 or @xmath44 ) .
lccccccccl + vertex & & & & & & & & & vertex factor + & & & & & & & & & + & & & & & & & & & + right selectron - electron - neutralino & & & & & & & & & @xmath100 + & & & & & & & & & + left selectron - electron - neutralino & & & & & & & & & @xmath101 + & & & & & & & & & + neutralino - @xmath42 - neutralino & & & & & & & & & @xmath102 + & & & & & & & & & + electron - @xmath42 - electron & & & & & & & & & @xmath103 $ ] + & & & & & & & & & + selectron - photon - selectron & & & & & & & & & @xmath104 + & & & & & & & & & + electron - photon - electron & & & & & & & & & @xmath105 + & & & & & & & & & + + 99 # 1#2#3 nucl .
* b#1 * , # 3 ( # 2 ) # 1#2#3 nucl . phys . *
a#1 * , # 3 ( # 2 ) # 1#2#3 phys . lett . *
b#1 * , # 3 ( # 2 ) # 1#2#3 phys . rep . *
# 1 * , # 3 ( # 2 ) # 1#2#3 phys . rev .
* d#1 * , # 3 ( # 2 ) # 1#2#3 phys .
# 1 * , # 3 ( # 2 ) # 1#2#3 z. phys . * c#1 * , # 3 ( # 2 ) # 1#2#3 jhep * c#1 * , # 3 ( # 2 ) # 1#2#3 int . j. mod .
a#1 * , # 3 ( # 2 ) # 1#2#3 j. math . phys . * # 1 * , # 3 ( # 2 ) g. t hooft , in `` recent developments in gauge theories .
proceedings , nato advanced study institute , cargese , france , august 26 - september 8 , 1979 , '' _ new york , usa : plenum ( 1980 ) 438 p. ( nato advanced study institutes series : series b , physics , 59 ) _ ; e. witten , nucl .
b * 188 * , 513 ( 1981 ) ; r. k. kaul , phys .
b * 109 * , 19 ( 1982 ) ; r. k. kaul and p. majumdar , nucl . phys .
b * 199 * , 36 ( 1982 ) ; r. k. kaul , pramana * 19 * , 183 ( 1982 ) . h. p. nilles , phys .
rept . * 110 * , 1 ( 1984 ) .
m. drees , r. godbole and p. roy , `` theory and phenomenology of sparticles : an account of four - dimensional n=1 supersymmetry in high energy physics , '' _ hackensack , usa : world scientific ( 2004 ) 555 p. _ j. a. aguilar - saavedra _ et al .
_ [ ecfa / desy lc physics working group ] , `` tesla technical design report part iii : physics at an e+e- linear collider , '' arxiv : hep - ph/0106315 . t. abe _ et al .
_ [ american linear collider working group ] , `` linear collider physics resource book for snowmass 2001 . 1 : introduction , '' in _ proc . of the aps / dpf / dpb summer study on the future of particle physics ( snowmass 2001 ) _ ed .
n. graf , arxiv : hep - ex/0106055 . k. abe _ et al . _
[ acfa linear collider working group ] , `` particle physics experiments at jlc , '' arxiv : hep - ph/0109166 .
g. weiglein _ et al .
_ [ lhc / lc study group ] , `` physics interplay of the lhc and the ilc , '' arxiv : hep - ph/0410364 .
j. a. aguilar - saavedra _ et al .
_ , eur .
j. c * 46 * , 43 ( 2006 ) [ arxiv : hep - ph/0511344 ] .
p. fayet , nucl .
b * 90 * , 104 ( 1975 ) ; h. p. nilles , m. srednicki and d. wyler , phys .
b * 120 * , 346 ( 1983 ) ; j. p. derendinger and c. a. savoy , nucl .
b * 237 * , 307 ( 1984 ) .
j. r. ellis , j. f. gunion , h. e. haber , l. roszkowski and f. zwirner , phys .
d * 39 * , 844 ( 1989 ) .
m. drees , int .
j. mod .
phys . a * 4 * , 3635 ( 1989 ) .
n. pandita , phys .
b * 318 * , 338 ( 1993 ) ; p.
n. pandita , z. phys .
c * 59 * , 575 ( 1993 ) ; u. ellwanger , phys .
b * 303 * , 271 ( 1993 ) [ arxiv : hep - ph/9302224 ] ; u. ellwanger , m. rausch de traubenberg and c. a. savoy , phys . lett .
b * 315 * , 331 ( 1993 ) [ arxiv : hep - ph/9307322 ] ; s. f. king and p. l. white , phys .
d * 53 * , 4049 ( 1996 ) [ arxiv : hep - ph/9508346 ] . b. ananthanarayan and p. n. pandita , int .
j. mod .
a * 12 * , 2321 ( 1997 ) [ arxiv : hep - ph/9601372 ] .
b. ananthanarayan and p. n. pandita , phys .
b * 371 * , 245 ( 1996 ) [ arxiv : hep - ph/9511415 ] .
b. ananthanarayan and p. n. pandita , phys .
b * 353 * , 70 ( 1995 ) [ arxiv : hep - ph/9503323 ] .
m. chemtob and p. n. pandita , phys .
d * 73 * , 055012 ( 2006 ) [ arxiv : hep - ph/0601159 ] .
m. chemtob and p. n. pandita , arxiv:0708.1284 [ hep - ph ] . for a review and references , see e. g. u. ellwanger , c. hugonie and a. m. teixeira , phys .
rept . * 496 * , 1 ( 2010 ) [ arxiv:0910.1785 [ hep - ph ] ] .
u. ellwanger , j. f. gunion and c. hugonie , jhep * 0507 * , 041 ( 2005 ) [ arxiv : hep - ph/0503203 ] .
g. a. moortgat - pick , s. hesselbach , f. franke and h. fraas , _ in the proceedings of 2005 international linear collider workshop ( lcws 2005 ) , stanford , california , 18 - 22 mar 2005 , pp 0206 _ [ arxiv : hep - ph/0508313 ] .
j. f. gunion , d. hooper and b. mcelrath , phys .
d * 73 * , 015011 ( 2006 ) [ arxiv : hep - ph/0509024 ] .
n. pandita , phys . rev .
d * 50 * , 571 ( 1994 ) .
n. pandita , z. phys .
c * 63 * ( 1994 ) 659 .
s. y. choi , d. j. miller and p. m. zerwas , nucl .
b * 711 * , 83 ( 2005 ) [ arxiv : hep - ph/0407209 ] .
h. goldberg , phys .
lett . *
50 * , 1419 ( 1983 ) .
j. r. ellis , j. s. hagelin , d. v. nanopoulos , k. a. olive and m. srednicki , nucl .
b * 238 * , 453 ( 1984 ) .
a. bartl , h. fraas and w. majerotto , nucl .
b * 278 * , 1 ( 1986 ) .
j. r. ellis , j. m. frere , j. s. hagelin , g. l. kane and s. t. petcov , phys .
b * 132 * , 436 ( 1983 ) ; e. reya , phys .
b * 133 * , 245 ( 1983 ) ; p. chiappetta , j. soffer , p. taxil , f. m. renard and p. sorba , nucl .
b * 262 * , 495 ( 1985 ) , [ erratum - ibid .
b * 279 * , 824 ( 1987 ) ] .
fayet , phys .
b * 117 * , 460 ( 1982 ) .
j. r. ellis and j. s. hagelin , phys .
b * 122 * , 303 ( 1983 ) .
k. grassie and p. n. pandita , phys .
d * 30 * , 22 ( 1984 ) .
t. kobayashi and m. kuroda , phys .
b * 139 * , 208 ( 1984 ) .
j. d. ware and m. e. machacek , phys .
b * 142 * , 300 ( 1984 ) .
l. bento , j. c. romao and a. barroso , phys .
d * 33 * , 1488 ( 1986 ) .
m. chen , c. dionisi , m. martinez and x. tata , phys .
* 159 * , 201 ( 1988 ) .
t. kon , prog .
phys . * 79 * , 1006 ( 1988 ) .
a. heister _ et al .
_ [ aleph collaboration ] , eur .
j. c * 28 * , 1 ( 2003 ) .
j. abdallah _ et al .
_ [ delphi collaboration ] , eur .
j. c * 38 * , 395 ( 2005 ) [ arxiv : hep - ex/0406019 ] .
p. achard _ et al .
_ [ l3 collaboration ] , phys .
b * 587 * , 16 ( 2004 ) [ arxiv : hep - ex/0402002 ] .
b. c. allanach _ et al .
_ , in _ proc . of the aps / dpf / dpb summer study on the future of particle physics ( snowmass 2001 ) _ ed . n. graf , eur .
j. c*25 * , 113 ( 2002 ) [ econf * c010630 * , p125 ( 2001 ) ] [ arxiv : hep - ph/0202233 ] .
a. pukhov , arxiv : hep - ph/0412191 .
f. a. berends , g. j. h. burgers , c. mana , m. martinez and w. l. van neerven , nucl .
b * 301 * , 583 ( 1988 ) .
f. boudjema _ et al .
_ , arxiv : hep - ph/9601224 . g. montagna , m. moretti , o. nicrosini and f. piccinini , nucl .
b * 541 * , 31 ( 1999 ) [ arxiv : hep - ph/9807465 ] .
h. e. haber and d. wyler , nucl .
b * 323 * , 267 ( 1989 ) .
s. ambrosanio and b. mele , phys .
d * 53 * , 2541 ( 1996 ) [ arxiv : hep - ph/9508237 ] .
s. ambrosanio and b. mele , phys .
d * 55 * , 1399 ( 1997 ) [ erratum - ibid .
d * 56 * , 3157 ( 1997 ) ] [ arxiv : hep - ph/9609212 ] .
h. baer and t. krupovnickas , jhep * 0209 * , 038 ( 2002 ) [ arxiv : hep - ph/0208277 ] . j. ( . ) .
brau _ et al . _
[ ilc collaboration ] , `` ilc reference design report : ilc global design effort and world wide study , '' arxiv:0712.1950 [ physics.acc-ph ] .
h. k. dreiner , o. kittel and u. langenfeld , eur .
j. c * 54 * , 277 ( 2008 ) [ arxiv : hep - ph/0703009 ] . | we study the production of the lightest neutralinos in the radiative process @xmath0 in low energy supersymmetric models for the international linear collider energies with longitudinally polarized electron and positron beams .
for this purpose we consider the case of nonminimal supersymmetric standard model as well as the case of minimal supersymmetric standard model . at the first stage of a linear collider , with @xmath1 gev
, the radiative production of the lightest neutralinos may be a viable channel to study supersymmetric partners of the standard model particles , especially if the other supersymmetric particles are too heavy to to be pair - produced .
we consider in detail the effect of beam polarization on the production cross section . we compare and contrast the dependence of the signal cross section on the parameters of the neutralino sector of the nonminimal and minimal supersymmetric standard model when
the electron and positron beams are longitudinally polarized . in order to assess the feasibility of experimentally observing the radiative neutralino production process
, we consider the background to this process coming from the standard model process @xmath2 with longitudinally polarized electron and positron beams .
we also consider the supersymmetric background to the radiative neutralino production process coming from the radiative production of the scalar partners of the neutrinos ( sneutrinos ) @xmath3 , with longitudinally polarized beams
. this process can be a a background to the radiative neutralino production when the sneutrinos decay invisibly . |
the study of connections between random matrix theory and properties of the riemann zeta function , @xmath5 , has known recently significant developments @xcite .
a central point is hugh montgomery s ( generalized ) conjecture @xcite that in the asymptotic limit ( high on the critical line for @xmath6 ) the fluctuation properties of non - trivial riemann zeros are the same as for the circular unitary ensemble ( @xmath7 ) of @xmath8 random unitary matrices ( with haar measure ) in the limit of large dimensionality , @xmath9 .
in particular , the normalized pair correlation function of the riemann zeros with @xmath10 is conjectured to be @xmath11 where @xmath12 is the unfolded distance between zeros ( i.e. the mean spacing is set to one ) .
andrew odlyzko , since the late 70 s , started accurate and extensive numerical computations of riemann zeros in order to check this and other conjectures @xcite .
his main result is that in the limit @xmath10 correlation functions of riemann zeros do agree with random matrix predictions .
for instance , in fig .
[ fig1](a ) the density distribution @xmath13 of spacings among consecutive zeros of @xmath5 ( called the nearest
neighbour spacing distribution in random matrix literature ) is plotted for a billion zeros around the @xmath14-th zero .
the agreement with the cue asymptotic prediction @xmath15 ( represented by a solid line in that figure ) is remarkable .
it is interesting to look at the difference @xmath16 between the computed and the conjectured distributions ( see fig .
[ fig1](b ) ) . though the difference is small ( of order @xmath17 ) it has a clear structure with a nontrivial @xmath12 dependence .
one may wonder how this difference compares with the one obtained , within random matrix theory , between the asymptotic and a finite @xmath18 calculation .
to proceed , one needs a criterion for the size of the matrix to compare with .
the simplest assumption is to choose @xmath19 , where @xmath20 this matrix size is obtained by equating the local density of zeros at height @xmath2 along the critical line to the density of eigenvalues of the @xmath21 unitary matrix @xcite .
= 6.7 in = 2.5 in [ fig1 ] this dimensional correspondence has been successful when comparing statistical properties of @xmath22 at finite @xmath2 with those of characteristic polynomials of @xmath7 matrices of size @xmath19 @xcite .
the riemann zeros in fig .
[ fig1 ] are located in a window around @xmath23 , which gives @xmath24 .
the difference between the finite @xmath18 ( with @xmath19 ) and the asymptotic nearest neighbour spacing distributions is represented by a dashed line in fig .
[ fig1](b ) .
though the functional form of the correction is qualitatively correct , its amplitude is clearly too small ( by a factor of order 20 ) . in his paper @xcite ,
a. odlyzko commented : clearly there is structure in the difference graph , and the challenge is to understand where it comes from. the purpose of this note is to provide some elements in this direction .
let us first analyze the corrections to the asymptotic two point function ( [ r2gue ] ) for @xmath21 random matrices from the circular unitary ensemble @xmath25 and for zeros of the riemann zeta function .
the comparison between them will provide an effective set of rules or correspondences between both sequences .
we start first with random matrix theory . for @xmath18-dimensional unitary matrices
the @xmath26point correlation functions are given by @xcite @xmath27 @xmath28 is the corresponding kernel , @xmath29 where @xmath30 and @xmath31 in particular , the unfolded two point correlation function takes the form @xmath32 the expansion ( [ kern ] ) leads to @xmath33 this formula expresses the correlation function as the asymptotic result eq .
( [ r2gue ] ) plus corrections proportional to inverse even powers of the matrix dimension .
notice that the corresponding expansion for random hermitian matrices ( gaussian unitary ensemble ) is different , it is non analytic and includes a term proportional to @xmath34 ( cf @xcite ) .
we now need an equivalent result for the riemann zeros .
an heuristic formula for the two - point correlation function for these zeros was obtained by bogomolny and keating in ref.@xcite using the hardy - littlewood conjecture for the distribution of prime pairs ( for more details see @xcite ) .
it states that the two - point correlation function of riemann zeros , @xmath35 , is the sum of three terms @xmath36 where the smooth density of zeros , @xmath37 , is asymptotically @xmath38 and the diagonal , @xmath39 , and off - diagonal , @xmath40 , parts are given by the following convergent expressions @xmath41\ ] ] and @xmath42 + { \rm c. c. } \ .\ ] ] here the summation and the product are taken over all primes @xmath43 . in @xcite it was checked that these formulas agree very well with odlyzko s results for the two - point correlation function of riemann zeros .
the unfolded two point correlation function is obtained by measuring distances between zeros in units of the local mean spacing , @xmath44 we are interested in the corrections to the asymptotic behavior of @xmath45 in the limit when @xmath1 . in this limit @xmath46 and the argument of @xmath47 in eq.([r2u ] ) becomes small ( keeping @xmath12 finite ) .
therefore , one can expand @xmath35 for @xmath48 . to perform the expansion
it is convenient to use the known series @xmath49 where @xmath50 are the stieljes constants , and the following auxiliary expansions : @xmath51 where @xmath52 and @xmath53 = 1 + c_0 \epsilon^2 + { { \rm i}}q \ \epsilon^3 + { \cal o } ( \epsilon^4)\ ] ] with @xmath54 collecting the different terms in the expansion , one gets @xmath55 and @xmath56 { { \rm e}}^{{{\rm i}}2 \pi { \overline{\rho}}\epsilon } + { \rm c. c.}\ .\ ] ] the unfolded two point correlation function takes therefore the form @xmath57 equation ( [ r2r1 ] ) expresses the two point correlation function of riemann zeros as the asymptotic random matrix result given by eq .
( [ r2gue ] ) , plus corrections that are proportional to inverse powers of the average density of zeros .
the comparison of eqs.([r2cue ] ) and ( [ r2r1 ] ) shows that the leading terms coincide , as conjectured by montgomery . to relate sub - leading terms we proceed as follows .
up to @xmath58 in eq .
( [ r2r1 ] ) , the term of @xmath59 can be absorbed in the term of @xmath60 by rescaling the @xmath12 variable .
equation ( [ r2r1 ] ) now takes the form @xmath61 where @xmath62 @xmath63 .
the comparison of eqs .
( [ r2cue ] ) and ( [ r2r2 ] ) leads to the following conclusions : * to leading order in @xmath64 the two point correlation function of the riemann zeros coincides with that of eigenvalues of random @xmath7 matrices of effective dimension @xmath65 , where @xmath66 where @xmath67 . *
the next - to - leading order is obtained by rescaling the variable @xmath12 in the first correction term according to @xmath68 where @xmath69 is given by eq.([alpha ] ) .
the effective matrix size obtained from our analysis is , therefore , different from @xmath70 ( eq . ( [ n0 ] ) ) , the relation between them being a multiplicative factor @xmath71 .
in general , to compute the nearest neighbour spacing distribution it is necessary to know correlation functions of arbitrary number of points .
though for the riemann zeros the hardy - littlewood conjecture is sufficient to obtain all correlation functions @xcite , the computations are cumbersome and for simplicity we shall use another method . for cue , the nearest
neighbour spacing distribution may be expressed as @xcite @xmath72 where @xmath73 \ , 1 \leq j , k \leq n \ .\ ] ] some physical arguments @xcite indicate that to leading order , under quite general conditions deviations from standard random matrix theory in systems with no time reversal symmetry are reduced to a change of the kernel only , and that the correction term is the same as for @xmath7 matrices with an effective matrix size .
= 6.6 in = 2.4 in [ figg2 ] if we accept this conjecture then from the results of the previous section it follows that to leading order _ all _ correlation functions of the riemann zeros are the same as those of @xmath7 matrices with effective dimension given by eq .
( [ neff ] ) .
in particular , it means that the nearest
neighbour spacing distribution of the riemann zeros can be calculated as follows .
first , find the expansion of the nearest
neighbour distribution for @xmath7 random unitary matrices in inverse powers of @xmath18 , namely @xmath74 ( the fact that the first correction is of order @xmath75 follows from results in @xcite ) .
the expansion ( [ pscue ] ) is difficult to derive analytically .
we have therefore computed numerically the spacing distribution @xmath76 from eqs .
( [ psdef]-[enofs ] ) .
the correction term is then obtained by computing @xmath77 $ ] for increasing values of @xmath18 .
an alternative method would be to use a non - linear differential equation for @xmath78 as derived in @xcite .
second , to leading order , at a given height @xmath2 above the real axis , replace in eq .
( [ pscue ] ) @xmath18 by @xmath79 given by eq .
( [ neff ] ) .
finally , we shall approximate the next - to - leading correction of @xmath80 according to the rule ( [ sca ] ) .
in such an approximation the nearest
neighbour spacing distribution of the riemann zeros equals the universal random matrix result , @xmath81 , plus the correction @xmath82 , where @xmath83 @xmath79 and @xmath69 are given by eq .
( [ neff ] ) and ( [ alpha ] ) , respectively .
figure 2(a ) shows the comparison between the numerical results and eq.([psz2 ] ) for zeros located on a window around @xmath23 ( as in fig .
[ fig1 ] ) .
the effective matrix size is @xmath84 ( instead of @xmath24 ) , and @xmath85 .
the agreement is quite good , and shows that @xmath79 is the correct matrix size in this case . for comparison ,
we have plotted as a dashed curve the theoretical formula ( [ psz2 ] ) without the rescaling of the variable @xmath12 .
figure 2(b ) is a plot of the difference between odlyzko s results and the prediction ( [ psz2 ] ) .
there is still some structure visible , which might be attributed to the @xmath86 correction . to test the convergence ,
we have made the same plot but now using one billion zeros located on a window around @xmath87 , which corresponds to @xmath88 ( instead of @xmath89 ) and @xmath90 ( fig .
[ fig3](a ) ) .
now the agreement is clearly improved .
the difference between the prediction ( [ psz2 ] ) and the numerical results , plotted in fig .
[ fig3](b ) , shows a structureless remain .
= 2.5 in [ fig3 ]
for zeros of the riemann zeta function located around @xmath0 , we derived an heuristic formula for the nearest neighbour spacing distribution that contains finite@xmath2 corrections .
we argued that to leading order the corrections are the same as for random matrices from the circular unitary ensemble of size @xmath91 .
we proposed to describe the next to leading order correction as a simple scaling of the dominant term .
two main conjectures were used .
the most important is the explicit expression for the two point correlation function for the riemann zeros ( eqs .
( [ diag ] ) and ( [ off ] ) ) obtained in @xcite .
the second is the statement that to leading order deviations from random matrix predictions reduce to a change of the kernel ( [ kern ] ) .
it follows that our procedure to compute the finite@xmath2 corrections of the complex riemann zeros from unitary matrices should be valid for any local statistics .
it is known that at finite heights @xmath2 above the real axis the appropriate symmetry group of the riemann zeta function is not @xmath7 .
it has been conjectured @xcite that @xmath92 is a member of a family obeying the symplectic symmetry @xmath93 .
the possible connections of our results with finite@xmath2 corrections associated to this symmetry deserve to be investigated .
p. j. forrester , n. c. snaith and j. j. m. verbaarschot , eds , _ special issue : random matrix theory _ , j. phys a * 36 * ( 2003 ) 2859 - 3646 ; f. mezzadri , n. c. snaith , eds , _ recent perspectives in random matrix theory and number theory _ , cambridge university press , cambridge , 2005 .
a. m. odlyzko , _ the @xmath94-nd zero of the riemann zeta function _ , in dynamical , spectral , and arithmetic zeta functions , m. van frankenhuysen and m. l. lapidus , eds . ,
soc . , contemporary math .
series * 290 * ( 2001 ) 139 - 144 .
e. b. bogomolny , _ spectral statistics and periodic orbits _ , in new directions in quantum chaos , proceedings of the international school of physics `` enrico fermi '' , course cxliii , g. casati , i. guarneri and u. smilansky , eds , varenna , 2000 , 333 - 368 .
e. b. bogomolny and j. p. keating , _ random matrix theory and the riemann zeros i : three and four correlations _ , nonlinearity * 8 * ( 1995 ) 1115 - 1131 , _ ibid _ _ ii : n point correlations _ , nonlinearity * 9 * ( 1995 ) 911 - 935 .
a. v. andreev and b. l. altshuler , _ spectral statistics beyond random matrix theory _ , phys .
* 75 * ( 1995 ) 902 - 905 ; s. m. nishigaki , _ level spacings at the metal - insulator transition _ , phys .
e * 59 * ( 1999 ) 2853 - 2862 . | it has been conjectured that the statistical properties of zeros of the riemann zeta function near @xmath0 tend , as @xmath1 , to the distribution of eigenvalues of large random matrices from the unitary ensemble . at finite @xmath2 numerical results
show that the nearest - neighbour spacing distribution presents deviations with respect to the conjectured asymptotic form .
we give here arguments indicating that to leading order these deviations are the same as those of unitary random matrices of finite dimension @xmath3 , where @xmath4 is a well defined constant .
24.5 cm |
when low- and intermediate - mass stars evolve along the asymptotic giant branch where double shell burning of he and h takes place in the interior , the he burning becomes unstable and gives rise to periodic thermonuclear runaway ( `` thermal pulse '' or `` he shell flash '' ) , which induces mixing of newly synthesized @xmath9c and other processed materials at the surface ( third dredge - up , iben 1975 ; sugimoto & nomoto 1975 ) . the mixing gradually enhances the surface abundance of carbon , which will eventually turn a star originally oxygen - rich in the surface chemical composition into a carbon star ( cf . iben & renzini 1983 ) . among cool luminous carbon stars , there are a group of stars that show silicate dust emission features in the mid - infrared at 10 and 18 @xmath10 m ( little - marenin 1986 ; willems & de jong 1986 ) ; these are the so - called `` silicate carbon '' stars .
silicates are the signature of oxygen - rich chemistry in their circumstellar dust shells , while their optical spectra dominated by molecular absorption bands of c@xmath11 and cn show that their atmospheres are carbon - rich .
the oxygen - rich chemistry in their circumstellar envelopes is also confirmed in the gas - phase by the detection of water vapor masers at 22 ghz ( benson & little - marenin 1987 ; nakada et al .
even carbon - rich objects showing crystalline silicate features have been discovered ( waters et al .
1998 ; molster et al .
2001 ) .
the intriguing co - existence of oxygen - rich and carbon - rich chemistries in a single system prompted two hypothetical explanations .
one was the binary system consisting of a carbon star and a dust - enshrouded m - type giant ( oh / ir ) star ( benson & little - marenin 1987 ; little - marenin , benson , & dickinson 1988 ) .
the other was more interesting , i.e. , that we are witnessing a brief evolutionary stage where the star is in transition from an oxygen - rich star to a carbon star by the third dredge - up while the remnant oxygen - rich dust shell is still visible ( willems & de jong 1986 , 1988 ) .
these earlier pictures were , however , discarded based on the absence of both spectroscopic signatures of a luminous m - type ( oh / ir ) companion ( noguchi et al . 1990 ; lambert , hinkle & smith 1990 ) and variability of the color - indices and silicate features ( chan & kwok 1988 ; lloyd - evans 1990 ) as well as a stringent requirement that primary and secondary must have very similar masses ( lambert et al . 1990 ) .
near - infrared speckle interferometry as well as water maser monitoring also ruled out the presence of a luminous m - type companion to the primary carbon star ( engels & leinert 1994 ) among silicate carbon stars .
morris ( 1987 ) suggested that a binary system of a mass - losing red giant star and either a main - sequence dwarf or a white dwarf companion can develop an accretion disk around the companion , sometimes a circumbinary disk , and even a circum - primary disk , depending on the system configuration .
lloyd - evans ( 1990 ) was inspired by morris s picture to propose that in silicate carbon stars the oxygen - rich material accumulated in a disk around a hypothetical companion when the primary was an oxygen - rich mass - losing star , which later turned into a carbon star through the third dredge - up .
this picture reconciles the observed red infrared color and the relatively small extinction in the optical .
the disk should be somewhat thickened and may extend to the circumbinary region to provide a sufficient mid - infrared flux .
engels & leinert ( 1994 ) inferred for v778 cyg and eu and a minimum radius of the molecular reservoir where water masers reside of 45 sin @xmath12 au , where @xmath12 denotes the inclination of the reservoir , which is 90@xmath13 when seen edge - on , based on the assumed mass of 1 @xmath14 for the primary , keplerian motion for the maser components , and the constancy of radial velocities of the maser lines .
kahane et al .
( 1998 ) and jura & kahane ( 1999 ) proposed the existence of a circumbinary reservoir in keplerian motion in silicate carbon stars , bm gem and eu and , on the basis of the detection of co j=1 - 0 and j=2 - 1 emission with very narrow widths at their systemic velocities .
they suggested that the silicate grains reside in the reservoir , which was presumably built by the influence of a postulated unseen companion when the primary was an oxygen - rich giant .
waters et al .
( 1998 ) noted the similarity to silicate carbon stars of red rectangle that possesses a circumbinary disk showing crystalline silicate features , an extended carbon - rich outflow , and narrow co emission lines .
however , yamamura et al . ( 2000 ) suggested a picture similar to that reported by lloyd - evans ( 1990 ) based on analysis of the iso spectrum of another silicate carbon star , v778 cyg .
they found that dust grains in the circumbinary region where the grains can be warm enough to emit the silicate features will be blown out in less than one orbital period , and thus unable to form a circumbinary reservoir .
they concluded that the source of the silicate features must be the oxygen - rich material continuously blown out from the disk around a companion by the primary s wind and radiation pressure .
the above scenarios all postulate a low - mass , low - luminosity companion , although no observational evidence has yet been provided . to gain further understanding of silicate carbon stars , it is essential to determine whether they indeed have a companion star .
it would be very difficult , however , to detect a postulated low - luminosity companion in the optical or longer wavelength bands because the primary carbon star must outshine the companion by many orders of magnitude in these wavelength regions . in this respect
, it has long been known that cool carbon stars exhibit extreme violet flux deficiency ( shane 1928 ) , the agent of which has not yet been unambiguously identified ( `` violet opacity problem '' ) . indeed ,
cool carbon stars are very dim in the violet region ; e.g. , @xmath15 of the carbon star y cvn is 8.9 ( nicolet 1978 ) , while those of m5 giants and dwarfs are 4.2 and 2.8 , respectively ( allen 1976 ) .
this phenomenon can be exploited to search for signatures of companions to silicate carbon stars in the violet spectral region .
therefore , we performed high sensitivity spectroscopic observations of the visually brightest silicate - carbon stars bm gem , v778 cyg , and eu and , in the violet spectral region .
while we saw no significant violet emission in v778 cyg or eu and , we have detected a featureless continuum below 4000 in bm gem , where the balmer continuum is higher than the paschen continuum at the balmer limit , which is very unusual for a cool luminous carbon star .
in addition , the balmer series lines were traced from h@xmath0 up to h@xmath1 and they showed distinct p cygni - type profiles .
the profiles give a gas expansion velocity of at least 400 km s@xmath2 .
such a high velocity outflow has never been observed in any type of currently mass - losing carbon star .
additional observations 75 days later further revealed considerable time variability of the balmer lines .
in addition , the bm gem system was found to be similar to the mira ( @xmath8 cet ) system , which is the prototype of a binary system consisting of a luminous agb star ( mira a ) and a low - mass , low - luminosity companion ( mira b ) . in the following sections , we discuss the origin of the continuum emission and the p cygni - type balmer lines , and argue for the presence of a companion to bm gem .
observations of bm gem , the brightest silicate carbon star in the visual region , were made with the high dispersion spectrograph ( hds ; noguchi et al .
2002 ) at the nasmyth focus of the subaru telescope ( kaifu et al .
2000 ; iye et al .
2004 ) on january 29 and april 14 , 2001 .
we used the atmospheric dispersion corrector for the optical nasmyth focus .
the entrance slit width was set to 360 @xmath10 m ( @xmath16 ) , which corresponded to a resolution of @xmath450,000 .
a quartz glass filter was inserted behind the slit .
hds is equipped with two mosaiced ccds from eev with 2048@xmath174100 pixels of 13.5 @xmath10 m square .
we used the 250 grooves mm@xmath2 cross - disperser grating to observe the range between 35505200 ( 35504350 and 44005200 ) .
we made two 30-minute exposures and one 30-minute exposure in january and april , respectively .
a bright j - type carbon star , y cvn , was also observed for 15 minutes on february 1 , 2001 ( ut ) with the same settings for comparison , because all the silicate carbon stars examined spectroscopically were known to be j - type stars ( lloyd - evans 1990 ) , which have low @xmath9c/@xmath18c abundance ratios ( @xmath19 ) in the atmosphere ( bouigue 1954 ) .
we also observed the second and third brightest silicate carbon stars , v778 cyg and eu and , for 40 minutes each on july 30 , 2001 with almost the same settings but we employed a slit width of @xmath20 and 2@xmath172 binning for the ccd readout to attain as high a signal to noise ratio as possible , which gave an effective spectral resolution of @xmath438,000 .
a summary of the observations is given in table 1 .
the last column shows the signal to noise ratio at around 4000 in the reduced one - dimensional spectrum .
the photometric and astrometric data of the targets taken from the literature are summarized in table 2 .
the second and third columns show the hipparcos parallax and the uncertainty in the parallax ( esa 1997 ) , respectively . near infrared data in the sixth through tenth columns
are from noguchi et al .
( 1981 ) for bm gem and y cvn and from noguchi et al .
( 1990 ) for v778 cyg and eu and . @xmath21 and @xmath22 magnitudes are from the hipparcos and tycho catalogs ( esa 1997 ) for bm gem and y cvn , and from alksnis & zaime(1993 ) for v778 cyg and eu and .
data reduction was performed using the echelle package on noao iraf .
standard procedures were followed : bias subtraction , flat fielding , scattered light subtraction , extraction of one - dimensional spectra , dispersion correction by th - ar lamp spectra , and removal of the echelle blaze profile by continuum lamp spectra .
strong cosmic ray events were removed manually .
we detected significant emission in bm gem throughout the observed wavelength range .
we then attempted to apply flux calibration to the observed spectra of bm gem .
a star , feige 34 , observed on january 28 , 2001 under @xmath23 seeing with @xmath24 slit width to let virtually all the incident light enter the spectrograph , was used for our approximate flux calibration .
we derived the system response from the observed spectra of the hot white dwarf feige 34 .
we applied an atmospheric extinction correction based on the mauna kea extinction curve ( beland , boulade & davidge 1988 ) to the spectra , and compared them with the calibrated magnitudes at 50 interval of feige 34 prepared in iraf , which is based on massey et al .
we then corrected the spectra of bm gem for the atmospheric extinction and the system response .
finally , we applied a correction for light losses from the entrance slit due to seeing .
seeing size was measured to be @xmath25 and @xmath26 for january and april observations , respectively .
we approximated the seeing image by a single gaussian and obtained values of 0.71 and 0.68 for the slit transmission efficiency for january and april , respectively , against @xmath16 slit width assuming perfect telescope guiding .
the spectra of y cvn were also calibrated in the same manner for comparison .
the seeing size at the observation of y cvn was @xmath25 , which gave a slit efficiency of 0.71 .
the uncertainty in the absolute flux densities thus obtained is non - negligible because the data were taken on different nights with somewhat different seeings with a slit width close to the size of the seeing image , a flux standard was observed on only one night , and the violet region is near the atmospheric cutoff , which is sensitive to extinction corrections .
the uncertainty due to seeing and guiding , which applies to bm gem , is estimated to be @xmath270.25 mag in each spectrum , adopting maximum possible errors of @xmath28 in both seeing estimate and guiding
. the uncertainty in the extinction correction which applies to both bm gem and feige 34 can give rise to an underestimate in their brightness of about 0.2 mag .
it is obtained by assuming the employed extinction coefficient of 0.37 mag airmass@xmath2 at 3600 ( beland et al .
1988 ) , with smaller values for longer wavelengths , was subject to a possible increase of 50% , although all our observations were made under clear sky conditions and the airmasses were only 1.01 , 1.20 , and 1.10 for bm gem(jan ) , bm gem(apr ) , and feige 34 , respectively .
taking all these uncertainties into account , each flux density derived is likely to be accompanied by an uncertainty range of ( -0.45 , + 0.25 ) , while the relative uncertainty between bm gem(apr ) and bm gem(jan ) can be as large as 0.7(=0.25 + 0.25 + 0.2 ) .
any difference between the two observations smaller than this magnitude is dominated by a calibration error and should be treated as insignificant .
we did not apply the above flux calibration to either v778 cyg or eu and because we did not see any significant emission in the region shortward of 4000 .
these stars are 3 magnitudes fainter than bm gem in @xmath21 ( table 2 ) .
if we put bm gem farther away from us so that its @xmath21 becomes 3 mag fainter , then any signal of the violet continuum on the raw ccd image becomes only one - quarter or less of the ccd read noise and merely equal to it or less even in the case of 2@xmath172 binning .
then , the non - detection of the violet continua here in v778 cyg and eu and indicates that they are not as bright as that of bm gem for their @xmath21 magnitudes . nevertheless , it does not necessarily mean that they do not have a similar violet continuum .
it is necessary to achieve much higher sensitivities for v778 cyg and eu and to distinguish between the presence and absence of a continuum like that in bm gem .
we will not discuss the violet spectra of v778 cyg or eu and further in this paper .
the observed spectrum was corrected for the doppler shift and transformed to the wavelength scale with respect to the local standard of rest ( lsr " ) using the tasks rvcorrect and dopcor in iraf .
the stellar systemic velocity with respect to lsr was also subtracted in the final spectra shown in this paper .
thus , matter stationary to the center of mass of the system should appear at zero velocity .
we employed the systemic velocity of 73.2 km s@xmath2 ( v@xmath29 ) for bm gem based on mm - wave co emission lines ( kahane et al .
1998 ; jura & kahane 1999 ) .
this is in agreement with v@xmath29 of 74.7 km s@xmath2 derived from the heliocentric radial velocity of 85.3 @xmath270.4 km s@xmath2 obtained for the co first overtone bands by lambert et al .
( 1990 ) . for y
cvn , we used 21.2 km s@xmath2 , based on the mm - wave co j=1 - 0 emission ( izumiura , ukita , & tsuji 1995 ) , as the systemic velocity with respect to lsr .
it should be noted that radial velocities of observed spectral lines given below will carry a typical uncertainty of @xmath40.5 km s@xmath2 on the basis of the observed line widths and strengths , achieved s / n ratio , and employed spectral resolution .
the resulting spectra of bm gem taken with the ccd that observes the shorter wavelength portion , reduced to a resolution of 1 , are shown in figure 1 along with those of y cvn .
we have adjusted the vertical scales so that the slopes of the spectra between 4050 and 4200 look similar to one another .
the spectra of bm gem show significant balmer and paschen continuum emission in the violet region , which is evident when compared with y cvn , which shows no detectable violet continuum . what is more intriguing is that the level of the balmer continuum is 1.3 times higher than that of the paschen continuum when compared around the balmer limit in both observation occasions .
this indicates that the continuum emission comes from an ionized gas region .
they also show a series of emission lines that are identified with hydrogen balmer series lines from h@xmath0 through h@xmath1 .
moreover , the balmer lines show p cygni - type line profiles , suggesting the presence of an outflow being accelerated outward against the central continuum source .
h@xmath30 emission was not identified at all in our spectra recorded on the ccd that covers the longer wavelength part , probably because the spectral range is dominated by the strong carbon star spectra .
broad k emission with a blueward depression was also detected but h emission was absent .
the h emission could be absorbed by the hydrogen in the outflow because the expected wavelength range of h line ( 3968.470 , moore 1959 ) coincides with the absorption core of the p cygni profile of h@xmath3 line ( 3970.074 , moore 1959 ) .
note here that the features that mimic emission lines , marked with small vertical ticks seen shortward of 4150 in y cvn , are contaminations of ghost spectra of very strong red light of the star , probably due to the cross - disperser grating .
the same ghost features are also present in the spectra of bm gem at the same positions , by which only h@xmath31 of the balmer lines is significantly contaminated
. the fluxes per unit wavelength at wavelength @xmath32 , @xmath33 , of the balmer continuum of bm gem after the approximate flux calibration in 2 are 3.0@xmath1710@xmath34 erg @xmath35 s@xmath2 @xmath2 in the january spectra and 4.3@xmath1710@xmath34 erg @xmath35 s@xmath2 @xmath2 in the april spectra immediately shortward of the balmer limit at 3646 .
these correspond to 16.1 mag and 15.7 mag following the definition of @xmath36 log(@xmath37 ) @xmath38 48.59 , where @xmath39 is the magnitude ,
@xmath40 is frequency , and @xmath37 is the flux per unit frequency at @xmath40 in erg @xmath35 s@xmath2 hz@xmath2 . here , @xmath37 is calculated by @xmath41 ( i.e. , @xmath42 ) , where @xmath43 is the speed of light .
this definition gives 0.048 for the magnitude of vega at 5556 ( hayes & latham 1975 ) .
the apparent difference in the continuum level of 0.4 mag is not significant , considering the possible uncertainty of 0.7 mag given in 2 .
thus , we argue that the continuum level is unchanged , and adopt a geometrical mean of the two values , 3.6@xmath1710@xmath34 erg @xmath35 s@xmath2 @xmath2 or 15.9 mag .
we take the differences of @xmath270.2 mag between this 15.9 mag and the two direct values as an additional uncertainty . adding this to that given in 2
, the total uncertainty range has a magnitude of ( -0.65 , + 0.45 ) .
similarly , the paschen continuum at 4000 has 16.2 mag and 15.6 mag in the january and april spectra , respectively , and they give 15.9 and ( -0.75 , + 0.55 ) in magnitude as the flux level and uncertainty range , respectively .
note , however , that bm gem shows light variation of 0.20.4 mag in 100 days in @xmath21 and @xmath22 bands ( alksnis & zaime 1993 ) .
whether bm gem is indeed variable in the violet continuum is an interesting issue for future studies . in the original high - resolution spectra of bm gem ,
cn red system lines between 48305250 give average radial velocities around @xmath44 km s@xmath2 with respect to the systemic velocity adopted in 2 in the two observations .
absorption features of triplet at 4030 - 4034 were identified , while low - excitation lines of were observed in emission shortward of 4040 , both of which are typical for late - n type carbon stars , as described by gilra ( 1976 ) .
the lines at the two epochs give average radial velocities around @xmath45 km s@xmath2 , which are compatible with the mm wave co outflow velocity of @xmath46 km s@xmath2 and may suggest that they form in the mass - loss flow of the carbon star .
the lines have apparent fwhms of @xmath410 km s@xmath2 and show average radial velocities near 0 km s@xmath2 in the two observations , suggesting that they originate in the extended atmosphere of the carbon star .
the uncertainties in the average radial velocities given above are as large as 1 km s@xmath2 for all of the cn red system , triplet , and lines . whether this magnitude of uncertainties , which is slightly larger than the internal uncertainties of @xmath47 km s@xmath2 given in 2 , is due to their intrinsic variabilities or not is an issue for a future work .
the continuum shortward of 4040 is almost featureless except for the balmer and lines .
no absorption features typical of stellar photospheres are seen .
figure 2 shows higher resolution views of the vicinity of h@xmath48 , h@xmath49 , and h@xmath50 lines together with k line observed on the two nights 75 days apart .
the zero systemic velocity corresponds to the middle of the p cygni profile , which argues for the outflowing gas being associated with bm gem .
the line profiles show that the emitting gas , which is likely ionized , is expanding at a velocity as large as 400 km s@xmath2 in the sight line . this is greater than the highest velocity outflows observed to date in agb and post - agb stars , v hya ( 200 km s@xmath2 , knapp , jorissen , & young 1997 ) , crl 618 ( 200 km s@xmath2 , cernicharo et al .
1989 ) , oh231.8 + 4.2 ( 330 km s@xmath2 , alcolea , bujarrabal , & snchez contreras 1996 ) , and r coronae borealis stars ( 200350 km s@xmath2 , clayton , geballe , & bianchi 2003 ) , with the exception of mira b ( 250 and 400 km s@xmath2 , wood , karovska , & hack 2001 , wood , karovska , & raymond 2002 ) and v854 cen ( 390 km s@xmath2 , clayton et al .
it is , however , much smaller than those of jets seen in symbiotic objects ( e.g. , @xmath46000 km s@xmath2 in mwc560 , tomov et al .
in addition , a change in the line shape in each of the balmer lines over a period of 75 days is evident .
the blue edge of the absorption features shifted redward by 200 km s@xmath2 , while the red edge did so by only 50 km s@xmath2 .
our discoveries of the balmer and paschen continua , the former being higher than the latter , the p cygni - type balmer series lines , and their line profile variability all suggest that bm gem is associated with a compact ionized gas region that is accelerated to form a rather spherical , high velocity outflow .
no mechanism is known for a single cool luminous carbon star to produce an ionized gas region as well as a high velocity outflow .
they can , however , be accounted for if we introduce an unseen , low - mass companion that captures matter in the stellar wind from the primary to give rise to an accretion disk .
this hypothesis is partially supported by similar spectral features observed in mira b ( @xmath8 cet b ) ( see 4 ) , which is a low - mass companion to the agb star mira a ( @xmath8 cet a ) and is considered to be associated with an accretion disk ( joy 1926 , 1954 ; deutsch 1958 ; warner 1972 ; reimers & cassatella 1985 ) .
low - mass companions with an accretion disk have also been suggested for r coronae borealis stars , which show outflows with similar high velocities and lines that require excitation sources ( rao et al .
1993 ; clayton et al .
2003 ) .
formation of an accretion disk has proven to be robust around a companion star to a mass - losing giant in various detached configurations ( mastrodemos & morris 1998 , 1999 ) .
the inner part of the accretion disk must be heated by the released gravitational potential energy from the accreting matter .
a hot ionized gas region should form between the innermost region of the disk and the surface of the companion .
the accretion phenomenon may also be responsible for the discovered high velocity , variable outflow , although the details of the acceleration mechanism is not yet settled for this type of outflow .
alternatively , the outflow may be due to dust formation in the circumcompanion region , analogous to the scenario proposed for rcb stars by clayton ( 1996 ) .
dust grains would be accelerated through radiation pressure , dragging gas to form the observed high velocity outflow .
the ionization could then be due to collisional ionization in the flow . in this case , to be compatible with the observed broad absorption core that extends from the systemic velocity to the terminal velocity in the p cygni profile , the ionization must occur before the gas acceleration is completed .
otherwise the absorber in front of the continuum source has terminal outflow velocity and a much narrower absorption component appears near the terminal velocity .
in either case of an outflow powered by accretion or driven by radiation pressure on dust grains , the observed spectral features favor the presence of a companion to bm gem .
the companion should be either a white dwarf or a low - mass dwarf star because hypotheses invoking a luminous companion have been rejected by previous studies ( noguchi et al .
1990 ; lambert et al . 1990 ; chan & kwok 1988 ; lloyd - evans 1990 ) .
the gas expansion velocity provides a hint to distinguish the candidates for the companion .
white dwarfs have escape velocities on the order of several thousand km s@xmath2 , while those of low - mass dwarf stars are of the order of several hundred km s@xmath2 .
if gas acceleration occurs near the surface of the secondary and the flow is aligned with the sight line , the latter is the case for bm gem .
however , the former could also be the case if the acceleration takes place at some point distant from the white dwarf , as suggested by warner ( 1972 ) , or the flow has a narrow opening angle and is markedly inclined with respect to the sight line .
below , we examine whether such a binary and accretion hypothesis is plausible energetically .
we first attempted to determine the absolute magnitudes of bm gem at the observed wavelengths by finding the distance to bm gem .
although it exists , the hipparcos parallax of bm gem ( table 2 ) is unreliable .
claussen et al .
( 1987 ) gave 1.51 kpc assuming that carbon stars have a constant absolute k - band magnitude of -8.1 .
we made another estimate by comparing the near - infrared flux densities of bm gem with those of y cvn , as they are @xmath18c - rich ( j - type ) carbon stars resembling each other in the spectral characteristics in the optical and near - infrared ( barnbaum 1991 ; ohnaka & tsuji 1999 , yamamura et al .
2000 ) , which may indicate they have similar intrinsic properties .
y cvn is the only j - type star that has a relatively reliable hipparcos parallax ( table 2 ) of 218 pc with 16% uncertainty .
it is evident from the data shown in table 2 that the differences between y cvn and bm gem in @xmath51 through @xmath52 bands are quite constant with a simple mean of 3.72 , which indeed shows the two stars to be similar and gives a distance to bm gem of 1.21 kpc .
in addition , the distance of 1.51 kpc reported by calussen et al . may be reduced to 1.14 kpc , considering the distance they gave of 0.29 kpc for y cvn should be readjusted to 0.218 kpc .
hence , we adopted 1.2 kpc with a conservative uncertainty factor of 1.2 as a reasonable estimate for the distance to bm gem .
the total luminosity then becomes 5.4@xmath53 ( cf .
groenewegen et al .
1992 ) , which implies that they are on the agb even with the distance uncertainty .
interstellar extinction toward bm gem ( @xmath54 and @xmath55 ) at 3650 ( @xmath56 band ) , @xmath57 , is estimated to be at most 0.6 mag .
it is obtained by @xmath58 where @xmath59 is the column density of atomic neutral hydrogen , @xmath60 is a conversion factor from @xmath59 to the total column density of neutral hydrogen @xmath61 including and h@xmath11 , @xmath62 , and @xmath63 and @xmath64 are the extinction at @xmath21 band and @xmath22 band , respectively .
we read @xmath65 in the direction of bm gem from heiles ( 1975 ) .
other quantities are found in cox ( 2000 ) and references therein : @xmath66 ( bohlin , savage , & drake 1978 ) ; @xmath67 , a standard value for diffuse interstellar matter ; @xmath68 for @xmath69 ( cardelli , clayton , & mathis 1989 ) .
then , we find @xmath70 , which should be an upper limit because the column denisity that heiles ( 1975 ) gave is an upper limit to bm gem , which lies somewhere between the boundary of the hi gas distribution in that direction and us . since there are no significant molecular clouds found in the direction of bm gem ( e.g. , dame , hartmann , & thaddeus 2001 ) , @xmath60 should be close to unity and the use of @xmath69 should be justified .
therefore we regard @xmath71 as an upper limit .
we have also obtained another estimate of @xmath72 for the interstellar extincion toward bm gem , using @xmath73 read from schlegel , finkbeiner , & davis ( 1998 ) by way of nasa / ipac infrared science archive , @xmath69 , and @xmath68 .
this is another upper limit because bm gem must lie somewhere between the boundary of the dust distribution in the sight line and us .
taking these into account we adopt the 0.3 mag directly from far - inrared dust observations as a nominal value , the 0.6 mag from the observations as a maximum value , and no extinction as a minimum value , for @xmath57 to bm gem , namely , @xmath74 .
then we find the absolute magnitudes at @xmath75 ( @xmath76 ) and 4000 ( @xmath77 ) to be 5.2@xmath78 and 5.2@xmath79 , respectively .
these are obtained from the observed magnitudes of @xmath76=15.9@xmath80 and @xmath77=15.9@xmath81 , the adopted distance of 1.2 kpc , its uncertainty factor of 1.2 , and the same interstellar extinction correction of [email protected] mag to both of them . the distance and
uncertainty factor corresponds to [email protected] mag .
the violet continuum is not simply explained by the photosphere of a postulated companion . for the case of a dwarf companion ,
the probable range of @xmath77 between + 6.3 and + 3.9 corresponds to a photosphere with a spectral type between late g and late f , which should show numerous strong absorption lines in the violet region .
the violet continuum of bm gem is , however , featureless except for the emission lines due to the primary carbon star and the balmer series lines due to the outflow , as mentioned in section 3.1 .
the featureless continuum requires that the companion s photosphere is smeared by veiling with / without obscuration and contributes only a small fraction of the observed violet continuum .
considering the s / n ratios in the original high - resolution spectra of @xmath415 , we should be able to detect absorption lines of which central depths are as weak as 20% of the continuum level at 3 @xmath82 confidence . in this violet region
, there are many absorption lines the central depths of which are as strong as 80% of the continuum level in late f through late k dwarfs . for such strong lines to appear weaker than the 3 @xmath82 upper limit of 20%
, the veiling should be at least 4 times the companion s photosphere .
possible cases are an f- or g - type dwarf obscured by continuous absorption and covered with veiling and a dwarf later than g - type either obscured or not by continuous absorption and covered with veiling .
obscured b- or a - type photosphere is not plausible for the violet continuum because we do not see the balmer jump typical for such spectral types and because the companion should be less massive than the primary carbon star .
the observed violet continuum also can not be explained by the photosphere of a white dwarf companion , because the @xmath75 band brightness , @xmath76 , of a db white dwarf with effective temperature of 25000 k is 9.1 ( allen 1976 ) , and those classified as da and later are less luminous .
therefore , most of the observed violet flux in bm gem should have an origin other than the photosphere of the assumed companion , which is consistent with our accretion hypothesis .
the total flux of the observed continuum emission @xmath83 is approximately obtained by @xmath84 , where @xmath40 and @xmath85 mean frequency and the flux of the continuum at @xmath40 per unit frequency range ( flux density ) and @xmath86 and @xmath87 are their typical values .
we adopt the frequency at @xmath88 ( 1216 ) , @xmath89 for @xmath87 and the flux density at @xmath89 , @xmath90 for @xmath86 to obtain @xmath91 , where @xmath92 is the wavelength of @xmath93 and @xmath94 is the flux of the continuum per unit wavelength range at @xmath88 wavelength . here
, we assume @xmath94 is approximated by @xmath95 , which is the flux of the continuum per unit wavelength range immediately shortward of the balmer limit at 3646 , for which we obtained 3.6@xmath1710@xmath34 erg @xmath35 s@xmath2 @xmath2 in 3.1 .
these approximations should be valid because model spectra of ionized gas show that the continuum emission diminishes rapidly toward higher frequencies with respect to @xmath88 line and because the level of the continuum flux per unit wavelength from an ionized gas at @xmath96 wavelength is similar to that just shortward in wavelength of the balmer limit ( e.g. , harrington , lutz , & seaton 1981 ; pottasch et al .
then , the product @xmath97 should give a reasonable estimate of the total flux .
the above approximations , however , may overestimate the total flux , and another uncertainty of ( -0 , + 1 ) in magnitude should be added .
now , assuming spherical symmetry of the radiation field , the gas radiative luminosity , @xmath98 , can be estimated as @xmath99 where @xmath100 denotes the distance to bm gem .
the total uncertainty range has a magnitude of ( -1.2 , + 2.0 ) by summing those discussed previously , or 0.030.6
@xmath98 in this range is always larger than the gas kinematic luminosity , ( 1/2 ) @xmath101 , and thermal luminosity , @xmath102 , and dominates the total luminosity as far as @xmath103 and @xmath104 k. here , @xmath105 ,
@xmath106 , and @xmath107 are the mass ejection rate , outflow velocity , and total particle number density , respectively , of the high velocity outflow .
@xmath108 , @xmath109 , and @xmath110 are the boltzmann constant , temperature of the gas , and radius under consideration from the center of the outflow source , respectively .
maximum energy input , @xmath111 , expected from the mass - accretion by a companion , is constrained by the gravitational potential energy release from the accreting matter , i.e. , @xmath112 where @xmath113 , @xmath114 , @xmath115 , and @xmath116 are the gravitational constant , the companion s mass accretion rate in @xmath117 , its mass in @xmath14 , and its radius in @xmath118 , respectively . in this equation
@xmath119 is almost constant and near unity for dwarf stars of type late f to late k , and is 2070 for white dwarfs
( allen 1976 ) .
for a white dwarf of 0.5 @xmath14 with 0.01 @xmath118 ( allen 1976 ) , an accretion rate of @xmath120 can afford the observed continuum luminosity of @xmath121 , whereas a k5 dwarf of 0.69 @xmath14 and 0.74 @xmath118 ( allen 1976 ) can give rise to the observed luminosity when the mass accretion rate is @xmath122 .
it is not self - evident if the latter accretion rate is feasible for a dwarf companion in a binary system compatible with the observations , given the current mass loss rate of @xmath123 in bm gem ( kahane et al .
1998 ) . if we assume bm gem is in a binary system consisting of a carbon star of 1.5 @xmath14 ( claussen et al .
1987 ; groenewegen et al .
1992 ) and a dwarf of 0.5 @xmath14 , in a circular orbit with a separation of 30 au , which is one likely configuration , then the orbital velocity of the companion about the primary becomes 7.7 km s@xmath2 .
the outflow velocity of the primary carbon star is 7.5 km s@xmath2 and the current mass loss rate is @xmath132 in bm gem ( kahane et al .
then , equation ( 5 ) gives an accretion luminosity of 0.08 @xmath5 , which is at least compatible with the observed luminosity of bm gem when its uncertainty range is taken into account . in their smoothed particle
hydrodynamic calculations , mastrodemos & morris ( 1999 ) found that the bondi - hoyle accretion is not as efficient as initially thought .
the obtained efficiency ( @xmath114/@xmath127 ) spreads over a range between 0.1% and 10% for the cases they examined that are compatible with the configurations under consideration here .
the efficiencies they found differ not by an order but by a factor from those obtained using equation ( 4 ) for the same parameter sets .
a dwarf companion is thus at least compatible with the observations .
in the previous section , we showed that our results were compatible with bm gem being accompanied by a low - mass , low - luminosity companion with an accretion disk , for which both a main - sequence star and a white dwarf are viable . in this section
, we further consider the characteristics of the bm gem system .
the observed uv - optical ( violet ) spectral features of bm gem nearly parallel those of mira ( @xmath8 cet ) .
mira is known to consist of a long period variable on the asymptotic giant branch , mira a and a low - mass , low - luminosity companion separated by about @xmath133 ( @xmath470 au ) , mira b ( karovska et al . 1997 and references therein ) .
joy ( 1926 , 1954 ) reported the first detection of a uv - optical continuum and of complex profiles of hydrogen balmer lines with emission and absorption cores in mira b. figure 8 of reimers & cassatella ( 1985 ) showed that both balmer and paschen continua existed , and the level of the former was higher than that of the latter .
the balmer lines were found to be in p cygni - type and their profiles were shown to be highly variable on a time scale similar to that found for bm gem ( joy 1926 , yamashita & maehara 1977 ) , although the lines were not in p cygni - type in reimers & cassatella ( 1985 ) .
the p cygni profiles indicated that the outflow velocity was as large as 400 km s@xmath2 ( wood et al .
warner ( 1972 ) examined the energetics of mira b and reported that its total luminosity of @xmath134 could be accommodated by the bondi - hoyle type accretion of mira a s wind to mira b , if the mass loss rate from mira a is greater than 0.8@xmath135 yr@xmath2 .
the actual mass loss rate of mira a is @xmath136 yr@xmath2 ( ryde & sch@xmath137ier 2001 ) , which supports the bondi - hoyle type accretion .
jura & helfand ( 1984 ) also found the uv - optical luminosity of mira b to be 0.2 ( 0.051 ) @xmath5 .
the above parallels between bm gem and mira in the spectral features as well as the uv - optical luminosity support the presence of a companion to bm gem . based on the x - ray luminosity as well as of the outflow velocity of @xmath4400 km s@xmath2
, the companion of bm gem may not be a white dwarf but a dwarf . here
, we again present a parallel discussion with mira .
a very low x - ray luminosity , which is @xmath410@xmath138 of the uv - optical luminosity , in the mira system led jura & helfand ( 1984 ) to conclude that mira b was not a white dwarf but a dwarf .
they argued that the x - ray luminosity would be comparable to the uv - optical luminosity if the companion was a white dwarf .
reimers & cassatella ( 1985 ) noted that there was no direct evidence for the presence of a hot white dwarf companion to mira a , although they favored a white dwarf companion .
based on a similar discussion , ireland et al .
( 2007 ) recently concluded that mira b is a k5 dwarf of 0.7 @xmath14 .
if mira b is indeed a dwarf , then the x - ray luminosity of bm gem would be as small as that of mira in the case of a dwarf companion but could be 10@xmath139 times as much for a white dwarf companion .
mira has an x - ray photon flux of 7.6@xmath140 counts s@xmath2 in the 0.12.4 kev energy band in the second rosat source catalog of pointed observations ( rosat consortium 2000 )
. the x - ray flux from bm gem in the same energy band could be as much as @xmath40.08 counts s@xmath2 for a white dwarf companion , with the difference in their distances of a factor 10 taken into account .
it should have been detected in the rosat all - sky survey ( voges et al .
1999 , 2000 ) , the detection limit of which was about 0.05 counts s@xmath2 , which is actually not the case .
this argues for a dwarf as the companion of bm gem .
however , that both mira and bm gem possess a white dwarf companion can not be excluded completely , because there may exist a mechanism that suppresses the x - ray luminosity arising from the accretion process and one to accelerate the outflow from some distant point from the white dwarf simultaneously .
the latter mechanism is favorable for the picture suggested for mira b by warner ( 1972 ) .
it is thus difficult to choose exclusively between a dwarf and a white dwarf as the companion , although the data favor a dwarf .
the binary separation of the bm gem system is only loosely constrained .
the separation is found to be ( 60.3)@xmath141 au for a dwarf companion , while that for a white dwarf companion is ( 51)@xmath142 au to reproduce the observed luminosity of 0.030.6 @xmath5 using equation ( 5 ) .
we adopted @xmath143 and @xmath144 and assumed that @xmath145 ( the mass of the primary in @xmath14)=1.5 , @xmath146 , @xmath116=0.65 ( dwarf ) or 0.016 ( white dwarf ) , and the system is in a circular orbit .
further , for any combination of @xmath145 between 1.0 and 2.0 and @xmath115 between 0.3 and 1.2 but @xmath147 ( dwarf ) or between 0.3 and 0.7 ( white dwarf ) , there are upper limits of 210 au and 930 au for a dwarf and a white dwarf companion , respectively , where realistic @xmath116 corresponding to @xmath115 ( allen 1976 , lang 1999 ) is used .
as equation ( 3 ) gives an upper limit and equation ( 4 ) is suspected to overestimate the accretion rate ( mastrodemos & morris 1999 ) , the derived upper limits for the separation are relatively stringent .
mastrodemos & morris ( 1999 ) and soker & rappaport ( 2000 ) suggested from theoretical considerations that the separation should be @xmath14830 au to form an accretion disk around a companion , irrespective of whether it is a dwarf or a white dwarf , in the detached binary configurations under consideration here .
the observation that mira b , which is at least 70 au from mira a , is associated with an accretion disk despite the moderate mass loss rate of mira a of @xmath136 yr@xmath2 suggests that the current hydrodynamic simulations do not constrain the upper - limit of the separation very well .
as bm gem has been shown to be a system resembling mira , its binary separation could be as large as 70 au .
ohnaka et al .
( 2006 ) derived an upper limit of @xmath46080 au for the diameter of the inner dust - free region in the silicate carbon star iras08002 - 3803 from mid - infrared interferometry , which at the same time gives an upper limit for the postulated binary separation of @xmath480 au .
however , they suggested that silicate carbon stars with an optically thin dust reservoir , to which bm gem belongs , have binary separations wider than those of silicate carbon stars with an optically thick reservoir , to which iras08002 - 3803 belongs .
in addition , if we pose a constraint that we do not see a roche - lobe overflow , then we find the binary separation should be larger than 1.8 au for any combination of mass ratio between 1 and 7 and carbon star radius between 1 and 1.5 au using the formulation by paczyski ( 1971 ) .
luminosity of the silicate dust in bm gem , @xmath149 is estimated as @xmath150 where @xmath151 , @xmath152 , and @xmath153 are the flux density due to silicate dust , frequency at 9.7 @xmath10 m , and flux density due to silicate dust at 9.7 @xmath10 m peak , respectively .
we assume the emission is isotropic , the feature extends over 8 - 12 @xmath10 m , and has a flux density of about 1.2@xmath154 w m@xmath155 at the peak , which is read from the `` iras low resolution spectrograph '' atlas ( joint iras science working group 1986 ) .
the factor 1/5 is a roughly determined correction factor for substituting the integration with the multiplication obtained by a simple calculation of ( 1/2)(12 @xmath10m-8 @xmath10m)/(9.7 @xmath10 m ) .
the dust luminosity is then @xmath156 .
the accretion process discussed here can not be the energy source of the silicate emission lines .
the energy must be supplied by the radiation from the primary carbon star of @xmath157 ( see 3.2 ) .
the dust luminosity indicates that about 2% of the total stellar luminosity is captured by the dust grains responsible for the silicate features .
barnbaum et al .
( 1991 ) found similar values of 12 % in bm gem and v778 cyg .
silicate emission features have been speculated to arise from a circumstellar disk around a companion as well as from a thickened circumbinary disk ( lloyd - evans 1990 ) .
engels & leinert ( 1994 ) inferred a minimum radius for a circumbinary molecular reservoir of 45 sin @xmath12 au , where @xmath12 denotes the inclination of the reservoir , which is 90@xmath13 when seen edge - on , for v778 cyg and eu and , assuming a 1 @xmath14 primary and circular keplerian motion of the reservoir .
they based their inference on constancy better than 0.06 km s@xmath2 yr@xmath2 of the radial velocity of the water maser lines .
kahane et al .
( 1998 ) and jura & kahane ( 1999 ) detected a very narrow emission component in the mm - wave co lines in bm gem and eu and .
they interpreted the narrow component as due to a circumbinary reservoir in keplerian motion formed by some binary interaction with an unseen companion .
they argue that the oxygen - rich material responsible for the silicate emission is stored in the long - lived circumbinary reservoir of 1001000 au size , which was formed when the primary star was an oxygen - rich mass - losing star .
this picture is further reinforced by the study of red rectangle by waters et al .
( 1998 ) who noted the similarity to silicate carbon stars of red rectangle , which possesses a circumbinary disk showing crystalline silicate features , an extended carbon - rich outflow , and narrow co emission lines .
the circumbinary reservoir of 1001000 au size is compatible with both the binary separation of ( 0.36)@xmath141 au ( as well as a stringent upper limit of 210 au ) derived for a dwarf companion to bm gem and the separation of ( 15)@xmath142 au ( as well as a stringent upper limit of 930 au ) for a white dwarf companion
. however , yamamura et al . ( 2000 ) found that silicate dust grains responsible for the emission features in v778 cyg should have temperature between 600 and 300 k , which implies their location to be between about 25 and 100 au ( 1250 stellar radii ) from the primary .
they concluded that a circumstellar disk around an invisible companion should be the reservoir of oxygen - rich material , because they found it difficult to locate a circumbinary disk in the vicinity of the companion s orbit .
dust grains there will be swept outward in less than one orbital period by radiation pressure from the primary .
they speculated that the silicate features originate from oxygen - rich material continuously blown out by radiation pressure of the primary from the circumcompanion disk that was built when the primary was an oxygen - rich mass - losing star .
furthermore , ohnaka et al .
( 2006 ) suggested that silicate carbon stars may be classified into two groups : one with an optically thick circumbinary dust reservoir with a smaller binary separation , and another with an optically thin dust outflow from the companion s disk with a wider binary separation .
following their criteria , bm gem belongs to the latter group , and thus may show dust outflow from the companion , possibly extending out in a region at 50100 au from the primary .
therefore , the region at 30100 au is a likely location of the circumstellar molecular / dust reservoir according to both yamamura et al .
and ohnaka et al .
this location is compatible with the binary separation for a dwarf companion , but not with the separation for a white dwarf companion .
finally , it should be noted that all of the silicate carbon stars examined spectroscopically are known to be j - type stars ( lloyd - evans 1990 , 1991 ) , while the opposite is not true .
it has been argued that j - type carbon stars are the direct descendants of r - type carbon stars ( lloyd - evans 1990 ; lambert et al .
1990 ) , which form through a mechanism unrelated to the third dredge - up , as both groups of stars are @xmath18c - rich and lack s - process enhancements in the surface chemical compositions ( utsumi 1985 , dominy 1985 ) .
abia & isern ( 2000 ) , however , favor another scenario in which the low - mass j - type stars discussed here form via a combination of non - standard extra mixing and cool bottom processing ( wasserburg , boothroyd , & sackmann 1995 ) early on the agb , which is consistent with their luminosities being similar to those of normal cool carbon stars on the agb ( wallerstein & knapp 1998 ; alksnis et al .
as there seems to be a connection between the j - type nature and the silicate features , the presence of a companion to bm gem would suggest a possible connection between binarity and j - type phenomena , which is worth investigating further .
radial velocity monitoring of j - type stars would be important , which could be carried out by observing the emission lines in the violet region as well as numerous photospheric molecular absorption lines .
we observed the violet spectra of the three brightest silicate carbon stars , bm gem , v778 cyg , and eu and , using the high dispersion spectrograph on the subaru telescope , and used a prototypical j - type carbon star y cvn for comparison .
balmer continuum and paschen continuum emission typical for ionized gas were found to be prominent in the region shortward of 4000 in bm gem , while no significant emission was observed in the same region in v778 cyg , eu and , or y cvn . in bm gem ,
balmer series lines were also detected from h@xmath0 through h@xmath1 , and they showed p cygni - type line profiles .
broad k emission with blueward depression was also found , while the h line is missing perhaps because it is absorbed by hydrogen in the outflow .
the p cygni profiles give the gas outflow velocity of at least 400 km s@xmath2 .
such spectral features have not been observed in other cool luminous carbon stars .
furthermore , the p cygni profiles changed significantly within a period of 75 days , suggesting a compact geometry of the outflow .
the overall spectral features mimic those of mira b. all these features in the cool carbon star bm gem suggest the presence of a companion that gives rise to an accretion disk , which is responsible for the ionized gas region and the observed high velocity outflow .
we investigated the energetics of the observed emission assuming a binary system and bondi - hoyle type mass accretion process .
the luminosity of the observed continuum emission is estimated to be @xmath40.2 ( 0.030.6 ) @xmath5 , while the silicate dust features convey about 110 @xmath5 , which shows that the silicate dust grains are heated not by the phenomenon discovered here but by radiation from the carbon star .
we found that the violet continuum luminosity is accommodated if an accretion rate on the order of @xmath158 yr@xmath2 is achieved to a dwarf companion , while @xmath159 yr@xmath2 is sufficient to a white dwarf companion .
although the required accretion efficiency seems rather high , a main - sequence companion is favored based on the observed outflow velocity of 400 km s@xmath2 and the non - detection of x - ray flux in the rosat all sky survey .
the possibility of a white dwarf companion , however , can not be ruled out , because the outflow could be narrow and inclined with respect to the sight line or be accelerated at some distant point from the surface of the white dwarf .
based on the above findings combined with those of previous studies , we conclude that bm gem is associated with a low - mass , low - luminosity companion giving rise to an accretion disk .
we note that the spectral features and uv - optical luminosity obtained for the bm gem system closely resemble those of the mira system , which should represent the actual similarity between the binary configurations of the two systems .
we derived the binary separations based on the observed uv - optical luminosity as ( 0.36)@xmath141 au and ( 15)@xmath142 au for a dwarf and a white dwarf companion , respectively , assuming the primary mass of 1.5 @xmath14 , companion mass of 0.5 @xmath14 , circular orbits , and bondi - hoyle type accretion .
we also derived another stringent upper limit of the separation of 210 au and 930 au for a dwarf and a white dwarf companion , respectively .
the separations for a dwarf are compatible with both the circumcompanion dust reservoir proposed by yamamura et al .
( 2000 ) and ohnaka et al .
( 2006 ) and the circumbinary dust reservoir proposed by kahane et al .
( 1998 ) and jura & kahane ( 1999 ) , while those for a white dwarf are plausible only for the circumbinary reservoir . however , it is still difficult to choose between the two scenarios for the dust reservoir responsible for the silicate emission features in bm gem
. a very high angular resolution imaging in the ultraviolet would be interesting to depict the reservoir through the violet continuum emission scattered by the silicate grains .
if the silicates are located in a circumbinary reservoir , we will see a ring - like structure , while we may see a spiral structure if they are blown out continuously from a circumcompanion disk .
it is also important to carry out high sensitivity , medium resolution spectroscopy in the ultraviolet of silicate carbon stars other than bm gem to determine whether the violet spectral features are common among silicate carbon stars .
theoretical studies of the formation mechanism of the dust reservoir are also required .
the authors thank dr .
geoff clayton , the referee of this paper , for valuable comments and careful reading that helped improve the manuscript considerably .
the authors are also grateful to all the staff members of the subaru telescope .
h.i . and k.n
were supported by a grant - in - aid for scientific research ( c ) ( no.13640247 ) from the japan society for the promotion of science ( jsps ) .
m.y . were also partly supported by a jsps grant for scientific research ( a ) ( no.14204018 ) .
this research made use of simbad and vizier databases , maintained by cds ( strasbourg ) , france , and of the nasa / ipac infrared science archive , which is operated by the jet propulsion laboratory , california institute of technology , under contract with the national aeronautics and space administration .
abia , c. & isern , j. 2000 , , 536 , 438 alcolea , j. , bujarrabal , v. , & snchez contreras , c. 1996 , , 312 , 560 alksnis , a. , & zaime , d. 1993 , baltic astronomy , 2 , 281 alksnis , a. , balklavs , a. , dzervitis , u. , & eglitis , i. 1998 , , 338 , 209 allen , c. w. 1976 , astrophysical quantities ( 3rd ed . ;
london : the athlone press , univ . of london ) barnbaum , c. , morris , m. , likkle , l. , & kastner , j. h. 1991 , , 251 , 79 bohlin , r. c. , savage , b. d. , & drake , j. f. 1978 , , 224 , 132 beland , s. , boulade , o. , & davidge , t. 1988 , cfht info .
, 19 , 16 benson , p.j .
, & little - marenin , i. r. 1987 , , 316 , l37 bouigue , r. 1954 , ann .
, 17 , 104 cardelli , j. a. , clayton , g. c. , & mathis , j. s. 1989 , , 345 , 245 cernicharo , j. , gulin , m. , martin - pintado , j. , pealver , j. , & mauersberger , r. 1989 , , 222 , l1 chan , s. j , & kwok , s. 1988 , , 334 , 362 claussen , m. j. , kleinmann , s. g. , joyce , r. r. , & jura , m. 1987 , , 65 , 385 clayton , g. c. , lawson , w. a. , whitney , b. a. , & pollacco , d. l. 1993 , , 264 , l13 clayton , g. c. 1996 , , 108 , 225 clayton , g. c. , geballe , t. r. , & bianchi , l. 2003 , , 595 , 412 cox , a. n. , ed .
2000 , allen s astrophysical quantities ( 4th ed . ;
new york : aip ) dame , t. m. , hartmann , d. , & thaddeus , p. 2001
, , 547 , 792 deutsch , a. j. 1958 , , 63 , 49 dominy , j. f. 1985 , , 97 , 1104 engels , d. , & leinert , ch .
1994 , , 282 , 858 esa 1997 , the hipparcos and tycho catalogues ( esa sp-1200 ) & frank , a. 2004 , , 600 , 992 gilra , d. p. 1976
, memoires societe royale des sciences de liege , 9 , 77 groenewegen , m. a. t. , de jong , t. , van der bliek , n. s. , slijkhuis , s. , & willems , f. j. 1992 , , 253 , 150 harrington , j. p. , lutz , j. h. , & seaton , m. j. 1981 , , 21p hayes , d. s. , & latham , d. w. 1975 , , 197 , 593 heiles , c. 1975 , , 20 , 37 iben , i. , jr . 1975 , , 196 , 525 iben , i. , jr . , & renzini , a. 1983 , , 21 , 271 ireland , m. j. , monnier , j. d. , tuthill , p. g. , cohen , r. w. , de buizer , j. m. , packham , c. , ciardi , d. , hayward , t. , & lloyd , j. p. 2007
, , 662 , 651 iye , m. , karoji , h. , ando , h. et al .
2004 , , 56 , 381 izumiura , h. , ukita , n. , & tsuji , t. 1995 , , 440 , 728 jura , m. & helfand , d. j. 1984 , , 287 , 785 , 485 , 341 jura , m. & kahane , c. 1999 , , 521 , 302 joint iras science working group , 1986 , iras catalogues and atlasses , low resolution spectrograph ( lrs ) , , 65 , 607 joy , a. h. 1926 , , 63 , 281 joy , a. h. 1954 , , 1 , 39 kahane , c. , barnbaum , c. , uchida , k. , balm , s.p . , & jura , m. 1998 , , 500 , 466 kaifu , n. , usuda , t. , hayashi , s. s. et al .
2000 , , 52 , 1 karovska , m. , hack , w. , raymond , j. , & guinan , e. 1997 , , 482 , l175 knapp , g. j. , jorissen , a. , & young , k. 1997 , , 326 , 318 lambert , d.l . ,
hinkle , k. h. , & smith , v. v. 1990 , , 99 , 1612 lang , k. r. 1999 , astrophysical formulae , vol . 1 , ( berlin : springer ) little - marenin , i.r .
1986 , , 307 , l15 little - marenin , i. r. , benson , p. j. , & dickinson , d. f. 1988 , , 330 , 828 lloyd - evans , t. 1990 , , 243 , 336 lloyd - evans , t. 1991 , , 249 , 409 massey , p. , strobel , k. , barnes , j. v. , & anderson , e. 1988 , , 328 , 315 mastrodemos , n. & morris , m. 1998 , , 497 , 303 mastrodemos , n. & morris , m. 1999 , , 523 , 357 molster , f.j . , yamamura , i. , waters , l. b. f .m . , nyman , l .- a . , kaufl , h .- u . , de jong , t. , & loup , c. 2001 , , 366 , 923 moore , c. e. 1959 , a multiplet table of astrophysical interest , nbs technical note , ( washington , dc : us department of commerce ) morris , m. 1987 , , 99 , 1115 nakada , y. , izumiura , h. , onaka , t. , hashimoto , o. , ukita , n. , deguchi , s. , & tanabe , t. 1987 , , 323 , l77 nicolet , b. 1978 , , 34 , 1 noguchi , k. , kawara , k. , kobayashi , y. , okuda , h. , sato , s. , & oishi , m. 1981 , , 33 , 373 noguchi , k. , murakami , h. , matsuo , h. , noda , m. , hamada , h. , & watabe , t. 1990 , , 42 , 441 noguchi , k. , aoki , w. , kawanomoto , s. , ando , h. , honda , s. , izumiura , h. , kambe , e. , okita , k. , sadakane , k. , sato , b. , takada - hidai , m. , tanaka , w. , & watanabe , e. 2002 , , 54 , 855 ohnaka , k. & tsuji , t. 1999 , , 345 , 233 ohnaka , k. , driebe , t. , hofmann , k .- h .
, leinert , ch . ,
morel , s. , paresce , f. , preibisch , th . ,
richichi , a. , schertl , d. , sch@xmath137ller , m. , waters , l. b. f. m. , weigelt , g. , & wittkowski , m. 2006 , , 445 , 1015 paczyski , b. 1971 , , 9 , 183 pottasch , s. r. , gathier , r. , gilra , d. p. , & wesselius , p. r. 1981 , , 102 , 237 reimers , d. & cassatella , a. 1985 , , 297 , 275 rao , n. k. , lambert , d. l. , adams , m. t. , doss , d. r. , gonzalez , g. , hatzes , a. p. , james , c. r. , johns - krull , c. m. , luck , r. e. , pandey , g. , reinsh , k. , tomkin , j. , & woolf , v. m. 1999 , , 310 , 717 rosat consortium , 2000 , rosat news , 72 , 25-may-2000 ryde , n. & sch@xmath137ier , f. l. 2001 , , 547 , 384 schlegel , d. j. , finkbeiner , & d. g. , davis , m. 1998 , , 500 , 525 shane , c. d. 1928 , lick obs . bull . , 13 , 123 soker , n. & rappaport , s. 2000 , , 538 , 241 sugimoto , d. & nomoto , k. 1975 , , 27 , 193 tomov , t. , kolev , d. , zamanov , r. , georgiev , l. , & antov , a. 1990 , , 346 , 637 utsumi , k. 1985 , in _ cool stars with excesses of heavy elements , _ eds .
m. jaschek , & p. c. keenan ( dordrecht : reidel ) , 243 voges , w. , aschenbach , b. , boller , t. , braeuninger , h. , briel , u. , burkert , w. , dennerl , k. , englhauser , j. , gruber , r. , haberl , f. , hartner g. , hasinger , g. , kuerster , m. , pfeffermann , e. , pietsch , w. , predehl , p. , rosso , c. , schmitt , j. h. m. m. , truemper , j. , & zimmermann , h.u .
1999 , , 349 , 389 voges , w. , aschenbach , b. , boller , th . ,
brauninger , h. , briel , u. , burkert , w. , dennerl , k. , englhauser , j. , gruber , r. , haberl , f. , hartner , g. , hasinger , g. , pfeffermann , e. , pietsch , w. , predehl , p. , schmitt , j. , trumper , j. , & zimmermann , u. 2000b , iau circ .
7432 , 1 wallerstein , g. & knapp , g. r. 1998 , , 36 , 369 warner , b. 1972 , , 159 , 95 wasserburg , g. j. , boothroyd , a. i. , & sackmann , i .- j .
1995 , , 447 , l37 waters , l. b. f. m. , cami , j. , de jong , t. , molster , f. j. , van loon , j. th . ,
bouwman , j. , de koter , a. , waelkens , c. , van winckel , h. , & morris , p. w. 1998 , , 391 , 868 willems , f.j . &
de jong , t. 1986 , , 309 , l39 willems , f.j . & de jong , t. 1988 , , 196 , 173 wood , b.e . ,
karovska , m. , & hack , w. 2001 , , 556 , l51 wood , b.e . , karovska , m. , & raymond , j.c .
2002 , , 575 , 1057 yamamura , i. , dominik , c. , de jong , t. , waters , l.b.f.m . , & molster , f.j .
2000 , , 363 , 629 yamashita , y. & maehara , h. 1977 , , 29 , 319 lllrllr bm gem & 07 20 58.9 & + 25
00 07.28 & 3600 & 50,000 & 2001 jan 29 & @xmath415 + & & & 1800 & 50,000 & 2001 apr 14 & @xmath415 + v778 cyg & 20 36 07.4 & + 60 05 26.2 & 2400 & 38,000 & 2001 jul 30 & @xmath1601 + eu and & 23 19 58.2 & + 47 14 28 & 2400 & 38,000 & 2001 jul 30 & @xmath1601 + y cvn & 12 45 07.8 & + 45 26 24.9 & 900 & 50,000 & 2001 feb 1 & @xmath41 + feige 34 & 10 39 36.7 & + 43 06 09.3 & 300 & 9,000 & 2001 jan 28 & @xmath480 + lllllrrrrr bm gem & 1.83 & 1.24 & 10.72 & 8.40 & 4.96 & 4.28 & 3.32 & 2.75 & 2.47 + v778 cyg & & & 13.75 & 10.26 & & 5.25 & 4.20 & 3.54 & 3.18 + eu and & & & 13.70 & 10.47 & & 5.41 & 4.37 & 3.79 & 3.33 + y cvn & 4.59 & 0.73 & 8.41 & 5.42 & 1.25 & 0.59 & -0.37 & -0.89 & -1.42 + | balmer and paschen continuum emission as well as balmer series lines of p cygni - type profile from h@xmath0 through h@xmath1 are revealed in the violet spectra of bm gem , a carbon star associated with an oxygen - rich circumstellar shell ( `` silicate carbon star '' ) observed with the high dispersion spectrograph ( hds ) on the subaru telescope .
the blue - shifted absorption in the balmer lines indicates the presence of an outflow , the line of sight velocity of which is at least 400 km s@xmath2 , which is the highest outflow velocity observed to date in a carbon star .
the balmer lines showed a significant change in profile over a period of 75 days .
strong k emission was also detected , while h emission , where h@xmath3 overlapped , was absent on both observation occasions .
violet spectra of the other two silicate carbon stars , v778 cyg and eu and , and of the prototypical j - type carbon star , y cvn , were also observed , but none of these were detected in either continuum emission below 4000 or balmer lines .
we argue that the observed unusual features in bm gem are strong evidence for the presence of a companion , which should form an accretion disk that gives rise to both an ionized gas region and a high velocity , variable outflow .
the estimated luminosity of @xmath40.2 ( 0.03 - 0.6 ) @xmath5 for the ionized gas can be maintained by a mass accretion rate to a dwarf companion of @xmath6 yr@xmath2 , while @xmath7 yr@xmath2 is sufficient for accretion to a white dwarf companion .
these accretion rates are feasible for some detached binary configurations on the basis of the bond - hoyle type accretion process .
therefore , we concluded that the carbon star bm gem is in a detached binary system with a companion of low mass and low luminosity .
however , we are unable to determine whether this companion object is a dwarf or a white dwarf , although the gas outflow velocity of 400 km s@xmath2 as well as the non - detection in the x - ray survey favor its identity as a dwarf star .
the upper limits for binary separation are 210 au and 930 au for a dwarf and a white dwarf , respectively , in the case of circular orbit .
we also note that the observed features of bm gem mimic those of mira ( @xmath8 cet ) , which may suggest actual similarities in their binary configurations and circumstellar structures . |
monolayers of transition metal dichalcogenides ( tmdcs ) have attracted recently much attention due to their exceptional properties , such as coupling of spin and valley degrees of freedom , which allows for the valley polarization with a circularly polarized light in these materials @xcite .
recent magneto - photoluminescence experiments revealed the significant zeeman splitting of emission lines associated with optical recombination of excitons and trions in different valleys .
the effective exciton @xmath4-factors of this `` valley zeeman effect '' in a magnetic field directed normal to a monolayer plane were found to be @xmath6 for a wide range of investigated materials , including selenides @xcite , sulphides @xcite and tellurides @xcite .
the multi band @xmath0 method is perfectly suited for theoretical investigation of magneto - optical and transport effects , including zeeman effect , optical absorption and photogalvanics @xcite .
the available two - band @xmath0 models describing electronic spectra in tmdcs @xcite account for the bottom conduction and topmost valence bands and are parameterized by density functional theory ( dft ) calculations .
the simple two - band model is , however , insufficient for calculation of exciton zeeman effect since the exciton @xmath4-factor @xmath5 vanishes in the two - band @xmath0 approximation [ ] .
therefore to obtain nonzero @xmath5 additional terms describing contributions of remote bands should be included in the two - band @xmath0 model @xcite , which make the model less transparent .
moreover parametrization of the @xmath0 model by fitting the dft band structure only in the vicinity of distinct points of the brillouin zone is less reliable and allows for a much freedom in the choice of parameters . in this work
we use an alternative approach based on the idea proposed in ref . [ ] .
this approach comprises three steps , namely , ( i ) use dft - based calculations of electronic states in tmdcs as a starting point , ( ii ) use atomistic tight - binding ( tb ) model to fit the electronic spectrum and wave functions along the high - symmetry paths of the brillouin zone , and ( iii ) derive @xmath0 hamiltonian by decomposing the tb model in the vicinity of a given wave vector in the brillouin zone . as a result
, we obtain a multi band @xmath0 hamiltonian ( the number of bands is equal to the number of atomic orbitals included in the tb model ) with a set of parameters that are uniquely determined by the tb parametrization .
this hamiltonian can be further used for calculating the valley zeeman effect , landau levels , etc . as a starting point
we use several existing eleven - band tb models @xcite .
these models include @xmath9-type orbitals of metal atoms and @xmath10-type orbitals of chalcogen atoms , and capture all symmetries of the studied system .
the resulting @xmath0 hamiltonian , which describes dispersion of the bottom conduction and topmost valence bands , contains six bands that have even parity with respect to mirror reflection in the monolayer plane .
we present a systematic parametrization of the six - band @xmath0 hamiltonian by expansion of tb hamiltonians @xcite in the vicinity of @xmath2 points of the brillouin zone for different materials ( mos@xmath3 , ws@xmath3 , mose@xmath3 , wse@xmath3 ) and different tb models .
we also do the whole three - step procedure on our own , i.e. , perform post - dft calculations by applying @xmath7 formalism to obtain reliable band gaps and accurate band dispersion , fit it with the tb model and do the @xmath0 decomposition , which allows us , as a result , to obtain an independent @xmath0 parameterization .
we want to stress the novelty of using @xmath7 set of data to extract tb parameters since in previous studies the main focus was made on dft calculations using mainly local or semi - local exchange - correlation functionals @xcite , whereas hybrid functional was also used @xcite to partially correct the severe underestimation of band gap values usually observed at the dft level for mos@xmath3 monolayers , see ref .
[ ] for a recent review .
the use of @xmath7 correction strongly affects the effective charge carrier masses @xcite too .
then , we use the derived six - band @xmath0 hamiltonian to calculate effective masses , landau levels , and the effective exciton @xmath4-factor @xmath5 in different tmdcs . we analyze the main contributions to @xmath5 , which result from the mixing with excited conduction and deep valence bands .
we show that the calculated values of @xmath5 and effective masses vary in a wide range for different tb parameterizations .
the calculated values of @xmath5 for available in literature tb models ( @xmath11 ) are rather far from experimental values .
however using our dft+@xmath7 calculations and the tb fitting procedure we were able to find additional @xmath0 parameterization sets , which well describe the experimental exciton @xmath4-factor ( @xmath6 ) , as well as conduction and valence band effective masses and the wave functions coefficients .
this result underlies the importance of the careful choice for the dft starting point calculations and tb parameterizations : along with effective masses and energy gaps the @xmath4-factor value may serve as a test for improving both the dft calculations and tb models .
as a starting point for construction of an effective @xmath0-hamiltonian we will use eleven - band tight - binding models developed in refs . .
these tight - binding models include three @xmath10-type orbitals on each of the two chalcogen atoms ( x ) and five @xmath9-type orbitals on a metal atom ( m ) .
the electron wave function within the tight - binding approximation is presented as a linear combination of atomic orbitals @xmath12 @xcite @xmath13 where @xmath14 is a number of an electronic band , @xmath15 is a wave vector , @xmath16 denotes the type of an atom , @xmath17 runs through the atoms of a given type , @xmath18 enumerates the set of orbitals , @xmath19 gives the position of atoms in a two - dimensional lattice , and @xmath20 are coefficients .
the basis orbitals of the eleven - band tight - binding model are @xcite @xmath21 where @xmath22 , @xmath23 , @xmath24 denotes the orbital with a @xmath9-like symmetry of the m atom and @xmath25 , @xmath26 are the symmetric and asymmetric combinations of the @xmath10-type orbitals of the top ( @xmath27 ) and bottom ( @xmath28 ) x atoms in the unit cell , @xmath29 and @xmath30 axes lie in the monolayer plane , and @xmath31 is the monolayer normal . the spin orbit interaction between electron spin and orbital momenta of atomic orbitals @xcite
is neglected in this work .
if the @xmath32 mirror symmetry is conserved , i.e. for a free - standing monolayer in the absence of external electric field and strain , the hamiltonian , which describes the energy spectrum of a monolayer electron with a wave vector @xmath15 , written in the basis eq .
has the form @xmath33 here @xmath34 is the 6@xmath356 block acting on the orbitals with even with respect to @xmath32 symmetry , and @xmath36 is the 5@xmath355 block acting on the orbitals with odd symmetry .
the exact form of blocks @xmath34 and @xmath36 depends on a particular tight - binding model @xcite .
it is known that the bloch functions of the bottom conduction and topmost valence bands , which are of the main interest in this work , are even with respect to @xmath32 reflection @xcite , and therefore these bands are described by the @xmath34 block .
we note that magnetic field normal to a monolayer does not break the parity of wave functions , and hence we do not need the @xmath36 block in the calculation of @xmath5 .
we also note , that the mixing of the @xmath34 and @xmath36 blocks by a perturbation that breaks @xmath32 symmetry does not affect @xmath5 in the first order in this perturbation . to construct an effective @xmath0 hamiltonian in the vicinity of @xmath37 points , where @xmath38 is the lattice constant
, we will decompose the tight - binding hamiltonian @xmath39 over a small wave vector @xmath40 . up to the second - order terms
this decomposition yields @xmath41 electron wave functions at @xmath2 valleys of mx@xmath3 transform according to irreducible representations ( irreps ) of the c@xmath42 point group .
we denote six wave functions that diagonalize @xmath43 at @xmath44 as @xmath45 , @xmath46 , @xmath47 , @xmath48 , @xmath49 , @xmath50 , where a superscript names the electronic band and a subscript denotes the corresponding irreducible representation ( @xmath51 , @xmath52 and @xmath53 ) , see tab .
[ tab : table1 ] .
note that in addition to two conduction ( @xmath54 and @xmath55 ) and two valence ( @xmath56 and @xmath57 ) bands , known from the four - band @xmath0 models @xcite , the six - band model contains two deep valence bands @xmath58 and @xmath59 , which transform at @xmath60-point via @xmath51 and @xmath52 representations , respectively . in the new basis @xmath61 can be written as @xmath62 where the first term contains linear in @xmath63 terms : @xmath64 and the second one @xmath65_{nl } = \frac{\hbar^2 q^2}{2 m'_{n } } \delta_{nl}\:,\:\ : n , l = 1 .. 6\ ] ] is a diagonal matrix with quadratic in @xmath63 elements . in eqs . , @xmath66 ( @xmath67 ) are the band energies at @xmath2-points , @xmath68 and @xmath69 are parameters , @xmath70 , and @xmath71
. the effective masses @xmath72 describe contributions to the band dispersion arising from the mixing with remote bands , which are not present in the @xmath0 model @xcite .
note that in the decomposition of the off - diagonal elements of the tight - binding hamiltonian we retained only the linear in @xmath63 terms .
the phases of wave functions at @xmath60-point ( tab .
[ tab : table1 ] ) are chosen in such a way that parameters @xmath68 and @xmath69 are real .
parameters of the @xmath0 hamiltonians and for different parametrizations , materials and tight - binding models are listed in tab .
[ tab : table3 ] of the appendix [ app : kp ] . to derive the @xmath0 hamiltonian @xmath73 at the @xmath74-point of the brillouin zone one
should replace @xmath75 by @xmath76 and vice versa in eq . ] .
note that at @xmath77 basis wave functions @xmath78 and @xmath79 transform according to @xmath53 irreducible representation whereas @xmath80 and @xmath81 transform according to @xmath52 .
.[tab : table1 ] the nonzero coefficients @xmath82 of wave functions
eq . at @xmath60-point and corresponding irreducible representations of the @xmath83 point group .
the phases of the wave functions are chosen in such a way , that @xmath84 and @xmath85 are real numbers , @xmath86 , and @xmath87 .
we denote @xmath88 . [ cols="^,^,^",options="header " , ] dgt+gw and tb calculations for electronic energy dispersion in mos@xmath3 .
( a ) the fit of tb model of ref .
[ ] , ( b ) the fit of tb model of ref .
the zero energy was aligned to the top of the valence band @xmath56 .
, scaledwidth=47.0% ] 43ifxundefined [ 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 http://link.aps.org/doi/10.1103/physrevlett.105.136805 [ * * , ( ) ] http://dx.doi.org/10.1038/nnano.2012.193 [ * * , ( ) ] http://link.aps.org/doi/10.1103/physrevlett.108.196802 [ * * , ( ) ] http://dx.doi.org/10.1038/nnano.2012.96 [ * * , ( ) ] http://dx.doi.org/10.1038/ncomms1882 [ * * , ( ) ] link:\doibase 10.1103/physrevb.86.081301 [ * * , ( ) ] http://link.aps.org/doi/10.1103/physrevlett.113.266804 [ * * , ( ) ] http://link.aps.org/doi/10.1103/physrevlett.114.037401 [ * * , ( ) ] http://dx.doi.org/10.1038/nphys3203 [ * * , ( ) ] http://dx.doi.org/10.1038/nphys3201 [ * * , ( ) ] http://stacks.iop.org/2053-1583/2/i=3/a=034002 [ * * , ( ) ] http://dx.doi.org/10.1021/acs.nanolett.5b00626 [ * * , ( ) ] http://dx.doi.org/10.1038/ncomms10643 [ * * ( ) ] link:\doibase 10.1103/physrevb.93.165412 [ * * , ( ) ] http://dx.doi.org/10.1021/acs.nanolett.6b00748 [ * * , ( ) ] @noop _ _ ( , ) @noop _ _ ( , ) http://stacks.iop.org/2053-1583/2/i=2/a=022001 [ * * , ( ) ] link:\doibase 10.1103/physrevx.4.011034 [ * * , ( ) ] link:\doibase 10.1103/physrevb.88.045416 [ * * , ( ) ] link:\doibase 10.1103/physrevb.88.085440 [ * * , ( ) ] link:\doibase 10.1103/physrevb.88.075409 [ * * , ( ) ] http://stacks.iop.org/2053-1583/1/i=3/a=034003 [ * * , ( ) ] http://link.aps.org/doi/10.1103/physrevb.92.205108 [ * * , ( ) ] http://stacks.iop.org/0953-8984/27/i=36/a=365501 [ * * , ( ) ] http://link.aps.org/doi/10.1103/physrevb.92.195402 [ * * , ( ) ] http://www.sciencedirect.com/science/article/pii/s016757291500028x [ * * , ( ) ] http://journals.aps.org/prb/abstract/10.1103/physrevb.88.045412 [ * * , ( ) ] http://journals.aps.org/prb/abstract/10.1103/physrevb.87.155304 [ * * , ( ) ] http://dx.doi.org/10.1002/pssb.201552211 [ * * , ( ) ] @noop * * , ( ) , link:\doibase 10.1103/physrevlett.113.026803 [ * * , ( ) ] http://link.aps.org/doi/10.1103/physrevlett.113.076802 [ * * , ( ) ] http://scitation.aip.org/content/aip/journal/apl/108/14/10.1063/1.4945268 [ * * , ( ) ] @noop _ _ ( , ) http://journals.aps.org/prb/abstract/10.1103/physrevb.47.558 [ * * , ( ) ] http://journals.aps.org/prb/abstract/10.1103/physrevb.54.11169 [ * * , ( ) ] http://journals.aps.org/prl/abstract/10.1103/physrevlett.77.3865 [ * * , ( ) ] link:\doibase 10.1103/physrevb.50.17953 [ * * , ( ) ] http://journals.aps.org/prb/abstract/10.1103/physrevb.59.1758 [ * * , ( ) ] http://journals.aps.org/prb/abstract/10.1103/physrevb.88.245309 [ * * , ( ) ] http://journals.aps.org/prb/abstract/10.1103/physrevb.74.035101 [ * * , ( ) ] | we developed a six - band @xmath0 model that describes the electronic states of monolayer transition metal dichalcogenides ( tmdcs ) in @xmath1-valleys .
the set of parameters for the @xmath0 model is uniquely determined by decomposing tight - binding ( tb ) models in the vicinity of @xmath2-points .
first , we used tb models existing in literature to derive systematic parametrizations for different materials , including mos@xmath3 , ws@xmath3 , mose@xmath3 and wse@xmath3 .
then , by using the derived six - band @xmath0 hamiltonian we calculated effective masses , landau levels , and the effective exciton @xmath4-factor @xmath5 in different tmdcs .
we showed that tb parameterizations existing in literature result in small absolute values of @xmath5 , which are far from the experimentally measured @xmath6 . to further investigate this issue we derived two additional sets of @xmath0 parameters by developing our own tb parameterizations based on simultaneous fitting of ab - initio calculated , within the density functional ( dft ) and @xmath7 approaches , energy dispersion and the value of @xmath5 .
we showed that the change in tb parameters , which only slightly affects the dispersion of higher conduction and deep valence bands , may result in a significant increase of @xmath8 , yielding close - to - experiment values of @xmath5 .
such a high parameter sensitivity of @xmath5 opens a way to further improvement of dft and tb models . |
hard - sphere ( hs ) particles provide a basic structural model for a wide variety of particulate systems ranging from the macroscale ( e.g. , packings of ball bearings ) to the microscale ( e.g. , colloidal suspensions or metallic glasses ) @xcite . at the microscale ,
thermal motion is the equilibrium configuration sampling mechanism , whereas randomly applied external perturbations ( e.g. , vibrations ) can play an analogous role for macroscale systems .
what is common across scales is the fundamental role of excluded volume due to the hard cores , which intrinsically limits the available configurational space of the system via strict geometric constraints .
while such constraints have been extensively studied to gain insights into jamming phenomena @xcite and the glass transition @xcite , there is still much to be learned regarding the extent to which the hard - core interaction limits the possible structural arrangements of the particles and hence the properties of fluid systems .
perhaps the most important measure of structure for homogeneous fluids is the radial distribution function , @xmath1 .
the product @xmath2 quantifies the expected number of particle centers to be found in a shell of differential width @xmath3 a distance @xmath4 away from a given particle center in a fluid of number density @xmath5 .
the core prevents particles from approaching closer than a particle diameter @xmath6 ; i.e. , @xmath7 for @xmath8 . in dense fluids ,
packing constraints provided by the cores also give rise to a decaying oscillatory structure in @xmath1 for @xmath9 , signifying the coordination shells that naturally form around the particles .
the coordination - shell structure quantified by the radial distribution function is critical for the thermodynamic properties ( e.g. , the equation of state ) of the fluid , but it can influence transport properties as well @xcite .
one question of fundamental interest and potential practical importance for materials design @xcite is how much ` pair disorder ' can be accommodated by a fluid of particles with hard cores of diameter @xmath6 at a given packing fraction @xmath10 ?
in other words , to what extent is it possible to suppress the core - induced oscillations in @xmath1 , e.g. , via incorporation of auxiliary pair interactions ?
a particularly suitable metric for pair disorder in equilibrium systems , and the one adopted in this paper , is the two - body contribution that emerges in the multi - particle expansion of the excess ( i.e. , configurational , over ideal gas ) entropy , @xmath11 @xcite . for a fluid of @xmath12-dimensional particles with hard - core diameter @xmath6 and isotropic pair interactions
, this quantity can be expressed in _ dimensionless _
form , where boltzmann s constant , @xmath13 , is absorbed in the definition , as @xmath14 \end{split}\]]in equation [ eqn : pair_excess_entropy ] , the first term in the sum , @xmath15 , is trivially due to the excluded volume of the hard core , while the second term accounts for any coordination - shell structure present in @xmath1 for @xmath9 . in seeking to find systems that maximize @xmath11 , it is important to ensure that known requirements for realizability are met . here , we follow convention and require the static structure factor @xmath16 , related to @xmath1 via the fourier transform ( ft ) @xmath17 ft@xmath18 $ ] , to be positive definite for all @xmath19 .
thus , we formulate the two - body disorder maximization problem of interest as @xmath20 \\ & g(r)=0 , \
\ r\leq d \\ & s(k)\geq 0 , \ \ 0\leq k < \infty \end{split}\]]for packing fractions @xmath21 , the analytical solution of equation [ eqn : positive_definite ] is simply the unit heaviside step function @xmath22 , i.e. , identical to the limiting radial distribution of the hs fluid as @xmath23 , with @xmath24 . in other words , a radial distribution function without _
any _ oscillatory coordination - shell structure appears realizable for fluids of particles with hard cores under these conditions . precisely at @xmath25 , the solution @xmath22 leads to hyperuniform structure ( long ranged density fluctuations are suppressed , @xmath26 @xcite ) and thus the system is incompressible . for packing fractions @xmath27
, the step function solution would violate the realizability constraint , i.e. , exhibiting @xmath28 for some range of @xmath19 @xcite .
hence , the maximally disordered pair structure for @xmath27 must generally be obtained via a numerical solution of equation [ eqn : positive_definite ] . the low - density , analytical solution to the @xmath11 maximization problem of equation
[ eqn : positive_definite ] described above also provides information pertaining to a so - called iso-@xmath0 [ i.e. , iso-@xmath29 process @xcite .
an iso-@xmath0 process is one where an equilibrium system at an initial packing fraction , @xmath30 ( in the above example , the hs fluid at @xmath31 ) , can realize an identical radial distribution function at a higher packing fraction @xmath10 by adding a @xmath10-dependent auxiliary pairwise interaction which counteracts any pairwise structural evolution that would have otherwise occurred . for a given iso-@xmath0 process
, there is also a singular packing fraction , @xmath32 ( in the above example , @xmath33 ) , above which the pair correlation of the original state can no longer be realized because it would unphysically result in @xmath34 for some values of @xmath19 .
irrespective of the chosen value for @xmath30 , iso-@xmath0 processes initiating from an equilibrium hs structure ( 1 ) necessarily grow in auxiliary , soft pair interactions external to the hard core at @xmath35 to prevent pair structure evolution and ( 2 ) terminate at an upper singular packing fraction @xmath36 at which @xmath37 , corresponding to an incompressible , hyperuniform state .
since the pair structure of such an iso-@xmath0 process is , by definition , identical that of the hs fluid at the lower initial packing fraction ( @xmath38 ) , its pair disorder ( e.g. , as measured by @xmath11 ) at @xmath32 is always higher than that of an equi - density equilibrium hs fluid , but less than or equal to that obtained by a formal maximization of @xmath11 via equation [ eqn : positive_definite ] .
since an iso-@xmath0 process is technically simpler to analyze than a full @xmath39 maximization , the former is a helpful construct for studying how the enhancement of two - body disorder of a hard - core fluid affects the total _
excess entropy , @xmath40 ( inclusive of all many - body correlation contributions ) and other related properties .
navely , one might anticipate that muting pair structure would also increase the overall structural disorder of the system , which would conceptually align with factorization approximations in equilibrium statistical mechanics including the kirkwood superposition approximation that estimates the three - body distribution according to : @xmath41 @xcite . however , upon further examination , this logic directly contradicts expectations based on thermodynamics . for example , the auxiliary interactions that are added to the hard - core potential to preserve structure in an iso-@xmath0 process , as discussed more formally in section [ results : step_isog ] , must necessarily cause a _ decrease _ in @xmath40 compared to the hs reference state ( despite the relative increase in @xmath11 ) .
the magnitude of this decrease , which is due to corresponding ( and more than compensating ) enhanced structure in higher - body static correlations compared to the hs reference fluid , has not yet been explored and is one focal point of this work .
probing the corresponding changes of @xmath40 and @xmath39 along iso-@xmath0 processes is also of interest given the empirically observed links between these quantities and dynamic properties of fluids @xcite . for many fluid systems , changes that increase excess entropy ( or its two - body contribution )
result in faster dynamical relaxation processes and vice - versa , even for systems that exhibit `` anomalous '' trends in dynamics @xcite with respect to quantities like number density .
while @xmath40 and @xmath39 are typically strongly correlated ( the latter is usually the dominant contribution to the former ) , they would yield conflicting dynamical predictions as a function of @xmath10 for iso-@xmath0 processes starting from a hs state at @xmath42 . from the perspective of @xmath39 ,
an iso-@xmath0 process should produce enhanced dynamics relative to the hs fluid at the same packing fraction .
this at odds with expectations based on behavior of @xmath40 , a quantity which is _ necessarily _ lowered relative to the hard sphere - fluid via an iso-@xmath0 process . while the magnitude of the @xmath40 reduction along an iso-@xmath0 path is unknown , a large decrease ( and implied dynamical slowdown ) might seem a reasonable expectation given the tendency toward incompressibility of an iso-@xmath0 fluid upon approach to the corresponding singular packing fraction @xmath32a topic we address in this paper .
along similar lines , strong correlations between dynamical relaxation times and the excess compressibility factor have been observed in simulations of fluids with hard cores ( see , e.g. , @xcite ) .
whether such correlations also hold for iso-@xmath0 processes , where they would predict a pronounced dynamic slowdown approaching @xmath32 , remains an open question and a stringent test of their generality .
also critical to understanding the static and dynamic properties displayed by systems of an iso-@xmath0 process are the @xmath10- ( and @xmath42- ) dependent auxiliary interactions which must be incorporated to keep the pair structure constant as packing fraction is increased .
these interactions are known to be of a repulsive yukawa form at large inter - particle separations and strongly believed , in accordance with predictions from approximate integral equation theories , to be of a simple repulsive ramp - like form near contact for the step - function hs iso-@xmath0 process @xcite .
interestingly , a ramp - like interparticle potential was also recently shown to remove the coordination - shell structure of a tracer solute in a hs solvent @xcite , and a ramp - like fluid - wall interaction is known to similarly ` flatten ' the density profile of a confined hs fluid @xcite .
to clarify the above issues , we seek to address the general accuracy of integral equation theory predictions for an iso-@xmath0 process , as tested by molecular dynamics ( md ) simulations , and confirm the possibly generic ramp - like interparticle potential form that apparently emerges for iso-@xmath0 processes that initiate from hs fluid reference states .
moreover , while iso-@xmath0 processes provide a simple means by which to find interactions that increase @xmath39 ( relative to the hs fluid at the same packing fraction ) and hence help to explore the ensuing thermodynamic , dynamic , and potential interaction consequences , we also want to examine the impact of formally maximizing two - body disorder at arbitrary density ( see equation [ eqn : positive_definite ] ) . here
, we tackle this problem via numerical optimizations performed with a genetic algorithm to reveal the absolute limits of two - body disorder in hard - core fluids . of key interest
is the residual structure that remains after @xmath11 maximization and the potentially unique changes in properties that the corresponding fluids display .
in particular , we demonstrate a unique correspondence between minimizing pair disorder and suppressing structure not only @xmath43 in @xmath44 , but a continuous range in @xmath19 wavevectors .
systems with a range of complete suppression in @xmath44 are termed `` stealthy '' since they are transparent ( do not scatter ) over a continuum of @xmath19 .
the balance of this paper is organized into two primary sections .
section [ methods ] describes our liquid - state theoretical calculations , md simulations , and numerical optimizations .
section [ results ] outlines our thermodynamic , dynamic and auxiliary potential predictions for an array of iso-@xmath0 processes while also discussing our @xmath39 maximization results . section [ conclusions ] concludes the paper and discusses some interesting future directions .
we also provide a review in appendix a of iso-@xmath0 processes and the corresponding mathematical framework for their analysis ( used implicitly and referenced throughout the text ) .
integral equation theory ( iet ) is a powerful tool of equilibrium statistical mechanics that is most frequently used for the forward prediction of @xmath1 and related thermodynamic properties based on knowledge of the dimensionless pair potential between particles , @xmath45 @xcite [ implicitly contains the inverse thermal energy @xmath46 where @xmath47 is temperature and @xmath13 is boltzmann s constant ] .
however , iet can also be used for the inverse problem as studied for three - dimensional fluid systems in this paper of deriving the necessary @xmath48 to form a system exhibiting a desired target @xmath1 .
iet employs a formally exact partitioning of the total correlation function , @xmath49 , into direct and multibody contributions via the direct correlation function , @xmath50 , and the ornstein - zernike ( oz ) relation @xmath51 the oz relation can be used together with approximate closures relating @xmath48 , @xmath1 and @xmath50 .
three popular closures are explored in this paper , namely , the hypernetted chain ( hnc ) @xcite , percus - yevick ( py ) @xcite , and martinov - sarkisov ( ms ) @xcite approximations @xmath52-c(r),\ \
\ \ \text{py}\]]@xmath54}-\text{ln}[g(r)]-1,\ \ \ \text{ms}\]]we also consider the random phase approximation ( rpa ) @xcite @xmath55where @xmath56 is an assumed known ( here , hard - sphere ) direct correlation function . to determine @xmath56
, we use the hs structure predicted from a forward iet calculation with the modified verlet ( mv ) closure @xcite @xmath57^{2}}{2+(8/5)[h_{0}(r)-c_{0}(r ) ] } \bigg],\ \ \ \text{mv } \end{split}\]]combining equation [ eqn : oz ] with one of equations [ eqn : hnc_closure]-[eqn : rpa_closure ] , either @xmath1 or @xmath48 can be obtained given knowledge of the other .
for iet , the inverse problem is technically simpler than the forward problem as numerical solution is not necessary for most simple closures like equations [ eqn : hnc_closure]-[eqn : ms_closure ] .
instead , solution follows by inputting @xmath58 , a ft of the known @xmath59 , into eqn .
[ eqn : oz ] and then solving for @xmath60 @xmath61which , after an inverse fourier transform , yields the direct correlation function via @xmath62 . provided the @xmath50 corresponding to the target @xmath1 , eqns .
[ eqn : hnc_closure]-[eqn : rpa_closure ] can be used to determine the corresponding @xmath48 .
additionally , within the hnc and rpa approximations , the _ dimensionless _ excess helmholtz free energy per particle @xmath63 can be calculated using @xcite @xmath64 \\ & + \dfrac{1}{4\pi^{2}\rho}\int_{0}^{\infty}dkk^{2}\big[\rho \hat{c}(k)+\text{ln}[1-\rho \hat{c}(k)]\big ] , \ \
\text{hnc } \end{split}\]]and @xmath65\big ] , \ \
\text{rpa } \end{split}\]]respectively , where @xmath66 is the hs reference , dimensionless excess helmholtz free energy per particle ( we employ the accurate carnahan - starling form @xcite ) .
free energy calculations provide @xmath40 via @xmath67 where @xmath68 is the dimensionless excess ( potential ) internal energy per particle .
for selected states , we conduct md simulations of monodisperse particles governed by pair potentials @xmath69 . here , @xmath70 is a normalized , steeply repulsive weeks - chandler - andersen ( wca ) interaction representing a hard - core exclusion volume @xcite , which is given by @xmath71^{48 } - [ d / r]^{24 } ) + 1 $ ] for @xmath72 and @xmath73 for @xmath74 .
the remaining term , @xmath75 , represents the density - dependent `` flattening '' potentials derived via iet to actualize iso-@xmath0 processes , where this term is set to @xmath76 when approximating hs fluids .
when simulating systems with non - zero @xmath75 , we choose the cutoff distance @xmath77 for @xmath48 such that the force at this distance @xmath78 . using these potentials , we perform three - dimensional simulations in the canonical ensemble with periodic boundary conditions using gromacs 4.6.5 @xcite .
we use an integration time - step of @xmath79 and fix temperature via a nos - hoover thermostat with time - constant @xmath80 . for hs systems ,
we simulate @xmath81 particles ; for systems governed by non - zero @xmath75 , we simulate between @xmath82 particles , where @xmath83 is proportional to the potential cutoff distance @xmath77 , which for our state points of interest ranges between @xmath84 .
to characterize pair correlations , we calculate the radial distribution function @xmath1 in the usual way from particle configurations ; subsequently , we calculate @xmath44 using the identity @xmath85 $ ] .
to characterize dynamics , we calculate the self - intermediate scattering function @xmath86\}}\rangle$ ] , where @xmath87 is the instantaneous position of particle @xmath88 at some lag - time @xmath89 relative to its position at @xmath90 and @xmath91 represents an ensemble average .
given the systems are isotropic , we average statistics across all three spatial directions to obtain @xmath92 .
when calculating this quantity , we take care to set up simulations with box lengths @xmath93 appropriate for obtaining @xmath94 , where examining this wavevector magnitude ( corresponding to a real - space distance of @xmath6 ) characterizes the dynamic decay of correlations due to single - particle diffusion . as described in the introduction , equation [ eqn : positive_definite ] has an analytical solution for @xmath95 in @xmath96 .
the radial distribution function with @xmath97 for all @xmath19 that maximizes @xmath11 under these conditions is the unit step function @xmath22 . for @xmath98
, we used a genetic algorithm @xcite to numerically solve equation [ eqn : positive_definite ] . in order to avoid restriction to any particular functional form , we directly optimized the radial distribution function at 50 different values of @xmath99 , interpolating between the points via a cubic spline to construct the full correlation function . based on preliminary optimizations ,
the knots for the spline were spaced unevenly ( weighted towards smaller values of @xmath99 ) , giving more flexibility to the @xmath1 near contact . to enforce the hard - core constraint , @xmath1 for @xmath100
was fixed to be zero .
additionally , @xmath1 for @xmath101 was set to unity . between these two bounds ,
@xmath1 was optimized , and the associated @xmath44 was calculated using the fftw program @xcite . if @xmath44 was not positive definite , then the minimum value of @xmath44 , weighted by a positive empirical factor ( @xmath102 ) , was added as a penalty to the figure of merit , and optimization cycles proceeded with an increasing @xmath102 until the constraint was satisfied .
the optimization was performed with a micro - genetic algorithm with a population size of 5 using tournament selection , 50% uniform crossover and 50% single point crossover , a mutation rate of 5% , elitism , binary encoding , and each cycle comprised 50,000 steps .
@xcite in order to restrict parameter space such that a solution could be found efficiently , the randomly generated @xmath1 was smoothed once with a single triangular smoothing function . as a practical consideration
, a genetic algorithm requires discretization of the solution space . as a consequence ,
several solutions exist that are very close in @xmath39 .
therefore , 10 independent optimizations were carried out at each density and level of discretization .
optimizations were performed iteratively , where the allowed precision of the @xmath1 values was incrementally increased as the range of values allowed in the optimization were concurrently decreased as determined from the spread of the data in the previous step .
if this range was exceeded in a subsequent optimization , the range was widened in the previous step until all optimizations fell inside the predetermined range .
once a discretization in the @xmath1 of @xmath103 was achieved , the 10 solutions were averaged to arrive at the final optimized @xmath1 .
, required for the iso-@xmath0 process initiated from the hs fluid at @xmath31 , as function of packing fraction @xmath10 .
the solid yellow line represents the exact result from eqn .
[ eqn : exact_step_cr ] and the shaded region denotes the upper and lower bound approximations derived in appendix a. ( b ) representative soft auxiliary potentials near the singular packing fraction @xmath104 predicted using iet and ( from bottom to top ) py , ms , rpa , and hnc closures as well as the asymptotic yukawa form [ derived in appendix a ] .
( c ) hnc predictions for full ( lower blue curves ) and two - body ( upper red curves ) excess entropy for this iso - g process [ solid lines ] as compared to the unperturbed hard sphere fluid [ dashed lines ] . circles and diamonds are simulation results from @xcite and @xcite respectively . ]
we begin our discussion by considering three - dimensional fluids at relatively low packing fractions ( @xmath105 ) , where as mentioned in the introduction the radial distribution function that maximizes @xmath11 subject to a hard - core constraint corresponds to that of the hs fluid at vanishing density [ i.e. , a step function , @xmath22 ] .
the range of equilibrium fluid states that can exhibit this ideal pair structure via the addition of auxiliary pair interactions to the hard - core potential corresponds to an iso-@xmath0 process ( initiating from a hs fluid with density @xmath107 ) that terminates at the singular density @xmath108 , at which @xmath26 .
we examine this relatively simple case to address the following questions , with a focus on behavior at @xmath36 , where the most pronounced hs structure is suppressed .
what is the nature of the auxiliary interactions required to eliminate the hard - core pair correlations for @xmath109 ?
how are thermodynamic quantities , including the two - body and total excess entropy , affected by these changes ? are incompressible states that must exist at @xmath36 dynamically slow or jammed in any sense ? as shown in fig .
1 , preventing pair correlation evolution as the the singular density is approached requires pronounced growth in the lengthscale of the auxiliary potential , which modifies system thermodynamics relative to that of bare hard spheres . first ,
as outlined in appendix a , the large-@xmath99 asymptotic portion of the auxiliary potential [ @xmath110 becomes long ranged [ @xmath111 in a power law fashion [ @xmath112 , serving as a sort of `` inverted '' critical phenomenon @xcite . here , the potential range can be calculated exactly from the imaginary portion , @xmath113 , of the @xmath43 complex root in the denominator of @xmath114 for the unit step @xmath1 @xmath115for which the solution ( yellow line ) is shown in fig .
[ sch : auxiliary_range]a . also shown in fig .
[ sch : auxiliary_range]a is a shaded region denoting the space between two approximations derived in appendix a ( equations a3 and a4 ) .
these generally applicable analytical forms ( 1 ) serve as useful upper and lower bounds and ( 2 ) indicate the onset of `` inverted '' critical - like behavior by their collapse .
collapse is realized once the potential range grows larger than the particle diameter , an intuitive result given that @xmath6 is the only relevant length scale to exceed .
some iet auxiliary potential predictions within this critical limit are presented in fig .
[ sch : auxiliary_range]b along with the asymptotic contribution ( equation [ eqn : yukawa_k0 ] ) .
while all four theories correctly merge to the asymptotic result as @xmath116 ( not shown ) and display a ramp - like form near contact , notable quantitative differences are apparent .
the accuracy of each theory is tested later in this section , but universal amongst each is the predicted divergent excess internal energy , @xmath68 , that arises upon approach to @xmath36 .
more precisely , @xmath117 .
the helmholtz energy @xmath63 also diverges upon approach to @xmath36 but , as we show below , @xmath67 remains finite implying an exact cancellation .
ultimately , it is the divergence in @xmath118 , and not entropic factors contained within @xmath63 , that drives the vanishing of the compressibility at @xmath36 . along an iso-@xmath0 process initiating from a hs fluid state at @xmath42 ,
the ` pair disorder ' [ characterized by @xmath39 ] progressively increases with @xmath10 relative to that of the hs fluid at the same packing fraction .
but , what are the consequences for the overall configurational disorder [ quantified via @xmath40 ] ?
this can be understood by beginning with the rigorous gibbs - bogoliubov ( gb ) inequality @xcite @xmath121where @xmath66 is the dimensionless , excess helmholtz free energy per particle of the hs fluid , @xmath1 is the radial distribution function of the iso-@xmath0 process with @xmath5-dependent auxiliary potential @xmath45 , and @xmath122 is the radial distribution function of the hs fluid . using the relations , @xmath123 , @xmath68 and @xmath124 in equation [ eqn : gb_inequality_i ] yields @xmath125 u(r ) \leq s \leq s_{0}\]]proving that @xmath40 necessarily decreases along an iso-@xmath0 process ( relative to the equidensity hs fluid ) .
this result also makes intuitive sense , given that one applies auxiliary interactions in the iso-@xmath0 process to modify the structure of a fluid whose equilibrium state is , by definition , one of maximum entropy subject to the hard - core excluded volume constraint .
importantly , the decrease in @xmath40 is _ finite bounded _ because @xmath126 will generally decay exponentially with @xmath99preventing unbounded growth of the integral in equation [ eqn : gb_inequality_ii ] , despite the growing range of @xmath48 .
more specific to iso-@xmath0 processes initiating from a hs fluid state , we expect the decay in the integral to be quite strong since hs radial distribution functions decay rapidly with @xmath99 at moderate densities @xcite , suggesting a fairly tight lower bound on @xmath40 .
two consequences for iso-@xmath0 processes follow from equation [ eqn : gb_inequality_ii ] : ( 1 ) approach to the incompressible fluid at @xmath32 is qualitatively different from approach to an incompressible , jammed or close - packed state , which would exhibit a negative divergence in @xmath40 @xcite , and ( 2 ) fluid relaxation times might be predicted to increase _ or _ decrease depending on whether @xmath40 or @xmath39 correlates with dynamics . to assess the expected changes in @xmath39 and @xmath40 along the unit - step function iso-@xmath0 process from @xmath127
, we calculate these properties based on @xmath48 and @xmath1 obtained from hnc frameworks . as shown in fig .
[ sch : auxiliary_range]c , a nominal increase in @xmath39 over hard spheres is seen . for @xmath40 ,
the change is opposite in magnitude and slightly larger than that observed for @xmath39though in an absolute sense the changes are still small , even at the singular point . given the modest changes in @xmath40 and @xmath11 along the iso-@xmath0 process shown in fig .
[ sch : auxiliary_range ] , one would not anticipate a pronounced change in dynamics . however , such predictions contradict expectations based on dynamical correlations with the excess compressibility factor @xcite ) , which would anticipate a pronounced dynamical slowdown approaching @xmath32 .
figure-skgrur-v7.png given these analytical results , we next demonstrate via md simulations that interactions such as those shown in fig .
[ sch : auxiliary_range]b can indeed minimize pair correlations and that upon approaching the singular density , the fluid states approach incompressibility while still exhibiting finite relaxation times comparable to equidensity hs . to wit , fig .
[ sch : skgrur - lowdens]a shows structure factors @xmath44 for simulated iso-@xmath0 state points ( no hs results shown ) between @xmath128 , while fig .
[ sch : skgrur - lowdens]b shows corresponding rdfs for both iso-@xmath0 and hs simulations . here
, the iso-@xmath0 systems are governed by the potentials @xmath48 shown in fig .
[ sch : skgrur - lowdens]c , which are a 3:1 weighted combinations of potentials derived via hnc and py closures for target step - function @xmath1 profiles .
these mixed potentials are excellent at generating the target step - function @xmath1 pair structure up to densities very close to the singular limit , and , as expected , the low-@xmath19 behaviors in @xmath44 indicate that the simulated compressibility drops rapidly with increasing @xmath10 .
figure-fskt-v3.png despite this precipitous drop in compressibility upon approach to @xmath32 , we show in fig .
[ sch : fskt - lowdens ] that the single - particle dynamics of this series of systems as captured by the decay of the self - intermediate scattering function @xmath129are virtually identical to those of hs systems at the same packing fractions .
this result can be rationalized on the basis of thermodynamics and the form of the auxiliary interactions .
first , we recall from fig .
[ sch : auxiliary_range]c that for this range of @xmath10 , changes in both @xmath39 and @xmath40 upon suppressing otherwise emergent hs structuring are finite and quite small .
thus , the similarity in dynamics may have been anticipated based on the quasi - universal correlation between excess entropy and relaxation discussed above ( though does not suggest whether either quantity is the _
better _ predictor ) . secondly , because the auxiliary interactions are smooth , monotonic , and increasingly long - ranged , they predominately establish a nearly additive renormalization of the system enthalpy relative to equidensity hs , introducing no new meaningful barriers to diffusion beyond those already present in the hs fluid .
accordingly , the rapidly decreasing compressibility near @xmath36 does not coincide with any jamming - like transition , but simply with the growing mean - field - like repulsive tail @xmath118 contribution to @xmath63 . as demonstrated for the step function iso-@xmath0 process ,
both @xmath39 and @xmath40 undergo only minor departures from hs fluid values , despite divergence of the auxiliary potential range and some thermodynamic measures at @xmath36 . to test the generality of these observations , iet - based calculations for a variety of finite density iso-@xmath0 processes
are explored with @xmath130 in 0.05 increments ( 11 iso-@xmath0 processes ) .
the first quantities of interest , shown in fig .
[ sch : continuum_s2]a , are the two - body and full excess entropy differences between the iso-@xmath0 processes and equidensity hs results , which we notate @xmath131 , where @xmath132 or @xmath40 .
while quantitative differences between the hnc and rpa theories emerge at higher density , both theories predict all 11 iso-@xmath0 processes furnish small overall changes in both @xmath39 and @xmath40 .
in addition to being small , the decrease in @xmath40 at the various @xmath36 points is nearly constant ( roughly @xmath133 ) , eventually being surpassed in magnitude by the change in @xmath39 with increasing @xmath10 .
the latter observation implies higher densities require less of a decrease in total disorder to suppress otherwise emergent pair structure . to get a better sense for how large the thermodynamic departures are from hs along an iso-@xmath0 process , we also consider @xmath134 , which expresses the changes as percents relative to the absolute hs value . for @xmath40
, we find a monotonic decrease ( in magnitude ) , which at the highest density iso-@xmath0 process only achieves about a 4% change . on the other hand ,
the @xmath39 predictions are non - monotonic and even more weakly varying ; evidently , the upper singular density permits only a meager 5 - 8% change in @xmath39 along a given hs iso-@xmath0 process . and ( b ) @xmath135 computed along a variety of hard sphere iso-@xmath0 processes within the rpa ( blue and teal ) and hnc ( red and orange ) approximations .
solid curves correspond to the 11 individual iso-@xmath0 processes ( hnc not shown for clarity ) while long dashed curves indicate the singular density values .
( c ) @xmath136 for the 11 iso-@xmath0 processes ( solid ) bare hard spheres ( small dashed ) and locus of singular density points ( long dashed ) .
open diamond symbols are simulation results from @xcite . ]
the quantities @xmath137 and @xmath138 lend insight into the change in @xmath39 and @xmath40 along an iso-@xmath0 process , but it is also interesting to consider how an iso-@xmath0 process _ repartitions _
two- versus higher - body correlations relative to hs . to address this issue , we use the fact that @xmath39 is the purely two - point contribution to @xmath40 ( i.e. , @xmath139 where @xmath140 contains all `` many - body '' terms not trivially reducible to purely two - point correlations ) .
thus , the ratio ( expressed as percentages ) @xmath141 intuitively expresses the percent of disorder due to two - body correlations . as shown in fig .
[ sch : continuum_s2]c , @xmath142 exhibits a non - monotonic variation that echoes that of the underlying behavior of the hs fluid ( i.e. , this is reflected by the singular points ) .
for the lowest density iso-@xmath0 process , the drop in @xmath142 is fairly strong ( 100% @xmath143 70% between @xmath144 ) , but higher density iso-@xmath0 processes are much less dramatic .
overall , the hnc and rpa theories suggest that no hs - reference iso-@xmath0 process is capable of swapping the dominant contributions towards `` disorder '' from two - body to many - body . whether the same is true
more generally for iso-@xmath0 processes referenced against non - hs systems remains to be seen .
potentials at @xmath145 given the reference states @xmath107 , 0.10 , 0.20 , 0.30 , 0.40 , 0.50 ( lightest to darkest ) .
hnc predictions are nearly identical and are thus not shown for clarity . ] in closing this section , we note that the auxiliary interactions required for the step - function iso-@xmath0 process are all ramp - like and purely - repulsive , a scenario that holds true for all of the hs iso-@xmath0 processes . as shown in fig .
[ sch : continuum_potentials ] , the rpa theory predicts repulsive ramp - like potentials for each of the iso-@xmath0 processes at @xmath145 .
the interaction range decreases with density , reflecting that higher density iso-@xmath0 processes must approach closer ( in density ) to the singularity for symptoms of the divergence to be detected .
all such potentials are slowly and smoothly varying , which is consistent with minimal changes in dynamics relative to the hs fluid upon approach to the respective singular packing fractions in accord with the modest changes in the excess entropy metrics shown in fig .
[ sch : continuum_s2 ] .
above @xmath146 , a unit step @xmath1 is not physically realizable because the corresponding @xmath44 is negative for some values of @xmath19 .
to find the @xmath1 of a fluid with hard cores that maximizes @xmath39 subject to the constraint that @xmath44 be positive definite for all @xmath19 , we use a genetic algorithm , beginning the optimization from the hs radial distribution functions ( shown in fig .
[ sch : optgr]a ) .
the resulting equi - density radial distribution functions are shown in fig .
[ sch : optgr]b . in general , the optimized structures bear some similarity to the hs fluid correlation functions , with muted features in the former , particularly at larger @xmath99 . just above the terminal density of the unit - step iso-@xmath0 process ( at @xmath147 ) , the optimized @xmath1 is predominantly flat with a small , slowly varying ramp near contact .
as @xmath10 increases , more structuring appears : the value of @xmath1 at contact grows , and a depletion region following the first coordination shell becomes apparent that grows in magnitude and shifts to lower @xmath99 with increasing @xmath10 . at the highest values of @xmath10 , a second coordination shell appears .
all @xmath11-maximized pair - correlation functions are relatively short - ranged ( much more so than the hs counterparts ) .
in increments of @xmath148 .
( b ) equi - density radial distribution functions optimized to maximize @xmath39 . ] fig .
[ sch : opts2]a shows @xmath39 for the hs fluid , @xmath11 from the optimized radial distribution functions ( with the @xmath44 positivity constraint ) , and @xmath11 of an ( unrealizable ) fluid with a unit - step radial distribution function .
of course , at lower densities where the @xmath1 most closely resembles a unit step , the bulk of @xmath39 comes from the core .
but non - hard core contributions due to the coordination shells grow in with increasing @xmath10 .
together , @xmath39 for the hs radial distribution function and @xmath39 for the unit - step radial distribution function provide a range in which the optimized @xmath39 must lie . in absolute terms , at the higher densities , the optimization has a greater effect to maximize @xmath39 .
we show this in fig .
[ sch : opts2]b , where the increase in @xmath39 from the hs to the optimized radial distribution function is plotted as a function of @xmath10 .
however , if we consider the change in @xmath39 upon optimization as a percentage of what could have otherwise been achieved without the @xmath149 realizability constraint , we see that this quantity decreases with @xmath10 ; see fig .
[ sch : opts2]c . in other words
, it is primarily the realizability constraint that limits pair disorder at higher @xmath10 . ) , maximal @xmath39 subject to the structure factor realizability constraint ( @xmath150 ) , and @xmath39 of an unrealizable unit - step radial distribution function ( @xmath151 ) as a function of packing fraction @xmath10 .
( b ) increase in @xmath39 relative to hard spheres : @xmath152 ( c ) percentage of @xmath11 gap between @xmath151 and @xmath153 recovered by the fluid upon @xmath11 optimization . ] the structure factors corresponding to the optimized radial distribution functions , shown in fig .
[ sch : optsk ] , are characteristic of disordered , stealthy , hyperuniform materials as they are transparent ( do not scatter ) over a finite range at low @xmath19 @xcite . due to the constrained optimization
, any solution above the terminal density must have at least one value of @xmath19 where @xmath44 is zero ; we find a whole range of low @xmath19 values , beginning at @xmath154 , where @xmath155 .
small oscillations above zero in the low @xmath19 region are due to the inexactness of the numerical optimizations and decrease in magnitude as the optimization progresses ( not shown ) . the range of @xmath19 values for which @xmath155 increases with @xmath10 .
this suggests an intimate relation between maximizing @xmath39 and the fabrication of hyperuniform , `` stealthy '' materials that possess minimal scattering at low wavevectors . this is a potentially useful relationship because optimizing materials in real space ( via interactions and radial distribution functions ) allows for a simpler encoding of the hard - core constraint ( or other real space constraints ) than does directly manipulating @xmath44 .
these calculations do not strictly prove the existence of such stealthy hard - core packings but , if realizable by explicit construction , would provide a complement to the library of currently existent _ point _ particle stealthy packings @xcite .
profiles in fig . 6 . note that oscillations at low @xmath19 are minor artifacts attributable to the numerical optimization process , which can in principle be refined via further iterative optimizations . ]
hard spheres are the archetypical model for a diverse array of systems at both the macroscale and microscale , but universal to all is the role of the hard core to restrict the available phase space and to limit the amount of disorder that is realizable .
as discussed in this work , thermalized hard spheres are maximally disordered when the excess entropy , @xmath40 , is considered ; introduction of any other interaction beyond the hard core necessarily lowers @xmath40 .
however , maximizing the disorder stored in the two body correlations , @xmath1 , is non - trivial . as our metric for pair disorder , we employed the two - body excess entropy , @xmath39 , which naturally arises from the multi - body expansion of @xmath40 .
to investigate the role of increasing @xmath39 , we explored a variety of hs iso-@xmath0 processes : from a given starting density , auxiliary interactions are introduced as the density is increased so that the initial hs @xmath1 is maintained .
such processes terminate when the singular density is reached ; these singular points are characterized by incompressibility and hyperuniformity features commonly associated with jamming or close - packed states .
unlike jammed packings though which would exhibit negative divergences in @xmath40 and @xmath39 at the jamming transition our iet calculations predict the iso-@xmath0 singularities have @xmath40 and @xmath39 values close to equidensity hard spheres .
this suggests the singular points are of virtually the same mobility as underlying hard spheres a prediction we confirmed with molecular dynamics simulations for the step function @xmath1 ( or zero starting density ) iso-@xmath0 process .
incompressibility at the singular densities is solely a manifestation of the divergent - ranged interactions required to maintain the lower - density radial distibution function , as opposed to a literal jamming transition .
the smooth , slowly varying nature of the interactions seems insufficient for generating the frustrated energy landscape needed for slow glassy dynamics . as a consequence of increasing the pair disorder along each iso-@xmath0 process
, stronger correlations are imbued into the @xmath156 body structure .
iet predicts that the increase in @xmath39 is comparable is magnitude to the decrease in @xmath40 , meaning that the increase in @xmath39 is accompanied by roughly twice as large of a decrease in the many body correlations in @xmath139 . despite transferring order from @xmath39 to @xmath140 , @xmath157 of @xmath40 of the iso-@xmath0 process
is predicted to result from @xmath39 .
this effect is rather mild , particularly considering that bare hard spheres have a contribution to @xmath40 from @xmath39 of @xmath158 .
thus , we conclude that hs iso-@xmath0 processes are not capable of swapping the dominant contribution to @xmath40 from two- to many - body . whether other systems can be found where @xmath159 is an interesting avenue for future research .
while iso-@xmath0 processes produce hard - core fluids of augmented @xmath39 relative to pure hard spheres , a literal maximization of @xmath39 subject to the constraint of a positive - definite @xmath44 was performed to uncover the true limits of pair disorder . below @xmath160 ,
the solution is the trivial step function @xmath1 ( degenerate with iso-@xmath0 processes beginning at @xmath31 ) . above @xmath160 ,
some degree of structure in @xmath1 must remain for realizability . as density
is increased , the optimized @xmath39 is increasingly greater than the @xmath39 of equidensity hard spheres ; however , this is a reflection of the greater available ( i.e , non - hard core ) @xmath39 at higher densities for hard spheres . expressed as a percentage of @xmath39 available to the optimization , we see that relative change in @xmath39 actually decreases as a function of density , reflecting that physical realizability is a more difficult constraint to satisfy at higher densities .
finally , we found that the resulting radial distribution functions are hyperuniform , with a range in low @xmath19 corresponding to @xmath155 that broadens with increasing density , indicating a relationship between maximizing pair disorder and so - called `` stealthly '' materials .
this relationship provides a convenient route to directly generate stealthy packing pair structure in a system of particles with hard cores .
explicit construction of such packings is an interesting avenue for future research .
this work was partially supported by the national science foundation ( 1247945 ) and the welch foundation ( f-1696 ) .
we acknowledge the texas advanced computing center ( tacc ) at the university of texas at austin for providing hpc resources
for completeness , this section provides a concise review of general analytical results regarding any iso-@xmath0 process , though the presented mathematical formalism follows a slightly different path than in the current literature @xcite .
general analytical results regarding any iso-@xmath0 process rely on approximating @xmath50 and exploiting the intimate relation it possesses with @xmath45 , in particular , the physical interpretation of @xmath161 as a renormalized pair potential between particles and the exact result , @xmath162 . from this relation
it is clear that knowledge of the asymptotic @xmath50 behavior
extractable from the oz relation alone provides the asymptotic form of @xmath45 .
all three of the considered closures ( equations [ eqn : hnc_closure]-[eqn : rpa_closure ] ) obey this limit via the trivial relation , @xmath163an expected consistency .
analytical statements regarding the long range @xmath50 and @xmath45 behavior are most readily found near the singular density , @xmath164 , at which @xmath165 for some @xmath166 . at @xmath167 , a divergence in @xmath114 is produced since @xmath168 yielding @xmath169 since @xmath170 . from equation [ eqn : ck_divergence ] ,
analytical results regarding the nature of the divergence , both in fourier and real space , are elucidated through a taylor expansion of @xmath58 in the denominator about @xmath167 yielding @xmath171 and @xmath172}\tag{a3}\label{eqn : correlation_length_i}\ ] ] where the primes indicate the order of differentiation , and @xmath173 is the auxiliary potential range .
interestingly , an alternative expression ( though identical at the singular density ) can be derived for @xmath174 by working with @xmath114 directly instead of @xmath175 yielding @xmath176}\tag{a4}\label{eqn : correlation_length_ii}\ ] ] both results furnish a divergence , @xmath177 , and together they appear to serve as bounds on the exact @xmath173 ( discussed in section [ results : step_isog ] ) . greater physical interpretation for @xmath173
as the auxiliary interaction range comes from the fourier transformed versions of equation [ eqn : ck_lorentzian ] , which for @xmath178 yields @xmath179}{r}\tag{a5}\label{eqn : yukawa_k0}\ ] ] arbitrarily close to the singular density @xmath180 which is well defined as @xmath181 remains finite .
this work was partially supported by the national science foundation ( 1247945 ) and the welch foundation ( f-1696 ) .
we acknowledge the texas advanced computing center ( tacc ) at the university of texas at austin for providing hpc resources .
62ifxundefined [ 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 10.1103/revmodphys.82.2633 [ * * , ( ) ] link:\doibase 10.1103/physreve.82.011403 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.103.135704 [ * * , ( ) ] @noop * * , ( ) link:\doibase 10.1103/physrevlett.84.2064 [ * * , ( ) ] link:\doibase 10.1103/physreve.68.011306 [ * * , ( ) ] link:\doibase 10.1103/physreve.71.011105 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.99.155501 [ * * , ( ) ] link:\doibase 10.1103/physreve.83.031503 [ * * , ( ) ] link:\doibase 10.1103/revmodphys.82.789 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.102.195701 [ * * , ( ) ] \doibase http://dx.doi.org/10.1063/1.3541248 [ * * , ( ) , http://dx.doi.org/10.1063/1.3541248 ] link:\doibase 10.1103/physreve.77.041201 [ * * , ( ) ] link:\doibase 10.1039/c1sm06932b [ * * , ( ) ] link:\doibase 10.1039/b814211b [ * * , ( ) ] link:\doibase 10.1002/aic.14491 [ * * , ( ) ] \doibase http://dx.doi.org/10.1063/1.1744724 [ * * , ( ) ] link:\doibase 10.1103/physreva.40.3817 [ * * , ( ) ] link:\doibase 10.1103/physreve.68.041113 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.95.090604 [ * * , ( ) ] link:\doibase 10.1021/jp010006q [ * * , ( ) ] , link:\doibase 10.1080/00268970500151528 [ * * , ( ) ] , @noop _ _ , ed .
( , , ) https://books.google.com/books?id=itcppndnjm0c[__ ] , ed .
( , , ) link:\doibase 10.1103/physreva.15.2545 [ * * , ( ) ] http://stacks.iop.org/0953-8984/11/i=28/a=303 [ * * , ( ) ] @noop * * , ( ) link:\doibase 10.1021/jp071369e [ * * , ( ) ] , , \doibase http://dx.doi.org/10.1063/1.2409932 [ * * , ( ) , http://dx.doi.org/10.1063/1.2409932 ] link:\doibase 10.1103/physreve.79.031203 [ * * , ( ) ] link:\doibase 10.1103/physreve.80.061205 [ * * , ( ) ] \doibase http://dx.doi.org/10.1063/1.3256235 [ * * , ( ) , http://dx.doi.org/10.1063/1.3256235 ] \doibase http://dx.doi.org/10.1063/1.3559676 [ * * , ( ) , http://dx.doi.org/10.1063/1.3559676 ] \doibase http://dx.doi.org/10.1063/1.1288804 [ * * , ( ) ] link:\doibase 10.1021/jp902934x [ * * , ( ) ] , , \doibase http://dx.doi.org/10.1063/1.4916053 [ * * , ( ) , http://dx.doi.org/10.1063/1.4916053 ] link:\doibase 10.1103/physrevlett.100.106001 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.8.462 [ * * , ( ) ] link:\doibase 10.1080/00268978300102111 [ * * , ( ) ] , link:\doibase 10.1080/00268978000102671 [ * * , ( ) ] , \doibase http://dx.doi.org/10.1063/1.1672048 [ * * , ( ) ] link:\doibase 10.1126/science.220.4599.787 [ * * , ( ) ] , link:\doibase 10.1021/ct700301q [ * * , ( ) ] , https://books.google.com/books?id=3_rqaaaamaaj[__ ] , ed . , artificial intelligence ( , , ) link:\doibase 10.1007/bf00175354 [ * * , ( ) ] in link:\doibase 10.1145/301618.301661 [ _ _ ] , ( , , ) pp .
link:\doibase 10.1515/zna-1991 - 1 - 205 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.96.177804 [ * * , ( ) ] link:\doibase 10.1103/physreve.62.993 [ * * , ( ) ] \doibase http://dx.doi.org/10.1063/1.2961314 [ * * , ( ) , http://dx.doi.org/10.1063/1.2961314 ] link:\doibase 10.1103/physrevx.5.021020 [ * * , ( ) ] \doibase http://dx.doi.org/10.1063/1.2897027 [ * * , ( ) , http://dx.doi.org/10.1063/1.2897027 ] | liquid - state theory , computer simulation , and numerical optimization are used to investigate the extent to which positional correlations of a hard - sphere fluid as characterized by the radial distribution function and the two - particle excess entropy can be suppressed via the introduction of auxiliary pair interactions .
the corresponding effects of such interactions on total excess entropy , density fluctuations , and single - particle dynamics are explored .
iso-@xmath0 processes , whereby hard - sphere - fluid pair structure at a given density is preserved at higher densities via the introduction of a density - dependent , soft repulsive contribution to the pair potential , are considered .
such processes eventually terminate at a singular density , resulting in a state that while incompressible and hyperuniform remains unjammed and exhibits fluid - like dynamic properties .
the extent to which static pair correlations can be suppressed to maximize pair disorder in a fluid with hard cores , determined via direct functional maximization of two - body excess entropy , is also considered .
systems approaching a state of maximized two - body entropy display a progressively growing bandwidth of suppressed density fluctuations , pointing to a relation between `` stealthiness '' and maximal pair disorder in materials . |
the study of stellar coronae historically has been relegated to x - ray wavelengths where high temperature plasmas predominately radiate .
unfortunately , observational limitations including low spectral resolution , lack of precise wavelength calibrations , small effective areas , line blending , nonnegligible continua , and unreliable atomic parameters have made it difficult to transfer powerful spectroscopic techniques commonly used in the ultraviolet ( uv ) to the x - ray spectral region .
this disparity is improving with the development of powerful x - ray telescopes , such as the _ chandra x - ray observatory _ ( _ cxo _ ) and the _ x - ray multi - mirror mission _ ( _ xmm - newton _ ) . with each new observation of a late - type star by these telescopes , our understanding of the structure of stellar
coronae is advancing ( e.g. ayres et al .
2001 ; brickhouse , dupree , & young 2001 ; gdel et al .
however , even the modern array of x - ray spectrometers fall far short of the resolving power needed to study the dynamics of coronal emission lines . in particular
, the spectral resolution of the _ cxo _ gratings is nominally about @xmath0/@xmath81000 , which corresponds to a velocity resolution of 300 km s@xmath9 .
the spectral resolution of the reflection grating spectrometer ( rgs ) onboard _ xmm - newton _ is at least a factor of two worse . in contrast , the _ far ultraviolet spectroscopic explorer _
( _ fuse _ ) has a resolving power of @xmath0/@xmath815,000 , which corresponds to a velocity resolution of 20 km s@xmath9 .
presently , the fortuitous presence of coronal forbidden lines in the uv provides a way to study fundamental aspects of the structure of stellar coronae , by circumventing the limitations of contemporary x - ray spectroscopy . over the course of the last century ,
coronal forbidden lines have repeatedly been observed and studied in the sun @xcite .
most recently , solar flare observations using the _ solar ultraviolet measurements of emitted radiation _ ( sumer ) instrument onboard the _ solar & heliospheric observatory _ ( _ soho _ ) have identified many coronal forbidden lines in the wavelength range from 500 - 1600 @xcite . by comparison ,
detections of ultraviolet coronal forbidden lines in other stars have been a relative novelty .
the initial discoveries were of @xmath01354 in the brightest coronal late - type stars , including the active m - dwarf au mic ( m0 v ; maran et al .
1994 ) , the hyperactive rs canum venaticorum ( rs cvn ) binary hr 1099 ( k0 iv + g5 v ; robinson et al .
1996 ) , and the giant - star binary system capella ( g8 iii + g1 iii ; linsky et al .
unfortunately , the line profile analysis is complicated because @xmath01354 is blended with a chromospheric emission line . besides , only one other coronal forbidden line species has been spotted in the wavelength range available from _ hubble space telescope _ ( _ hst _ ) : @xcite identified the relatively cool @xmath101242,1349 coronal lines in the spectrum of @xmath11 eri . @xcite
have presented a comprehensive survey of and coronal forbidden lines based on _ hst _ observations , primarily with the space telescope imaging spectrograph ( stis ) . the _ fuse _ satellite provides access to the important 900 - 1200 spectral region below the short wavelength cutoff of _ hst _ , covering a new spectral region to search for stellar coronal forbidden lines . again
, solar flare observations of coronal lines have previously been made at these wavelengths @xcite .
@xcite published the solar spectrum from 914 - 1177 , based on rocket observations , as a template for stellar observations to be made by _
fuse_. included were 18 coronal forbidden lines , the strongest of which are : @xmath01133 , @xmath0974 , @xmath01118 , and @xmath01153 .
various authors have noted the presence of coronal forbidden lines in _ fuse _ observations of active stars .
@xcite identified the line in the spectrum of capella , and found that the bulk of the emission arises from the g8 giant .
@xcite also used the and lines in their multi - wavelength analysis of the emission measure of au mic .
@xcite identified and in observations of ab dor , @xmath11 eri , and au mic , as part of a _ fuse _ survey of cool dwarf stars .
@xcite noted the and lines in @xmath12 cet and 31 com .
the present paper surveys a large body of _ fuse _ observations of late - type stars for coronal forbidden lines .
our objective is not only to enumerate positive detections , but also to demonstrate the utility of the coronal forbidden line diagnostics for understanding the dynamics of high temperature plasmas in stellar coronae .
table [ properties ] lists the 26 stars in our sample .
they include the targets of the _ fuse _ team s cool star survey , supplemented with a number of guest observer observations .
table [ summary ] summarizes the _ fuse _ observations .
most targets were observed in time - tagged mode , and all were observed either through the medium resolution slit ( mdrs , @xmath13 ) or the large aperture ( lwrs , @xmath14 ) .
the lwrs has the advantage of usually including light from all of the lif and sic channels for the duration of the exposure , at the expense of higher airglow contamination , especially at ly@xmath12 and ly@xmath15 , but also in many and lines .
the mdrs reduces airglow , but thermal effects often cause one or more of the channels to drift out of the aperture during an exposure , because only the lif1 channel is used for guiding .
we monitor the integrated flux of the strongest lines in each segment ( i.e. @xmath0977 , @xmath01031 , and @xmath01176 ) . during exposures
when the star drifts our of the aperture , the flux in these lines is significantly reduced from its nominal value .
these exposures are therefore not included in the reduction of the final spectrum .
the spectral images were processed with calfuse 2.0.5 .
compared to earlier versions of the pipeline , this new rendition provides a more accurate wavelength scale , removes large noise event bursts , corrects for misplaced photons at low sensitivity portions of the detector ( so - called `` walk '' problem ) , improves corrections for scattered light and astigmatism , extracts the spectrum using an `` optimal '' algorithm , and applies a more accurate flux calibration ( flux errors are @xmath1615% ; oegerle , murphy , & kriss [ 2000 ] ) .
although the scattered light correction and flux calibration are much improved , they remain approximations and can occasionally lead to negative flux levels , where there is no stellar emission .
the flux error incorporates the errors involved in these calibration steps .
individual exposures were reduced separately and the integrated flux in strong emission lines was used to test the consistency of the signal recorded in each segment , throughout the entire observation . for each segment ,
the exposures then were cross - correlated , using selected stellar emission features , and coadded , excepting those with significant drops in signal due to target drifts out of the aperture ( a common occurrence for mdrs observations ) .
redundant segments were not coadded , however , to avoid degrading the spectral resolution .
the correction for astigmatism ( curvature of spectrum perpendicular to the dispersion ) improves the spectral resolution by 510% compared with earlier versions of calfuse used to reduce the previously published spectra of ab dor @xcite and capella @xcite .
the relative wavelength scale within a given detector channel is accurate to @xmath17 5 over most of the far - ultraviolet ( fuv ) bandpass , except for some deterioration at the far edges of the individual channels @xcite . due to the absence of an onboard wavelength calibration lamp , we established the zero point of the wavelength scale according to interstellar absorption features in the 1036.34 and 977.02 lines , and by comparing the 1176 lines in _ fuse _ spectra with wavelength - calibrated observations by _
hst_. the procedure is described in detail by @xcite .
we searched for evidence of coronal forbidden lines in the 900 - 1200 region in our sample of stellar spectra . due to our relatively poor knowledge of the rest wavelengths of these lines , the relatively weak signal of the coronal lines , and the moderate likelihood of coincidental blends with low temperature lines ,
labeling spectral features located at wavelengths coincident with an expected coronal line can lead to misidentifications , as demonstrated below .
therefore , we used a comparative technique developed by @xcite to identify coronal forbidden lines in longer wavelength _ hst _ spectra .
we compared each sample spectrum with a typical spectrum of an active and inactive dwarf star ( au mic and @xmath3 cen a , respectively ) , and an active and inactive giant star ( @xmath12 cet and @xmath12 gem , respectively ) .
strong coronal emission lines will stand out as broad features in the spectra of active stars , but should be absent in the inactive stars .
coincidental low temperature lines in the active spectra can be identified because they will be relatively narrow , and also be present in the inactive stars .
the underlying assumptions in this comparative analysis are that active stars have larger coronal emission measures than inactive stars , and that giant stars all have reasonably similar emission measure distributions , and all dwarf stars also share similar emission measure distributions .
although we have chosen specific active stars to demonstrate this technique below , any star with its own unique emission measure distribution can be used as the standard .
this comparison proves useful at identifying coincident blends and should be used along with other line identification techniques when searching for coronal forbidden lines .
figure [ lines ] illustrates the comparative technique .
we selected the four strongest high temperature ( @xmath18 k ) coronal forbidden lines observed in solar flares @xcite and located in the _ fuse _ spectral range : @xmath0974 ( @xmath19 k ) , @xmath01118 ( @xmath20 k ) , @xmath01133 ( @xmath21 k ) , and @xmath01153 ( @xmath22 k ) .
as the strongest lines in the _ fuse _ spectral range detected in solar flares , these lines are an appropriate place to start looking for coronal forbidden lines in stellar spectra .
the top two panels of figure [ lines ] demonstrate that and are positively detected , because the features appear in the normalized spectra of both active stars , and are absent in the inactive stars .
, shown in the bottom panel is clearly not detected in any of the stars . at the expected location of , there is a prominent feature .
however , it is present only in the spectra of the giant stars , both active and inactive , and is narrow , inconsistent with the high line formation temperature . therefore , it has the characteristics of a low temperature line .
in fact , @xcite identified numerous fluorescent lines in the _ fuse _ spectra of @xmath3 tra , two of which fall at wavelengths 1133.675 and 1133.854 .
these lines are blended and appear as a single feature .
therefore , the emission detected in _ fuse _ spectra at @xmath23 is not , but rather a coincidental fluorescent line pumped by h ly@xmath3 .
the comparative technique was applied to an extensive list of more than 50 candidate coronal forbidden lines that lie in the _ fuse _ spectral range . of these ,
only two were positively detected : the @xmath0974 and @xmath01118 lines mentioned previously .
these features form at @xmath1 and @xmath24 , respectively . in table
[ observed ] , we list the observed integrated line strengths corresponding to direct flux integrations within @xmath17 150 km s@xmath9 of line center for stars with @xmath25 km s@xmath9 , and within @xmath17 250 km s@xmath9 for stars with @xmath26 km s@xmath9 .
x - ray fluxes , also included in the table , are based on the 0.2 - 2.0 kev band observed by the _ rntgensatellit _ ( _ rosat _ ) .
the addition of the @xmath0974 line , in particular , to the collection of uv coronal forbidden lines is very promising , due to its optimum spectral context . unlike other high temperature forbidden lines , such as @xmath01118 and @xmath01354 whose
profiles are compromised by coincident blends , the nearest significant blend with is @xmath27 km s@xmath9 from line center .
therefore , requires no additional assumptions or delicate deconvolutions , when trying to reconstruct the intrinsic coronal forbidden line profile .
furthermore , the ultraviolet continua of typical cool stars decrease dramatically toward the shorter wavelengths .
therefore , modeling the continuum , if present at all , is less challenging in the spectral region around , than , for example , at @xmath01354 . also , the strong local interstellar medium ( lism ) absorption in the prominent 977.02 line , adjacent to the line , provides a convenient means of calibrating the zero point of the local velocity scale .
figure [ fe18line ] presents the @xmath0974 lines , and single gaussian fits , in seven spectra of the slowly rotating stars with detections . table [ fits ] lists the derived parameters . in figure
[ fe18line ] , the solid horizontal line indicates the continuum level ( often zero ) , based on line - free regions on either side of the feature . the intrinsic stellar line profile ( prior to convolution with the instrumental line spread function [ wood et al .
2002 ] and the rotational line profile ) is shown by the dashed curve .
the solid curve represents the predicted profile including both instrumental and rotational broadening .
note that the intrinsic line widths are so broad that the contribution of the instrument and rotation often is negligible .
rotational broadening is most noticeable for 47 cas , which is the only star among these seven with @xmath28 km s@xmath9 ; the other six stars have relatively modest rotational velocities ( see table [ properties ] ) .
in fact , figure [ fe18line ] indicates very little variation in line width among the seven targets , consistent with domination by thermal broadening .
the line formation temperature for is @xmath29 , which corresponds to a full - width at half - maximum ( fwhm ) of 69 km s@xmath9 , but is likely to form over a modest spread in temperatures .
a reasonable range is @xmath30 , which corresponds to widths of @xmath31 km s @xmath32 . in column 7 of table
[ fits ] , we list the equivalent temperature required to account for the observed intrinsic line width if turbulent motions do not contribute to the line broadening .
figure [ fe18line ] also demonstrates that @xmath0974 displays no large velocity shifts , as the emission centroids are approximately at the rest velocity of the stellar photosphere in all cases . in a couple of the stars , 47 cas and ek dra , we were unable to determine an accurate zero point of the wavelength scale because either there is no obvious lism absorption in the 977.02 line , or no high - resolution wavelength - calibrated data ( e.g. , stis ) available to measure the velocity of lism absorption empirically . therefore , for these two stars , we assume that the centroid of the feature is at the rest velocity of the star .
@xcite showed that the centroid of can be shifted from line center by as much as 15 km s@xmath9 .
therefore we use 15 km s@xmath9 as our estimate of the systematic error . only in the two broadest profiles , 31 com and ab dor ,
do the statistical errors dominate the systematic errors .
spreading out the signal of an already weak coronal line , makes it extremely difficult to accurately determine the centroid of the feature .
higher s / n observations are required for the stars that exhibit broad line profiles in order to measure the centroids of the emission .
the widths in most of the stars are consistent with little to no turbulent broadening : thermal broadening , alone , is sufficient to account for the observed intrinsic line width .
the two fastest rotating stars in the sample , 31 com and ab dor show large intrinsic line widths , however , even when the photospheric rotational broadening is taken into account .
another fast rotator , 47 cas , also shows a large line width , marginally in excess of the expected thermal broadening .
although an additional broadening mechanism may also play a part in the coronal line profiles of 47 cas , we limit the following analysis to 31 com and ab dor .
the spectra of these two stars clearly indicates that thermal broadening alone is insufficient to account for the width of the coronal line profiles . in figure
[ fe18rot ] , the top panels show the initial single gaussian fit for which the rotational broadening was fixed at the observed photospheric value .
it is clear that the inferred intrinsic line profile is much wider than those of the slow rotators , shown in figure [ fe18line ] .
it seems unlikely that nonthermal , turbulent broadening in 31 com and ab dor could explain the very broad profiles , because such broadening is clearly of little importance in the other stars .
instead , we interpret the additional broadening to be the result of `` super - rotation '' , extended co - rotating coronal emission high above the stellar photosphere . in the bottom panels of figure [ fe18rot ] ,
the intrinsic line width was fixed at the thermal value .
an alternative broadening mechanism is now required to derive a satisfactory fit to the data .
if we allow the rotational velocity to vary , the additional broadening can now be attributed to a value of @xmath4 that is larger than the photospheric value .
both models produce equally satisfactory fits , but the top panels indicate a model that assumes that the rotational velocity is fixed at the photospheric value and turbulent broadening must account for the line width , whereas the bottom panels assume that turbulent broadening is negligible , based on the results of the slowly rotating stars , and the rotational velocity is instead the free parameter . for 31 com , @xmath33 for the fixed rotational velocity model and @xmath34 for the variable rotational velocity model . for ab dor , @xmath35 for the fixed rotational velocity model and @xmath36 for the variable rotational velocity model .
unfortunately , the s / n of the data is not high enough to differentiate between the two assumptions based solely on the subtle differences in the resulting line profile .
however , the unanimous result that turbulent broadening is nonexistent in our sample of slowly rotating stars , leads us to propose that the additional broadening is due to the rotation of the coronal material at velocities greater than the photospheric @xmath4 .
the best fit @xmath4 for 31 com is @xmath37 km s@xmath9 and for ab dor , @xmath38 km s@xmath9 .
although the s / n of the data prohibits an accurate determination of the coronal rotational velocity , it is clear that it must be greater than the photospheric value .
based on the enhanced rotational velocities , we can estimate how far above the stellar surface the presumably co - rotating emitting plasma extends . for 31 com , we infer that coronal material extends to @xmath5r @xmath39 r@xmath7 above the stellar surface , and for ab dor , @xmath5r @xmath40 r@xmath7 .
similar calculations have been made using transition region lines @xcite and coronal forbidden lines @xcite , observed with _ hst_. extended , but cooler , structures also have been inferred through time series of h@xmath3 absorption spectra of ab dor @xcite .
@xcite have used zeeman - doppler maps of ab dor to extrapolate the topology of the extended coronal magnetic field . an alternative interpretation to explain the observed additional broadening is by the presence of high speed flows .
measurements of density sensitive lines in spectra observed by the _ extreme ultraviolet explorer _ satellite ( _ euve _ ) , seem to indicate that the emitting volume of the coronal plasma is small compared to the stellar radius @xcite .
this appears to imply compact coronal structures as opposed to extended co - rotating coronal emission .
such high speed flows in coronal plasmas have recently been observed on the sun @xcite .
however , we only detect additional broadening in the fastest rotating stars , not necessarily in the most active . for example
, slow rotators , such as @xmath11 eri and au mic , which are much more active than the sun , do not exhibit additional broadening , although they may be expected to have prominent high speed flows .
therefore , we feel that emission from plasma confined in extended but thin magnetic loops is a more likely explanation for the observed additional broadening in the fastest rotators , although a specific model should be built with densities consistent with the euve measurements to support this conclusion .
to estimate whether plasma confined in extended magnetic flux tubes can explain the observed broad fe xviii emission lines , we consider this simple example .
we assume that all of the fe xviii emission occurs in half - circular flux loops extending up to a height @xmath41 above the stellar surface with a constant radius @xmath42 and electron density @xmath43 .
if there are @xmath44 such loops , then the emission measure @xmath45 .
we can now solve for the product , @xmath46 as an example , we consider the corona of ab dor . @xcite analyzed the nonflare spectrum of ab dor observed by the rgs instrument on xmm . in their three component model
, the middle temperature component has a temperature @xmath47 kev , similar to the formation temperture of fe xviii , and the corresponding emission measure @xmath48 @xmath49
. ratios of the helium - like triplet lines of o vii are consistent with a coronal electron density @xmath50 @xmath49 . if `` super - rotation '' is the explanation for the widths of the fe xviii line , then @xmath51 cm for @xmath52 @xcite .
these parameters yield @xmath53 . if we assume
n = 10 or 100 , then @xmath3 = 0.03 or 0.009 , and the loops are thin as is true for the solar corona @xcite . we conclude that the `` super - rotation '' model with many thin flux loops is consistent with the coronal data for this test case .
also the flux loop thickness parameter @xmath3 will likely be small , as expected , for a wide range of coronal parameters .
interpretation of the coronal @xmath0974 feature of the giant binary capella is complicated by the fact that both stars ( the g8 giant primary [ @xmath3 aur aa ] and the g1 giant secondary [ @xmath3 aur ab ] ) can contribute to the emission .
_ hst _ observations of the @xmath01354 line profile of capella have demonstrated that the relative contribution by each star changes with time .
based on the analysis of capella observations taken on 1995 september 9 , @xcite determined that each star provided approximately half of the total emission flux . @xcite and @xcite showed in stis observations of capella taken four years later , on 1999 september 12 , that the emission from the g8 primary star was reduced to @xmath54% of its previous level , although the g1 component remained about the same .
@xcite used a single gaussian model and applied it to the first _ fuse _ observation of capella ( 2000 nov 5 ) to evaluate the individual stellar contributions to the emission flux .
they estimated that the g8 primary contributed @xmath55% of the total flux .
we go beyond that first stage analysis by fitting all three available _ fuse _ observations in an effort to better understand the nature of capella s coronal feature .
the dates of the capella observations are listed in table [ summary ] , while individual stellar velocities , phases , and the ephemeris reference are listed in table [ properties ] .
first , we modeled the @xmath0974 emission profile with a single gaussian , appropriate if the emission were coming principally from one of the stars .
figure [ fe18cap ] illustrates the fit to each individual exposure .
the first two observations were taken near orbital quadrature ( maximum radial velocity separation ) , and the third near conjunction ( minimum velocity separation ) .
the velocity scale was adjusted to the reference frame of the g8 primary star , known to @xmath56 km s@xmath9 , using the strong lism absorption in the neighboring 977.02 line @xcite .
the fit parameters corresponding to figure [ fe18cap ] are listed in table [ fitscapella ] .
it is unclear why there is a decrease in integrated flux between the first and second two observations .
instrumental effects , such as thermal drifts , can cause a decrease in flux , as discussed in section 2 .
although stellar variability may also explain the decrease in flux @xcite , it is unlikely because the first two discrepant observations were are taken only two days apart .
however , the flux discrepancy does not effect the line width , line centroid , or flux ratio of the binary components , and therefore does not effect the conclusions of this work .
the bulk of the emission clearly is associated with the g8 star , based on the centroid of the emission , as was noted by @xcite .
note , however , that the line width decreases significantly between quadrature and conjunction .
this indicates that although emission from the g8 primary star may dominate , there is a nonnegligible contribution from the faster rotating g1 star , and we are compelled to fit the capella emission feature with two components , one for the contribution of each star . in order to estimate the contribution of the g1 secondary star
, we modeled the @xmath0974 emission profile with two gaussians , one for each star , including instrumental and rotational broadening as before . in this case , however , the rotational velocity of the g8 star is small ( @xmath57 km s@xmath9 ) , but that of the g1 star is much larger ( @xmath58 km s@xmath9 ; strassmeier et al .
[ 1988 ] ) .
although there is discernible emission from the g1 secondary star , it is still relatively weak , and the gaussian modeling benefits from fixing several of the free parameters . in particular , we constrained the velocity centroid of each component to be at the respective orbital velocity , and the fwhm of the g8 component to the value determined from the single gaussian fit near conjunction ( essentially the thermal width ) .
both constraints are motivated by the empirical behavior of active single stars , discussed in section [ secfe18 ] : velocity centroids are consistent with the photospheric radial velocities ; and line widths in slow rotators are well characterized by thermal broadening alone .
the results of the double gaussian modeling are depicted in figure [ fe18_2cap ] , and listed in table [ fitscapella ] .
the @xmath59 goodness - of - fit metric is also listed in table [ fitscapella ] .
the fits are noticeably improved for the two observations near quadrature , particularly redward of the main component ( 80 - 140 km s@xmath9 ) .
the g1 secondary apparently contributes @xmath60% of the flux in .
the width of the secondary component is broader than expected from thermal broadening alone . as for 31 com and ab
dor discussed previously , we interpret the excess width in terms of an extended coronal zone . in this case
, we estimate that the emission extends @xmath5r @xmath61r@xmath7 above the stellar photosphere .
@xcite reached similar conclusions in the @xmath01354 line of the g1 star .
the other coronal forbidden line detected in _ fuse _ observations of late - type stars is @xmath01118 .
in contrast to @xmath0974 , there are several severe blends that can obscure the feature .
at least eight lines are scattered between 1117.20 and 1118.49 . in some stars , however , these lines are weak relative to the coronal emission . the interval surrounding is shown in figure [ fe19line ] for the 10 stars that exhibit a significant feature . if we assume that the emission measure is the same for the formation of and , we can predict the line profile of based on theoretical emissivities and the empirical gaussian parameters . we compute the emissivities with chianti version 4 @xcite .
each ion is assumed to be emitting at its most probable temperature .
the resulting ratio of emissivities is insensitive to assumptions of pressure , abundances , or ionization equilibrium . in figure
[ fe19line ] , the single gaussian profiles of , presented in figures [ fe18line ] , [ fe18rot ] , and [ fe18cap ] , have been scaled by the relative emissivities . in some cases , particularly 47 cas , @xmath12 cet , au mic , and @xmath62 vel , there is excellent agreement between the observed spectra and the predicted line profile . the shapes and doppler shifts of the line profiles relatively free of blending ,
seem to be identical to . here
the lines are weak , although some excess carbon emission is present , particularly blueward of the coronal feature .
in other cases , especially capella and 31 com , the emission wreaks havoc with the spectra .
the line is clearly detected in many cases , nevertheless , severe blending typically limits our ability to analyze it . in some cases , particularly , ek dra , dk uma , and @xmath11 eri
, we significantly overestimate the flux of the feature .
our first order assumption that the emission measure is identical for ( @xmath19 k ) and ( @xmath20 k ) was motivated by the idea that the emission measure distribution is unlikely to change drastically over a small change in formation temperature .
this is supported by the differential emission measure analysis of au mic by @xcite , as well as the other stars that show excellent agreement between the observed and predicted levels of emission in figure [ fe19line ] . because stellar flux was not detected in the sic channels of the 1999 october 20 observation of ab dor , the emission profile was only observed in the 1999 december 14 observation ( see table [ summary ] ) . however , both visits successfully observed the spectral region surrounding the coronal forbidden line . the coadded spectrum is shown in figure [ fe19line ] .
the full _ fuse _ spectrum of ab dor indicates that emission lines are present @xcite .
therefore , we expect some contamination in the profile .
the broad emission in the line of ab dor appears to be consistent with the broad emission detected in , although it is difficult to differentiate between the contribution of flux and emission .
this is an excellent example of the powerful use of a coronal forbidden line that is uncontaminated by low temperature lines , such as @xmath0974 , and the difficulty in analyzing coronal lines that are severely blended , such as @xmath01118 .
all of the stars in our sample have soft x - ray fluxes observed by _
rosat_. because the forbidden lines observed in the uv and the soft x - rays both sample high temperature coronal plasmas in these stars , a correlation between the two diagnostics is expected .
figure [ lfelx ] compares @xmath0974 and the 0.2 - 2.0 kev soft x - rays .
the coronal fluxes are normalized by the bolometric fluxes given in table [ properties ] , to remove biases of stellar size and distance .
those stars with detected emission are displayed as open circles , and those with upper limits are shown as small filled circles .
a unit slope power law is indicated by the thick dashed line .
a good correlation exists over two orders of magnitude , down to a sensitivity threshold of @xmath63 .
all of the upper limits are consistent with this correlation .
@xcite demonstrated a similar unit power law correlation for @xmath01354 . in active stars , at least , the flux can serve as a surrogate for the broad - band soft x - ray emission .
_ fuse _ provides an opportunity to study a new collection of high temperature ( @xmath64 k ) forbidden lines in nearby coronal stars .
the goal of the present paper was to extensively examine _ fuse _ observations of late - type stars for such features , and to analyze any detections as potential diagnostics of the physical characteristics of the high temperature plasmas .
our results are as follows : * after searching the _ fuse _ spectra for candidate coronal forbidden lines identified in solar flares @xcite , only two were found : @xmath0974 and @xmath01118 .
detections were obtained in 10 of the 26 stars in our sample . the two _ fuse _ lines join @xmath101242,1349 and @xmath01354 coronal lines from the stis 1200 - 1600 region .
* of the five known ultraviolet coronal forbidden lines , @xmath0974 has important advantages as a spectroscopic diagnostic . unlike @xmath01354 and @xmath01118 ,
is not blended with low temperature lines that can corrupt the intrinsic line profile .
likewise , the stellar continuum is almost nonexistent at the location of , whereas , at longer wavelengths it can create difficulties for isolating the weak coronal forbidden emission .
although _ fuse _ does not have an internal wavelength calibration system , lism absorption in the strong 977.02 line , just 660 km s@xmath9 redward of , often can be used as a wavelength fiducial ( in those cases where @xmath65 is known independently ) .
* we find that a single gaussian component is successful for modeling the @xmath0974 profile for all single stars .
the line is fully resolved , so instrumental broadening is negligible . *
the centroid of @xmath0974 is approximately at rest relative to the radial velocity of the star , in contrast to lower temperature ( @xmath66 k ) transition region lines which are often redshifted 10 - 15 km s @xmath9 @xcite .
therefore , the high temperature plasma likely is confined in stable magnetic structures ; not participating in any large systematic mass motions . * for all of the slowly rotating stars , thermal broadening dominates the widths of the coronal forbidden lines , consistent with predicted formation temperatures @xmath67 k. turbulent broadening appears to be negligible , which is additional evidence that the hot coronal material might be stably entrained in coronal magnetic structures in slowly rotating stars . *
three of the fastest rotating stars in our sample ( ab dor , 31 com , and the g1 star of capella ) , all show excess broadening in @xmath0974 . owing to the overwhelming similarity of line widths in all of the slowly rotating stars of our sample ,
we attribute the excess to `` super - rotational '' broadening .
namely , the hot gas is co - rotating in extended coronal structures rising well above the stellar surface .
we estimate that this material is located approximately @xmath5r @xmath68-@xmath69r@xmath7 beyond the stellar photosphere .
high speed flows could also potentially explain the excess broadening , although it is not clear why such flows would be observed in the fastest rotating stars . a detailed discussion concerning the cause of the excess broadening is given in section [ secfe18 ] . * the @xmath0974 profile of capella is complicated by contributions from both stars of the binary .
as demonstrated with the @xmath01354 line , the relative contribution of each star can vary @xcite .
using three different observations , at quadrature and conjunction , we find that the g8 primary contributes @xmath70% of the emission , and the g1 secondary @xmath60% . *
we find an excellent correlation between @xmath0974 and the 0.2 - 2.0 kev _ rosat _ soft x - ray flux .
although upper limits mask any trends at lower activity levels , we expect that the correlation will exhibit a `` break '' at a critical activity level where the flux will drop faster than the unit slope power law for stars of progressively lower activity .
in particular , @xcite have demonstrated that the coronal temperature drops with decreasing activity , and eventually the coronal emission measures will peak below the formation temperature , suppressing the forbidden line flux relative to the soft x - ray band .
this work is based on data obtained for the guaranteed time team by the nasa - cnes - csa _ fuse _ mission , which is operated by the johns hopkins university .
financial support to u. s. participants has been provided by nasa contract nas5 - 32985 , as well as individual _ fuse _ guest investigator grants .
this research has made use of the simbad database , operated at cds , strasbourg , france , and the _ rosat _ x - ray archive at the heasarc of the nasa goddard space flight center .
ake , t.b . ,
dupree , a.k .
, young , p.r . ,
linsky , j.l . ,
malina , r.f . , griffiths , n.w . ,
siegmund , o.h.w . , & woodgate , b.e .
2000 , , 538 , l87 ambruster , c.w .
, sciortino , s. , & golub , l. 1987 , , 65 , 273 audard , m. , gdel , m. , drake , j.j . , & kashyap , v.l .
2000 , , 541 , 396 ayres , t.r .
, brown , a. , harper , g. , osten , r.a . , linsky , j.l . ,
wood , b.e . , & redfield , s. 2003 , apj , in press ayres , t.r . , brown , a. , osten , r.a .
, huenemoerder , d.p . ,
drake , j.j . ,
brickhouse , n.s . , & linsky , j.l .
2001 , , 549 , 554 ayres , t.r . ,
simon , t. , stern , r.a . , drake , s.a . , wood , b.e . , & brown , a. 1998 , , 496 , 428 brandt , j.c .
2001 , , 121 , 2173 brickhouse , n. s. , dupree , a. k. , & young , p. r. 2001 , , 562 , l75 collier cameron , a. 1996 , in iau symp .
176 , stellar surface structure , ed .
strassmeier & j.l .
linsky ( dordrecht : kluwer ) , 449 del zanna , g. , landini , m. , & mason , h.e .
2002 , , 385 , 968 de medeiros , j.r . , & mayor , m. 1999 , , 139 , 433 dere , k.p . ,
landi , e. , mason , h.e . ,
monsignori fossi , b.c . ,
& young , p.r .
1997 , , 125 , 149 dixon , v. 2001 , introduction to calfuse v2.0 dupree , a. k. , brickhouse , n. s. , doschek , g. a. , green , j. c. , & raymond , j. c. 1993 , , 418 , l41 dupree , a. k. , lobel , a. , young , p. r. , ake , t. b. , linsky , j. l. , & redfield , s. 2003 , , submitted edln , b. 1945 , , 105 , 323 fekel , f.c .
1997 , , 109 , 514 feldman , u. , curdt , w. , landi , e. , wilhelm , k. 2000 , , 544 , 508 feldman , u. , & doschek , g. a. 1991 , , 75 , 925 franchini , m. , covino , e. , stalio , r. , terranegra , l. , & chavarria - k , c. 1992 , , 256 , 525 gray , d.f .
1989 , , 347 , 1021 gray , d.f .
2000 , , 532 , 487 gdel , m. et al .
2001 , , 365 , l336 gdel , m. et al . 2003 , in the twelfth cool stars , stellar systems and the sun , eds .
a. brown , t.r .
ayres , & g.m .
harper ( boulder : university of colorado ) , in press gdel , m. , guinan , e.f . , & skinner , s.l .
1997 , , 483 , 947 hale , a. 1994 , , 107 , 306 harper , g.m . ,
wilkinson , e. , brown , a. , jordan , c. , & linsky , j.l .
2001 , , 551 , 486 hummel , c.a . ,
armstrong , j.t . ,
quirrenbach , a. , buscher , d.f . ,
mozurkewich , d. , elias , n.m . ,
ii , & wilson , r.e .
1994 , , 107 , 1859 jardine , m. , collier cameron , a. , & donati , j .- f . 2002 , , 333 , 339 johnson , o. , drake , j.j . , kashyap , v. , brickhouse , n.s . ,
dupree , a.k .
, freeman , p. , young , p.r . , & kriss , g.a .
2002 , , 565 , l97 jordan , c. , mcmurry , a. d. , sim , s. a. , & arulvel , m. 2001 , , 322 , l5 linsky , j. l. , wood , b. e. , brown , a. , & osten , r. a. 1998 , , 492 , 767 maran , s.p et al .
1994 , , 421 , 800 montes , d. , lpez - santiago , j. , fernndes - figueroa , m.j . , & glvez , m.c .
2001 , , 379 , 976 oegerle , b. , murphy , e. , & kriss , j. 2000 , the _ fuse _ data handbook v1.1 pagano , i. , linsky , j.l . , carkner , l. , robinson , r.d . , woodgate , b. , & timothy , g. 2000 , , 532 , 497 pallavicini , r. , golub , l. , rosner , r. , vaiana , g.s . ,
ayres , t. , & linsky , j.l .
1981 , , 248 , 279 peter , h. , & judge , p.g .
1999 , , 522 , 1148 pettersen , b.r .
1980 , , 82 , 53 redfield , s. , linsky , j.l . ,
ake , t.b . ,
ayres , t.r . ,
dupree , a.k . ,
robinson , r.d . ,
wood , b.e . , & young , p.r .
2002 , , in press robinson , r.d . ,
airapetian , v.s . ,
maran , s.p . , & carpenter , k.g .
1996 , , 469 , 872 royer , r. , gerbaldi , m. , faraggiana , r. , & gmex , a.e .
2002 , , 381 , 105 sanz - forcada , j. , brickhouse , n. s. , & dupree , a. k. 2002 , , 570 , 799 schrijver , c. j. , et al .
1999 , , 187 , 261 simon , t. , drake , s.a .
, & kim , p.d .
1995 , , 107 , 1034 smith , g. , edvardsson , b. , & frisk , u. 1986 , , 165 , 126 strassmeier , k.g . , hall , d.s . ,
zeilik , m. , nelson , e. , eker , z. , & fekel , f.c .
1988 , , 72 , 291 van belle , g. t. , ciardi , d. r. , thompson , r. r. , akeson , r. l. , & lada , e. a. 2001 , , 559 , 1155 winebarger , a. r. , warren , h. , van ballegooijen , a. , deluca , e. e. , & bolub , l. 2002 , , 567 , l89 wood , b.e . , brown , a. , linksy , j.l . ,
kellett , b.j . ,
bromage , g.e . ,
hodgkin , s.t . , & pye , j.p .
1994 , , 93 , 287 wood , b.e . ,
linsky , j.l . , & ayres , t.r .
1997 , , 478 , 745 wood , b.e . ,
linsky , j.l . ,
hbrard , g. , vidal - madjar , a. , lemoine , m. , moos , h.w . ,
sembach , k.r . , & jenkins , e.b .
2002 , , 140 , 91 wood , b.e . ,
redfield , s. , linsky , j.l .
, & sahu , m.s .
2003 , , submitted young , p.r .
, del zanna , g. , landi , e. , dere , k.p . ,
mason , h.e . , & landini , m. 2002 , , in press young , p. r. , dupree , a. k. , wood , b. e. , redfield , s. , linsky , j. l. , ake , t. b. , & moos , h. w. 2001 , , 555 , l121 lccccccccc & 23628 & a4 v & 0 & 7.71 & 0.18 & 0.20 & 130 & 215 & 1 + hr 5999 & 144668 & a7 iv & 1.9 & 6.98 & 0.29 & 0.39 & 210 & 204 & 2 + @xmath3 aql & 187642 & a7v & 26.1 & 0.77 & 0.22 & 118 & 5.15 & 210 & 3 + @xmath3 car & 45348 & f0 ii & 20.5 & 0.72 & 0.15 & 478 & 95.8 & 9 & 2 + 47 cas & 12230 & f0 v & 26 & 5.28 & 0.30 & 1.89 & 33.6 & 62.1 & 4 + @xmath3 cmi & 61421 & f5iv - v&3.2 & 0.34 & 0.40 & 185 & 3.50 & 5.0 & 5 + @xmath71 peg & 220657 & f8 iv & 11.1 & 4.40 & 0.61 & 5.2 & 53 & 33.7 & 6 + ek dra & 129333 & f8 & 30.5 & 7.61 & 0.59 & 0.24 & 33.9 & 17.3 & 7 + 31 com & 111812 & g0 iiip & 1.4 & 4.94 & 0.63 & 3.4 & 94.2 & 66.5 & 6 + hr 9024 & 223460 & g1 iii & 0.7 & 5.90 & 0.79 & 1.6 & 130 & 21.5 & 6 + @xmath3 aur ab & 34029 & g1iii & 52.4 , 51.3 , 25.4 & 0.76 & 0.65 & 138 & 12.9 & 36 & 8 + @xmath12 dra & 159181 & g2 iab & 20.0 & 2.79 & 0.98 & 24 & 110.9 & 10.7 & 6 + @xmath3 aqr & 209750 & g2 ib & 7.5 & 2.95 & 0.96 & 20 & 230 & 7.8 & 6 + @xmath3 cen a & 128620 & g2v & 24.6 & -0.01 & 0.71 & 283 & 1.35 & 3 & 9 + dk uma & 82210 & g4 iii & 27.2 & 4.57 & 0.75 & 4.6 & 32.4 & 5.5 & 6 + @xmath62 vel & 93497 & g5 iii & 6.2 & 2.72 & 0.91 & 28 & 35.5 & 6.4 & 6 + @xmath3 aur aa & 34029 & g8iii & 7.1 , 8.1 , 32.8 & 0.91 & 0.88 & 140 & 12.9 & 3 & 8 + @xmath12 cet & 4128 & k0iii & 13.0 & 2.04 & 1.02 & 56 & 29.4 & 3.0 & 10 + @xmath12 gem & 62509 & k0 iiib & 3.3 & 1.15 & 1.00 & 121 & 10.3 & @xmath16 1.0 & 6 + ab dor & 36705 & k1v & 29.2 & 6.93 & 0.80 & 0.51 & 14.9 & 91 & 11 + @xmath3 cen b & 128621 & k2v & 20.7 & 1.33 & 0.88 & 93 & 1.35 & 2 & 12 + @xmath11 eri & 22049 & k2v & 15.5 & 3.73 & 0.88 & 10.4 & 3.22 & 2.0 & 13 + @xmath3 tau & 29139 & k5 iii & 54.3 & 0.85 & 1.54 & 343 & 20.0 & 2.0 & 6 + tw hya & & k8 v & 13.5 & 11.1 & 0.7 & 0.010 & 56 & @xmath16 15 & 14 + au mic & 197481 & m0v & 5.7 & 8.61 & 1.44 & 0.3 & 10.0 & @xmath16 6.2 & 15 + @xmath3 ori & 39801 & m2 iab & 21.0 & 0.58 & 1.77 & 1065 & 131.1 &
@xmath16 11 & 16 + proxima cen & & m5.5 v & 15.7 & 11.05 & 1.97 & 0.3 & 1.295 & 1.4 & 17 + lccccc hd 23628 & p1860501 & 2000 dec 29 & lwrs & 6 & 6.9 + hd 144668 & p1860201 & 2001 aug 20 & lwrs & 4 & 4.8 + @xmath3 aql & p1180701 & 2001 sep 14 & lwrs & 8 & 4.2 + @xmath3 car & p1180101 & 2000 dec 11 & lwrs & 2 & 5.6 + @xmath3 car & p2180101 & 2001 oct 25 & lwrs & 3 & 6.1 + @xmath3 car & p2180102 & 2001 oct 26 & lwrs & 9 & 11.1 + 47 cas & p1860601 & 2000 nov 26 & lwrs & 3 & 10.9 + procyon & p1041801 & 2001 oct 21 & mdrs & 7 & 5.3 + @xmath71 peg&b0720301 & 2001 jul 1 & lwrs & 4 & 11.8 + ek dra & p1860401 & 2000 mar 1 & lwrs & 3 & 10.3 + 31 com & p1180401 & 2001 apr 20 & lwrs & 8 & 12.9 + hr 9024 & b0720101 & 2001 jul 10 & lwrs & 19 & 37.2 + @xmath3 aur & p1040301 & 2000 nov 5 & lwrs & 10 & 14.2 + @xmath3 aur & p1040302 & 2000 nov 7 & lwrs & 10 & 12.3 + @xmath3 aur & p1040303 & 2001 jan 11 & mdrs & 28 & 21.4 + @xmath12 dra & p1180301 & 2000 may 9 & lwrs & 2 & 5.6 + @xmath12 dra & p2180301 & 2001 jun 30 & lwrs & 8 & 16.4 + @xmath3 aqr & p1180201 & 2000 jun 29 & lwrs & 1 & 3.3 + @xmath3 aqr & p1180202 & 2001 oct 8 & lwrs & 7 & 15.9 + @xmath3 aqr & p2180201 & 2001 jun 16 & lwrs & 11 & 34.4 + @xmath3 aqr & p2180202 & 2001 oct 7 & lwrs & 5 & 10.5 + @xmath3 cen a & p1042601 & 2001 jun 25 & mdrs & 8 & 15.3 + dk uma & c1070102 & 2002 feb 14 & lwrs & 6 & 24.6 + @xmath62 vel & c1070201 & 2002 mar 5 & lwrs & 8 & 25.2 + @xmath12 cet & p1180501 & 2000 dec 10 & lwrs & 10 & 12.6 + @xmath12 gem & p1180601 & 2000 nov 11 & lwrs & 7 & 21.9 + ab dor & x0250201&1999 oct 20 & lwrs & 1 & 22.1 + ab dor & x0250203 & 1999 dec 14 & lwrs & 6 & 24.2 + @xmath3 cen b & p1042501 & 2001 jun 24 & mdrs & 13 & 22.7 + @xmath11 eri & p1040701 & 2000 dec 8 & mdrs & 15 & 34.8 + @xmath3 tau & p1040901 & 2001 jan 14 & mdrs & 8 & 12.2 + tw hya & p1860101 & 2000 jun 3 & lwrs & 3 & 2.1 + au mic & p1180801 & 2000 aug 26 & lwrs & 9 & 17.3 + au mic & p2180401 & 2001 oct 10 & lwrs & 13 & 26.5 + @xmath3 ori & p1180901 & 2000 nov 3 & lwrs & 7 & 10.4 + proxima cen & p1860701 & 2000 may 16 & lwrs & 2 & 6.9 + lcccc hd 23628 & @xmath16 1.8 & @xmath16 1.0 & 90 @xmath17 20 & 1 + hd 144668 & @xmath16 1.2 & @xmath16 3.0 & 76 @xmath17 4 & 1 + @xmath3 aql & @xmath16 3.1 & @xmath16 7.7 & 71 @xmath17 2 & 1 + @xmath3 car & @xmath16 1.5 & @xmath16 1.3 & 420 @xmath17 10 & 1 + 47 cas & 1.0 @xmath17 0.3 & 1.2 @xmath17 0.2 & 1310 @xmath17 30 & 1 + @xmath3 cmi & @xmath16 3.0 & @xmath16 4.9 & 1300 @xmath17 20 & 1 + @xmath71 peg & @xmath16 1.0 & @xmath16 0.6 & 600 @xmath17 100 & 2 + ek dra & 1.0 @xmath17 0.3 & 0.6 @xmath17 0.1 & 730 @xmath17 190 & 1 + 31 com & 0.8 @xmath17 0.3 & 1.5 @xmath17 0.2 & 800 @xmath17 100 & 2 +
hr 9024 & @xmath16 0.8 & @xmath16 0.6 & 1000 @xmath17 500 & 2 + @xmath12 dra & @xmath16 0.9 & @xmath16 1.1 & 190 @xmath17 10 & 1 + @xmath3 aqr & @xmath16 0.5 & @xmath16 0.3 & @xmath16 20 & 1 + @xmath3 cen a & @xmath16 1.1 & @xmath16 2.4 & 700 @xmath17 200 & 2 + dk uma & 1.1 @xmath17 0.2 & 1.0 @xmath17 0.1 & 1400 @xmath17 100 & 2 + @xmath62 vel & 1.4 @xmath17 0.2 & 1.9 @xmath17 0.2 & 1600 @xmath17 100 & 2 + @xmath3 aur & 11.5 @xmath17 1.3 & 31.0 @xmath17 4.7 & 12700 @xmath17 1600 & 2 + @xmath12 cet & 3.8 @xmath17 0.4 & 4.1 @xmath17 0.4 & 2000 @xmath17 200 & 2 + @xmath12 gem & @xmath16 0.9 & @xmath16 0.6 & 30 @xmath17 2 & 1 + ab dor & 1.6 @xmath17 0.3 & 1.6 @xmath17 0.2 & 5500 @xmath17 2000 & 1 + @xmath3 cen b & @xmath16 1.3 & @xmath16 2.4 & @xmath6 1200 & 1 + @xmath11 eri & 0.8 @xmath17 0.2 & 0.7 @xmath17 0.1 & 1500 @xmath17 500 & 2 + @xmath3 tau & @xmath16 0.9 & @xmath16 0.6 & @xmath16 4 & 2 + tw hya & @xmath16 1.8 & @xmath16 0.5 & 300 @xmath17 10 & 1 + au mic & 0.6 @xmath17 0.1 & 0.7 @xmath17 0.1 & 2200 @xmath17 1000 & 1 + @xmath3 ori & @xmath16 0.8 & @xmath16 0.6 & @xmath16 13 & 1 + proxima cen & @xmath16 0.4 & @xmath16 0.5 & 700 @xmath17 200 & 2 + lcccccc 47 cas & 12230 & 4 @xmath17 15 & 1.0 @xmath17 0.1 & 113 @xmath17 10 & 87 @xmath17 10 & @xmath72 + ek dra & 129333 & 3 @xmath17 15 & 0.96 @xmath17 0.10 & 78 @xmath17 10 & 74 @xmath17 10 & @xmath73 + 31 com & 111812 & 22 @xmath17 35 & 0.70 @xmath17 0.07 & 149 @xmath17 14 & 125 @xmath17 14 & @xmath74 + dk uma & 82210 & 4 @xmath17 10 & 1.19 @xmath17 0.13 & 75.3 @xmath17 9.6 & 75.3 @xmath17 9.6 & @xmath75 + @xmath62 vel & 93497 & 2.6 @xmath17 5.0 & 1.5 @xmath17 0.2 & 77 @xmath17 10 & 77 @xmath17 10 & @xmath76 + @xmath12 cet & 4128 & 1.1 @xmath17 5.0 & 3.53 @xmath17 .35 & 83.4 @xmath17 6.8 & 83.4 @xmath17 6.8 & 6.93 @xmath17 0.05 + ab dor & 36705 & 16 @xmath17 19 & 1.92 @xmath17 0.32 & 309 @xmath17 35 & 288 @xmath17 35 & @xmath77 + @xmath11 eri & 22049 & 7 @xmath17 10 & 0.97 @xmath17 0.10 & 75.8 @xmath17 8.8 & 75.8 @xmath17 8.8 & @xmath78 + au mic & 197481 & 5 @xmath17 10 & 0.56 @xmath17 0.08 & 81 @xmath17 14 & 81 @xmath17 14 & @xmath79 + lccccccccc + + @xmath3 aur ( g8 ) & 34029 & 2000 nov 5 & 0.34 & 5.5 @xmath17 5.0 & 12.7 @xmath17 1.3 & 97.5 @xmath17 4.3 & 97.5 @xmath17 4.3 & 7.06 @xmath17 0.03 & 3.31 + @xmath3 aur ( g8 ) & 34029 & 2000 nov 7 & 0.35 & 3.0 @xmath17 5.0 & 9.76 @xmath17 0.98 & 92.8 @xmath17 6.8 & 92.8 @xmath17 6.8 & @xmath80 & 6.21 + @xmath3 aur ( g8 ) & 34029 & 2001 jan 11 & 0.98 & 8.1 @xmath17 5.0 & 9.62 @xmath17 0.96 & 80.1 @xmath17 5.0 & 80.1 @xmath17 5.0 & 6.89 @xmath17 0.04 & 9.99 + + @xmath3 aur ( g8 ) & 34029 & 2000 nov 5 & 0.34 & 0.0 & 9.09 @xmath17 0.91 & 80.1 & 80.1 & 6.89 & 1.12 + ( g1 ) & & & & 0.0 & 4.44 @xmath17 0.72 & 169 @xmath17 25 & 164 @xmath17 25 & 7.51@xmath81 & 1.12 + @xmath3 aur ( g8 ) & 34029 & 2000 nov 7 & 0.35 & 0.0 & 7.47 @xmath17 0.75 & 80.1 & 80.1 & 6.89 & 1.14 + ( g1 ) & & & & 0.0 & 2.86 @xmath17 0.66 & 158 @xmath17 28 & 151 @xmath17 30 & 7.44@xmath82 & 1.14 + @xmath3 aur ( g8 ) & 34029 & 2001 jan 11 & 0.98 & 0.0 & 7.12 @xmath17 0.62 & 80.1 & 80.1 & 6.89 & 2.27 + ( g1 ) & & & & 0.0 & 2.15 @xmath17 0.34 & 164 & 158 & 7.48 & 2.27 + | we present a survey of coronal forbidden lines detected in _
far ultraviolet spectroscopic explorer _
( _ fuse _ ) spectra of nearby stars .
two strong coronal features , @xmath0974 and @xmath01118 , are observed in 10 of the 26 stars in our sample .
various other coronal forbidden lines , observed in solar flares , also were sought but not detected .
the feature , formed at @xmath1 , appears to be free of blends , whereas the line can be corrupted by a multiplet .
_ fuse _ observations of these forbidden iron lines at spectral resolution @xmath2 provides the opportunity to study dynamics of hot coronal plasmas .
we find that the velocity centroid of the feature deviates little from the stellar rest frame , confirming that the hot coronal plasma is confined .
the observed line widths generally are consistent with thermal broadening at the high temperatures of formation and show little indication of additional turbulent broadening .
the fastest rotating stars , 31 com , @xmath3 aur ab , and ab dor , show evidence for excess broadening beyond the thermal component and the photospheric @xmath4 .
the anomalously large widths in these fast rotating targets may be evidence for enhanced rotational broadening consistent with emission from coronal regions extending an additional @xmath5r @xmath6 0.4 - 1.3 r@xmath7 above the stellar photosphere or represent the turbulent broadening caused by flows along magnetic loop structures .
for the stars in which is detected , there is an excellent correlation between the observed _ rntgensatellit _ ( _ rosat _ ) 0.2 - 2.0 kev soft x - ray flux and the coronal forbidden line flux . as a result , is a powerful new diagnostic of coronal thermal conditions and dynamics that can be utilized to study high temperature plasma processes in late - type stars .
in particular , _ fuse _ provides the opportunity to obtain observations of important transition region lines in the far - uv , as well as , simultaneous measurements of soft x - ray coronal emission , using the coronal forbidden line . |
the @xmath7 isobar is a well established nucleon resonance with spin - parity @xmath8 and isospin @xmath9 . due to its coupling to the nucleon - nucleon @xmath10 system via the transitions @xmath11 and @xmath12 this resonance plays an important role in the dynamics of the @xmath13 interaction @xcite as well as in electromagnatic and pionic processes involving the @xmath13 system , cf . ,
e.g. , refs . @xcite and references therein .
for example , unpolarized cross sections of @xmath14-meson production at energies of several hundred mev in many reactions involving the @xmath15 system can be reasonably described by using theoretical models with the @xmath16-excitation explicitely included @xcite .
but even near the pion - production threshold the contributions from @xmath11 transitions are essential for a quantitative description of the reaction @xmath17 @xcite .
also an important part of three - nucleon forces can be related to the @xmath16-isobar excitation , as is widely discussed in the literature , see , for example , refs .
these forces allow to remove the so - called sagara descrepancy @xcite in the unpolarized @xmath18 elastic cross section in the region of the minimum of the cross section @xcite at beam energies 60 - 200 mev .
the contribution of these forces increases with beam energy , since the intermediate @xmath16-isobar comes closer to its mass - shell , and they dominate the unpolarized cross section of @xmath19 elastic backward scattering at 400 - 600 mev , as is seen from model calculations @xcite . finally , the influence of the @xmath16-excitation was shown for in - medium three - body forces too where the @xmath16 generates an repulsive effective interaction @xcite . despite of this important role one has to concede that the spin dependence of the @xmath11 transition amplitude is not yet well known .
indeed the @xmath11 amplitude was studied in several experimental @xcite and theoretical papers @xcite . however , though a rather satisfactory overall description of available data at intermediate energies could be achieved , there are several unresolved problems concerning specific spin observables in the @xmath20 transition @xcite .
also , while three - body forces related to the @xmath16 contribution are capable of explaining the unpolarized @xmath21 cross section , corresponding investigations of polarization observables show no systematic improvement when such three - body forces are included @xcite .
improved information on the spin dependence of the @xmath11 amplitude might allow to shed light on this issue too .
furthermore , a better knowledge of the spin dependence of the @xmath11 amplitude might be useful for understanding the remarkable variation in the energy dependence of the spin - dependend @xmath22 cross sections @xmath23 @xcite and @xmath24 @xcite , at beam energies 600 - 800 mev .
this variation is considered by some authors @xcite as an indication of dibaryon resonances ( for a review see @xcite ) .
but since the observed structure lies in the proximity of the nominal @xmath25 threshold it could be closely linked with the properites of the @xmath11 and/or @xmath26 amplitudes @xcite .
the general binary reaction @xmath27 is described by 16 independent spin amplitudes and the determination of all of them is an extremely complicated and challenging experimental task . in collinear kinematics , however , the spin structure of any binary reaction is simplified considerably because only a single physical direction exists in this case , namely the direction of the beam momentum . as a consequence ,
the number of invariant spin amplitudes describing the reactions reduces drastically .
a similar simplification occurs in the threshold regime with the s - wave dominance in the final state of the reaction . under these conditions
it seems rather realistic to perform a complete polarization experiment for the process @xmath20 and , as a result , provide stringent constraints on the pertinent transition amplitudes .
note that the analysis of polarization effects in collinear kinematics with explicit axial symmetry can not be considered as a limiting case of the general formalism developed for binary reactions where the scattering plane is well defined @xcite . for the case of collinear kinematics the corresponding analytical expressions have to be derived anew .
for example , for backward elastic @xmath28 scattering the formalism was developed in @xcite .
the near threshold formalism for polarization phenomena has been recently documented in ref .
@xcite .
the main goal of this paper is to derive the formalism for spin observables of binary reactions of the types @xmath29 and @xmath30 in collinear kinematics and to find on this basis the corresponding complete polarization experiments which would allow one to measure all invariant amplitude describing these processes . for this aim
we use the technique of clebsch - gordan coefficients @xcite applied recently for collinear kinematics of a binary reaction in ref .
@xcite ( see for details also ref .
our phenomenological analysis of the reactions @xmath11 and @xmath31 is model - independent and based only on parity and angular momentum conservation .
however , in addition , for the reaction @xmath31 we develop also a model within the impulse approximation .
this allows us to connect the three invariant spin amplitudes of the reaction @xmath31 with the four invariant amplitudes of the process @xmath11 . as a result
, the knowledge of all three independent amplitudes of the reaction @xmath31 obtained from a dedicated complete polarizations experiment for this reaction allows one to determine the three independent combinations of the four amplitudes of the reaction @xmath11 .
earlier , in ref .
@xcite a similar formalism was developed for the charge - exchange reaction @xmath32 .
this formalism was succesfully applied for the determination of some of the spin - flip amplitudes in the quasi - free charge - exchange scattering @xmath33 @xcite .
the paper is organized in the following way . in section 2
we present the general formalism for spin observables of a binary reaction in collinear kinematics . in section 3
we present the full set of spin observables for the reaction @xmath29 and suggest a complete polarization experiment for this reaction .
section 4 deals with the reaction @xmath30 . a complete polarization experiment for this reaction
is described in subsections 4.a and 4.b considering a transversally as well as longitudinally polarized beam .
the relations between the amplitudes of the reactions @xmath34 and @xmath35 are given in subsection 4.c in the impulse approximation .
within this approximation we find formulae for three independent linear combination of the amplitudes of the reaction @xmath35 in terms of the invariant amplitudes of the reaction @xmath34 and transition form factor . in the appendix formulae for spin observables of the reaction @xmath34
are given .
the most general expression for the amplitude of the binary reaction @xmath36 in collinear kinematics can be written as @xcite @xmath37 here @xmath38 and @xmath39 are the spin of the _ k_-th particle and its z - projection , @xmath40 and @xmath41 are the spin and its z - projection for initial ( final ) particles .
the summation over the total angular momentum and orbital angular momenta in the initial and final states is included into the definition of the invariant spin amplitudes @xmath42 ( see ref .
the orbital momentum @xmath43 in eq .
( [ tfi ] ) is restricted by parity conservation via @xmath44 , where @xmath45 is the intrinsic parity of i-_th _ particle .
the triple - spin correlation coefficients for the case of two polarized initial particles and one polarized final particle can be determined as @xmath46 where @xmath47 is the transition operator determined by eq .
( [ tfi ] ) , and @xmath48 denotes the tensor operator of rank @xmath49 and magnetic quantum number @xmath50 @xmath51 for the _ i_-th particle .
this operator is normalized as @xmath52 using eqs.([tfi ] ) , ( [ ttt ] ) and properties of the operators @xmath53 ( see , for example , ref .
@xcite ) , one can find the following formula @xcite @xmath54 due to the presence of the clebsch - gordan coefficient @xmath55 in eq .
( [ ktttgen ] ) and parity conservation only even @xmath56 contribute to the right side of eq .
( [ ktttgen ] ) @xmath57 the coefficient @xmath58 is nonzero only for @xmath59 . from eq .
( [ ktttgen ] ) one can find the following property @xmath60 eq .
( [ minusm ] ) shows that , in the particular case of @xmath61 only an even sum of @xmath62 is allowed for nonzero triple correlations .
( [ ktttgen ] ) is rather general and describes also double spin correlation if the rank @xmath49 for one ( initial or final ) particle is zero and it gives tensor polarizations ( or analyzing powers ) for @xmath63 if the ranks of the two other particles are zero .
for the reaction @xmath64 in collinear kinematics only four invariant amplitudes @xmath65 are allowed by parity and angular momentum conservation : @xmath66 where @xmath67 results from the initial spin - singlet state and the other amplitudes are from the spin - triplet state .
the unpolarized cross section can be written as @xmath68 where @xmath69 and @xmath70 is the phase - space factor . using eq .
( [ ktttgen ] ) one can find the full set of spin observables of the reaction @xmath64 as @xmath71,\\ \label{k000020 } k_{00,00}^{20}\sigma= \frac{1}{8}\bigl [ -10|b_1|^2 - 5|b_2|^2 + 2\sqrt{15 } re b_2^*(b_3+\sqrt{2}b_4 ) -2|\sqrt{2}b_3-b_4|^2+|b_3+\sqrt{2}b_4|^2\bigr ] , \\
\label{k101020 } k_{10,10}^{20}\sigma= \frac{1}{8}\bigl [ 10|b_1|^2 - 5|b_2|^2 + 2\sqrt{15 } re b_2^*(b_3+\sqrt{2}b_4 ) + 2|\sqrt{2}b_3-b_4|^2+|b_3+\sqrt{2}b_4|^2\bigr ] .\end{aligned}\ ] ] @xmath72,\\ \label{k100030 } k_{10,00}^{30}\sigma= \frac{1}{4}\bigl [ -2\sqrt{5}|b_2|^2 + 6re b_1^*(\sqrt{2}b_3-b_4)+2\sqrt{3}re b_2^*(b_3+\sqrt{2}b_4 ) \bigr ] , \\
\label{k110031 } k_{11,00}^{3 - 1}\sigma= \frac{1}{4}\bigl [ -2\sqrt{5}re b_1\,b_2^*+ 4re b_2^*(\sqrt{2}b_3-b_4)+2\sqrt{3}re b_1^*(b_3+\sqrt{2}b_4 ) \bigr ] , \\
\label{sigma } \sigma= 5|b_1|^2 + 5|b_2|^2+|b_3+\sqrt{2}b_4|^2 + |\sqrt{2}b_3-b_4|^2).\end{aligned}\ ] ] for observables with the sum @xmath62 being odd ( i.e. t - odd observables ) we find the following formulae : @xmath73,\\ \label{k110021 } k_{11,00}^{2 - 1}\,\sigma= \frac{i}{16}\bigl [ -10\sqrt{2}imb_1b_2^ * % -6\sqrt{6}im b_3b_4^ * + 2\sqrt{6}im(b_3+\sqrt{2}b_4)(\sqrt{2}b_3-b_4)^ * -2\sqrt{30}im b_1(b_3+\sqrt{2}b_4)^*+\nonumber \\
+ 2\sqrt{10}im b_2(\sqrt{2}b_3-b_4)^ * \bigl].\end{aligned}\ ] ] one can see that the amplitudes @xmath74 and @xmath75 enter the above formulae for observables only in the two combinations @xmath76 and @xmath77 .
in order to determine completely the matrix element of this reaction one has to find four moduli of the amplitudes and three relative phases ( the overall phase is arbitrary ) .
the choice of a minimal set of experiments is not unique and depends on the experimental conditions .
here we describe one possible minimal set .
we use here and below the following relations : @xmath78 where @xmath79 is the phase of the amplitude @xmath80 ( @xmath81 ) .
the four moduli @xmath82 and @xmath83 can be found from eqs .
( [ k111100 ] ) , ( [ k101020 ] ) , ( [ k000020 ] ) and ( [ k111122 ] ) in the following form @xmath84 where @xmath85 is given by eq.([sigma ] ) and can be written as @xmath86 thus , a measurement of the observables @xmath87 , @xmath88,@xmath89 , @xmath90 and @xmath91 is sufficient for determining those moduli .
simultaneously , one can find from these observables the value of @xmath92 . in order to determine @xmath93 one could measure @xmath94 in addition to the above observables . as a result , for the two real parts we find the following @xmath95 furthermore , a measurement of the two t - odd observables @xmath96 and @xmath97 allows us to determine @xmath98 and @xmath99 , respectively , as it seen from eqs .
( [ k11113 - 2 ] ) and ( [ k111130 ] ) .
note that if the modulus and the real part of a complex number is known , then it would be sufficient to measure only the sign of the imaginary part in order to determine the phase of this number .
thus , measurement of the t - odd observables will not require a high accuracy .
the knowledge of the real and imaginary parts of the products @xmath100 and @xmath101 completely determines the relative phases @xmath102 and @xmath103 . the last relative phase , for example , @xmath104 , can be found by a measurement of the observables @xmath105 and @xmath106 .
indeed , one can see from eqs .
( [ k111031 ] ) and ( [ k110031 ] ) , that this measurement provides two linear equations for the two unknown variables @xmath107 and @xmath108 . thus , a complete polarization experiment in this version requires to measure ten observables : @xmath87 , @xmath88,@xmath89 , @xmath90 , @xmath91 , @xmath94 , @xmath109 , @xmath96 , @xmath97 and @xmath110 .
for the reaction @xmath112 in collinear kinematics the following three invariant amplitudes @xmath65 are allowed by parity and angular momentum conservation : @xmath113 in order to determine a strategy for a complete polarization experiment we derive here all non - zero double and triple spin observables , which are given in appendix a. ( note that some observables are not presented but can be easily derived via eq .
( [ minusm ] ) . ) in the subsequent discussion of a complete polarization experiment we consider a particular case , namely those observables which require only transversally polarized particles in the initial state .
the case with longitudinal polarizations is considered below separately . using the following six observables , @xmath87 , @xmath114 , @xmath115 , @xmath91 , @xmath116 , and @xmath117
, one can find moduli of the three amplitudes @xmath118 , @xmath119 , @xmath120 and cosines of two phases : @xmath121 where @xmath122 @xmath123 in order to find the sines of these phases one should measure the three t - odd observables @xmath124 , @xmath125 , and @xmath126 . with those observables
one obtains @xmath127 note that for the sum @xmath62 being odd , the value of @xmath128 is purely imaginary .
the cartesian components of these coefficients are purely real @xcite .
again , only the signs of the imaginary parts of @xmath129 and @xmath130 are required .
thus , one needs nine observables related to transversally polarized beam and/or target in order to completely determine the three spin amplitudes of the reaction @xmath111 .
measurements with longitudinally polarized beam / target allows one to diminish the number of observables in a complete polarization experiment as compared to measurements with transversal polarizations .
for example , using @xmath131 and the longitudinal spin correlation parameter @xmath132 from eq .
( [ s101000 ] ) together with the transversal observables @xmath115 , @xmath133 , @xmath117 , given by eqs .
( [ i0 ] ) , ( [ s111100 ] ) , ( [ s002121 ] ) and ( [ s002222 ] ) , respectively , one can find three moduli @xmath118 , @xmath119 , and @xmath120 .
in addition , a measurement of the coefficient @xmath91 given by eq .
( [ s000020 ] ) allows to determine the cosines of three phases from @xmath134 , @xmath135 and @xmath136 .
finally , a knowledge of the two t - odd observables @xmath124 and @xmath125 , presented by eqs . ( [ s001121 ] ) and ( [ s110021 ] ) , gives @xmath137 as @xmath138 in view of the following relation between the relative phases , @xmath139 , where @xmath140 and @xmath141 is the phase of the amplitude @xmath142 , one can determine unambiguously the three complex numbers @xmath143 , @xmath144 and @xmath145 from the above eight observables .
assuming that single @xmath2 scattering ( see fig . [ mechanism ] ) dominates in the reaction @xmath111 one can express the transition amplitude of this reaction in terms of invariant amplitudes of the elementary subprocess @xmath146 in the following way : @xmath147 here @xmath148 and @xmath149 are the @xmath150-projections of the spins of the beam proton , deuteron and @xmath16-isobar , respectively , and @xmath151 denotes the nucleon mass . @xmath152
( @xmath153=0 and 2 ) are the transition form factors @xmath154 at the momentum q transferred from the beam proton to the final @xmath16 isobar , @xmath155 where @xmath156 and @xmath157 are the s- and d - components of the deuteron wave function normalized by @xmath158=1,\end{aligned}\ ] ] the @xmath13 scattering wave function in the @xmath159 state is normalized at @xmath160 as @xmath161 where @xmath162 is the momentum of the nucleon in the cms of the @xmath13 system and @xmath163 is the @xmath159 phase shift
. the single scattering mechanisms of the @xmath164 reaction . ] using eq .
( [ tfi ] ) one can find the invariant amplitudes ( [ aaa ] ) from the following relations : @xmath165 where the lower indices in @xmath166 correspond to the proton ( @xmath167 ) and deuteron ( @xmath168 ) spin projections on the @xmath150-axis .
the above amplitudes are related to the unpolarized c.m.s .
cross section by @xmath169 where @xmath170 denotes the invariant mass of the @xmath19 system and @xmath171 ( @xmath172 ) is the initial ( final ) momentum in the c.m.s . of the binary reaction @xmath173 .
using eqs .
( [ tfi ] ) , ( [ iaampl ] ) and ( [ a3andm ] ) one can find @xmath174 solving the system of equations eq . ( [ asb ] ) , one can find finally the following combinations of four amplitudes of the process @xmath175 @xmath176 where @xmath177 as one can see , the amplitudes @xmath75 , @xmath178 , and @xmath179 are determined by the three invariant amplitudes @xmath180 , the form factor @xmath181 and the ratio @xmath182 .
the amplitudes @xmath180 can be measured by the complete polarization experiment described by eqs .
( [ fourpd])-([2todd ] ) .
the transition form factor @xmath181 and the ratio @xmath183 are reasonably well constrained by existing @xmath13 data at moderate transferred momenta @xmath184 mev / c ( corresponding to the kinetic energy of the proton beam of @xmath185 mev )
we have developed a general formalism for double and triple spin - correlations of the reactions @xmath11 and @xmath186 in collinear kinematics in terms of , respectively , four and three independent spin amplitudes , describing these reactions .
a complete polarization experiment is suggested for the reaction @xmath11 .
one possible set of observables is found in eqs .
( [ b1sqr2 ] ) - ( [ realnnnd2 ] ) , ( [ k110031 ] ) , ( [ k11113 - 2 ] ) , ( [ k111130 ] ) and ( [ k111031 ] ) which includes seven t - even observables and three t - odd ones . for the reaction
@xmath186 a complete polarization experiment is described in eqs .
( [ fourpd ] ) - ( [ 2todd ] ) in terms of nine observables related to transversally polarized initial particles .
we showed also that longitudinal observables used in combination with transversal ones could reduce the total number of required measurements for a complete polarization experiment .
on the basis of the impulse approximation for the reaction @xmath186 , three combinations of the invariant amplitudes of the process @xmath146 are expressed in eqs .
( [ 3combin ] ) and ( [ y ] ) in terms of the invariant amplitudes of the reaction @xmath186 .
the formalism can be applied to the planned spin - physics at cosy @xcite . in particular , the applicability of the impulse approximation for the reaction @xmath186 is expected to be valid at beam energies sufficiently high above the threshold , i.e. for @xmath187 gev , because then the transferred momentum is with @xmath188 mev / c fairly small .
several spin observables which are necessary to perform the complete polarization experiments for the reactions @xmath27 and @xmath189 , require a measurement of the polarization of the final @xmath16 .
such a measurement can be performed by measuring the angular distribution of the final particles in the decay @xmath190 , which are determined by the spin - density matrix of the @xmath16 isobar @xcite .
some of the spin - density matrix elements and single spin - correlations , measured in the reaction @xmath191 at beam momenta 3 - 11.8 gev / c , were presented in ref .
besides of physics related to the @xmath16 excitation , obviously this formalism can be applied to other baryon resonances with the spin - parity @xmath192 , in particular , to reactions involving the production of the @xmath193(1385 ) baryon .
here we present the full set of nonzero spin observables for the reaction @xmath111 in collinear kinematics . using @xmath131 determined by eq .
( [ i0 ] ) , one can find from eq .
( [ ktttgen ] ) @xmath194,\\ \label{s002000 } k_{00,20}^{00}\ , i_0=\frac{2\sqrt{3}}{12}\bigl [ re(a_1a_3^ * -a_1a_2^*+a_2a_3^*)\bigr ] , \\
\label{s101000 } k_{10,10}^{00}\,i_0=-\frac{1}{12}\bigl [ 2|a_1|^2-|a_2|^2-|a_3|^2 - 2 re(a_1a_2^*-a_1a_3^*+2a_2a_3^*)\bigr ] , \\
\label{s100010 } k_{10,00}^{10 } \,i_0=-\frac{\sqrt{30}}{180}\bigl [ |a_1|^2 - 5|a_2|^2 - 5|a_3|^2 -4re(a_1a_2^*- a_1a_3^*+2a_2a_3^*)\bigr ] , \\
\label{s001010 } k_{00,10}^{10 } \,i_0=\frac{\sqrt{5}}{60}\bigl [ 2|a_1|^2 + 5|a_2|^2 + 5|a_3|^2 + 2re(-a_1a_2^*+ a_1a_3^*+4a_2a_3^*)\bigr ] , \\
\label{s102010 } k_{10,20}^{10 } \,i_0=-\frac{\sqrt{15}}{180}\bigl [ 4|a_1|^2 - 2|a_2|^2 - 2|a_3|^2 + 2re(a_1a_2^*- a_1a_3^*-7a_2a_3^*)\bigr ] , \\ \label{s1 - 1221 - 1 } k_{1 - 1,22}^{1 - 1}\ , i_0=\frac{\sqrt{10}}{60}\bigl [ 4|a_1|^2+|a_2|^2+|a_3|^2 + 2re(-2a_1a_2^*+ 2a_1a_3^*-a_2a_3^*)\bigr ] , \\ \label{s112 - 110 } k_{11,2 - 1}^{10 } \,i_0=-\frac{\sqrt{5}}{60}\bigl [ -2|a_1|^2+|a_2|^2+|a_3|^2 -re(a_1a_2^*- a_1a_3^*+2a_2a_3^*)\bigr ] , \\ \label{s112011 } k_{11,20}^{1 - 1 } \,i_0=\frac{\sqrt{15}}{180}\bigl [ 4|a_1|^2+|a_2|^2 + 7|a_3|^2 + re(14a_1a_2^ * -2a_1a_3^*-8a_2a_3^*)\bigr ] , \\
\label{s110011 } k_{1 - 1,00}^{11 } \,i_0=-\frac{\sqrt{30}}{180}\bigl [ 2|a_1|^2 + 5|a_2|^2-|a_3|^2 -2re(a_1a_2^ * + 5a_1a_3^*+2a_2a_3^*)\bigr ] , \\
\label{s001111 } k_{00,1 - 1}^{11 } \,i_0=\frac{\sqrt{5}}{120}\bigl [ 8|a_1|^2 - 10|a_2|^2 + 2|a_3|^2 -2re(a_1a_2^ * + 5a_1a_3^ * -4a_2a_3^*)\bigr ] , \\
\label{s102111 } k_{10,21}^{1 - 1 } \,i_0=-\frac{\sqrt{5}}{60}\bigl [ 4|a_1|^2+|a_2|^2 - 5|a_3|^2 + re(5a_1a_2^*+ a_1a_3^*+4a_2a_3^*)\bigr ] , \end{aligned}\ ] ] @xmath195,\\ \label{s002020 } k_{00,20}^{20 } \ , i_0=\frac{\sqrt{3}}{12}\bigl [ |a_2|^2+|a_3|^2 + 2re(a_1a_2^*- a_1a_3^*)\bigr ] , \\
\label{s101020 } k_{10,10}^{20 } \ , i_0=\frac{1}{6}\bigl [ |a_1|^2+|a_2|^2+|a_3|^2 + re(-a_1a_2^ * + a_1a_3^*+a_2a_3^*)\bigr ] , \\
\label{s111120 } k_{1 - 1,11}^{20 } \ , i_0=-\frac{1}{12}\bigl
[ 2|a_1|^2-|a_2|^2-|a_3|^2 + re(a_1a_2^*- a_1a_3^*+2a_2a_3^*)\bigr ] , \\
\label{s002121 } k_{00,2 - 1}^{21 } \ , i_0=\frac{\sqrt{3}}{12}\bigl [ -|a_2|^2+|a_3|^2 -re(a_1a_2^ * + a_1a_3^*)\bigr ] , \\
\label{s111021 } k_{11,10}^{2 - 1 } \ , i_0 = \frac{\sqrt{3}}{12}\bigl [ -|a_2|^2+|a_3|^2 + 2re(a_1a_2^*+ a_1a_3^*)\bigr ] , \\
\label{s101121 } k_{10,11}^{2 - 1 } \ , i_0=-\frac{\sqrt{3}}{12}\bigl [ |a_2|^2-|a_3|^2 + re(a_1a_2^*+ a_1a_3^*)\bigr ] , \\ \label{s002222 } k_{00,2 - 2}^{22 } \ , i_0=-\frac{\sqrt{3}}{12}\bigl [ -|a_2|^2+|a_3|^2 + 2re(a_1a_2^ * + a_1a_3^*)\bigr ] , \\ \label{s111122 } k_{11,11}^{2 - 2 } \ , i_0 = \frac{\sqrt{6}}{12}\bigl [ |a_2|^2-|a_3|^2 + re(a_1a_2^*+ a_1a_3^*)\bigr ] , \\ \label{s112130 } k_{11,2 - 1}^{30 } \ , i_0=\frac{\sqrt{5}}{20}\bigl [ -2|a_1|^2+|a_2|^2+|a_3|^2 -re(a_1a_2^*- a_1a_3^*+2a_2a_3^*)\bigr ] , \\ \label{s102030 } k_{10,20}^{30 } \ , i_0=\frac{\sqrt{15}}{60}\bigl [ 4|a_1|^2 + 3|a_2|^2 + 3|a_3|^2 + 2re(a_1a_2^*- a_1a_3^*-2a_2a_3^*)\bigr ] , \\ \label{s001030 } k_{00,10}^{30 } \ , i_0 = \frac{\sqrt{5}}{10}\bigl [ -|a_1|^2 + re(a_1a_2^*- a_1a_3^*+a_2a_3^*)\bigr ] , \\ \label{s100030 } k_{10,00}^{30 } \ , i_0=\frac{\sqrt{30}}{60}\bigl [ |a_1|^2 -2re(2a_1a_2^*- 2a_1a_3^*-a_2a_3^*)\bigr ] , \\ \label{s001131 } k_{00,1 - 1}^{31 } \ , i_0 = -\frac{\sqrt{30}}{30}\bigl [ |a_1|^2-|a_3|^2 + re(a_1a_2^*+a_2a_3^*)\bigr ] , \\ \label{s110031 } k_{1 - 1,00}^{31 } \ , i_0 = \frac{\sqrt{5}}{30}\bigl [ |a_1|^2 + 2|a_3|^2 + 2re(2a_1a_2^*-a_2a_3^*)\bigr ] , \end{aligned}\ ] ] @xmath196,\\ \label{s102 - 131 } k_{10,2 - 1}^{31 } \ , i_0 = \frac{\sqrt{30}}{30}\bigl [ |a_1|^2-|a_2|^2- re(a_1a_3^ * - a_2a_3^*)\bigr ] , \\ \label{s1 - 12 - 132 } k_{1 - 1,2 - 1}^{32 } \ , i_0=\frac{\sqrt{6}}{12}\bigl [ |a_2|^2-|a_3|^2+re(a_1a_2^*+a_1a_3^*)\bigr ] , \\ \label{s102 - 232 } k_{10,2 - 2}^{32 } \ , i_0=\frac{\sqrt{3}}{12}\bigl [ |a_2|^2-|a_3|^2-re(a_1a_2^*+a_1a_3^*)\bigr ] , \\ \label{s1 - 12 - 233 } k_{1 - 1,2 - 2}^{33 } \ , i_0=-\frac{1}{4}\bigl [ |a_2|^2+|a_3|^2 + 2re(a_2a_3^*)\bigr ] .\end{aligned}\ ] ] for the t - odd observabes we find @xmath197 in addition , we have @xmath198 r. machleidt , k. holinde and c. elster , phys . rept .
* 149 * ( 1987 ) 1 . c. elster , k. holinde , d. schtte , and r. machleidt , phys .
c * 38 * ( 1988 ) 1828 ; c. elster , w. ferchlnder , k. holinde , d. schtte , and r. machleidt , phys .
c * 37 * ( 1988 ) 1647 .
j. haidenbauer , k. holinde , and m.b .
johnson , phys .
c * 48 * ( 1993 ) 2190 .
eyser , r. machleidt , and w. scobel , eur .
j. * a22 * ( 2004 ) 105 .
r. machleidt and i. slaus , j. phys .
* 27 * ( 2001 ) r69 .
h. arenhvel , w. leidemann and e. l. tomusiak , arxiv : nucl - th/0407053 .
m. jetter and h. w. fearing , phys .
c * 51 * ( 1995 ) 1666 [ arxiv : nucl - th/9410040 ] ; f. de jong , k. nakayama and t. s. h. lee , phys .
c * 51 * ( 1995 ) 2334 [ arxiv : nucl - th/9412013 ] ; g.h .
martinus , o. scholten , and j.a .
tjon , phys .
c * 58 * ( 1998 ) 686 .
h. garcilazo and t. mizutani , _
@xmath199 systems _ , ( singapore , world scientific , 1990 ) . j.a .
niskanen , phys .
b * 289 * ( 1992 ) 227 ; phys .
c * 53 * ( 1996 ) 526 .
h. witala et al . , phys .
lett . * 81 * ( 1998 ) 1183 .
kondratyuk , f.m .
lev , and l.v .
schevchenko , yad . phys . * 33 * ( 1981 ) 1208 .
a. boudard and m. dillig , phys .
c * 31 * ( 1985 ) 302 .
uzikov , phys .
elem . chast . at
* 29 * ( 1998 ) 1405 ( phys .
* 29 * ( 1998 ) 583 ) .
f. samarrucca and e.j .
stephenson , phys .
* c64 * ( 2001 ) 034006 .
w. zuo , z.h .
li , a. li , and g.c .
lu , phys .
c * 69 * ( 2003 ) 064001 . c. fuchs , a.fssler , and m.el-shabshiry , phys .
c * 64 * ( 2001 ) 024003 .
wicklund et al . , phys .
d * 34 * ( 1986 ) 19 .
shypit et .
* 60 * ( 1988 ) 901 ; phys .
c * 40 * ( 1989 ) 2203 .
bugg , proceedings of the 3rd international symposium on @xmath200 and @xmath13 physics , lnpi , gatchina ( 1989 ) p.57 .
anisovich , a.v .
sarantsev , and d.v .
bugg , nucl . phys .
* a537 * ( 1992 ) 501 .
n. hoshizaki , phys .
c * 45 * ( 1992 ) r1424 .
f. sammarruca , phys .
c * 50 * ( 1994 ) 652 .
t. mizutani , c. fayard , g.h .
lamot , and b. saghai , phys .
c * 47 * ( 1993 ) 56 .
h. sakai et al .
* 84 * ( 2000 ) 5288 .
k. sekiguchi et al . , phys .
* c 65 * ( 2002 ) 034003 .
s. yaschenko et al .
* 91 * ( 2005 ) 072304 .
w. de boer et al .
* 34 * ( 1975 ) 558 .
ditzler , et al . ,
phys . rev .
d * 27 * ( 1983 ) 680 .
f. perrot et al . , nucl . phys .
* b278 * ( 1986 ) 881 .
auer et al . , phys .
* 41 * ( 1978 ) 354 ; i.p .
auer et al . , phys .
* 51 * ( 1983 ) 1411 .
n. hoshizaki , prog.theor . phys .
* 57 * ( 1977 ) 1099 .
* 61 * ( 1979 ) 129 .
r.bhandari , r.a .
arnd , l.d .
roper , b.j .
verwest , phys .
46 * ( 1981 ) 1111 . c. lechanoine - leluck and f. lehar , rev .
* 65 * ( 1993 ) 47 .
kloet and r.r .
silbar , phys .
* 45 * ( 1980 ) 970 ; nucl .
a364 * ( 1981 ) 346 . j. niskanen , phys .
lett.*112 b * ( 1982 ) 17 .
e. tomasi - gustaffson and m.p .
rekalo , fiz .
elem . chat . at.yadr . * 33 * ( 2002 ) 436 ( phys . part.nucl . * 33 * ( 2002 ) 220 ) .
rekalo , n.m .
piskunov , and i.m .
sitnik , few-bod.syst .
* 23 * ( 1998 ) 187 .
bilenky , l.l .
lapidus , l.d .
puzikov and r.m .
ryndin , nucl . phys .
* 7 * ( 1958 ) 646 .
uzikov , phys .
b * 595 * ( 2004 ) 277 . yu.n .
uzikov , arxiv : nucl - th/0411113 .
bugg and c. wilkin , nucl .
phys . a*467 * ( 1987 ) 575 . c. ellegaard et al .
* 59 * ( 1987 ) 974 .
t. sams et al . , phys .
c * 51 * ( 1995 ) 1945 .
varschalovich , a.n .
moskalev , and v.k .
khersonskiy , kvantovaya teoriya uglovogo momenta , `` nauka '' , leningrad , 1975 , 439 p. ( in russian ) ; singapore , world scientific , 1988 ( translated to english ) .
a. kacharava , f. rathmann , and c. wilkin for the anke collaboration , proposal # 152 : _
spin physics from cosy to fair _ ; available from . | a general formalism for double and triple spin - correlations of the reaction @xmath0 is developed for the case of collinear kinematics . a complete polarization experiment allowing to reconstruct all of the four amplitudes describing this process is suggested .
furthermore , the spin observables of the inelastic charge - exchange reaction @xmath1 are analyzed in collinear kinematics within the single @xmath2 scattering mechanism involving the subprocess @xmath3 .
the full set of spin observables related to the polarization of one or two initial particles and one final particle is obtained in terms of three invariant amplitudes of the reaction @xmath4 and the transition form factor @xmath5 . a complete polarization experiment for the reaction @xmath6
is suggested which allows one to determine three independent combinations of the four amplitudes of the elementary subprocess @xmath0 . |
the complete @xmath0-dependence formulas for the nuclear recoil corrections to the energy levels of hydrogenlike atoms in the case of a point nucleus were first obtained by a quasipotential method [ 1 ] and subsequently rederived by different approaches [ 2 - 4 ] . according to [ 4 ] ,
the nuclear size corrections to the recoil effect can be partially included in these formulas by a replacement of the pure coulomb potential with the potential of an extended nucleus . the total recoil correction for a state @xmath4 of a hydrogenlike atom is conveniently written as the sum of a low - order term @xmath5 and a higher - order term @xmath6 [ 1 ] , where ( @xmath7 ) @xmath8|a\rangle\,,\\ \delta e_{h}&=&\frac{i}{2\pi m}\int_{-\infty}^{\infty}d\omega\ , \langle a|\bigl({\bf d}(\omega)- \frac{[{\bf p},v]}{\omega+i0 } \bigr)\nonumber\\ & & \times g(\omega + \varepsilon_{a } ) \bigl({\bf d}(\omega)+\frac{[{\bf p},v]}{\omega+i0}\bigr ) unperturbed state of the dirac electron in the nuclear potential @xmath9 , @xmath10 is the momentum operator , @xmath11 is the relativistic coulomb green function , @xmath12 , @xmath13 , @xmath14 are the dirac matrices , @xmath15 and @xmath16 is the transverse part of the photon propagator in the coulomb gauge : @xmath17 in equation ( 2 ) , the scalar product is implicit . for point - like nuclei , @xmath18 .
if extended nuclei are considered , @xmath9 is the potential of the extended nucleus in eq .
( 2 ) and in calculating @xmath19 , @xmath20 , and @xmath21 .
therefore , the nuclear size corrections are completely included in the coulomb part of the recoil effect . in the one - transverse - photon part and the two - transverse - photon part ( see ref . [ 4 ] ) , they are only partially included . at least for high @xmath1
we expect that this procedure accounts for the dominant part of the nuclear size effect since using the extended nucleus wave function and the extended nucleus green function strongly reduces the singularities of the integrands in ( 1 ) and ( 2 ) in the nuclear region .
the term @xmath5 contains all the recoil corrections within the @xmath22 approximation .
its calculation for a point nucleus , based on the virial relations for the dirac equation [ 5 - 7 ] , yields [ 1 ] @xmath23 where @xmath24 is the dirac electron energy for the point nucleus case .
@xmath6 contains the contribution of order @xmath25 and all contributions of higher order in @xmath0 .
to lowest order in @xmath0 , this term represents the salpeter correction [ 8 ] .
the calculation of this term to all orders in @xmath0 was performed in [ 9,10 ] for the case of a point nucleus . according to these calculations , the recoil correction to the lamb shift of the 1s state in hydrogen
constitutes 7.1(9 ) khz , in addition to the salpeter term .
this value is close to the @xmath26 correction ( 7.4 khz ) found in [ 3 ] and is clearly distinct from a recent result for the @xmath26 correction ( 16.4 khz ) obtained in [ 11 ] .
( the @xmath27 corrections cancel each other [ 12,13 ] . )
the total recoil correction to the ground state energy in @xmath2u@xmath3 was calculated in [ 9 ] to be 0.51 ev . in this work
we calculate the recoil correction to the ground state energy of hydrogenlike atoms in the range @xmath1=1110 using the formulas ( 1 ) and ( 2 ) employing the potential of an extended nucleus .
using the virial relations for the dirac equation in a central field [ 7 ] , the formula ( 1 ) can be transformed to ( see appendix ) @xmath28\,,\end{aligned}\ ] ] where @xmath19 and @xmath24 are the dirac electron energies for an extended nucleus and the point nucleus , respectively , @xmath29 is the relativistic angular quantum number , @xmath30 is the deviation of the nuclear potential from the pure coulomb potential , and @xmath31 and @xmath32 are the pauli matrices . here
, the notations for the radial matrix elements from [ 7 ] are used : @xmath33 u(r)\ , dr\,,\\ ( a|\sigma_{z}u|b)&=&\int_{0}^{\infty}[g_{a}(r)g_{b}(r)-f_{a}(r)f_{b}(r ) ] u(r)\ , dr\,,\\ ( a|\sigma_{x}u|b)&=&\int_{0}^{\infty}[g_{a}(r)f_{b}(r)+f_{a}(r)g_{b}(r ) ] u(r)\ , dr\,.\end{aligned}\ ] ] @xmath34 and @xmath35 are the radial components of the dirac wave function for the extended nucleus , which are defined by @xmath36 the first term on the right side of equation ( 4 ) corresponds to the low - order recoil correction for the point nucleus ( see eq .
the second term gives the nuclear size correction .
we calculate this term for the uniformly charged nucleus . in table
i , we display the results of this calculation for the the 1s state .
the values are expressed in terms of the function @xmath37 which is defined by @xmath38 in order to compare the nuclear size correction to the low - order term with the corresponding correction to the higher - order term ( see the next section ) , in the last column of the table i we display the value @xmath39 which is defined by @xmath40 using eq .
( 4 ) , one easily finds for an arbitrary @xmath41s state and for very low @xmath1 ( @xmath42 ) @xmath43\,,\end{aligned}\ ] ] where @xmath44 is the radius of the uniformly charged nucleus .
the first term in ( 10 ) is a pure nonrelativistic one .
it describes the reduced mass correction to the nonrelativistic nuclear size effect .
so , if the nuclear size correction to the energy level is calculated using the reduced mass , this term must be omitted in equation ( 10 ) . the second term , which is dominant , arises from the coulomb part ( @xmath45 ) . for the standard parametrization of the proton form factor @xmath46 which corresponds to @xmath47 and @xmath48\,,\end{aligned}\ ] ]
the contribution of this term to @xmath49 is @xmath50 we will see in the next section that this term cancels with the corresponding correction to the coulomb part of the higher - order term .
this implies that the sum of the low - order and higher - order contributions is more regular at @xmath51 than each of them separately .
to calculate the higher - order term ( 2 ) we transform it in the same way as it was done in [ 9 ] .
the final expressions are given by the equations ( 41)-(54 ) of ref .
[ 9 ] where the pure coulomb potential ( @xmath52 ) in the equations ( 42 ) and ( 48 ) has to replaced by the potential of the extended nucleus @xmath9 .
we calculate these expressions for the uniformly charged nucleus by using the finite basis set method with the basis functions constructed from b - splines [ 14 ] .
the algorithm of the numerical procedure is the same as it is described in [ 9 ] .
the results of the calculation for the 1s state are presented in the second column of the table ii .
they are expressed in terms of the function @xmath53 defined by @xmath54 for comparison , in the third column of this table we list the point - nucleus results ( @xmath55 ) that are obtained by the corresponding calculation for @xmath56 .
these point - nucleus results are in good agreement with our previous results from [ 9 ] . in the fourth column of the table ,
the difference @xmath57 is listed .
finally , in the last column the salpeter contribution [ 8,15 ] @xmath58 is displayed . for low @xmath1
the nuclear size correction to the higher - order term is mainly due to the coulomb contribution @xmath59g(\omega+\varepsilon_{a } ) [ { \bf p},v]|a\rangle\,.\end{aligned}\ ] ] it is comparable with the deviation of the complete @xmath0-dependence value from the salpeter contribution ( in the case of hydrogen @xmath60 while @xmath61 ) . to check this result let us calculate the finite nuclear size correction to the coulomb part of the @xmath62 contribution . taken to the lowest order in @xmath0 , formula ( 17 ) yields @xmath63 where @xmath64 is the non - relativistic wave function at @xmath65 and @xmath66 is the nuclear potential in the momentum representation . using the standard parametrization of the proton form factor @xmath67 and separating the point nucleus result from ( 18 )
, we can write for an @xmath41s state @xmath68 where @xmath69\,.\end{aligned}\ ] ] evaluation of this integral to the lowest order in @xmath70 yields @xmath71 as we noted above , the correction ( 22 ) cancels with the corresponding correction to the low - order term ( see eq . ( 14 ) ) . for @xmath72 fm
[ 16 ] , which corresponds to @xmath73 mev , the formula ( 22 ) yields @xmath74 while the exact calculation of the integral ( 21 ) amounts to @xmath75 .
these results are in good agreement with the corresponding result ( @xmath76 ) from the table ii .
in this work we have calculated the recoil correction to the ground state energy of hydrogenlike atoms for extended nuclei in the range @xmath77 .
this correction is conveniently written in the form @xmath78 the function @xmath79 is shown in fig .
1 . for comparison ,
the point nucleus function @xmath55 and the salpeter function @xmath80 are also presented in this figure .
the table iii displays the values of the recoil corrections ( in ev ) in the range @xmath1=10110 . in the case of hydrogen
we find that the recoil correction amounts to @xmath81 khz beyond the salpeter contribution and the nonrelativistic nuclear size correction to the recoil effect ( the first term in eq .
it almost coincides with the point nucleus result .
this is caused by the fact that the nuclear size correction to the higher - order term ( eq .
( 22 ) ) and the relativistic nuclear size correction to the low - order term ( eq .
( 14 ) ) cancel each other . for high @xmath1 , where the @xmath0 expansion as well as the reduced mass approximation are not valid any more
, we should not separate any contributions from the total recoil effect . in the case of hydrogenlike uranium ( @xmath2u@xmath3 )
, the total recoil correction constitutes @xmath82 ev and is by 10 % smaller than the corresponding point nucleus value ( @xmath83 ev ) found in [ 9 ] .
this improvement affects the current numbers of the lamb shift prediction [ 17 ] .
finally , we note a very significant amount of the nuclear size effect for @xmath1=110 . according to the table iii ,
the finite nuclear size modifies the point nucleus result by more than 40% .
valuable conversations with s.g .
karshenboim , p. mohr , k. pachucki , and a.s .
yelkhovsky are gratefully acknowledged .
v. m. s. thanks the institut fr theoretische physik at the technische universitt dresden for the kind hospitality .
the work of v. m. s. , a. n. a. , and v. a. y. was supported in part by grant no .
95 - 02 - 05571a from rfbr .
also we gratefully acknowledge support by bmbf , daad , dfg , and gsi .
t. b. and .
g. p. express their gratitude to the department of physics at the st .
petersburg state university , where they have been welcome in a very friendly atmosphere .
using the identity @xmath84 , the coulomb part of the low - order term can be written as @xmath85\,.\end{aligned}\ ] ] as described in detail in [ 18 ] , the breit part of the low - order term can be transformed to @xmath86|a\rangle \nonumber\\ & = & -\frac{1}{2m}\langle a|\frac{\alpha z}{r } \bigl(\mbox{\boldmath $ \alpha$}+\frac{(\mbox{\boldmath $ \alpha$ } \cdot { \bf r})}{r^2}\bigr)\cdot{\bf p}|a\rangle\nonumber\\ & = & -\frac{\alpha z}{2m}\langle a|\frac{1}{r } \bigl(2\varepsilon_{a}-2\beta m-2v+\frac{i \kappa}{r } \alpha_{r}\beta\bigr)|a\rangle\nonumber\\ & = & \frac{1}{2m}\bigl[2\alpha z\langle a|v / r|a\rangle -2\alpha z\varepsilon_{a } \langle a|1/r|a\rangle \nonumber\\ & & + \,2\alpha z\langle a|m\beta / r|a\rangle+ 2\kappa \alpha z\int_{0}^{\infty}g_{a}f_{a}dr\,,\end{aligned}\ ] ] where @xmath87 and @xmath29 is the relativistic angular quantum number of the state @xmath4 . in the following
we will use the notations of ref . [
7 ] , @xmath88 where @xmath34 and @xmath35 are the radial components of the dirac wave function for the extended nucleus , and the radial scalar product defined by the equations ( 5)-(7 ) . using the equation ( 2.9 ) of ref .
[ 7 ] , we find @xmath89\,,\end{aligned}\ ] ] where @xmath90 .
> from the equations ( 2.8)-(2.10 ) of ref .
[ 7 ] , one obtains @xmath91 substituting ( 30 ) and ( 31 ) into ( 29 ) , we find @xmath92\,.\end{aligned}\ ] ] separating the point nucleus result from the right side of ( 32 ) , we get the equation ( 4 ) . 15 v.m .
shabaev , teor . mat .
fiz . * 63 * , 394 ( 1985 ) ( theor .
phys . * 63 * , 588 ( 1985 ) ) ; in : papers at first soviet - british symposium on spectroscopy of multicharged ions ( academy of sciences , troitsk , 1986 ) , pp .
238 - 240 .
yelkhovsky , preprint binp 94 - 27 ( budker inst . of nuclear physics ,
novosibirsk , 1994 ) ; hep - th/9403095 ( 1994 ) .
k. pachucki and h. grotch , phys .
rev . a * 51 * , 1854 ( 1995 ) .
shabaev , phys .
rev . a ( in press ) ; physics/9705039 .
j.h . epstein and s.t .
epstein , am .
j. phys . * 30 * , 266 ( 1962 ) .
shabaev , vestn .
n4 , 15 ( 1984 ) .
shabaev , j. phys .
b * 24 * , 4479 ( 1991 ) .
salpeter , phys . rev . * 87 * , 328 ( 1952 ) ; h.a .
bethe and e.e .
salpeter , quantum mechanics of one- and two - electron atoms ( springer , berlin , 1957 ) .
artemyev , v.m .
shabaev , and v.a .
yerokhin , phys .
a * 52 * , 1884 ( 1995 ) .
artemyev , v.m .
shabaev , and v.a .
yerokhin , j. phys .
b * 28 * , 5201 ( 1995 ) .
yelkhovsky , zhetf ( in press ) , physics/9706014 .
khriplovich , a.i . milstein and a.s .
yelkhovsky , phys .
t * 46 * , 252 ( 1993 ) .
r.n . fell , i.b .
khriplovich , a.i . milstein and a.s .
yelkhovsky , phys .
a * 181 * , 172 ( 1993 ) .
johnson , s.a .
blundell , and j. sapirstein , phys .
rev . a * 37 * , 307 ( 1988 ) .
erickson and d.r .
yennie , ann.phys .
( ny ) * 35 * , 271 ( 1965 ) ; g.w .
erickson , in : physics of one- and two - electron atoms , eds .
f. bopp and h. kleinpoppen ( north - holland , amsterdam , 1970 ) .
simon , c. schmidt , f. borkowski , and v.h .
walther , nucl .
a * 333 * , 381 ( 1980 ) .
t. beier , g. plunien , and g. soff , hyperfine interaction * 108 * , 19 ( 1997 ) ; t. beier , p.j .
mohr , h. persson , g. plunien , m. greiner , and g. soff , phys .
a , in press .
v.m . shabaev and a.n .
artemyev , j. phys .
b * 27 * , 1307 ( 1994 ) . g. fricke , c. bernhardt , k. heilig , l.a .
schaller , l. schellenberg , e.b .
shera , and c.w .
de jager , at . data and nucl .
data tables * 60 * , 177 ( 1995 ) . h. de vries , c.w .
de jager , and c. de vries , at .
data and nucl .
data tables * 36 * , 495 ( 1987 ) .
johnson and g. soff , at .
data and nucl .
data tables * 33 * , 405 ( 1985 ) .
zumbro , r.a .
naumann , m.v .
hoehn , w. reuter , e.b .
shera , c.e .
bemis jr . , and y. tanaka , phys . lett . * 167 b * 383 ( 1986 ) .
zumbro , e.b .
shera , y. tanaka , c. e. bemis jr .
naumann , m.v .
hoehn , w. reuter , and r.m .
steffen , phys .
* 53 * , 1888 ( 1984 ) .
review of particle properties , phys .
d * 50 * , 1173 ( 1994 ) .
nuclear size correction to the low - order term for the 1s state expressed in terms of the functions @xmath37 and @xmath39 , defined by equations ( 8) and ( 9 ) , respectively .
the values of the nuclear radii employed in the calculation are taken from [ 16,19 - 23 ] . [
cols="^,<,<,<",options="header " , ] | the recoil correction to the ground state energy of hydrogenlike atoms is calculated to all orders in @xmath0 in the range @xmath1 = 1110 .
the nuclear size corrections to the recoil effect are partially taken into account . in the case of hydrogen ,
the relativistic recoil correction beyond the salpeter contribution and the nonrelativistic nuclear size correction to the recoil effect , amounts to 7.2(2 ) khz .
the total recoil correction to the ground state energy in hydrogenlike uranium ( @xmath2u@xmath3 ) constitutes 0.46 ev .
recoil correction to the ground state energy of hydrogenlike atoms + v.m .
shabaev + _ max - planck - institut fr physik komplexer systeme , nthnitzer str .
38 , d-01187 , dresden , germany _ a.n .
artemyev + _ department of physics , st.petersburg state university , _ + _ oulianovskaya 1 , petrodvorets , st.petersburg 198904 , russia _
t. beier , g. plunien , v.a .
yerokhin , and g. soff + _ institut fr theoretische physik , technische universitt dresden , mommsenstrasse 13 , d-01062 dresden , germany _ pacs
number(s ) : 12.20.-m , 31.30.jv , 31.30.gs |
green [ gr ] ( also see [ ch ] ) showed that the basic structure theory for finite - dimensional algebras carries over to coalgebras , with injective indecomposable comodules replacing projective indecomposables .
define the ( _ gabriel- _ or _ _
ext- _ _ ) _ quiver _ of a coalgebra @xmath3 to be the directed graph @xmath20 with vertices @xmath21 corresponding to isoclasses of simple comodules and @xmath22 arrows from @xmath23 to @xmath24 , for all @xmath25 .
the blocks of @xmath3 are ( the vertices of ) components of the undirected version of @xmath20 . in other words ,
the blocks are the equivalence classes of the equivalence relation on @xmath21 generated by arrows .
the indecomposable or block coalgebras are the direct sums of injective indecomposables having socles from a given block .
write @xmath26 as a direct sum of indecomposable injectives with multiplicities .
let @xmath27 denote the direct sum of the indecomposables where each indecomposable occurs with multiplicity one .
the comodule @xmath27 is called a basic injective for @xmath3 .
@xmath27 can be written as @xmath28 where @xmath29 is an idempotent in the dual algebra .
define the _ basic coalgebra _ to be the coendomorphism coalgebra @xmath30 basic coalgebras were first constructed by simson [ sim ] and were rediscovered in [ cmo ] and [ wo ] . as observed in [ cg ] , @xmath31 is more simply described as the coalgebra @xmath32 ( or just @xmath33 ) , which is a noncounital homomorphic image of @xmath3 .
of course @xmath31 is morita - takeuchi equivalent to @xmath3 , meaning that their comodule categories are equivalent . if _
k _ is algebraically closed , then @xmath31 is a pointed coalgebra .
let @xmath34 be a quiver ( not necessarily finite ) with vertex set @xmath35 and arrow set @xmath36 for a path @xmath37 , let @xmath38 denote the start ( or source ) of @xmath37 and let denote @xmath39 its terminal ( or target ) .
the * path coalgebra * @xmath40 of @xmath34 is defined to be the span of all paths in @xmath34 with coalgebra structure @xmath41{c}\delta(p)=\sum_{p = p_{2}p_{1}}p_{2}\otimes p_{1}+t(p)\otimes p+p\otimes s(p)\\ \varepsilon(p)=\delta_{|p|,0}\end{array}\ ] ] where @xmath42 is the concatenation @xmath43of the subpaths @xmath44 and @xmath45 ( @xmath46)@xmath47 here @xmath48 denotes the length of @xmath37 and the starting vertex of @xmath49 is required to be the end of @xmath50 the paths @xmath51 occuring in the sum are subpaths of @xmath37 with the same terminal @xmath52 the span of these subpaths ( together with @xmath39 ) span the coideal ( i.e. subcomodule ) generated by @xmath53 similarly , any subcoalgebra containing a path @xmath37 contains all subpaths .
thus @xmath54 are group - like elements , and if @xmath55 is an arrow @xmath56 , with @xmath57 then @xmath55 is a @xmath58- skew primitive , i.e. , @xmath59 it is apparent that @xmath40 is pointed with coradical @xmath60and the degree one term of the coradical filtration is @xmath61 a subcoalgebra @xmath3 of the path coalgebra k@xmath34 is said to be _ admissible _ if @xmath3 contains @xmath35 and @xmath36 there is a coalgebraic version ( with no finiteness restrictions ) of a fundamental result of gabriel for finite dimensional algebras : ( [ cmo][wo ] ) every coalgebra @xmath3 over an algebraically closed field is morita - takeuchi equivalent to an admissible subcoalgebra of @xmath62 .
let @xmath63be an admissible subcoalgebra .
@xmath65 @xmath31 is a k - linear combination of distinct paths @xmath66 with @xmath67 for all @xmath68 , we say that @xmath69 is a _ reduced element _ , and we denote the support of @xmath69 by @xmath70 we shall also say that @xmath66 _ appears in _ @xmath31 and write @xmath71 if @xmath66 @xmath65@xmath72 for some @xmath73 clearly @xmath74 if and only if @xmath75 for all paths @xmath76 exactly in the case where @xmath31 is the span of a set of paths . definition : : let @xmath3 be a coalgebra with gabriel quiver @xmath34 and basic coalgebra @xmath77@xmath78 @xmath3 is said to be a * string coalgebra * if 1 .
every vertex of @xmath34 is the start of at most two arrows and the end of at most two arrows . 2 .
given any arrow @xmath79 , there is at most one arrow @xmath80 such that @xmath81 @xmath31 and at most one arrow @xmath82 such that @xmath83 @xmath84 @xmath85 3 .
@xmath31 is spanned by a set of paths ( containing the vertices and arrows of @xmath34 ) .
if the last condition s3 is dropped , we say that @xmath3 is a * special biserial coalgebra .
* string algebras and special biserial algebras are defined dually , given a possibly infinite bound quiver @xmath86 , see [ br ] , [ erd ] , [ sw ] .
string coalgebras have been defined for coalgebras specified by an ideal of relations in the path algebra in [ sim3 ] .
we make a few more elementary remarks .
let @xmath87 be a pointed coalgebra .
we let @xmath88 be an injective indecomposable in @xmath89 , having socle @xmath90 corresponding to the vertex ( group - like ) @xmath91 it is useful to note that an injective hull of the simple right k@xmath34 comodule @xmath90 is k@xmath92 where @xmath93 is the idempotent dual to @xmath24 in the dual algebra @xmath94k@xmath95 using the right hit action .
it is easy to see that k@xmath92 is simply the span of all paths ending at @xmath24 , and also that @xmath96 @xmath97k@xmath98 if @xmath31 is special biserial , then we can be specific about the form of @xmath88 .
let @xmath87 be a special biserial coalgebra and let @xmath88 be a finite - dimensional injective indecomposable right @xmath31-comodule .
then one of the following cases holds:(a ) @xmath88 is a uniserial comodule , generated by a path.(b ) @xmath88 is the sum of two distinct uniserial comodules , each generated by a path , where the only common vertex of the two paths is @xmath24 ( c ) @xmath88 is the sum of two distinct uniserial comodules , each generated by a path , where the only common vertices of the two paths is @xmath24 and a common starting vertex .
( d ) @xmath88 is generated by @xmath99 @xmath100 where @xmath37 and @xmath101 are distinct parallel paths , both ending at @xmath102 with the same starting vertex , and no other common vertices .
suppose there is only one arrow only ending at @xmath24 .
then condition s2 in the definition implies ( a ) .
now assume that there are two arrows ending at @xmath24 . if there are no common vertices other than @xmath24 , we are in case ( b ) .
otherwise there are two paths @xmath103 in @xmath88 having a common vertex other than @xmath24 , and then condition s2 forces it to be the common starting vertex , as in ( c ) and ( d ) .
if @xmath88 is spanned by paths , then @xmath88 is as in ( c ) .
consider the case where @xmath88 is not spanned by paths . in this case
@xmath31 contains a linear combination , say @xmath104 of at least two paths not in @xmath105 where each path involved has terminal vertex @xmath24 .
conditions s1 and s2 imply that the support @xmath106 of this element consists of exactly two paths ( say ) @xmath37 and @xmath101 with @xmath107 and @xmath108 .
thus @xmath88 contains an element @xmath109 as in the statement .
it follows easily that @xmath110 ( otherwise we obtain the contradiction @xmath111 ) , and as already stated , @xmath38 is the only common vertex of @xmath37 and @xmath112 it remains to see that @xmath88 is generated by @xmath113 if @xmath114 is a linear combination of paths not in @xmath115 with support size greater than @xmath116 , then it follows as just argued that the supp@xmath72 consists of exactly two paths .
furthermore these two paths must be precisely @xmath37 and @xmath112 this forces @xmath117 to be a scalar multiple of @xmath118 on the other hand , if @xmath119 is a path , then @xmath120 and s1 forces the terminal arrow of @xmath121 to equal the terminal arrow of @xmath37 or @xmath101 ( unless @xmath122 ) .
now s2 forces @xmath121 to be a proper subpath of @xmath37 or @xmath112 this means that , say , @xmath123 for a nonntrivial path @xmath124 , and thus @xmath125 this shows that @xmath88 is generated by @xmath99 completing the proof of ( d ) .
it is easy to see that all cases in the proposition above can occur . by duality and the symmetry in the definition of special biserial coalgebras
, it follows that the comodules in case ( d ) of the proposition above are finite - dimensional and projective . in case ( b ) and ( c ) ,
the top of the injective comodule is _ spanned _ by the two generating paths ( and thus is not projective ) . in case ( d ) , the top is spanned by the image of the reduced element @xmath109 .
the only other case where @xmath88 might be projective is in the finite - dimensional uniserial case , where @xmath88 is generated as a comodule by a single path , as in part ( a ) .
it is a simple matter to construct examples of such uniserial injectives . for a comodule @xmath126 , we let @xmath127 denote the intersection of all maximal subcomodules of @xmath126 .
since submodules of rational @xmath128modules are rational , this coincides with the standard usage .
the next lemma may be well - known .
let @xmath3 be a coalgebra with finite - dimensional right c - comodule @xmath126 .
let @xmath129be the jacobson radical of the algebra @xmath130 then @xmath131 it is elementary that @xmath132 , so we shall prove the other inclusion .
obviously @xmath133 is annihilated by @xmath134 .
therefore since @xmath133 is a rational @xmath135-module , it follows that the coefficent space cf@xmath136 . thus @xmath133 is a completely reducible @xmath3-comodule , so we conclude that @xmath137 as desired .
the following lemma is a coalgebraic version of results that seem to be well - known for finite - dimensional algebras , cf .
[ erd , ii.1.3 ] .
it is used in the proof of the subsequent result .
let @xmath138 be a coalgebra where @xmath139 and @xmath34 are injective right subcomodules and @xmath139 is the direct sum of finite - dimensional indecomposable projective right comodules.(a ) assume @xmath34 has no projective summands .
then @xmath140 is a subcoalgebra of @xmath3.(b ) let @xmath126 be an indecomposable right @xmath3-comodule , and assume @xmath139 is indecomposable with @xmath141 then @xmath142 \(a ) let @xmath143rad@xmath144 and suppose that @xmath145 is not a subcoalgebra of @xmath146 then @xmath145 is not a left @xmath3-comodule , so it not a right @xmath135-submodule and there exists @xmath147 such that @xmath148 this right hit action extends to a right comodule ( and left @xmath135-module ) map @xmath149 such that @xmath150 write @xmath151 and @xmath152 for the respective krull - schmidt decompositions of @xmath139 and @xmath34 . by the previous lemma
, we have @xmath153rad@xmath154 by projecting to @xmath139 we see that there is a comodule map from @xmath34 to @xmath139 whose image is not contained in @xmath155@xmath156 . therefore , for some indices
@xmath157 there exists a nonzero projection of @xmath158 onto @xmath159 this implies that @xmath34 has projective summand @xmath160 .
this concludes the proof of ( a ) .
\(b ) we shall show that there exists @xmath161 such that @xmath162 induces a surjection @xmath163 indeed , the hypothesis yields @xmath164 with @xmath165 with the @xmath166 linearly independent and @xmath167rad@xmath139.@xmath168 for some @xmath169 .
let @xmath170 , in a dual basis for @xmath171 obtained by extending the @xmath166 to a basis for @xmath172 then the right @xmath3-comodule mapping @xmath173 corresponds to @xmath162 , and we have @xmath174 composing with the projection of @xmath3 onto @xmath139 yields a surjection of @xmath126 onto @xmath175 since rad@xmath139 is the unique maximal subcomodule of @xmath139 .
since @xmath126 is indecomposable , this forces @xmath176 as desired .
this completes the proof of the lemma .
the representation theory of special biserial coalgebras is closely related to that of string coalgebras .
if @xmath3 is special biserial , we can remove the reduced linear combinations of paths and pass to a slightly smaller string coalgebra @xmath177 losing only the obvious representations in the process . to be precise ,
let @xmath178 @xmath179 is projective } and put @xmath180 let @xmath87 be a special biserial coalgebra.(a ) then @xmath181 is a string coalgebra ; in addition : @xmath182(@xmath183 @xmath184 ( b ) the ar quiver of @xmath31 is obtained from the ar quiver for @xmath181 by attaching an arrow from @xmath185 to @xmath88 and an arrow from @xmath88 to @xmath186 for all @xmath187 .
let @xmath31 be as in the hypothesis .
inspection of the list of possible forms for the @xmath88 in proposition 1.2.1 shows that the ones of type ( d ) are exactly the injective - projectives which are not uniserial and which are not spanned by paths .
it is easy to see that the radical of these @xmath88 s are spanned by the proper subpaths of @xmath37 and @xmath101 ending at @xmath24 .
the uniserial projectives in case ( a ) which might occur are generated by a single path , and clearly their radicals are also spanned by the proper subpaths .
thus we see that @xmath188 is spanned by paths , and therefore that @xmath181 is a string coalgebra .
part ( b ) of the preceding lemma shows that the @xmath105 @xmath189 are the only indecomposables not already in @xmath190 it remains to prove that @xmath181 is a subcoalgebra of @xmath31 .
this assertion follows directly from part ( a ) of the preceding lemma .
part ( b ) follows from a standard argument ( see e.g. [ ars ] ) that shows that the only almost split sequences with a projective - injective occurring as a summand of the middle term is of the form @xmath191 this concludes the proof .
one can replace @xmath192 by @xmath193 to obtain a larger string coalgebra in @xmath31 .
[ erd ] , ii.1.3 , ii.6.4 ) .
if @xmath31 is self - projective ( i.e. quasi - cofrobenius [ din ] ) to begin with , then we obtain simply rad(@xmath31 ) as a string coalgebra .
consider a semiperfect string coalgebra @xmath3 . by the theory of string algebras mentioned in the introduction , the category of finite - dimensional comodules @xmath194 consists of _ string _ modules and _ band _ modules which are locally nilpotent as quiver representations .
evidently , the semiperfect assumption is unneeded for the classification of indecomposables , though it is relevant to the discussion of the ar quiver .
here is a very basic example .
let @xmath34 be a single loop with vertex @xmath24 and arrow @xmath80 .
then @xmath195 is a string coalgebra .
however , the path algebra k@xmath196 $ ] is not a string algebra ( as defined in [ br ] , p. 157 ) because of the assumption that amounts to the coalgebra being semiperfect .
the finite dimensional modules are given by string and band modules .
@xmath197the band modules corresponding to the band @xmath80 and nonzero scalar parameters @xmath198 do not correspond to comodules .
these are the indecomposable modules annihilated by some power of @xmath199 @xmath200 .
for the main example concerning quantum @xmath1 in this article in section 2 below , there are no bands , and we shall only be concerned with string modules , which we presently describe .
the string modules that will arise in section 2 will be relatively simple to describe , but for completeness we give the general set - up .
* _ letters _ are arrows in @xmath201 or formal inverses of arrows , denoted by @xmath202 for @xmath203 we also set @xmath204 @xmath205 and @xmath206 .
a _ word _ is defined to be a sequence @xmath207 of letters with @xmath208 @xmath209 and @xmath210 for @xmath211 , @xmath212 .
note that the empty word of length @xmath213 is allowed .
the formally inverted word @xmath214 is defined to be @xmath215 . * _ strings _ are defined to be the equivalence class of words where each subpath ( or its inverse ) is in @xmath31 , under the relation that identifies each word with its inverse . *
while we will not need them in section 2 ( except to say that there are not any of them ) , we mention that _ bands _ are defined to be equivalence classes of words , under cyclic permutation , whose powers @xmath216 , @xmath217 are defined ( but are not themselves powers of words ) and such that every subpath of the of every power @xmath216 is in @xmath31 .
one - parameter families of modules are associated to bands ( [ erd ] , [ br ] , [ ri ] , [ gp ] ) . * a _ string module _
st(@xmath218 ) is associated to each string representative @xmath218 by associating a quiver of type @xmath219 to the string @xmath218 with edges labeled by the letters @xmath220 and having the arrows pointing to the left if @xmath218 is an arrow and to the right otherwise .
the obvious representation of @xmath34 corresponding to the diagram is specified as follows .
* let @xmath221 be a string representative where the @xmath222 are arrows and the @xmath223 are @xmath224 or @xmath116 .
consider the k - vector space @xmath225 with basis @xmath226 ... @xmath227 define @xmath228 for each @xmath229 to be the span of the following basis elements @xmath230 together with @xmath231 if either : @xmath232 and @xmath233 or @xmath234 and @xmath235 in other words , we partition the basis so that @xmath236 @xmath237 and @xmath238 the arrows act by @xmath239 if @xmath240 and @xmath241 if @xmath242 .
all arrows not yet defined act as zero .
it is not hard to see that the representation does not depend on the representative of the string , i.e. st@xmath243st@xmath244 .
it is clear from the construction that string modules are locally nilpotent and hence correspond to @xmath31-comodules .
we shall refer to such comodules as _ string comodules_. * as an example consider the kronecker quiver @xmath245 with arrows @xmath246 and vertices @xmath247 as shown .
the string @xmath248 corresponds the @xmath249 quiver @xmath250 or more suggestively , @xmath251{lllllllll}$v_{4}$ & & & & $ v_{2}$ & & & & $ v_{0}$\\ & $ \overset{b}{\searrow}$ & & $ \overset{a}{\swarrow}$ & & $ \overset { b}{\searrow}$ & & $ \overset{a}{\swarrow}$ & \\
& & $ v_{3}$ & & & & $ v_{1}$ & & \end{tabular } \ \ ] ] or even more briefly @xmath252 just using subscripts and omitting the arrows that are recoverable with the obvious convention . here
the sink vertices correspond to the socle @xmath253 and the source vertices correspond to the top @xmath254 .
the @xmath255-dimensional string module @xmath225 decomposes as @xmath256 as a k - vector space , where @xmath228 is spanned by @xmath257 and @xmath254 is spanned by @xmath258 with the indicated action of the arrows . here
the sink vertices correspond to the socle @xmath253 and the source vertices correspond to the top @xmath259 the basic result of gelfand - ponomarev [ gp ] as reiterated by ringel [ ri ] is let @xmath260@xmath86 be a string algebra
. then every finite dimensional indecomposable module is isomorphic to a string module or a band module .
the following proposition shows that string ( special biserial ) algebras and string ( resp .
special biserial ) coalgebras are dual notions and that being a string ( resp .
special biserial ) coalgebra is local property .
this will be used in the specific example of quantum @xmath1 in the next section .
\(a ) if @xmath31 is a finite - dimensional pointed coalgebra , then @xmath31 is a special biserial coalgebra if and only the dual algebra @xmath261 is a special biserial algebra.(b ) if @xmath31 is a finite - dimensional pointed coalgebra , then @xmath31 is a string coalgebra if and only if the dual algebra @xmath261 is a string algebra.(c ) every finite - dimensional subcoalgebra of a pointed special biserial coalgebra is a special biserial coalgebra.(d ) every finite - dimensional subcoalgebra of a pointed string coalgebra is contained in a finite - dimensional subcoalgebra that is a string coalgebra .
consider the finite - dimensional pointed coalgebra @xmath31 and view it as is an admissible subcoalgebra of the path algebra of its quiver k@xmath34 .
set @xmath262 .
then @xmath263 is an admissible ideal of the path algebra k@xmath264 we have @xmath265 by e.g. [ abe ] , lemma 2.2.1 , and @xmath266 .
parts ( a ) and ( b ) follow from the observation that for a path @xmath267 , @xmath268 if and only if @xmath269 now let @xmath31 be a finite - dimensional subcoalgebra of a finite - dimensional special biserial coalgebra . then @xmath31 is an admissible subcoalgebra of k@xmath270 where @xmath270 is the finite full subquiver @xmath34 with vertices corresponding to the group - likes of @xmath31 . the fact that @xmath31 is special biserial follows directly from the definition ( applied to the subquiver @xmath270 ) . part ( d ) is similar , using the observation that one can enlarge a special biserial subcoalgebra @xmath181 of the string coalgebra @xmath31 by adjoining all paths @xmath37 in @xmath31 such that @xmath271 this yields a finite dimensional string algebra containing @xmath181 .
every finite dimensional indecomposable comodule over a string coalgebra is either a string comodule or a band comodule .
the statement follows using duality from the theorem and proposition above .
assume the base field is of characteristic zero .
henceforth , let @xmath272 $ ] be the @xmath101-analog of the coordinate hopf algebra of @xmath1 , where @xmath101 is specialized to a root of unity @xmath273 of odd order @xmath274 .
we study the category _ _ m__@xmath275 of finite - dimensional comodules of @xmath3 , which is equivalent to the category of finite - dimensional modules over the lusztig hyperalgebra @xmath2 modulo the relation @xmath276 ( so only type 1 modules are considered ) , cf .
[ lu ] , [ apw ] .
a fundamental fact is the existence of a nondegenerate hopf pairing @xmath277 which yields a hopf algebra isomorphism @xmath278 ( see [ ck2 , section 5 ] for further references ) where @xmath279 denotes the finitary dual [ mo ] .
quantized coordinate algebras can be defined by taking appropriate duals of quantized enveloping algebras , and as in [ apw , appendix ] , it can be shown in type @xmath280 that the resulting hopf algebra agrees with other definitions in the literature .
for each nonnegative integer @xmath121 , there is a unique simple module @xmath284 of highest weight @xmath121 .
these comodules exhaust the simple comodules ( [ apw ] , [ lu ] , also [ cp , 11.2 ] ) .
we shall write @xmath285 where @xmath286 , the short @xmath37-adic decomposition of @xmath121 .
define an @xmath287-reflection @xmath288by @xmath289 if @xmath290 , and @xmath291 if @xmath292 .
put @xmath293 so that @xmath294 for @xmath121 with @xmath290 modular reduction from the generic lusztig form yields the highest weight module @xmath295 with @xmath296 . when @xmath297 with @xmath290 and @xmath298 this known as a _ weyl module _ having socle series @xmath299 the _ dual weyl module _ has socle series @xmath300for all such @xmath301 on the other hand we have @xmath302 if @xmath303 or @xmath304 ( cf .
[ cp , 11.2.7 ] ) \(a ) ( [ apw ] ) if @xmath312then @xmath313 .
( b ) ( [ ck ] ) if @xmath314 , then @xmath315 has socle series with factors @xmath316 ( c)([ck ] ) if @xmath317 ( and @xmath290 ) , then @xmath315 has socle series with factors @xmath318 as computed in [ ch ] and [ ck ] the cartan matrix for each nontrivial block is the symmetric matrix @xmath319 indexed by @xmath320 with nonzero entries @xmath321 and @xmath322 @xmath323 for all @xmath324 fixing @xmath325 this is interpreted as saying that the multiplicity of @xmath326 in @xmath327 equals @xmath116 , the multiplicity of @xmath328 in @xmath329 equals 1 , and the multiplicity of @xmath326 in @xmath329 equals @xmath330 . taking the socle series of weyl modules and dual weyl modules into account , we see that the second socles of the injective hulls have simple factors as in the statements
. the desired conclusions will follow immediately once we show that there are no nonsplit self - extensions of simples .
this fact follows from the following argument ( cf .
e.g. [ ja , p. 205 ] ) .
let @xmath331 suppose @xmath332 is a short exact sequence , and let @xmath333 be a weight vector of weight @xmath121 generating @xmath284
. then there exists @xmath334 of weight @xmath121 with @xmath335 now let @xmath336 be the subcomodule of @xmath27 generated by @xmath337 . since @xmath338 the simplicity of @xmath339 forces the sequence to split .
this result determines the quiver as having vertices labelled by nonnegative integers and with arrows @xmath340 in case @xmath341 if and only if @xmath342 or @xmath343 . in case @xmath344 ,
the simple block of @xmath121 is a singleton ( equivalently @xmath284 is injective ) and we call this block `` trivial '' .
thus the nontrivial block containing @xmath345 @xmath346 has quiver @xmath347 to simplify notation , we shall drop all but the exponents on @xmath348 and write @xmath349 the injectives indecomposable comodules are all finite - dimensional , in contrast to the injectives for the ordinary ( nonquantum ) modular coordinate coalgebra .
the result also shows that the coradical filtration is of length 2@xmath47 it follows that @xmath3 is a _ semiperfect _
coalgebra ( [ lin ] ) in the sense that every finite dimensional comodule has a finite - dimensional injective envelope ( or , equivalently , projective cover ) on both the left and the right .
thus by [ ckq , corollary 5.4 ] , we know that the category @xmath350 of finite - dimensional right comodules has almost split sequences .
since @xmath3 a hopf algebra , we also have that @xmath3 is quasi - cofrobenius ( or `` self - projective '' ) , i.e. @xmath3 is projective as both a right and left @xmath3-comodule .
see [ dnr ] for discussion of these coalgebraic properties . we can write @xmath352 with @xmath353 and @xmath354 the left hand side @xmath315 has a weyl module filtration @xmath355 computing the dimension of yields @xmath315 @xmath356 on the other hand , @xmath357 has filtration @xmath358and @xmath359 .
thus the two sides have the same dimension .
the lusztig tensor product theorem implies that @xmath360 has @xmath361 as a direct summand .
the desired conclusion follows .
let @xmath362 denote a nontrivial block of @xmath3 as specified by a simple module with highest weight @xmath363 , i.e. @xmath362=@xmath364 as above ( exponent is the multiplicity ) . by inspecting maps between indecomposable injectives
we obtain the basic coalgebra @xmath31 of @xmath362 is the subcoalgebra of path coalgebra of the quiver @xmath365 spanned by the by group - likes @xmath366 corresponding to vertices and arrows @xmath367 together with coradical degree two elements @xmath368 let @xmath27 denote the basic injective @xmath369 for @xmath362 .
then the basic coalgebra @xmath370is defined to be the coendomorphism coalgebra @xmath371 and we see easily that @xmath372 set @xmath373 for all @xmath374 with the convention that @xmath375 if @xmath376 . from the description of the indecomposable injectives above
, we see that @xmath377 by inspecting socle series we observe that @xmath378 and @xmath379 are both one - dimensional . also @xmath380 is one - dimensional being the composition of @xmath379 and @xmath381 thus there is a basis of @xmath382 consisting of maps @xmath383 such that @xmath384 for all @xmath385 also an easy induction shows that we may choose @xmath386 so that @xmath387 for all @xmath388 we choose dual basis elements in @xmath389 by setting @xmath390 and we obtain the skew primitive elements corresponding to the arrows as in the statement .
it is not hard to check that the elements satisfy the claimed comultiplications in the statement .
we have @xmath391 and @xmath392 . now , for example , evaluating @xmath393 on the products @xmath394 we get @xmath395 and @xmath396 the standard embedding of @xmath31 into the path coalgebra is given as follows ( cf .
write @xmath397 for a coideal @xmath263 .
define the embedding @xmath398 by @xmath399 for all @xmath400 here @xmath401 , @xmath402 is the cotensor coalgebra , @xmath403 is taken to be the identity , and @xmath404 is the projection onto @xmath134 .
the conclusions concerning the form of the ( images of ) elements @xmath393 follow immediately .
the remaining cases are simpler and are left to the reader . by a result in [ rad ]
, there is a distinguished group - like element in @xmath3 ( see also [ dnr , p. 197 ] ) , which is trivial if and only if @xmath3 is unimodular . there is a unique one - dimensional comodule , namely the trivial comodule @xmath405 , so @xmath3 is unimodular .
this implies that the distinguished group - like element in @xmath3 is just the identity element .
the square of the antipode of @xmath2 is inner by the element @xmath406 and it follows by duality that the square of the antipode of @xmath3 is given by @xmath407 ( a fact that can be checked directly ) . thus @xmath3 is unimodular and is inner by an element of @xmath408 and the conclusion now follows from [ din ] .
the conclusion can be similarly shown to hold for quantized coordinate algebras generally , using the fact that @xmath409 is inner in the generic quantized enveloping algebra @xmath410 where @xmath411 is complex semisimple lie algebra , see e.g. , [ ja2 ] 4.4 . by definition [ din ] , there is a @xmath135-bicomodule embedding @xmath412 by [ cg ] the basic coalgebra @xmath31 of @xmath3 is of the form @xmath413 where @xmath29 .
one can easily check that @xmath414 so that the coalgebra @xmath413 embeds in @xmath415 as a @xmath261-bicomodule by restriction . from the description of the nontrivial basic block @xmath420 above it
is immediate that @xmath31 is a special biserial coalgebra .
the indecomposable injective right @xmath31-comodules are @xmath421 @xmath422 and they are also projective . clearly rad@xmath423 and rad@xmath424 .
so by deleting all the basis elements @xmath425 @xmath212 we reduce to an associated string coalgebra @xmath181 spanned by the arrows and group - likes @xmath426 and arrows @xmath427 , as in 1.4 . by 1.4 , the ar - quiver of the string coalgebra @xmath181 is the same as for @xmath31 ( and for @xmath428 except for the injective - projectives .
the noninjective indecomposable comodules for @xmath181 are determined by strings .
the simples correspond to the group - likes and the length two comodules ( weyl and dual weyl modules ) are determined by arrows .
more generally , indecomposables are given by concatenations of arrows and their formal inverses .
there are strings of the form ( up to equivalence ) @xmath429 where the subscripts form an interval @xmath430 $ ] of nonnegative integers strictly increasing from right to left in the string .
we adopt the following notation for string modules @xmath431 specifying each type of module by its starting and ending vertices @xmath124 and @xmath432 respectively .
we declare that @xmath433 is the simple module @xmath434 for all @xmath68 .
the comodules @xmath435 , @xmath436with @xmath437 @xmath438even , @xmath438together with the @xmath439 with @xmath437 odd exhaust the indecomposable @xmath181-comodules .
the remaining @xmath31-comodules are the projective - injectives @xmath440 @xmath385 the string comodule represented by @xmath441 ( or @xmath442 ) has composition length two and has structure described by the diagram @xmath251{lll}$t$ & & \\ & $ \searrow$ & \\ & & $ t-1$\end{tabular } \ \ \ \ \ \ \ ] ] where the weyl module with highest weight @xmath443 corresponds ( via morita - takeuchi equivalence ) to @xmath444 .
dually , the string comodule represented by @xmath445 has structure described by the diagram @xmath251{lll } & & $ t$\\ & $ \swarrow$ & \\
$ t+1 $ & & \end{tabular } \ \ \ \ \ \ \ ] ] where the dual weyl comodule corresponds to @xmath446 .
the string comodule represented by @xmath447 is constructed by the pullback @xmath41{ccc}\mathrm{st}(a_{t}b_{t-1}^{-} ... a_{i-1}b_{s}^{- } ) & \longrightarrow & \mathrm{st}(a_{t})\\ \downarrow & & \downarrow\\ \mathrm{st}(b_{t-1}^{-} ... a_{i-1}b_{s}^{- } ) & \longrightarrow & s(t ) \end{array}\ ] ] where we recall that @xmath448 denotes the simple comodule at the vertex @xmath449 .
the other string comodules can be constructed inductively similarly by pushouts and pullbacks .
therefore , by category equivalence , the indecomposable @xmath3 -comodules can be thus constructed .
\(a ) we must show that for each dimension vector @xmath450 , there exist only finitely many indecomposable right @xmath3-comodules @xmath126 with dimension vector @xmath451 if @xmath452 is an indecomposable with @xmath453 then @xmath452 corresponds to one of the two string comodules @xmath435 , @xmath436 ( if @xmath437 @xmath197is even)@xmath438and to one of @xmath439 ( if @xmath437 is odd ) , or @xmath452 is injective .
this proves ( a ) .
\(b ) fix @xmath388
let @xmath454 denote the subcoalgebra of @xmath31 spanned by @xmath455 @xmath454 is a subcoalgebra of path coalgebra of the finite quiver @xmath456 clearly @xmath454 is a special biserial coalgebra , with dual special biserial algebra @xmath457 the associated string coalgebra @xmath458 is the span of @xmath459 one sees easily that the dual algebra @xmath460 is a string algebra isomorphic to the path algebra modulo the ideal of paths of length greater than one .
now the strings are ones already determined for all of @xmath461 and , again , there are no bands .
so by the theory of special biserial algebras , see [ erd ] , the string algebra @xmath460 and the special biserial algebra @xmath462 are both of finite representation type . by the standard category equivalence between modules for a finite - dimensional algebra and comodules over its dual
, the same is true for the dual coalgebras @xmath458 and @xmath454 .
obviously @xmath463 , so the proof of ( b ) is finished .
there is a duality on @xmath464 which is obtained on strings by interchanging the letters @xmath465 and @xmath466 for all i. the duality interchanges @xmath467 and @xmath468 and interchanges @xmath469 and @xmath470 .
this is just the usual duality on @xmath350 ( using the antipode on @xmath3 ) , transported via category equivalence to @xmath471 and restricted to @xmath472 alternatively one can observe that interchanging the @xmath55 s and @xmath69 s yields a coalgebra anti - isomorphism of @xmath31 .
the statements concerning the @xmath126 s and @xmath336 s are obvious from the definitions .
this is verified by directly computing @xmath474 : @xmath475 this can be done as follows .
the diagram @xmath251{lllll } & & $ \genfrac{}{}{0pt}{}{\genfrac{}{}{0pt}{1}{t}{t+1\text { } t-1}}{t}\oplus\genfrac{}{}{0pt}{}{\genfrac{}{}{0pt}{1}{t-2}{t-1\text { } t-3}}{t-2}\oplus ... \oplus\genfrac{}{}{0pt}{}{\genfrac{}{}{0pt}{1}{s}{s+1\text { } s-1}}{s}$ & & \\ & $ \overset{}{\nearrow}$ & & $ \overset{}{\searrow}$ & \\
$ \genfrac{}{}{0pt}{}{t-1\quad ... \quad s-2\quad s+1}{t\quad\quad t-2\quad ... \quad s+2\quad s}$ & & \multicolumn{1}{c}{$\overset{\omega^{-1}}{\rightsquigarrow}$ } & & $ \genfrac{}{}{0pt}{}{t\quad t-1\quad ...
\quad\quad\quad s}{t+1\quad t-2\quad\quad ... \quad s+1\quad s-1}$\end{tabular } \ \ \ \ \ \ ] ] with the injective hull on top and cokernel on the right@xmath197demonstates @xmath476 and similarly @xmath251{lllll } & & $ \genfrac{}{}{0pt}{}{\genfrac{}{}{0pt}{1}{t}{t+1\text { } t-1}}{t}\oplus\genfrac{}{}{0pt}{}{\genfrac{}{}{0pt}{1}{t-2}{t-1\text { } t-3}}{t-2}\oplus ... \oplus\genfrac{}{}{0pt}{}{\genfrac{}{}{0pt}{1}{2}{3\text { } 1}}{2}\oplus\genfrac{}{}{0pt}{}{\genfrac{}{}{0pt}{1}{0}{1}}{0}$ & & \\ & $ \overset{}{\nearrow}$ & & $ \overset{}{\searrow}$ & \\
$ \genfrac{}{}{0pt}{}{t-1\quad ... \quad\quad1}{t\quad t-2\quad ... \quad2\quad0}$ & & \multicolumn{1}{c}{$\overset{\omega^{-1}}{\rightsquigarrow}$ } & & $ \genfrac{}{}{0pt}{}{\quad t\quad\quad ... \quad\quad2\quad0}{t+1\quad
t-1\quad ... \quad\quad1\quad}$\end{tabular } \ \ \ \ \ \ ] ] yields @xmath477
lastly , we obtain [ 3 ] : @xmath251{lllll } & & $ \genfrac{}{}{0pt}{}{\genfrac{}{}{0pt}{1}{t}{t+1\text { } t-1}}{t}\oplus\genfrac{}{}{0pt}{}{\genfrac{}{}{0pt}{1}{t-2}{t-1\text { } t-3}}{t-2}\oplus ... \oplus\genfrac{}{}{0pt}{}{\genfrac{}{}{0pt}{1}{s+1}{s+2\text { \ } s}}{s+1}$ & & \\ & $ \overset{}{\nearrow}$ & & $ \overset{}{\searrow}$ & \\ $ \genfrac{}{}{0pt}{}{\quad t-1\quad
... \quad s+2\quad s}{t\quad t-2\quad ... \quad s+1}$ & & \multicolumn{1}{c}{$\overset{\omega^{-1}}{\rightsquigarrow}$ } & & $ \genfrac{}{}{0pt}{}{\quad t\quad ... \quad\quad s+3\quad s+1}{t+1\quad t-1\quad ... \quad s+2\quad}$\end{tabular } \ \ \ \ \ \ ] ] from the formulae above for @xmath474 we get the following expressions for string comodules as elements in the @xmath473-orbits of simples : @xmath478 the first equation follows directly from [ 1 ] .
the second comes from computing @xmath479(m(\frac{t - s-1}{2},\frac{t - s-1}{2}))\\ & = [ \omega^{-s}\circ\omega^{-1}](m(t - s-1,0))\\ & = \omega^{-s}(n(t - s,0))\\ & = n(t , s),\end{aligned}\ ] ] using [ 1],[2 ] and [ 3 ] in succession .
the second pair of equations are obtained dually .
the @xmath473-orbits of indecomposable comodules for a nontrivial block can be graphed by identifying the string comodules @xmath480 and @xmath481 with the cartesian points @xmath482 and identifying the dual comodules @xmath483 and @xmath484 with the points @xmath485 , reversing the coordinates
. the figure below shows the orbits containing the simples @xmath486 and @xmath487 .
the almost split sequences and ar quiver for the category of finite - dimensional @xmath31-comodules are described starting with the sequences having an injective - projective comodule @xmath488 in the middle term .
these are precisely the sequences @xmath489 with @xmath374 .
these sequences can be rewritten as @xmath490 where @xmath491 is declared to be @xmath492 . since @xmath31 is self - projective ( i.e. quasi - cofrobenius ) , the usual arguments for modules ( e.g. [ ars , iv.3 ] ) apply to show that the functors @xmath495 provide autoequivalences of the stable category @xmath496 ( modulo injective - projectives ) . applying @xmath497 to the ends of an almost split sequence @xmath498 yields the exact commutative diagram @xmath251{ccccccccc } & & $ 0 $ & & $ 0 $ & & $ 0 $ & & \\ & & $ \downarrow$ & & $ \downarrow$ & & $ \downarrow$ & & \\ $ 0 $ & $ \longrightarrow$ & $ \omega m$ & $ \longrightarrow$ & $ \omega
e\oplus j$ & $ \longrightarrow$ & $ \omega n$ & $ \longrightarrow$ & $ 0$\\ & & $ \downarrow$ & & $ \downarrow$ & & $ \downarrow$ & & \\ $ 0 $ & $ \longrightarrow$ & $ i_{m}$ & $ \longrightarrow$ & $ i_{m}\oplus i_{n}$ & $ \longrightarrow$ & $ i_{n}$ & $ \longrightarrow$ & $ 0$\\ & & $ \downarrow$ & & $ \downarrow$ & & $ \downarrow$ & & \\ $ 0 $ & $ \longrightarrow$ & $ m$ & $ \longrightarrow$ & $ e$ & $ \longrightarrow$ & $ n$ & $ \longrightarrow$ & $ 0$\\ & & $ \downarrow$ & & $ \downarrow$ & & $ \downarrow$ & & \\ & & $ 0 $ & & $ 0 $ & & $ 0 $ & & \end{tabular } \ \ \ ] ] where @xmath499 and @xmath500 are projective covers and @xmath134 is a projective - injective comodule .
the arguments of [ ars , x.1 , propositions 1.3 , 1.4 , p. 340
] show that the top row @xmath501 is an almost split sequence .
similarly , there is an almost split sequence @xmath502 for some projective - injective @xmath263 . by the proposition above
, we obtain all almost split sequences in this manner . since the sequences in [ 5 ] are precisely the almost split sequences with an injective - projective in the middle term , the other sequences @xmath503 we obtain when we apply @xmath504 for a nonzero integer @xmath505 do not have an injective - projective summand in the middle term .
thus the sequences as in the statement are all almost split and exhaust all almost split sequences .
we define the auslander - reiten ( ar ) quiver of a coalgebra to be the quiver whose vertices are isomorphism classes of indecomposable comodules and whose ( here multiplicity - free ) arrows are defined by the existence of an irreducible map between indecomposables . when the coalgebra @xmath3 is right semiperfect , then the results of [ ckq ] guarantee that the category of finite - dimensional right @xmath3-comodules has almost split sequences , and thus that the ar quiver is determined by them .
we now describe the ar quiver for @xmath471.@xmath197the stable ar quiver ( with all @xmath488 s deleted ) consists of two components of type @xmath506 which are transposed by @xmath497 . to get the ar quiver we use proposition 1.2.4 ( b ) .
we adjoin the injective - projectives @xmath488 with @xmath507 odd to the component containing @xmath508 with @xmath507 odd , and similarly we adjoin the @xmath488 with @xmath507 even to the other component containing the @xmath508 with @xmath507 even .
the two components for a nontrivial block are shown below . @xmath509{ccccccccccccccc } & \omega^{3}s(0 ) & & & & \omega s(0 ) & \rightarrow & i_{0 } & \rightarrow & \omega^{-1}s(0 ) & & & & \omega^{-3}s(0 ) & \\ \nearrow & & \searrow & & \nearrow & & \searrow & & \nearrow & & \searrow & & \nearrow & & \searrow\\ & & & \omega^{2}s(1 ) & & & & s(1 ) & & & & \omega^{-2}s(1 ) & & & \\ \searrow & & \nearrow & & \searrow & & \nearrow & & \searrow & & \nearrow & & \searrow & & \nearrow\\ & \omega^{3}s(2 ) & & & & \omega s(2 ) & \rightarrow & i_{2 } & \rightarrow & \omega^{-1}s(2 ) & & & & \omega^{-3}s(2 ) & \\ \nearrow & & \searrow & & \nearrow & & \searrow & & \nearrow & & \searrow & & \nearrow & & \searrow\\ & & & \omega^{2}s(3 ) & & & & s(3 ) & & & & \omega^{-2}s(3 ) & & & \\ \searrow & & \nearrow & & \searrow & & \nearrow & & \searrow & & \nearrow & & \searrow & & \nearrow\\ & \omega^{3}s(4 ) & & & & \omega s(4 ) & \rightarrow & i_{4 } & \rightarrow & \omega^{-1}s(4 ) & & & & \omega^{-3}s(4 ) & \end{array } \cdot\cdot\cdot\\ \cdot\\ \cdot\\ \cdot\end{gathered}\ ] ] @xmath509{ccccccccccccccc } & \omega^{2}s(0 ) & & & & s(0 ) & & & & \omega^{-2}s(0 ) & & & & \omega^{-4}s(0 ) & \\ \nearrow & & \searrow & & \nearrow & & \searrow & & \nearrow & & \searrow & & \nearrow & & \searrow\\ & & & \omega s(1 ) & \rightarrow & i_{1 } & \rightarrow & \omega^{-1}s(1 ) & & & & \omega^{-3}s(1 ) & & & \\ \searrow & & \nearrow & & \searrow & & \nearrow & & \searrow & & \nearrow & & \searrow & & \nearrow\\ & \omega^{2}s(2 ) & & & & s(2 ) & & & & \omega^{-2}s(2 ) & & & & \omega^{-4}s(2 ) & \\ \nearrow & & \searrow & & \nearrow & & \searrow & & \nearrow & & \searrow & & \nearrow & & \searrow\\ & & & \omega s(3 ) & \rightarrow & i_{3 } & \rightarrow & \omega^{-1}s(3 ) & & & & \omega^{-3}s(3 ) & & & \\ \searrow & & \nearrow & & \searrow & & \nearrow & & \searrow & & \nearrow & & \searrow & & \nearrow\\ & \omega^{2}s(4 ) & & & & s(4 ) & & & & \omega^{-2}s(4 ) & & & & \omega^{-4}s(4 ) & \end{array } \cdot\cdot\cdot\\ \cdot\\ \cdot\\ \cdot\end{gathered}\ ] ] let @xmath514 and let @xmath515 ( resp .
@xmath516 ) denote the @xmath3-comodule in the block of @xmath517 corresponding to the string comodule @xmath467 ( resp .
@xmath518 ) the quantum dimensions are @xmath519_{\zeta}\sum_{i = s}^{t}(-1)^{i}(i+1)\\ \operatorname*{qdim}n_{r_{0}}(s , t ) & = [ r_{0}+1]_{\zeta}\sum_{i = s}^{t}(-1)^{i}(i+1).\end{aligned}\ ] ] the composition series of @xmath515 is @xmath520 and is of odd length , the comodule @xmath516 is of even length has composition series given by the same expression .
we compute @xmath521 by first observing that @xmath522 @xmath523 and more generally @xmath524 note that @xmath525_{\zeta}=-[r_{0}+1]_{\zeta}$ ] . by the lusztig tensor product theorem ( see [ lu ] or [ cp ] ) we have @xmath526 where @xmath527 is the classical nonquantum simple module of highest weight @xmath169 and the superscript @xmath528 is the quantum frobenius twist .
the grouplike @xmath512 acts diagonally on the tensor product and acts trivially on the second factor .
therefore @xmath529_{\zeta}.$ ] thus @xmath530_{\zeta}\text { if } j\text { odd}}{(j+1)[r_{0}+1]_{\zeta}\text { if } j\text { even}}\text{.}\ ] ] the result follows .
the green ring for @xmath350 has as a @xmath531-basis of isomorphism classes @xmath532 $ ] of finite dimensional indecomposable comodules @xmath533 , with addition given by direct sum and multiplication given by the tensor product .
the _ stable green ring _ is the green ring modulo the ideal of projective - injectives .
this ring was studied for certain finite - dimensional hopf algebras ( generalizing the frobenius - lusztig kernel for quantum @xmath1 ) in [ egst ] . while the basic coalgebra @xmath31 is not a hopf algebra
, @xmath471 inherits a quasi - tensor structure from @xmath3 ( and from the quantized hyperalgebra @xmath534 ) . by a well - known result [ cp , 11.3 ] ( cf . [ and ] , see also [ ck ] ) we have a formula for the tensor product of two simples as a direct sum of simples and injectives .
it is used in the proof of the next result .
we shall also use _ chebyshev polynomials of the second kind_. they are defined to be polynomials @xmath535 given recursively by @xmath536 with @xmath537 , @xmath538 note that @xmath535 is of degree @xmath507 and is an element of @xmath539 $ ] . the clebsch - gordon decomposition for the tensor product of simple modules @xmath546 and @xmath547 with for positive integers @xmath246 with @xmath548 is @xmath549 where @xmath449 runs over integers such that @xmath550 @xmath551 @xmath552 and @xmath553 let @xmath554 $ ] and @xmath555 .
this first case of the decomposition yields the recursive formula @xmath556 for @xmath507 with @xmath557 .
since @xmath558 is projective , it is obvious that @xmath559 it follows that @xmath5 generates the subalgebra spanned by @xmath560 and that @xmath5 satisfies the rescaled chebyshev polynomial of the second kind @xmath561 of degree @xmath562 the simple modules @xmath284 are of the form @xmath563 by the lusztig tensor product theorem [ cp , 11.2 ] , cf .
thus the remaining ( nonrestricted ) simples are generated by @xmath5 along with the frobenius twists of simple @xmath564-modules @xmath306 .
these modules are generated by @xmath565 , in view of the classical clebsch - gordon formula .
thus the class of every simple module is uniquely of the form @xmath566 where @xmath567 and @xmath568 note that @xmath569 for all @xmath570 .
this says that in the stable green ring @xmath571 the operator induced by @xmath497 , which we denote by @xmath572 , we have @xmath573 for all @xmath574 .
thus by elementary ring theory @xmath575 is an @xmath571-module map , which equals multiplication by @xmath576=[v(1)].\ ] ] also @xmath218 is invertible with inverse @xmath577 $ ] . for all @xmath578 and integers @xmath579 we
now have @xmath580 thus by proposition 2.4.1 every indecomposable module has image in @xmath571 of the form @xmath581 @xmath582 where @xmath583 is the class of a simple module . by the proof of proposition 2.4.1 , all @xmath584-orbits are infinite ; thus we see that the class of every finite - dimensional indecomposable comodule is uniquely represented by a monomial of the form @xmath585 where @xmath586 and @xmath587 . it remains to see that the ideal @xmath588 of injective - projectives of the green ring is generated by @xmath589=u_{\ell-1}(\frac{1}{2}x).$ ] for the non - obvious inclusion ,
let @xmath361 be an indecomposable injective comodule with socle @xmath284 .
this injective can be written as @xmath590 ( lemma 2.0.3 ) .
thus it suffices to show that @xmath591 .
this follows from the decomposition @xmath592 for @xmath593 , which is special case of the formula at the beginning of the proof .
one observes iteratively that the classes of all summands which occur on the right hand side ( i.e. @xmath594 @xmath595 @xmath596 are in @xmath588 .
w. chin , a brief introduction to coalgebra representation theory , _ in _ : proceedings from an international conference held at depaul university , j. bergen , s. catoiu , w. chin , eds .
marcel dekker ( 2003 ) . | we develop the theory of special biserial and string coalgebras and other concepts from the representation theory of quivers .
these theoretical tools are then used to describe the finite dimensional comodules and auslander - reiten quiver for the coordinate hopf algebra of quantum sl(2 ) at a root of unity .
we also compute quantum dimensions and describe the stable green ring .
let @xmath0 $ ] denote the quantized coordinate hopf algebra of @xmath1 as defined in [ apw ] and [ lu2 ] at a root of unity of odd order over a field of characteristic zero .
in this article we study the category of finite - dimensional comodules , a category that is equivalent to the category of finite - dimensional modules over ( a suitable quotient of ) the quantized hyperalgebra @xmath2 .
our approach uses representations of the gabriel quiver associated to @xmath3 , methods from the representation theory of quivers , and most notably , the theory string algebras and special biserial algebras .
the methods of this paper demonstrate the utility of coalgebraic methods , which will hopefully be applicable to other categories of interest .
we discuss some required coalgebra representation theory in section 1 and then use the more general results to concentrate on the case of the quantized coordinate hopf algebra @xmath3 at a root of unity over a field of characteristic zero in section 2 . in [ ck ]
we determined the structure of the injective indecomposable comodules along with the block decomposition of @xmath3 . here
we completely determine the finite - dimensional indecomposable comodules , the auslander - reiten quiver and almost split sequences .
we introduce the notions of a special biserial coalgebra and string coalgebra , the latter modifying the definition in [ sim2 ] , [ sim3 ] .
these notions are coalgebraic versions of machinery known for algebras [ ri ] , [ br ] , [ erd ] .
it turns out that @xmath3 is special biserial and that its representations are closely related to an associated string subcoalgebra , and this allows a listing of indecomposables and a description of the auslander - reiten quiver .
one consequence is that the coalgebra @xmath3 is of tame of discrete comodule type [ sim2 ] and is the directed union of finite - dimensional coalgebras of finite type .
this is in contrast with the situation for restricted representations , see [ xia ] , [ sut ] , [ rt ] .
we compute the green ring modulo projective - injectives , or _ stable green ring _ , in which the syzygy operation is given by multiplication by the isomorphism class of a certain weyl module . here again the structure is simpler than for restricted representations ( cf .
[ egst ] ) ; the stable green ring is a polynomial ring @xmath4 $ ] three indeterminates modulo a rescaled chebyshev polynomial of the second kind in @xmath5 .
the notion of quantum dimension for comodules over @xmath6was studied in [ and ] . quantized tilting modules were of use in [ os ] , where module categories over the tensor category of finite - dimensional comodules @xmath7 of @xmath3 were characterized . here
we compute the quantum dimensions for finite - dimensional @xmath3-comodules using string modules and the lusztig tensor product theorem . as a result
we find that the only comodules of quantum dimension zero are the nonsimple injective comodules , a fact observed for quantized tilting modules in [ and ] .
let us provide an outline of the contents of this article .
the first section contains relevant coalgebra theory .
we begin in 1.1 by reviewing the notions of path coalgebras , basic coalgebras and the dual gabriel theorem .
we introduce special biserial and string coalgebras in 1.3 and their representation theory in 1.3 .
section 2 begins with a summary of facts concerning the coordinate hopf algebra @xmath8@xmath9 $ ] at a root of unity where is of characteristic zero .
here it is shown that @xmath3 is a symmetric coalgebra . in 2.1
we compute the basic coalgebra of @xmath3 as a subcoalgebra of the path coalgebra of its gabriel quiver .
we show in passing that @xmath3 and its basic coalgebra are symmetric as coalgebras in the sense of [ din ] in 2.2 .
using string coalgebra theory we list the finite - dimensional indecomposable @xmath3-comodules in 2.3 . in 2.4
we show that every simple comodule is in the syzygy - orbit of a simple comodule by direct computation .
this yields the description of almost split sequences , and the ar quiver in 2.5 .
the stable green ring and quantum dimensions are computed in 2.6 and 2.7 .
* notation * let @xmath3 denote a coalgebra over the fixed base field k. set the following @xmath10 the category of right @xmath3-comodules .
@xmath7 the category of finite - dimensional right @xmath3 -comodules @xmath11 the category of quasifinite right @xmath3 -comodules ind@xmath12 the category finite - dimensional indecomposable left @xmath3-comodules @xmath13 the cohom functor @xmath14 the cotensor ( over @xmath3 ) the linear dual @xmath15 @xmath16 the functor @xmath17 @xmath18 the finitary dual coalgebra of the algebra @xmath19 the author wishes to acknowledge the advice of k. erdmann concerning special biserial algebras and d. simson for his comments . |
the calice collaboration @xcite is a group of 140 people from 24 institutes in europe , us and asia .
the collaboration is planning a beam test program starting in 2004 to study both electromagnetic ( ecal ) and hadronic ( hcal ) calorimeters for a future linear collider ( lc ) .
the ecal will be a silicon - tungsten sandwich sampling calorimeter @xcite and there are various options under investigation for the hcal , namely using tile scintillators with analogue readout @xcite or rpcs or gems with digital readout @xcite as the sensors .
the main driver for these choices is the requirement of fine granularity for the `` energy flow '' algortihms used to obtain the jet resolutions necessary for lc physics @xcite .
the beam test program will test the ecal and all versions of the hcal in an integrated data acquisition system so the data from both calorimeters can be analysed together .
it is planned to do a systematic study of the dependence of the shower structure on particle energy , type and incident angle , as well as adding a `` preshower '' in front of the calorimeter to generate narrow cones of particles to simulate jets .
the total number of different configurations which will be tested is of order @xmath0 , and to see the fine detail required , of order @xmath1 events is estimated to be needed for each configuration .
the calice beam test ecal will consist of 30 layers of silicon and tungsten .
each silicon layer comprises nine wafers , each being a @xmath2 array of diode pads and each of these is a readout channel .
this gives 324 channels per layer , or 9720 channels in total .
these need to be read for every event , preferably without any threshold or other data suppression so as to allow pedestal and noise studies offline .
the signals expected in the diodes range from the energy deposited by one minimum ionising particle ( mip ) to around 1000 mips in the core of high energy electron showers , a range of 10 bits . the intrinsic noise from the on - detector preamplifier @xcite is expected to be equivalent to around 0.1 mip or less .
it is of interest to measure this noise level , so the dynamic range required is 14 bits in total . at the highest energies of around 1000 mips ,
the resolution is degraded if the precision is below a mip , so 10 bits of precision are required
. the on - detector preamplifier has a cr - rc shaper with a peaking time of 180ns .
this defines the trigger latency needed to ensure the waveform is sampled at this peak .
a trigger jitter of 10ns is acceptable ; this leads to a maximum shift below the peak of 0.1% , which is within the 10 bits precision requirement .
it consists of 15 identical readout boards and a single trigger board .
the readout boards handle the on - detector preamplifier signals and all digitisation for two layers of the ecal , or 648 channels .
these signals are received on six cables per readout board .
each cable connects to an electrically independent on - detector board which holds the preamplifier and which handles three silicon wafers of 108 channels .
the trigger board is a simple board to prevent further triggers until readout is complete and to distribute the trigger and the system clock across the backplane to the readout boards .
it contains very few components and will not be described further below .
an schematic diagram of the readout board is also shown in figure [ fig : overview ] .
the board is controlled by a `` master fpga '' which handles the vme interface and distributes the clock and trigger from the backplane across the board .
it interacts with the `` slave fpgas '' by distributing configuration data ( written to the board via vme ) and collects the event data prior to vme readout .
there is one slave fpga for each cable from the on - detector electronics and so six per readout board .
each operates independently of the others although they contain identical firmware so , in standard operation , they would effectively be synchronised together .
the slave fpgas send the control and timing signals needed by the on - detector electronics and receive six analogue signal lines .
the 108 channels are multiplexed onto these six lines , with 18 channels per line .
the slave fpgas control six 16-bit , 500khz adcs , one per signal line .
the extra bits give robustness against loss due to non - optimal range matching .
the adc speed results in a time to digitise all 18 channels of less than 50@xmath3s , or only 5% of the allowed 1ms event time .
the trigger is also distributed to the on - detector electronics via the slave fpgas , rounded to a 100 mhz clock , giving the required 10ns maximum jitter . to save any parallel development of similar electronics for the tile hcal option , the on - detector electronics for
this system is attempting to become similar enough to the ecal such that the same readout boards could be used for both .
there will be around 1500 channels in the tile scintillator hcal .
the 16-bit adc would be more than ample for the range and precision required and would result in 3kbytes per event .
the number of channels multiplexed per board for the hcal is also not known , so it is not clear how many extra readout boards would need to be fabricated .
however , it is clear that a common solution would save significant effort and is being actively pursued by the groups involved .
the maximum rate of data readout over vme is around 30mbytes / s . for a desired 1khz rate , this implies the event size needs to be below 30kbytes .
there will be no threshold suppression of the data on the readout board , so the ecal data size will be 19kbytes per event .
both options of the hcal will be around 3kbytes and the size of any beam monitoring , particle idenitification and trigger data are uncertain , but likely to be around 1kbyte .
hence , the total event size should be around 25kbytes .
this makes a 1khz peak rate tight but not impossible .
the readout board vme interface will be optimised for speed , using dma transfers and asynchronous vme access .
a fallback solution will be to split the system between two or more crates , with two vme - pci bus convertors . | the aims of the calice test beam program are presented .
the proposed electronics readout system for the calice ecal is described .
it is a purpose - built vme - based , unbuffered system . |
spectropolarimetry offers a direct probe of early - time sn geometry since a hot , young sn atmosphere is dominated by electron scattering , which by its nature is highly polarizing . for an unresolved source that has a spherical distribution of scattering electrons ,
the directional components of the electric vectors of the scattered photons cancel exactly , yielding zero net linear polarization .
any asymmetry in the distribution of the scattering electrons , or of absorbing material overlying the electron - scattering atmosphere , results in incomplete cancellation , and produces a net polarization ( see , e.g. , * ? ? ?
type ii - plateau supernovae ( sne ii - p ) are the canonical core - collapse events that arise from isolated red supergiant stars @xcite .
the precise nature of the mechanism responsible for the stellar explosion , however , remains the subject of considerable debate @xcite .
one key diagnostic is explosion geometry .
a global asphericity of sn ejecta will manifest itself through two main spectropolarimetric signatures .
first , there will be significant polarization in spectral regions not dominated by line opacity .
for sne ii - p , this includes the broad spectral region 68008200 .
strong emission lines should generally be unpolarized , since directional information is lost as photons are absorbed and reemitted in a line @xcite ( see , however , * ? ? ?
finally , polarization increases are anticipated in the troughs of strong p - cygni absorption lines .
this results primarily from selective blocking of more forward - scattered and , hence , less polarized , light in the trough regions .
a critical point to bear in mind is that the longer one waits after the explosion , the deeper into the ejecta one can see . for sne ii - p ,
the story unfolds over an extended time period as the photosphere recedes ( in mass coordinates ) back through the ejecta as the hydrogen recombination front progresses from the outside inwards through the thick hydrogen envelope .
the transition from the opaque , `` photospheric '' phase to the transparent , `` nebular phase '' is signaled by the sudden drop in luminosity that marks the end of the photometric plateau . in initial studies ,
sne ii - p were found to have minimal intrinsic polarizations at the generally early times at which they were observed .
low - contrast line features were sometimes seen in the data , but no obvious strong continuum polarization was found ( e.g. , * ? ? ?
* ; * ? ? ?
* ) ; for a recent review of sn spectropolarimetry of all types , see @xcite .
the situation changed dramatically with observations of sn 2004dj , however , for which a sudden `` spike '' in polarization was seen during the steepest part of the descent off of the plateau @xcite .
using existing models @xcite , it was concluded that the late onset of the polarization suggested a fundamentally non - spherical explosion , with the asphericity cloaked at early times by the massive , opaque , hydrogen envelope .
an obvious observational question presented itself : was sn 2004dj a unique ( or rare ) event ? or , would other similarly well - observed sne ii - p display similar behavior ?
@xcite and @xcite provide partial answers from analysis of four additional sne ii - p : three ( sne 2006my , 2006ov , and 2007aa ) exhibit large , late - time ( end of photometric plateau and beyond ; ) polarization @xcite , while one ( sn 2004et ) exhibits strong and temporally increasing polarization that began at least three months before the end of its plateau phase @xcite .
detailed comparison of these datasets with that of sn 2004dj is hindered , however , by modest temporal sampling or phase coverage .
here we show initial results from our program to obtain critically sampled spectropolarimetry of sne ii - p with the european southern observatory very large telescope .
figure 1 presents nine spectropolarimetric epochs for our most well observed sn ii - p , sn 2008bk , along with the data of sn 2004dj from @xcite . unlike the case for sn 2004dj ,
sn 2008bk was strongly polarized well _ before _ the end of the plateau ( much like sn 2004et ) , with the level rising through the plateau . despite the different temporal behavior ,
it is intriguing that at the point of maximum recorded polarization ( i.e. , day 31 with respect to the end of the plateau for sn 2004dj and day -21 for sn 2008bk ) the spectropolarimetric _ signatures _ of sn 2008bk are virtually identical to those of sn 2004dj .
this similarity may suggest a common cause or , at least , geometry . following the plateau drop - off
, the polarization of sn 2008bk remains remarkably constant .
in fact , although not shown in figure 1 , the spectropolarimetry from days 9 through 120 post - plateau ( epochs 5 through 9 ) are virtually indistinguishable from one another .
this is quite different from the behavior of sn 2004dj .
-band magnitude of sn 2004dj ; the expected polarization derived from the models of @xcite and discussed in the text is also shown ( filled triangles connected by a dashed line ) . *
bottom : * same as the top , but for sn 2008bk .
note that an interstellar polarization ( isp ) of @xmath0 has been removed from the sn 2004dj data @xcite , whereas no isp has been removed from the sn 2008bk data , because it is estimated to be @xmath1 from the depolarizations seen across strong emission lines ( e.g. , h@xmath2 , ca ii ) at all epochs . ]
we have developed non - local - thermodynamic equilibrium , time - dependent radiative - transfer simulations specifically crafted for sne ii - p @xcite .
the simulations , based on the one - dimensional sn ii - p models of @xcite , compute the expected polarization ( in both lines and continuum ) as a function of time for an asphericity generated by a latitudinal density enhancement .
a significant finding is that temporal polarization changes in sne ii - p do not _ necessarily _ implicate a change in the asymmetry of the ejecta .
for instance , in figure 1 ( top , right ) we include the expected temporal evolution of the continuum polarization for an sn ii - p with an oblate equator - to - pole density enhancement of 1.67 , viewed edge - on .
this model reproduces what is observed for sn 2004dj quite well , with the sudden `` spike '' in polarization now explained not by a dramatic increase in asphericity but , rather , by the sharp reduction in optical depth ( i.e. , fewer electron scatters for outgoing photons ) that occurs at the plateau s end . the post - plateau polarization decline is also reproduced as a consequence of the expected @xmath3 dependence of the ejecta optical depth during the nebular phase @xcite .
while models with a dominant and fixed large - scale asymmetry fit the polarization behavior of sn 2004dj , the substantial early time polarization seen in sn 2008bk ( and sn 2004et ) is not replicated ; in fact , current models with equator - to - pole density enhancements as extreme as 5.0 are unable to produce significant ( i.e. , > 0.1% ) polarization during the mid - plateau @xcite .
the persistent late - time polarization seen in sn 2008bk also presents a modeling challenge .
it is worth noting that , in addition to their spectropolarimetric differences , sne 2008bk , 2004dj , and 2004et also exhibit photometric and spectroscopic differences @xcite , some of which are clearly evident for sn 2004dj and sn 2008bk in figure 1 , such as the rapidity of the plateau drop - off .
these differences may be associated with explosion physics but might also be related to the progenitor s core structure and the @xmath4-mixing properties ( morphology , amount , depth below the hydrogen - rich envelope ; see @xcite ) as well .
it is hoped that with the benefit of further modeling we may begin to connect all of these parameters and develop a consistent interpretation of the asymmetries indicated by the polarization of sne ii - p .
we thank for bruno leibundgut , stephane blondin , tom matheson , robert kirshner , malcom hicken , and brian schmidt for assistance with the observations .
dcl acknowledges support from nsf grant ast-1009571 , under which part of this research was carried out .
based on observations collected at the european organisation for astronomical research in the southern hemisphere , chile , under observing programs 081.d-0128 , 082.d-0151 , and 085.d-0391 ( pi : dessart ) . | the type ii - plateau supernova ( sn ii - p ) sn 2004dj was the first sn
ii - p for which spectropolarimetry data were obtained with fine temporal sampling before , during , and after its fall off of the photometric plateau the point that marks the transition from the photospheric to the nebular phase in sne ii - p .
unpolarized during the plateau , sn 2004dj showed a dramatic spike in polarization during the descent off of the plateau , and then exhibited a smooth polarization decline over the next two hundred days .
this behavior was interpreted by leonard et al .
( 2006 ) as evidence for a strongly non - spherical explosion mechanism that had imprinted asphericity only in the innermost ejecta . in this brief report , we compare nine similarly well - sampled epochs of spectropolarimetry of the type ii - p sn 2008bk to those of sn 2004dj .
in contrast to sn 2004dj , sn 2008bk became polarized well before the end of the plateau and also retained a nearly constant level of polarization through the early nebular phase . curiously , although the onset and persistence of polarization differ between the two objects , the detailed spectropolarimetric characteristics at the epochs of recorded maximum polarization for the two objects are extremely similar , feature by feature .
we briefly interpret the data in light of non - local - thermodynamic equilibrium , time - dependent radiative - transfer simulations specifically crafted for sn ii - p ejecta .
address = department of astronomy , san diego state university , san diego , ca
92182 , usa ; [email protected]
address = laboratoire dastrophysique de marseille , universit de provence , cnrs , 38 rue frdric - joliot - curie , f-13388 marseille cedex 13 , france address = department of physics and astronomy , university of pittsburgh , 3941 ohara street , pittsburgh , pa 15260 , usa address = departamento de ciencias fisicas , universidad andres bello , avda . republica 252 , santiago , chile , altaddress = departamento de astronomia , universidad de chile , casilla 36-d , santiago , chile |
let @xmath0 be a centrally - symmetric convex compact set with non - empty interior ( body " ) in euclidean space @xmath13 .
we write @xmath2 for the norm induced on @xmath3 by @xmath0 and @xmath14 for the support function of @xmath0 ; this is precisely the dual norm @xmath15 .
the parameters : @xmath16 where @xmath17 denotes the rotationally invariant probability measure on the unit euclidean sphere @xmath18 , play a central role in the asymptotic theory of finite dimensional normed spaces .
let @xmath19 denote the volume - radius of @xmath0 , where @xmath20 denotes lebesgue measure in the linear hull of @xmath21 and @xmath22 denotes the unit euclidean ball .
it is easy to check that : @xmath23 where @xmath24 is the polar body to @xmath0 , i.e. the unit - ball of the dual norm @xmath25 .
indeed , the left - hand side is a simple consequence of jensen s inequality after we express the volume of @xmath0 as an integral in polar coordinates , while the right - hand side is the classical urysohn inequality . in particular
, one always has @xmath26 . in the other direction , it is known from results of figiel tomczak - jaegermann @xcite , lewis @xcite and pisier s estimate @xcite on the norm of the rademacher projection , that for any centrally - symmetric convex body @xmath0 , there exists @xmath27 such that : @xmath28 where @xmath29 is a universal constant . throughout this note ,
unless otherwise stated , all constants @xmath30 denote universal numeric constants , independent of any other parameter , whose value may change from one occurrence to the next .
we write @xmath31 if there exist absolute constants @xmath32 such that @xmath33 .
the role of the linear map @xmath34 in ( [ eq : mm * ] ) is to put the body in a good position " , since without it @xmath35 can be arbitrarily large .
the purpose of this note is to obtain good upper bounds on the parameter @xmath36 , when @xmath0 is already assumed to be in a good position - the isotropic position .
a convex body @xmath0 in @xmath3 is called isotropic if it has volume @xmath37 , its barycenter is at the origin , and there exists a constant @xmath38 such that : @xmath39 it is not hard to check that every convex body @xmath0 has an isotropic affine image which is uniquely determined up to orthogonal transformations @xcite . consequently , the isotropic constant @xmath11 is an affine invariant of @xmath0
. a central question in asymptotic convex geometry going back to bourgain @xcite asks if there exists an absolute constant @xmath7 such that @xmath40 for every ( isotropic ) convex body @xmath0 in @xmath1 and every @xmath41 .
bourgain @xcite proved that @xmath42{n}\log n$ ] for every centrally - symmetric convex body @xmath0 in @xmath43 .
the currently best - known general estimate , @xmath42{n}$ ] , is due to klartag @xcite ( see also the work of klartag and e. milman @xcite and a further refinement of their approach by vritsiou @xcite ) . it is known that if @xmath0 is a centrally - symmetric isotropic convex body in @xmath3 then @xmath44 , and hence trivially @xmath45 .
it seems that , until recently , the problem of bounding @xmath36 in isotropic position had not been studied and there were no other estimates besides the trivial one .
the example of the normalized @xmath46 ball shows that the best one could hope is @xmath47 .
note that obtaining a bound of the form @xmath48 immediately provides a non - trivial upper bound on @xmath11 , since @xmath49 , and hence @xmath50 .
the current best - known upper bound on @xmath11 suggests that @xmath51 might be a plausible goal .
paouris and valettas ( unpublished ) proved that for every isotropic centrally - symmetric convex body @xmath0 in @xmath3 one has : @xmath52{\log ( e+n)}}{\sqrt[12]{n}l_k } .\ ] ] subsequently , this was extended by giannopoulos , stavrakakis , tsolomitis and vritsiou in @xcite to the case of the @xmath10-centroid bodies @xmath53 of an isotropic log - concave probability measure @xmath54 on @xmath55 ( see section 5 for the necessary definitions ) .
the approach of @xcite was based on a number of observations regarding the local structure of @xmath56 ; more precisely , lower bounds for the in - radius of their proportional projections and estimates for their dual covering numbers ( we briefly sketch an improved version of this approach in section 7 ) . in this work
we present a different method , applicable to general centrally - symmetric convex bodies , which yields better quantitative estimates . as always
, our starting point is dudley s entropy estimate ( see e.g. ( * ? ? ? * theorem 5.5 ) ) : @xmath57 where @xmath58 are the entropy numbers of @xmath0 . recall that the covering number @xmath59 is defined to be the minimal number of translates of @xmath60 whose union covers @xmath0 , and that @xmath61 .
our results depend on the following natural volumetric parameters associated with @xmath0 for each @xmath62 : @xmath63 where @xmath64 denotes the grassmann manifold of all @xmath9-dimensional linear subspaces of @xmath1 , and @xmath65 denotes orthogonal projection onto @xmath66 .
note that by the blaschke sanatal inequality and its reverse form due to bourgain and v. milman ( see section [ sec : prelim1 ] ) , it is immediate to verify that @xmath67 .
[ thm : covering2-intro ] for every centrally - symmetric convex body @xmath0 in @xmath1 and @xmath68 : @xmath69 by invoking carl s theorem ( see section [ sec : prelim1 ] ) , a slightly weaker version of theorem [ thm : covering2-intro ] may be deduced from the following stronger statement : [ thm : main - s - intro ] let @xmath0 be a centrally - symmetric convex body in @xmath1
. then for any @xmath70 there exists @xmath71 so that : @xmath72 and dually , there exists @xmath71 so that : @xmath73 a weaker version of theorem [ thm : main - s - intro ] , with the parameters @xmath74 , @xmath8 above replaced by : @xmath75 respectively , was obtained by v. milman and g. pisier in @xcite ( see theorem [ thm : milman - pisier ] ) .
our improved version is crucial for properly exploiting the corresponding properties of isotropic convex bodies . by ( essentially )
inserting the estimates of theorem [ thm : covering2-intro ] into ( with @xmath0 replaced by @xmath76 ) , we obtain that if @xmath0 is a centrally - symmetric convex body in @xmath1 with @xmath4 then : @xmath77 in the case of the centroid bodies @xmath56 of an isotropic log - concave probability measure @xmath78 on @xmath3 , one can obtain precise information on the growth of the parameters @xmath79 .
we recall the relevant definitions in section [ sec : prelim2 ] , and use to deduce in section [ sec : results ] that : @xmath80{q } } .\ ] ] in particular , since @xmath81 and @xmath82 , where @xmath83 denotes the uniform probability measure on @xmath21 , we immediately obtain : [ th : current - m - estimate ] if @xmath0 is a centrally - symmetric isotropic convex body in @xmath55 then : @xmath84{n}l_k}.\ ] ] it is clear that is not optimal .
note that if ( [ eq : mzq - intro ] ) were to remain valid until @xmath85 , we would obtain the bound @xmath86 , which as previously explained would in turn imply that @xmath87 , in consistency with the best - known upper bound on the isotropic constant .
we believe that it is an interesting question to extend the range where ( [ eq : mzq - intro ] ) remains valid . in section [ sec : results ] , we obtain such an extension when @xmath54 is in addition assumed to be @xmath88 ( see section [ sec : results ] for definitions ) . our entire method is based on pisier s regular versions of v. milman s @xmath89-ellipsoids associated to a given centrally - symmetric convex body @xmath0 , comparing between volumes of sections and projections of @xmath0 and those of its associated regular ellipsoids .
this expands on an approach already employed in @xcite .
we conclude the introduction by remarking that the dual question of providing an upper bound for the mean - width @xmath90 of an isotropic convex body @xmath0 has attracted more attention in recent years . until recently ,
the best known estimate was @xmath91 , where @xmath7 is an absolute constant ( see ( * ? ? ?
* chapter 9 ) for a number of proofs of this inequality ) .
the second named author has recently obtained in @xcite an essentially optimal answer to this question - for every isotropic convex body @xmath0 in @xmath3 one has @xmath92 .
let us introduce some further notation . given @xmath93 ,
we denote @xmath94 and @xmath95 .
a centrally - symmetric convex body @xmath0 in @xmath3 is a compact convex set with non - empty interior so that @xmath96 .
the norm induced by @xmath0 on @xmath3 is given by @xmath97 .
the support function of @xmath0 is defined by @xmath98 , with @xmath99 denoting the unit - ball of the dual - norm . by the blaschke santal inequality (
the right - hand side below ) and its reverse form due to bourgain and v. milman @xcite ( the left - hand side ) , it is known that : @xmath100 recall that the @xmath9-th entropy number is defined as @xmath101 a deep and very useful fact about entropy numbers is the artstein
szarek duality of entropy theorem @xcite , which states that : @xmath102 for every centrally - symmetric convex body @xmath0 and @xmath103 . in
what follows , a crucial role is played by g. pisier s regular version of v. milman s @xmath89-ellipsoids .
it was shown by pisier ( see @xcite or ( * ? ? ?
* chapter 7 ) ) that for any centrally - symmetric convex body @xmath0 in @xmath1 and @xmath104 , there exists an ellipsoid @xmath105 so that : @xmath106 where @xmath107 is a positive constant depending only on @xmath108 . given a pair of centrally - symmetric convex bodies @xmath109 in @xmath1 , the gelfand numbers @xmath110 are defined as : @xmath111 where @xmath112 .
we denote @xmath113 and @xmath114 .
carl s theorem @xcite relates any reasonable lorentz norm of the sequence of entropy numbers @xmath115 with that of the gelfand numbers @xmath116 . in particular , for any @xmath117 , there exist constants @xmath118 such that for any @xmath68 : @xmath119 and : @xmath120 in fact , pisier deduces the covering estimates of from an application of carl s theorem , after establishing the following estimates : @xmath121 our estimates depend on a number of volumetric parameters of @xmath0 , already defined in the introduction , which we now recall : @xmath122 and @xmath123 note that @xmath124 by ( [ eq : bs ] ) . also observe that @xmath125 is non - increasing by the alexandrov inequalities and kubota s formula , and that @xmath126 is non - decreasing by polar - integration and jensen s inequality .
we refer to the books @xcite and @xcite for additional basic facts from the local theory of normed spaces .
the main result of this section provides a general upper bound for the entropy numbers @xmath58 .
[ thm : covering1 ] let @xmath0 be a centrally - symmetric convex body in @xmath1 , and let @xmath68 . then : @xmath127 we combine this fact with dudley s entropy estimate @xmath128 ( see ( * ? ? ? * theorem 5.5 ) for this formulation ) . as an immediate consequence ,
we obtain : [ cor : covering1 ] let @xmath0 be a centrally - symmetric convex body in @xmath1 with @xmath129 .
then : @xmath130 dually , let @xmath0 be a centrally - symmetric convex body in @xmath1 with @xmath4 .
then : @xmath131 the first claim follows by a direct application of if we estimate @xmath58 using theorem [ thm : covering1 ] and the observation that @xmath132 for all @xmath103 ( recall that @xmath133 ) .
then , the second claim follows by duality since @xmath134 .
we will see in the next section that the supremum over @xmath135 above is unnecessary and that one may always use @xmath136 , only summing over @xmath62 .
but we proceed with the proof of theorem [ thm : covering1 ] , as it is a simpler approach .
_ proof of theorem [ thm : covering1]_. assume without loss of generality that @xmath9 is divisible by 3 , and use the estimate : @xmath137 where @xmath138 is pisier s @xmath139-regular @xmath89-ellipsoid associated to @xmath0 , with @xmath140 to be determined .
the first term is controlled directly by pisier s regular covering estimate ( [ eq : pisier - ek ] ) . for the second term we use the following simple fact about covering numbers of ellipsoids ( see e.g. ( * ? ? ?
* remark 5.15 ) ) : @xmath141 the latter equivalence follows since @xmath142 is the geometric average of the @xmath135 largest principal radii of @xmath143 , and so @xmath144 is non - increasing
. now recall that @xmath145 to estimate @xmath146 , we use a trivial volumetric bound : for any @xmath147 , @xmath148 for @xmath149 to be determined . consequently : @xmath150 and plugging this back into ( [ eq : step1 ] ) , we deduce : @xmath151 and hence : @xmath152 setting @xmath153 , we conclude that : @xmath154 combining everything , we obtain : @xmath155 setting @xmath156 , the assertion follows .
[ rem : duality ] theorem [ thm : covering1 ] implies the following dual covering estimate : @xmath157 indeed , this is immediate from the duality of entropy theorem and the fact that @xmath158 .
alternatively , one may simply repeat the proof of theorem [ thm : covering1 ] with the roles of @xmath0 and @xmath22 exchanged .
[ sec : diam ] this section may be read independently of the rest of this work , and contains a refinement of the following result of v. milman and g. pisier from @xcite , as exposed in ( * ? ? ?
* lemma 9.2 ) : [ thm : milman - pisier ] let @xmath0 be a centrally - symmetric convex body in @xmath1 .
then , for any @xmath159 : @xmath160 in other words , there exists @xmath71 so that : @xmath161 and dually , there exists @xmath71 so that : @xmath162 our version refines these estimates by replacing @xmath163 and @xmath164 above by the stronger @xmath74 and @xmath8 parameters , respectively ; this refinement is crucial for our application in this paper .
[ thm : main - s ] let @xmath0 be a centrally - symmetric convex body in @xmath1 .
then for any @xmath159 : @xmath165 in other words , there exists @xmath71 so that : @xmath166 and dually , there exists @xmath71 so that : @xmath167 our refinement will come from exploiting the full strength of pisier s result on the existence of regular @xmath89-ellipsoids .
in contrast , the milman
pisier result is based on v. milman s quotient - of - subspace theorem , from which it seems harder to obtain enough regularity to deduce our proposed refinement .
_ proof of theorem [ thm : main - s]_. given @xmath159 , let @xmath138 denote pisier s
@xmath139-regular @xmath89-ellipsoid , for some @xmath140 to be determined . by the second estimate in ( [ eq : pisier - ck ] ) , we know that there exists @xmath168 so that : @xmath169 for the ellipsoid @xmath170 , we may always find a linear subspace @xmath171 of codimension @xmath135 in @xmath172 so that : @xmath173 where @xmath174 is the grassmannian of all @xmath135-dimensional linear subspaces of @xmath172 . indeed , this is immediate by choosing @xmath175 to be the subspace spanned by the @xmath135 shortest axes of @xmath176 , and setting @xmath177 to be its orthogonal complement . consequently , there exists a subspace @xmath178 so that : @xmath179 we now deviate from the proof of our refined version , to show how one may recover the milman
pisier estimate ; the reader solely interested in the proof of our refinement may safely skip this paragraph .
assume for simplicity that @xmath180 .
by the first estimate in ( [ eq : pisier - ck ] ) , we know that there exists @xmath181 so that : @xmath182 given @xmath183 and denoting @xmath184 with @xmath185 $ ] , it follows that : @xmath186 setting @xmath187 , it follows from ( [ eq : basic ] ) that there exists @xmath188 so that : @xmath189 \right\ } } b_f .\ ] ] noting that the sequence @xmath190 is non - decreasing , and setting @xmath156 , we have found @xmath188 such that @xmath191 as asserted in ( [ eq : mp-2 ] ) ( with perhaps an immaterial constant @xmath192 instead of @xmath193 )
. the assertion of ( [ eq : mp-1 ] ) follows by duality . to obtain our refinement
, we will use instead of the first estimate in ( [ eq : pisier - ck ] ) , the covering estimate ( [ eq : pisier - ek ] ) ( which pisier obtains from ( [ eq : pisier - ck ] ) by an application of carl s theorem , requiring the entire sequence of @xmath194 estimates , not just the one for our specific @xmath9 ) .
setting @xmath136 , we use a trivial volumetric estimate to control @xmath195 , exactly as in the proof of theorem [ thm : covering1 ] : for any @xmath196 , @xmath197 together with ( [ eq : pisier - ek ] ) , we obtain : @xmath198 plugging this into ( [ eq : basic ] ) and setting as usual @xmath156 , the asserted estimate ( [ eq : pf - estimate ] ) follows . the other estimate ( [ eq : f - estimate ] ) follows by duality . as immediate corollaries
, we have : [ cor : covering2 ] for every centrally - symmetric convex body @xmath0 in @xmath1 , @xmath62 and @xmath199 : @xmath200 where @xmath201 is a constant depending only on @xmath108 .
this is immediate from theorem [ thm : main - s ] and carl s theorem .
note that @xmath202 is non - increasing , and so there is no difference whether we take the supremum on the right - hand - side just on the even integers .
[ cor : m ] for every centrally - symmetric convex body @xmath0 in @xmath1 so that @xmath129 , we have : @xmath203 dually , for every centrally - symmetric convex body @xmath0 in @xmath1 so that @xmath4 , we have : @xmath5 let us verify the first claim , the second follows by duality .
indeed , this is immediate from dudley s entropy estimate ( [ eq : dudley ] ) coupled with carl s theorem ( [ eq : carl2 ] ) : @xmath204 obviously @xmath205 for all @xmath9 , and so the assertion follows from the estimates of theorem [ thm : main - s ] . both corollaries should be compared with the results of the previous section .
[ rem : alternative]it may be insightful to compare theorem [ thm : main - s ] to some other known estimates on diameters of @xmath9-codimensional sections , besides the milman pisier theorem [ thm : milman - pisier ] .
one sharp estimate is the pajor tomczak - jaegermann refinement @xcite of v. milman s low-@xmath206 estimate @xcite : @xmath207 for any origin - symmetric convex @xmath60 and @xmath208 .
however , for our application , we can not use this to control @xmath209 since we do not a - priori know @xmath210 .
a type of dual low-@xmath89 estimate was observed by klartag @xcite : @xmath211 since @xmath212 is now well understood for an isotropic origin - symmetric convex body @xcite , this would give good estimates for low - dimensional sections ( large codimension @xmath9 ) , but unfortunately this is not enough for controlling @xmath36 .
klartag obtains the latter estimate from the following one , which is more in the spirit of the estimates we obtain in this work : @xmath213 again , this seems too rough for controlling the diameter of high - dimensional sections .
an absolutely continuous borel probability measure @xmath54 on @xmath55 is called @xmath214-concave if its density @xmath215 is of the form @xmath216 with @xmath217 convex .
note that the uniform probability measure on @xmath0 , denoted @xmath218 , is log - concave for any convex body @xmath0 .
the barycenter of @xmath54 is denoted by @xmath219 .
the isotropic constant of @xmath54 , denoted @xmath220 , is the following affine invariant quantity : @xmath221 where @xmath222 denotes the covariance matrix of @xmath54 .
we say that a @xmath223-concave probability measure @xmath78 on @xmath3 is isotropic if @xmath224 and @xmath225 is the identity matrix .
note that a convex body @xmath0 of volume @xmath37 is isotropic if and only if the log - concave probability measure @xmath226 is isotropic , and that @xmath227 indeed coincides with @xmath11 .
it was shown by k. ball @xcite that given @xmath41 : @xmath228 where the suprema are taken over all log - concave probability measures @xmath54 and convex bodies @xmath0 in @xmath1 , respectively ( see e.g. @xcite for the non - even case ) .
klartag s bound on the isotropic constant @xcite thus reads @xmath229 for all log - concave probability measures @xmath54 on @xmath1 .
given @xmath230 , we denote by @xmath231 the push - forward of @xmath54 via @xmath65 .
obviously , if @xmath54 is centered or isotropic then so is @xmath232 , and by the prkopa leindler theorem , the same also holds for log - concavity .
given a log - concave probability measure @xmath78 on @xmath3 and @xmath233 , the @xmath10-centroid body of @xmath54 , denoted @xmath53 , is the centrally - symmetric convex body with support function : @xmath234 observe that @xmath78 is isotropic if and only if it is centered and @xmath235 . by jensen
s inequality @xmath236 for all @xmath237 .
conversely , it follows from work of berwald @xcite or by employing borell s lemma ( see ( * ? ? ?
* appendix iii ) ) , that : @xmath238 when @xmath239 is the uniform probability measure on a centrally - symmetric convex body @xmath0 in @xmath1 , it is easy to check ( e.g. @xcite ) using the brunn minkowski inequality that : @xmath240 let @xmath54 denote an isotropic log - concave probability measure @xmath54 on @xmath1 .
it was shown by paouris @xcite that @xmath241 and that : @xmath242 conversely , it was shown by klartag and e. milman in @xcite that : @xmath243 this determines the volume radius of @xmath53 for all @xmath244 . for larger values of @xmath245 one can still use the lower bound : @xmath246 obtained by lutwak , yang and zhang @xcite via symmetrization .
we refer to the book @xcite for further information on isotropic convex bodies and log - concave measures .
[ sec : results ] let @xmath54 denote an isotropic log - concave probability measure on @xmath1 , and fix @xmath247 .
a very useful observation is that : @xmath248 it follows from that : @xmath249 furthermore , using , we see that : @xmath250{k},\frac{\sqrt{\min(q ,
k)}}{l_{\pi_h \mu}}\right ) } .\ ] ] unfortunately , we can only say in general that @xmath251{k}$ ] , and so the estimate ( [ eq : a4 ] ) is not very useful , unless we have some additional information on @xmath54 . recalling the definition of @xmath79 , we summarize this ( somewhat sloppily ) in : [ lem : vk - zq ]
let @xmath78 be an isotropic log - concave probability measure on @xmath3 .
for any @xmath233 and @xmath62 we have : @xmath252 assuming that @xmath253 we have : @xmath254 plugging these lower bounds for @xmath255 into either theorem [ thm : covering1 ] or corollary [ cor : covering2 ] coupled with remark [ rem : duality ] , we immediately obtain estimates on the entropy numbers @xmath256 .
similar estimates on the maximal ( with respect to @xmath257 ) in - radius of @xmath258 are obtained by invoking theorem [ thm : main - s ] .
[ thm : rqk - variant ] given @xmath259 and an integer @xmath208 , denote : @xmath260{k } ) } \frac{n}{k } \log\left ( e+ \frac{n}{k}\right ) \right\ } } .\ ] ] then , for any isotropic log - concave probability measure @xmath54 on @xmath1 : @xmath261 and there exists @xmath262 so that : @xmath263 from and lemma [ lem : vk - zq ] we have : @xmath264{m } ) } \right\ } } .\ ] ] then , it suffices to observe that : @xmath265{m})}\right\ } } & \simeq \sup_{1 { \leqslant}m { \leqslant}k } { \left\ { 2^{-\frac{k}{3 m } } \left ( \frac{1}{\sqrt{q}}+\frac{1}{\sqrt[4]{m}}\right ) \right\}}\\ & { \leqslant}c\left ( \frac{1}{\sqrt{q}}+\frac{1}{\sqrt[4]{k}}\right ) \simeq \frac{1}{\min ( \sqrt{q},\sqrt[4]{k})},\end{aligned}\ ] ] because @xmath266 for all @xmath267 , and @xmath268{m}$ ] attains its minimum at @xmath269 , so that @xmath270{m}){\leqslant}c/\sqrt[4]{k}$ ] .
we also use the fact that in a certain range of values for @xmath259 and @xmath68 , we might as well use the trivial estimates : @xmath271 which hold since @xmath272 .
an elementary computation based on corollary [ cor : m ] then yields a non - trivial estimate for @xmath273 .
it is interesting to note that without using the trivial information that @xmath274 ( or equivalently , the trivial estimates in ( [ eq : trivial ] ) ) , corollary [ cor : m ] would not yield anything meaningful .
[ thm : m - zq ] for any isotropic log - concave probability measure @xmath54 on @xmath1 : @xmath275{q } } .\ ] ] we use the estimate : @xmath276{k } ) } \frac{n}{k } \log\left ( e+ \frac{n}{k}\right ) \right\ } } , \ ] ] which follows from corollary [ cor : m ] combined with theorem [ thm : rqk - variant ] .
we set @xmath277 .
note that if @xmath278 then @xmath279 .
therefore , we may write : @xmath280{q}}.\end{aligned}\ ] ] the result follows . for larger values of @xmath245
, we obtain no additional information beyond the trivial monotonicity : @xmath281 if @xmath0 is an isotropic centrally - symmetric convex body in @xmath1 , using that @xmath282 is isotropic log - concave and that @xmath283 is isomorphic to @xmath284 , one immediately translates the above results to corresponding estimates for @xmath0 .
given @xmath208 , set : @xmath285{k } } \frac{n}{k } \log\left ( e+ \frac{n}{k}\right ) \right\ } } .\ ] ] then , for any isotropic centrally - symmetric convex body @xmath0 in @xmath1 : @xmath286 and there exists @xmath262 so that : @xmath287 moreover : @xmath288 it is interesting to perform the same calculations under the assumption that @xmath289 for any log - concave probability measure @xmath54 ( regardless of dimension ) . in that case : @xmath290 this would yield the following conditional result : given @xmath259 and an integer @xmath208 , denote : @xmath291 assuming that @xmath289 for any log - concave probability measure ( regardless of dimension ) , then for any isotropic log - concave probability measure @xmath54 on @xmath1 : @xmath261 and there exists @xmath262 so that : @xmath263 furthermore : @xmath292{q}}\;\;\hbox{for all}\;\ ; 2{\leqslant}q{\leqslant}(n\log n)^{2/3 } .\ ] ] consequently , for every isotropic convex body @xmath0 in @xmath3 one would have : @xmath293 finally , rather than assuming that @xmath220 is always bounded , we repeat the calculations for a log - concave measure @xmath54 which is assumed to be @xmath294-regular . recall that @xmath54 is called @xmath294 with constant @xmath295 ( @xmath296 $ ] ) if : @xmath297 note that this property is inherited by all marginals of @xmath54 , and that any log - concave measure is @xmath298 with @xmath299 a universal constant .
it was shown by klartag and e. milman @xcite that when @xmath54 is a @xmath294 log - concave probability measure on @xmath1 with constant @xmath295 , then : @xmath300 and : @xmath301 this implies that for such a measure , for any @xmath247 : @xmath302 by , we know that : @xmath303 unfortunately , since we only know that : @xmath304 we again see that the maximum in ( [ eq : lyz ] ) is always attained by the @xmath305 term .
summarizing , we have : let @xmath78 be an isotropic log - concave probability measure on @xmath3 which is @xmath294 with constant @xmath295 for some @xmath296 $ ] .
then for any @xmath233 and @xmath62 we have : @xmath306 plugging this estimate into the general results of sections [ sec : covering ] and [ sec : diam ] , we obtain : let @xmath54 denote an isotropic log - concave probability measure on @xmath1 which is @xmath294 with constant @xmath295 for some @xmath296 $ ] .
given @xmath259 and an integer @xmath208 , denote : @xmath307 then : @xmath261 and there exists @xmath262 so that : @xmath263 furthermore : @xmath292{q}}\;\;\hbox{for all}\;\ ; 2{\leqslant}q{\leqslant}c \frac{(n \log ( e+n))^{\frac{2\alpha } { \alpha + 4}}}{b_\alpha^\frac{4 \alpha}{\alpha+4 } } .\ ] ] consequently , for every isotropic convex body @xmath0 in @xmath3 so that @xmath218 is @xmath308 with constant @xmath309 , one has : @xmath310 better estimates for the entropy - numbers @xmath311 and gelfand numbers @xmath312 may be obtained for various ranges of @xmath9 by employing the alternative known estimates mentioned in remark [ rem : alternative ] .
however , these do not result in improved estimates on @xmath313 , which was our ultimate goal .
we therefore leave these improved estimates on the entropy and gelfand numbers to the interested reader .
we only remark that even the classical low-@xmath206 estimate ( [ eq : low - m * ] ) coupled with our estimate on @xmath313 yield improved estimates for @xmath314 and @xmath194 in a certain range - a type of bootstrap " phenomenon .
in this section we briefly describe an improved and simplified version of the arguments from @xcite and compare the resulting improved estimates to the ones from the previous section . following the general approach we employ in this work , the arguments are presented for general centrally symmetric convex bodies , and
this in fact further simplifies the exposition of @xcite . for the proof of theorem [ th : projzq - main ] , we use a sort of converse to carl s theorem ( [ eq : carl ] ) on the diameter of sections of a convex body satisfying @xmath193-regular entropy estimates , which is due to v. milman @xcite ( see also ( * ? ? ?
* chapter 9 ) ) .
_ proof of theorem [ th : projzq - main]_. given @xmath320 , let @xmath138 denote pisier s
@xmath139-regular @xmath89-ellipsoid , for some @xmath140 to be determined . instead of directly using pisier s estimate ( [ eq : pisier - ck ] ) on the gelfand numbers as in the proof of theorem [ thm : main - s ] to deduce the existence of @xmath168 so
that : @xmath321 the starting point in @xcite are the more traditional covering estimates ( [ eq : pisier - ek ] ) : @xmath322 in @xcite , the following estimate was used ( see ( * ? ? ? * theorem 5.14 ) ) : @xmath323 however , this estimate does not take into account the regularity of the covering .
consequently , a significantly improved estimate is obtained by employing lemma [ lem : milman - m*-covering ] ( and the subsequent remark ) which exploits this regularity : @xmath324 even with this improvement , note that this is where the current approach incurs some unnecessary logarithmic price with respect to the approach in the previous sections : instead of using ( [ eq : compare1 ] ) directly , one uses ( [ eq : compare2 ] ) which pisier obtains from ( [ eq : compare1 ] ) by applying carl s theorem , and then uses the converse to carl s theorem ( lemma [ lem : milman - m*-covering ] ) to pass back to gelfand number estimates . the rest of the proof is identical to that of theorem [ thm : main - s ] . for the ellipsoid @xmath327
we may always find a linear subspace @xmath171 of codimension @xmath9 in @xmath172 so that : @xmath328 estimating @xmath329 by comparing to @xmath330 via the dual covering estimate on @xmath331 ( note that there is no need to use the duality of entropy theorem here ) , we obtain : @xmath332 combining all of the above , we deduce the existence of @xmath71 so that : @xmath333 this concludes the proof .
having obtained a rather regular estimate on the gelfand numbers , the next goal is to obtain an entropy estimate . to this end
, one can use carl s theorem ( [ eq : carl ] ) or ( [ eq : carl2 ] ) , as we do in section [ sec : diam ] .
the approach in @xcite proceeds by employing an entropy extension theorem of litvak , v. milman , pajor and tomczak - jaegermann @xcite .
we remark that this too may be avoided , by employing the following elementary covering estimate ( see e.g. ( * ? ? ?
* chapter 9 ) ) : [ lem:6.5.3 ] let @xmath0 be a symmetric convex body in @xmath55 and assume that @xmath334 for some @xmath335 .
let @xmath336 be a subspace of @xmath55 with @xmath337 and @xmath338 .
then , we have @xmath339 finally , having a covering estimate at hand , the estimate on @xmath36 is obtained by dudley s entropy bound ( [ eq : dudley ] ) . plugging in the lower bounds on @xmath255 given in section [ sec : results ] , the results of @xcite are recovered and improved . as the reader may wish to check , the improved approach of this section over the arguments of @xcite yields estimates which are almost as good as the ones obtained in section [ sec : results ] , and only lose by logarithmic terms .
the first named author acknowledges support from the programme api@xmath340teia ii " of the general secretariat for research and technology of greece . the second named author is supported by isf ( grant no .
900/10 ) , bsf ( grant no .
2010288 ) , marie - curie actions ( grant no .
pcig10-ga-2011 - 304066 ) and the e. and j. bishop research fund .
, _ a note on a low @xmath346-estimate _ , in geometry of banach spaces , proceedings of a conference held in strobl , austria , 1989 " ( p. f. mller and w. schachermayer , eds . ) , lms lecture note series , vol .
158 , cambridge university press ( 1990 ) , 219229 .
, _ isotropic position and interia ellipsoids and zonoids of the unit ball of a normed @xmath347-dimensional space _ , in _ geometric aspects of functional analysis _
, volume 1376 of _ lecture notes in mathematics _ , pages 64104 .
springer - verlag , 1987 - 1988 . | let @xmath0 be a centrally - symmetric convex body in @xmath1 and let @xmath2 be its induced norm on @xmath3 . we show that if @xmath4 then : @xmath5 where @xmath6 is the mean - norm , @xmath7 is a universal constant , and @xmath8 denotes the minimal volume - radius of a @xmath9-dimensional orthogonal projection of @xmath0 .
we apply this result to the study of the mean - norm of an isotropic convex body @xmath0 in @xmath3 and its @xmath10-centroid bodies . in particular
, we show that if @xmath0 has isotropic constant @xmath11 then : @xmath12{n}l_k } .\ ] ] |
the calculation of the viscosity coeffcient in a hot and dense system is of interest both in the fields of high energy heavy ion collisions and astrophysics [ 1 - 4 ] .
the dissipative processes in a quark - gluon plasma(qgp ) supposed to be formed in ultrarelativstic heavy collision can be described by the viscosity coefficient .
because of the estimated very short collision time of the ions in the heavy ion collisions , the equilibrium statistical approaches are probably inappropriate for an adequate description of such a process , and the dissipative phenomena are important , at least for the equilibration and the expansion phases of the system . in principle , the dissipative processes in heavy ion collisions should be described by non - equilibrium dynamical theory that is based on qcd .
there are two methods to calculate the shear viscosity in qgp .
the first one is by use of the relativstic kinetic theory .
starting from the boltzmann equation , the viscosity coefficient @xmath1 can be derived containing the transport cross section[2,3 ] .
the second one is from the kubo formulas[2 ] .
for example , in the relaxation time approximation , the shear viscosity coefficient was found to be @xmath2 , the constant @xmath3 was estimated to be @xmath4,@xmath5,or @xmath6,whereas a variational calculation gave @xmath7 with two quark flavors in high temperature limit[2,3 ] . and ref.3 also gave the result beyond the leading logarithm approximation , @xmath8.\ ] ] on the other hand , basing on the kubo formula , ref.2 gave @xmath9 and the lattice calculations obtained the value near the phase transition as @xmath10 [ 2 ] .
it is known that the ensential nature of particles in a plasma at finite temperature is that they no longer have a infinite lifetimes , but have finite widths becaue of the collective effects .
that is quite different from the behavior at zero temperature[6,7 ] . in qgp debye screening and damping of collective modes
occur because of the interactions among particles and the heat bath .
therefore it is reasonable to take into account both the debye screening and finite width effects when study the collective behaviors of the qgp , , which might also influence the transport processes of qgp . on the other hand , so far most all of the results of the shear viscosity cofficients in qgp are calculated at finite temperature without taking into account of the baryon density . certainly by lattice simulation
it is difficult to include the baryon density , which is related to the baryo - chemical potential @xmath11 through @xmath12 and measures the deviation from the balance of quarks and and antiquarks .
however at rhic energies , there might still exist a considerable amount of stopping to @xmath13[5 ] , @xmath14 is the quark chemical potential . in this paper
we will investigate the damping rate effect of the transverse gluon on the shear viscosity coefficient of qgp at both finite temperature and chemical potential quark chemical potential .
we will show explicitly both the finite width and chemical potential cause considerable enhancements to the shear viscoisity .
the paper will be organized as follows . in sec .
ii we will get the shear viscosity of quarks and gluons via real time qcd by considering both the debye screening and finite width effects at finite chemical potential .
in sec.iii we will consider the running behavior of the coupling constant ane present numerical results .
finally we will summarize our results in sec.iv .
we will calculate the shear viscosity coefficient by using the relativistic kinetic theory for a massless qgp in the so - called relaxation time approximation[2,3 ] : @xmath15 where @xmath16 is the energy density of particles of type @xmath17 in the qgp and @xmath18 is the mean free path , which is the inverse interaction rate : @xmath19 .
considering the large angle scattering dominate the transport processes , we should multiply the interaction rate by a factor @xmath20 , @xmath21 which is the so - called transport interaction rate , and @xmath22 is the scattering angle in the center of mass system .
there are two equivalent ways of computing the interaction rates either using the matrix elements or from imaginary parts of the quark or gluon self - energies [ 3 ] .
@xmath23 let us consider first the quark self - energy shown in fig.1 , where we have included the screening effects and the damping rate effects by using the effective gluon propagator[8 ] . for hard quark momentum
@xmath24 it is sufficient to use the bare quark propagator and bare vertices .
we will calculate the imaginary part of the self - energy at finite temperature and chemical potential using the thermo field dynamics .
for the hard particles one can show that the main contributions to the interaction rate comes from the soft momentum transfer range , i.e.@xmath25 in the weak coupling limit[3 ] . using the usual feynman rules of qcd to evaluate the diagram of fig.1 , one can obtain : @xmath26 where @xmath27 is the effective gluon propagator , in covariant gauge[9 ] , @xmath28 where @xmath29 @xmath30 where @xmath31 is the gauge parameter . in the static limit the contribution of the hard thermal loops for hot dense qcd are[5 ] , @xmath32 for the longitudinal part and @xmath33 for the transverse part of gluon polarization tensor .
substituting eqs.(7),(8 ) into eq(6),one gets @xmath34 in a thermal system , because of the thermal effects and the particles interactions , the particles ( quasi - particles ) will no longer have infinite lifetime , namely there are damping for the particles in qgp , which is an important feature of qgp .
if one take into account the damping rate of the transverse gluon @xmath35[3,8 ] , the transverse gluon propagator can be written as [ 6,7 ] @xmath36 by using eqs(5)-(10),one can cast eq(4 ) into : @xmath37 where @xmath38 .
for finite temperature qcd , we will use the hot dense propagators for quarks and gluons in @xmath39 matrix in tfd[9 ] . among them the hot propagator of quark has the form @xmath40 @xmath41 where @xmath42 and for the gluons @xmath43 @xmath44 @xmath45 @xmath46 @xmath47 @xmath48n_b(q)e^{-\beta|q_0|/2 } , \eqno(13)\ ] ] where @xmath49 is the bose distribution function and for soft mementum @xmath50 @xmath51
it is known that one can get the imginary part of the retard green function of the quark self - energy from the imaginary part of @xmath52 type green function , which is very convenient to calculate in tfd[10 ] @xmath53 where @xmath54 @xmath55 then one can obtain @xmath56 substituting eq(16),into eq(2 ) , and assuming the transfer momentum @xmath50 is soft : @xmath57 , @xmath58 we obtain @xmath59 @xmath60 which shows that the gauge dependent part is suppressed by the square of the coupling constant @xmath61 by evaluating the integration over @xmath50 in @xmath57 , we obtain : @xmath62 the transport interaction rate of gluon can be obtained in a similar way from the gluon self - energy , and it was found that @xmath63 one can evaluate the quark and gluon energy density in qgp via the vaccum graphs in hot qcd both at finite temperature and chemical potential:[1 ] @xmath64 @xmath65 by making use of eqs(1),(2),(19 ) , and assuming @xmath66 , @xmath35[8 ] we can get the shear viscosity coefficient of quarks for two flavors and gluons respectively @xmath67 @xmath68 @xmath69 in order to compare with previous calculations , one may come back to vanishing quark chemical potential case and not take the gluon damping rate into account , namely substituting @xmath70 in eqs(20)-(22),then one obtains the shear viscosity coefficient of qgp with two quark flavors at vanishing quark chemical potential without including the finite width effect : @xmath71,\eqno(23)\ ] ] which agrees with the previous result in ref.3 beyond the leading logarithm approximation in the weak coupling limit except a factor @xmath72 due to the definition . at leading order and leading logarithm approximation
, one can cast eq(23 ) into @xmath73 the coefficient @xmath74 is remarkably close to the one found by thoma @xmath75 [ 3 ] and the one found by baym et al.[2 ] , the reason of the difference is we used @xmath76 other than @xmath77 , @xmath78 .
before our numerical analysis , we consider the running behavior of the coupling constant g. in principle the running coupleing constant is governed by the renormalarization equations of qcd at finite temperture and chemical potential , which is still under investigation .
hereby we only use the following result[12 ] : @xmath79 where is the qcd papameter , and we choose @xmath80 as usual .
and @xmath81 is the number of flavor , we will only consider two flavors . substituting the running coupling constant eq(25 ) into egs(20)-(22 ) , we obtain the experessions of the shear viscosity coefficients as functions of temperature and chemical potential .
so we can plot the curves of the quark and gluon viscosity coefficients against temperature in fig.2(a ) and fig .
3(a ) for different quark chemical potentials @xmath82 with and without including the gluon damping effect in qgp respectively . from fig.2(a ) one can find the quark chemical potential effect may enhance the quark viscosity coefficient tremendously because the increase of quark energy density in qgp , while the chemical potential effect on the gluon viscosity is almost negligible . from fig .
3(a ) we find that the three curves of gluon viscosity coefficient against @xmath83 at different chemical potential almost meet together both with and without taking into acount the gluon finite width . on the other hand
the gluon damping effect causes large enhancements to the shear viscosities both for quark and gluon .
the reason is that the damping effect has decreased their transport interaction rates , and hence increase their mean free paths.this qualitatively agrees with the analyses in ref.4 for the nuclear matter .
fig.2(b ) and fig.3(b ) show the temperature dependence of the scaled viscosity coefficients @xmath84 for different quark chemical potentials @xmath82 with and without including the gluon damping effect .
our result agrees with the kubo formula result in ref.2 @xmath85 at vanish baryon density , but our result is much bigger than the lattice estimate near the critical temperature in ref.2 .
fig.2(b ) and fig.3(b ) showes at vanish chemical potential the curves of scaled viscoisity coefficients are no longer flat lines because of the running coupling constant . on the other hand ,
the tendencies are quite different between the scaled viscoisity of quark and gluon .
the reason is the quark chemical potential plays different roles in them .
neverthless , the finite width still cause considerable enhancement to both the scaled viscosity coefficients now we study how the shear viscosity coefficients vary with the coupling constant
. we plot the curves of the scaled viscosity coefficients @xmath84 against the coupling constant at a fixed temperature @xmath86 and different quark chemical potentials @xmath82 with and without including the gluon damping effect in fig.2(c ) and fig.3(c ) .
the curves are almost above the navier - stockes regime and in agreement with thoma s result .
also the tendencies are quite different between the scaled viscoisity of quark and gluon via the coupling constant because of quark chemical potential plays different roles in them . however , the finite width still cause considerable enhancement to both the scaled viscosity coefficients .
one interesting thing is that , although our calculations are valid for weak coupling limit @xmath87 , fig.2(c ) and fig.3(c ) show that the results can be extrapolated to higher coupling constant at fixed temperatur @xmath86 , as predicted in ref.3 at least for quantities of energetic particles with @xmath88 .
we have examined the quark and gluon shear viscosity behaviors by real time qcd both at finite temperature and baryon density in the quark gluon plasma in which both debye screening and damping or finite width are regarded as important underlying features .
also the running feature of the coupling constant is accounted for .
we have seen that both the quark and gluon shear viscosity coefficients are increased substantially , due to the effects of the finite width of transverse gluons .
meanwhile all the scaled shear viscosity coefficients are increased in comparison with the previors results , in which the finite widtheffect was not accounted for .
we have also examined the quark chemical potential effects to the shear viscosity coefficients in a straight forward way .
it also enhances the shear viscosity coefficients of qgp consiberably , but the enhancements are smaller than that caused by the finite width effect . on the other hand , we have seen the tendencies of the quark viscosity coefficients , especially the scaled ones , differ quite substantially from that of the gluon .
this is because the baryon density plays different roles in quark and gluon shear viscosity coefficients .
we note that if we remove the quark chemical potential and finite width the results will be in good agreement with previous calculations [ 2,3,13 ] .
we conclude that both the gluon damping rate and quark chemical potential effects have a considerable influence on the dissipative processes of qgp .
this dissipation can not be neglected in hydrodynamic descriptions of the expansion phase of qgp in ultrarelativistic heavy ion collisions .
this work is supported partly by the national natural science funds of china .
hou defu wants to thank the deutsche forschungsgemeinschaft ( dfg ) for supports .
the authors are indebted to peter henning and ulrich heinz for valuble discussion and comments , and whish to thank liu lianshou for his help .
[ 1 : : ] mller , in nato asi series b , physics vol.303(plenum press , newyork , london,1992 ) ; u.heinz , k.kajantie and t.toimela , annals of phys.*176*,218(1987 ) ; j .-
blaizot , nucl . phys . *
a566 * , ( 1994)333c .
[ 2 : : ] a.hosoya and k.k .
kajantie , nucl.phys.*b250*,666(1985 ) ; p. danielewicz and m.gyulassy , phys.rev.*d31*,53(1985 ) ; + g.baym , h. monien , c.j.pethick and d.g.ravenha , phys.rev.lett.*64 * , 1867(1990 ) .
[ 3 : : ] m.h.thoma , phys.rev.*d49*(1994)451 ; m.h.thoma , phys.lett*b*(1991)144 .
[ 4 : : ] p.a.henning , nucl.phys.*a567*,844(1994 ) .
[ 5 : : ] h. vija and m.h.thoma phys.lett*b342*,212(1995 ) .
[ 6 : : ] n.p.landsman , and ch.g.van weert , phys.rep . * 145*,141(1987 ) .
[ 7 : : ] p.henning phys.rep.*253*,235 ( 1995 ) ; hou defu , li jiarong , z.phys.*c71*,503(1996 ) hou defu , s. ochs , li jiarong , phys . rev.*d54*,(1996)(in press ) [ 8 : : ] r.d.pisarski , phys.rev.lett . * 63*,1129(1989 ) ; e.braaten , r.d.pisarski , phys .
rev.*d42 * , 2156(1990 ) , phys.rev.lett . , * 64*,1338(1990 ) .
[ 9 : : ] r.kobes , phys . rev .
* d42*,562 ( 1990 ) ; phys . rev . *
d43*,1269(1991 ) .
[ 10 : : ] y.fujimoto , m. morikawa and m. sasaki , phys . rev .
* d33*,590(1986 ) .
[ 11 : : ] y. fujimoto and hou defu , phys . lett . *
b335 * , ( 1994)87 .
[ 12 : : ] c.p.singh b.k.patra , and saeed uddin , phys.rev .
* d49*,4023(1994 ) .
[ 13 : : ] p.rehberg , s.p . klevansky and j.hfner , nucl.phys.a608 , 356(1996 ) .
fig.1 : : the resummed one - loop quark self - energy graph [ fig.1 ] fig.2(a ) : : the quark viscosity coefficient as a function of the temperture for + @xmath89 ( bottom to top ) without ( thin solid lines in lower part ) and with ( thick solid lines in upper part)including the gluon damping effect .
+ [ fig.2(a ) ] fig.2(b ) : : the scaled quark viscosity coefficient @xmath84 as a function of the temperture for @xmath89 ( bottom to top ) without ( thin solid lines in lower part ) and with ( thick solid lines in upper part)including the gluon damping effect .[fig.2(b ) ] fig.2(c ) : : the scaled quark viscosity coefficient @xmath84 as a function of coupling constant @xmath90 at fixed temperature @xmath86 for @xmath89 ( bottom to top ) without ( thin solid lines in lower part ) and with ( thick solid lines in upper part)including the gluon damping effect .[fig.2(c ) ] fig.3(a ) : : the gluon viscosity coefficient as a function of the temperture for @xmath89 ( meet together ) without ( thin solid lines in lower part ) and with ( thick solid lines in upper part)including the gluon damping effect .[fig.3(a ) ] fig.3(b ) : : the scaled gluon viscosity coefficient @xmath84 as a function of the temperture for @xmath89 ( bottom to top ) without ( thin solid lines in lower part ) and with ( thick solid lines in upper part)including the gluon damping effect .[fig.3(b ) ] fig.3(c ) : : the scaled gluon viscosity coefficient @xmath84 as a function of coupling constant @xmath90 at fixed temperature @xmath86 for @xmath89 ( bottom to top ) without ( thin solid lines in lower part ) and with ( thick solid lines in upper part)including the gluon damping effect .[fig.3(c ) ] | by considering the debye screening and the damping rate of gluons , the viscosity cofficient of the quark - gluon plasma was evaluated via the real time finite temperature qcd in the relaxiation time approximation at finite temperature and chemical potential .
the results show that both the damping rate and quark chemical potential cause considerable enhancements to the viscosity coefficient of the hot dense quark - gluon plasma . * pcac number(s ) : 12.38.mh , 12.38.bx * institute of particle physics , hua - zhong normal university + wuhan 430070 , china + e - mail : [email protected] + @xmath0 present address : institute fr theoretische physik , + 93040 uni regensburg , germany .
+ e - mail : [email protected] + |
optical atomic clocks have surpassed state - of - the art microwave clocks in terms of accuracy and stability and allow targeting fractional inaccuracies of 1 part in @xmath4 @xcite .
a crucial element in achieving this performance are ultrastable lasers which currently limit the short term stability of state - of - the - art optical clocks @xcite .
thus , development of techniques to improve the performance of the ultrastable interrogation lasers is a rapidly evolving field @xcite .
already today , ultrastable lasers are demanding and cost - intensive devices , and it is becoming apparent , that future 100 mhz linewidth lasers will even more increase in complexity and costs , making these systems affordable possibly only in a few laboratories .
thus , methods are required , to provide these ultrastable optical frequencies to remote laboratories over large distances with highly demanding short term stability . direct comparisons of optical clocks or even only state - of - the - art lasers are difficult , since today , the complex setup of an optical clock does not allow for transportation .
since the first demonstrations @xcite , phase - coherent long - range dissemination and remote characterization of ultrastable optical frequencies have therefore become an important tool in frequency metrology .
the most promising method with the highest demonstrated stability directly transmits the ultrastable optical carrier via phase - stabilized fibers @xcite and was applied to remote optical clock comparisons over distances of 3.5 km @xcite .
long - range comparisons recommend usage of commercial telecommunication fiber networks and transfer at 1.5 @xmath5 m @xcite . in this work
we realized a remote comparison of 73 km - distant , independent ultrastable optical frequencies on the @xmath6 level in fractions of a second via a long - distance telecommunication fiber network .
this represents a remote optical frequency comparison with an unprecedented resolution on short time scales .
the focus of our work was to explore the capability of long - distance fiber links for comparing optical frequencies with high short term stability .
thus , investigating possible limitations on the achievable short term performance of the whole dissemination chain .
the achieved short term stability at 100 ms is a tenfold improvement to previous remote comparisons of optical frequencies .
the demonstrated performance allows remote dissemination of the stability of 100 mhz linewidth lasers within seconds . with our setup
we are able to reveal the flicker floor of our ultrastable lasers of @xmath0 in only 0.1 s averaging time .
this represents a real - time application of a long - distance fiber link at the performance - level of state - of - the - art optical clocks . in combination with sophisticated fiber stabilization systems @xcite ,
the results confirm the potential of long - distance fiber links for comparing optical clocks at the level of @xmath2 or below in a few minutes .
the lasers compared in this work are separated by more than 50 km geographical distance .
the individual laser systems are used for a calcium optical frequency standard located at the physikalisch - technische bundesanstalt ( ptb ) in braunschweig and for a magnesium optical frequency standard at the institute of quantum optics ( iq ) at the university of hanover .
the experimental setup is schematically depicted in fig .
[ setup ] .
in hanover , currently two ultrastable laser systems are in operation , l1 and l2
. both are diode lasers systems at 914 nm , which are stabilized to two independent ultrastable optical resonators .
l1 uses a horizontal resonator ( finesse @xmath7 ) mounted in a multistage spring pendulum configuration for vibration isolation .
l2 is stabilized to a horizontal resonator ( @xmath8 ) mounted near the symmetry plane for reduced vibrational sensitivity , similar to a design used in @xcite .
these systems are compared to an ultrastable laser located at ptb via a 73 km long telecommunication fiber to iq .
the ultrastable master laser ( ml ) at ptb is a diode laser system at 657 nm stabilized to a vibrationally insensitive optical resonator reaching a demonstrated linewidth at hz level @xcite and a flicker floor of @xmath9 for @xmath10 s @xcite .
its stability is transferred to a 1542 nm fiber laser ( tl ) by means of a femtosecond frequency comb @xcite .
we inject about 5 mw of the light of tl into the 73 km long fiber link to iq , connected via the local computer center in hanover ( rz - h ) . at iq , the light
is amplified in a bidirectional erbium doped fiber amplifier ( edfa ) , frequency shifted with an acousto - optic modulator ( aom 2 ) and afterwards partially reflected back to ptb , using a faraday rotator mirror . at ptb ,
the back reflected light is used for active cancellation of phase noise of the fiber by servo feedback to aom 1 .
the servo bandwidth of the noise cancellation loop is limited to @xmath11 700 hz due to fiber length @xcite .
the single pass loss in the 73 km fiber is @xmath11 23 db .
information about the stabilization system and the link between ptb and rz - h can be found in @xcite .
part of the transmitted light at 1542 nm is coupled out at iq and used for comparison to l1 and l2 at 914 nm by means of a second frequency comb .
both frequency combs are based on femtosecond erbium fiber lasers which are frequency doubled into the visible .
thus , optical frequencies even far outside from the telecommunication window at 1.5 @xmath5 m can be remotely compared via the fiber link using the transfer laser at 1542 nm without degrading the stability or accuracy @xcite .
the remote comparison of the ultrastable optical frequencies is performed in the time and frequency domains . in the time domain ,
all frequencies are recorded simultaneously by using a multichannel frequency counter with synchronous readout and zero deadtime .
this can be operated either with a simple single average per gate time ( so called @xmath12 estimator ) or using overlapping multiple averages within the gate period ( overlapping @xmath13 estimator ) @xcite .
the frequency of the transfer beat @xmath14 between the delivered cw light @xmath15 at 1542 nm and the 914 nm cw light @xmath16 of the local ultrastable lasers is calculated using the relation @xmath17 by measuring the mode numbers @xmath18 , @xmath19 of the frequency comb , the carrier - envelope offset frequency @xmath20 , and the beat note frequencies @xmath21 , @xmath22 between the cw frequencies and the corresponding comb modes .
@xmath15 at 1542 nm is thereby measured using the fundamental comb spectrum , @xmath23 , and @xmath16 at 914 nm using the frequency doubled comb spectrum , @xmath24 , with @xmath25 the repetition rate of the frequency comb .
first , we performed a remote comparison between the three independent laser systems .
this enabled us to identify the system that provides the best stability for the investigation of the achievable short term performance of the remote comparison .
the results of the remote characterization of the three independent laser systems are depicted in fig .
[ stability ] . the allan deviation ( adev ) shown
is obtained by operating the counter with a simple single average per gate time ( @xmath12 estimator ) .
l2 shows a drift rate of several @xmath6/s and l1 of @xmath6/s , which we attribute to temperature fluctuations of the ultrastable resonators .
the drift rate of ml , and thus tl , was identified independently at ptb using a hydrogen maser as @xmath26/s . with linear drift correction , for @xmath27 s tl is identified as the most stable laser , and l2 as more stable than l1 ( fig .
[ stability ] ) . also shown in fig .
[ stability ] is the link instability of @xmath28 , estimated @xcite from independent ptb - iq - ptb round - trip measurements . according to these estimates ,
it is reasonable that the observed measurement performance between l2 and tl is limited by the link instability up to 1 s. to go beyond this limitation and to check for other possible noise sources that are not only introduced by the fiber link itself but also by other components of the whole dissemination chain , as e.g. the two independent frequency combs , we performed phase - averaged data acquisition .
we implemented a data evaluation based on the modified allan deviation ( modadev ) , which is obtained ( case ( c ) in @xcite ) by operating the counter with overlapping multiple averages per gate time ( overlapping @xmath13 estimator , with an internal gate time of 1 ms ) . while leaving the instability contributions of frequency modulation noise processes unaffected ( within 30% error @xcite ) , the contribution of white phase noise - which is typical for residual link fluctuations - falls off as @xmath29 .
figure [ modallan ] depicts the modadev for the l2-tl comparison .
also shown is the instability of the 73 km link , as estimated from independent round - trip measurements ptb - iq - ptb .
the modadev of the link behaves as @xmath30 .
thus , at 1 s measurement time , the link instability contribution is reduced by a factor of 30 .
the l2-tl trace coincides with the estimated link instability for very short time scales , i.e. no significant additional measurement noise or laser noise could be detected .
the performance of the remote laser comparison is already revealed after 100 ms and shows a flicker floor of @xmath0 .
the combined thermal noise contributions of the resonators of l2 and ml is calculated @xcite to be @xmath31 .
the remote measurement results are close to the performance of state - of - the - art ultrastable lasers , as represented in fig .
[ modallan ] by the previous results of the _ local _ comparison of ml to the ultrastable interrogation laser yl of the yb@xmath32 experiment at ptb via 300 m of stabilized fiber @xcite .
the achieved short term stability for the remote comparison is superior to any reported remote comparison of optical frequencies and shows an improvement at 100 ms of one order of magnitude to any published results . for analysis in the frequency domain , we additionally implemented the operation in eq .
( [ transfer ] ) ( divided by 4 ) by analog signal processing @xcite using a direct digital synthesizer and double - balanced mixers .
the spectrum of the remote transfer beat note @xmath33 , at 48.5 thz , between l2 and tl is depicted in fig .
[ transferbeat ] .
it shows a linewidth of 1 hz , limited by the spectrum analyzer used . by analyzing the spectrum of the remote beat signal , we optimized the laser system l2 in real - time .
the beat signal also enables remote phase locking of a laser to the transferred light at the noise level of the stabilized fiber .
to conclude , we performed a remote comparison of ultrastable optical frequencies on the @xmath6 level on sub - second time scales .
modadev based data evaluation suppresses residual fiber noise below @xmath34 in 1 s and enables rapid resolution of the transferred signal .
this displays the enormous potential of dissemination of ultrastable optical frequencies using long - distance telecommunication fiber networks for frequency metrology .
exciting applications are comparisons of optical clocks , remote high - precison spectroscopy or gravitational wave detectors .
this work underlines the feasibility of expanding the distances bridged by the optical fiber links .
thus , the vision of a national @xcite and european wide fiber network becomes a realistic scenario in the near future , which will strongly stimulate the field of fundamental physics and precison frequency metrology .
we gratefully acknowledge long - standing advice and support by w. ertmer . we thank t. rosenband ( nist , boulder ) for providing us the design for our new ultrastable resonators .
we thank u. sterr and s. kraft for providing the ultrastable calcium clock laser and thank s. falke for proof - reading .
we acknowledge financial support by the dlr and esa .
o. terra is supported by a scholarship from egypt and is a member of the braunschweig international graduate school of metrology , igsm .
this joint work was made possible by the framework of the sonderforschungsbereich 407 and the centre for quantum engineering and space time research ( quest ) , with financial support from the deutsche forschungsgemeinschaft . | we demonstrate a fully optical , long - distance remote comparison of independent ultrastable optical frequencies reaching a short term stability that is superior to any reported remote comparison of optical frequencies .
we use two ultrastable lasers , which are separated by a geographical distance of more than 50 km , and compare them via a 73 km long phase - stabilized fiber in a commercial telecommunication network . the remote characterization spans more than one optical octave and reaches a fractional frequency instability between the independent ultrastable laser systems of @xmath0 in 0.1 s. the achieved performance at 100 ms represents an improvement by one order of magnitude to any previously reported remote comparison of optical frequencies and enables future remote dissemination of the stability of 100 mhz linewidth lasers within seconds .
10 t. rosenband , d. b. hume , p. o. schmidt , c. w. chou , a. brusch , l. lorini , w. h. oskay , r. e. drullinger , t. m. fortier , j. e. stalnaker , s. a. diddams , w. c. swann , n. r. newbury , w. m. itano , d. j. wineland , and j. c. bergquist , `` frequency ratio of al+ and hg+ single - ion optical clocks ; metrology at the 17th decimal place , '' science * 319 * , 1808 ( 2008 ) .
a. ludlow , t. zelevinsky , g. campbell , s. blatt , m. boyd , m. de miranda , m. martin , j. thomsen , s. foreman , j. ye , t. fortier , j. stalnaker , s. diddams , y. le coq , z. barber , n. poli , n. lemke , k. beck , and c. oates , `` sr lattice clock at @xmath1 fractional uncertainty by remote optical evaluation with a ca clock , '' science * 319 * , 1805 ( 2008 ) .
s. a. webster , m. oxborrow , s. pugla , j. millo , and p. gill , `` thermal - noise - limited optical cavity , '' phys .
rev .
a * 77 * , 033847 ( 2008 ) .
j. millo , d. v. magalhes , c. mandache , y. le coq , e. m. l. english , p. g. westergaard , j. lodewyck , s. bize , p. lemonde , and g. santarelli , `` ultrastable lasers based on vibration insensitive cavities , '' phys .
rev .
a * 79 * , 053829 ( 2009 ) .
a. amy - klein , a. goncharov , m. guinet , c. daussy , o. lopez , a. shelkovnikov , and c. chardonnet , `` absolute frequency measurement of a sf6 two - photon line by use of a femtosecond optical comb and sum - frequency generation , '' opt .
lett . * 30 * , 33203322 ( 2005 ) .
l .- s .
ma , p. jungner , j. ye , and j. l. hall , `` delivering the same optical frequency at two places : accurate cancellation of phase noise introduced by an optical fiber or other time - varying path , '' opt . lett . *
19 * , 17771779 ( 1994 ) .
j. ye , j .-
l .
peng , r. j. jones , k. w. holman , j. l. hall , d. j. jones , s. a. diddams , j. kitching , s. bize , j. c. bergquist , l. w. hollberg , l. robertsson , and l .- s .
ma , `` delivery of high - stability optical and microwave frequency standards over an optical fiber network , '' j. opt .
soc .
am .
b * 20 * , 14591467 ( 2003 ) .
g. grosche , b. lipphardt , h. schnatz , g. santarelli , p. lemonde , s. bize , m. lours , f. narbonneau , a. clairon , o. lopez , a. amy - klein , and c. chardonnet , `` transmission of an optical carrier frequency over a telecommunication fiber link , '' in `` conference on lasers and electro - optics / quantum electronics and laser science conference and photonic applications systems technologies , '' ( optical society of america , 2007 ) , p. cmkk1 .
s. m. foreman , k. w. holman , d. d. hudson , d. j. jones , j. ye , `` remote transfer of ultrastable frequency references via fiber networks , '' rev .
sci .
instrum . * 78 * , 021101 ( 2007 ) .
s. m. foreman , a. d. ludlow , m. h. g. de miranda , j. e. stalnaker , s. a. diddams , and j. ye , `` coherent optical phase transfer over a 32-km fiber with 1 s instability at @xmath2 , '' phys .
rev .
lett . * 99 * , 153601 ( 2007 ) .
n. r. newbury , p. a. williams , and w. c. swann , `` coherent transfer of an optical carrier over 251 km , '' opt
. lett . * 32 * , 30563058 ( 2007 ) .
h. jiang , f. kflian , s. crane , o. lopez , m. lours , j. millo , d. holleville , p. lemonde , c. chardonnet , a. amy - klein , and g. santarelli , `` long - distance frequency transfer over an urban fiber link using optical phase stabilization , '' j. opt .
soc . am .
b * 25 * , 20292035 ( 2008 ) .
g. grosche , o. terra , k. predehl , r. holzwarth , b. lipphardt , f. vogt , u. sterr , and h. schnatz , `` optical frequency transfer via 146 km fiber link with @xmath3 relative accuracy , '' opt .
lett . * 34 * , 22702272 ( 2009 ) .
f. kflian , o. lopez , h. jiang , c. chardonnet , a. amy - klein , and g. santarelli , `` high - resolution optical frequency dissemination on a telecommunications network with data traffic , '' opt .
lett . * 34 * , 15731575 ( 2009 ) .
f .- l .
hong , m. musha , m. takamoto , h. inaba , s. yanagimachi , a. takamizawa , k. watabe , t. ikegami , m. imae , y. fujii , m. amemiya , k. nakagawa , k. ueda , and h. katori , `` measuring the frequency of a sr optical lattice clock using a 120 km coherent optical transfer , '' opt .
lett .
* 34 * , 692694 ( 2009 ) .
o. terra , g. grosche , k. predehl , r. holzwarth , t. legero , u. sterr , b. lipphardt , and h. schnatz , `` phase - coherent comparison of two optical frequency standards over 146 km using a telecommunication fiber link , '' appl .
phys .
b : lasers and optics * 97 * , 541551 ( 2009 ) .
s. a. webster , m. oxborrow , and p. gill , `` vibration insensitive optical cavity , '' phys .
rev .
a * 75 * , 011801 ( 2007 ) .
h. stoehr , f. mensing , j. helmcke , and u. sterr , `` diode laser with 1 hz linewidth , '' opt .
lett .
* 31 * , 736738 ( 2006 ) .
t. nazarova , f. riehle , and u. sterr , `` vibration - insensitive reference cavity for an ultra - narrow - linewidth laser , '' appl .
phys .
b : lasers and optics * 83 * , 531536 ( 2006 ) .
b. lipphardt , g. grosche , u. sterr , c. tamm , s. weyers , and h. schnatz , `` the stability of an optical clock laser transferred to the interrogation oscillator for a cs fountain , '' ieee trans .
. meas . *
58 * , 12581262 ( 2009 ) .
g. grosche , b. lipphardt , and h. schnatz , `` optical frequency synthesis and measurement using fibre - based femtosecond lasers , '' the european physical journal d * 48 * , 2733 ( 2008 ) .
s. dawkins , j. mcferran , and a. luiten , `` considerations on the measurement of the stability of oscillators with frequency counters , '' ieee tr .
uffc * 54 * , 918925 ( 2007 ) .
p.
a. williams , w. c. swann , and n. r. newbury , `` high - stability transfer of an optical frequency over long fiber - optic links , '' j. opt .
soc .
am .
b * 25 * , 12841293 ( 2008 ) .
k. numata , a. kemery , and j. camp , `` thermal - noise limit in the frequency stabilization of lasers with rigid cavities , '' phys
. rev .
lett . *
93 * , 250602 ( 2004 ) . |
for a smooth algebraic curve @xmath11 of genus @xmath1 we denote by @xmath12 the space of two - torsion points in the jacobian of @xmath11 , viewed as an @xmath6-vector space .
we recall the definition of the _ weil pairing _ @xmath13 , cf .
@xcite lemma 2 : let @xmath14 and write @xmath15 and @xmath16 , for certain divisors @xmath17 and @xmath18 on @xmath11 , such that @xmath19 . pick rational functions @xmath20 and @xmath1 on @xmath11 , such that @xmath21 and @xmath22 .
then define @xmath23 by the formula @xmath24 the definition of @xmath25 is independent of the choice of the divisors @xmath17 and @xmath18 and the rational functions @xmath20 and @xmath1 .
the weil pairing is a nondegenerate symplectic form .
the set of _ theta characteristics _ of @xmath11 , defined as @xmath26 , is an affine space over @xmath27 .
it was coble s insight @xcite to realize that in order to acquire an abstract understanding of the geometry of @xmath28 and clarify the distinction between the period characteristics and the theta characteristics of @xmath11 , it is advantageous to view theta characteristics as quadrics in the vector space @xmath27
. let @xmath29 be a symplectic vector space over @xmath6 .
the set @xmath30 of quadratic forms on @xmath31 with fixed polarity given by the symplectic form @xmath32 , consists of all functions @xmath33 satisfying the identity @xmath34 if @xmath35 is a quadratic form and @xmath36 , we can define a new quadratic form @xmath37 by setting @xmath38 , for all @xmath39 similarly one can add two quadratic forms .
if @xmath40 , then there exists a uniquely determined element @xmath36 such that @xmath41 , and we set @xmath42 . in this way , the set @xmath43 becomes a @xmath44-dimensional vector space over @xmath6 .
there is a natural action of the symplectic group @xmath45 on @xmath30 . for a transformation @xmath46 and a quadratic form @xmath35
, one defines @xmath47 this action has two orbits @xmath48 and @xmath49 respectively , which can de distinguished by the _ arf invariant _
@xcite of a quadratic form .
let us choose a symplectic base @xmath50 of @xmath31 .
then define @xmath51 the invariant @xmath52 is independent of the choice of a symplectic basis of @xmath31 .
we set @xmath53 to be the space of even quadratic form , and @xmath54 that of odd quadratic forms .
note that there are @xmath55 even quadratic forms and @xmath56 odd ones .
every theta characteristic @xmath57 defines a form @xmath58 , by setting @xmath59 it follows from the _ riemann - mumford relation _
@xcite or @xcite theorem 1.13 , that @xmath60 , that is , the polar of @xmath61 is the weil form . for @xmath14 ,
the following relation holds : @xmath62 thus one has the following identification between theta characteristics and quadrics : under this isomorphism , even ( respectively odd ) theta characteristics correspond to forms in @xmath48 ( respectively @xmath49 ) .
furthermore , @xmath63 . using this identification
, one can translate frobenius s @xcite , @xcite entire _ theory of fundamental systems _ of theta characteristics into an abstract setting . of great importance
is the following : a system of three theta characteristics @xmath64 is called _ syzygetic _
( respectively _ azygetic _ ) if @xmath65 ( respectively @xmath66 ) .
here the additive notation @xmath67 refers to addition in the extended vector space @xmath68 . in terms of line bundles , @xmath69 .
it is an easy exercise to show that @xmath70 is a syzygetic triple if and only if @xmath71 .
if the system @xmath70 is syzygetic , then any three elements of the set @xmath72 form a syzygetic triple .
in this case we say that the four theta characteristics form a _
syzygetic tetrad_. four odd theta characteristics @xmath73 corresponding to contact contact divisors @xmath74 such that @xmath75 form a syzygetic tetrad , if and only if there exists a quadric @xmath76 , such that @xmath77 for instance , when @xmath78 , four bitangents to a quartic @xmath79 form a syzygetic tetrad exactly when the @xmath80 points of tangency are the complete intersection of @xmath11 with a conic . to understand the geometry of the configuration @xmath28
one defines appropriate systems of coordinates .
following @xcite p. 283
, one says that @xmath81 theta characteristics @xmath82 form a _ fundamental system _ if any triple @xmath83 where @xmath84 is azygetic .
the sum of the @xmath81 elements of a fundamental system equals @xmath85 , see @xcite p. 284 , and the number of fundamental systems in @xmath86 is known , cf .
@xcite p. 285
: @xmath87 a smooth plane quartic @xmath79 has precisely @xmath88 fundamental systems @xmath89 , where the first @xmath90 theta characteristics are odd , and @xmath91 is then necessarily even .
all elements in @xmath28 can be expressed in the `` coordinate system '' given by @xmath92 : the remaining odd theta characteristics are @xmath93 .
the @xmath94 even theta characteristics are of the form @xmath95 .
one can also consider systems of syzygetic theta characteristics , that is subsets , @xmath96 such that all triples @xmath97 are syzygetic .
then @xmath98 , see @xcite p. 299 . following frobenius @xcite , a maximal system of such theta characteristics
is called a _ gpel system _ and corresponds to @xmath99 theta characteristics such any three of them form a syzygetic triple .
these definitions can be immediately extended to cover general principally polarized abelian varieties not only jacobians .
we now link the realization of theta characteristics in abstract finite geometry , to the theory of theta functions with characteristics . there are established classical references , above all @xcite , @xcite , @xcite , @xcite , as well as modern ones , for instance @xcite , @xcite .
we fix an integer @xmath100 and denote by @xmath101 the siegel upper half - space of period matrices for abelian varieties of dimension @xmath1 ; hence @xmath102 is the moduli space of principally polarized abelian varieties of dimension @xmath1 . for a vector @xmath103=\left[\begin{array}{ccc } \epsilon_1 \ldots \epsilon_g\\ \delta_1 \ldots \delta_g\\ \end{array}\right]\in \ff_2^{2g}$ ] one defines the _ riemann theta function with characteristics _ as the holomorphic function @xmath104 given by @xmath105(\tau , z):=\sum_{m\in \mathbb z^g } \mathrm{exp } \bigl(\pi i \ ^t(m+\frac{\epsilon}{2})\tau ( m+\frac{\epsilon}{2})+2\pi i \ ^t(m+\frac{\epsilon}{2})(z+\frac{\delta}{2})\bigr).\ ] ] for any period matrix @xmath106 , the pair @xmath107(\tau , z)=0\bigr\}\bigr]\ ] ] defines a principally polarized abelian variety , that is , @xmath108\in \ca_g$ ] .
there is an identification of symplectic vector spaces @xmath109,\ \mbox { } \mbox { given by } \
\mbox { } \mbox { } \left[\begin{array}{c } \epsilon\\ \delta\\ \end{array}\right ] \mapsto \frac{\tau\cdot \epsilon+\delta}{2}\in a_{\tau}[2].\ ] ] this isomorphism being understood , points in @xmath110 $ ] were classically called _ period characteristics _ , see @xcite section vii.2 .
the theta function @xmath111(\tau , z)$ ] is the unique section of the translated line bundle @xmath112 .
krazer @xcite goes to great lengths to emphasize the difference between the _ period _ and the _ theta _ characteristics , even though at first sight , both sets of characteristics can be identified with the vector space @xmath113 .
the reason for this is the transformation formula for theta functions under the action of the symplectic group , see @xcite theorem 8.6.1 .
the theta constants with characteristics are modular forms of weight one half with respect to the subgroup @xmath114 .
to define the action of the full symplectic group @xmath115 on @xmath113 , we consider the quadratic form associated to a characteristic @xmath116 $ ] , that is , @xmath117}\bigl(x , y\bigr)=x\cdot y+\epsilon\cdot x+ \delta\cdot y,\ ] ] with @xmath118 .
we define an action of @xmath115 on the set of characteristics which factors through the following action of @xmath119 : if @xmath120 , set @xmath121 where we recall that we have already defined @xmath122 . it can be shown that this non - linear action of the symplectic group on the set of characteristics is compatible with the transformation rule of theta constants , see also @xcite , @xcite . in this context ,
once more , a theta characteristic @xmath123 $ ] appears as an element of @xmath30 .
this interpretation makes the link between the classical definition of theta characteristics found in @xcite and the more modern one encountered for instance in @xcite , @xcite .
the algebra of period and theta characteristics was developed by frobenius @xcite , @xcite in order to derive general identities between theta constants and describe the structure of the set of such identities .
recently , the action of @xmath5 on the set of characteristics and the algebra of characteristic systems has been used by grushevsky @xcite , salvati manni @xcite , cacciatori , dalla piazza and van geemen @xcite and others , in order to find an explicit formula for the _ chiral superstring amplitudes_. this is an important foundational question in string theory and we refer to the cited papers for background and further references . loosely speaking
, dhoker and phong conjectured that there exists a modular form @xmath124 of weight @xmath80 with respect to the group @xmath125 , satisfying the following two constraints : \(1 ) _ factorization : _ for each integer @xmath126 , the following factorization formula @xmath127 holds , when passing to the locus @xmath128 of decomposable abelian varieties .
\(2 ) _ initial conditions : _ for @xmath129 , one must recover the standard chiral measure , that is , @xmath130 ^ 8 \theta \left[\begin{array}{c } 0\\ 1\\ \end{array}\right]^4 \theta \left[\begin{array}{c } 1\\ 0\\ \end{array}\right]^4.\ ] ] a unique solution has been found in @xcite when @xmath78 and in @xcite when @xmath131 .
this was placed in @xcite in a general framework that works in principle for arbitrary @xmath1 . for a set of theta characteristics @xmath132 ,
one defines the product of theta constants @xmath133(\tau , 0).\ ] ] note that @xmath134 vanishes if @xmath135 contains an odd characteristic .
then for all integers @xmath136 , we set @xmath137 it is pointed out in @xcite proposition 13 , that this function is a modular form of weight @xmath80 with respect to @xmath138 .
this result can be traced back to frobenius . observe that in the definition of @xmath139 the only non - zero summands correspond to totally syzygetic systems of characteristics , for else such a system @xmath135 contains an odd characteristic and the corresponding term @xmath140 is identically zero .
then it is showed that the expression @xmath141 satisfies the factorization rules when @xmath142 .
note that for higher @xmath1 the definition of @xmath124 leads to a multivalued function due to the impossibility of choosing consistently the roots of unity for the various summands .
can one find the superstring amplitudes for higher @xmath1 by working directly with the space @xmath7 and constructing via algebro - geometric rather than theta function methods a system of effective divisors of slope ( weight ) @xmath80 satisfying the factorization formula ?
we now concern ourselves with describing cornalba s @xcite compactification @xmath9 of the moduli space @xmath4 of theta characteristics .
we recall that @xmath4 is the parameter space of pairs @xmath143 $ ] , where @xmath11 is a smooth curve of genus @xmath1 and @xmath57 .
following atiyah @xcite such a pair is called a _ spin curve _ of genus @xmath1 .
mumford @xcite by algebraic means ( and atiyah @xcite with analytic methods ) showed that the parity of a spin curve is locally constant in families : if @xmath144 is a flat family of smooth curves of genus @xmath1 and @xmath145 is a line bundle on @xmath146 together with a morphism @xmath147 such that @xmath148 is an isomorphism for all @xmath149 , then the map @xmath150 defined as @xmath151 is constant on connected components of @xmath152 .
one can speak of the parity of a family of spin curves and according to the value of @xmath153 , the moduli space @xmath4 splits into two connected components @xmath154 and @xmath155 .
the forgetful map @xmath156 , viewed as a morphism of deligne - mumford stacks , is unramified
. a compactified moduli space @xmath9 should be the coarse moduli space of a deligne - mumford stack , such that there is a _
morphism @xmath157 fitting into a commutative diagram : @xmath158_{\pi } \ar@{^{(}->}@<-0.5ex>[r ] & \ss_g \ar[d]_{\pi } & \\ & \cm_g \ar@{^{(}->}@<-0.5ex>[r ] & \mm_g & \\ } \ ] ] as an algebraic variety , @xmath9 is the normalization of @xmath159 in the function field of @xmath4 .
it is the main result of @xcite that the points of @xmath9 have a precise modular meaning in terms of line bundles on curves belonging to a slightly larger class than that of stable curves . a reduced , connected , nodal curve @xmath146
is called _ quasi - stable _ , if for any component @xmath160 that is isomorphic to @xmath161 , one has that ( i ) @xmath162 , and ( ii ) any two rational components @xmath163 with @xmath164 , are disjoint .
smooth rational components @xmath160 for which @xmath165 are called _
exceptional_. the class of quasi - stable curve is a very slight enlargement of the class of stable curves . to obtain a quasi - stable curve ,
one takes a stable curve @xmath166\in \mm_g$ ] and a subset of nodes @xmath167 which one `` blows - up '' ; if @xmath168 denotes the normalization map and @xmath169 , then we define the nodal curve @xmath170 where @xmath171 for each node @xmath172 and @xmath173 .
the _ stabilization map _
@xmath174 is a partial normalization and contracts all exceptional components @xmath175 , that is , @xmath176 , for each @xmath172 . why extend the class of stable curves , after all @xmath159 is already a projective variety ?
one can define compactified moduli spaces of theta characteristics working with stable curves alone , if one is prepared to allow sheaves that are not locally free at the nodes of the curves .
allowing semi - stable curves , enables us to view a degeneration of a theta characteristic as a line bundle on a curve that is ( possibly ) more singular .
these two philosophies of compactifying a parameter space of line bundles , namely restricting the class of curves but allowing singularities for the sheaves , vs. insisting on local freeness of the sheaves but enlarging the class of curves , can also be seen at work in the two ( isomorphic ) compactifications of the universal degree @xmath177 jacobian variety @xmath178 over @xmath179 constructed in @xcite and @xcite .
we now describe all points of @xmath9 : [ spinstructures ] a _ stable spin curve _ of genus @xmath1 consists of a triple @xmath180 , where @xmath146 is a quasi - stable curve of arithmetic genus @xmath1 , @xmath181 is a line bundle of total degree @xmath182 such that @xmath183 for every exceptional component @xmath160 , and @xmath184 is a sheaf homomorphism which is not zero along each non - exceptional component of @xmath146 . when @xmath146 is a smooth curve , then @xmath185 is an ordinary theta characteristics and @xmath186 is an isomorphism .
note that in this definition , the morphism @xmath187 vanishes with order @xmath2 along each exceptional component @xmath160 .
cornalba @xcite proved that stable spin curves form a projective moduli space @xmath9 endowed with a regular stabilization morphism @xmath157 , set - theoretically given by @xmath188):=[c]$ ] ; here @xmath11 is obtained from @xmath146 by contracting all the exceptional components .
the space @xmath9 has two connected components @xmath8 and @xmath7 depending on the parity @xmath189 of the spin structure .
[ thc ] for a stable curve @xmath166\in \mm_g$ ] , we denote by @xmath190)$ ] the zero - dimensional scheme of length @xmath191 classifying stable spin structures on quasi - stable curves whose stable model is @xmath11 .
the scheme @xmath28 has an interesting combinatorial structure that involves the dual graph of @xmath11 . before describing it for an arbitrary curve @xmath166\in \mm_g$ ]
, it is helpful to understand the boundary structure of @xmath9 , which amounts to describe @xmath28 when @xmath11 is a @xmath66-nodal curve .
we shall concentrate on the space @xmath8 and leave @xmath7 as an exercise ( or refer to @xcite , @xcite for details ) .
the boundary @xmath192 of the moduli space of curves decomposes into components @xmath193}$ ] .
a general point of @xmath194 corresponds to a @xmath66-nodal irreducible curve of arithmetic genus @xmath1 , and for @xmath195 $ ] the general point of @xmath196 is the class of the union of two components of genera @xmath197 and @xmath198 respectively , meeting transversely at a point .
we fix an integer @xmath195 $ ] and a general point @xmath199\in \delta_i$ ] , where @xmath200\in \cm_{i , 1}$ ] and @xmath201\in \cm_{g - i , 1}$ ] are smooth curves .
we describe all stable spin curve @xmath202\in \pi^{-1}([c\cup_y d])$ ] . for degree reasons , @xmath203 , that is , one must insert an exceptional component @xmath18 at the node @xmath204 and then @xmath205 where @xmath206 and @xmath207 .
moreover @xmath208 and since @xmath209 , it follows that @xmath210 and @xmath211 , that is , a theta characteristic on a curve of compact type is simply a collection of theta characteristics on each of its ( necessarily smooth ) components .
the condition that @xmath212 be odd implies that @xmath213 and @xmath214 have opposite parities .
accordingly , the pull - back divisor @xmath215 splits in two components depending on the choice of the respective arf invariants : we denote by @xmath216 the closure of the locus corresponding points for which @xmath217 and @xmath218 , that is , @xmath219 and @xmath220 .
we denote by @xmath221 the closure of the locus of spin curves for which @xmath222 and @xmath223 . at the level of divisors
, the following relation holds @xmath224 moreover one has that @xmath225 we fix a general @xmath2-pointed smooth curve @xmath226\in \cm_{g-1 , 2}$ ] and identify the points @xmath227 and @xmath228 .
the resulting stable curve @xmath229 , where @xmath230 , corresponds to a general point of the boundary divisor @xmath194 . unlike in the case of curves of compact type ,
two possibilities do occur , depending on whether @xmath146 possesses an exceptional component or not .
if @xmath231 , then the locally free sheaf @xmath232 is a root of the dualizing sheaf @xmath233 . setting @xmath234 , from the condition @xmath235 by counting degrees , we obtain that @xmath236
for each choice of @xmath237 as above , there is precisely one choice of gluing the fibres @xmath238 and @xmath239 in a way that if @xmath232 denotes the line bundle on @xmath240 corresponding to this gluing , then @xmath212 is odd .
let @xmath241 denote the closure in @xmath8 of the locus of such points .
then @xmath242 .
since we expect the fibre @xmath243 to consist of @xmath191 points ( counting also multiplicities ) , we see that one can not recover all stable spin curves having @xmath240 as their stable model by considering square roots of the dualizing sheaf of @xmath240 alone .
the remaining spin curves correspond to sheaves on @xmath240 which are not locally free at the node , or equivalently , to spin structures on a strictly quasi - stable curve .
assume now that @xmath244 , where @xmath18 is an exceptional component .
since @xmath209 it follows that @xmath245 must vanish at both @xmath227 and @xmath228 and then for degree reasons @xmath210 . for parity reasons , @xmath219 .
we denote by @xmath246 the closure in @xmath8 of the locus of such points .
a local analysis carried out in @xcite shows that @xmath247 is simply ramified over @xmath246 .
since @xmath248 is not ramified along any other divisors of @xmath8 , one deduces that @xmath246 is the ramification divisor of the forgetful map and the following relation holds : @xmath249 a general point of @xmath246 is determined by specifying an odd theta characteristic on @xmath11 , thus @xmath250 . by direct calculation one
checks that @xmath251 which confirms that @xmath247 is simply ramified along @xmath246 . after this preparation , we are now ready to tackle the case of an arbitrary stable curve @xmath11 .
we denote by @xmath168 the normalization map and by @xmath252 the _ dual graph _ whose vertices are in correspondence with components of @xmath11 , whereas an edge of @xmath252 corresponds to a node which lies at the intersection of two components ( note that self - intersections are allowed ) .
a set of nodes @xmath253 is said to be _ even _ , if for any component @xmath254 , the degree @xmath255 is an even number .
for instance , if @xmath256 is a union of two smooth curves meeting transversally , a set of nodes @xmath257 is even if and only if @xmath258 .
assume @xmath202\in \ss_g$ ] and let @xmath259 be the stabilization morphism . if @xmath167 is the set of exceptional nodes of @xmath11 , that is , nodes @xmath172 having the property that @xmath260 , and we write that @xmath261 , then the set @xmath262 of non - exceptional nodes of @xmath11 is even , see both @xcite and @xcite .
one has the following description of the scheme @xmath28 , cf .
@xcite proposition 5 : [ components ] let @xmath166\in \mm_g$ ] and @xmath263 be the betti number of the dual graph .
then the number of components of the zero - dimensional scheme @xmath264 is equal to @xmath265 a component corresponding to an even set @xmath266 appears with multiplicity @xmath267 .
one can easily verify that the length of the scheme @xmath28 is indeed @xmath191 . from proposition [ components ] it follows for instance that @xmath28 is a reduced scheme if and only if @xmath11 is of compact type . in the case we studied above , when @xmath11 is irreducible with a single node , proposition [ components ] gives that @xmath268 has @xmath269 irreducible components .
precisely , @xmath270 and @xmath271 .
the structure of the schemes @xmath28 for special singular curves @xmath11 has been used in @xcite to describe the singularities of @xmath9 and in @xcite to prove that a general curve @xmath166\in \cm_g$ ] is uniquely determined by the set of contact hyperplanes @xmath272 in spite of these important applications , a systematic study of the finite geometry of the set @xmath28 when @xmath11 is singular ( e.g. a theory of fundamental and gpel systems , syzygetic tetrads ) has not yet been carried out and is of course quite interesting .
it is customary to denote the divisor classes in the picard group of the moduli stack by a result of putman s @xcite shows that for @xmath273 , the divisor classes @xmath274 , @xmath275 , @xmath276}$ ] , @xmath277}$ ] freely generate the rational picard group @xmath278 .
a similar result holds for @xmath7 .
the space @xmath8 is a normal variety with finite quotient singularities ; an tale neighbourhood of an arbitrary point @xmath279\in \ss_g$ ] is of the form @xmath280 where @xmath281 can be identified via deformation theory with the cotangent space to the moduli stack @xmath159 at the point @xmath166:=\pi([x , \eta , \beta])$ ] . for a ( predictable ) definition of an automorphism of a triple @xmath282
, we refer to @xcite section 1 . using kodaira - spencer deformation theory
, one describes the cotangent bundle of the stack @xmath159 as the push - forward of a rank @xmath66 sheaf on the universal curve over @xmath159 .
a famous and at the time , very innovative use in @xcite of the grothendieck - riemann - roch theorem for the universal curve , yields the formula @xmath283 } \delta_i\in \mbox{pic}(\mm_g).\ ] ] from the riemann - hurwitz theorem applied to the finite branched cover @xmath248 , we find the formula for the canonical class of the spin moduli stack : an identical formula holds for @xmath7 .
unfortunately both spaces @xmath7 and @xmath8 have non - canonical singularities , in particular there exist _ local _ obstructions to extending pluri - canonical forms defined on the smooth part of @xmath9 to a resolution of singularities .
however , an important result of ludwig @xcite shows that this obstructions are not of global nature .
the following result holds for both @xmath7 and @xmath8 : ( ludwig ) for @xmath284 fix a resolution of singularities @xmath285 .
then for any integer @xmath286 there exists an isomorphism of vector spaces @xmath287 therefore , in order to conclude that @xmath9 is of general type , it suffices to show that the canonical class @xmath288 lies in the interior of the effective cone @xmath289 of divisors , or equivalently , that it can be expressed as a positive linear combination of an ample and an effective class on @xmath9 .
this becomes a question on slopes of the effective cone of @xmath9 , which can be solved in a spirit similar to @xcite and @xcite , where it has been proved with similar methods that @xmath159 is of general type for @xmath290 .
the standard references for effective divisors on moduli spaces of stable curves are @xcite , @xcite , @xcite and @xcite .
in the papers @xcite , @xcite and @xcite we have initiated a study of effective divisors on @xmath9 .
the class of the locus @xmath291 of vanishing theta - nulls on @xmath7 is computed in @xcite ; in the paper @xcite we study the space @xmath8 with the help of the divisor of spin curves with an everywhere tangent hyperplane in the canonical embedding having higher order contact than expected . finally in @xcite
we define effective divisors of brill - noether type in more general setting on both spaces @xmath8 and @xmath7 .
we survey these constructions , while referring to the respective papers for technical details .
we begin with the moduli space @xmath8 : an odd theta characteristic @xmath57 with @xmath292 determines a unique effective divisor @xmath293 such that @xmath294 .
we write in this case that @xmath295 .
the assignment @xmath296 can be viewed as a rational map between moduli spaces @xmath297 and it is natural to use this map and the well - understood divisor theory @xcite on the universal symmetric product @xmath298 , in order to obtain promising effective divisors on @xmath8 . in particular , the boundary divisor @xmath299 on @xmath298 with general point being a pair @xmath300 where @xmath293 is a divisor with non - reduced support , is known to be extremal .
its pull - back to @xmath8 parametrizes ( limits of ) odd spin curves @xmath143\in \cs_g^-$ ] such that there exists a point @xmath301 with @xmath302 .
the calculation of the class of the closure of this locus is one of the main results of @xcite : [ degen ] we fix @xmath303 .
the locus consisting of odd spin curves @xmath304\in \cs_g^-:\theta=\oo_c\bigl(2x_1+\sum_{i=2}^{g-2 } x_i\bigr),\ \mbox { with } x_i\in c \mbox { for } i=1 , \ldots , g-2 \bigr\}\ ] ] is a divisor on @xmath155 .
the class of its compactification inside @xmath8 equals @xmath305= ( g+8)\lambda-\frac{g+2}{4}\alpha_0 - 2\beta_0-\sum_{i=1}^{[\frac{g}{2 } ] } 2(g - i)\ \alpha_i-\sum_{i=1}^{[\frac{g}{2 } ] } 2i\ \beta_i\in \mathrm{pic}(\ss_g^-).\ ] ] for low genus , @xmath306 specializes to well - known geometric loci .
for instance @xmath307 is the divisor of hyperflexes on plane quartics , classifying pairs @xmath308\in
\cs_3 ^ -$ ] , where @xmath309 is such that @xmath310 .
then @xmath311 and @xmath309 is a hyperflex point .
the divisor @xmath312 together with pull - backs of effective divisors on @xmath159 can be used to determine the range in which @xmath8 is of general type .
this application also comes from @xcite : [ sg- ] the moduli space @xmath8 is a variety of general type for @xmath313 .
the passing from theorem [ degen ] to theorem [ sg- ] amounts to simple linear algebra . by comparing the class
@xmath314 $ ] against that of the canonical divisor , we note that @xmath315 , \lambda , \alpha_i , \beta_i , \
i=0 , \ldots , \bigl[\frac{g}{2}\bigr]\bigr\rangle,\ ] ] for the coefficient of @xmath316 in the expression of @xmath314 $ ] is too small .
however one can combine @xmath314 $ ] with effective classes coming from @xmath159 , and there is an ample supply of such , see @xcite , @xcite , @xcite . to avoid technicalities , let us assume that @xmath317 is composite and consider the brill - noether divisor @xmath318 of curves @xmath166\in \cm_g$ ] with a @xmath319 , where the brill - noether number @xmath320 .
the class of the closure @xmath321 of @xmath318 in @xmath159 has been computed in @xcite ( and in @xcite for @xmath322 ) and plays a crucial role in the proof by harris , mumford , eisenbud that the moduli space @xmath159 is of general type for @xmath290 .
there exists an explicit constant @xmath323 such that the following relation holds @xcite , @xmath324= c_{g , d , r}\bigl((g+3)\lambda-\frac{g+1}{6}\delta_0-\sum_{i=1}^{[\frac{g}{2 } ] } i(g - i)\delta_i\bigr)\in \mathrm{pic}(\mm_g).\ ] ] by interpolation , one find a constant @xmath325 such that the effective linear combination @xmath326+c'_{g , d , r}[\pi^*(\mm_{g , d}^r)]=\frac{11g+37}{g+1}\lambda-2\alpha_0 - 3\beta_0-\sum_{i=1}^{[g/2 ] } ( a_i\cdot \alpha_i+ b_i),\ ] ] where @xmath327 for @xmath328 and @xmath329 are explicitly known rational constants . by comparison , whenever the inequality @xmath330 is satisfied ( and @xmath317 is composite ) , the class @xmath331 is big , that is , @xmath8 is of general type .
the case @xmath332 is rather difficult and we refer to the last section of @xcite .
on @xmath7 we consider the locus of even spin curves with a vanishing theta characteristic .
the following comes from @xcite : [ thetanull ] the closure in @xmath333 of the divisor @xmath334\in \cs_g^{+ } : h^0(c , \eta)\neq 0\bigr\}\ ] ] of curves with an effective even theta characteristics has class equal to @xmath335= \frac{1}{4}\lambda-\frac{1}{16}\alpha_0-\frac{1}{2}\sum_{i=1}^{[\frac{g}{2 } ] } \beta_i\in \ \mathrm{pic}(\ss_g^{+}).\ ] ] in the paper @xcite it is shown that the class @xmath336\in \mbox{eff}(\ss_g^+)$ ] is extremal when @xmath337 .
it is an open question whether this is the case for arbitrary @xmath1 , certainly there is no known counterexample to this possibility . combining @xmath336 $ ] with pull - backs of effective classes from @xmath159 like in the previous case , we find a constant @xmath338 such that @xmath339+c^{''}_{g , d , r}[\pi^*(\mm_{g , d}^r)]=\frac{11g+29}{g+1}\lambda-2\alpha_0 - 3\beta_0-\sum_{i=1}^{[\frac{g}{2 } ] } ( a_i'\cdot \alpha_i+b_i'\cdot \beta_i)\in \mathrm{eff}(\ss_g^+),\ ] ] where the appearing coefficients satisfy the inequalities @xmath340 for @xmath341 and @xmath342 .
restricting ourselves again to the case when @xmath317 is composite ( while referring to @xcite for the remaining cases ) , we obtain that whenever @xmath343 the space @xmath7 has maximal kodaira dimension .
we summarize these facts as follows : [ sg+ ] the moduli space @xmath7 is a variety of general type for @xmath344 .
next we present ways of proving the unirationality of @xmath4 in small genus .
the basic references are the papers @xcite and @xcite .
we recall that a normal @xmath345-factorial projective variety @xmath146 is said to be _ uniruled _ if through a very general point @xmath346 there passes a rational curve @xmath347 .
uniruled varieties have negative kodaira dimension .
conversely , if the canonical class @xmath348 is not a limit of effective divisor classes ( which implies that the kodaira dimension of @xmath146 is negative ) , then @xmath146 is uniruled , see @xcite .
the classification by kodaira dimension of both @xmath8 and @xmath7 is governed by @xmath10 surfaces , in the sense that @xmath4 is uniruled precisely when a general spin curve @xmath143\in \cs_g$ ] can be represented as a section of a special @xmath10 surface @xmath152 . by varying @xmath11 in a pencil on @xmath152 , we induce a rational curve in the moduli space @xmath9 passing through a general point
. the @xmath10 surface must have special properties that will allow us to assign a theta characteristic to each curve in the pencil . in the case of even spin curves ,
the @xmath10 surface in question must be of nikulin type .
we refer to @xcite for an introduction to nikulin surfaces .
a polarized _ nikulin surface _ of genus @xmath349 consists of a triple @xmath350 , where @xmath152 is a smooth @xmath10 surface , @xmath351 is a non - trivial line bundle with the property that @xmath352 , where @xmath353 are pairwise disjoint @xmath354 curves on @xmath152 , and @xmath355 is a numerically effective curve class such that @xmath356 and @xmath357 , for @xmath358 .
since the line bundle @xmath359 is divisible by two , there exists a double cover @xmath360 branched exactly along the smooth rational curves @xmath353 .
the curve @xmath355 does not meet the branch locus of @xmath20 , hence the restriction @xmath361 is an unramified double covering which induces a non - trivial half - period on @xmath11 .
each curve in the linear system @xmath362 acquires a half - period in its jacobian in this way .
the following result is quoted from @xcite : [ eventheta ] the even spin moduli space @xmath363 is uniruled for @xmath364 .
let us choose a general spin curve @xmath143\in \cs_g^+$ ] and a non - trivial point of order two @xmath365 $ ] , such that @xmath366 .
because of the generality assumption it follows that @xmath367 and the support of @xmath368 consists of @xmath182 distinct points @xmath369 . to simplify matters assume that @xmath370
. then it is proved in @xcite that the general prym curve @xmath371\in \cr_g$ ] is a section of a genus @xmath1 polarized nikulin surface @xmath372 , that is , @xmath355 and @xmath373 .
we consider the map induced by the linear system @xmath374 the points @xmath375 span a codimension @xmath2 linear subspace .
let @xmath376 be the pencil of curves on @xmath152 induced by the hyperplanes in @xmath377 through @xmath375 .
each curve @xmath378 contains the divisor @xmath379 as an _ odd _ theta characteristic .
the line bundle @xmath380 is an _ even _ theta characteristic on each curve @xmath381 , because as already discussed , the arf invariant remains constant in a family of spin curves .
this procedure induces a rational curve in moduli @xmath382,\ ] ] which passes through the general point @xmath143\in \ss_g^+$ ] and finishes the proof .
observe that in this proof , if instead of being a nikulin surface , @xmath152 is an arbitrary @xmath10 surface containing @xmath11 , the same reasoning can be used to construct a rational curve in @xmath8 that passes through a general point , provided the curve @xmath11 we started with , has general moduli .
a general curve of genus @xmath1 lies on a @xmath10 surface if an only if @xmath337 or @xmath383 , see @xcite .
the case @xmath384 can be handled via a slightly different idea , see @xcite theorem 3.10 .
thus one also has the following result : the odd spin moduli space @xmath8 is uniruled for @xmath385 .
we explain the strategy pursued in @xcite to construct alternative models of moduli spaces of odd spin curves which can then be used to establish unirationality of the moduli space : [ sgunir ] @xmath8 is unirational for @xmath386 .
the main idea is to construct a dominant map over @xmath8 from the total space of a projective bundle over a space parametrizing spin curves on nodal curves of smaller genus .
we begin by recalling that mukai , in a series of well - known papers @xcite , @xcite , @xcite , @xcite , has found ways of representing a general canonical curve of genus @xmath337 as a linear section of a certain @xmath387-dimensional rational homogeneous variety @xmath388 , which we shall call the _ mukai variety _ of genus @xmath1 .
one has the following list : _ - @xmath389 : the plcker embedding of the symplectic grassmannian @xmath390 , _ - @xmath391 : the plcker embedding of the grassmannian @xmath392 , - @xmath393 : the plcker embedding of the orthogonal grassmannian @xmath394 . inside the hilbert scheme of curvilinear sections of @xmath395 ,
we denote by @xmath396 the open subset classifying nodal sections @xmath397 by a linear space of dimension @xmath182 .
the automorphism group @xmath398 acts on @xmath396 and we call the git quotient @xmath399 the _ mukai model _ of the moduli space of curves of genus @xmath1
. note that with our definition , the variety @xmath400 is only quasi - projective and the picard number of @xmath400 is equal to @xmath66 .
the moduli map @xmath401 being @xmath398-invariant , it induces a regular map @xmath402 .
we can paraphrase mukai s results as stating that the map @xmath403 is a birational isomorphism , or equivalently , the general @xmath66-dimensional linear section of @xmath395 is a curve with general moduli .
the map @xmath404 deserves more study and in principle it can be used to answer various questions concerning the cohomology of @xmath159 or the minimal model program of the moduli space of curves ( see @xcite for a case in point when @xmath405 ) .
the following concept is key to our parametrization of @xmath8 using mukai models .
let @xmath406 be the space of _ clusters _ , that is , @xmath85-dimensional schemes @xmath407 of length @xmath408 with the following properties : * @xmath409 is a hyperplane section of a smooth curve section @xmath166 \in \cu_g$ ] , * @xmath409 has multiplicity two at each point of its support , * @xmath410 consists of @xmath182 linearly independent points .
a general point of @xmath411 corresponds to a @xmath85-cycle @xmath412 satisfying @xmath413 furthermore , @xmath411 is birational to the subvariety of the grassmannian variety @xmath414 parametrizing @xmath415-dimensional planes @xmath416 such that @xmath417 , where @xmath418 .
then we consider the incidence correspondence : the first projection map @xmath419 is finite of degree @xmath56 ; its fibre at a general point @xmath166\in \cu_g$ ] corresponding to a smooth curve classifies odd theta characteristics of @xmath11 .
the spin moduli map @xmath420 induces a birational isomorphism @xmath421 let us fix now an integer @xmath422 .
we define the locally closed set of pairs consisting of clusters and @xmath423-nodal curvilinear sections of @xmath395 , that is , @xmath424 the quotient of @xmath425 under the action of the automorphism group of @xmath395 is birational to the locus @xmath426 with general point given by an odd spin structure on a curve whose stable model is an irreducible @xmath423-nodal curve where each of the nodes is `` blown - up '' and an exceptional component is inserted . let us fix a general point @xmath427 and suppose that @xmath428 and denote the @xmath429 the remaining points in the support of @xmath409 . if @xmath430 is the normalization map , then we observe that @xmath431 , which gives rise to a point in the locus @xmath432 .
the important point now is that over @xmath425 one can consider an incidence correspondence that takes into account not only a @xmath423-nodal curve together with a cluster , but also all linear sections of @xmath395 that admit the same cluster .
precisely : the variety @xmath433 comes equipped with projection maps @xmath434 it is shown in @xcite that @xmath433 is birational to a projective bundle over @xmath425 and furthermore , the quotient @xmath435 is a projective bundle over @xmath432 .
moreover , it is proved that the projection map @xmath436 is dominant if and only if @xmath437 to summarize these considerations , we have reduced the unirationality of @xmath8 to two conditions .
one is numerical and depends solely on the mukai variety @xmath395 , the other has to do with the geometry of the spin moduli space @xmath432 of nodal curves of smaller geometric genus : for @xmath337 , if @xmath387 denotes the dimension of the corresponding mukai variety @xmath395 , the moduli space @xmath8 is unirational provided there exists an integer @xmath438 such that 1 .
@xmath439 , 2 .
@xmath432 is unirational .
it turns out that the locus @xmath440 is unirational for @xmath441 ( see @xcite theorem 4.16 ) .
however condition ( i ) is only satisfied when @xmath386 , and this is the range for which theorem [ sgunir ] is known at the moment .
in this section we discuss a few major themes related to aspects of the geometry of @xmath4 other than birational classification .
one can stratify theta characteristics by their number of global sections . for
an integer @xmath442 let us denote by @xmath443\in \cs_g : h^0(c , \theta)\ge r+1,\ \ h^0(c , \theta ) \equiv r+1 \mbox { mod } 2\bigr\}.\ ] ] the variety @xmath444 has a lagrangian determinantal structure discussed in @xcite theorem 1.10 , from which it follows that each component of @xmath444 has codimension at most @xmath445 inside @xmath4 .
this bound also follows from nagaraj s @xcite interpretation of the tangent space to the stack @xmath444 which we briefly explain .
fix a point @xmath143\in \cs_g^r$ ] and form the gaussian map @xmath446 more intrinsically , the projectivization of the map @xmath447 assigns to a pencil @xmath448 its ramification divisor . recalling the identification provided by kodaira - spencer theory @xmath449}(\cs_g)=h^0(c , k_c^{\otimes 2})^{\vee},\ ] ] it is shown in @xcite that the following isomorphism holds : this description is consistent with the bound @xmath450 from @xcite .
we now ask what is the actual dimension of the strata @xmath444 ? using hyperelliptic curves one can observe that @xmath451}\neq \emptyset$ ] even though the expected dimension of this locus as a determinantal variety is very negative .
moreover , the locus @xmath452 is non - empty and consists of ( theta characteristics on ) curves @xmath453 which are extremal from the point of view of castelnuovo s bound . therefore one can not hope that the dimension of @xmath444 be always @xmath454
. however this should be the case , and the locus @xmath444 should enjoy certain regularity properties , when @xmath455 is relatively small with respect to @xmath1 .
we recall the following precise prediction from @xcite : the conjecture is proved in @xcite for all integers @xmath458 and @xmath459 .
we point out that @xmath460 coincides with the divisor @xmath461 studied in @xcite and @xcite . to prove conjecture [ strat ] it suffices to exhibit a single spin curve @xmath143\in \cs_g^r$ ] with an injective gaussian map @xmath447 . as further evidence ,
we mention the following result , see @xcite proposition 2.4 : one could ask whether in the range @xmath457 , the locus @xmath444 is pure - dimensional , or even irreducible .
not much evidence in favor of this speculation exists , but there are no counterexamples either .
we mention however that @xmath465 is pure of codimension @xmath466 in @xmath155 , and when the locus @xmath467 has pure codimension @xmath468 in @xmath154 for @xmath469 , see @xcite .
[ [ syzygies - of - theta - characteristics - and - canonical - rings - of - surfaces . ] ] syzygies of theta characteristics and canonical rings of surfaces .
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ some tentative steps in this direction appear in @xcite , where it is shown that with a few exceptions , @xmath472 is generated in degree at most @xmath466 .
the interest in this question comes to a large extent from the study of surfaces of general type .
let @xmath473 be a canonically embedded surface of general type and assume for simplicity that @xmath474 and @xmath475 .
then a general hyperplane section @xmath476 comes equipped with an @xmath455-dimensional theta characteristics , that is , @xmath477\in \cs_g^r$ ] . by restriction
there is a surjective morphism of graded rings @xmath478 and the syzygies of the two rings are identical , that is , @xmath479 it is worth mentioning that using green s duality theory @xcite , one finds the isomorphism @xmath480 between the various koszul cohomology groups . in a departure from the much studied case of syzygies of canonical curves ,
the graded betti diagram of a theta characteristic has three non - trivial rows . for two - torsion points
@xmath481 a precise _ prym - green _ conjecture concerning the groups @xmath482 has been formulated ( and proven for bounded genus ) in @xcite .
there is no clear prediction yet for the vanishing of @xmath483 .
[ [ the - scorza - correspondence - on - the - moduli - space - of - even - spin - curves . ] ] the scorza correspondence on the moduli space of even spin curves .
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ to a non - effective even theta characteristic @xmath143\in \cs_g^+$ ] one can associate the _ scorza correspondence _
@xmath484 denoting by @xmath485 the two projections and by @xmath486 the diagonal , the cohomology class of the scorza curve can be computed : @xmath487 by the adjunction formula , @xmath488 .
the curve @xmath489 , first considered by scorza @xcite , reappears in the modern literature in the beautiful paper @xcite , where it plays an important role in the construction of an explicit birational isomorphism between @xmath490 and @xmath491 .
it is shown in @xcite that @xmath489 is smooth for a general even spin curve , hence one can consider the scorza map at the level of moduli space , that is , @xmath492:=[r_{\theta}].\ ] ] since @xmath7 is a normal variety , the rational map @xmath493 extends to a regular morphism outside a closed set of @xmath7 of codimension at least two .
it is of interest to study this map , in particular to answer the following questions : to give one example , we explain one of the results proved . for a general point @xmath143\in \thet$ ] , we denote by @xmath497 the _ trace curve _ induced by the pencil @xmath498 , that is , @xmath499 we set @xmath500 and it is easy to see that for a generic choice of @xmath143\in \thet$ ] , the set @xmath423 consists of @xmath501 distinct points .
we consider a family @xmath502 of even theta characteristics over a @xmath66-dimensional base , such that for a point @xmath503 we have that @xmath504=[c , \theta]\in \thet$ ] and @xmath505 for @xmath506 .
in particular , the cycle @xmath507 is defined for @xmath508 .
we prove the following result : the flat limit of the family of scorza curves @xmath509 corresponding to @xmath510 , is the non - reduced cycle @xmath511 the associated stable curve @xmath512\in \mm_{1 + 3g(g-1)}$ ] can be described as the transverse union @xmath513 , where @xmath514 is the double cover of @xmath262 branched over @xmath423 .
ems e. arbarello , m. cornalba , p. a. griffiths , j. harris , _ geometry of algebraic curves _ , grundlehren der mathematischen wissenschaften 267 ( 1985 ) , springer verlag . c. arf , _ untersuchungen ber quadratische formen in krpern der charakteristik @xmath2 _ , j. reine angewandte mathematik * 183 * ( 1941 ) , 148 - 167 .
m. atiyah , _ riemann surfaces and spin structures _ , annales scient .
cole normale sup .
* 4 * ( 1971 ) , 47 - 62 .
baker , _ abelian functions : abel s theorem and the allied theory of theta functions _ , 1897 , reissued in the cambridge mathematical library 2006 .
s. boucksom , j.p .
demailly , m. pun and t. peternell , _ the pseudo - effective cone of a compact khler manifold and varieties of negative kodaira dimension _ , arxiv : math/0405285 .
c. birkenhake and h. lange , _ complex abelian varieties _ , grundlehren der mathematischen wissenschaften 302 , 2nd edition 2004 , springer verlag .
s. cacciatori , f. dalla piazza and b. van geemen , _ modular forms and three loop superstring amplitudes _ , nuclear physics b * 800 * ( 2008 ) , 565 - 590 .
l. caporaso , _ a compactification of the universal picard variety over the moduli space of stable curves _ , journal of the american mathematical society * 7 * ( 1994 ) , 589 - 660 .
l. caporaso and c. casagrande _ combinatorial properties of stable spin curves _ , communications in algebra * 359 * ( 2007 ) , 3733 - 3768 . l. caporaso and e. sernesi , _ characterizing curves by their theta characteristics _ ,
j. reine angewandte mathematik * 562 * ( 2003 ) , 101 - 135 .
a. coble , _ algebraic geometry and theta functions _ ,
american mathematical society colloquium publications , volume x 1929 , reprinted and revised in 1961 .
m. cornalba , _ moduli of curves and theta characteristics _ , in : lectures on riemann surfaces ( trieste , 1987 ) , world sci .
publ . , 560 - 589
. i. dolgachev and d. ortland , _ point sets in projective space and theta functions _ , astrisque , volume 165 ( 1988 ) , socit mathmatique de france .
i. dolgachev and v. kanev , _ polar covariants of plane cubics and quartics _ , advances in mathematics * 98 * ( 1993 ) , 216 - 301 .
d. eisenbud and j. harris , _ the kodaira dimension of the moduli space of curves of genus @xmath515 _ , inventiones math .
* 90 * ( 1987 ) , 359387 .
g. farkas , _
gaussian maps , gieseker - petri loci and large theta characteristics _ , j. reine angewandte mathematik , * 581 * ( 2005 ) , 151 - 173 . g. farkas , _ the birational type of the moduli space of even spin curves _ ,
advances in mathematics * 223 * ( 2010 ) , 433 - 443 .
g. farkas , _ brill - noether geometry on moduli spaces of spin curves _ , in : classification of algebraic varieties , schiermonnikoog 2009 ( c. faber , g. van der geer , e. looijenga editors ) , series of congress reports , european mathematical society 2011 , 259 - 276 .
g. farkas , _ koszul divisors on moduli spaces of curves _ , american journal of mathematics * 131 * ( 2009 ) , 819 - 869 .
g. farkas and e. izadi , _ the scorza correspondence on the moduli space of stable spin curves _ , in preparation .
g. farkas and k. ludwig , _ the kodaira dimension of the moduli space of prym varieties _ , journal of the european mathematical society * 12 * ( 2010 ) , 755 - 795 . g. farkas and a. verra , _ the geometry of the moduli space of odd spin curves _ ,
g. farkas and a. verra , _ moduli of theta characteristics via nikulin surfaces _ , arxiv:1104.0273 , to appear in mathematische annalen .
m. fedorchuk , _ the log canonical model program on the moduli space of stable curves of genus four _ , arxiv:1106.5012 , to appear in international mathematical research notices
. g. frobenius , _ ber thetafunktionen mehrerer variablen _ , j. reine angewandte mathematik * 96 * ( 1884 ) , 100 - 122 . g. frobenius , _ ber gruppen von thetacharakteristiken _ , j. reine angewandte mathematik * 96 * ( 1884 ) , 81 - 99 .
g. frobenius , _ antrittsrede _ , sitzungsberichte der kniglich preussischen akademie der wissenschaften zu berlin 1893 , 626 - 632 .
b. van geemen and a. sarti , _ nikulin involutions on @xmath10 surfaces _ , mathematische zeitschrift * 255 * ( 2007 ) , 731 - 753 .
a. gpel , _ theoriae transcendentium abelianarum primi ordinis adumbratio levis _ , j. reine angewandte mathematik * 35 * ( 1877 ) , 277 - 312 . m. green , _ koszul cohomology and the cohomology of projective varieties _
, journal of differential geometry * 19 * ( 1984 ) , 125 - 171 .
b. gross and j. harris , _ on some geometric constructions related to theta characteristics _ , in : contributions to automorphic forms , geometry and number theory , johns hopkins university press , baltimore 2004 , 279 - 311 .
s. grushevsky , _ superstring scattering amplitudes in higher genus _ , communications in mathematical physics * 294 * ( 2010 ) , 343 - 352 .
s. grushevsky and r. salvati manni , _ the scorza correspondence in genus @xmath466 _ , arxiv:1009.0375 . j. harris , _ theta characteristics on algebraic curves _
, transactions american mathematical society * 271 * ( 1982 ) , 611 - 638 .
j. harris and d. mumford , _ on the kodaira dimension of @xmath159 _ , inventiones math .
* 67 * ( 1982 ) , 23 - 88 .
a. krazer , _ lehrbuch der thetafunktionen _ , verlag von b.g .
teubner 1903 , leipzig .
a. logan , _ the kodaira dimension of moduli spaces of curves with marked points _ , american journal of mathematics * 125 * ( 2003 ) , 105 - 138 .
k. ludwig , _ on the geometry of the moduli space of spin curves _ , journal of algebraic geometry * 19 * ( 2010 ) , 133 - 171 .
d. nagaraj , _ on the moduli of curves with theta characteristics _ , compositio math .
* 75 * ( 1990 ) , 287 - 297 .
a. mattuck , _ arthur bryan coble _ , bulletin of the american mathematical society * 76 * ( 1970 ) , 693 - 699
. s. mukai , _ curves , @xmath10 surfaces and fano @xmath466-folds of genus @xmath516 _ , in : algebraic geometry and commutative algebra 357 - 377 , 1988 kinokuniya , tokyo .
s. mukai , _ curves and grassmannians _ , in : algebraic geometry and related topics , eds .
yang , y. namikawa , k. ueno , 19 - 40 , 1992 international press .
s. mukai , _ curves and symmetric spaces i _ , american journal of mathematics * 117 * ( 1995 ) , 1627 - 1644 .
s. mukai , _ curves and symmetric spaces ii _ , annals of mathematics * 172 * ( 2010 ) , 1539 - 1558
. d. mumford , _ theta characteristics of an algebraic curve _
, annales scient .
cole normale sup .
* 4 * ( 1971 ) , 181 - 192 .
r. pandharipande , _ a compactification over @xmath159 of the universal moduli space of slope - semistable vector bundles _ , journal of the american mathematical society * 9 * ( 1996 ) , 425 - 471 .
a. putman , _ the picard group of the moduli space of curves with level structure _ , to appear in duke mathematical journal .
m. reid , _ infinitesimal view of extending a hyperplane section - deformation theory and computer algebra _ , in : algebraic geometry laquila 1988 , lecture notes in mathematics * 1417 * , 214 - 286 , springer verlag .
g. rosenhain , _ mmoire sur les fonctions de deux variables , qui sont les inverses des intgrales ultra - elliptiques de la premire classe _ , paris 1851 .
r. salvati manni , _ modular varieties with level @xmath2 theta structure _ , american journal of mathematics * 116 * ( 1994 ) , 1489 - 1511 . r. salvati manni , _ remarks on superstring amplitudes in higher genus _ , arxiv:0804.0512 , nuclear physics b , * 801 * ( 2008 ) , 163 - 173 .
g. scorza , _ sopra le curve canoniche di uno spazio lineare qualunque e sopra certi loro covarianti quartici _ , atti accad .
reale sci .
torino * 35 * ( 1900 ) , 765 - 773 .
m. teixidor i bigas , _ half - canonical series on algebraic curves _
, transactions of the american mathematical society * 302 * ( 1987 ) , 99 - 115 .
m. teixidor i bigas , _ the divisor of curves with a vanishing theta - null _ , compositio math . * 66 * ( 1988 ) , 15 - 22 . w. wirtinger , _ untersuchungen ber thetafunktionen _ , verlag von b.g .
teubner , leipzig 1895 . | we discuss topics related to the geometry of theta characteristics on algebraic curves .
they include the birational classification of the moduli space @xmath0 of spin curves of genus @xmath1 , interpretation of theta characteristics as quadrics in a vector space over the field with two elements as well as the connection with modular forms and superstring scattering amplitudes .
special attention is paid to the historical development of the subject .
[ multiblock footnote omitted ] theta characteristics appeared for the first time in the context of characteristic theory of odd and even theta functions in the papers of gpel @xcite and rosenhain @xcite on jacobi s inversion formula for genus @xmath2 .
they were initially considered in connection with _
riemann s bilinear addition _ relation between the degree two monomials in theta functions with characteristics . later , in order to systematize the relations between theta constants , frobenius @xcite , @xcite developed an algebra of characteristics ; he distinguished between _ period _ and _ theta _ characteristics .
the distinction , which in modern terms amounts to the difference between the _ prym _ moduli space @xmath3 and the _ spin _ moduli space @xmath4 , played a crucial role in elucidating the transformation law for theta functions under a linear transformation of the moduli , and it ultimately led to a correct definition of the action of symplectic group @xmath5 on the set of characteristics .
an overwiew of the 19th century theory of theta functions can be found in krazer s monumental book @xcite .
it is a very analytic treatise in character , with most geometric applications either completely absent relegated to footnotes .
the remarkable book @xcite by coble represents a departure from the analytic view towards a more abstract understanding of theta characteristics using configurations in finite geometry .
coble viewed theta characteristics as quadrics in a vector space over @xmath6 . in this language frobenius
earlier concepts ( syzygetic and azygetic triples , fundamental systems of characteristics ) have an elegant translation .
with fashions in algebraic geometry drastically changing , the work of coble was forgotten for many decades .
the modern theory of theta characteristic begins with the works of atiyah @xcite and mumford @xcite ; they showed , in the analytic ( respectively algebraic ) category , that the parity of a theta characteristic is stable under deformations .
in particular , mumford s functorial view of the subject , opened up the way to extending the study of theta characteristics to singular curves ( achieved by harris @xcite ) , to constructing a proper deligne - mumford moduli space of stable spin curves ( carried out by cornalba @xcite ) , or to reinterpreting coble s work in modern terms ( see the book @xcite ) .
the aim of this paper is to survey various developments concerning the geometry of moduli spaces of spin curves .
particular emphasis is placed on the complete birational classification of both the even and the odd spin moduli spaces , which has been carried out in the papers @xcite , @xcite and @xcite .
precisely , we shall explain the following result : the birational type of the moduli spaces @xmath7 and @xmath8 of even and odd spin curves of genus @xmath1 can be summarized as follows : @xmath7 : [ cols="^,^",options="header " , ] we describe the structure of the paper . in the first two sections we recall the interpretation of theta characteristics as quadrics in an @xmath6-vector space , then link this description both with the classical theory of characteristics of theta functions and modern developments inspired by string theory . in the next three sections we explain some features of the geometry of the moduli space @xmath9 of stable spin curves of genus @xmath1 , discuss ways of constructing effective divisors on @xmath9 and computing their cohomology classes , then finally present unirational parametrizations of the moduli space in small genus , by using mukai models and special @xmath10 surfaces .
we close by surveying a few open problems related to syzygies of theta characteristics and stratifications of the moduli space . |
support vector machines ( boser et al . , 1992 ) belong to core machine learning techniques for binary classification .
given a large number of training samples characterized by a large number of features , a linear svm is often the _ go - to _ approach in many applications . a handy collection of software packages ,
e.g. , ` liblinear ` ( fan et al . ,
2008 ) , ` pegasos ` ( shalev - shwartz et al . , 2011 ) , ` svm^{\texttt{perf } } ` ( joachims , 2006 ) , ` scikit - learn ` ( pedregosa et al . , 2011 ) provide practitioners with efficient algorithms for fitting linear models to datasets .
finding optimal hyperparameters of the algorithms for model selection is crucial though for good performance at test - time .
a vanilla cross - validated grid - search is the most common approach to choosing satisfactory hyperparameters .
however , grid search scales exponentially with the number of hyperparameters while choosing the right sampling scheme over the hyperparameter space impacts model performance ( bergstra & bengio , 2012 ) .
linear svms typically require setting a single @xmath0 hyperparameter that equally regularizes the training ` loss ` of misclassified data .
( klatzer & pock , 2015 ) propose bi - level optimization for searching several hyperparameters of linear and kernel svms and ( chu et al . , 2015 ) use warm - start techniques to efficiently fit an svm to large datasets but both approaches explore the hyperparameter regularization space partially . the algorithm proposed in ( hastie et al . , 2004 )
builds the entire regularization path for linear and kernel svms that use single , symmetric cost for misclassifying negative and positive data .
the stability of the algorithm was improved in ( ong et al . , 2010 ) by augmenting the search space of feasible event updates from one- to multi - dimensional hyperparameter space . in this paper
, we also show that a one - dimensional path following method can diverge to unoptimal solution wrt kkt conditions .
many problems often require setting multiple hyperparameters ( karasuyama et al . , 2012 ) .
they arise especially when dealing with imbalanced datasets ( japkowicz & stephen , 2002 ) and require training an svm with two cost hyperparameters assymetrically attributed to positive and negative examples .
( bach et al . , 2006 ) builds a pencil of one - dimensional regularization paths for the assymetric - cost svms . on the other hand , ( karasuyama et al . , 2012 )
build a one - dimensional regularization path but in a multidimensional hyperspace .
in contrast to algorithms building one - dimensional paths in higher - dimensional hyperparameter spaces , we describe a solution path algorithm that explores the entire regularization path for an assymetric - cost linear svms .
hence , our path is a two - dimensional path in the two - dimensional hyperparameter space .
our main contributions include : * development of the entire regularization path for assymetric - cost linear support vector machine ( ac - lsvm ) * algorithm initialization at arbitrary location in the @xmath1 hyperparameter space * computationally and memory efficient algorithm amenable to local parallelization .
our binary classification task requires a _ fixed _ input set of @xmath2 training examples @xmath3 , where @xmath4 , @xmath5 , @xmath6 , to be annotated with corresponding binary labels @xmath7 denoting either class . then , the objective is to learn a decision function @xmath8 that will allow its associated classifier @xmath9 $ ] to predict the label @xmath10 for new sample @xmath11 at test - time .
[ [ active - sets ] ] active sets + + + + + + + + + + + solving the primal qp is often approached with the help of lagrange multipliers @xmath29 , where @xmath30^t$ ] , which are associated with @xmath2 constraints in .
let @xmath31 $ ] and @xmath32^t $ ] .
then , the dual problem takes the familiar form : @xmath33 the immediate consequence of applying the lagrange multipliers is the expression for the lsvm parameters @xmath34 yielding the decision function @xmath35 .
the optimal solution @xmath36 of the dual problem is dictated by satisfying the usual karush - kuhn - tucker ( kkt ) conditions .
notably , the kkt conditions can be algebraically rearranged giving rise to the following _ active sets _ : @xmath37 @xmath38 @xmath39 firstly , the @xmath40 sets @xmath41 cluster data points @xmath42 to the margin @xmath43 , to the left @xmath44 , and to the right @xmath45 of the margin along with their associated scores @xmath46 .
secondly , the sets indicate the range within the @xmath1 space for lagrange multipliers @xmath47 over which @xmath36 is allowed to vary thereby giving rise to a convex polytope in that space . [ [ convex - polytope ] ] convex polytope + + + + + + + + + + + + + + + a unique region in @xmath1 satisfying a particular configuration of the @xmath40 set is bounded by a convex polytope . the first task in path exploration
is thus to obtain the boundaries of the convex polytope . following ( hastie , 2004 ) , we obtain linear inequality constraints from @xmath41 : @xmath48 \label{eq : h_alpha0}\ ] ] @xmath49^t \label{eq : h_alphac1}\ ] ] @xmath50^t \label{eq : h_alphac2}\ ] ] @xmath51 \label{eq : h_l}\ ] ] @xmath52 \label{eq : h_r}\ ] ] where , @xmath53 is the orthogonal projector onto the orthogonal complement of the subspace spanned by @xmath54 and @xmath55 is the moore - penrose pseudoinverse if @xmath54 has full column rank .
specifically , let @xmath56 be a matrix composed of constraints @xmath41 .
@xmath57 \label{eq : h}\ ] ] then , the boundaries of the convex polytope in the @xmath1 space are indicated by a subset of active constraints in @xmath58 , which evaluate to @xmath59 for some @xmath60 $ ] .
the boundaries can be determined in linear time @xmath61 with efficient convex hull ( ` ch ` ) routines ( avis et al . , 1997 ) .
now , in order grow the entire regularization path in @xmath1 , the @xmath40 sets have to updated at @xmath62-th step such that the kkt conditions will hold , thereby determining an @xmath63-th convex polytope
. the polytope constraints @xmath64 for which @xmath65 indicate that a point @xmath42 has to go from @xmath43 to @xmath45 in order to satisfy the kkt conditions .
likewise , @xmath66 for which @xmath67 indicate updating the point from @xmath43 to @xmath44 , @xmath68 indicate point transition from @xmath45 to @xmath43 , and @xmath69 indicate point transition from @xmath44 to @xmath43 .
these set transitions are usually called events , while the activated constraints are called breakpoints . therefore , at breakpoint , we determine the event for @xmath70-th point by a function that updates the @xmath71 set from @xmath62 to @xmath72 as : @xmath73 where the direction of the transition depends on the current @xmath71 set configuration . following ( hastie et al . , 2004 )
, our algorithm requires @xmath71 set to proceed to the next set by .
however , unlike ( hastie et al . , 2004 )
, the constraints @xmath41 are independent of the previous computations of @xmath47 and @xmath20 .
this has several implications .
firstly , our algorithm does not accumulate potential numerical errors in these parameters .
secondly , the algorithm can be initialized from an arbitrary location in the @xmath1 space .
the evolution of @xmath36 is continuous and piecewise linear in the @xmath1 space ( bach et al . , 2006 ) .
an immediate consequence is that the active constraints have to flip during the set update .
[ [ flipping - constraints ] ] flipping constraints + + + + + + + + + + + + + + + + + + + + suppose we have a single event that @xmath74 goes to @xmath75 thereby forcing the set update rule @xmath76 . then , the @xmath70-th constraint in can be rearranged using the _ matrix inversion lemma _
wrt @xmath42 as : @xmath77 where @xmath78 .
this constraint is equal to its corresponding , sign - flipped counterpart in at @xmath72 as : @xmath79 with @xmath80 .
the same argument holds for update type @xmath81 .
furthermore , ( hastie et al . , 2004 ) express the evolution of @xmath36 with a single cost parameter @xmath0 .
this case is equivalent to @xmath82 that yields the identity line in the @xmath1 space .
( ong et al . , 2010 ) observe that one - dimensional path exploration over a line can lead to incorrect results and resort to searching for alternative set updates over a higher - dimensional hyperparameter space .
notably , when two points hit the margin at the same time at @xmath62 , the matrix updated by both points @xmath83 not necessarily needs to become singular .
however , the @xmath71 sets can be incorrectly updated .
we formalize this by introducing the notion of _ joint events _ that may co - occur at some point on the line . in our setting of the @xmath84d path exploration
, this is always the case when a vertex of a polytope coincides with a line in the @xmath1 space .
[ [ joint - events ] ] joint events + + + + + + + + + + + + at the vertices of the convex polytope at least two events occur concurrently . in this case
, the @xmath40 set can be updated twice from @xmath62 to @xmath72 .
hence , this vertex calls @xmath85 different updates of the @xmath71 set , _
i.e. _ two single updates for both edges and a joint update . note that the piecewise continuous @xmath84d path of @xmath36 also implies piecewise continuous @xmath21d path of the events .
moreover , as each vertex is surrounded by @xmath86 different @xmath40 sets , two events at the vertex have to satisfy the following _ vertex loop property _ : @xmath87 stating that the events have to flip at the vertex such that a sequence of up to @xmath85 single updates reaches each @xmath88 set from any other @xmath89 set associated with that vertex .
we now describe our algorithm .
we represent the entire regularization path for the ac - lsvm by the set of vertices @xmath90 , edges @xmath91 , and facets @xmath92 .
let @xmath93 .
then : @xmath94 @xmath95 @xmath96 where @xmath97 and @xmath98 denote attribute and connectivity , respectively , of each element in the @xmath99 sets . [ [ ordering ] ] ordering + + + + + + + + the @xmath99 sets admit the following connectivity structure .
let @xmath100 , @xmath101 , and @xmath102 be partitioned into @xmath103 subsets where @xmath104 , @xmath105 , and @xmath106 .
the subsets admit a sequential ordering , where @xmath107 , @xmath108 , and @xmath109 , such that edges @xmath110 determine the adjacency of facet pairs @xmath111 or @xmath112 while vertices @xmath113 determine the intersection of edges @xmath110 or @xmath114 . in effect , our algorithm orders facets into a _ layer_-like structure .
we define a vertex @xmath115 as an _
open vertex _
@xmath116 when @xmath117 or @xmath118 , where @xmath119 is set cardinality , if the vertex does not lie on neither @xmath0 axis .
we define a _ closed vertex _ when @xmath120 .
when @xmath121 , the vertex is also closed if it lies on either @xmath0 axis .
similarly , an edge @xmath122 is called an _ open edge _
@xmath123 when @xmath124 and a _ closed edge _
@xmath125 when @xmath126 .
then , a facet @xmath127 is called an _
open facet _
when first and last edge in @xmath128 are unequal ; otherwise it is a _ closed facet_. finally , @xmath129 are called either _ single _
@xmath130 when they are unique or _ replicated _
@xmath131 , otherwise .
we propose to explore the ac - lsvm regularization path in a sequence of layers @xmath132 .
we require that @xmath133 facet attributes are given at the beginning , where @xmath134 , @xmath135 , and @xmath136 .
an @xmath137-th layer is then composed of four successive steps . *
closing open edges and facets * - ` cef ` for each @xmath71 set , which is attributed to @xmath138 , the algorithm separately calls a convex hull routine ` ch ` . the routine uses to compute linear inequality constraints @xmath139 creating a convex polytope at @xmath62 .
the ordered set of edges @xmath140 , where the first and last edge are open , serve as initial , mandatory constraints in the ` ch ` routine .
after completion , the routine augments the set @xmath141 by closed edges @xmath142 and the set @xmath143 by open vertices @xmath144 . *
2 . merging closed edges and open vertices * - ` mev ` as the ` ch ` routine runs for each facet @xmath138 separately , some edges @xmath142 and/or vertices @xmath144 may be replicated , thereby yielding @xmath145 and @xmath146 . notably , a vertex @xmath147 is replicated when another vertex @xmath148 ( or other vertices ) has the same attribute , i.e. @xmath149 .
however , we argue that merging vertices into a single vertex based on the distance between them in some metric space may affect the numerical stability of the algorithm . on the other hand , a closed edge @xmath150 is replicated by another closed edge @xmath151 , when both edges connect a pair of vertices that are both replicated .
replicated edges can not merge solely by comparing their event attributes @xmath152 .
as they are piecewise continuous in the @xmath1 space , they are not unique . similarly to vertices
though , the edges might be merged by numerically comparing their associated linear constraints , which are only sign - flipped versions of each other , as shown in @xmath41 .
however , this again raises concerns about the potential numeric instability of such a merging procedure . in view of this
, we propose a sequential merging procedure that leverages @xmath153 sets , which are both unique in @xmath154 and discrete . to this end , we first introduce two functions that act on attributes and connectivity of objects @xmath155 .
let @xmath156 be an indexing function that groups @xmath157 by assigning labels from set @xmath158 to @xmath159 based on @xmath160 indexed over @xmath161 : @xmath162 let @xmath163 then be a relabeling function that assigns labels from @xmath164 to @xmath165 indexed over @xmath166 : @xmath167 the algorithm commences the merging procedure by populating initially empty set @xmath168 with facets @xmath169 that are obtained by separately updating the facets @xmath170 through the events attributed to each edge @xmath171 .
note , however , that replicated edges @xmath172 will produce facet attributes in @xmath168 that replicate facet attributes from the preceding layer .
moreover , single edges @xmath173 may as well produce replicated facet attributes in the current layer .
hence , we have that @xmath174 . in order to group facets into single and replicated @xmath40 sets , the algorithm indexes facet attributes with @xmath175 based on their equality .
then , relabeling facet - edge connectivities of edges @xmath142 with @xmath176 allows for indexing the connectivities with @xmath177 also based on their equality . having indicated single and replicated edges , the algorithm relabels edge - vertex connectivities of vertices @xmath144 with @xmath178 . note that there are two general vertex replication schemes
. vertices , which indicate the occurrence of joint events , can be replicated when two facets connect ( i ) through an edge ( i.e. , vertices share replicated edge ) or ( ii ) through a vertex ( i.e. , vertices connect only to their respective , single edges ) . at this point of the merging procedure ,
a vertex @xmath144 is associated indirectly through edges @xmath179 with two facets @xmath180 , when it lies on either @xmath0 axis , or three facets @xmath181 , otherwise .
two vertices lying on , say , @xmath27 axis are replicated when their respective edges share a facet and are attributed the events that refer to the same , negative point @xmath182 , but yet that have opposite event types , i.e. @xmath183 .
this condition follows directly from @xmath41 , as @xmath184 at @xmath185 .
conversely , when vertices lie on @xmath28 axis , they are replicated when their edges have events referring to the positive point , @xmath186 .
then , two vertices lying on neither @xmath0 axis are replicated when their respective edges are associated with two common facets and equal joint events .
hence , vertices are indexed with @xmath187 based on the equality of edge - facet connectivities along with edge attributes .
alternatively , should the joint events be unique , the vertices could then be merged solely by comparing these events .
showing that joint events being unique is true or false , i.e. two @xmath21d event paths can intersect only at a single point in the entire @xmath1 space , is an interesting future work . having grouped facets @xmath169 ,
edges @xmath188 , and vertices @xmath189 , now the algorithm can merge facet - edge connectivities @xmath190 of the replicated facets , prune replicated edges @xmath191 , and merge vertex - edge connectivites @xmath192 of the replicated vertices .
being left with only single facets @xmath193 , edges @xmath194 , and vertices @xmath195 , the algorithm relabels with @xmath196 the edge - vertex connectivities of single , open edges @xmath197 and single , closed edges @xmath179 intersecting with @xmath197 .
finally , the algorithm relabels with @xmath198 the facet - edge connectivites of facets from the preceding and current layer . *
3 . closing open vertices * - ` cv ` in this step , the algorithm closes the vertices @xmath199 by attaching open edges . specifically , by exploiting the piecewise continuity of events at vertices , the algorithm populates the @xmath200 set with open edges @xmath201 , such that a vertex @xmath202 now connects either to ( i ) @xmath85 edges , when it lies on , say , @xmath27 axis and connects to event edge with associated positive point , or to ( ii ) @xmath86 edges when it lies on neither axis . using the vertex loop property
, the algorithm then augments the set @xmath203 with additional facets @xmath204 such that now the closed vertex @xmath202 connects indirectly through its edges @xmath205 to facets @xmath206 and additionally up to @xmath21 new facet @xmath207 .
there are several advantages for generating open edges .
firstly , augmenting the initialization list of edges during the call to the ` ch ` routine reduces the number of points for processing , with computational load @xmath208 .
secondly , each vertex generates up to two single open edges .
however , there can be two single vertices that generate the same open edge thereby merging the @xmath21d path of an event .
in this case , both open edges are merged into a single closed edge and the facet is closed without processing it with ` ch ` routine .
this merging step is described next . *
merging open edges and facets * - ` mef ` as open edges and their facets , which are generated in step @xmath85 , can also be single or replicated , step @xmath86 proceeds similarly to step @xmath84 .
the algorithm indexes additional facets with @xmath209 and relabels the open edge connectivities with @xmath210 .
then , the algorithm indexes these connectivites with @xmath211 and merges edge - vertex @xmath212 and facet - edge @xmath190 connectivities .
finally , the algorithm relabels with @xmath213 the facet - edge connectivity of all facets in @xmath168 and returns to step @xmath21 . [
[ termination ] ] termination + + + + + + + + + + + the algorithm terminates at @xmath103-th layer , in which the ` ch ` routine for @xmath40 sets of all facets in @xmath214 , where @xmath215 , produces _ open polytopes _ in the @xmath1 space .
[ [ special - cases ] ] special cases + + + + + + + + + + + + + as mentioned in ( hastie et al . , 2004 )
, two special case events may occur after closing facet @xmath127 and set updating . when ( i ) replicated data points @xmath216 exist in the dataset and enter the margin , or ( ii ) single points simultaneously project onto the margin such that @xmath217 , then the matrix @xmath83 becomes singular and thus not invertible , yielding non - unique paths for some @xmath47 .
in contrast to ( hastie et al . , 2004 ) , note that the case ( ii ) is likely to occur in the considered lsvm formulation ( 1)@xmath41 as the positive and negative data points span up to @xmath218 subspace after being affine transformed , yielding e.g. parameters @xmath219 $ ] . in the context of our algorithm , both cases ( i)@xmath41(ii )
are detected at @xmath127 when the matrix @xmath220 formed of @xmath221 constraints @xmath41 associated with these points either has @xmath222 , producing multiple events at an edge denoted by constraints that are identical up to positive scale factor , or has @xmath223 , producing multiple joint events at a vertex denoted by constraints that intersect at the same point .
we propose the following procedure for handling both special cases .
namely , when some facets @xmath224 close with edges having multiple events or with vertices having multiple joint events that would lead to cases ( i)@xmath41(ii ) , the algorithm moves to step @xmath84 , as it can obtain facet updates in these special cases . however , it skips step @xmath85 for these particular facets .
while we empirically observed that such vertices close with @xmath84 edges having multiple joint events , it is an open issue how to generate open edges in this case .
instead , during successive layers , step @xmath84 augments the list of facets , edges , and vertices by the ones associated to ( i)@xmath41(ii ) for indexing and relabeling them with respect to successive ones that will become replicated in further layers . in effect
, our algorithm goes around these special case facets and attempts to close them by computing adjacent facets .
however , the path for @xmath36 in these cases is not unique and remains unexplored .
nevertheless , our experiments suggest that unexplored regions occupy relatively negligibly small area in the hyperparameter space .
when the algorithm starts with all points in @xmath44 and either case ( i)@xmath41(ii ) occurs at the initial layers , the exploration of the path may haltd multiple event paths referring to these cases will go to both @xmath0 axis , instead of to one @xmath0 axis and to infinity . ] due to the piecewise continuity of the ( multiple ) events .
a workaround can then be to run a regular lsvm solver at yet unexplored point @xmath1 , obtain @xmath40 sets , and extract convex polytope to restart the algorithm .
our future work will focus on improving our tactics for special cases .
we posit that one worthy challenge in this regard is to efficiently build the entire regularization path in @xmath2-dimensional hyperparameter space . [
[ computational - complexity ] ] computational complexity + + + + + + + + + + + + + + + + + + + + + + + + let @xmath225 be the average size of a margin set for all @xmath62 , let @xmath226 be the average size of @xmath154 .
then , the complexity of our algorithm is @xmath227 , where @xmath228 is the number of computations for solving ( without inverse updating / downdating ( hastie et al . , 2004 ) ) and we hid constant factor @xmath84 related to convex hull computation .
however , note that typically we have @xmath229 .
in addition , we _ empirically _ observed that @xmath230 ( but cf .
( gartner et al . , 2012 ) ) , so that the number of layers @xmath137 approximates dataset size .
our algorithm is sequential in @xmath103 but parallel in @xmath226 .
therefore , the complexity of a parallel implementation of the algorithm can drop to @xmath231 . finally , at each facet
, it is necessary to evaluate .
but then the evaluation of constraints @xmath41 can be computed in parallel , as well .
while this would lead to further reduce the computational burden , memory transfer remains the main bottleneck on modern computer architectures .
our algorithm partitions the sets @xmath92 , @xmath91 , @xmath90 into a _ layer_-like structure such that our two - step merging procedure requires access to objects only from layer pairs @xmath137 and @xmath232 and not to preceding layers , it requires access to @xmath127 , @xmath122 , @xmath115 objects related to these cases even after @xmath233 layers , but the number of these objects is typically small . ] . in effect , the algorithm only requires @xmath234 memory to cache the sets at @xmath137 , where @xmath235 and @xmath236 are average edge and vertex subset sizes of @xmath141 and @xmath143 , respectively .
in this section , we evaluate our ac - lsvmpath algorithm described in section [ sec : algo ] .
we conduct three numerical experiments for exploring the two - dimensional path of assymetric - cost lsvms on synthetic data .
we generate samples from a gaussian distribution @xmath237 for ( i ) a small dataset with large number of features @xmath238 , ( ii ) a large dataset with small number of features @xmath239 , and ( iii ) a moderate size dataset with moderate number of features @xmath240 .
we also build two - dimensional regularization path when input features are sparse ( iv ) .
we use off - the - shelf algorithm for training flexible part mixtures model ( yang & ramanan , 2013 ) , that uses positive examples from parse dataset and negative examples from inria s person dataset ( dalal & triggs , 2006 ) . the model is iteratively trained with hundreds of positive examples and millions of hard - mined negative examples .
we keep original settings .
the hyperparameters are set to @xmath241 and @xmath242 to compensate for imbalanced training ( akbani et al . , 2004 ) . for experiments ( i)@xmath41(iv )
, we have the following settings : ( i ) @xmath243 , @xmath244 , @xmath245 , ( ii ) @xmath246 , @xmath247 , @xmath248 , ( iii ) @xmath249 , @xmath247 , @xmath248 , ( iv ) @xmath250 , @xmath251 , @xmath245 .
we set @xmath252 in all experiments , as in ( yang & ramanan , 2013 ) .
the results are shown in fig . 1 and fig
this work proposed an algorithm that explores the entire regularization path of asymmetric - cost linear support vector machines .
the events of data concurrently projecting onto the margin are usually considered as special cases when building one - dimensional regularization paths while they happen repeatedly in the two - dimensional setting . to this end
, we introduced the notion of joint events and illustrated the set update scheme with vertex loop property to efficiently exploit their occurrence during our iterative path exploration .
finally , as we structure the path into successive layers of sets , our algorithm has modest memory requirements and can be locally parallelized at each layer of the regularization path . finally , we posit that extending our algorithm to the entire @xmath2-dimensional regularization path would facilitate processing of further special cases .
hsieh , c. j. , chang , k. w. , lin , c. j. , keerthi , s. s. , sundararajan , s. ( 2008 ) . a dual coordinate descent method for large - scale linear svm . in _ proceedings of the 25th international conference on machine learning _
, 408 - 415 pedregosa , f. , varoquaux , g. , gramfort , a. , michel , v. , thirion , b. , grisel , o. , et al . ,
duchesnay , e. ( 2011 ) .
scikit - learn : machine learning in python .
_ the journal of machine learning research _ , 12 , 2825 - 2830 chu , b. y. , ho , c. h. , tsai , c. h. , lin , c. y. , lin , c. j. ( 2015 ) .
warm start for parameter selection of linear classifiers .
_ acm sigkdd international conference on knowledge discovery and data mining _ , 149 - 158 | we propose an algorithm for exploring the entire regularization path of asymmetric - cost linear support vector machines .
empirical evidence suggests the predictive power of support vector machines depends on the regularization parameters of the training algorithms .
the algorithms exploring the entire regularization paths have been proposed for single - cost support vector machines thereby providing the complete knowledge on the behavior of the trained model over the hyperparameter space .
considering the problem in two - dimensional hyperparameter space though enables our algorithm to maintain greater flexibility in dealing with special cases and sheds light on problems encountered by algorithms building the paths in one - dimensional spaces .
we demonstrate two - dimensional regularization paths for linear support vector machines that we train on synthetic and real data . |
the search for one or more higgs bosons is the central task of the cern large hadron collider ( lhc ) , with @xmath1 tev and a luminosity of 100 @xmath2 per year . in the standard model ( sm ) ,
the higgs boson mass is a free parameter with an upper bound of @xmath3 800gev @xcite . beyond the sm , the minimal supersymmetric standard model ( mssm ) ,
whose higgs sector is a special case of the two higgs doublet model ( 2hdm ) @xcite , is of particular theoretical interest , and contains five physical higgs bosons : two neutral cp - even bosons @xmath4 and @xmath5 , one neutral cp - odd boson @xmath6 , and two charged bosons @xmath7 .
the @xmath4 is the lightest , with a mass @xmath8 gev when including the radiative corrections @xcite , and is a sm - like higgs boson especially in the decoupling region ( @xmath9 ) .
the other four are non - sm - like ones , and the discovery of them may give the direct evidence of the mssm .
it has been shown in @xcite that the @xmath4 boson of mssm can not escape detection at the cern large hadron collider ( lhc ) and that more than one neutral higgs particle can be found in large area of the supersymmetry ( susy ) parameter space at the lhc , the neutral higgs bosons can be produced through following mechanisms : gluon fusion @xmath10 @xcite , weak boson fusion @xmath11 @xcite , associated production with weak bosons @xcite , associated production with a heavy quark - antiquark pair @xmath12 @xcite and pairs production @xcite .
studying the associated production process of a neutral higgs boson and a vector boson at future hadron colliders may be an interesting way in searching for neutral higgs bosons , since the total cross section may be large and also the leptonic decay of the vector boson can be used as a spectacular event trigger . in the sm ,
the process @xmath13 has been studied both at the leading order ( lo ) @xcite and the next - to - leading order ( nlo ) @xcite in qcd . in the 2dhm and mssm , the associated production of @xmath14 and
@xmath15 has been studied only at tree level for drell - yan process and at one - loop level for gluon fusion in @xcite and @xcite , respectively .
it was shown in ref .
@xcite that the @xmath15 associated production rate at the lhc strongly depends on the susy parameters @xmath16 ( the ratio of two vacuum expectation values ) and @xmath17 ( the mass of @xmath6 ) .
the total cross section increases with increment of @xmath16 , and decreases with increment of @xmath17 . in this paper
, we present the complete nlo qcd , including supersymmetric qcd , calculation for the cross section of the associated production of @xmath15 through @xmath18 annihilation process at the lhc .
for simplicity , in our calculation , we neglect the bottom quark mass except in the yukawa couplings .
such approximations are valid in all diagrams , in which the bottom quark appears as an initial state parton , according to the simplified aivazis - collins - olness - tung ( acot ) scheme @xcite . to regularize the ultraviolet ( uv ) , soft and collinear divergences , two regularization schemes are used in our calculations for cross check , i.e. the dimensional regularization ( dreg ) scheme @xcite ( with naive @xmath19 scheme @xcite ) and the dimensional reduction ( dred ) scheme @xcite , and their results
are compared .
the arrangement of this paper is as follows . in sect .
ii , we show the lo results and define the notations . in sect .
iii , we present the details of the calculations of both the virtual and real parts of the nlo qcd corrections , and compare the results in dreg with those in dred . in sect .
iv , by a detailed numerical analysis , we present the predictions for the inclusive and differential cross sections of the @xmath15 associated production at the lhc .
v contains a brief conclusion . for completeness
, the relevant feynman rules are collected in appendix a , and the lengthy analytic expressions of the result of our calculation are summarized in appendices b and c.
the related feynman diagrams which contribute to the lo amplitude of the partonic process @xmath20 are shown in fig . 1 .
the lo amplitude in @xmath21 dimensions is @xmath22\nonumber\end{aligned}\ ] ] with @xmath23 where @xmath24 is the color tensor ( @xmath25 are color indices for the initial state quarks ) , @xmath26 is a mass parameter introduced to keep the couplings dimensionless , @xmath27 and @xmath28 are mandelstam variables , which are defined as @xmath29 @xmath30 , @xmath31 and @xmath32 denote the coefficients appearing in the relevant @xmath33 , @xmath34 and @xmath35 couplings , respectively , and their explicit expressions are given in appendix a. in order to simplify the expressions , we further introduce the following mandelstam variables : @xmath36 after the @xmath37-dimensional phase space integration , the lo partonic differential cross sections are given by @xmath38 \nonumber \\ & & \hspace{1.8 cm } \times \delta ( s+ t + u - m_{z^0}^2 -m_{a^0}^2 ) \overline{\sum}|m^b|^2\end{aligned}\ ] ] with @xmath39\bigg(\frac{a_1f_1}{s - m^2_{h^0}}+\frac{a_2f_2}{s - m^2_{h^0}}\bigg)^2\nonumber \\&&\hspace{2.0 cm } + a_3 ^ 2(|c_v|^2+|c_a|^2)\frac{2m^2_{z^0}(1-\epsilon)(tu - m^2_{z^0}m^2_{a^0})+st^2}{m^2_{z^0}t^2 } \nonumber\\&&\hspace{2.0cm}+a_3 ^ 2(|c_v|^2+|c_a|^2)\frac{2m^2_{z^0}(1-\epsilon)(tu - m^2_{z^0}m^2_{a^0})+su^2}{m^2_{z^0}u^2}\nonumber \\&&\hspace{2.0cm}+\frac{4a_3c_as(t+u)(tu - m^2_{z^0}m^2_{a^0})}{tum^2_{z^0}}\bigg[\frac{a_1f_1}{(s - m^2_{h^0 } ) } + \frac{a_2f_2}{(s - m^2_{h^0})}\bigg ] \nonumber\\&&\hspace{2.0cm}+2a_3 ^ 2(|c_v|^2-|c_a|^2)\frac{2(1-\epsilon)m^2_{z^0}(tu - m^2_{z^0}m^2_{a^0})+s(2m^2_{z^0}m^2_{a^0}-tu)}{m^2_{z^0}tu}\bigg\},\end{aligned}\ ] ] where @xmath40 and the @xmath41 function is the heaviside step function .
@xmath42 is the lo squared matrix element of @xmath43 , in which the colors and spins of the outgoing particles have been summed , and the colors and spins of the incoming ones have been averaged over .
the lo total cross section at the lhc is obtained by convoluting the partonic cross section with the parton distribution functions ( pdfs ) @xmath44 in the proton : @xmath45\hat{\sigma}^{b},\label{born0}\end{aligned}\ ] ] where @xmath46 is the factorization scale and @xmath47 is the born level constituent cross section of @xmath48 . obviously , the above lo results in the dreg scheme are equal to the ones in the dred scheme since the lo calculations are finite and free of any singularity .
the nlo contributions to the associated production of @xmath6 and @xmath49 can be separated into the virtual corrections arising from loop diagrams of colored particles and the real corrections arising from the radiation of a real gluon or a massless ( anti)quark . for both the virtual and real corrections , we will first present the results in the dreg scheme , and then compare them with the ones obtained in the dred scheme .
the virtual corrections to @xmath50 arise from the feynman diagrams shown in fig . 2 and fig .
they consist of self - energy , vertex and box diagrams , which represent the sm qcd corrections , arising from quarks and gluons , and supersymmetric qcd corrections , arising from squarks and gluinos .
we carried out the calculation in thooft - feynman gauge and used the dimensional regularization in @xmath51 dimensions to regularize the ultraviolet ( uv ) , soft and collinear divergences in the virtual loop corrections . in order to remove the uv divergences , we renormalize the bottom quark masses in the yukawa couplings and the wave function of bottom quark , adopting the on - shell renormalization scheme @xcite . denoting @xmath52 and @xmath53 as the bare bottom quark mass and the bare wave function , respectively , the relevant renormalization constants @xmath54 and @xmath55 are then defined as @xmath56 after calculating the self - energy diagrams in fig . 2 , we obtain the explicit expressions of all the renormalization constants as follows : @xmath57(m_b^2,m_{\tilde{g}}^2,m_{\tilde{b}_i}^2)\bigg\},\nonumber \\ & & \delta z_{bl}= -\frac{\alpha_s}{2\pi}c_f\sum_{i=1}^2(r^{\tilde b}_{i1})^2 ( b_0+b_1)(0,m_{\tilde{b}_i}^2,m_{\tilde{g}}^2),\nonumber \\ & & \delta z_{br}= -\frac{\alpha_s}{2\pi}c_f\sum_{i=1}^2(r^{\tilde b}_{i2})^2 ( b_0+b_1)(0,m_{\tilde{b}_i}^2,m_{\tilde{g}}^2),\nonumber\end{aligned}\ ] ] where @xmath58 , @xmath59 are the scalar two - point integrals @xcite , @xmath60 are the sbottom masses , @xmath61 is the gluino mass , and @xmath62 is a @xmath63 matrix shown as below , which is defined to transform the sbottom current eigenstates to the mass eigenstates @xcite : @xmath64 with @xmath65 , by convention .
correspondingly , the mass eigenvalues @xmath66 and @xmath67 ( with @xmath68 ) are given by @xmath69 with @xmath70 here , @xmath71 is the sbottom mass matrix .
@xmath72 and @xmath73 are soft susy breaking parameters and @xmath74 is the higgsino mass parameter . @xmath75 and @xmath76 are the third component of the weak isospin ( i.e. @xmath77 ) and the electric charge of the bottom quark @xmath78 ( i.e. @xmath79 ) , respectively . the renormalized virtual amplitudes can be written as @xmath80 here , @xmath81 contains the radiative corrections from the one - loop self - energy , vertex and box diagrams , as shown in fig . 2 , and @xmath82 is the corresponding counterterm .
moreover , @xmath81 can be separated into two parts : @xmath83 where @xmath84 and @xmath85 denote the corresponding diagram indexes in fig . 2 and fig . 3 , respectively .
they can be further expressed as @xmath86 where @xmath87 and @xmath88 are the form factors , which are given explicitly in appendix b , and the @xmath89 are the standard matrix elements defined as @xmath90 the counterterm contribution @xmath82 is separated into @xmath91 , @xmath92 and @xmath93 , i.e. the counterterms for s , t and u channels , respectively , which are given by @xmath94(m_1+m_2 ) \ , , \nonumber\\ & & m^{con(t)}=\frac{-ia_3}{t } \bigg[\delta m_b+\frac{m_b}{2}(\delta z_{bl}+\delta z_{br})\bigg][2(c_v - c_a)m_1 - 4c_vm_3\nonumber\\ & & \hspace{4.3cm}-(c_v - c_a)m_5 + 2c_vm_6 ] \
, , \nonumber \\ & & m^{con(u)}=\frac{ia_3}{u } \bigg[\delta m_b+\frac{m_b}{2}(\delta z_{bl}+\delta z_{br})\bigg][2(c_v+c_a)m_2 - 4c_vm_4\nonumber\\ & & \hspace{4.3cm}+(c_v+c_a)m_5 - 2c_vm_6 ] \ , .\nonumber\end{aligned}\ ] ] the @xmath95 virtual corrections to the differential cross section can be expressed as @xmath96 \nonumber \\ & & \hspace{1.8 cm } \times \delta ( s+ t + u - m_{a^0}^2 -m_{z^0}^2 ) \ 2 \ { \rm re } \bigg[\overline{\sum}{(m^v m^{b\ast})}\bigg],\end{aligned}\ ] ] where the renormalized amplitude @xmath97 is uv finite , but it still contains the infrared ( ir ) divergences : @xmath98 where @xmath99 here , the infrared divergences include the soft divergences and the collinear divergences .
the soft divergences are cancelled after adding the real emission corrections , and the remaining collinear divergences can be absorbed into the redefinition of pdf @xcite , which will be discussed in the following subsections .
note that the coefficients @xmath100 and @xmath101 of the infrared divergence terms are constants , similar to the drell - yan type processes .
needless to say that the susy qcd corrections do not generate infrared divergences , for squarks and gluinos are massive particles . in the above calculation , we have adopted the naive @xmath19 prescription in the dreg scheme to calculate the @xmath15 associated production rate . to cross check the above calculation
, we shall also adopt the dred scheme to carefully treat the @xmath19 factor in the amplitude calculation .
we shall show that the total inclusive rate is independent of the regularization scheme , though the individual contributions , from either virtual or real emission corrections , can be scheme - dependent .
in the dred scheme , @xmath102 and @xmath103 remain unchanged , however , @xmath104 is different , and @xmath105 similarly , the form factors are found to be different , and @xmath106 thus , it is easy to obtain the following relations from the above results : @xmath107 the feynman diagrams for the real gluon emission process @xmath108 are shown in fig . 4 . the phase space integration for the real gluon emission will produce infrared singularities , which can be either soft or collinear and can be conveniently isolated by slicing the phase space into different regions defined by suitable cut - offs .
in this paper , we use the two - cutoff phase space slicing method @xcite which introduces two small cut - offs to decompose the three - body phase space into three regions .
first , the phase space can be separated into two regions by an arbitrary small soft cut - off @xmath109 , according to whether the energy ( @xmath110 ) of the emitted gluon is soft , i.e. @xmath111 , or hard , i.e. @xmath112 .
correspondingly , the partonic real cross section can be written as @xmath113 where @xmath114 and @xmath115 are the contributions from the soft and hard regions , respectively .
@xmath114 contains all the soft divergences , which can be explicitly obtained after analytically integrating over the phase space of the emitted soft gluon .
second , in order to isolate the remaining collinear divergences from @xmath115 , we should introduce another arbitrary small cut - off , called collinear cut - off @xmath116 , to further split the hard gluon phase space into two regions , according to whether the mandelstam variables satisfy the collinear condition @xmath117 or not .
thus , we have @xmath118 where the hard collinear part @xmath119 contains the collinear divergences , which can be explicitly obtained after analytically integrating over the phase space of the emitted collinear gluon .
the hard non - collinear part @xmath120 is finite and can be numerically computed using standard monte - carlo integration techniques @xcite , and can be written in the form : @xmath121 here , @xmath122 is the hard non - collinear region of the three - body phase space . in the next two subsections , we will discuss in detail the soft and hard collinear gluon emission . in the soft limit
, i.e. when the energy of the emitted gluon is small , with @xmath123 , the matrix element squared @xmath124 for the process @xmath125 can be simply factorized into the born matrix element squared times an eikonal factor @xmath126 : @xmath127 where the eikonal factor @xmath126 is given by @xmath128 moreover , the phase space in the soft limit can also be factorized as @xmath129 where @xmath130 is the integration over the phase space of the soft gluon , which is given by @xcite @xmath131 hence , the parton level cross section in the soft region can be expressed as @xmath132 using the approach of ref .
@xcite , after analytically integrating over the soft gluon phase space , eq .
( [ soft ] ) becomes @xmath133 \bigg(\frac{a_2^s}{\epsilon^2 } + \frac{a_1^s}{\epsilon } + a_0^s\bigg)\end{aligned}\ ] ] with @xmath134 in the hard collinear region , i.e. @xmath112 and @xmath135 , the emitted hard gluon is collinear to one of the incoming partons .
as a consequence of the factorization theorems @xcite , the squared matrix element for @xmath136 can be factorized into the product of the born squared matrix element and the altarelli - parisi splitting function for @xmath137 @xcite , i.e. @xmath138 where @xmath139 denotes the fraction of incoming parton @xmath140 s momentum carried by parton @xmath140 with the emitted gluon taking a fraction @xmath141 , and @xmath142 are the unregulated splitting functions in @xmath21 dimensions for @xmath143 , which can be related to the usual altarelli - parisi splitting kernels @xcite as @xmath144 .
explicitly @xmath145 moreover , the three - body phase space can also be factorized in the collinear limit , and , for example , in the limit @xmath146 it has the following form @xcite : @xmath147^{-\epsilon}.\end{aligned}\ ] ] here , the two - body phase space should be evaluated at the squared parton - parton energy @xmath148 .
thus , the three - body cross section in the hard collinear region is given by @xcite @xmath149 ( -\frac{1}{\epsilon } ) \delta_c^{-\epsilon } \bigg[p_{bb}(z,\epsilon)g_{b / p}(x_1/z)g_{\bar{b}/p}(x_2 ) \nonumber \\ & & \hspace{1.4 cm } + p_{\bar{b}\bar{b } } ( z,\epsilon)g_{\bar{b}/p}(x_1/z ) g_{b / p}(x_2 ) + ( x_1\leftrightarrow x_2)\bigg ] \frac{dz}{z } ( \frac{1 -z}{z})^{-\epsilon } dx_1 dx_2,\end{aligned}\ ] ] where @xmath150 is the bare pdf .
in addition to the real gluon emission , a second set of real emission corrections to the inclusive production rate of @xmath151 at
the nlo involves the processes with an additional massless ( anti)quark in the final states : @xmath152 the relevant feynman diagrams for massless ( anti)quark emission ( the diagrams for the antiquark emission are similar and omitted here ) are shown in fig . 5 since the contributions from the real massless ( anti)quark emission contain the initial state collinear singularities , we also need to use the two - cutoff phase space slicing method @xcite to isolate those collinear divergences . because there is no soft divergence in the splitting of @xmath153
, we only need to separate the phase space into two regions : the collinear region and the hard non - collinear region .
thus , according to the approach shown in ref .
@xcite , the cross sections for the processes with an additional massless ( anti)quark in the final states can be expressed as @xmath154 dx_1dx_2 \nonumber \\ & & \hspace{1.0 cm } + \hat{\sigma}^b \bigg[\frac{\alpha_s}{2\pi } \frac{\gamma(1-\epsilon ) } { \gamma(1 - 2\epsilon ) } ( \frac{4\pi\mu^2_r}{s})^\epsilon\bigg ] ( -\frac{1}{\epsilon } ) \delta_c^{-\epsilon } \bigg[p_{bg}(z,\epsilon)g_{g / p}(x_1/z)g_{\bar{b}/p}(x_2 ) \nonumber \\ & & \hspace{1.0 cm } + g_{b / p}(x_1)p_{\bar{b}g}(z,\epsilon)g_{g / p}(x_2/z)+(x_1\leftrightarrow x_2)\bigg ] \frac{dz}{z } \bigg(\frac{1 -z}{z}\bigg)^{-\epsilon } dx_1 dx_2,\end{aligned}\ ] ] where @xmath155 , \hspace{2.0 cm } p_{bg}'(z)=p_{\bar{b}g}'(z)=-z(1-z).\label{pp2}\end{aligned}\ ] ] the first term in eq .
( [ accd ] ) represents the non - collinear cross sections for the two processes , which can be written in the form : @xmath156 where @xmath84 and @xmath85 denote the incoming partons in the partonic processes , and @xmath122 is the three body phase space in the non - collinear region . the second term in eq . ( [ accd ] ) represents the collinear singular cross sections . as mentioned above , after adding the renormalized virtual corrections and the real corrections , the partonic cross sections still contain the collinear divergences , which can be absorbed into the redefinition of the pdf at nlo , in general called mass factorization @xcite .
this procedure in practice means that first we convolute the partonic cross section with the bare pdf @xmath157 , and then rewrite @xmath157 in terms of the renormalized pdf @xmath158 in the numerical calculations . in the @xmath159 scheme ,
the scale dependent pdf @xmath158 is given by @xcite @xmath160 \int_x^1 \frac{dz}{z } p_{\alpha\beta } ( z ) g_{\beta
/ p}(x / z).\end{aligned}\ ] ] after replacing the bare pdf by the renormalized @xmath159 pdf and integrating out the collinear region of the phase space defined in the two - cutoff phase space slicing method @xcite , the resulting sum of eq . ( [ eq:40 ] ) and the collinear part ( the second term ) of eq .
( [ accd ] ) yields the remaining @xmath161 collinear contribution as @xcite : @xmath162 \{\tilde{g}_{b / p}(x_1,\mu_f ) g_{\bar{b}/p}(x_2,\mu_f ) + g_{b / p}(x_1,\mu_f ) \tilde{g}_{\bar{b}/p}(x_2,\mu_f ) \nonumber \\ & & \hspace{1.2 cm } + \sum_{\alpha = b,\bar{b}}\bigg[\frac{a_1^{sc}(\alpha\rightarrow \alpha g)}{\epsilon } + a_0^{sc}(\alpha\rightarrow \alpha g)\bigg]g_{b / p}(x_1,\mu_f ) g_{\bar{b}/p}(x_2,\mu_f ) \nonumber \\ & & \hspace{1.2 cm } + ( x_1\leftrightarrow x_2)\ } dx_1dx_2,\label{11}\end{aligned}\ ] ] where @xmath163 with @xmath164 the nlo total cross section for @xmath151 in the @xmath165 factorization scheme is obtained by summing up the born , virtual , soft , collinear and hard non - collinear contributions . in terms of the above notations ,
we have @xmath166(\hat{\sigma}^{b } + \hat{\sigma}^{v}+ \hat{\sigma}^{s } + \hat{\sigma}^{\overline{hc}})\ } + \sigma^{coll } \nonumber \\ & & \hspace{0.4 cm } + \sum_{(\alpha = g,\beta = b,\bar{b})}\int dx_1dx_2 \bigg[g_{\alpha / p}(x_1,\mu_f ) g_{\beta / p}(x_2,\mu_f ) + ( x_1\leftrightarrow x_2)\bigg ] \hat{\sigma}^{\overline{c}}(\alpha\beta\rightarrow a^0z^0 + x ) \ , .\label{t}\end{aligned}\ ] ] we note that the above expression contains no singularities , for @xmath167 and @xmath168 .
namely , all the @xmath169 and @xmath170 terms cancel in @xmath171 .
the apparent logarithmic @xmath109 and @xmath116 dependent terms also cancel with the the hard non - collinear cross section @xmath120 after numerically integrating over its relevant phase space volume . in the end of sec .
iii a , cf .
( [ v1])([0 ] ) , we discussed the results of virtual corrections in the dred scheme . here , we examine the real emission corrections and the nlo total cross section in the dred scheme and compare them with those obtained in the dreg scheme .
we find that the contributions from soft gluon emission remain the same , while the ones from hard collinear gluon emission and massless ( anti)quark emission are different due to the difference in the parton splitting functions and the perturbative pdfs . first , the splitting functions in the dred scheme contain no @xmath172 parts , so that @xmath173 thus , from eqs . ( [ 11 ] ) and ( [ 12 ] ) , we find the difference @xmath174 secondly , the perturbative pdfs defined in the dred and dreg schemes are different , and @xcite : @xmath175 after substituting them into the formula for calculating the born level cross sections , cf .
( [ born0 ] ) , we find the difference arising from the perturbative pdfs , at the @xmath176 level , as : @xmath177 except the upper limit of the integral over y , the two expressions in eqs .
( [ 1 ] ) and ( [ 2 ] ) are the same . after substituting eqs .
( [ 1 ] ) , ( [ 2 ] ) and ( [ 0 ] ) into eq . ( [ t ] )
, we find the relation between the two nlo total cross sections , separately calculated in the dreg and dred schemes , as follows : @xmath178 using the explicit expressions of the @xmath172 parts of the splitting functions @xmath179 , cf .
( [ pp1 ] ) and ( [ pp2 ] ) , we find @xmath180 as expected , both schemes yield the same nlo total cross sections , up to @xmath181 . in this subsection
, we present the differential cross section in the transverse momentum of @xmath49 and @xmath6 bosons , respectively , and the invariant mass of the @xmath15 pair . using the notations defined in ref .
@xcite , the differential distribution of the transverse momentum ( @xmath182 ) and rapidity ( @xmath183 ) of @xmath49 boson for the processes @xmath184\end{aligned}\ ] ] is given by @xmath185 where @xmath186 is the total center - of - mass energy of the collider , and @xmath187 with @xmath188 and @xmath189 .
the limits of integral over @xmath183 and @xmath182 are @xmath190 with @xmath191 the differential distribution with respect to @xmath182 and @xmath183 of @xmath6 is similar to the one of @xmath49 . the differential distribution with respect to the invariant mass @xmath192 is given by @xmath193 where @xmath194 is the parton luminosity , defined as : @xmath195,\end{aligned}\ ] ] with @xmath196
in the numerical calculations , we used the following set of sm parameters@xcite : @xmath197 the running qcd coupling @xmath198 is evaluated at the two - loop order @xcite , and the cteq6 m pdfs @xcite is used throughout this paper to calculate various cross sections , either at the lo or nlo .
as to the yukawa coupling of the bottom quark , we shall first use the @xmath159 bottom quark mass , @xmath199gev , to evaluate the event rate , then compare it with the one calculated using the qcd improved running mass to reduce the higher order qcd radiative corrections , therefore improve the perturbative calculations .
the qcd improved running mass @xmath200 , evaluated by the nlo formula @xcite , is : @xmath201 where the evolution factor @xmath202 is @xmath203 , \nonumber \\ d^{(f)}=\frac{12}{33 - 2f } , \hspace{1.0 cm } j^{(f)}=-\frac{8982 - 504f+40f^2}{3(33 - 2f)^2 } \
, , \end{aligned}\ ] ] and @xmath204 is the number of the active light quarks . for comparison
, we list the qcd improved running bottom quark mass in table [ tc ] for various energy scale @xmath205 . for large @xmath206 ,
the susy threshold correction to the bottom quark yukawa couplings could be large , and it can be resummed by making the following replacement in the tree - level couplings to improve the perturbation calculations @xcite : @xmath207 \label{deltamb1 } \ , , \end{aligned}\ ] ] where @xmath208 @xmath209 and @xmath210 and @xmath211 are the rotation matrices for defining the mass eigenstates of @xmath212 and @xmath213 , respectively .
we set @xmath214 in @xmath215 to @xmath61 in our numerical calculations .
needless to say that when using the running bottom quark yukawa coupling to evaluate cross sections , we shall subtract the corresponding ( susy-)qcd corrections at the order @xmath216 from the renormalization constant @xmath104 to avoid double counting in perturbative expansion of the strong coupling constant .
the values of the mssm parameters taken in our numerical calculations were constrained within the minimal supergravity scenario ( msugra ) @xcite , in which there are only five free input parameters at the grand unification ( gut ) scale .
they are @xmath217 and the sign of @xmath74 , where @xmath218 are , respectively , the universal gaugino mass , scalar mass and the trilinear soft breaking parameter in the superpotential . given those parameters ,
all the mssm parameters at the weak scale are determined in the msugra scenario by using the program package suspect 2.3 @xcite .
in particular , we used the running higgs masses at the @xmath219 scale , defined in the modified dimensional reduction ( @xmath220 ) scheme , which have included the full one - loop corrections , as well as the two - loop corrections controlled by the strong gauge coupling and the yukawa couplings of the third generation fermions @xcite . in our numerical calculations , we used the two - loop renormalization group equations ( rges ) presented in that program for calculating all the gauge couplings , the ( third generation ) yukawa couplings and the gaugino masses , while using one - loop rge for the other supersymmetric parameters . in the following
, we shall present our numerical studies based on the five sets of susy input parameters listed in table [ ta ] , which are consistent with all the existing experiment data @xcite .
we will also vary @xmath221 , @xmath222 and @xmath223 to examine their effects to various cross sections .
for completeness , we also show the relevant susy output parameters in table [ tb ] .
the qcd plus susy - qcd and susy - ew improved bottom quark running mass are listed in table[td ] , which should be compared with those given in table [ tc ] , in which only qcd running effect is included . for comparison ,
the qcd plus susy - qcd improved bottom quark running mass are separately listed in table[te ] .
.the qcd improved running bottom quark mass , evaluated at @xmath224 , and @xmath225gev .
the @xmath226 bottom quark mass is taken to be @xmath199gev . [
cols="^,^,^,^",options="header " , ] as for the renormalization and factorization scales , we always chose @xmath227 and @xmath228 , unless specified otherwise . in fig .
6 and fig .
7 , we first compare the lo total cross sections of @xmath151 via @xmath18 annihilation with the ones via gluon fusion and drell - yan processes , respectively . here
, we use the @xmath229 bottom quark mass @xmath199gev , without including the effect from qcd running .
our numerical results are different from the ones presented in ref .
@xcite , because the updated susy parameters are used instead of the earlier input parameters used in ref .
@xcite which have already been ruled out by recent experiments . as shown in figs . 6 and 7 , the lo total cross sections via @xmath18 annihilation and drell - yan processes increase with @xmath221 , while the ones via gluon fusion process are relatively larger for low and high values of @xmath221 , but become smaller for intermediate values of @xmath221 .
moreover , all the lo rates decrease when @xmath230 increases .
6 and 7 also show that in most of the chosen parameter range , @xmath18 contributions are much larger than the ones from gluon fusion and drell - yan processes , especially for large @xmath221 and small @xmath230 , where the total cross sections from the @xmath18 contributions can reach a few hundred fb . in fig .
8 , we show the dependence of the nlo qcd predictions on the two arbitrary theoretical cutoff scales @xmath109 and @xmath116 , introduced in the two - cutoff phase space slicing method , where we have set @xmath231 to simplify the study and used qcd plus susy improved bottom quark yukawa coupling .
the nlo total cross section can be separated into two classes of contributions .
one is the @xmath232 rate contributed by the born level , and the @xmath233 virtual , soft and hard collinear real emission corrections , denoted as @xmath47 , @xmath234 , @xmath114 and @xmath235 in eq .
( [ t ] ) .
another is the @xmath236 rate contributed by the @xmath233 hard non - collinear real emission corrections , denoted as @xmath120 and @xmath237 in eq .
( [ t ] ) .
as noted in the previous section , the @xmath238 and @xmath236 rates depend individually on @xmath109 and @xmath116 , but their sum should not depend on any of the theoretical cutoff scales .
this is clearly illustrated in fig .
8 for two different sets of susy parameters .
we find that @xmath239 is almost unchanged for @xmath109 between @xmath240 and @xmath241 , which is about 200fb and 28fb , respectively for the two different sets of susy parameters .
therefore , we take @xmath242 and @xmath231 in the numerical calculations below . in fig . 9
, we show the total cross sections of @xmath243 at the lhc as a function of @xmath230 for @xmath244 and @xmath245 , respectively , assuming @xmath246gev , @xmath247gev , and @xmath248 .
we considered the lo total cross sections in three different cases , i.e. using ( i ) @xmath159 bottom quark mass at the scale @xmath249 , ( ii ) qcd improved bottom quark running mass at the scale @xmath230 , and ( iii ) qcd plus susy improved bottom quark running mass at the scale @xmath230 , respectively .
we also considered the nlo total cross sections for the cases of ( ii ) and ( iii ) .
9 shows that the lo and nlo total cross sections get smaller with the increasing @xmath230 , and the results for @xmath244 in fig .
9(2 ) are much smaller than the ones for @xmath250 in fig .
9(1 ) . for small @xmath230 ( @xmath251gev ) the lo total cross sections in fig .
9(1 ) can be larger than @xmath252fb .
the contributions from the qcd running @xmath249 mass effects and the susy improved @xmath249 corrections are significant , for example , in fig .
9(1 ) when @xmath253gev and @xmath250 , the lo total cross sections are about 270fb , 120fb and 185fb for the three cases , respectively . moreover , fig .
9 shows that the nlo qcd corrections can either enhance or suppress the total rate , and the @xmath176 contribution is in general a few tens percent of the total rate , as described below .
define the k - factor as the ratio of the nlo to lo total cross sections , calculated using the cteq6 m pdfs .
we shown in fig .
10 the dependence of the k factor on @xmath230 for @xmath15 production , based on the results of case ( ii ) in fig .
namely , the qcd improved bottom quark running mass is used for calculating the total cross section at the lo and the nlo .
10 shows that in general the k factor becomes smaller with the increasing @xmath230 .
for example , the curve ( a ) in fig .
10(1 ) shows that when @xmath230 varies from 108gev to 900gev , the k factor varies from 1.72 to 0.82 , and the curve ( a ) in fig .
10(2 ) shows that when @xmath230 varies from 235gev to 860gev , the k factor varies from 0.91 to 0.68 .
the contributions to the k factors , shown as curve ( a ) , in both fig .
10(1 ) and fig .
10(2 ) come from the pure qcd corrections , shown as curve ( b ) , and susy qcd corrections , shown as curve ( c ) .
the former includes both the virtual and real emission contributions originated from pure qcd corrections , while the latter consists of only virtual corrections .
as expected , the @xmath254-factor contributed by the pure qcd corrections is under controlled , of a few tens percent , when the qcd improved bottom quark running mass is used to evaluate the yukawa coupling of the bottom quark . on the other hand
, the susy qcd corrections could become large as @xmath230 decreases , especially for large @xmath221 .
for example , in fig .
10(1 ) , for @xmath250 , when @xmath255gev , the @xmath254 factor of susy qcd corrections is about 0.8 which dominates the overall k factor . hence , to improve the convergence of the perturbation calculations in the case of large @xmath221 , we could use the susy improved bottom quark running mass to evaluate the yukawa coupling of bottom quark .
more on susy qcd corrections will be discussed below .
we have also examined the contributions from the box diagrams shown in fig .
the pure qcd box diagram contribution , arising from fig .
3(a ) and fig .
3(c ) , is ultraviolet finite but not infrared finite . for @xmath250
the finite part of the pure qcd box diagram contribution becomes more important for large @xmath230 , and its effect is to decreases the total rate . on the contrary , the susy qcd box diagram contribution , arising from fig .
3(b ) and fig .
3(d ) , is free of any singularity , and is small numerically .
11 shows the dependence of the k factors on @xmath230 for @xmath15 production , based on the results of case ( iii ) in fig .
namely , the qcd plus susy improved bottom quark running yukawa coupling is used for calculating the total cross section at the lo and the nlo .
generally , the k factor decreases with @xmath230 .
for example , for @xmath250 , when @xmath230 varies from 108gev to 900gev , the k factor corresponding to curve ( a ) ranges from 0.98 to 0.61 , which contains two parts : the pure qcd corrections , shown as curve ( b ) , and susy qcd corrections , shown as curve ( c ) . as compared to the results in fig .
10(1 ) , we find that the susy qcd correction , shown as curve ( c ) , has been largely suppressed .
for instance , the k factor of susy qcd corrections drops from 0.8 , in fig .
10(1 ) , to 0.05 , in fig . 11 , for @xmath255gev , while the other susy parameters are identical in both calculations .
this is because using the susy improved running @xmath249 to evaluate the lo cross section , we have already included the dominant nlo susy qcd corrections .
therefore , we shall use the qcd plus susy improved bottom quark running mass in the following numerical analysis for both the lo and nlo calculations , unless specified otherwise .
it is instructive to examine the results of figs .
10(1 ) and fig .
11 in the heavy mass limit , where all the susy mass parameters except @xmath230 are of the same size and tend to be heavy , i.e. @xmath256 @xmath257 @xmath258 @xmath257 @xmath74 @xmath257 @xmath259 @xmath257 @xmath260 @xmath261 @xmath262 . in the heavy mass limit , the susy qcd box diagram contribution , arising from fig . 3(b ) and
3(d ) , is suppressed by powers of @xmath214 and can be neglected .
this is confirmed by our numerical calculation which shows that the susy qcd box contribution is generally below @xmath263 of the total rate .
hence , we shall examine the effect of susy qcd corrections in the heavy mass limit to the virtual diagrams shown in figs .
2(a)(g ) , and compare the analytical result with our numerical calculations .
since our aim is to examine the nlo susy qcd effect in this part of study , we shall use the lo bottom quark yukawa coupling ( with @xmath264gev ) to evaluate the relevant tree level vertices . keeping only terms at @xmath176 that are not suppressed by negative powers of heavy mass @xmath214 in the heavy mass limit , the one loop susy qcd correction to the individual diagram in fig . 2 yields the following corrections .
after stripping off the born level matrix element ( including all the vertex and propagator factors ) , the multiplicative factor of the s - channel diagram with the @xmath4 propagator , cf .
2(a ) , is given by @xmath265 where @xmath84 is the mixing angle of the two cp - even higgs bosons @xcite .
note that eq .
( [ haber ] ) is in agreement with the one shown in ref .
similarly , the multiplicative factor of the s - channel diagram with the @xmath5 propagator , cf .
2(a ) , is given by @xmath266 the multiplicative factor of the t and u - channel diagrams , cf .
2(c ) or ( d ) , is given by @xmath267 the multiplicative factor for the sum of figs .
2(b ) and 2(g ) is zero .
this is because after adding the wavefunction renormalization factor for the external bottom quark line , the renormalized @xmath268 vertex vanishes in the heavy mass limit .
( again , we have dropped any term that is suppressed by negative powers of the heavy mass scale @xmath214 . ) similarly , the multiplicative factor for the sum of figs .
2(e ) and 2(f ) is zero . given the above multiplicative factors , we can calculate the susy qcd correction to the total cross section for @xmath243 production at the lhc , and compare it with the complete numerical calculation described in sec .
the results in the heavy mass limit are shown in fig .
12 , which show that the agreement becomes better for larger value of @xmath214 .
hence , this provides a consistent check on our complete numerical calculations . in figs .
13 ( 1 ) and 13 ( 2 ) , the total cross sections for @xmath151 at the lhc are plotted as a function of @xmath221 for two representative values of @xmath269 and @xmath222 , respectively . in fig .
13(2 ) , when @xmath221 ranges between 4 and 40 , @xmath230 varies from 330gev to 223gev , and from 660gev to 458gev for @xmath270gev and @xmath271gev , respectively . from fig .
13(2 ) we can clearly see that the lo and nlo total cross sections are enhanced with the increasing @xmath221 and decreased with the increasing @xmath269 . for large @xmath221 ( @xmath272 ) and @xmath270gev
, the lo and nlo total cross sections can be over 30fb .
the features in fig .
13(1 ) are similar to the ones in fig .
13(2 ) , but in general the total cross sections are larger than later .
for example , for large @xmath273 and @xmath274gev , both of the lo and nlo total cross sections can reach about hundreds of fb .
14 shows the dependence of the k factors on @xmath221 , based on the results in fig . 13 , where the k factor increases with the increasing @xmath221 .
for the results of fig .
13(1 ) , the k factor varies from 0.69 to 0.90 and from 0.65 to 0.92 for @xmath275gev and @xmath271gev , respectively . for the results of fig .
13(2 ) , the k factor varies from 0.70 to 0.74 and from 0.62 to 0.63 for @xmath270gev and @xmath271gev , respectively .
[ scale ] shows the dependence of the total cross sections for @xmath151 production at the lhc on the renormalization scale ( @xmath26 ) and the factorization scale ( @xmath46 ) , with @xmath277 .
the case ( 1 ) is for @xmath248 , and the case ( 2 ) is for @xmath278 . in both cases ,
the scale dependence of the nlo total cross section is smaller than that of the lo cross section . for example , the lo cross sections vary from 65fb to 261fb and 25fb to 45fb when @xmath277 ranges between @xmath279 and @xmath280 , while the nlo ones vary from 230fb to 232fb and 38fb to 39fb , in the case ( 1 ) and ( 2 ) , respectively . here , the qcd plus susy improved bottom quark yukawa coupling is used . for comparison
, we also show the results of other two calculations .
the case ( 3 ) is similar to the case ( 1 ) , but in ( 3 ) the pure qcd running bottom quark mass is used instead .
the case ( 4 ) is similar to the case ( 1 ) , but in ( 4 ) the contribution from the susy - ew correction in the running bottom quark yukawa coupling is not included , namely , only the pure qcd and suysy - qcd corrections are included . to further investigate the scale dependence in case ( 1 ) , with @xmath281 , we study the scale dependence of the total cross section on the renormalization scale @xmath282 and the factorization scale @xmath283 seperatedly in fig .
[ scale2 ] . here , the qcd plus susy improved bottom quark yukawa coupling is used .
we find that in either case , whether we fixed @xmath26 and let @xmath46 vary , or vice versa , the nlo rate is less dependent on the scale than the lo rate . hence , when applying the usual prescription to estimate the scale dependence , i.e. varying the scale around @xmath284 by a factor of 2 , the nlo cross sections vary by around 10% to 20% , cf . figs .
[ scale ] and [ scale2 ] , to estimate the uncertainties in the total cross sections due to the uncertainty of pdfs , we take the 41 sets of cteq6.1 pdfs to calculate the lo and nlo rates @xcite . as shown in fig .
[ cteq ] , the lo result of using the cteq6 m pdf lies between the maximum ( @xmath285 ) and minimum ( @xmath286 ) lo rates .
the nlo total cross sections are then calculated using three different pdf sets , one of which is cteq6 m , the other two are the ones that give the maximum and minimum lo rates , respectively .
the total cross sections for @xmath151 production at the lhc , as a function of the trilinear coupling @xmath223 , for the above mentioned pdfs are shown in fig .
[ cteq ] , where we have used the qcd running mass to evaluate the bottom quark yukawa coupling .
it turns out that the pdf uncertainties ( defined here as @xmath287 ) in the lo and nlo total cross sections are about the same , when the qcd running @xmath249 is used .
for example , when @xmath247gev , the pdf uncertainties are @xmath288 at the lo , and @xmath289 at the nlo , respectively .
[ pdfu ] shows the pdf uncertainties ( defined here as the eq .
( 3 ) in ref .
@xcite ) in the lo and nlo total cross sections for @xmath151 production at the lhc , as a function of @xmath230 . here
, we also used the qcd running mass to evaluate the bottom quark yukawa coupling .
it turns out that the nlo rate has a slightly larger uncertainty than the lo rate due to the pdf uncertainties , especially at large @xmath230 .
also , the uncertainty in the total cross section at the lhc increases as @xmath230 increases . fig .
[ pt1 ] shows the differential cross section as a function of the transverse momentum @xmath182 of @xmath49 and @xmath6 in the associated production of the @xmath15 pairs at the lhc .
we find that the nlo qcd correction could change the shape of transverse momentum distribution .
the nlo qcd correction enhances the lo differential cross section in low and high @xmath182 region , but reduces in medium @xmath182 region .
[ pt2 ] shows the differential cross section as a function of the invariant mass @xmath290 of the @xmath15 pairs produced at the lhc .
the nlo qcd corrections reduce the lo differential cross sections more in the medium values of @xmath290 , and much less in low or high values of @xmath290 .
in conclusion , we have calculated the complete nlo qcd corrections to the inclusive total cross sections of the @xmath15 pairs produced at the lhc in the mssm .
we have preformed the calculations using both the dreg and dred schemes , and found that the nlo total cross sections in the above two schemes are the same , which provides a cross check to our calculations .
our results show that the lo total cross sections are a few tens fb in most of the susy parameter space , and can exceed 100fb for @xmath230 below 160gev with large @xmath221(@xmath291 ) . the nlo correction can either enhance or reduce the total cross sections , but it generally efficiently reduces the dependence of the total cross sections on the renormalization / factorization scale . for small @xmath230 and large @xmath221 ,
the @xmath254-factor of susy qcd corrections could become large , and using the qcd plus susy improved yukawa coupling in the calculation could reduce the size of the overall @xmath254-factor .
we have also examined the uncertainty in total cross sections due to the pdf uncertainties , and found that the uncertainty in nlo cross sections is slightly larger than that in lo ones , especially at large @xmath230 .
finally , we also examined a few differential distributions and found that the nlo qcd corrections could change the shape of transverse momentum and invariant mass distributions .
we thank qing - hong cao and tao han for useful discussion .
this work was supported in part by the national natural science foundation of china and specialized research fund for the doctoral program of higher education .
the work of cpy was supported in part by the usa nsf grant phy-0244919 .
in this appendix , we give the relevant feynman rules .
@xmath292 @xmath293 where @xmath84 is the mixing angle in the cp even neutral higgs boson sector @xcite . here
, we use the abbreviations @xmath294 and @xmath295 .
@xmath296 @xmath297 + 3 .
@xmath298 @xmath299 here , we define the outgoing four - momenta of @xmath300 and @xmath6 to be negative and positive , respectively .
@xmath301 @xmath302 5 .
@xmath303_{\alpha\beta}$ ] @xmath304\\ \frac{1}{\sqrt{2}}h_b \left[a_b s_\alpha + \mu c_\alpha \right ] & \frac{gm_{z}}{\cos\theta_w}(-\frac{1}{3}\sin\theta^2_w)\sin(\alpha+\beta)+\sqrt{2}m_bh_bs_\alpha \end{array } \right),\nonumber\end{aligned}\ ] ] @xmath305 \\ -\frac{1}{\sqrt{2}}h_b\left[a_b c_\alpha -\mu s_\alpha \right ] & \frac{gm_{z}}{\cos\theta_w}(\frac{1}{3}\sin\theta^2_w)\cos(\alpha+\beta)-\sqrt{2}m_bh_b c_\alpha \end{array } \right),\nonumber\end{aligned}\ ] ] @xmath306 with @xmath307 .
@xmath308 @xmath309 where @xmath310 and @xmath311 are the four - momenta of @xmath312 and @xmath313 in direction of the charge flow .
in this appendix , we collect the explicit expressions of the nonzero form factors in eq .
( [ fvertex ] ) and eq .
( [ fbox ] ) . since @xmath314 ,
only the form factors of the first six matrix elements are presented here . for simplicity ,
we introduce the following abbreviations for the passarino - veltman three - point integrals @xmath315 and four - point integrals @xmath316 , which are defined similar to ref . @xcite except that we take internal masses squared as arguments : @xmath317 , + @xmath318 , + @xmath319 , + @xmath320 , + @xmath321 , + @xmath322 , + @xmath323 , + @xmath324 , + @xmath325 , + @xmath326 , + @xmath327 , + @xmath328 , + @xmath329 , + @xmath330 , + @xmath331 , + @xmath332 , + @xmath333 , + @xmath334 , + @xmath335 , + @xmath336 , + @xmath337 , + @xmath338 , + @xmath339 , + @xmath340 , + @xmath341 , + @xmath342 , + @xmath343 , + @xmath344 , + @xmath345 , + @xmath346 , + @xmath347 , + @xmath348 . +
many of the above functions contain the soft and/or collinear singularities .
since all the passarino - veltman integrals can be written as a combination of the scalar functions @xmath223 , @xmath349 , @xmath350 and @xmath351 , we present here the explicit expressions for the @xmath350 and @xmath351 functions used in our calculations : @xmath352,\nonumber\\ & & c^d_0=c^h_0=c^j_0=\frac{c_\epsilon}{u - m^2_{a^0}}\bigg[\frac{1}{\epsilon}\ln\bigg(\frac{-u}{m^2_{a^0}}\bigg ) + \frac{1}{2}\ln^2\bigg(\frac{s}{m^2_{a^0}}\bigg)-\frac{1}{2}\ln^2\bigg(\frac{s}{-u}\bigg)-\frac{\pi^2}{2}\bigg],\nonumber\\ & & c^c_0=c^g_0=c^l_0=c^v_0=\frac{c_\epsilon}{u - m_z^2}\bigg[\frac{1}{\epsilon}\ln\bigg(\frac{-u}{m^2_z}\bigg ) + \frac{1}{2}\ln^2\bigg(\frac{s}{m_z^2}\bigg)-\frac{1}{2}\ln^2\bigg(\frac{s}{-u}\bigg)-\frac{\pi^2}{2}\bigg],\nonumber\\ & & c^e_0=c^f_0=c^i_0=c^y_0=\frac{c_\epsilon}{t - m_z^2}\bigg[\frac{1}{\epsilon}\ln\bigg(\frac{-t}{m^2_z}\bigg ) + \frac{1}{2}\ln^2\bigg(\frac{s}{m_z^2}\bigg)-\frac{1}{2}\ln^2(\frac{s}{-t})-\frac{\pi^2}{2}\bigg ] , \nonumber\\ & & c^b_0=c^k_0=c^x_0=\frac{c_\epsilon}{t - m^2_{a^0}}\bigg[\frac{1}{\epsilon}\ln\bigg(\frac{-t}{m^2_{a^0}}\bigg ) + \frac{1}{2}\ln^2\bigg(\frac{s}{m^2_{a^0}}\bigg)-\frac{1}{2}\ln^2\bigg(\frac{s}{-t}\bigg)-\frac{\pi^2}{2}\bigg],\nonumber\\ & & d^a_0=d^b_0= \frac{c_\epsilon}{st}\bigg[\frac{1}{\epsilon^2}+\frac{2}{\epsilon}\ln\bigg(\frac{m_zm_{a^0}}{-t}\bigg)+\frac{\pi^2}{3}\bigg ] -2\frac{c_\epsilon}{st}\bigg \{{\rm li}\bigg(\frac{m^2_{a^0}-u}{s}\bigg ) -{\rm li}\bigg(\frac{s - m_z^2}{s}\bigg ) \nonumber\\ & & \ \ \ \ \ \ \ -{\rm li}\bigg[\frac{-st}{(s - m_z^2)(m_z^2-t)}\bigg ] + { \rm li}\bigg(\frac{-t}{s - m_z^2}\bigg)+{\rm li}\bigg(\frac{m^2_{a^0}}{m^2_{a^0}-t}\bigg ) -\frac{1}{2}\ln^2\bigg[\frac{-st}{(s - m_z^2)(m_z^2-t)}\bigg]\nonumber\\ & & \ \ \ \ \ \ \ + \frac{1}{2}\ln^2\bigg(\frac{-t}{s - m_z^2}\bigg ) + \ln\bigg(\frac{m_z^2-t}{s}\bigg)\ln\bigg(\frac{m^2_{a^0}-u}{t}\bigg ) -\frac{1}{2}\ln\bigg(\frac{m_z^2-t}{s}\bigg)\ln\bigg(\frac{sm^2_{a^0}}{t^2}\bigg)\nonumber\\ & & \ \ \ \ \ \ \ + \frac{1}{4}\ln^2\bigg(\frac{sm^2_{a^0}}{t^2}\bigg ) + \frac{1}{4}\ln^2\bigg(\frac{m^2_{z}}{s}\bigg ) + \frac{1}{2}\ln\bigg(\frac{m^2_{a^0}}{s}\bigg)\ln\bigg(\frac{m_z^2-t}{m^2_{z}}\bigg ) + \frac{1}{2}\ln^2\bigg(\frac{m^2_{a^0}}{m^2_{a^0}-t}\bigg ) \bigg \},\nonumber \\ & & d^c_0=d^d_0=d^a_0(t\leftrightarrow u , m^2_{a^0}\leftrightarrow m_z^2),\nonumber \end{aligned}\ ] ] where @xmath353 .
+ for diagrams(a)-(g ) in fig . 2
, we get the form factors as following , respectively , @xmath354c_0^a\nonumber\\ & & \hspace{0.6cm}+f_1m_{\tilde g}(s - m^2_{h_0})\hat{g}^{\tilde b}_1(a , b)r^{\tilde b}_{a,1}r^{\tilde b}_{b,2}c^m_0 + f_2m_{\tilde g}(s - m^2_{h_0})\hat{g}^{\tilde b}_2(a , b)r^{\tilde b}_{a,1}r^{\tilde b}_{b,2}c^m_0\}\nonumber \\&&\hspace{0.6cm}+\frac{4\alpha_s\epsilon}{3\pi(s - m^2_{h_0})(s - m^2_{h_0})}[a_1f_1(s - m^2_{h_0 } ) + a_2f_2(s - m^2_{h_0})]b_0(s,0,0),\nonumber \\ & & f^a_2=f^a_1,\nonumber \\ & & f^a_3=\frac{4m_{\tilde g}\alpha_s(r^{\tilde b}_{a,2}r^{\tilde b}_{b,1}-r^{\tilde b}_{a,1}r^{\tilde b}_{b,2})}{3\pi(s - m^2_{h_0})(s - m^2_{h_0})}[f_1\hat{g}^{\tilde b}_1(a , b)(s - m^2_{h_0})+f_2\hat{g}^{\tilde b}_2(a , b)(s - m^2_{h_0})]c^m_0,\nonumber \\ & & f^a_4=f^a_3,\nonumber \\ & & f^b_1=\frac{4im_ba_3(c_v - c_a)\alpha_s}{3\pi t}[(1-\epsilon)(2c^f_{00}-uc^f_{12}+(t - m_{a_0}^2)c^f_{11}+m^2_zc^e_1 , \nonumber\\ & & \hspace{0.9cm}-(s - m^2_z - m^2_{a_0})c^f_{12}+m^2_{a_0}c^f_{11}-b_0(t,0,0))+m_z^2c^i_0+tc^f_1 ] \nonumber\\ & & \hspace{0.9cm}-\frac{4m_bg\alpha_sa_3t_{z}(a , b)r^{\tilde b}_{a1}r^{\tilde b}_{b1}}{3\pi tcos\theta_{w } } [ 2c^q_{00}+uc^q_2+uc^q_{22}+(t - m^2_{a_0})(c^q_1+c^q_{11}+c^q_2+c^q_{22})\nonumber\\ & & \hspace{0.9 cm } + ( s - m^2_z - m^2_{a_0})(c^q_2+c^q_{22})+m^2_{a_0}(c^q_1+c^q_{11}+c^q_2+c^q_{22})+tc^q_{12 } ] , \nonumber \\&&f^b_3=\frac{-4im_bc_va_3\alpha_s}{3\pi t}\{(1-\epsilon)[2c^f_{00}-uc^f_{12}+(t - m_{a_0}^2)c^f_{11}+m^2_zc^e_1 \nonumber\\ & & \hspace{0.9cm}-(s - m^2_z - m^2_{a_0})c^f_{12}+m^2_{a_0}c^f_{11}-b_0(t,0,0)]+m_z^2c^i_0+tc^f_1\ } \nonumber\\ & & \hspace{0.9cm}+\frac{4m_bg\alpha_sa_3t_{z}(a , b)r^{\tilde b}_{a1}r^{\tilde b}_{b1}}{3\pi tcos\theta_{w } } [ 2c^q_{00}+uc^q_2+uc^q_{22}+(t - m^2_{a_0})(c^q_1+c^q_{11}+c^q_2+c^q_{22})\nonumber\\ & & \hspace{0.9 cm } + ( s - m^2_z - m^2_{a_0})(c^q_2+c^q_{22})+m^2_{a_0}(c^q_1+c^q_{11}+c^q_2+c^q_{22})+tc^q_{12 } ] , \nonumber \\&&f^b_5=\frac{-2im_b\alpha_sa_3(c_v - c_a)}{3\pi t}\{(1-\epsilon)[2c^f_{00}+(t - m^2_{a_0})c^f_1+m_z^2c^e_1\nonumber\\ & & \hspace{0.9cm}+(s - m^2_z - m^2_{a_0})(c^e_1-c^f_2 ) + m^2_{a_0}(2c^f_1-c^e_2)-b_0(t,0,0)]+m_z^2c^i_0\}\nonumber\\ & & \hspace{0.9cm}+\frac{2gm_b\alpha_sa_3t_{z}(ab)r^{\tilde b}_{a1}r^{\tilde b}_{b1}}{3\pi cos\theta_{w } t}c^q_{00},\nonumber \\&&f^b_6=\frac{4im_b\alpha_sc_va_3}{3\pi t}\{(1-\epsilon)[2c^f_{00}+(t - m^2_{a_0})c^f_1+m_z^2c^e_1\nonumber\\ & & \hspace{0.9cm}+(s - m^2_z - m^2_{a_0})(c^e_1-c^f_2 ) + m^2_{a_0}(2c^f_1-c^e_2)-b_0(t,0,0)]+m_z^2c^i_0\}\nonumber\\ & & \hspace{0.9cm}+\frac{-4gm_b\alpha_sa_3t_{z}(a , b)r^{\tilde b}_{a1}r^{\tilde b}_{b1}}{3\pi cos\theta_{w } t}c^q_{00},\nonumber \\ & & f^c_1=\epsilon\frac{-4im_ba_3\alpha_s}{3\pi } ( c_v+c_a)(c^d_1-c^h_2),\nonumber \\ & & f^c_2=\frac{-4im_ba_3\alpha_s}{3\pi u}(c_v+c_a)[m^2_{a_0}c^j_0-b_0(u,0,0)]+\frac{4m_{\tilde{g}}\alpha_s}{3\pi u}\hat{g}^{\tilde b } _ 3(a , b)r^{\tilde b}_{a1}r^{\tilde b}_{b2}c^s_0,\nonumber\\ & & \hspace{0.9cm}+\epsilon\frac{-4im_ba_3\alpha_s}{3\pi u}(c_v+c_a)[uc^d_1+uc^d_2-uc^h_1-uc^h_2+(u - m^2_z)c^h_1\nonumber\\ & & \hspace{0.9cm}-m_z^2(c^d_2 - 2c^h_1 ) + ( s - m_z^2-m_{a^0}^2)(c^d_1-c^h_2)+m_{a^0}^2c^d_1+b_0(u,0,0)],\nonumber \\ & & f^c_3=\epsilon\frac{8im_ba_3\alpha_s}{3\pi } c_v(c^d_1-c^h_2),\nonumber \\ & & f^c_4=\frac{8im_ba_3\alpha_sc_v}{3\pi u}[m^2_{a_0}c^j_0-b_0(u,0,0)]\nonumber\\ & & \hspace{0.9cm}+\frac{4m_{\tilde{g}}\hat{g}^{\tilde b } _ 3(a , b)\alpha_s } { 3\pi u}[r^{\tilde b}_{a2}r^{\tilde b}_{b1}(c_v - c_a)c^s_0-r^{\tilde b}_{a1}r^{\tilde b}_{b2}(c_v+c_a)c^s_0]\nonumber\\ & & \hspace{0.9cm}+\epsilon\frac{8im_ba_3\alpha_s}{3\pi u}c_v[uc^d_1+uc^d_2-uc^h_1-uc^h_2+(u - m^2_z)c^h_1\nonumber\\ & & \hspace{0.9cm}-m_z^2(c^d_2 - 2c^h_1 ) + ( s - m_z^2-m_{a^0}^2)(c^d_1-c^h_2)+m_{a^0}^2c^d_1+b_0(u,0,0)],\nonumber \\ & & f^c_5=\frac{-2im_ba_3\alpha_s}{3\pi u}(c_v+c_a)[m^2_{a_0}c^j_0-b_0(u,0,0)]+\frac{4m_{\tilde{g}}\alpha_s}{3\pi u}\hat{g}^{\tilde b } _ 3(a , b)r^{\tilde b}_{a1}r^{\tilde b}_{b2}c^s_0\nonumber\\ & & \hspace{0.9cm}+\epsilon\frac{-2im_ba_3\alpha_s}{3\pi u}(c_v+c_a)[uc^h_1+(s - m_z^2-m_{a^0}^2)(c^d_1-c^h_2)\nonumber\\ & & \hspace{0.9cm}-m^2_zc^d_2+m_{a^0}^2c^d_1+b_0(u,0,0 ) ] , \nonumber \\ & & f^c_6=\frac{4im_ba_3\alpha_sc_v}{3\pi u}[m^2_{a_0}c^j_0-b_0(u,0,0)]\nonumber\\ & & \hspace{0.9cm}+\frac{4m_{\tilde{g}}\hat{g}^{\tilde b } _ 3(a ,
b)\alpha_s } { 3\pi u}[r^{\tilde b}_{a2}r^{\tilde b}_{b1}(c_v - c_a)c^s_0-r^{\tilde b}_{a1}r^{\tilde b}_{b2}(c_v+c_a)c^s_0]\nonumber\\ & & \hspace{0.9cm}+\epsilon\frac{4im_ba_3\alpha_s}{3\pi u}c_v[uc^h_1+(s - m_z^2-m_{a^0}^2)(c^d_1-c^h_2)\nonumber\\ & & \hspace{0.9cm}-m^2_zc^d_2+m_{a^0}^2c^d_1+b_0(u,0,0 ) ] , \nonumber \\ & & f^d_1=\frac{4i(c_v - c_a)m_ba_3\alpha_s}{3\pi t}[c^k_0m^2_{a_0}-b_0(t,0,0)]+\frac{-4(c_v - c_a)m_{\tilde g } \alpha_s\hat{g}^{\tilde b } _ 3(a , b)r^{\tilde b}_{a1}r^{\tilde b}_{b2}}{3\pi t}c^t_0\nonumber\\ & & \hspace{0.9cm}+\epsilon\frac{4i(c_v - c_a)m_ba_3\alpha_s}{3\pi t}[c^b_1m^2_{a_0}+b_0(t,0,0)],\nonumber \\ & & f^d_3=\frac{-8ic_vm_ba_3\alpha_s}{3\pi t}[c^k_0m^2_{a_0}-b_0(t,0,0 ) ] + \frac{-4(c_v+c_a)m_{\tilde g}\alpha_s\hat{g}^{\tilde b } _ 3(a , b)r^{\tilde b}_{a2}r^{\tilde b}_{b1}}{3\pi t}c^t_0\nonumber\\ & & \hspace{0.9cm}+\frac{4(c_a - c_v)m_{\tilde g}\alpha_s\hat{g}^{\tilde b } _ 3(a , b)r^{\tilde b}_{a1}r^{\tilde b}_{b2}}{3\pi t}c^t_0\nonumber\\ & & \hspace{0.9cm}+\epsilon\frac{-8ic_vm_ba_3\alpha_s}{3\pi t}[c^b_1m^2_{a_0}+b_0(t,0,0)],\nonumber\\ & & f^d_5=\frac{-f^d_1}{2},\nonumber \\ & & f^d_6=\frac{-f^d_3}{2},\nonumber \\ & & f^e_2=\frac{-4i(c_v+c_a)m_ba_3\alpha_s}{3\pi u}[2c^g_{00}+uc^g_{11}+uc^g_{12}+m^2_z(c^c_1+c^l_0)]\nonumber\\ & & \hspace{0.9cm}+\frac{4gm_ba_3\alpha_s}{3\pi ucos\theta_{w}}t_{z}(a , b)r^{\tilde b}_{a2}r^{\tilde b}_{b2}(2c^o_{00}-uc^o_{12}),\nonumber\\ & & f^e_4=\frac{8ic_vm_ba_3\alpha_s}{3\pi u}[2c^g_{00}+uc^g_{11}+uc^g_{12}+m^2_z(c^c_1+c^l_0)]\nonumber\\ & & \hspace{0.9cm}+\frac{-4gm_ba_3\alpha_s}{3\pi ucos\theta_{w}}t_{z}(a , b)(2c^o_{00}-uc^o_{12})(r^{\tilde b}_{a1}r^{\tilde b}_{b1}+r^{\tilde b}_{a2}r^{\tilde b}_{b2}),\nonumber\\ & & f^e_5=\frac{-2i(c_v+c_a)m_ba_3\alpha_s}{3\pi u}[2c^g_{00}+uc^g_{11}+uc^g_{12}+m^2_z(c^c_1+c^l_0)]\nonumber\\ & & \hspace{0.9cm}+\frac{4gm_ba_3\alpha_s}{3\pi ucos\theta_{w}}t_{z}(a , b)r^{\tilde b}_{a2}r^{\tilde b}_{b2}c^o_{00},\nonumber \\ & & f^e_6=\frac{4ic_vm_ba_3\alpha_s}{3\pi u}[2c^g_{00}+m^2_z(c^c_1+c^l_0 ) ] + \frac{-4gm_ba_3\alpha_s}{3\pi ucos\theta_{w}}t_{z}(a , b)c^o_{00}(r^{\tilde b}_{a1}r^{\tilde b}_{b1}+r^{\tilde b}_{a2}r^{\tilde b}_{b2}),\nonumber \\ & & f^f_1=\frac{2i(c_v - c_a)m_ba_3\alpha_s}{3\pi t^2}{r^{\tilde b } _ { a1}}^2[-2m^2_{\tilde g}b_0(0,m^2_{\tilde g},m^2_{\tilde g } ) + m^2_{\tilde g}b_0(t , m^2_{\tilde g},m^2_{{\tilde b}_a})-2m^2_{\tilde g}\nonumber\\ & & \hspace{0.9cm}+ 2m^2_{{\tilde b}_a}b_0(0,m^2_{{\tilde b}_a},m^2_{{\tilde b}_a})-m^2_{{\tilde b}_a}b_0(t , m^2_{{\tilde g}},m^2_{{\tilde b}_a } ) -2m^2_{{\tilde b}_a}+tb_0(t , m^2_{{\tilde g}},m^2_{{\tilde b}_a})\nonumber\\ & & \hspace{0.9cm}+ 2m^2_{{\tilde b}_a}b_0(0,m^2_{{\tilde g}},m^2_{{\tilde b}_a})]+\frac{2i(c_v - c_a)m_ba_3\alpha_s}{3\pi t^2}(1-\epsilon)tb_0(t,0,0),\nonumber \\ & & f^f_3=\frac{-2im_ba_3\alpha_s}{3\pi t^2}[(c_v - c_a){r^{\tilde b } _
{ a1}}^2+(c_v+c_a){r^{\tilde b } _
{ a2}}^2][-2m^2_{\tilde g}b_0(0,m^2_{\tilde g},m^2_{\tilde g } ) + m^2_{\tilde g}b_0(t , m^2_{\tilde g},m^2_{{\tilde b}_a})\nonumber\\ & & \hspace{0.9cm}+2m^2_{{\tilde b}_a}b_0(0,m^2_{{\tilde b}_a},m^2_{{\tilde b}_a})-2m^2_{\tilde g}-m^2_{{\tilde b}_a}b_0(t , m^2_{{\tilde g}},m^2_{{\tilde b}_a } ) -2m^2_{{\tilde b}_a}+tb_0(t , m^2_{{\tilde g}},m^2_{{\tilde
b}_a } ) \nonumber\\&&\hspace{0.9cm}+ 2m^2_{{\tilde b}_a}b_0(0,m^2_{{\tilde g}},m^2_{{\tilde b}_a } ) ] + \frac{-2im_ba_3c_v\alpha_s}{3\pi t^2}(1-\epsilon)tb_0(t,0,0),\nonumber \\ & & f^g_2=\frac{-2i(c_v+c_a)m_ba_3\alpha_s}{3\pi t^2}{r^{\tilde b } _
{ a2}}^2[-2m^2_{\tilde g}b_0(0,m^2_{\tilde g},m^2_{\tilde g } ) + m^2_{\tilde g}b_0(t , m^2_{\tilde g},m^2_{{\tilde b}_a})+ 2m^2_{{\tilde b}_a}b_0(0,m^2_{{\tilde b}_a},m^2_{{\tilde b}_a})\nonumber\\ & & \hspace{0.9cm}-2m^2_{\tilde g}-m^2_{{\tilde b}_a}b_0(u , m^2_{{\tilde g}},m^2_{{\tilde b}_a } ) -2m^2_{{\tilde b}_a}+ub_0(u , m^2_{{\tilde g}},m^2_{{\tilde b}_a})+ 2m^2_{{\tilde b}_a}b_0(0,m^2_{{\tilde g}},m^2_{{\tilde b}_a } ) ] \nonumber\\ & & \hspace{0.9cm}+\frac{-2i(c_v+c_a)m_ba_3\alpha_s}{3\pi t^2}(1-\epsilon)ub_0(u,0,0),\nonumber \\ & & f^g_4=\frac{2im_ba_3\alpha_s}{3\pi u^2}[(c_v - c_a){r^{\tilde b } _ { a1}}^2+(c_v+c_a){r^{\tilde b } _ { a2}}^2][-2m^2_{\tilde g}b_0(0,m^2_{\tilde g},m^2_{\tilde g } ) + m^2_{\tilde g}b_0(u , m^2_{\tilde g},m^2_{{\tilde b}_a})\nonumber\\ & & \hspace{0.9cm}+ 2m^2_{{\tilde b}_a}b_0(0,m^2_{{\tilde b}_a},m^2_{{\tilde b}_a})-2m^2_{\tilde g}-m^2_{{\tilde b}_a}b_0(u , m^2_{{\tilde g}},m^2_{{\tilde b}_a } ) -2m^2_{{\tilde b}_a}+ub_0(u , m^2_{{\tilde g}},m^2_{{\tilde b}_a})\nonumber\\&&\hspace{0.9cm}+ 2m^2_{{\tilde b}_a}b_0(0,m^2_{{\tilde g}},m^2_{{\tilde b}_a } ) ] + \frac{2im_ba_3c_v\alpha_s}{3\pi u^2}(1-\epsilon)ub_0(u,0,0),\nonumber \\ & & f^g_5=\frac{-f^g_4}{2},\nonumber \\ & & f^f_6=\frac{-f^g_2}{2}.\nonumber\end{aligned}\ ] ] for the box diagrams(a)-(d ) in fig . 3
, we find , respectively , @xmath355\nonumber\\ & & \hspace{1.6cm}+\epsilon\frac{-4im_ba_3(c_v - c_a)\alpha_s}{3\pi } [ c^e_2+c^u_1+(u - m^2_{a^0})(d^a_{11}+d^a_{13})-(t - m^2_{a^0})(d^b_2+d^b_3)\nonumber\\ & & \hspace{1.6 cm } -m^2_{a^0}(d^a_1+d^a_{13}+d^a_3+d^a_{33})],\nonumber\\ & & f^{box(a)}_2=\frac{4im_ba_3(c_v - c_a)\alpha_s}{3\pi } [ m^2_{a^0}(d^a_0+d^b_2)-(t - m^2_{a^0})(d^a_1+d^a_2+d^a_3)-(u - m^2_{a^0})d^b_3\nonumber\\ & & \hspace{1.6cm}-c^a_0+c^z_0-c^x_0 ] \nonumber\\ & & \hspace{1.6cm}+\epsilon\frac{4im_ba_3(c_v - c_a)\alpha_s}{3\pi } [ 2d^a_{00}-(u - m^2_{a^0})(d^a_{11}+d^a_{12}+d^a_{13}+d^b_3)\nonumber\\ & & \hspace{1.6cm}+m^2_{a^0 } ( d^a_{13}+d^a_{23}+d^a_{33}+d^b_2)],\nonumber\\ & & f^{box(a)}_3=\frac{-2c_vf^{box(a)}_1}{c_v - c_a},\nonumber\\ & & f^{box(a)}_4=\frac{-2c_vf^{box(a)}_2}{c_v - c_a},\nonumber\\ & & f^{box(a)}_5=\frac{2im_ba_3(c_v - c_a)\alpha_s}{3\pi } [ -sd^a_0+(u - t)d^a_0+c^y_0+c_0^z-2c_0^x]\nonumber\\ & & \hspace{1.6cm}+\epsilon\frac{2im_ba_3(c_v - c_a)\alpha_s}{3\pi } [ 2d^a_{00}-(t - m^2_{a^0})d^b_2-m^2_{a^0}d^a_3-c^x_0],\nonumber\\ & & f^{box(a)}_6=\frac{-2c_vf^{box(a)}_5}{c_v - c_a},\nonumber \\ & & f^{box(b)}_1=\frac{4igm_{\tilde g}\alpha_s}{3\pi cos\theta_{w } cos\theta_{w } } t_z(l , a){\hat g}_3^{\tilde b}(l , b)r^{\tilde b}_{a1}r^{\tilde b}_{b2}(d^e_1+d^e_3+d^f_0),\nonumber\\ & & f^{box(b)}_2=\frac{4igm_{\tilde g}\alpha_s}{3\pi cos\theta_{w } cos\theta_{w } } t_z(l , a){\hat g}_3^{\tilde b}(l , b)r^{\tilde b}_{a1}r^{\tilde b}_{b2}(d^e_1+d^e_2+d^e_3+d^f_0),\nonumber\\ & & f^{box(b)}_3=\frac{4igm_{\tilde g}\alpha_s}{3\pi cos\theta_{w } cos\theta_{w } } t_z(l , a){\hat g}_3^{\tilde b}(l , b)(r^{\tilde b}_{b1}-r^{\tilde b}_{a1}r^{\tilde b}_{b2})(d^e_1+d^e_3+d^f_0),\nonumber\\ & & f^{box(b)}_4=\frac{4igm_{\tilde g}\alpha_s}{3\pi cos\theta_{w } cos\theta_{w } } t_z(l , a){\hat g}_3^{\tilde b}(l , b)(r^{\tilde b}_{b1}-r^{\tilde b}_{a1}r^{\tilde b}_{b2})(d^e_1+d^e_2+d^e_3+d^f_0),\nonumber \\ & & f^{box(c)}_1=\frac{-4im_ba_3(c_v+c_a)\alpha_s}{3\pi}[-sd^c_1 + ( u - m^2_z)(d^d_3-d^c_1)-m^2_zd^d_2-c^v_0]\nonumber\\ & & \hspace{1.6cm}+\epsilon\frac{-4im_ba_3(c_v+c_a)\alpha_s}{3\pi } [ c^d_1-c^u_1-c^x_1-(t - m^2_z)d^c_{11}+(u - m^2_z)d^d_3\nonumber\\ & & \hspace{1.6cm}+m^2_z(d^c_1+d^c_{13})],\nonumber\\ & & f^{box(c)}_2=\frac{-4im_ba_3(c_v+c_a)\alpha_s}{3\pi}[c^v_1-sd^c_1-sd^c_2 -(t - m^2_z)d^d_3-(u - m^2_z)(d^c_1+d^c_2)\nonumber\\ & & \hspace{1.6cm}+m^2_z(d^c_0+d^d_2)-c^a_0]\nonumber\\ & & \hspace{1.6cm}+\epsilon\frac{-4im_ba_3(c_v+c_a)\alpha_s}{3\pi}[c^d_1+c^d_2-c^x_1 -(t - m^2_z)(d^c_{11}+d^c_{12}+d^d_{3})\nonumber\\ & & \hspace{1.6cm}+m^2_z(d^c_{13}+d^c_{23}+d^d_2)+c^v_0],\nonumber \\ & & f^{box(c)}_3=\frac{-2c_vf^{box(b)}_1}{c_v+c_a},\nonumber\\ & & f^{box(c)}_4=\frac{-2c_vf^{box(b)}_2}{c_v+c_a},\nonumber\\ & & f^{box(c)}_5=\frac{2im_ba_3(c_v+c_a)\alpha_s}{3\pi}[-sd^c_0+(t - u)d^d_2+c^l_0+c^v_0 - 2c^x_0]\nonumber\\ & & \hspace{1.6cm}+\epsilon\frac{2im_ba_3(c_v+c_a)\alpha_s}{3\pi}[2d^c_{00}-(u - m^2_{a^0})d^d_2-m^2_zd^c_{3}-c^x_0],\nonumber \\ & & f^{box(c)}_6=\frac{-2c_vf^{box(b)}_5}{c_v+c_a},\nonumber\\ & & f^{box(d)}_1=\frac{4igm_{\tilde g}\alpha_s}{3\pi cos\theta_{w } cos\theta_{w } } t_z(l , a){\hat g}_3^{\tilde b}(l , b)r^{\tilde b}_{a1}r^{\tilde b}_{b2}d^h_1,\nonumber\\ & & f^{box(d)}_2=\frac{4igm_{\tilde g}\alpha_s}{3\pi cos\theta_{w } cos\theta_{w } } t_z(l , a){\hat g}_3^{\tilde b}(l , b)r^{\tilde b}_{a1}r^{\tilde b}_{b2}(d^h_1+d^h_2),\nonumber\\ & & f^{box(d)}_3=\frac{4igm_{\tilde g}\alpha_s}{3\pi cos\theta_{w } cos\theta_{w } } t_z(l , a){\hat g}_3^{\tilde b}(l , b)(r^{\tilde b}_{b1}-r^{\tilde b}_{a1}r^{\tilde b}_{b2})d^h_1,\nonumber\\ & & f^{box(b)}_4=\frac{4igm_{\tilde g}\alpha_s}{3\pi cos\theta_{w } } t_z(l , a){\hat g}_3^{\tilde b}(l , b)(r^{\tilde b}_{a2}r^{\tilde b}_{b1}-r^{\tilde b}_{a1}r^{\tilde b}_{b2})(d^h_1+d^h_2).\nonumber\end{aligned}\ ] ] m. spira et al . ,
b318 , 347(1993 ) ; nucl .
b453 , 17(1995 ) ; s. dawson et al .
77 , 16(1996 ) ; robert v. harlander and matthias steinhauser , phys .
b574 , 258(2003 ) , phys .
d68 , 111701(2003 ) , j. high energy phys .
0409 , 066(2004 ) ; a. djouadi and m. spira , phys .
d62 , 014004(2000 ) .
r. v. harlander and w. kilgore , phys .
lett.88 , 201801(2002 ) ; j. high energy phys .
0210 , 017(2002 ) ; c. anastasiou and k. melnikov , nucl .
b646 , 220(2002 ) ; phys .
d67 , 037501(2003 ) ; v. ravindran et al .
b665 , 325(2003 ) . r. n. cahn and s. dawson , phys .
b136 , 196(1984 ) ; g. altarelli et al .
b287 , 205(1987 ) ; t. han et al .
lett.69 , 3274(1992 ) .
s. glashow et al . , phys .
d18 , 1724(1978 ) .
r. kleiss , z. kunszt and w. j. stirling , phys .
b253 , 269(1991 ) .
t. han and s. willenbrock , phys .
b273 , 167(1991 ) .
z. kunszt , nucl .
b247 , 339(1984 ) ; w. beenakker et al .
lett.87 , 201805(2001 ) ; nucl .
b653 , 151(2003 ) ; s. dawson et al .
lett.87 , 201804(2001 ) ; phys .
d67 , 071503(2003 ) ; c. balazs , et al . , phys .
d59 , 055016(1999 ) .
a. a. barrientos bendez and b. a. kniehl , phys .
d64 , 035006(2001 ) .
j. ohnemus et al . , phys .
d47 , 2722(1993 ) ; h.baer et al .
d47 , 2730(1993 ) .
l. l. yang et al .
, j. phys .
g30 , 1821(2004 ) .
y. jun et al .
d66 , 095008(2002 ) .
chung kao and shankar sachithanandam , .
chung kao , phys .
d46 4907(1992 ) , fsu - hep-911205 . m. a. aivazis et al . ,
d50 , 3102(1994 ) ; j. c. collins , phys .
d58 , 094002(1998 ) ; m. krmer et al . ,
d62 , 096007(2000 ) .
g. t hooft and m. j. g. veltman , nucl .
b44 , 189(1972 ) .
m. chanowitz et al . , nucl .
b159 , 225(1979 ) .
z. bern et al . , phys .
d66 , 085002(2002 ) .
a. sirlin , phys .
d22 , 971(1980 ) ; w. j. marciano and a. sirlin , phys .
d22 , 2695(1980 ) ; phys .
d31 , 213(1985)(e ) ; a. sirlin and w. j. marciano , nucl .
b189 , 442(1981 ) ; k. i. aoki et al .
suppl.73 , 1(1982 ) .
a. denner , fortschr .
41 , 307(1993 ) .
s. kraml , phd thesis , ; j. ellis and s. rudaz , phys .
b128 , 248(1993 ) .
g. altarelli et al . , nucl .
b157 , 461(1979 ) ; j. c. collins et al . , in : perturbative quantum chromodynamics , ed .
a. h. mueller ( world scientific , 1989 ) .
b. w. harris and j. f. owens , phys . rev .
d65 , 094032(2002 ) .
g. p. lepage , j. comp .
27 , 192(1978 ) .
j. c. collins et al .
b261 , 104(1985 ) ; g. t. bodwin , phys .
d31 , 2616(1985 ) ; phys .
d34 , 3932(1986)(e ) .
g. altarelli and g. parisi , nucl .
b126 , 298(1977 ) .
r. k. ellis et al .
b178 , 421(1981 ) ; l. j. bergmann , in : next - to - leading - log qcd calculation of symmetric dihadron production ( ph.d .
thesis , florida state university , 1989 ) ; z. kunszt and d. e. soper , phys .
d46 192(1992 ) ; m. l. mangano et al .
b73 , 295(1992 ) .
b. kamal , phys .
d53 , 1142(1996 ) .
w. beenakker et al . , nucl .
b492 , 51(1997 ) .
particle data group , s. eidelman , et al .
b592 1(2004 ) .
s. g. gorishny et al . ,
a5 , 2703(1990 ) ; phys . rev .
d43 , 1633(1991 ) ; a. djouadi et al .
, z. phys .
c70 , 427(1996 ) ; a. djouadi et al .
108 , 56(1998 ) ; m. spira , fortschr .
46 , 203(1998 ) .
j. pumplin et al . , j. high energy phys . 0207 , 012(2002 ) .
m. carena et al . , nucl .
b577 , 88(2000 ) .
m. beneke and a. signer , phys .
b471 , 233(1999 ) ; a. h. hoang , phys .
d61 , 034005(2000 ) .
m. drees and s. p. martin , mad - ph-879,um - th-95 - 02 , a. djouadi et al . , hep - ph/0211331 . b. c. allanach et al . ,
j. high energy phys .
0409 , 044(2004 ) .
tilman plehn , phys .
d67 , 014018(2003 ) .
howard e. haber et al .
d63 , 055004(2001 ) .
d. stump et al .
, j. high energy phys .
0310 , 046(2003 ) .
j. pumplin et al . , j. high energy phys . 0207 , 012(2002 ) . | we present the calculations of the complete next - to - leading order ( nlo ) qcd corrections ( including supersymmetric qcd ) to the inclusive total cross sections of the associated production processes @xmath0 in the minimal supersymmetric standard model at the cern large hadron collider .
both the dimensional regularization scheme and the dimensional reduction scheme are used to organize the calculations which yield the same nlo rates .
the nlo correction can either enhance or reduce the total cross sections , but it generally efficiently reduces the dependence of the total cross sections on the renormalization / factorization scale .
we also examine the uncertainty of the total cross sections due to the parton distribution function uncertainties . |
the absence of a gunn - peterson trough in the majority of the observed absorption spectra towards high - redshift quasars suggests that the reionisation of intergalactic hydrogen was completed at a redshift @xmath8 ( see @xcite for a recent review ) and that it remained highly ionised ever since .
current observational estimates of the ultra - violet ( uv ) luminosity density at redshifts @xmath9 ( e.g. @xcite ; @xcite ; @xcite ; @xcite ; @xcite ; @xcite ; @xcite ; @xcite ; @xcite ; @xcite ) , on the other hand , may imply star formation rate ( sfr ) densities several times lower than the critical sfr density required to keep the intergalactic medium ( igm ) ionised ( but see , e.g. , @xcite ; @xcite ; @xcite ) .
taken at face value , these low sfr densities pose a severe challenge to commonly employed theoretical models in which the observed population of star - forming galaxies is the only source of ionising radiation in the high - redshift universe .
there are , however , large uncertainties associated with both the observationally inferred ( see , e.g. , the comprehensive analysis of @xcite ) and the critical sfr densities . the critical sfr density , @xmath10 here rescaled to match the most recent wmap estimate for the cosmic baryon density ( @xcite ) , has been derived by @xcite using an early version of the @xcite population synthesis code , assuming a salpeter initial stellar mass function ( imf ) and solar metallicity .
it results from simply equating the spatially averaged rate at which ionising photons are emitted into the igm to the spatially averaged rate at which the intergalactic gas recombines .
[ eq : criticalsfr ] is therefore incapable of addressing a number of potentially important physical effects .
some ionising photons will , for instance , redshift below the ionisation threshold before ionising and some ionising photons will have been emitted longer than a recombination time ago upon impact with a neutral atom , so that equating instantaneous rates is not appropriate and one even may have to take source evolution into account .
it is therefore important to keep in mind that eq .
[ eq : criticalsfr ] is likely only accurate within factors of a few .
the critical sfr is inversely proportional to the escape fraction @xmath11 , i.e. the fraction of ionising photons produced by star - forming galaxies that escape the interstellar medium ( ism ) to ionise the igm , and proportional to the average recombination rate in the igm .
the latter is parametrised using the dimensionless clumping factor @xmath12 , where @xmath1 is the baryon density , @xmath13 is the mean baryon density of the universe and the brackets @xmath3 indicate spatial averaging over the gas constituting the recombining igm . under the assumption of a uniformly ionised igm ,
the clumping factor expresses the spatially averaged number of recombinations occurring per unit time and unit volume in the ionised igm , relative to that in gas at the cosmic mean density @xmath13 .
a larger escape fraction implies a smaller critical sfr density , as more photons are available to ionise the igm . on the other hand , a larger clumping factor implies a larger critical sfr density since more ionising photons are required to compensate for the increased number of recombinations .
most observational studies that compare ( @xcite ) . ]
the sfr density derived from estimates of the uv luminosity density at redshift @xmath6 to the critical sfr density assume an escape fraction @xmath14 and a clumping factor @xmath15 .
while a variety of both observational and theoretical studies ( e.g. @xcite and references therein ; @xcite ; @xcite ) have ruled out larger escape fractions , the estimate for the clumping factor comes from a single cosmological simulation performed more than 10 years ago ( @xcite ) .
it is on the basis of these values for the escape fraction and the clumping factor that the observed population of galaxies has been found to be incapable of keeping the intergalactic hydrogen ionised , forming massive stars at a rate which is up to an order of magnitude lower than required by eq .
[ eq : criticalsfr ] .
it has been pointed out that this discrepancy between the inferred and critical sfr densities could be resolved if the employed clumping factor were too high ( e.g. @xcite ; see also the discussion in @xcite ) . indeed , in most ( but not all ) of the more recent theoretical theoretical studies ( e.g. @xcite ; @xcite ; @xcite ; @xcite ; @xcite ; @xcite ; @xcite ; @xcite ; @xcite ; @xcite ; @xcite ; @xcite ) significantly lower clumping factors were derived .
on the other hand , it is sometimes emphasised that simulations underestimate the clumping factor , due to a lack of resolution ( see , e.g. , @xcite ) . in this work
we perform a set of cosmological smoothed particle hydrodynamics ( sph ) simulations that include radiative cooling and photo - ionisation by a uniform uv background in the optically thin limit to study the clumping factor of the igm .
we focus on the effect of photo - ionisation heating on the evolution of the clumping factor .
previous investigations of the impact of photo - heating on the reionisation of the igm have almost exclusively come to the conclusion that it acts as to provide a negative feedback .
photo - heating boils the gas out of the potential wells of dark matter ( dm ) halos with virial temperatures @xmath16 ( e.g. @xcite ; @xcite ; @xcite ; @xcite ; @xcite ; @xcite ; @xcite ; @xcite ; @xcite ; @xcite ; @xcite ; @xcite ) .
this inhibits the formation of stars in these low - mass halos and thus decreases the ionising emissivity , which makes it more difficult to reionise the universe .
the same mechanism that reduces the number of ionising photons that are emitted into the igm does , however , also affect the evolution of the clumping factor ( e.g. @xcite ; @xcite ; @xcite ; @xcite ; @xcite ; @xcite ) . in this paper
we demonstrate that photo - heating significantly lowers the clumping factor and hence the average recombination rate in the igm .
while photo - ionisation heating undoubtedly impedes the production of ionising photons , our results imply that it also makes it much easier to keep the igm ionised .
the paper is structured as follows . in section [ sec : simulations ] we give a detailed description of our set of simulations . in section [ sec : results ] we use our simulations to compute the clumping factor of the igm . finally , in section [ sec : conclusions ]
, we discuss our results and their implications for the value of the critical sfr density .
[ cols="^,^,^,^,^,^,^,^,^ " , ] @xcite provided a convenient four - parameter fit to the volume - weighted probability density function ( pdf ) of the gas density at redshifts @xmath17 and @xmath18 obtained from the l10 hydrodynamical simulation described in @xcite .
in addition , they provided a prescription for extrapolating this fit to higher redshifts .
here we compare the volume - weighted pdf obtained from our default simulation ( _ r9l6n256 _ ) at @xmath19 to their fit . because we find that their set of fitting parameters provides a somewhat poor description of this pdf
, we also provide an updated set of parameters that yields a fit which more accurately describes the one - point distribution of gas densities at @xmath19 in our default simulation . in fig .
[ fig : pdf : fit ] we show the volume - weighted pdf of the gas density per unit @xmath20 .
we have investigated the convergence of this pdf with respect to the resolution and the size of the simulation box , using the same set of simulations that we employed to study the convergence of the clumping factor in section [ sec : convergence ] .
we find that the pdf is converged with respect to changes in the resolution .
it is converged with respect to changes in the size of the simulation box for overdensities @xmath21 . for larger overdensities ,
the values of the pdf obtained from our default simulation may not yet be fully converged .
we note that at @xmath19 the pdfs obtained from the simulations _
r[@xmath22l6n256z _ that employ a reheating redshift @xmath23 are almost undistinguishable from that obtained from our default simulation .
we compare the pdf to the fit given by @xcite , @xmath24 \delta^{-\beta } d\delta , \label{eq : pdf : fit}\ ] ] using their values for the parameters . ]
@xmath25 and @xmath26 .
as can be seen from fig .
[ fig : pdf : fit ] , for overdensities @xmath27 the pdf obtained from our default simulation is significantly steeper than predicted by the fit . using the simulations _
r9l6n256w3 _ and _ r9l6n256w1 _ , we verified that the steepness of the slope of the pdf over this range of overdensities is sensitive to the cosmological parameters employed .
note that the pdf does also not asymptote to @xmath28 , as predicted by eq .
[ eq : pdf : fit ] . for overdensities larger than the overdensity for the onset of star formation , the pdf is instead governed by the effective equation of state characteristic for the multi - phase star - forming gas .
we fitted eq .
[ eq : pdf : fit ] to our pdf over the range @xmath29 , constraining the parameters @xmath30 and @xmath31 by the requirement that the total integral over the volume- and mass - weighted pdf must be normalized to unity .
this yields values @xmath32 and @xmath33 .
note that the fit based on these parameter values still does not provide a good description of the pdf for overdensities @xmath34 .
( per unit @xmath20 ) at @xmath19 obtained from our default simulation _ r9l6n256 _
( black solid histogram ) . for comparison ,
the red solid line shows the four - parameter fit given by eq .
[ eq : pdf : fit ] , with the parameter values taken from @xcite .
the blue dashed line shows our best fit of eq .
[ eq : pdf : fit ] to the pdf .
bottom panel : absolute value of the relative differences of the fits with respect to the pdf . in both panels ,
the vertical line indicates the overdensity corresponding to the onset of star formation .
, scaledwidth=49.0% ] | the critical star formation rate density required to keep the intergalactic hydrogen ionised depends crucially on the average rate of recombinations in the intergalactic medium ( igm ) .
this rate is proportional to the clumping factor @xmath0 , where @xmath1 and @xmath2 are the local and cosmic mean baryon density , respectively and the brackets @xmath3 indicate spatial averaging over the recombining gas in the igm .
we perform a suite of cosmological smoothed particle hydrodynamics simulations that include radiative cooling to calculate the volume - weighted clumping factor of the igm at redshifts @xmath4 .
we focus on the effect of photo - ionisation heating by a uniform ultra - violet background and find that photo - heating strongly reduces the clumping factor because the increased pressure support smoothes out small - scale density fluctuations .
photo - ionisation heating is often said to provide a negative feedback on the reionisation of the igm because it suppresses the cosmic star formation rate by boiling the gas out of low - mass halos .
however , because of the reduction of the clumping factor it also makes it easier to keep the igm ionised .
photo - heating therefore also provides a positive feedback which , while known to exist , has received much less attention . we demonstrate that this positive feedback is in fact very strong .
using conservative assumptions , we find that if the igm was reheated at @xmath5 , the observed population of star - forming galaxies at @xmath6 may be sufficient to keep the igm ionised , provided that the fraction of ionising photons that escape the star - forming regions to ionise the igm is larger than @xmath7 .
[ firstpage ] cosmology : theory - methods : numerical - hydrodynamics - radiative transfer - intergalactic medium - galaxies : formation |
* sle and its conformal markov property . * oded schramm s sle processes introduced in @xcite have deeply changed the way mathematicians and physicists understand critical phenomena in two dimensions .
recall that a chordal sle is a random non - self - traversing curve in a simply connected domain , joining two prescribed boundary points of the domain .
modulo conformal invariance hypotheses that have been proved to hold in several cases , the scaling limit of an interface that appears in various two - dimensional models from statistical physics , when boundary conditions are chosen in a particular way , is one of these sle curves .
for instance , in the ising model on a triangular lattice , if one connected arc @xmath9 of the boundary of a simply connected region @xmath10 is forced to contain only @xmath11 spins whereas the complementary arc @xmath12 contains only @xmath13 spins , then there is a random interface that separates the cluster of @xmath11 spins attached to @xmath9 from the cluster of @xmath13 spins attached to @xmath12 ; this random curve has recently been proved by chelkak and smirnov to converge in distribution to an sle curve ( sle@xmath14 ) when one lets the mesh of the lattice go to zero ( and chooses the critical temperature of the ising model ) @xcite .
note that sle describes the law of one particular interface , not the joint law of all interfaces ( we will come back to this issue later ) . on the other hand , for a given model
, one expects all macroscopic interfaces to have similar geometric properties , i.e. , to locally look like an sle .
the construction of sle curves can be summarized as follows : the first observation , contained in schramm s original paper @xcite , is the `` analysis '' of the problem : assuming that the two - dimensional models of statistical physics have a conformally invariant scaling limit , what can be said about the scaling limit of the interfaces ?
if one chooses the boundary conditions in a suitable way , one can identify a special interface that joins two boundary points ( as in the ising model mentioned above ) .
schramm argues that if this curve has a scaling limit , and if its law is conformally invariant , then it should satisfy an `` exploration '' property in the scaling limit .
this property , combined with conformal invariance , implies that it can be defined via iterations of independent random conformal maps . with the help of loewner s theory for slit mappings ,
this leads naturally to the definition of the ( one parameter ) family of sle processes , which are random increasing families of compact sets ( called loewner chains ) , see @xcite for more details .
recall that loewner chains are constructed via continuous iterations of infinitesimal conformal perturbations of the identity , and they do not a priori necessarily correspond to actual planar curves .
a second step , essentially completed in @xcite , is to start from the definition of these sle processes as random loewner chains , and to prove that they indeed correspond to random two - dimensional curves .
this constructs a one - parameter family of sle random curves joining two boundary points of a domain , and the previous steps shows that if a random curve is conformally invariant ( in distribution ) and satisfies the exploration property , then it is necessarily one of these sle curves .
one can study various properties of these random loewner chains .
for instance , one can compute critical exponents such as in @xcite , determine their fractal dimension as in @xcite , derive special properties of certain sle s locality , restriction as in @xcite , relate them to discrete lattice models such as uniform spanning trees , percolation , the discrete gaussian free field or the ising model as in @xcite , or to the gaussian free field and its variants as in @xcite etc .
indeed , at this point the literature is far too large for us to properly survey here . for conditions that ensure that discrete interfaces converge to sle paths ,
see the recent contributions @xcite .
* conformal markov property for collections of loops . * a natural question is how to describe the `` entire '' scaling limit of the lattice - based model , and not only that of one particular interface . in the present paper , we will answer the following question : supposing that a discrete random system gives rise in its scaling limit to a conformally invariant collection of loops ( i.e. , interfaces ) that remain disjoint ( note that this is not always the case ; we will comment on this later ) , what can these random conformally invariant collections of loops be ?
more precisely , we will define and study random collections of loops that combine conformal invariance and a natural restriction property ( motivated by the fact that the discrete analog of this property trivially holds for the discrete models we have in mind ) .
we call such collections of loops conformal loop ensembles ( cle )
. the two main results of the present paper can be summarized as follows : [ mainresults ] * for each cle , there exists a value @xmath15 $ ] such that with probability one , all loops of the cle are sle@xmath1-type loops .
* conversely , for each @xmath15 $ ] , there exists exactly one cle with sle@xmath1 type loops .
in fact , these statements will be derived via two almost independent steps , that involve different techniques : 1 .
we first derive the first of these two statements together with the uniqueness part of the second one .
this will involve a detailed analysis of the cle property , and consequences about possible ways to `` explore '' a cle . here
, sle techniques will be important .
we derive the existence part of the second statement using clusters of poisson point processes of brownian loops ( the brownian loop - soups ) . in the end
, we will have two remarkably different explicit constructions of these conformal loop ensembles cle@xmath1 for each @xmath6 in @xmath16 $ ] ( one based on sle , one based on loop - soups ) .
this is useful , since many properties that seem very mysterious from one perspective are easy from the other .
for example , the ( expectation ) fractal dimensions of the individual loops and of the set of points not surrounded by any loop can be explicitly computed with sle tools @xcite , while many monotonicity results and fkg - type correlation inequalities are immediate from the loop - soup construction @xcite .
one illustration of the interplay between these two approaches is already present in this paper : one can use sle tools to determine exactly the value of the critical intensity that separates the two percolative phases of the brownian loop - soup ( and to our knowledge , this is the only self - similar percolation model where this critical value has been determined ) . in order to try to explain the logical construction of the proof ,
let us outline these two parts separately in the following two subsections .
let us now describe in more detail the results of the first part of the paper , corresponding to sections [ s.2 ] through [ s.8bis ] .
we are going to study random families @xmath17 of non - nested simple disjoint loops in simply connected domains .
for each simply connected @xmath10 , we let @xmath18 denote the law of this loop - ensemble in @xmath10 .
we say that this family is * conformally invariant * if for any two simply connected domains @xmath10 and @xmath19 ( that are not equal to the entire plane ) and conformal transformation @xmath20 , the image of @xmath18 under @xmath21 is @xmath22 .
we also make the following `` local finiteness '' assumption : if @xmath10 is equal to the unit disc @xmath23 , then for any @xmath24 , there are @xmath25 almost surely only finitely many loops of radius larger than @xmath26 in @xmath27 .
consider two simply connected domains @xmath28 , and sample a family @xmath29 according to the law @xmath30 in the larger domain @xmath31 .
then , we can subdivide the family @xmath32 into two parts : those that do not stay in @xmath33 , and those that stay in @xmath33 ( we call the latter @xmath34 ) .
let us now define @xmath35 to be the random set obtained when removing from the set @xmath33 all the loops of @xmath27 that do not fully stay in @xmath33 , together with their interiors .
we say that the family @xmath18 * satisfies restriction * if , for any such @xmath33 and @xmath31 , the conditional law of @xmath34 given @xmath35 is @xmath36 ( or more precisely , it is the product of @xmath18 for each connected component @xmath10 of @xmath35 ) .
when a family is conformally invariant and satisfies restriction , we say that it is a conformal loop ensemble ( cle ) .
the goal of the paper is to characterize and construct all possible cles .
( the grey wedge on the left of the right figure ) the conditional law of the remaining loops is an independent cle in each component of the ( interior of the ) complement of this set.,title="fig:",width=172 ] ( the grey wedge on the left of the right figure ) the conditional law of the remaining loops is an independent cle in each component of the ( interior of the ) complement of this set.,title="fig:",width=172 ] by conformal invariance , it is sufficient to describe @xmath18 for one given simply connected domain .
let us for instance consider @xmath10 to be the upper half - plane @xmath37 .
a first step in our analysis will be to prove that for all @xmath38 , if @xmath27 is a cle , then the conditional law of the unique loop @xmath39 that surrounds @xmath40 , conditionally on the fact that @xmath41 intersects the @xmath42-neighborhood of the origin , converges as @xmath43 to a probability measure @xmath44 on `` pinned loops '' , i.e. , loops in @xmath37 that touch the real line only at the origin .
we will derive various properties of @xmath44 , which will eventually enable us to relate it to sle .
one simple way to describe this relation is as follows : [ t1 ] if @xmath27 is a cle , then the measure @xmath44 exists for all @xmath38 , and it is equal to the limit when @xmath43 of the law of an sle@xmath1 from @xmath42 to @xmath45 in @xmath37 conditioned to disconnect @xmath40 from infinity in @xmath37 , for some @xmath15 $ ] ( we call this limit the sle@xmath1 * bubble measure * ) .
this shows that all the loops of a cle are indeed in some sense `` sle@xmath1 loops '' for some @xmath6 .
in fact , the way in which @xmath44 will be described ( and in which this theorem will actually be proved ) can be understood as follows ( this will be the content of proposition [ p44 ] ) in the case where @xmath46 : consider @xmath47 the lowest point on @xmath48 \cap \gamma ( i)$ ] , and @xmath49 the unbounded connected component of the domain obtained by removing from @xmath50 $ ] all the loops of the cle that intersect @xmath51 . consider the conformal map @xmath52 from @xmath49 onto @xmath37 with @xmath53 and @xmath54 .
then , the law of @xmath55 is exactly @xmath56 .
( sketch).,width=480 ] theorem [ t1 ] raises the question of whether two different cle distributions can correspond to the same measure @xmath44 .
we will prove that it is not possible , i.e. , we will describe a way to reconstruct the law of the cle out of the knowledge of @xmath44 only , using a construction based on a poisson point process of pinned loops : [ t2 ] for each @xmath15 $ ] , there exists at most one cle such that @xmath44 is the sle@xmath1 bubble measure . in a way
, this reconstruction procedure can be interpreted as an `` excursion theory '' for cles .
it will be very closely related to the decomposition of a bessel process via its poisson point process of excursions .
in fact , this will enable us to relate our cles to the random loop ensembles defined in @xcite using branching sle processes , which we now briefly describe . recall that when @xmath57 , @xmath58 is a random simple curve from one marked boundary point @xmath59 of a simply connected domain @xmath10 to another boundary point @xmath60 . if we now compare the law of an sle from @xmath59 to @xmath60 in @xmath10 with that of an sle from @xmath59 to @xmath61 in @xmath10 when @xmath62 , then they clearly differ , and it is also immediate to check that the laws of their initial parts ( i.e. , the laws of the paths up to the first time they exit some fixed small neighborhood of @xmath59 ) are also not identical .
we say that sle@xmath1 is not target - independent .
however , a variant of sle(@xmath6 ) called sle(@xmath6 , @xmath63 ) has been shown by schramm and wilson @xcite ( see also @xcite ) to be target - independent .
this makes it possible to couple such processes starting at @xmath59 and aiming at two different points @xmath60 and @xmath61 in such a way that they coincide until the first disconnection point .
this in turn makes it possible to canonically define a conformally invariant `` exploration tree '' of sle ( @xmath6 , @xmath64 ) processes rooted at @xmath59 , and a collection of loops called conformal loop ensembles in @xcite .
it is conjectured in @xcite that this one - parameter collection of loops indeed corresponds to the scaling limit of a wide class of discrete lattice - based models , and that for each @xmath6 , the law of the constructed family of loops is independent of the starting point @xmath59 .
the branching sle ( @xmath65 ) procedure works for any @xmath66 $ ] , but the obtained loops are simple and disjoint loops only when @xmath57 . in this paper , we use the term cle to refer to any collection of loops satisfying conformal invariance and restriction , while using the term cle@xmath1 to refer to the random collections of loops constructed in @xcite .
we shall prove the following : [ t3 ] every cle is in fact a cle@xmath1 for some @xmath0 $ ] .
let us stress that we have not yet proved at this point that the cle@xmath1 are themselves cles ( and this was also not established in @xcite ) nor that the law of cle@xmath1 is root - independent .
in fact , it is not proved at this point that cles exist at all .
all of this will follow from the second part .
we now describe the content of sections [ s.2bis ] to [ s.4 ] .
the brownian loop - soup , as defined in @xcite , is a poissonian random countable collection of brownian loops contained within a fixed simply - connected domain @xmath10 .
we will actually only need to consider the outer boundaries of the brownian loops , so we will take the perspective that a loop - soup is a random countable collection of simple loops ( outer boudaries of brownian loops can be defined as sle@xmath67 loops , see @xcite ) .
let us stress that our conformal loop ensembles are also random collections of simple loops , but that , unlike the loops of the brownian loop - soup , the loops in a cle are almost surely all disjoint from one another .
the loops of the _ brownian loop - soup _ @xmath68 in the unit disk @xmath23 are the points of a poisson point process with intensity @xmath69 , where @xmath2 is an _ intensity _ constant , and @xmath70 is the _
brownian loop measure in @xmath23_. the brownian loop - soup measure @xmath71 is the law of this random collection @xmath72 . when @xmath47 and @xmath73 are two closed bounded subsets of a bounded domain @xmath10 , we denote by @xmath74 the @xmath70-mass of the set of loops that intersect both sets @xmath47 and @xmath73 , and stay in @xmath10 .
when the distance between @xmath47 and @xmath73 is positive , this mass is finite @xcite .
similarly , for each fixed positive @xmath26 , the set of loops that stay in the bounded domain @xmath10 and have diameter larger than @xmath26 , has finite mass for @xmath70
. the conformal restriction property of the brownian loop measure @xmath70 ( which in fact characterizes the measure up to a multiplicative constant ; see @xcite ) implies the following two facts ( which are essentially the only features of the brownian loop - soup that we shall use in the present paper ) : 1 .
conformal invariance : the measure @xmath75 is invariant under any moebius transformation of the unit disc onto itself .
this invariance makes it in fact possible to define the law @xmath76 of the loop - soup in any simply connected domain @xmath77 as the law of the image of @xmath72 under any given conformal map @xmath52 from @xmath23 onto @xmath10 ( because the law of this image does not depend on the actual choice of @xmath52 ) .
restriction : if one restricts a loop - soup in @xmath23 to those loops that stay in a simply connected domain @xmath78 , one gets a sample of @xmath79 .
we will work with the usual definition ( i.e. , the usual normalization ) of the measure @xmath70 ( as in @xcite note that there can be some confusion about a factor @xmath80 in the definition , related to whether one keeps track of the orientation of the brownian loops or not ) . since we will be talking about some explicit values of @xmath2 later
, it is important to specify this normalization . for a direct definition of the measure @xmath70 in terms of brownian loops , see @xcite . as mentioned above , @xcite pointed out a way to relate brownian loop - soups clusters to sle - type loops : two loops in @xmath72 are said to be adjacent if they intersect .
denote by @xmath81 the set of clusters of loops under this relation .
for each element @xmath82 write @xmath83 for the closure of the union of all the loops in @xmath84 and denote by @xmath85 the family of all @xmath83 s .
we write @xmath86 for the _ filling _ of @xmath84 , i.e. , for the complement of the unbounded connected component of @xmath87 .
a cluster @xmath84 is called _ outermost _ if there exists no @xmath88 such that @xmath89 .
the _ outer boundary _ of such an outermost cluster @xmath84 is the boundary of @xmath86 .
denote by @xmath27 the set of outer boundaries of outermost clusters of @xmath72 .
let us now state the main results of this second step : [ loopsoupsatisfiesaxioms ] suppose that @xmath72 is the brownian loop - soup with intensity @xmath2 in @xmath23 .
* if @xmath90 $ ] , then @xmath27 is a random countable collection of disjoint simple loops that satisfies the conformal restriction axioms . * if @xmath4 , then there is almost surely only one cluster in @xmath81
. it therefore follows from our markovian characterization that @xmath27 is a cle@xmath1 ( according to the branching sle(@xmath91 ) based definition in @xcite ) for _ some _ @xmath0 $ ] .
we will in fact also derive the following correspondence : [ kappacorrespondence ] fix @xmath90 $ ] and let @xmath72 be a brownian loop - soup of intensity @xmath2 on @xmath23 .
then @xmath27 is a cle@xmath1 where @xmath0 $ ] is determined by the relation @xmath92 . since every @xmath0 $ ]
is obtained for exactly one value of @xmath90 $ ] in theorem [ kappacorrespondence ] , we immediately get thanks to theorem [ t3 ] that the random simple loop configurations satisfying the conformal restriction axioms are precisely the cle@xmath1 where @xmath0 $ ] , which completes the proof of theorem [ mainresults ] and of the fact that the following three descriptions of simple loop ensembles are equivalent : 1 .
the random loop ensembles traced by branching schramm - loewner evolution ( sle@xmath1 ) curves for @xmath6 in @xmath7 $ ] .
2 . the outer - cluster - boundary ensembles of brownian loop - soups for @xmath93 .
3 . the ( only ) random loop ensembles satisfying the cle axioms . let us now list some further consequences of these results .
recall from @xcite that the hausdorff dimension of an sle@xmath1 curve is almost surely @xmath94 .
our results therefore imply that the boundary of a loop - soup cluster of intensity @xmath95 has dimension @xmath96 note that just as for mandelbrot s conjecture for the dimension of brownian boundaries @xcite , this statement does not involve sle , but its proof does .
in fact the result about the dimension of brownian boundaries can be viewed as the limit when @xmath97 of this one .
furthermore we may define the _ carpet _ of the cle@xmath1 to be the random closed set obtained by removing from @xmath23 the interiors ( i.e. the bounded connected component of their complement ) of all the loops @xmath98 of @xmath27 , and recall that sle methods allowed @xcite to compute its `` expectation dimension '' in terms of @xmath6 .
the present loop - soup construction of cle@xmath1 enables to prove ( see @xcite ) that this expectation dimension is indeed equal to its almost sure hausdorff dimension @xmath99 , and that in terms of @xmath2 , @xmath100 the critical loop - soup ( for @xmath5 ) corresponds therefore to a carpet of dimension @xmath101 .
another direct consequence of the previous results is the `` additivity property '' of cle s : if one considers two independent cle s in the same simply connected domain @xmath10 with non - empty boundary , and looks at the union of these two , then either one can find a cluster whose boundary contains @xmath102 , or the outer boundaries of the obtained outermost clusters in this union form another cle .
this is simply due to the fact that each of the cle s can be constructed via brownian loop soups ( of some intensities @xmath103 and @xmath104 ) so that the union corresponds to a brownian loop - soup of intensity @xmath105 .
this gives for instance a clean direct geometric meaning to the general idea ( present on various occasions in the physics literature ) that relates in some way two independent copies of the ising model to the gaussian free field in their large scale limit : the outermost boundaries defined by the union of two independent cle@xmath14 s in a domain ( recall @xcite that cle@xmath14 is the scaling limit of the ising model loops , and note that it corresponds to @xmath106 ) form a cle@xmath107 ( which corresponds to `` outermost '' level lines of the gaussian free field , see @xcite and to @xmath108 ) . in order to put our results in perspective
, we briefly recall some closely related work on conformally invariant structures . there exist several natural ways to construct conformally invariant structures in a domain @xmath10 .
we have already mentioned the brownian loop - soup that will turn out be instrumental in the present paper when constructing explicitely cles .
another natural conformally invariant structure that we have also just mentioned is the gaussian free field .
this is a classical basic object in field theory .
it has been shown ( see @xcite ) that it is very closely related to sle processes , and that one can detect all kinds of sles within the gaussian free field .
in particular , this indicates that cles ( at least when @xmath109 ) can in fact also be defined and found as `` geometric '' lines in a gaussian free field .
a number of discrete lattice - based models have been conjectured to give rise to conformally invariant structures in the fine - mesh limit .
for some of these models , these structures can be described by random collections of loops .
we have already mentioned that smirnov @xcite has proved this conjecture for some important models ( percolation , ising model see also @xcite for some other cases ) . those models that will be directly relevant to the present paper ( i.e. , with disjoint simple loops ) include the ising model and the discrete gaussian free field level lines ( @xcite ) .
the scaling limits of percolation and of the fk - model related to the ising model give rise to interfaces that are not disjoint .
these are of course also very interesting objects ( see @xcite for the description of the percolation scaling limit ) , but they are not the subject of the present paper .
conjecturally , each of the cles that we will be describing corresponds to the scaling limit of one of the so - called @xmath110 models , see e.g. @xcite , which are one simple way to define discrete random collections of non - overlapping loops .
in fact , if one starts from a lattice - based model for which one controls the ( conformally invariant ) scaling limit of an observable ( loosely speaking , the scaling limit of the probability of some event ) , it seems possible ( see smirnov @xcite ) to use this to actually prove the convergence of the entire discrete `` branching '' exploration procedure to the corresponding branching sle(@xmath65 ) exploration tree .
it is likely that it is not so much harder to derive the `` full '' scaling limit of all interfaces than to show the convergence of one particular interface to sle .
another quite different family of discrete models that might ( more conjecturally ) be related to the cles that we are studying here , are the `` planar maps '' , where one chooses at random a planar graph , defined modulo homeomorphisms , and that are conjecturally closely related to the above ( for instance via their conjectured relation with the gaussian free field ) .
it could well be that cles are rather directly related to random planar maps chosen in a way to contain `` large holes '' , such as the ones that are studied in @xcite .
in fact , cles , planar maps and the gaussian free field should all be related to each other via liouville quantum gravity , as described in @xcite .
note finally that conformal field theory , as developed in the theoretical physics community since the early eighties @xcite , is also a setup to describe the scaling limits of all correlation functions of these critical two - dimensional lattice models .
this indicates that a description of the entire scaling limit of the lattice models in geometric sle - type terms could be useful in order to construct such fields .
cle - based constructions of conformal field theoretical objects `` in the bulk '' can be found in benjamin doyon s papers @xcite .
it may also be mentioned that aspects of the present paper ( infinite measures on `` pinned configurations '' ) can be interpreted naturally in terms of insertions of boundary operators .
[ s.2 ] a * simple loop * in the complex plane will be the image of the unit circle in the plane under a continuous injective map ( in other words we will identify two loops if one of them is obtained by a bijective reparametrization of the other one ; note that our loops are not oriented ) .
note that a simple loop @xmath98 separates the plane into two connected components that we call its interior @xmath111 ( the bounded one ) and its exterior ( the unbounded one ) and that each one of these two sets characterizes the loop .
there are various natural distances and @xmath112-fields that one can use for the space @xmath113 of loops .
we will use the @xmath112-field @xmath114 generated by all the events of the type @xmath115 when @xmath116 spans the set of open sets in the unit plane .
note that this @xmath112-field is also generated by the events of the type @xmath117 where @xmath118 spans a countable dense subset @xmath119 of the plane ( recall that we are considering simple loops so that @xmath120 as soon as @xmath121 ) . in the present paper
, we will consider ( at most countable ) collections @xmath122 of simple loops .
one way to properly define such a collection is to identify it with the point - measure @xmath123 note that this space of collections of loops is naturally equipped with the @xmath112-field generated by the sets @xmath124 , where @xmath125 and @xmath126 .
we will say that @xmath127 is a * simple loop configuration * in the bounded simply connected domain @xmath10 if the following conditions hold : * for each @xmath128 , the loop @xmath129 is a simple loop in @xmath10 . * for each @xmath130 , the loops @xmath129 and @xmath131 are disjoint . * for each @xmath130
, @xmath129 is not in the interior of @xmath131 : the loops are not nested .
* for each @xmath24 , only finitely many loops @xmath129 have a diameter greater than @xmath42 .
we call this the * local finiteness * condition . all these conditions are clearly measurable with respect to the @xmath112-field discussed above
. we are going to study random simple loop configurations with some special properties .
more precisely , we will say that the random simple loop configuration @xmath132 in the unit disc @xmath23 is a * conformal loop ensemble * ( cle ) if it satisfies the following properties : * non - triviality : the probability that @xmath133 is positive .
* conformal invariance : the law of @xmath27 is invariant under any conformal transformation from @xmath23 onto itself .
this invariance makes it in fact possible to define the law of the loop - ensemble @xmath134 in any simply connected domain @xmath77 as the law of the image of @xmath27 under any given conformal map @xmath52 from @xmath23 onto @xmath10 ( this is because the law of this image does not depend on the actual choice of @xmath52 ) .
we can also define the law of a loop - ensemble in any open domain @xmath135 that is the union of disjoint open simply connected sets by taking independent loop - ensembles in each of the connected components of @xmath10 .
we call this law @xmath18 . *
restriction : to state this important property , we need to introduce some notation .
suppose that @xmath136 is a simply connected subset of the unit disc .
define @xmath137 and @xmath138 .
define the ( random ) set @xmath139 this set @xmath140 is a ( not necessarily simply connected ) open subset of @xmath23 ( because of the local finiteness condition ) .
the restriction property is that ( for all @xmath136 ) , the conditional law of @xmath141 given @xmath142 ( or alternatively given the family @xmath143 ) is @xmath144 .
this definition is motivated by the fact that for many discrete loop - models that are conjectured to be conformally invariant in the scaling limit , the discrete analog of this restriction property holds .
examples include the @xmath110 models ( and in particular the critical ising model interfaces ) .
the goal of the paper is to classify all possible cles , and therefore the possible conformally invariant scaling limits of such loop - models .
the non - nesting property can seem surprising since the discrete models allow nested loops .
the cle in fact describes ( when the domain @xmath10 is fixed ) the conjectural scaling limit of the law of the `` outermost loops '' ( those that are not surrounded by any other one ) . in the discrete models
, one can discover them `` from the outside '' in such a way that the conditional law of the remaining loops given the outermost loops is just made of independent copies of the model in the interior of each of the discovered loops .
hence , the conjectural scaling limit of the full family of loops is obtained by iteratively defining cles inside each loop . at first sight
, the restriction property does not look that restrictive .
in particular , as it involves only interaction between entire loops , it may seem weaker than the conformal exploration property of sle ( or of branching sle(@xmath145 ) ) , that describes the way in which the path is progressively constructed .
however ( and this is the content of theorems [ t1 ] and [ t2 ] ) , the family of such cles is one - dimensional too , parameterized by @xmath0 $ ] .
we now list some simple consequences of the cle definition .
suppose that @xmath32 is a cle in @xmath23 . 1 .
then , for any given @xmath146 , there almost surely exists a loop @xmath129 in @xmath27 such that @xmath147 . here
is a short proof of this fact : define @xmath148 to be the probability that @xmath40 is in the interior of some loop in @xmath27 . by moebius invariance , this quantity @xmath149 does not depend on @xmath40 .
furthermore , since @xmath150 , it follows that @xmath151 ( otherwise the expected area of the union of all interiors of loops would be zero ) .
hence , there exists @xmath152 such that with a positive probability @xmath153 , the origin is in the interior of some loop in @xmath27 that intersects the slit @xmath154 $ ] ( we call @xmath47 this event ) .
we now define @xmath155 and apply the restriction property .
if @xmath47 holds , then the origin is in the interior of some loop of @xmath27 .
if @xmath47 does not hold , then the origin is in one of the connected components of @xmath156 and the conditional probability that it is surrounded by a loop in this domain is therefore still @xmath149 .
hence , @xmath157 so that @xmath158 .
2 . the previous observation implies immediately that @xmath159 is almost surely infinite . indeed ,
almost surely , all the points @xmath160 are surrounded by a loop , and any given loop can only surround finitely many of these points ( because it is at positive distance from @xmath161 ) .
3 . let @xmath162 denote the set of configurations @xmath17 such that for all @xmath128 , the radius @xmath163 $ ] is never locally `` touched without crossing '' by @xmath129 ( in other words , @xmath164 is a local extremum of none of the @xmath165 s ) .
then , for each given @xmath164 , @xmath27 is almost surely in @xmath166 .
indeed , the argument of a given loop that does not pass through the origin can anyway at most have countably many `` local maxima '' , and there are also countably many loops . hence , the set of @xmath164 s such that @xmath167 is at most countable .
but the law of the cle is invariant under rotations , so that @xmath168 does not depend on @xmath164 . since its mean value ( for @xmath169 $ ] ) is @xmath3 , it is always equal to @xmath3 .
+ if we now define , for all @xmath170 , the moebius transformation of the unit disc such that @xmath171 , @xmath172 and @xmath173 , the invariance of the cle law under @xmath21 shows that for each given @xmath174 , almost surely , no loop of the cle locally touches @xmath175)$ ] without crossing it .
4 . for any @xmath176
, the probability that @xmath177 is entirely contained in the interior of one single loop is positive : this is because each simple loop @xmath98 that surrounds the origin can be approximated `` from the outside '' by a loop @xmath178 on a grid of rational meshsize with as much precision as one wants .
this implies in particular that one can find one such loop @xmath178 in such a way that the image of one loop @xmath98 in the cle under a conformal map from @xmath179 onto @xmath23 that preserves the origin has an interior containing @xmath177 .
hence , if we apply the restriction property to @xmath180 , we get readily that with positive probability , the interior of some loop in the cle contains @xmath181 .
since this property will not be directly used nor needed later in the paper , we leave the details of the proof to the reader .
the restriction property continues to hold if we replace the simply connected domain @xmath78 with the union @xmath136 of countably many disjoint simply connected domains @xmath182 .
that is , we still have that the conditional law of @xmath141 given @xmath142 ( or alternatively given the family @xmath183 ) is @xmath144 . to see this ,
note first that applying the property separately for each @xmath184 gives us the marginal conditional laws for the set of loops within each of the @xmath184 .
then , observe that the conditional law of the set of loops in @xmath184 is unchanged when one further conditions on the set of loops in @xmath185 .
hence , the sets of loops in the domains @xmath186 are in fact independent ( conditionally on @xmath143 ) .
suppose that @xmath32 is a cle in the unit disc @xmath23 .
suppose that @xmath24 is given .
cut out from the disc a little given shape @xmath187 of radius @xmath42 around @xmath3 .
if @xmath188 is a point on the unit circle , then we may write @xmath189 for @xmath188 times the set @xmath190 i.e. , @xmath190 rotated around the circle via multiplication by @xmath188 .
the precise shape of @xmath190 will not be so important ; for concreteness , we may at this point think of @xmath190 as being equal to the @xmath42-neighborhood of @xmath3 in the unit disc .
let @xmath191 denote the connected component that contains the origin of the set obtained when removing from @xmath192 all the loops that do not stay in @xmath193 .
if the loop @xmath194 in the cle that surrounds the origin did not go out of @xmath195 , then the ( conditional ) law of the cle restricted to @xmath191 ( given the knowledge of @xmath191 ) is again a cle in this domain ( this is just the cle restriction property ) .
we then define the conformal map @xmath196 from @xmath191 onto @xmath23 with @xmath197 and @xmath198 .
now we again explore a little piece of @xmath191 : we choose some point @xmath199 on the unit circle and define @xmath200 to be the domain obtained when removing from @xmath191 the preimage ( under @xmath196 ) of the shape @xmath190 centered around @xmath199 ( i.e. , @xmath201 ) .
again , we define the connected component @xmath202 that contains the origin of the domain obtained when removing from @xmath200 the loops that do not stay in @xmath200 and the conformal map @xmath203 from @xmath202 onto @xmath23 normalized at the origin .
we then explore in @xmath202 if @xmath204 , and so on .
one can iterate this procedure until we finally `` discover '' the loop @xmath194 that surrounds the origin .
clearly , this will happen after finitely many steps with probability one , because at each step the derivative @xmath205 is multiplied by a quantity that is bounded from below by a constant @xmath206 ( this is because at each step , one composes @xmath207 with a conformal map corresponding to the removal of at least a shape @xmath208 in order to define @xmath209 ) .
hence , if we never discovered @xmath194 , it would follow from koebe s @xmath210-theorem that @xmath211 as @xmath212 , and this would contradict the fact that @xmath194 is almost surely at positive distance from @xmath45 .
we call @xmath213 the random finite step after which the loop @xmath194 is discovered i.e. , such that @xmath214 but @xmath215
. it it important to notice that at each step until @xmath213 , one is in fact repeating the same experiment ( up to a conformal transformation ) , namely cutting out the shape @xmath190 from @xmath23 and then cutting out all loops that intersect @xmath190 . because of the cle s conformal restriction property , this procedure defines an i.i.d .
sequence of steps , stopped at the geometric random variable @xmath213 , which is the first step at which one discovers a loop surrounding the origin that intersects @xmath216 .
this shows also that the conditional law of the cle in @xmath23 given the fact that @xmath217 is in fact identical to the image under @xmath218 of the cle in @xmath219 . in the coming sections
, we will use various sequences @xmath220 .
one natural possibility is to simply always choose @xmath221
. this will give rise to the @xmath42 radial - explorations that will be discussed in section [ s.5 ] .
however , we first need another procedure to choose @xmath222 that will enable to control the behavior of @xmath223 , of @xmath224 and of @xmath225 as @xmath42 tends to @xmath45 .
this will then allow us show that the conditional law of the cle in @xmath23 given the fact that @xmath226 has a limit when @xmath43 .
the precise shape of the sets @xmath190 that we will use will in fact not be really important , as long as they are close to small semi - discs
. it will be convenient to define , for each @xmath188 on the unit circle , the set @xmath227 to be the image of the set @xmath228 , under the conformal map @xmath229 from the upper half - plane @xmath37 onto the unit disc such that @xmath230 and @xmath231 .
note that @xmath232 , so that when @xmath42 is very small , the set @xmath233 is close to the intersection of a small disc of radius @xmath234 around @xmath188 with the unit disc .
this set @xmath235 will play the role of our set @xmath236 .
suppose that @xmath32 is a given ( deterministic ) simple loop - configuration in @xmath23 .
( in this section , we will derive deterministic statements that we will apply to cles in the next section . )
we suppose that : 1 . in @xmath27 , one loop ( that we call @xmath194 ) has @xmath45 in its interior
@xmath237 is a given closed simply connected set such that @xmath238 is simply connected , @xmath47 is the closure of the interior of @xmath47 , and the length of @xmath239 is positive .
the loop @xmath194 does not intersect @xmath47 .
all @xmath129 s in @xmath27 that intersect @xmath240 also intersect the interior of @xmath47 .
our goal will be to explore almost all ( when @xmath42 is small ) large loops of @xmath27 that intersect @xmath47 by iterating explorations of @xmath42-discs .
when @xmath24 is given , it will be useful to have general critera that imply that a subset @xmath241 of the unit disc contains @xmath227 for at least one @xmath242 : consider two independent brownian motions , @xmath243 and @xmath244 started from the origin , and stopped at their first hitting times @xmath245 and @xmath246 of the unit circle .
consider @xmath247 and @xmath248 the two connected components of @xmath249 \cup b^2 [ 0 , t_2])$ ] that have an arc of @xmath161 on their boundary .
note that for small enough @xmath26 , the probability @xmath250 that both @xmath247 and @xmath248 contain some @xmath227 is clearly close to @xmath3 .
suppose now that @xmath241 is a closed subset of @xmath23 such that @xmath251 is simply connected , and let @xmath252 be the probability that one of the two random sets @xmath247 or @xmath248 is a subset of @xmath241 .
then : [ uaub ] for all @xmath253 , there exists a positive @xmath254 such that there exists @xmath255 with @xmath256 as soon as @xmath257 .
the definition of @xmath258 and of @xmath252 shows that @xmath241 contains some @xmath227 with a probability at least @xmath259 . since this is a deterministic fact about @xmath241
, we conclude that the set @xmath241 does indeed contain some set @xmath227 for some @xmath255 as soon as @xmath260 .
it therefore suffices to choose @xmath261 in such a way that @xmath262 .
define now a particular class of iterative exploration procedures as follows : let @xmath263 and @xmath264 . for @xmath265 :
* choose some @xmath266 on @xmath161 in such a way that @xmath267 .
* define @xmath268 as the connected component that contains the origin of the set obtained by removing from @xmath269 all the loops in @xmath27 that do not stay in @xmath270 . *
let @xmath271 be the conformal map from @xmath268 onto @xmath23 such that @xmath272 and @xmath273 .
there is only one way in which such an iterative definition can be brought to an end , namely if at some step @xmath274 , it is not possible anymore to find a point @xmath188 on @xmath161 such that @xmath275 ( otherwise it means that at some step @xmath213 , one actually has discovered the loop @xmath194 , so that @xmath276 is not well - defined but we know that this can not be the case because we have assumed that @xmath277 ) .
such explorations @xmath278 will be called _
@xmath42-admissible explorations of the pair @xmath279_. , sketch of a completed @xmath42-admissible exploration of @xmath279,title="fig:",width=240 ] , sketch of a completed @xmath42-admissible exploration of @xmath279,title="fig:",width=240 ] our goal is to show that when @xmath42 gets smaller , the set @xmath280 is close to @xmath156 , where @xmath156 is the connected component containing the origin of @xmath281 ( here @xmath282 ) .
the local finiteness condition implies that the boundary of @xmath156 consists of points that are either on @xmath283 or on some loop @xmath129 ( in this case , we say that this loop @xmath129 contributes to this boundary ) . for every @xmath284 , there exists @xmath285 such that for all @xmath286 , every loop of diameter greater than @xmath287 that contributes to @xmath288 is discovered by any @xmath42-admissible exploration of @xmath279 .
suppose now that @xmath129 is a loop in @xmath27 that contributes to the boundary of @xmath156 .
our assumptions on @xmath27 and @xmath47 ensure that @xmath129 therefore intersects both the interior of @xmath47 and @xmath238 .
this implies that we can define three discs @xmath289 , @xmath290 and @xmath291 in the interior of @xmath129 such that @xmath292 , and @xmath293 .
suppose that for some @xmath294 , this loop @xmath129 has not yet been discovered at step @xmath295 .
since @xmath296 and @xmath297 , we see that @xmath298 .
since this loop has a positive diameter , and since @xmath27 is locally finite , we can conclude that with a positive probability @xmath149 that depends on @xmath299 , two brownian motions @xmath243 and @xmath244 started from the origin behave as follows : , the three discs and the brownian motions ( sketch),width=307 ] * they both enter @xmath289 without hitting @xmath47 or @xmath283 or any of the other loops @xmath300 for @xmath301 . *
they both subsequently enter @xmath290 without going out of @xmath302 . *
they both subsequently disconnect @xmath290 from the boundary of @xmath291 before hitting it .
( this in particular guarantees that the curves hit one another within the annulus @xmath303 . ) * they both subsequently hit @xmath161 without going out of @xmath47 .
this shows that one of the sets @xmath247 and @xmath248 as defined before lemma [ uaub ] is contained in @xmath47 with probability at least @xmath149 .
in fact , if we stop the two brownian motions at their first exit of @xmath304 instead on the hitting time of @xmath161 , the same phenomenon will hold : one of the two sets @xmath305 and @xmath306 ( with obvious notation ) will be contained in @xmath307 with probability at least @xmath149 . by conformal invariance of planar brownian motion , if we apply lemma [ uaub ] to the conformal images of these two brownian motions under @xmath207 , we get that if @xmath42 is chosen to be sufficiently small , then it is always possible to find an @xmath42-admissible point @xmath308 .
hence , @xmath309 , i.e. @xmath295 is not the final step of the exploration . as a consequence
, we see that the loop @xmath129 is certainly discovered before @xmath274 , i.e. , that @xmath310 , for all @xmath311 .
the lemma follows because for each positive @xmath287 , there are only finitely many loops of diameter greater than @xmath287 in @xmath27 . loosely speaking
, this lemma tells us that indeed , @xmath280 converges to @xmath156 as @xmath43 .
we now make this statement more precise , in terms of the conformal maps @xmath312 and @xmath313 , where @xmath313 denotes the conformal map from @xmath156 onto @xmath23 with @xmath314 and @xmath315 .
let us first note that @xmath316 because the construction of @xmath280 implies that before @xmath274 one can only discover loops that intersect @xmath47 .
let us now consider a two - dimensional brownian motion @xmath317 started from the origin , and define @xmath318 ( respectively @xmath319 ) the first time at which it exits @xmath136 ( resp .
@xmath156 ) .
let us make a fifth assumption on @xmath47 and @xmath27 : * \5 .
almost surely , @xmath320 .
note that this is indeed almost surely the case for a cle ( because @xmath27 is then independent of @xmath321 so that @xmath321 is a.s . in the interior of some loop
if it is not on @xmath161 ) .
this assumption implies that almost surely , either @xmath322 ( and @xmath323 is on the boundary of some loop of positive diameter ) or @xmath324 .
the previous result shows that if @xmath325 denotes the exit time of @xmath280 ( for some given @xmath42-admissible exploration ) , then @xmath326 for all small enough @xmath42 .
it therefore follows that @xmath312 converges to @xmath313 in the sense that for all proper compact subsets @xmath327 of @xmath23 , the functions @xmath328 converge uniformly to @xmath329 in @xmath327 as @xmath43 .
we shall use this notion of convergence on various occasions throughout the paper . note that this it corresponds to the convergence with respect to a distance @xmath99 , for instance @xmath330
we have therefore shown that : for each given loop configuration and @xmath47 ( satisfying conditions 1 - 5 ) , @xmath331 tends to @xmath45 as @xmath43 , uniformly with respect to all @xmath42-admissible explorations of @xmath279 .
suppose now that @xmath332 intersects the interior of @xmath47 .
exactly the same arguments show that there exists @xmath333 such that for all `` @xmath42-admissible choices '' of the @xmath266 s for @xmath334 , one discovers @xmath194 during the exploration ( and this exploration is then stopped in this way ) . for each small @xmath335
, we define the wedge @xmath336 . for each positive @xmath174 ,
let @xmath337 denote the image of the positive half disc @xmath338 under the moebius transformation of the unit disc with @xmath171 , @xmath172 and @xmath173 .
note that @xmath339 is continuously increasing on @xmath340 $ ] from @xmath341 to the positive half - disc @xmath342 .
for all non - negative integer @xmath343 , we then define @xmath344 suppose that @xmath335 is fixed , and that @xmath27 is a loop - configuration satisfying conditions 1 - 5 for all set @xmath345 for @xmath346 , where @xmath347 we are going to define the conformal maps @xmath348 corresponding to the conformal map @xmath313 when @xmath47 is respectively equal to @xmath349 . for each given @xmath335 and @xmath42 , it is possible to define an @xmath42-admissible chain of explorations of @xmath27 and @xmath350 as follows : let us first start with an @xmath42-admissible exploration of @xmath351 .
if @xmath352 , then such an exploration does not encounter @xmath194 , and we then continue to explore until we get an @xmath42-admissible exploration of @xmath353 , and so on , until the last value @xmath354 of @xmath355 for which the exploration of @xmath356 fails to discovers @xmath194 . in this way
, we define conformal maps @xmath357 corresponding to the sets discovered at each of these @xmath354 explorations . note that @xmath358 .
one can then also start to explore the set @xmath359 until one actually discovers the loop @xmath194 .
( sketch),width=288 ] this procedure therefore defines a single @xmath42-admissible exploration ( via some sequence @xmath360 ) , that explores the sets @xmath361 s in an ordered way , and finally stops at some step @xmath213 , i.e. , the last step before one actually discovers @xmath194 .
we call this an @xmath42-admissible exploration chain of @xmath362 .
our previous results show that uniformly over all such @xmath42-admissible exploration - chains ( for each given @xmath27 and @xmath335 ) : * @xmath363 for all sufficiently small @xmath42 . * @xmath364 .
we now suppose that @xmath27 is a _
random _ loop - configuration .
then , for each @xmath335 , @xmath365 is random .
we assume that for each given @xmath335 , the conditions 1 - 5 hold almost surely for each of the @xmath327 sets @xmath366 .
the previous results therefore hold almost surely ; this implies for instance that for each @xmath367 , there exists @xmath368 such that for all such @xmath42-admissible exploration - chain of @xmath369 with @xmath370 , @xmath371 we will now wish to let @xmath335 go to @xmath45 ( simultaneously with @xmath42 , taking @xmath372 sufficiently small ) so that we will ( up to small errors that disappear as @xmath42 and @xmath335 vanish ) just explore the loops that intersect the segment @xmath373 $ ] `` from @xmath3 to @xmath45 '' up to the first point at which it meets @xmath194 .
we therefore define @xmath374 \ , \
r \in \gamma_0 \}.\ ] ] we define the open set @xmath375 as the connected component containing the origin of the set obtained by removing from @xmath376 $ ] all the loops that intersect @xmath377 $ ] .
note that @xmath378 .
we let @xmath379 denote the conformal map from @xmath375 onto @xmath23 such that @xmath380 and @xmath381 .
we also define @xmath382 .
( sketch),width=288 ] [ pwellchosen ] for a well - chosen function @xmath383 ( that depends on the law of @xmath27 only ) , for any choice of @xmath372-admissible exploration - chain of the random loop - configuration @xmath27 and @xmath384 with @xmath385 , the random pair @xmath386 converges almost surely to the pair @xmath387 as @xmath388 .
note first that our assumptions on @xmath27 imply that almost surely @xmath389 as @xmath388 ( this is a statement about @xmath27 that does not involve explorations ) .
we know that @xmath194 intersects @xmath390 . the local finiteness of @xmath27 and the fact that any two loops are disjoint therefore implies that the diameter of the second largest loop ( after @xmath194 ) of @xmath27 that intersects this set almost surely tends to @xmath45 as @xmath388 .
this in particular implies that almost surely , the distance between @xmath391 and @xmath392 tends to @xmath45 as @xmath388 ( uniformly with respect to the choice of the exploration , as long as @xmath372 tends to @xmath45 sufficiently fast ) .
recall finally that for each given @xmath335 , @xmath393 as @xmath43 .
hence , if @xmath372 is chosen small enough , the map @xmath391 indeed converges almost surely to @xmath379 as @xmath388 .
it now remains to show that @xmath394 .
note that @xmath395 ( because at that step , @xmath194 has not yet been discovered ) , that @xmath396 ( with high probability , if @xmath372 is chosen to be small enough ) . on the other hand
, the definition of the exploration procedure and of @xmath354 shows that @xmath397 so that if we choose @xmath398 small enough , then the euclidean distance between @xmath399 and @xmath400 tends to @xmath45 almost surely .
let us look at the situation at step @xmath213 : the loop @xmath401 in the unit disc is intersecting @xmath402 ( by definition of @xmath213 ) , and it contains also the point @xmath403 ( because @xmath404 ) .
suppose that @xmath405 does not almost surely tend to @xmath45 ( when @xmath388 ) ; then , with positive probability , we could find a sequence @xmath406 such that @xmath407 and @xmath408 converge to different points on the unit circle along this subsequence .
in particular , the harmonic measure at the origin of any of the two parts of the loop between the moment it visits @xmath407 and the @xmath42-neighborhood of @xmath408 in @xmath23 is bounded away from @xmath45 .
hence , this is also true for the preimage @xmath194 under @xmath391 : @xmath194 contains two disjoint paths from @xmath400 to @xmath409 such that their harmonic measure at @xmath45 in @xmath23 is bounded away from @xmath45 .
recall that @xmath410 . in the limit when @xmath388 , we therefore end up with a contradiction , as we have two parts of @xmath194 with positive harmonic measure from the origin , that join @xmath400 to some point of @xmath411 $ ] , which is not possible because @xmath412 \not= \emptyset$ ] and @xmath194 is a simple loop .
hence , we can conclude that @xmath413 almost surely .
finally , let us observe that @xmath414 almost surely ( this follows for instance from the fact that a continuous path that stays inside @xmath194 and joins the origin to @xmath400 stays both in all @xmath219 s and in @xmath375 ) .
it follows that @xmath408 converges almost surely to @xmath415 .
we will use the previous exploration mechanisms in the context of cles .
it is natural to define the notion of markovian explorations of a cle .
suppose now that @xmath27 is a cle in the unit disc and that @xmath42 is fixed .
when @xmath416 , define just as before the set @xmath191 and the conformal map @xmath196 obtained by discovering the set of loops @xmath417 of @xmath27 that intersect @xmath418 .
then we choose @xmath199 and proceed as before , until we discover ( at step @xmath419 ) the loop @xmath194 that surrounds the origin .
we say that the exploration is markovian if for each @xmath295 , the choice of @xmath420 is measurable with respect to the @xmath112-field generated by @xmath421 , i.e. , the set of all already discovered loops .
a straightforward consequence of the cle s restriction property is that for each @xmath295 , conditionally on @xmath422 ( and @xmath423 ) , the law of the set of loops of @xmath27 that stay in @xmath304 is that of a cle in @xmath304 . in other words ,
the image of this set of loops under @xmath207 is independent of @xmath422 ( on the event @xmath424 .
in fact , we could have used this independence property as a definition of markovian explorations ( it would allow extra randomness in the choice of the sequence @xmath420 ) .
in other words , an exploration is markovian if we can choose @xmath420 as we wish using the information about the loops that have already been discovered , but we are not allowed to use any information about the yet - to - be - discovered loops .
this ensures that one obtains an iteration of i.i.d .
explorations as argued in subsection [ heuristics ] .
in particular , if an exploration is markovian , the random variable @xmath213 is geometric @xmath425 and @xmath426 is distributed according to the conditional law of @xmath194 given @xmath427 .
recall that a cle is a random loop configuration such that for any given @xmath335 and @xmath428 , almost surely , all loops that intersect @xmath345 also intersect its interior .
we can therefore apply proposition [ pwellchosen ] , and use markovian @xmath372-admissible successive explorations of @xmath429 combining this with our description of the conditional law of @xmath194 given @xmath427 , we get the following result : [ p44 ] when @xmath43 , the law of @xmath194 conditioned on the event @xmath430 converges to the law of @xmath431 ( using for instance the weak convergence with respect to the hausdorff topology on compact sets ) . note
that local finiteness of the cle ensures that @xmath432 is a simple loop in @xmath433 that intersects @xmath161 only at @xmath415 , so that @xmath434 is indeed a loop in @xmath433 that touches @xmath161 only at @xmath3 . conditioned on the event @xmath430 converges ( sketch),title="fig:",width=220 ] conditioned on the event @xmath430 converges ( sketch),title="fig:",width=220 ] this limiting law will inherit from the cle various interesting properties .
the loop @xmath194 in the cle can be discovered along the ray @xmath435 $ ] in the unit disc as in this proposition , but one could also have chosen any other smooth continuous simple curve from @xmath161 to @xmath45 instead of that ray and discovered it that way .
this fact should correspond to some property of the law of this pinned loop .
conformal invariance of the cle will also imply some conformal invariance properties of this pinned loop .
the goal of the coming sections is to derive and exploit some of these features .
we are now going to associate to each cle a natural measure on loops that can loosely be described as the law of a loop `` conditioned to touch a given boundary point '' . in the previous subsection ,
we have constructed a probability measure on loops in the unit disc that was roughly the law of @xmath194 ( the loop in the cle that surrounds the origin ) `` conditioned to touch the boundary point @xmath3 '' .
we will extend this to an infinite measure on loops that touch the boundary at one point ; the measure will be infinite because we will not prescribe the `` size '' of the boundary - touching loop ; it can be viewed as a cle `` excursion measure '' ( `` bubble measure '' would also be a possible description ) .
( sketch),title="fig:",width=220 ] ( sketch),title="fig:",width=220 ] we find it more convenient at this stage to work in the upper half - plane rather than the unit disc because scaling arguments will be easier to describe in this setting .
we therefore first define the probability measure @xmath436 as the image of the law of @xmath437 under the conformal map from @xmath23 onto the half - plane that maps @xmath45 to @xmath438 , and @xmath415 to @xmath45 ( we use the notation @xmath436 instead of @xmath56 as in the introduction , because we will very soon be handling measures that are not probability measures ) . in other words ,
the probability measure @xmath436 is the limit as @xmath43 of the law of the loop @xmath439 that surrounds @xmath438 in a cle in the upper half - plane , conditioned by the fact that it intersects the set @xmath440 defined by @xmath441 for a loop - configuration @xmath27 in the upper half - plane and @xmath38 , we denote by @xmath41 the loop of @xmath27 that surrounds @xmath40 ( if this loop exists ) .
when @xmath442 for @xmath443 , scale invariance of the cle shows that the limit as @xmath43 of the conditional law of @xmath444 given that it intersects the disc of radius @xmath445 exists , and that it is just the image of @xmath436 under scaling .
let us denote by @xmath446 the probability that the loop @xmath439 intersects the disc of radius @xmath42 around the origin in a cle . the description of the measure @xmath436 ( in terms of @xmath379 ) derived in the previous section shows that at least for almost all @xmath447 sufficiently close to one , the loop that surrounds @xmath438 also surrounds @xmath448 and @xmath449 with probability at least @xmath450 under @xmath436 , and that @xmath449 as well as @xmath451 are a.s .
not on @xmath439 ( when @xmath447 is fixed ) .
let @xmath452 denote the interior of the loop @xmath439 .
we know that @xmath453 on the other hand , the scaling property of the cle shows that when @xmath43 , @xmath454 hence , for all @xmath447 sufficiently close to @xmath3 , we conclude that @xmath455 if we call @xmath456 this last quantity , this identity clearly implies that this convergence in fact holds for all positive @xmath457 , and that @xmath458 .
furthermore , we see that @xmath459 as @xmath460 . hence
: there exists a @xmath461 ( @xmath462 can not be negative since @xmath463 is non - decreasing ) such that for all positive @xmath447 , @xmath464 this has the following consequence : @xmath465 as @xmath466 .
note that for any @xmath467 , there exists @xmath468 such that for all @xmath469 , @xmath470 hence , for all @xmath471 it follows that @xmath472 it follows that @xmath473 when @xmath212 .
since @xmath463 is an non - decreasing function of @xmath42 , the corollary follows . because of scaling
, we can now define for all @xmath474 , a measure @xmath475 on loops @xmath41 that surround @xmath40 and touch the real line at the origin as follows : @xmath476 for any measurable set @xmath477 of loops .
this is also the limit of @xmath478 times the law of @xmath41 in an cle , restricted to the event @xmath479 .
let us now choose any @xmath40 in the upper half - plane .
let @xmath480 now denote the moebius transformation from the upper half - plane onto itself with @xmath481 and @xmath482 .
let @xmath483 .
clearly , for any given @xmath484 , for any small enough @xmath42 , the image of @xmath440 under @xmath21 is `` squeezed '' between the circles @xmath485 and @xmath486 .
it follows readily ( using the fact that @xmath487 as @xmath488 ) that the measure @xmath475 defined for all measurable @xmath477 by @xmath489 can again be viewed as the limit when @xmath43 of @xmath490 times the distribution of @xmath41 restricted to @xmath491 .
finally , we can now define our measure @xmath70 on pinned loops .
it is the measure on simple loops that touch the real line at the origin and otherwise stay in the upper half - plane ( this is what we call a pinned loop ) such that for all @xmath38 , it coincides with @xmath475 on the set of loops that surround @xmath40 .
indeed , the previous limiting procedure shows immediately that for any two points @xmath40 and @xmath492 , the two measures @xmath475 and @xmath493 coincide on the set of loops that surround both @xmath40 and @xmath492 . on the other hand , we know that a pinned loop necessarily surrounds a small disc :
thus , the requirement that @xmath70 coincides with the @xmath475 s ( as described above ) fully determines @xmath70 . let us sum up the properties of the pinned measure @xmath70 that we will use in what follows : * for any conformal transformation @xmath21 from the upper half - plane onto itself with @xmath482 , we have @xmath494 this is the _ conformal covariance property _ of @xmath70 .
note that the maps @xmath495 for real @xmath496 satisfy @xmath497 so that @xmath70 is invariant under these transformations .
* for each @xmath40 in the upper half - plane , the mass @xmath498 is finite and equal to @xmath499 , where @xmath21 is the conformal map from @xmath37 onto itself with @xmath500 and @xmath501 .
* for each @xmath40 in the upper half - plane , the measure @xmath70 restricted to the set of loops that surround @xmath40 is the limit as @xmath466 of @xmath490 times the law of @xmath41 in a cle restricted to the event @xmath502 . in other words , for any bounded continuous ( with respect to the hausdorff topology , say ) function @xmath503 on the set of loops , @xmath504 since this pinned measure is defined in a domain with one marked point , it is also quite natural to consider it in the upper half - plane @xmath37 , but to take the marked point at infinity . in other words , one takes the image of @xmath70 under the mapping @xmath505 from @xmath37 onto itself .
this is then a measure @xmath506 on loops pinned at infinity , i.e. on double - ended infinite simple curves in the upper - half plane that go from and to infinity .
it clearly also has the scaling property with exponent @xmath462 , and the invariance under the conformal maps that preserve infinity and the derivative at infinity is just the invariance under horizontal translations .
we will now ( and also later in the paper ) use the exploration mechanism corresponding to the case where all points @xmath420 are chosen to be equal to @xmath3 , instead of being tailored in order for the exploration to stay in some a priori chosen set @xmath47 as in section [ explo ] .
let us describe this discrete radial exploration ( this is how we shall refer to it ) in the setting of the upper half - plane @xmath37 : we fix @xmath24 , and we wish to explore the cle in the upper half - plane by repeatedly cutting out origin - centered semi - circles @xmath507 of radius @xmath42 ( and the loops they intersect ) and applying conformal maps .
the first step is to consider the set of loops that intersect @xmath440 .
either we have discovered the loop @xmath439 ( i.e. , the loop that surrounds @xmath438 ) , in which case we stop , or we havent , in which case we define the connected component of the complement of these loops in @xmath508 that contains @xmath438 , and map it back onto the upper half - plane by the conformal map @xmath509 such that @xmath510 and @xmath511 .
we then start again , and this defines an independent copy @xmath512 of @xmath509 . in this way
, we define a random geometric number @xmath513 of such conformal maps @xmath514 .
the @xmath419-th map can not be defined because one then discovers the loop that surrounds @xmath438 .
the probability that @xmath515 is equal to @xmath516 .
the difference with the exploration procedure of section [ explo ] is that we do not try to explore `` along some prescribed curve '' , but we just iterate i.i.d .
conformal maps in such a way that the derivative at @xmath438 remains a positive real , i.e. we consistently target the inner point @xmath438 .
another natural exploration procedure uses a discrete chordal exploration that targets a boundary point : we first consider the set of loops in a cle in @xmath37 that intersect @xmath440 .
now , we consider the unbounded connected component @xmath517 of the complement in @xmath508 of the union of these loops .
we map it back onto the upper half - plane using the conformal map @xmath509 normalized at infinity , i.e. @xmath518 as @xmath38 .
then we iterate the procedure , defining an infinite i.i.d .
sequence @xmath519 of conformal maps , and a decreasing family of domains @xmath520 ( unlike the radial exploration , this chordal exploration never stops ) .
let us make a little heuristic discussion in order to prepare what follows .
when @xmath42 is very small , the law of the sets that one removes at each step in these two exploration mechanisms can be rather well approximated thanks to the measure @xmath70 .
let us for instance consider the chordal exploration . because of @xmath70 s scaling property , the @xmath70-mass of the set of loops of half - plane capacity ( which scales like the square of the radius ) larger than @xmath178 should decay as @xmath521 . in particular , in the discrete exploration , the average number of exploration steps where one removes a set @xmath522 with half - plane capacity larger than @xmath118 is equal ( in the @xmath43 limit ) to @xmath523 times the average number of exploration steps where one removes a set with half - plane capacity larger than one .
readers familiar with lvy processes will probably have recognized that in the @xmath43 limits , the capacity jumps of the chordal exploration process will be distributed like the jumps of a @xmath524-stable subordinator .
it is important to stress that we know a priori that the limiting process has to possess macroscopic capacity jumps ( corresponding to the discovered macroscopic loops ) .
since @xmath287-stable subordinators exist only for @xmath525 , we expect that @xmath526 .
proving this last fact will be the main goal of this section .
in fact , we shall see that the entire discrete explorations ( and not only the process of accumulated half - plane capacities ) converge to a continuous `` lvy - exploration '' defined using a poisson point process of pinned loops with intensity @xmath70 .
we however defer the more precise description of these lvy explorations to section [ s.8 ] , where we will also make the connection with branching sle(@xmath145 ) processes and show how to reconstruct the law of a cle using the measure @xmath70 only , i.e. , that the measure @xmath70 characterizes the law of the cle .
in fact , the rest of this part of the paper ( until the constructive part using brownian loop - soups ) has two main goals : the first one is to show that the cle definition yields a description of the pinned measure @xmath70 in terms of sle@xmath1 for some @xmath15 $ ] . the second one is to prove that the pinned measure @xmath70 characterizes the law of the entire cle and also to make the connection with sle(@xmath145 ) .
we choose to start with the sle - description of @xmath70 ; in the next subsection , we will therefore only derive those two results that will be needed for this purpose , leaving the more detailed discussion of continuous chordal explorations for later sections . for
readers who are not so familiar with lvy processes or loewner chains , let us now briefly recall some basic features of poisson point processes and the stability of loewner chains that will hopefully help making the coming proofs more transparent .
* stability of loewner chains : * let @xmath527 denote the class of conformal maps @xmath528 from a subset @xmath49 of the complex upper half - plane back onto the upper half - plane @xmath37 such that @xmath529 , @xmath530 and @xmath531 is a positive real . note that ( because @xmath532 ) , @xmath533 .
let us then define @xmath534 .
this is a decreasing function of the domain @xmath49 ( the smaller @xmath49 , the larger @xmath59 ) , which is closely related to the conformal radius of the domain ( one can for instance conjugate with @xmath535 in order to be in the usual setting of the disc ) .
it is immediate to check that for some universal constant @xmath84 , for all @xmath536 and for all @xmath537 such that the diameter of @xmath538 is smaller than @xmath174 , @xmath539 on the other hand , for some other universal constant @xmath84 , for all @xmath536 and for all @xmath49 such that @xmath540 , @xmath541 suppose now that @xmath542 are @xmath213 given conformal maps in @xmath527 .
define @xmath543 to be the composition of these conformal maps .
it is an easy fact ( that can be readily deduced from simple distortion estimates , or via loewner s theory to approximate these maps via loewner chains for instance ) that for any family @xmath544 of conformal maps in @xmath527 such that @xmath545 the conformal maps @xmath546 converge in carathodory topology ( viewed from @xmath438 ) to @xmath52 as @xmath388 . in other words , putting some perturbations of the identity between the iterations of @xmath537 s does not change things a lot , as long as the accumulated `` size '' ( measured by @xmath59 ) of the perturbations is small . *
poisson point processes discrete random sequences : * we will approximate poisson point processes via discrete sequences of random variables .
we will at some point need some rather trivial facts concerning poisson random variables , that we now briefly derive .
suppose that we have a sequence of i.i.d .
random variables @xmath547 that take their values in some finite set @xmath548 with @xmath549 .
let @xmath213 denote the smallest @xmath295 value at which @xmath550 . for each @xmath551 ,
let @xmath552 denote the cardinality of @xmath553 .
we want to control the joint law of @xmath554 .
one convenient way to represent this joint law is to consider a poisson point process @xmath555 on @xmath556 with intensity @xmath557 , where @xmath558 is some @xmath112-finite measure on a space @xmath559 .
consider disjoint measurable sets @xmath560 in @xmath559 with @xmath561 , @xmath562 ,
, @xmath563 .
we define @xmath564 i.e. , loosely speaking , if we interpret @xmath565 as a time - variable , @xmath318 is the first time at which one observes an @xmath566 in @xmath47 . since @xmath567 ,
the law of @xmath318 is exponential with parameter @xmath3 .
in particular , @xmath568 .
we now define , for @xmath569 , @xmath570 i.e. , the number of times one has observed an @xmath566 in @xmath571 before the first time at which one observes an @xmath566 in @xmath47 .
then the law of @xmath572 is the same as before , where @xmath573 . because of the independence properties of poisson point processes , the conditional law of @xmath554 given @xmath318 is that of @xmath327 independent poisson random variables with respective means @xmath574 .
hence , @xmath575 . furthermore , if we condition on the event @xmath576 , the ( joint ) conditional distribution of @xmath572 `` dominates '' that of @xmath355 independent poisson random variables of parameter @xmath577 .
let us denote the `` radius '' of a loop @xmath98 in the upper half - plane by @xmath578 [ lr1 ] the @xmath70-measure of the set of loops with radius greater than @xmath3 is finite .
let us use the radial exploration mechanism .
the idea of the proof is to see that if the @xmath70-mass of the set of loops of radius greater than @xmath3 is infinite , then one `` collects '' too many macroscopic loops before finding @xmath439 in the exploration mechanism , which will contradict the local finiteness of the cle .
recall that , at this moment , we know that for any given point @xmath38 , the measure @xmath579 is the limit when @xmath43 of @xmath490 times the law of @xmath41 restricted to the event that @xmath41 intersects the @xmath42-neighborhood of the origin .
furthermore , for any given point @xmath38 , the distance between @xmath40 and @xmath98 is @xmath70-almost always positive .
this implies that for any given finite family of points @xmath580 in @xmath37 , the measure @xmath70 restricted to the set @xmath581 of loops that surrounds at least one of these points is the limit when @xmath43 of @xmath490 times the ( sum of the ) laws of loops in @xmath477 that intersect the @xmath42-neighborhood of the origin .
suppose now that @xmath582 .
this implies clearly that for each @xmath583 , one can find a finite set of points @xmath580 at distance greater than one of the origin such that @xmath584 .
hence , it follows that when @xmath42 is small enough , at each exploration step , the probability to discover a loop in @xmath477 is at least @xmath585 times bigger than @xmath446 .
the number of exploration steps before @xmath213 at which this happens is therefore geometric with a mean at least equal to @xmath585 .
it follows readily that with probability at least @xmath586 ( for some universal positive @xmath586 that does not depend on @xmath42 ) , this happens at least @xmath587 times .
note that the harmonic measure in @xmath37 at @xmath438 of a loop of radius at least @xmath3 that intersects also @xmath588 is bounded from below by some universal positive constant @xmath103 .
if at some step @xmath589 , the radius of @xmath590 is greater than @xmath3 , then it means that there is a loop in the cle in @xmath37 that one explores at the @xmath591-th step that has a radius at least @xmath3 and that intersects @xmath588 .
its preimage under @xmath592 ( which is a loop of the original cle that one is discovering ) has therefore also has a harmonic measure ( in @xmath37 and at @xmath438 ) that is bounded from below by @xmath103 .
hence , we conclude that with probability at least @xmath586 , the original cle has at least @xmath587 different loops such that their harmonic measure seen from @xmath438 in @xmath37 is bounded from below by some universal constant @xmath103 .
this statement holds for all @xmath585 , so that with probability at least @xmath586 , there are infinitely many loops in the cle such that their harmonic measure seen from @xmath438 in @xmath37 is bounded from below by @xmath103 . on the other hand , we know that @xmath439 is almost surely at positive distance from @xmath438 , and this implies that for some positive @xmath287 , the probability that some loop in the cle is a distance less than @xmath287 of @xmath438 is smaller than @xmath593 .
hence , with probability at least @xmath593 , the cle contains infinitely many loops that are all at distance at least @xmath287 from @xmath438 and all have harmonic measure at least @xmath103 .
a similar statement is therefore true for the cle in the unit disc if one maps @xmath438 onto the origin .
it is then easy to check that the previous statement contradicts the local finiteness ( because the diameter of the conformal image of all these loops is bounded from below ) .
hence , the @xmath70-mass of the set of pinned loops that reach the unit circle is indeed finite .
let us now list various consequences of lemma [ lr1 ] : * for all @xmath170 , let us define @xmath594 . because of scaling , we know that for all @xmath170 , @xmath595
. clearly , this can not be a constant finite function of @xmath174 ; this implies that @xmath596 .
also , we get that for each fixed @xmath174 , @xmath597 .
* we can now define the function @xmath598 as the probability that in the cle , there exists a loop that intersects @xmath440 and @xmath599 .
we know that @xmath600 ( because @xmath601 ) .
hence , it follows that for any @xmath602 , @xmath603 as @xmath43 .
this will be useful later on .
* we can rephrase in a slightly more general way our description of @xmath70 in terms of limits of cle loops .
note that @xmath604 , where @xmath605 is some given dense sequence on @xmath606 .
+ for each simple loop configuration @xmath27 , and for each @xmath42 , let us define @xmath607 to be the loop in the configuration @xmath27 that intersects the disc of radius @xmath42 and with largest radius ( in case there are ties , take any deterministic definition to choose one ) .
we know from ( [ vandu ] ) that the probability that @xmath608 decays like @xmath609 as @xmath43 ( for each fixed @xmath610 ) .
+ furthermore , we note that the probability that there exist two different loops of radius greater than @xmath174 that intersect the circle @xmath440 decays like @xmath611 as @xmath43 .
indeed , otherwise , for some sequence @xmath612 , the probability that in our exploration procedure , two macroscopic loops are discovered simultaneously remains positive and bounded from below , which is easily shown to contradict the fact that almost surely , any two loops in our cle are at positive distance from each other . + hence , for each @xmath295 , the measure @xmath70 restricted to @xmath613 is the limit when @xmath43 of @xmath490 times the law of @xmath607 , restricted to the event that it surrounds at least one of the points @xmath580 .
+ we conclude that the measure @xmath70 restricted to @xmath614 can be viewed as the weak limit when @xmath43 of @xmath490 times the law of @xmath607 restricted to @xmath614 . in other words , if we consider the set of pinned loops of strictly positive size ( i.e. , the loop of zero length is not in this set ) endowed with the hausdorff metric , we can say that @xmath70 is the vague limit of @xmath490 times the law of @xmath607 .
let us finally state another consequence of this result , that will turn out to be useful in the loop - soup construction part of the paper . consider a cle in @xmath37 and let us consider the set of loops that intersect the unit circle .
define @xmath615 to be the radius of the smallest disc centered at the origin that contains all these loops .
note that scaling shows that @xmath616 for @xmath617 .
[ boundedness ] if @xmath618 , then @xmath619 .
just note that @xmath620 because @xmath621 . with the @xmath622 identification that we will derive later , @xmath623 corresponds to @xmath624 and then @xmath625 .
the next proposition corresponds to the fact that @xmath287-stable subordinators exist only for @xmath525 .
the scaling exponent @xmath462 described above lies in @xmath626 .
we use the radial exploration mechanism again .
let us now assume that @xmath627 and focus on the contribution of the `` small '' loops that one discovers before actually discovering @xmath439 .
let @xmath628 be @xmath355 fixed disjoint ( measurable ) sets of loops that do not surround @xmath438 , with @xmath629 .
we suppose that all @xmath630 s for @xmath631 belong to the algebra of events generated by the events of the type @xmath632 .
we let @xmath633 denote the number of loops in @xmath630 that one has discovered in this way before one actually discovers the loop that surrounds @xmath438 .
our previous results show that when @xmath43 , the joint law of @xmath634 converges to that of the @xmath572 that we described at the end of the previous subsection . for each integer @xmath591
, we define @xmath635 to be the set of loops that surround @xmath636 but that do not surround any point @xmath637 for @xmath638 .
the scaling property of @xmath70 implies that for all @xmath591 , @xmath639 and it is easy to check that @xmath640 ( because the loops that surround @xmath438 have a finite radius so that @xmath641 ) .
hence , for all positive @xmath591 , @xmath642 recall that if @xmath21 is a conformal map in @xmath527 from a simply connected subset @xmath49 of @xmath37 that does contain @xmath438 but not @xmath643 for some @xmath644 , then @xmath645 furthermore , for each fixed @xmath646 , when @xmath42 is small enough , we can compare the number of loops in @xmath647 that have been discovered before @xmath439 via the chordal exploration mechanism with i.i.d .
poisson random variables @xmath648 . note also that when one composes conformal maps in @xmath527 , the derivatives at @xmath438 get multiplied and the @xmath59 s therefore add up .
hence , it follows immediately that if @xmath627 , then for each @xmath646 , with a probability that is bounded from below independently of @xmath646 , when @xmath42 is small enough , @xmath649 hence , we conclude that there exists @xmath650 such that for each @xmath651 , if one chooses @xmath42 small enough , the probability that @xmath652 is at least @xmath104 . but
this contradicts the fact that @xmath439 is at positive distance from @xmath438 ( for instance using koebe s @xmath210 theorem ) .
hence , we conclude that @xmath462 is indeed smaller than @xmath80 .
we now investigate what sort of `` restriction - type '' property the pinned measure @xmath70 inherits from the cle .
note that @xmath70 is an infinite measure on single pinned loops ( rather than a probability measure on loop configurations ) , so the statement will necessarily be a bit different from the restriction property of cle .
suppose now that @xmath47 is a closed bounded set such that @xmath653 is simply connected , and that @xmath654 .
our goal is to find an alternative description of the infinite measure @xmath70 restricted to the set of loops @xmath98 that do not intersect @xmath47 .
suppose that a deterministic pinned ( at zero ) loop @xmath98 ( in the upper half - plane ) is given .
sample a cle @xmath655 in the upper half - plane .
this defines a random @xmath656 which is the connected component that has the origin on its boundary of the set obtained by removing from @xmath653 all loops of @xmath655 that intersect @xmath47 .
then , we define a conformal map @xmath657 from @xmath37 onto @xmath656 such that @xmath658 in order to fix @xmath657 , another normalization is needed . we can for instance take @xmath659 , but all of what follows would still hold if one replaced @xmath657 by the map @xmath660 such that @xmath661 as @xmath662 ( we will in fact also use this map in the coming sections ) .
finally , we define @xmath663 .
clearly , this is a pinned loop that stays in @xmath664 and therefore avoids @xmath47 almost surely .
suppose now that we use the product measure @xmath665 on pairs @xmath666 ( where @xmath667 is the law of the same cle that was used to define the pinned measure @xmath70 ) .
for each pair @xmath666 we define the loop @xmath668 as before , and we define @xmath669 to be the image measure of @xmath665 via this map .
this @xmath669 is an infinite measure on the set of loops that do not intersect @xmath47 .
( sketch),width=480 ] we are now ready to state the pinned measure s restriction property : [ mua ] the measure @xmath70 restricted to the set @xmath670 is equal to @xmath669 .
recall that we constructed the pinned measure by `` exploring '' a cle in @xmath23 until we discover @xmath194 .
if we started with a pinned loop together with a cle in the complement of the pinned loop we might try to `` explore '' this configuration until we hit the pinned loop .
this would be a way of constructing a natural measure on loops pinned at two points .
we will essentially carry out such a procedure later on , and the above proposition will be relevant .
let us now consider the set @xmath47 and a cle @xmath655 , and define @xmath656 and the map @xmath657 as before . for each @xmath42 , the event that no loop in @xmath655 intersects both @xmath440 and @xmath47 is identical to the event that @xmath671 .
let us now condition on the set @xmath656 ( for a configuration where @xmath671 ) ; the restriction property of the cle tells us that the conditional law of the cle - loops that stay in @xmath656 is exactly that of an `` independent cle '' defined in this random set @xmath656 .
hence , we get an identity between the following two measures : * the law of @xmath607 in the cle ( the loop with largest radius that intersects @xmath440 ) , restricted to the event that no loop in the cle intersects both @xmath440 and @xmath47 .
* sample first a cle @xmath655 , define @xmath656 , restrict ourselves to the event where @xmath671 , define @xmath657 , consider an independent cle @xmath672 in the upper half - plane and its image @xmath673 , and look at the law of the loop with largest radius in this family that intersects @xmath440 .
note that the total mass of these two measures is the probability that no loop in the cle intersects both @xmath440 and @xmath47 .
now , we consider the vague limits when @xmath43 of @xmath674 times these two measures .
it follows readily from our previous considerations that : * for the first construction , the limit is just @xmath70 restricted to the set of pinned loops that do not hit @xmath47 .
* for the second construction , the limit is just @xmath669 ( recall that @xmath657 has been chosen in such a way that @xmath675 , so that when @xmath42 is small @xmath676 is very close to @xmath440 ) .
in fact , it will be useful to upgrade the previous result to the space of `` pinned configurations '' : we say that @xmath677 is a _ pinned configuration _ if @xmath678 is a pinned loop in the upper half - plane and if @xmath679 is a loop - configuration in the unbounded connected component of @xmath680 .
let us define a first natural measure on the space of pinned configurations .
suppose that one is given a pinned loop @xmath98 and a loop - configuration @xmath27 ( that is not necessarily disjoint from @xmath98 ) .
define @xmath681 the unbounded connected component of the complement of @xmath98 , and @xmath682 the conformal map from @xmath37 onto @xmath681 that is normalized at infinity ( @xmath683 ) .
then @xmath684 is clearly a pinned loop configuration .
we now let @xmath685 denote the image of the product measure @xmath686 under this map @xmath687 .
we call it the pinned cle configuration measure . clearly , the marginal measure of @xmath678 ( under @xmath685 ) is @xmath70 ( because @xmath667 is a probability measure and @xmath688 ) .
( sketch),width=480 ] suppose now that @xmath47 is a set as before .
if @xmath689 is a pinned configuration , then we define : * @xmath690 to be the set of loops of @xmath679 that intersect @xmath47 . *
@xmath691 to be the unbounded connected component of @xmath692 * @xmath693 to be the set of loops of @xmath679 that stay in @xmath691 . hence ,
when @xmath694 , we define a triplet @xmath695 .
let @xmath696 denote the image of @xmath685 ( restricted to the set @xmath694 ) under this transformation .
we now construct another measure that will turn out to be identical to @xmath696 : start on the one hand with a pinned measure configuration @xmath677 and on the other hand with a loop - configuration in @xmath37 that we denote by @xmath655 .
we define @xmath657 as before ( using @xmath47 and @xmath655 ) .
we also let @xmath697 denote the loops in @xmath655 that intersect @xmath47 .
then we consider the triplet @xmath698 .
this triplet is a function of @xmath699 .
we now define @xmath700 to be the image of the product measure @xmath701 under this mapping .
( sketch),width=480 ] [ p72 ] the two measures @xmath696 and @xmath700 are identical .
if we consider the marginal measures on the pinned loops of @xmath696 and @xmath700 , we recover proposition [ mua ] .
the proof is basically identical to that of proposition [ mua ] : one just needs also to keep track of the remaining loops , and this in done on the one hand thanks to the cle s restriction property , and on the other hand ( in the limiting procedure ) thanks to the fact that loops are disjoint and at positive distance from the origin . in order to control the @xmath43 limiting procedure applied to loop configurations
, one can first derive the result for the law of finitely many loops in the loop configuration ( for instance those that surround some given point ) .
as in the case of the cle , we are going to progressively explore a pinned configuration , cutting out recursively small ( images of ) semi - circles , and trying to make use of the pinned measure s restriction property in order to define a `` two - point pinned measure '' .
however , some caution will be needed in handling ideas involving independence because @xmath685 is not a probability measure .
recall that @xmath685 is a measure on configurations in a simply connected domain with one special marked boundary point ( i.e. , the origin , where the pinned loop touches the boundary of the domain ) .
it will therefore be natural to work with chordal @xmath42-admissible explorations instead of radial ones
. this will be rather similar to the radial case , but it is nonetheless useful to describe it in some detail : it is convenient to first consider the domain to be the upper half - plane and to take @xmath702 as the marked boundary point instead of the origin .
suppose that @xmath27 is a loop configuration ( with no pinned loop ) in the upper half - plane @xmath37 and choose some bounded closed simply connected set @xmath703 such that 1 .
@xmath653 is simply connected , 2 .
@xmath47 is the closure of the interior of @xmath47 , 3 .
the interior of @xmath47 is connected , and 4 .
the length of @xmath704 is positive .
suppose furthermore that all @xmath129 s in @xmath27 that intersect @xmath240 also intersect the interior of @xmath47 . here ,
when @xmath705 , @xmath706 will denote the set of points in @xmath37 that are at distance less than @xmath42 from @xmath118 .
we choose some @xmath707 on the real line , such that @xmath708 .
then , we define the set @xmath517 to be the unbounded connected component of the set obtained when removing from @xmath709 all loops of @xmath27 that intersect @xmath710 .
we also define the conformal map @xmath711 from @xmath517 onto @xmath37 that is normalized at infinity ( @xmath712 as @xmath713 ) , and we let @xmath714 .
then , we proceed inductively : for each @xmath715 , we choose ( when it is possible ) @xmath716 on the real line such that @xmath717 .
then we consider the set @xmath718 to be the unbounded connected component of the set obtained when removing from @xmath719 all loops of @xmath720 that intersect @xmath721 .
we also define the conformal map @xmath722 from @xmath718 onto @xmath37 that is normalized at infinity ( @xmath723 as @xmath713 ) , and we let @xmath724 .
this procedure necessarily has to stop at some step @xmath213 , i.e. , at this step @xmath213 , it is not possible to find a point @xmath118 such that @xmath725 .
this is just because of additivity of the half - plane capacity under composition of the conformal maps ( and because the half - plane capacity of @xmath47 is finite ) .
we say that such an exploration is an _
@xmath42-admissible ( chordal ) exploration of @xmath279 rooted at infinity_. the results of subsection [ explodet ] and their proofs can be immediately adapted to the present setting .
it is for instance easy to check that for any given @xmath27 , for any given loop @xmath129 in @xmath27 that intersects such a given @xmath47 , there exists a positive @xmath726 such that any @xmath42-admissible chordal exploration of @xmath279 necessarily discovers the loop @xmath129 as soon as @xmath469 .
hence , after the last step @xmath213 of this @xmath42-exploration of @xmath279 , for all positive @xmath287 , when @xmath42 is chosen to be sufficiently small , one has necessarily discovered all loops of diameter greater than @xmath287 of @xmath27 that intersect @xmath47 .
let us now suppose that @xmath27 is a loop configuration and @xmath47 is a set satisfying the same conditions as before , except that we also assume that it is at positive distance of the origin .
we can now define chordal @xmath42-explorations of @xmath279 that are _ rooted at @xmath45 _ : these are just the image under @xmath505 of the previous explorations of @xmath727 , where @xmath728 denotes the loop configuration obtained when one considers the loops @xmath729 .
the difference between the exploration rooted at the origin and the exploration rooted at infinity in fact only lies in what we call @xmath730 , and how one renormalizes the conformal maps @xmath731 at each step . in the exploration rooted at the origin
, @xmath706 would have to be replaced by the conformal image of @xmath730 under the conformal mapping from @xmath37 onto @xmath37 that fixes @xmath118 , has derivative @xmath3 at @xmath118 and maps @xmath702 onto @xmath45 , and the conformal map @xmath732 would have to be replaced by the conformal map @xmath733 from @xmath37 onto @xmath734 such that @xmath735 when @xmath662 .
let us describe more precisely an iteration step in this case : let us define , for each @xmath736 and each small @xmath42 , @xmath737 to be the image of @xmath440 under the moebius transformation @xmath738 of @xmath37 that maps @xmath45 onto @xmath118 and @xmath702 onto @xmath45 .
note that for all @xmath118 and @xmath188 in @xmath739 , @xmath740 when @xmath662 and that @xmath741 .
when @xmath42 is very small , @xmath737 is very close to the semi - circle of radius @xmath742 around @xmath118 ( i.e. , it is squeezed between to semi - circles of radii close to @xmath742 ) because @xmath743 .
suppose that @xmath734 , @xmath733 are already defined .
the choice of @xmath744 is then said to be @xmath42-admissible if @xmath745 .
we then define @xmath718 to be the unbounded connected component the domain by removing from @xmath746 all the loops that intersect @xmath747 .
clearly , after the last step @xmath213 of this exploration rooted at the origin ( when no @xmath42-admissible point can be found ) , we again have a set @xmath748 that is in some sense close to @xmath49 ( which is the unbounded connected component of the domain obtained when removing from @xmath653 all loops that intersect @xmath47 ) .
one way to make this more precise is to use the carathodory topology `` seen from the origin '' . even if the origin is a boundary point of the simply connected domains @xmath748 and @xmath49 , we can symmetrize these domains ( by considering their union with their symmetric sets with respect to the real axis ) , and therefore view the conformal maps @xmath749 and @xmath750 as conformal maps normalized at the origin , which is now an inner point of the domain .
we will implicitely use this topology for domains in the upper half - plane `` viewed from the origin '' .
then , just as in subsection [ explodet ] , we get that ( for each given @xmath27 and @xmath47 ) @xmath748 converges to @xmath49 , when @xmath43 , uniformly with respect to all possible @xmath42-admissible explorations . we know that the @xmath70-mass of the set of pinned loops that intersect the segment @xmath751 $ ] is positive and finite . by scaling
, we can choose @xmath59 in such a way that the @xmath70-mass of the set of pinned loops that intersect @xmath752 $ ] is equal to @xmath3 .
on this set of configurations , @xmath70 can therefore be viewed as a probability measure . similarly , the pinned configuration measure @xmath685 restricted to the set @xmath753 ) = \ { ( \bar \gamma , \bar \gamma ) \ : \
\bar \gamma \cap [ a , a+ia ] \not= \emptyset \}\ ] ] is a probability measure that we will denote by @xmath754}$ ] .
let us now consider a pinned configuration @xmath755 in @xmath477 .
define @xmath756 so that @xmath757 is the lowest point of @xmath758 $ ] .
let @xmath21 denote the conformal map from the unbounded connected component @xmath759 of the set obtained by removing from @xmath760 $ ] all the loops of @xmath679 that intersect this segment back onto @xmath37 , normalized by @xmath761 , @xmath762 and @xmath763 .
we then define @xmath764 to be the distribution of @xmath765 , i.e. the image of the probability measure @xmath766}$ ] under the transformation @xmath767
. a by - product of the coming arguments will be that in fact : [ p81bis ] under @xmath754}$ ] , the loop @xmath768 is independent of @xmath21 .
define on the other hand the probability measure @xmath769 to be the measure @xmath70 restricted to the set of pinned loops that intersect @xmath770 , and renormalized in order to be a probability measure .
the main first goal of this subsection is to derive the following proposition ( which we will later prove together with proposition [ p81bis ] ) : [ p81 ] as @xmath43 , @xmath769 converges to @xmath764 .
as we have already indicated , the basic idea of the proof will be similar in spirit with the construction of the pinned measure itself : we will use discrete explorations of pinned configurations , and use the restriction property of the pinned measure in order to deduce some independence property between the already discovered loops and the yet - to - be - discovered ones along the exploration ; this will enable us to conclude .
suppose now that @xmath47 is a set that is at positive distance from the origin , and that satisfies the conditions 1 - 4 described in the previous subsection . clearly , if we define @xmath771 then @xmath772 is finite and positive .
we can therefore define @xmath773 to be @xmath685 restricted to this set , and renormalized in order to be a probability measure .
suppose now that @xmath532 is a simply connected domain with @xmath774 .
define @xmath775 to be the conformal map from @xmath49 onto @xmath37 such that @xmath776 when @xmath777 , and suppose that @xmath778 is a set such that @xmath779 satisfies the same conditions as before .
we then define @xmath780 to be the image of @xmath773 under @xmath781 .
basically , @xmath780 is the `` law '' of a pinned configuration in @xmath49 rooted at the origin and `` conditioned '' on the event that the pinned loop intersects @xmath84 .
suppose now that @xmath317 is another set at positive distance of the origin that satisfies the four conditions described in the previous subsection and such that @xmath782 .
we are now going to define , for each such @xmath317 and @xmath47 , the measure @xmath783 to be the measure @xmath685 on pinned configurations @xmath677 , restricted to the set @xmath784 of configurations such that @xmath678 intersects @xmath317 but not @xmath47 .
when @xmath785 , we then define @xmath786 to be @xmath787 normalized to be a probability measure . finally , when we are given @xmath679 and @xmath47 , we can define the set @xmath788 as before ( i.e. , the unbounded connected component of the set obtained by removing from @xmath653 all the loops of @xmath679 that intersect @xmath47 ) .
then , it follows immediately from the pinned measure s restriction property ( proposition [ p72 ] ) that : [ c555 ] suppose that @xmath677 is sampled according to @xmath789 .
then the conditional law of @xmath678 given @xmath790 ( i.e. , the loops of @xmath679 that intersect @xmath47 ) is @xmath791 where @xmath792 .
suppose now that we are given a configuration @xmath677 that is sampled according to @xmath793 .
for each given @xmath42 , we can perform a `` markovian '' @xmath42-admissible exploration as before ( markovian can simply mean here that we have chosen some deterministic procedure to choose each @xmath744 ) .
when @xmath678 intersects @xmath47 , this exploration procedure can discover @xmath678 at some ( random ) step that we call @xmath794 , and we know that when @xmath43 , the probability that @xmath795 exists tends to @xmath3 .
corollary [ c555 ] shows that conditionally on @xmath796 , on @xmath734 and on @xmath744 , the law of @xmath678 is @xmath797 .
this implies in particular that the conditional law of @xmath798 is @xmath769 ( recall that @xmath799 is just the conformal map from @xmath800 onto @xmath37 that maps the origin onto itself , has derivative at the origin equal to @xmath3 , and that maps @xmath801 onto @xmath3 ) . hence ( on the event @xmath802 ) , @xmath803 is in fact independent of @xmath804 and of @xmath800 .
it is worth stressing at this point that the law of @xmath805 is not geometric as in the cle case , and that the iteration steps are not i.i.d . anymore , but this will not prevent us from now proving propositions [ p81 ] and [ p81bis ] : the arguments are again close in spirit to the definition of the pinned measure itself .
suppose that @xmath327 is a large integer , that @xmath806 , and consider the rectangles @xmath807 \times [ 0 , ak \delta]$ ] for @xmath808 .
note that because of scale - invariance and translation invariance , for each given @xmath335 , up to a set of configurations of zero measure , all the loops of @xmath677 that intersect some @xmath809 also intersect its interior ( when @xmath335 is fixed ) .
furthermore , the probability measure @xmath810 converges to @xmath811}$ ] . for each fixed @xmath812 , we can start to explore each @xmath813 by a markovian chordal @xmath42-admissible exploration rooted at @xmath45 , then continue exploring @xmath814 and so on until we either complete a chordal @xmath42-admissible exploration of @xmath815 \times [ 0 , a]$ ] , or we have discovered @xmath678 .
when @xmath677 is sampled according to @xmath764 , the probability to discover @xmath678 is going to @xmath3 as @xmath466 .
let us call ( as before ) @xmath816 the ( random ) step at which @xmath678 is discovered .
furthermore , for each given @xmath335 , it is possible to choose @xmath42 small enough , so that the probability that the exploration processes misses no loop of @xmath679 of diameter greater than @xmath817 that intersects @xmath818 $ ] is as close to @xmath3 as we want . the very same arguments as in the proof of proposition [ pwellchosen ] then show that if @xmath398 is chosen to be sufficiently small , then the set @xmath800 converges to @xmath759 almost surely , and that @xmath819 converges almost surely to @xmath21 .
we then conclude using the independence between @xmath819 and @xmath820 .
we safely leave the details to the reader .
a consequence of proposition [ p81 ] is that the probability measure @xmath764 is invariant under all moebius transformations of @xmath37 onto itself that have @xmath45 and @xmath3 as fixed points .
this follows immediately from the conformal covariance property of @xmath70 and the description of @xmath764 as limits of @xmath769 given in the proposition .
in fact , our proof of proposition [ p81 ] works also if we replace the straight segment @xmath752 $ ] by any given other piecewise linear path starting at @xmath3 such that the @xmath70 mass of the set of pinned loops that it intersects is finite ( we can then just multiply @xmath70 by a constant to turn it into a probability measure on this set ) : suppose that @xmath178 is some finite piecewise linear path that starts on the real line , has no double points , and such that all of its segments are horizontal or vertical .
we parametrize it `` from the real axis to the tip '' . for each pinned configuration @xmath677 such that @xmath821
, we can define the first meeting time @xmath318 of @xmath178 with @xmath678 , i.e. @xmath822 .
let us now consider @xmath823 to be the set of loops of @xmath679 that intersect @xmath824 $ ] , and let @xmath21 denote the conformal map as before ( that maps the set obtained by removing from @xmath825 $ ] all the loops of @xmath679 that intersect @xmath824 $ ] back onto @xmath37 , normalized by @xmath761 , @xmath762 and @xmath826 .
note that @xmath827 is a pinned loop ( the map @xmath21 is smooth near the origin ) .
note also that @xmath828 is positive and finite .
then : if we consider the measure @xmath685 on the event @xmath829 and if we renormalize it to be a probability measure , then under this probability measure , @xmath765 is distributed according to the two - point pinned distribution @xmath764 .
furthermore , @xmath768 is independent of the conformal map @xmath21 .
we are now going to derive the key independence property of the two - point pinned measure .
suppose that @xmath768 is a two - point pinned loop in @xmath37 that touches the real line at @xmath45 and @xmath3 .
we can define two simple paths @xmath830 and @xmath831 from @xmath45 to @xmath3 in @xmath37 in such a way that their union is @xmath768 and that @xmath831 is `` below '' @xmath832 .
we now define @xmath833 to be the conformal map from the unbounded connected component @xmath834 of @xmath835 onto @xmath37 that fixes the three boundary points @xmath45 , @xmath3 and @xmath702 , and finally , we define @xmath836 .
( sketch),width=480 ] [ pkey ] if @xmath768 is chosen according to the two - point pinned measure @xmath764 , then @xmath837 and @xmath831 are independent .
recall on the one hand that @xmath764 is the limit of @xmath769 when @xmath43 .
let us now consider a hull @xmath703 that is attached to the interval @xmath838 that satisfies the conditions 1 - 4 as before .
let @xmath839 where @xmath42 is sufficiently small , so that @xmath840 .
let us now describe the law of @xmath768 conditioned to avoid @xmath47 , trying to use corollary [ c555 ] .
remark first that it is the limit when @xmath43 of the pinned measure `` renormalized in order to be a probability '' on the event @xmath841 i.e. on the event where @xmath98 intersects @xmath770 but not @xmath47 . but this measure can be described using corollary [ c555 ] , and we could use this to prove proposition [ pkey ] directly . let us however describe first the `` explicit restriction - type - property '' for the law @xmath764 of the two - point pinned loop @xmath768 .
suppose that @xmath47 is such that @xmath842 , and that @xmath653 is simply connected .
we would like describe the conditional law of @xmath768 given that it does not intersect @xmath47 . in order to do so ,
let us use a cle @xmath655 that is independent of @xmath768 .
we will use @xmath843 , @xmath844 and @xmath845 to denote the expectation with respect to @xmath768 , to @xmath655 and to both .
we define @xmath656 , the unbounded connected component of the set obtained by removing from @xmath653 all the loops of @xmath655 that intersect @xmath47 .
then , we define the conformal map @xmath846 from @xmath656 onto @xmath37 such that @xmath847 , @xmath848 and @xmath849 .
it is easy to check that @xmath850 .
the loop @xmath851 is clearly two - point pinned loop that avoids @xmath47 almost surely .
[ explicit ] the conditional law of @xmath768 given @xmath852 satisfies , for a bounded continuous function @xmath503 on the set of loops , @xmath853 proposition [ p81 ] shows that @xmath854 the measure @xmath70 restricted to the set of loops that do not intersect @xmath47 , is equal to @xmath669 by proposition [ mua ] . recall that in order to define @xmath669 , we use another cle @xmath655 , that we consider the set @xmath656 and the conformal map @xmath657 from @xmath37 onto @xmath656 such that @xmath855 , @xmath856 and @xmath857 , and that @xmath669 is the image of the product measure @xmath665 under the mapping @xmath858 .
hence , @xmath859 but when @xmath42 is very small , the @xmath42-neighborhood of @xmath3 is not much distorted by @xmath846 .
furthermore , @xmath860 is a conformal mapping from the upper half - plane into itself , that depends on @xmath655 only , that maps the origin onto itself and its derivative at the origin is equal to @xmath861 .
recall also that @xmath70 satisfies conformal covariance with the exponent @xmath462 .
it follows easily by dominated convergence that @xmath862 we now use this to derive proposition [ pkey ] : let us consider an independent cle @xmath655 as before , and for each @xmath47 as before , such that @xmath842 , define @xmath846 .
it is important to observe that if @xmath768 is a two - point pinned loop , then @xmath863 is always a two - point pinned loop , and that @xmath864 .
furthermore , note that the event @xmath865 is identical to the event that @xmath866 .
hence , we get that for all continuous bounded function @xmath503 on the space of loops ( recall that @xmath98 and @xmath846 are independent ) @xmath867 since this is true for all such @xmath47 , it follows that @xmath837 and @xmath831 are indeed independent . in the sequel , we call @xmath868 the law of @xmath837 .
this is a probability measure on paths joining @xmath3 to @xmath45 in the upper half - plane .
note here that @xmath868 is invariant under the moebius transformations from @xmath37 onto itself that preserve @xmath45 and @xmath3 .
this follows easily from the corollary and from the fact that the same is true for the two - point pinned measure .
let us define @xmath869 to be the image of @xmath868 under the conformal map @xmath870 of @xmath37 onto itself ( that maps @xmath3 to @xmath45 and @xmath45 to @xmath702 ) .
@xmath869 is then a probability measure on simple paths in @xmath37 from @xmath45 to infinity , that is scale - invariant ( because the multiplications by positive constants are the moebius transformations in @xmath37 that leave @xmath45 and @xmath702 invariant ) .
the goal of this section is show that the pinned measure @xmath70 associated to a cle is necessarily one of the `` sle excursion measures '' .
let us first combine the results of the previous section and reformulate them into a single tractable statement : consider the pinned configuration measure @xmath685 and some finite piecewise linear path @xmath178 in @xmath37 that starts on the positive real half - line , has no double points , and such that all of its segments are horizontal or vertical .
we parametrize it `` from the real axis to the tip '' .
we also suppose that @xmath871 so that we can view @xmath70 ( and @xmath685 on the corresponding set of configurations ) as a probability measure . for each pinned configuration @xmath677 such that @xmath821 , we can define the first meeting time @xmath318 of @xmath178 with @xmath678 , i.e. @xmath822 .
let us now consider @xmath823 to be the set of loops of @xmath679 that intersect @xmath824 $ ] .
we also call @xmath872 the part of @xmath98 ( when oriented counterclockwise ) between @xmath45 and @xmath873 , and @xmath874 the other part .
let us now define the set @xmath875 to be the unbounded connected component of the set obtained by removing from @xmath37 the union of @xmath872 , @xmath824 $ ] and the loops of @xmath823 .
we let @xmath876 denote the conformal map from @xmath875 onto @xmath37 such that @xmath877 , @xmath878 and @xmath879 .
note that @xmath876 , @xmath872 and @xmath823 are all deterministic functions of the triple @xmath880 .
we will sometimes write @xmath881 and @xmath882 , @xmath883 to indicate the dependence in @xmath178 .
[ propo ] @xmath884 is independent of @xmath876 and @xmath872 , and its law is @xmath868 .
this is a direct combination of our previous results .
let @xmath885 denote the conformal map from @xmath886 \cup \tilde \gamma_\eta)$ ] onto @xmath37 with @xmath887 , @xmath888 and @xmath889 .
conformal invariance of @xmath764 under the maps that preserve @xmath45 and @xmath3 , and the definition of the two - point pinned measure shows that @xmath890 is independent of @xmath885 and that its law is the two - point pinned measure @xmath764 .
if we now define @xmath891 and @xmath892 , and @xmath833 ( which depends on @xmath831 alone ) as before , we know from proposition [ pkey ] that @xmath893 is independent of @xmath831 , and its law is @xmath868 .
hence , @xmath894 is independent of @xmath895 .
we can now conclude , noting that @xmath896 and that @xmath897 .
suppose now that @xmath898 \to { \mathbb{h}}\cup \eta ( 0)$ ] is a finite piecewise linear path as before , except that @xmath899 \not= \emptyset \ } ) $ ] is not equal to one .
we can use scaling in order to find some constant @xmath447 , in such a way that we can apply the previous statement to @xmath900 , and it therefore follows that proposition [ propo ] still holds if we consider the measure @xmath70 on the event @xmath901 \not= \emptyset \ } $ ] , and renormalized in order to be a probability measure on this event . since this is true for all @xmath902 , it also follows that : [ finalco ] the same statement holds if we consider the measure @xmath70 restricted to the event that @xmath98 intersects only the last segment of @xmath178 ( and not the previous ones ) , and if we renormalize the measure @xmath70 to be a probability measure on this event .
we now want to deduce from this corollary that @xmath868 is the law of some sle@xmath1 from @xmath3 to @xmath45 in @xmath37 i.e. , that @xmath869 is the law of some chordal sle@xmath1 in @xmath37 .
we will use oded schramm s conformal markov property characterization of sle that we now briefly recall .
suppose that one is given a continuous simple curve @xmath903 in @xmath904 starting at @xmath45 .
let us parametrize @xmath903 according to its half - plane capacity , i.e. , in such a way that for each @xmath565 , there exists a conformal map @xmath905 from @xmath906 $ ] onto @xmath37 such that @xmath907 as @xmath713 .
suppose that the half - plane capacity of @xmath903 is not bounded ( i.e. , that @xmath903 is defined for all @xmath908 ) , and define @xmath909 , i.e. the conformal map from @xmath910 onto @xmath37 with @xmath911 and @xmath912 at infinity .
then : [ schmarkov ] if the law of @xmath903 is scale - invariant in distribution ( i.e. , the law of @xmath913 does not depend on @xmath447 ) , and if for all @xmath908 , the conditional law of @xmath914 given @xmath915 $ ] is identical to the initial law of @xmath903 , then @xmath903 is an sle curve .
recall that this result can be easily understood using loewner s theory of slit mappings that shows that the curve @xmath903 is characterized by the function @xmath916 .
indeed , the previous conditions imply on the one hand that the random function @xmath917 has the brownian scaling property ( because @xmath903 is scale - invariant in distribution ) and on the other hand that it is a continuous process with independent increments .
these two fact imply that @xmath918 it is a multiple of a standard brownian motion ( and one can call this multiplicative constant @xmath919 ) .
as the function @xmath918 fully characterizes @xmath903 , this determines the law of the path @xmath903 ( up to this one - parameter choice ) .
furthermore , sle computations ( see @xcite ) show that the fact that the curve is simple implies that @xmath57 . there are several natural ways that one can use in order to parametrize a pinned loop .
we are going to choose one that is tailored for our purpose , i.e. , to recognize sle excursions .
let us first define @xmath920 in such a way that @xmath921 then , the measure @xmath70 restricted to the set @xmath922 is a probability measure that we will call @xmath119 in the present section .
until the rest of the present subsection , we will assume that @xmath98 is sampled from this probability measure .
suppose now that @xmath923 and let us orient it `` anti - clockwise '' , i.e. , @xmath98 starts at @xmath45 , makes a simple anti - clockwise loop in @xmath37 and comes back to @xmath45 .
on the way , there is the first intersection point @xmath924 of @xmath98 with @xmath925 .
we define @xmath926 the beginning part of the loop between @xmath45 and @xmath924 ( with the anti - clockwise orientation ) and we call @xmath927 the other ( end-)part of the loop @xmath98 between @xmath924 and @xmath45 .
let @xmath928 denote the conformal mapping from @xmath929 onto @xmath37 normalized by @xmath930 , @xmath931 and @xmath932 ( where @xmath933 denotes the image of the left - limit of @xmath45 in @xmath934 ) .
the end - part @xmath935 of the loop between @xmath924 and @xmath45 is a continuous simple path from @xmath924 to @xmath936 in @xmath937 .
hence , its image under @xmath928 is a simple path from @xmath45 to infinity in the upper half - plane that we now call @xmath903 .
we parametrize @xmath903 according to its half - plane capacity as before ( and define the conformal maps @xmath905 and @xmath938 ) .
this therefore defines ( via the map @xmath939 ) a parametrization of the path @xmath940 . and
@xmath941 ( sketch),width=403 ] for each @xmath908 , we now define @xmath942 , @xmath943 \hbox { and } e_t : = e_t ( \gamma ) = e_0 [ t , \infty).\ ] ] let us insist on the fact that @xmath944 and @xmath945 are paths ( and not points ) .
@xmath946 will denote the conformal map from @xmath947 with @xmath948 , @xmath949 and @xmath950 .
we will now prove the following statement that we will then combine with lemma [ schmarkov ] : under the probability measure @xmath119 , for each given @xmath908 , @xmath951 is independent of @xmath941 .
furthermore , its law is @xmath869 .
let us fix @xmath908 .
define @xmath952 the conformal map from @xmath947 on @xmath37 with @xmath953 , @xmath954 , @xmath955 .
then , the lemma is clearly equivalent to the fact that @xmath956 is independent of @xmath941 , and that its law is @xmath957 .
define @xmath958 as before ( when @xmath98 is defined under the probability measure @xmath119 ) .
the proof will be based on the independence property of the two - point pinned measure . in order to use this
, we will need to discover the pinned loop via some grid - based path that turns out to be close to @xmath944 .
take @xmath715 .
let us try to associate to each @xmath959 a grid approximation of @xmath944 `` from the right side ''
, that we will call @xmath960 . first of all , let us note that the diameter of @xmath961 goes to @xmath45 almost surely when @xmath212 and that we can find some sequence @xmath612 such that @xmath962 ( this follows easily from the fact that @xmath98 is @xmath119-almost surely a simple curve ) , where here @xmath99 denotes the usual euclidean distance . then
, when @xmath335 is a mesh - size , we try to define a particular grid - approximation of @xmath963 on the grid @xmath964 as follows : delete all the edges from the grid that are distance less than @xmath335 from @xmath963 , and consider the unbounded connected component @xmath84 of the graph that one obtains .
let @xmath965 denote the first edge of @xmath84 that @xmath98 hits ( it is therefore after `` time '' @xmath966 ) .
it is clearly on the `` boundary of @xmath84 seen from @xmath963 '' , and one can find the simple nearest - neighbor path from the positive real axis to @xmath965 ( where @xmath965 is its last edge ) on @xmath84 that is `` closest '' to @xmath963 .
we call this path @xmath967 .
( sketch),width=384 ] clearly , when @xmath565 , @xmath295 and @xmath60 are fixed , then when @xmath388 , the ( hausdorff ) distance between @xmath462 and @xmath60 is going to @xmath45 .
hence , we can choose @xmath968 in such a way that @xmath969 ( where @xmath970 denotes the hausdorff distance ) .
once these sequences @xmath968 and @xmath971 are chosen , we denote for each given @xmath295 , the grid approximation @xmath972 by @xmath960 .
the previous estimates also ensure that @xmath973 furthermore , we can observe that if we are given @xmath944 and a finite grid - path @xmath178 on @xmath974 , it suffices to look at the path @xmath944 only until the moment it hits @xmath178 for the first time to check whether @xmath975 or not .
suppose now that @xmath178 is some given grid - path that has a positive probability to be equal to @xmath960 .
let us consider a pinned configuration @xmath976 such that @xmath98 intersects @xmath977 , and let us define @xmath978 , @xmath979 and @xmath980 as before ( recall that @xmath872 and @xmath874 are the two parts of @xmath98 , before and after it hits @xmath873 , which is the first point of @xmath178 that intersects @xmath98 , and that @xmath876 corresponds to the domain where one from @xmath981 $ ] the loops of @xmath679 that intersect @xmath982 ) .
then , we know on the one hand that @xmath983 is independent of @xmath876 and @xmath872 , and that its law is @xmath868 . on the other hand , we have just argued that the event @xmath984 is measurable with respect to @xmath872 ( this is because @xmath872 contains the part of @xmath98 up to the first time it hits @xmath178 ) , so that the conditional law of @xmath884 given @xmath985 is also @xmath868 . when @xmath986 , we define @xmath987 to be this map @xmath876 , and we also define @xmath988 and @xmath989 to be these @xmath874 and @xmath872 .
note that when @xmath295 is very large , the probability that @xmath960 intersects some macroscopic loop of @xmath679 is very small ( this is because they are all at positive distance of @xmath678 , and that @xmath960 is very close to @xmath98 ) .
this , and our definition of @xmath960 ensures readily that @xmath990 converges ( in hausdorff topology ) to @xmath944 , and that @xmath991 converges to @xmath992 .
we are now almost ready to conclude : for any continuous bounded functions @xmath503 and @xmath750 with respect to the hausdorff topology , we get that : @xmath993 where we have use the independence property of corollary [ finalco ] between the second and the third line .
this lemma has a number of consequences : it implies ( taking @xmath994 ) that the law of @xmath903 itself is @xmath869 .
we know also that @xmath869 is scale - invariant .
hence , we get that @xmath903 is a continuous simple curve from @xmath45 to infinity in the upper half - plane , that is scale - invariant in law , and such that for all @xmath565 , the conditional law of @xmath995 given @xmath915 $ ] is the same as the law of @xmath903 itself .
the previous characterization of sle therefore implies that : [ recognize ] the curve @xmath903 is an sle@xmath1 for some @xmath996 $ ] .
furthermore , @xmath903 is independent of @xmath997 .
the independence between @xmath998 and @xmath903 shows that the probability measure @xmath119 can be constructed as follows : first sample @xmath998 ; this defines its tip @xmath924 .
then draw an sle@xmath1 ( for some given value of @xmath6 ) from @xmath924 to @xmath936 in @xmath999 .
standard sle computations ( see @xcite ) show that the probability that an sle@xmath1 from @xmath3 to @xmath45 in @xmath37 hits the ( semi-)circle of radius @xmath1000 around the origin decays like @xmath1001 when @xmath1002 .
it follows immediately that if @xmath6 is the value associated to the pinned measure @xmath70 , @xmath1003 as @xmath1002 .
if we compare this with the scaling property of @xmath70 , i.e. , the fact that @xmath1004 we get that : one necessarily has @xmath1005 , @xmath1006 and @xmath15 $ ] . the last two statements in the corollary are just due to the fact that we know already that @xmath526 ( so that @xmath1007 ) and also that @xmath57 ( because @xmath98 does not touch the boundary elsewhere than at the origin ) , which implies that @xmath1008 . in order to show that corollary [ recognize ] in fact characterizes the measure @xmath70 , let us first recall the following rather standard facts concerning sle processes and excursions of bessel processes : [ l.besexc ] suppose that @xmath15 $ ] , and that @xmath1009 denotes the law of an sle@xmath1 from @xmath42 to @xmath45 in the upper half - plane .
then , the measures @xmath1010 converge ( vaguely ) to an infinite measure @xmath1011 on the set of pinned loops , that we call the sle@xmath1 excursion measure .
vague convergence here means that for all @xmath610 , when restricted on the set of paths with diameter greater than @xmath174 , @xmath1012 converges weakly ( with respect to the hausdorff topology for instance ) .
let us fix a positive @xmath42 and consider an sle@xmath1 curve ( for @xmath1013 $ ] ) from @xmath42 to @xmath45 in @xmath37 .
it is possible to parametrize this curve as a loewner chain in the upper half - plane , with time measured via the half - plane capacity .
if we use this parametrization , then the sle path is defined up to some ( random ) time @xmath1014 , and one can define for each @xmath908 , the conformal map @xmath905 from @xmath1015 $ ] normalized at infinity , and it satisfies loewner s equation @xmath1016 for all @xmath1017 and @xmath1018 $ ] , where @xmath1019 .
note also that @xmath1014 is the hitting time of @xmath45 by @xmath1020 ( because @xmath98 is almost surely a simple curve ) .
it is quite easy to argue that the process @xmath1021 is necessarily a markov process with the same scaling property as brownian motion , and therefore the multiple of some bessel process .
in fact , direct computations ( see for instance @xcite ) shows that @xmath98 is a so - called sle(@xmath145 ) process stopped at its first swallowing time of @xmath45 , i.e. , that @xmath1022 and @xmath1023 where @xmath317 is a standard brownian motion .
note that @xmath136 can be recovered from @xmath566 because @xmath1024 an important feature here is ( and this follows immediately from it s formula ) that for @xmath1025 , @xmath1026 is a local martingale up to its first hitting time of @xmath45 .
for instance , when @xmath1027 , then the probability that @xmath566 hits @xmath1028 before hitting @xmath45 is @xmath1029 .
this makes it possible ( in the standard procedure to define bessel excursions , see for instance @xcite ) to define a measure @xmath1011 on the space of one - dimensional excursions @xmath1030
( i.e. , such that @xmath1031 and @xmath1032 ) in such a way that the @xmath1011-mass of the set of paths that reach level @xmath42 ( at some time @xmath1033 ) is @xmath1034 , and that the @xmath1035-law of @xmath1036)$ ] is the law of @xmath566 as before started from level @xmath42 and stopped at its first hitting of @xmath45 .
furthermore , one can check ( for instance via scaling considerations ) that for @xmath1011-almost all @xmath1037 , @xmath1038 .
each such excursion @xmath1037 clearly defines ( via loewner s equation , replacing @xmath566 by @xmath1037 and using the fact that @xmath1039 is finite ) a two - dimensional pinned loop @xmath98 .
so , we can view the measure @xmath1011 as a measure on the set of loops , and it satisfies the following two statement : * the @xmath1011-mass of the set of pinned loops @xmath98 such that @xmath1037 hits @xmath42 is @xmath1034 .
* on this set , the measure @xmath1040 is a probability measure , and one can sample @xmath98 by first sampling @xmath1041 $ ] , and then finishing the curve by an sle@xmath1 in @xmath1042 $ ] from @xmath1043 to @xmath45 .
it follows that the difference between the two probability measures @xmath1044 and @xmath1009 is just due to the mapping by the random conformal mapping normalized at infinity from @xmath1042 $ ] onto @xmath37 .
but , when @xmath43 , this mapping clearly converges to the identity away from the origin , which proves the claim . [ muissleexcursion ] the pinned measure @xmath70 is the multiple of the sle@xmath1-excursion measure @xmath1011 for @xmath1045 ( where @xmath462 is the `` exponent '' associated to @xmath70 ) , that is normalized in such a way that @xmath1046 .
a consequence of this proposition is that two cle s with the same exponent @xmath462 have the same pinned measure .
recall that we have the following description of @xmath70 ( this is just the combination of the scaling property of @xmath70 and of the definition of @xmath920 ) : for each @xmath1047 , the @xmath70-mass of the set of loops with radius at least @xmath1048 is @xmath1034 .
if we restrict ourselves to this set of loops , and renormalize @xmath70 so that it is a probability measure on this set , we can first sample the part of @xmath98 up to the point at which it reaches the circle of radius @xmath42 , and then complete it with an sle@xmath1 back to the origin in the remaining domain .
we can note that for some absolute constant @xmath1049 , if we consider a loop that does not reach the circle of radius @xmath920 , then the corresponding loewner - type function @xmath1037 as before does not reach @xmath1049 .
let us now sample the beginning of the loop @xmath98 as before up to the first hitting point of the circle of radius @xmath1048 .
then continue the sle up to the first time ( if it exists ) at which the function @xmath1037 associated to the loop reaches @xmath1050 .
a fixed positive fraction @xmath1051 ( independent of @xmath42 because of scale - invariance ) of these sles succeed in doing so . then , after this point , we still are continuing with an sle@xmath1 in the remaining domain , so that @xmath98 is close to a sample of @xmath1009 .
hence , we conclude readily that @xmath70 is a constant multiple of @xmath1011 ( where the constant is given in terms of @xmath1049 , @xmath1051 and @xmath462 ) . another way to describe
this constant is to recall that @xmath1046 .
let us now make the following comment : suppose that a cle in @xmath37 is given .
this defines a random collection of loops @xmath1052 .
if we now define the symmetry @xmath190 with respect to the imaginary axis ( i.e. the map @xmath1053 ) , then it is clear from our cle axioms that the family @xmath1054 is also a cle in @xmath37 .
we have just seen that each cle defines a pinned measure @xmath70 that happens to be the ( multiple of an ) sle@xmath1 excursion measure .
but , by construction , the pinned measure associated to the cle @xmath1054 is simply the image of @xmath70 by the map @xmath190 .
therefore ( noting that both these cles correspond to the same exponent @xmath462 ) , the cles @xmath1054 and @xmath1052 correspond to the same sle excursion measure .
hence : [ reversibility ] if the pinned measure of a cle is the sle@xmath1 excursion measure for some @xmath6 , then it implies that the law of the corresponding sle@xmath1 bubble ( as defined in the introduction ) is reversible : the trace of sle@xmath1 bubble traced clockwise has the same law as that of an sle@xmath1 bubble traced anti - clockwise .
it has been proved by dapeng zhan that sle@xmath1 for @xmath57 is indeed reversible ( see @xcite ) , which in fact implies this last statement .
the present set - up therefore provides an alternative approach to reversibility of sle paths for these values of @xmath6 .
the goal of the present section is to show that if one knows @xmath70 , then one can recover the law of the initial cle .
this will conclude the proof of theorem [ t2 ] .
the rough idea is to use the radial exploration and to use the same idea as in the proof of the fact that @xmath526 , but `` backwards '' : the fact that @xmath526 will ensure that this exploration ( described in section [ s.tail ] ) can be approximated by keeping only those steps that discover `` large '' loops , because the cumulative contribution of the small ones vanish .
furthermore , these large loops can be described using a poisson point process of intensity @xmath70 .
this will lead to the description of a `` continuous exploration mecanism '' that we will relate to sle(@xmath145 ) processes in the next section .
it is useful to first show how one can recover the law of the loop in the cle that contains @xmath438 .
recall that the set - up is the following : we suppose that we are given the law @xmath667 of a cle .
the previous sections enable us to define its pinned measure @xmath70 and we have seen that it is necessarily equal ( up to a multiplicative constant ) to the sle@xmath1 excursion measure for some @xmath6 in @xmath7 $ ] .
[ l.mugivesgammailaw ] the pinned measure @xmath70 characterizes the law of @xmath1055 under @xmath667 . in other words , if two ( laws of ) cles define the same pinned measure , then @xmath1055 is distributed in the same way for both cles .
let us use on the one hand the radial exploration mechanism in @xmath37 .
this defines ( for each @xmath42 ) the geometric number @xmath213 of exploration steps , the conformal maps @xmath1056 that are normalized at @xmath438 ( i.e. , @xmath1057 and the derivative at @xmath438 is positive ) .
we also define for all @xmath423 , @xmath1058 and @xmath1059 .
recall that : * the random variable @xmath213 is geometric with mean @xmath1060 . *
conditionally on the value of @xmath213 , the maps @xmath1061 are i.i.d .
( and their law does not depend on @xmath213 these are the maps corresponding to the cle exploration conditioned on the fact that @xmath439 does not intersect @xmath440 ) . *
what one `` discovers '' at the @xmath1062-th step is independent of the value of @xmath213 and of the maps @xmath1061 .
it is just a cle conditioned by the event that the loop surrounding @xmath438 intersects @xmath440 . on the other hand ,
define a poisson point process of pinned loops @xmath1063 with intensity @xmath70 , and let @xmath318 be the first time at which one @xmath1064 surrounds @xmath438 .
the @xmath70-mass of the set of loops that surround @xmath438 is equal to @xmath3 , so that @xmath318 is an exponential random variable of parameter @xmath3 ( see for instance @xcite for background on poisson point processes ) . for each of the countably many @xmath565 in @xmath1065 such that @xmath1064 exists ,
we denote by @xmath938 the conformal map normalized at @xmath438 by @xmath1066 and @xmath1067 , from the unbounded connected component of @xmath1068 onto @xmath37 . the fact that @xmath526 shows that @xmath1069 since @xmath1070 , the number of times @xmath565 before @xmath318 at which @xmath1071 is almost surely finite , so that almost surely , @xmath1072 is finite ( here and in this section , we shall re - use notations and arguments that have been presented in the `` stability of loewner chains '' paragraph of section [ s43 ] ) .
hence , if for each @xmath1073 , we define the iteration @xmath1074 of the finitely many @xmath938 s for @xmath1075 ( in their order of appearance , let us call the corresponding times @xmath1076 ) that correspond to pinned loops @xmath1077 s of radius greater than @xmath174 , we get that as @xmath1078 , the maps @xmath1074 converge ( in carathodory topology , with respect to the marked point @xmath438 ) to some conformal map @xmath1079 that can be interpreted as the iteration of all the conformal maps @xmath1080 in their order of appearance .
note that @xmath1079 maps some open set onto the upper half - plane , but that we have no information at this point on the regularity of the boundary of this open set ( whether it is a curve or not , etc . ) .
our goal now is to prove that when @xmath42 is small , then @xmath1079 and @xmath1081 are very close ( in law with respect to the carathodory topology from @xmath438 ) .
this will ensure that the law of @xmath439 in the cle is identical to that of @xmath1082 , i.e. the image under @xmath1083 of an independent sample of @xmath436 . in order to do so
, we are going to introduce the cut - off @xmath1074 as before . for each @xmath1073 , we define @xmath1084 the steps before @xmath213 at which one discovers a loop intersecting @xmath1085 ( note that this number of steps @xmath355 and the values of the @xmath1086 s are random , and depend on @xmath42 as well ) .
the description of @xmath70 shows that for each fixed @xmath610 , the joint law of @xmath1087 converges precisely to that of the corresponding @xmath1088 ; the law of their composition @xmath1089 therefore converges to that of @xmath1074 .
it now just remains to control @xmath1090 as @xmath1091 , uniformly with respect to @xmath42 .
we will bound the expected value of this quantity .
recall that @xmath526 , that the half - plane capacity of a set contained in a disc of radius smaller than @xmath174 around the origin is bounded by a constant @xmath84 times @xmath1092 , that @xmath213 follows a geometric random variable of mean @xmath1060 , and that conditionally on @xmath213 , the maps @xmath514 are i.i.d.it follows that @xmath1093 and that @xmath1094 this last expectation will be controlled thanks to the bound @xmath526 : let us first fix @xmath1095 and choose some @xmath1096 , for instance @xmath1097 .
we know that @xmath1098 in particular , for some @xmath1099 , we get that for all @xmath1100 , @xmath1101 hence , there exists a constant @xmath88 such that for all @xmath1102 @xmath1103 hence , if @xmath1104 , we get that @xmath1105 } \delta^{2j } p ( r ( \tilde \gamma ( \delta^{n_0 } ) ) \ge \delta^{j+1 } ) \\ & \le & c '' \sum_{j \in [ n_1 , n_0 ] } \delta^{2j } \frac { v ( \delta^{n_0-j-1 } ) } { u ( \delta^{n_0 } ) } \\ & \le & c ' '' \sum_{j \ge n_1 } \delta^{2j } ( \delta^{j})^{-2+a } \\ & \le & c ' '' r^a\end{aligned}\ ] ] where all the constants do not depend on @xmath174 and @xmath42 , so that this quantity converges to @xmath45 as @xmath1078 independently of @xmath42 ; this completes the proof of the fact that the law of @xmath1081 converges to @xmath1079 ( in carathodory topology ) .
recall that for each @xmath24 , the law of @xmath1055 is described as follows : we consider on the one hand a sample of the conformal map @xmath1106 , and on the other hand , an independent sample of the loop @xmath1107 that surrounds @xmath438 in a cle in @xmath37 conditioned to intersect @xmath440 . then
, @xmath1055 is distributed exactly as @xmath1108 .
we have just seen that @xmath1081 converges in law to @xmath1079 and we have also proved that the ( conditional ) law of @xmath1107 converges to @xmath436 .
it follows therefore that the law of @xmath439 is described as the image under @xmath1083 of an independent sample of @xmath436 .
this description is based solely on the measure @xmath70 , so that the law of @xmath439 can indeed be fully recovered from @xmath70 .
we may observe that the previous proof also shows that if @xmath1109 are other points in @xmath37 , if we restrict ourselves to the event where all these @xmath1110 other points are surrounded by some macroscopic loops that one discovered in this radial exploration mechanism before ( or at ) the @xmath213-th step , the joint distribution of these @xmath1110 loops ( some of which may be the same if some loops surround several of these points ) in the @xmath43 limit is also described via the poisson point process of pinned loops i.e. via @xmath70 .
we now want to prove the more general result : two cle probability measures that define the same pinned measure @xmath70 are necessarily equal .
in other words , the pinned measure @xmath70 characterizes the law of the corresponding cle .
this clearly implies theorem [ t2 ] .
note that the law of the cle is characterized by its `` finite - dimensional marginals '' , i.e. , by the joint distribution of @xmath1111 for any finite set of points @xmath1112 in @xmath37 ( if we consider the cle defined in the upper half - plane ) .
let us first focus on the law of @xmath1113 for some given @xmath38 ( the general case will be very similar ) . in order to motivate what follows ,
let us informally describe what could happen if we simply try to use exactly the same procedure as in the previous subsection .
that is , we consider the conformal maps @xmath1114 , @xmath1115 for @xmath423 as before .
let @xmath327 denote the largest @xmath1116 such that @xmath1117 ( and write @xmath1118 if @xmath40 is not swallowed during the first @xmath213 steps ) .
we want to keep track of what happens to @xmath40 , so we define @xmath1119 ( which will be defined up to @xmath346 ) .
let us now fix some small positive @xmath335 and consider @xmath1120 .
we will also introduce another step @xmath1121 , which is the first step @xmath355 at which either @xmath1122 or @xmath1123 . note that
if @xmath1124 , this means that after the @xmath354-th step , neither @xmath41 nor @xmath439 have been discovered , but that the points @xmath40 and @xmath438 are `` conformally '' quite far away from each other in the domain @xmath1125 ( i.e. , the green s function @xmath1126 is small recall that @xmath335 is small ) .
this can happen if the exploration gets close to one of the two points @xmath438 or @xmath40 in @xmath37 , or if it almost disconnects one from the other in @xmath37 ( and this is in fact a scenario that might really occur with positive probability , even in the @xmath388 limit ) .
we can subdivide the case @xmath1127 into three possibilities : the loop @xmath41 has been discovered before the @xmath213-th step ( in other words : @xmath1128 ) , the loop @xmath41 is discovered at the @xmath213-th step ( then with high probability @xmath1129 when @xmath43 ) , or it has still to be discovered after then @xmath213-th step .
in all these three cases , the arguments of the previous subsection allow us to conclude that the joint law of @xmath1130 on the event that @xmath1127 can be described via the poisson point process of pinned loops , i.e. thanks to the knowledge of @xmath70 .
when @xmath1124 , we could just try to continue the exploration process after @xmath354 , but at some later step @xmath355 , it could happen that the exploration procedure captures @xmath40 because @xmath1131 . or more generally , the loop @xmath41 could be discovered in the discrete exploration process at some step @xmath355 via some loop that intersects @xmath440 and that has a very small radius ( this does not necessarily contradict the fact the @xmath41 is at positive distance of @xmath40 ) . in such a case ,
the previous argument will clearly not work , as the exploration step corresponding to the discovery of @xmath41 would not be apparent in the limiting poisson point process of `` macroscopic '' pinned loops .
here is a simple way of fixing this : at the step @xmath354 , instead of doing an exploration using the semi - circle @xmath440 , we use the semi - circle @xmath1132 of radius @xmath1133 .
note that we are only exploring the loops that hit the semicircle ( not all the loops hitting the semi - disc ) and that @xmath1134 .
the harmonic measure of this semi - circle in @xmath37 seen from @xmath1135 or from @xmath438 is small ( if @xmath335 is small ) .
after this @xmath354-th step , @xmath40 and @xmath438 are then in two different domains .
furthermore , the probability to discover a loop that intersects also @xmath1136 or @xmath1137 at this @xmath354-th step is very small . from @xmath438 ( sketch).,width=480 ] [ imx9 ] the same arguments as above show convergence ( as @xmath1138 ) of the laws of the @xmath1139 ( in the carathodory sense , with normalization at @xmath438 ) . with high probability ,
the @xmath354-th exploration step changes the log conformal radius viewed from either @xmath40 or @xmath438 by a fraction that goes to @xmath45 when @xmath388 .
if we keep in mind that the restriction property still holds for non - simply connected subsets of @xmath37 ( this is the final observation of section [ s.2 ] ) , we see that we can then continue the discrete radial exploration ( aiming towards @xmath438 ) with @xmath42-semi - circles in the connected component containing @xmath438 after the @xmath354-th step , and independently another discrete radial exploration ( aiming at @xmath40 ) in the domain containing @xmath40 . in both cases , we continue until we discover @xmath439 and @xmath41 , and the same arguments as above allow to approximate the laws of these two loops in the two respective domains via two poison point processes of loops
. we can therefore conclude that up to an error that tends to @xmath45 with @xmath335 , the joint law of @xmath1140 on the event where @xmath1124 is also fully described via a procedure that is based on the knowledge of @xmath70 only .
hence , the joint law of @xmath1130 is fully determined by the pinned measure @xmath70 .
let us now consider the case where there is more than one additional point , i.e. , when we look at the joint law of @xmath1141 .
we then use the same argument as above , stopping at the first time that _ at least one _ of the images of the other points gets to a distance less than @xmath335 from the origin , i.e. , at the step @xmath1142 where @xmath1143 .
note that at this step , it could happen that several @xmath1144 are in fact quite small simultaneously . if we would explore as before by cutting out the semi - circle @xmath1145
, we might be unlucky and have some @xmath1146 that lies very close to this semi - circle .
however , it is clear that for at least one semi - circle @xmath1147 out of the @xmath1110 semi - circles , @xmath1148 , @xmath1149 , , @xmath1150 , the harmonic measure of @xmath1151 in @xmath37 seen from _ any _ of the @xmath1110 points @xmath1152 is very small ( i.e. , tends uniformly to @xmath45 when @xmath388 ) .
we therefore choose to use this particular semi - circle for the @xmath1153-th exploration step .
this reduces the problem to two independent explorations in two new domains as before , and each of the two domains contains strictly less than @xmath1154 points .
we can then inductively continue the discrete exploration procedure . for each fixed @xmath335
, this shows that up to a small error ( that vanishes when @xmath388 ) , the @xmath43 limit of this exploration procedure is well - approximated via poisson point processes with intensity @xmath70 .
this enables to conclude that the joint law of the loops @xmath1141 can be fully described in terms of @xmath70 .
it now remains to make the connection with the cle@xmath1 families defined in @xcite via sle(@xmath145 ) branching trees .
this will prove of theorem [ t3 ] .
we first briefly review some of the properties of sle(@xmath145 ) , in the case where @xmath57 that we are focusing on in the present paper .
we refer to @xcite for the more precise statements and their proofs .
we have already encountered the sle(@xmath1155 ) process in section [ pinnedsle ] . recall that a process @xmath1156 started from @xmath1157
is called a bessel process of dimension @xmath1158 if it is the solution to the ordinary differential equation @xmath1159 and that this process is well - defined up to its first hitting time of the origin ( which is almost surely finite if @xmath1160 ) . when @xmath1161 , it is possible to define it ( uniquely ) also after it hits the origin , in such a way that @xmath1162 when @xmath908 and that the lebesgue measure of the time spent by @xmath566 at the origin is 0 .
this is the `` instantaneously reflected bessel process '' , that can also be defined by concatenating a poisson point process @xmath1163 of bessel excursions ( see for instance @xcite for background on bessel processes ) .
we will also use a variation of this process obtained when tossing a fair coin for each excursion of the bessel process in order to decide if it is positive or negative ( this process is therefore defined on a larger probability space than the brownian motion @xmath317 ) , which is the `` symmetrized bessel process '' . to each excursion @xmath1164 of the bessel process of dimension @xmath335 , we associate @xmath1165 the integral of @xmath1166 on this excursion . recall that @xmath1167 is a local martingale as long as @xmath1168 stays away from the origin ( for @xmath1169 ) ; it follows from a simple scaling argument that the @xmath1165 s are finite and that the intensity of the poisson point process @xmath1170 is a multiple of @xmath1171 .
it is already apparent in section [ pinnedsle ] that it will be essential to try to make sense of a quantity like @xmath1172 when @xmath566 is a bessel process .
in fact , we will need to focus on the case where @xmath1173 $ ] and @xmath1174 , where this integral is infinite as soon as @xmath566 starts touching the origin ( this is because @xmath1175 diverges so that almost surely , @xmath1176 ) .
there are two ways around it this problem .
the first one involves the notion of `` principal values '' and is described in @xcite .
the other one , which works for all @xmath1177 $ ] , is simply to consider the symmetrized bessel process that chooses its signs at random according to independent fair coin tosses on each excursion . in this case
, the integral @xmath1172 is not absolutely convergent , but the process @xmath1178 can be nevertheless easily defined ( this corresponds exactly to the existence of symmetric @xmath287-stable processes for @xmath1179 , even if @xmath287-stable subordinators do not exist when @xmath1180 ) .
we refer to @xcite for a more detailed description of the possible options .
what we will need here is that in all these cases , one defines a continuous markov process @xmath1181 such that : * @xmath566 is a solution to the sde ( [ sde ] ) when it is away from the origin . *
the lebesgue measure spent by @xmath566 at the origin is @xmath45 .
* on the set of times where @xmath1168 is away from the origin , @xmath1182 is differentiable and its derivative is @xmath1183 .
* @xmath1181 satisfies the brownian scaling property . the way to define an sle(@xmath145 ) process out of this couple @xmath1181 goes as follows : define for all @xmath908 , the continuous process @xmath1184 and construct the chordal loewner chain driven by this function , i.e. , solve the loewner differential equation for all @xmath38 @xmath1185 started from @xmath1186 in order to define a random loewner chain .
if we define @xmath1187 , then when @xmath1168 is away from the origin , i.e. , when @xmath1188 , one has @xmath1189 , i.e. , @xmath1190 follows the image of some boundary point under the loewner flow .
hence , putting our various equations together ( definitions of @xmath1168 , @xmath1191 , @xmath1190 , and the relation between @xmath6 and @xmath335 ) , we get that on the set of times where @xmath1192 , @xmath1193 this continuous driving function @xmath1194 defines a loewner chain , but it is not clear whether this chain is almost surely generated by a path . however , on those time - intervals where @xmath1195 , the law of @xmath1196 is ( locally ) absolutely continuous to that of a standard brownian motion , and the loewner chain is therefore tracing streches of simple paths during these intervals .
in fact , each of these intervals correspond to a _ simple loop _ traced by the sle(@xmath145 ) loewner chain .
we can say that the positive excursions of @xmath566 correspond to anti - clockwise loops while the negative ones correspond to clockwise loops .
we have already mentioned that such chordal sle(@xmath145 ) chains are of particular interest because of the following properties @xcite : they are target - independent , and if @xmath1197 , then the process locally continues exactly as an sle@xmath1 targeting @xmath1198 .
this makes it is possible ( see @xcite ) to construct a branching sle(@xmath1199 ) process whose law is invariant under conformal automorphisms of @xmath37 that fix the origin .
in fact , it is also possible to define a radial version of the sle(@xmath145 ) process ( see @xcite ) in order to also explore the connected components that are `` cut out '' without having a point on the boundary of @xmath37 .
this can be either defined using a `` radial - chordal '' equivalence ( like that of sle(6 ) ) or viewed as a rather direct consequence of the target - independence of the ( chordal ) sle(@xmath145 ) ( just interpreting the past of the sle as new boundary points ) .
the obtained set is then dense in the upper half - plane , and it is therefore possible to define the family of all loops that are traced in this way : these are exactly the cle@xmath1 defined in @xcite . in particular , for any given choice of the root ( say at the origin i.e. , the sle(@xmath145 ) is started from @xmath1200 ) , any given @xmath40 in the upper half - plane is almost surely surrounded by an ( outermost ) loop in this cle@xmath1 , and we will denote it by @xmath1201 . let us make the following simple observation that will be useful later on : suppose that @xmath1202 and @xmath1203 are two points in the upper half - plane , and that the ( random ) time @xmath1014 at which this ( branching ) sle(@xmath145 ) process separates @xmath1202 from @xmath1203 is finite
. then , there are two possible ways in which this can happen : * either @xmath1014 is the end - time of an excursion of @xmath566 , where the sle traced a closed loop that surrounds one of the two points but not the other . * or the sle has not yet traced a closed loop that surrounds one of the two points and not the other , but the loewner chain traced by the paths nevertheless separates the two points from one another .
this can for instance happen if the sle(@xmath145 ) touches the boundary of the domain at a limit time that occurs after the tracing of infinitely many small loops ( recall that the sle(@xmath145 ) is not a simple path ) .
the definition of sle(@xmath145 ) clearly shows that it is possible to reconstruct the process @xmath1156 starting from a poisson point process of ( positive ) bessel excursions @xmath1163 of the corresponding dimension .
more precisely ( for the symmetrized bessel process ) , one defines also an i.i.d .
family @xmath1204 ( with @xmath1205 ) of `` random signs '' and then defines @xmath566 by concatenating the excursions @xmath1206 .
each individual excursion corresponds to a pinned loop @xmath1207 in the upper - half plane .
in fact , this pinned loop is an sle excursion , and one has tossed the fair coin in order to decide whether to trace it clockwise or anti - clockwise . hence , an sle(@xmath145 ) defines the same poisson point process of pinned loops @xmath1208 with intensity @xmath70 as in section [ onepoint ] .
but there is a difference in the way this poisson point process is used in order to construct the `` composition '' of the corresponding conformal maps @xmath1209 . here
, one does not compose the conformal maps @xmath1209 that are normalized to have real derivative at @xmath438 .
instead , one keeps track of the previous location of the tip of the curve and continues growing from there .
more precisely , suppose that a pinned loop @xmath678 is given , and that one decides to trace it clockwise .
consider the unbounded connected component @xmath49 of its complement in @xmath37 and let @xmath1210 denote the boundary point of @xmath49 defined as `` the origin seen from the right '' .
define the conformal map @xmath1211 from @xmath49 onto @xmath37 such that @xmath1212 and @xmath1213 .
if the loop was traced counterclockwise , just replace @xmath1210 by @xmath936 in this definition of @xmath1211 . then ,
if one focuses on the radial sle(@xmath145 ) only at those times @xmath565 at which @xmath1214 and before it draws a loop surrounding @xmath438 , we have exactly the iteration of the conformal maps @xmath1215 . in other words ,
let us define @xmath1216 $ ] in such a way that @xmath1217 ; then the difference between the radial sle exploration procedure and the continuous iteration of maps @xmath1218 described in section [ onepoint ] is this `` moebius rotation '' at the end of each discovered loop .
more precisely , let @xmath1219 denote the moebius transformation of @xmath37 onto itself with @xmath1220 and @xmath1221 ; then @xmath1222 .
this gives a motivation to modify the discrete radial exploration mechanism of a cle that we used in lemma [ l.mugivesgammailaw ] as follows : after each step , instead of normalizing the maps @xmath1223 at @xmath438 by @xmath1224 and @xmath1225 , we use the modified map @xmath1226 defined by @xmath1227 and by tossing a fair coin in order to decide whether @xmath1228 or @xmath1229 . because the law of a cle in @xmath37 is invariant under moebius transformations ,
this does not change the fact that the @xmath1230 s are i.i.d .
, that @xmath213 is a geometric random variable with mean @xmath1231 , and that what one discovers at the @xmath1062-th step is independent of the value of @xmath213 and of the maps @xmath1232 .
we would now like to see that this discrete exploration mechanism converges ( as @xmath43 ) exactly to the radial sle(@xmath1233 ) procedure described before .
we already know from the proof of lemma [ l.mugivesgammailaw ] that if we focus on the discovered pinned loops of radius larger than @xmath174 ( for each fixed @xmath174 ) , then the corresponding discovered loops converge in law to the corresponding sle loops ( and that the laws of the corresponding conformal maps therefore converge too ) .
the point is therefore only to check that the effect of all these additional `` moebius rotations '' remains under control ( in other words , that it tends to @xmath45 when @xmath1091 , uniformly with respect to @xmath42 ) . but as we shall now see , this follows from the same arguments that allow us to define symmetrized bessel processes ( respectively symmetric @xmath287-stable processes ) as almost sure limits of processes obtained by the cut - off of all bessel excursions of small height or length ( resp .
the cut - off of all jumps of small size ) i.e. , it is a straightforward consequence of the fact that @xmath526 : for each @xmath423 , we define @xmath1223 and @xmath1226 as before , and choose @xmath1234 in such a way that @xmath1235 .
note that there exists a universal constant such that as soon as @xmath1236 , the corresponding @xmath1237 satisfies @xmath1238 ( because harmonic measure scales like the diameter ) .
it follows , using the independence between the exploration steps , and the symmetry ( i.e. , the fact that @xmath1239 and @xmath1240 have the same law ) that @xmath1241 we have already shown in the proof of lemma [ l.mugivesgammailaw ] that this last quantity goes to @xmath45 as @xmath1091 , uniformly with respect to @xmath42 .
note also that @xmath1242 is a martingale .
hence , doob s inequality ensures that uniformly with respect to @xmath42 , @xmath1243 from this , it follows that forgetting or adding these `` rotations '' to the exploration procedure does not affect it much ( in carathodory sense , seen from @xmath438 ) ; one can for instance use the loewner chain interpretation of this exploration procedures , and the above bound on the supremum of the cumulative rotations implies that the radial loewner driving functions ( with or without these rotations ) are very close . in this section
, we finally complete the proof of theorem [ t3 ] by showing that indeed the loops in cle necessarily have the same law as the set of loops generated by an sle(@xmath1155 ) ( as constructed in @xcite ) .
as before , it is enough to prove that for every finite set @xmath1244 of points in @xmath37 , the joint law of the @xmath1245 is the same for the cle as for the cle@xmath1 loops ( more precisely , as the outermost loops surrounding these points in the cle@xmath1 ) .
fix the points @xmath1244 in @xmath37 and consider a radial sle(@xmath145 ) targeting @xmath1202 , say
. we do not know whether sle(@xmath145 ) is a continuous curve , but we recall that it is continuous at times in the interior of the loop - tracing intervals ( during the excursions of the bessel process ) . for each @xmath1246 , the hull of the process
@xmath1247 is compact and its complement @xmath1248 contains @xmath1202 .
the chain @xmath1249 also traces loops as explained before .
we define @xmath1014 to be the first time at which at least one of the points @xmath1250 is `` swallowed '' by @xmath1251 , i.e. separated from @xmath1202 .
we will use @xmath1252 to denote @xmath1253 . note that by the target invariance property , the law of @xmath1254 does not depend on the fact that we singled out @xmath1202 i.e.
, @xmath1014 is just the first time at which @xmath1252 separates the set of @xmath1110 points into at least two parts . as mentioned earlier
, the time @xmath1014 can occur in two different ways : either the sle has traced a loop surrounding some @xmath1255 or it has simply disconnected the domain into two parts . [ l.sleclemix ] consider the following method of generating a random loop @xmath1256 surrounding each @xmath1255 .
1 . for each @xmath591 ,
if @xmath1255 is surrounded by one of the loops traced by @xmath1257 , then we let @xmath1256 be that loop .
2 . in each component of @xmath1258 that is not surrounded by a loop traced by @xmath1257 , we then construct an independent copy of the cle ( conformally mapped to that ensemble ) , and for each @xmath1255 in that component , we let @xmath1256 be the loop that surrounds @xmath1255 in this cle . then the set @xmath1259 agrees in law with @xmath1260 . given this
lemma , theorem [ t3 ] follows immediately by induction . indeed , suppose that the collections @xmath1261 and @xmath1262 agree in law for all collections of @xmath1263 . by construction ,
each component of @xmath1258 has fewer than @xmath1110 points ; moreover , conditioned on @xmath1264 , the remainder of the branching radial sle ( @xmath1155 ) consists of an independent radial sle(@xmath1265 ) in each component of @xmath1258 not surrounded by a loop by inductive hypothesis , the loops traced by such a process agree in law with the @xmath1256 .
we now proceed to prove lemma [ l.sleclemix ] .
first , we claim that it is enough to prove ( for each given @xmath1266 ) the statement of lemma [ l.sleclemix ] where the time @xmath1014 is replaced by @xmath1267 defined as follows .
let first @xmath1268 denote the first time at which at least one of the images of the @xmath1255 gets within distance @xmath335 of @xmath1269 .
note that typically , at time @xmath1268 , the bessel process @xmath566 will be in the middle of some excursion away from @xmath45 .
then , let @xmath1267 be the first time @xmath565 at which either : 1 .
the sle(@xmath145 ) completes a loop that surrounds some @xmath1255 .
2 . the sle completes the loop it is tracing at @xmath1268 ( i.e. , @xmath1270 and @xmath1271 ) . indeed , we know that each of the loops @xmath1245 is at finite distance from the @xmath1255 . hence , almost surely , the conformal radius of @xmath1272 from each of the @xmath1255 s remains bounded as @xmath388 ; thus , in the carathodory sense seen from @xmath1255 , the domain surrounding @xmath1272 tends to a limit as @xmath388 almost surely ( note that the @xmath1267 increase as @xmath335 decreases ) .
it is therefore clear that the law of the @xmath768 s ( conditioned on @xmath1273 ) tends to a limit as @xmath388 and also that ( by the arguments of section [ s.severalpoints ] ) this limit is indeed that of an independent cle in each of the component of @xmath1272 .
now it remains only to prove the claim for these @xmath1267 . to do this , it suffices to return to the modified radial @xmath26-exploration scheme ( the one including the @xmath1234 `` rotations '' ) that we have just defined and studied , and use the fact that for each fixed @xmath335 , up to @xmath1267 , it approximates the continuous mechanism corresponding to the sle(@xmath145 ) excursions .
we now begin the second part of the paper , focusing on properties of clusters of brownian loops .
the next three sections are structured as follows .
we first study some properties of the brownian loop - soup and of the clusters it defines .
the main result of the present section is that when @xmath2 is not too large ( i.e. is subcritical ) , the outer boundaries of outermost brownian loop - soup clusters form a random collection of disjoint simple loops that does indeed satisfy the conformal restriction axioms . by the main result of the first part
, this implies that they are cle@xmath1 ensembles for some @xmath6 . in section [ s.3 ] ,
we compare how loop - soups and sle@xmath1 curves behave when one changes the domain that they are defined in , and we deduce from this the relation between @xmath6 and @xmath2 in this subcritical phase . in section [ s.4 ] , we show that if the size of the clusters in a brownian loop - soup satisfy a certain decay rate property , then the corresponding @xmath2 is necessary _ strictly _ subcritical .
this enables to show that the loop - soup corresponding to @xmath1274 is the only possible critical one , and completes the identification of all cle@xmath1 s for @xmath15 $ ] as loop - soup cluster boundaries .
we will in fact directly use sle results on only three distinct occasions : we use the standard sle restriction properties from @xcite and the description of cle in terms of sle excursions in section [ s.3 ] , and we use an estimate about the size of an sle@xmath1 excursions in section [ s.4 ] . recall that we consider a brownian loop - soup @xmath72 in @xmath23 with intensity @xmath2 ( which is in fact a random countable collection of simple loops because we take the outer boundaries of brownian loops ) .
note that almost surely , for any two loops in the loop - soup , either the two loops are disjoint or their interiors are not disjoint .
we say that two loops @xmath902 and @xmath1275 in @xmath72 are in the same cluster of loops if one can find a finite chain of loops @xmath1276 in @xmath72 such that @xmath1277 , @xmath1278 and @xmath1279 for all @xmath1280 .
we then define @xmath85 to be the family of all closures of loop - clusters .
finally , we let @xmath27 denote the family of all outer boundaries of outermost elements of @xmath85 ( i.e. elements of @xmath85 that are surrounded by no other element of @xmath85 ) .
the goal of this section is to prove the following proposition : [ p.satisfiesconformal ] there exists a positive constant @xmath586 such that for all @xmath2 in @xmath1281 , the set @xmath27 satisfies the conformal restriction axioms , whereas when @xmath2 is ( strictly ) greater than @xmath586 , @xmath72 has only one cluster almost surely . throughout this section
, we will use neither sle - type results nor results derived earlier in the paper .
we will focus on properties of the collection @xmath85 of ( closures of ) the clusters defined by the loop - soup @xmath72 .
the proposition will follow immediately from a sequence of six lemmas that we now state and prove .
it is easy to see ( and we will justify this in a moment ) that when @xmath2 is very large , there almost surely exists just one cluster , and that this cluster is dense in @xmath23 , i.e. , that almost surely , @xmath1282 . [ l.confrest ]
suppose that @xmath1283 .
let @xmath78 denote some open subset of @xmath23 , and define @xmath142 to be the set obtained by removing from @xmath136 all the ( closures of ) loop - soup clusters @xmath83 that do not stay in @xmath136 .
then , conditionally on @xmath142 ( with @xmath1284 ) , the set of loops of @xmath72 that do stay in @xmath142 is distributed like a brownian loop - soup in @xmath142 .
are dashed ( sketch),title="fig:",width=192 ] are dashed ( sketch),title="fig:",width=192 ] note that we have not yet proved at this point that in this case , @xmath27 is a locally finite collection of disjoint simple loops ( this fact will be proved later in this section ) .
let us define for any @xmath715 , the set @xmath1285 to be the largest union of dyadic squares of side - lengths @xmath1286 that is contained in @xmath142 ( note that this is a deterministic function of @xmath142 ) . for each @xmath715 , and for each union @xmath1287 of such
dyadic squares the loop - soup restricted to @xmath1287 is independent of the event @xmath1288 .
it implies immediately that conditionally on @xmath142 , the set of loops that do stay in @xmath1289 is distributed like a brownian loop - soup in @xmath1289 . since this holds for all @xmath295 , the statement of the lemma follows .
[ l.biggerc-onecluster ] suppose that @xmath1290 and that there is a @xmath75 positive probability that @xmath85 contains an element intersecting the boundary of @xmath23 .
then for all positive @xmath1291 , @xmath1292 . assume the hypotheses of the lemma , and let @xmath1293 be the union of all elements of @xmath85 that intersect some prescribed boundary arc @xmath1294 of @xmath23 of positive length .
by invariance under rotation , @xmath1295 .
using the same argument as in the previous lemma , we get that if @xmath136 is a fixed open set such that @xmath1296 , then conditioned on the event @xmath1297 , the law of the set of loops in @xmath72 that are contained in @xmath136 is the same as its original law ( since changing the set of loops within @xmath136 has no effect on @xmath1293 ) .
since this holds for any @xmath136 , conformal invariance of the loop soup implies that conditioned on @xmath1293 , the law of the elements of @xmath72 that do not intersect @xmath1298 is that of independent copies of @xmath72 conformally mapped to each component of @xmath1299 . note that this in fact implies that the event that @xmath1293 is empty is independent of @xmath72 , and hence has probability zero or one ( but we will not really need this ) .
the conformal radius @xmath1300 of @xmath1299 seen from @xmath45 has a strictly positive probability to be smaller than one .
we now iterate the previous procedure : we let @xmath202 denote the connected component of @xmath1301 that contains the origin .
note that the harmonic measure of @xmath1302 at @xmath45 in @xmath191 is clearly not smaller than the harmonic measure of @xmath1294 in @xmath23 at @xmath45 ( a brownian motion started at the origin that exits @xmath23 in @xmath1294 with necessarily exit @xmath191 through @xmath1303 ) .
we now consider the loop - soup in this domain @xmath202 , and we let @xmath1304 denote the union of all loop - soup clusters that touch @xmath1303 .
we then iterate the procedure , and note that the conformal radius of @xmath304 ( from @xmath45 ) is dominated by a product of i.i.d .
copies of @xmath1300 .
this shows that for any positive @xmath335 one can almost surely find in @xmath27 a finite sequence of clusters @xmath1305 , such that @xmath1306 for all @xmath1307 , such that @xmath599 touches @xmath1294 and @xmath1308 . by conformal invariance ,
it is easy to check that the same is true if we replace the origin by any fixed point @xmath40 .
hence , the statement holds almost surely , simultaneously for all points @xmath40 with rational coordinates , for all rational @xmath335 , and all boundary arcs of @xmath161 of positive length .
note that almost surely , each loop of the loop soup surrounds some point with rational coordinates .
we can therefore conclude ( see figure [ loops1000 ] ) that almost surely , any two clusters in @xmath85 are `` connected '' via a finite sequence of adjacent clusters in @xmath85 .
if we now augment @xmath72 by adding an independent brownian loop - soup @xmath1309 of intensity @xmath1291 for any given positive @xmath1291 , the new loops will almost surely join together any two adjacent clusters of @xmath85 into a single cluster ( this is just because any given point @xmath1310 for instance one chosen contact point between two adjacent clusters of @xmath85 will almost surely be surrounded by infinitely many small loops of @xmath1309 ) . furthermore
, for an analogous reason , almost surely , any loop of @xmath1309 intersects some loop of @xmath72 .
it follows that for all @xmath1311 , @xmath1312 almost surely , there exists just one single cluster , i.e. , @xmath1282 .
[ l.firstpcdef ] there is a critical constant @xmath1313 $ ] such that 1 . if @xmath1314 , then @xmath1315 .
2 . if @xmath1316 , then @xmath75 almost surely 1 .
@xmath85 has infinitely many elements
no element of @xmath85 intersects the boundary of @xmath23 .
no two elements of @xmath85 intersect each other .
suppose that @xmath1317 ; if there is a @xmath75 positive probability that two elements of @xmath85 intersect each other , then ( applying lemma [ l.confrest ] to some @xmath136 that contains one but not the other with positive probability ) we find that there is a positive probability that an element of @xmath85 intersects @xmath161 . also , if @xmath1318 and @xmath85 has only finitely many elements with positive probability , then ( with the same probability ) at least one of these elements must intersect @xmath161 ( since the loops of @xmath72 are dense in @xmath23 a.s . ) .
thus , lemma [ l.biggerc-onecluster ] implies that if @xmath586 is the supremum of @xmath2 for which ( a ) , ( b ) , and ( c ) hold almost surely , then @xmath72 has only one cluster @xmath75 almost surely whenever @xmath1319 .
we say @xmath2 is _ subcritical _ if the ( a ) , ( b ) , and ( c ) of lemma [ l.firstpcdef ] hold @xmath75 almost surely .
we will later show that @xmath586 is subcritical , but we have not established that yet .
we remark that the proof of lemma [ l.firstpcdef ] shows that in order to prove that some @xmath2 is subcritical , it suffices to check ( b ) .
[ pc ] the @xmath586 of lemma [ l.firstpcdef ] lies in @xmath1320 .
moreover , when @xmath1318 is small enough , there are @xmath75 almost surely continuous paths and loops in @xmath10 that intersect no element of @xmath72 .
we first prove the latter statement : for small @xmath1318 there exist almost surely simple paths crossing @xmath23 that intersect no element of @xmath72 .
this will also imply @xmath1321 . to this end
we will couple the loop - soup with a well - known fractal percolation model .
the argument is similar in spirit to the one for multi - scale poisson percolation in @xcite , chapter 8 .
consider the unit square @xmath1322 instead of @xmath23 .
for each @xmath295 , we will divide it into @xmath1323 ( closed ) dyadic squares of side - length @xmath1286 . to each such square @xmath84 ,
associate a bernoulli random variable @xmath1324 equal to one with probability @xmath153 .
we assume that the @xmath1324 are independent .
then , define @xmath1325 ^ 2 \setminus \bigcup_{c \ : \
x(c)=0 } c .\ ] ] this is the fractal percolation model introduced by mandelbrot in @xcite ( see also the book @xcite ) .
it is very easy to see that the area of @xmath585 is almost surely zero as soon as @xmath1326 .
chayes , chayes and durrett @xcite have shown that this model exhibits a phase transition : there exists a @xmath1327 , such that for all @xmath1328 , @xmath585 connects the left and right sides of the unit square with positive probability , whereas for all @xmath1329 , this is a.s . not the case ( note that in fact , if @xmath1330 , then @xmath585 is almost surely empty by a standard martingale argument ) . here
, we will only use the fact that for @xmath153 large enough ( but less than one ) , @xmath585 connects the two opposite sides of the unit square with positive probability .
we remark that the proof in @xcite actually gives ( for large @xmath153 ) a positive probability that there exists a continuous path from the left to right side of the unit square in @xmath585 that can be parameterized as @xmath1331 where @xmath1332 $ ] and @xmath1333 is non - decreasing .
it also shows ( modulo a straightforward fkg - type argument ) that @xmath585 contains loops with positive probability .
now , let us consider a loop - soup with intensity @xmath2 in the unit square . for each loop
@xmath902 , let @xmath1334 denote its @xmath1335-diameter ( i.e. , the maximal variation of the @xmath118-coordinate or of the @xmath188-coordinate ) , and define @xmath1336 in such a way that @xmath1337 . note that @xmath902 can intersect at most 4 different dyadic squares with side - length @xmath1286 .
we can therefore associate in a deterministic ( scale - invariant and translation - invariant ) manner to each loop @xmath1338 , a dyadic site @xmath1339 such that @xmath902 is contained in the square @xmath1340 with side - length @xmath1341 and bottom - left corner at @xmath1342 .
note that @xmath1343 .
we are first going to replace all loops @xmath902 in the loop - soup by the squares @xmath1340 .
this clearly enlarges the obtained clusters . by the scale - invariance and the poissonian character of the loop - soup , for each fixed square @xmath1344 \times [ j ' 2^{-m } , ( j'+2 ) 2^{-m } ] \subset [ 0,2]^2,\ ] ]
the probability that there exists no loop @xmath902 in the loop - soup such that @xmath1345 is ( at least ) equal to @xmath1346 for some positive constant @xmath60 ( that is independent of @xmath1110 ) .
furthermore , all these events ( when one lets @xmath190 vary ) are independent . hence , we see that the loop - soup percolation process is dominated by a variant of mandelbrot s fractal percolation model : let @xmath1347 denote an independent percolation on squares of type , with each @xmath1348 equal to @xmath3 with probability @xmath1349 and define the random compact set @xmath1350 ^ 2 \setminus \bigcup_{s \ : \
\tilde x(s)=0 } s.\ ] ] note that in the coupling between the loop - soup and @xmath1351 described above , the distance between @xmath1351 and each fixed loop in the loop - soup is ( strictly ) positive almost surely .
in particular , @xmath1351 is contained in the complement of the union of the all the loops ( and their interiors ) .
we now claim that this variant of the percolation model is dominated by mandelbrot s original percolation model with a larger ( but still less than 1 ) value of @xmath1352 that we will choose in a moment . to see this ,
let @xmath1353 be a @xmath1354-percolation ( with @xmath1355 ) on the set of @xmath1356 pairs with @xmath84 a dyadic square and @xmath190 a square comprised of @xmath84 and three of its neighbors .
take @xmath1357 clearly @xmath1324 is a bernoulli percolation with parameter @xmath1358 and @xmath1348 is a bernoulli percolation with parameter @xmath1359 .
let us now choose @xmath153 in such a way that @xmath1360 .
note that @xmath153 tends monotonically to @xmath3 as @xmath1361 tends to @xmath3 .
hence , by taking @xmath1361 sufficiently close to @xmath3 , i.e. @xmath2 sufficiently small , we can ensure that @xmath153 is as close to @xmath3 as we want ( so that @xmath585 contains paths and loops with positive probability ) .
but in our coupling , by construction we have @xmath1362 and thus @xmath1363 .
we have now shown that @xmath1364 , but we still have to show that @xmath1365 .
we use a similar coupling with the fractal percolation model . for any dyadic square that does not touch the boundary of @xmath373 ^ 2 $ ] , we let @xmath1324 be @xmath45 if @xmath84 is surrounded by a loop in @xmath72 that is contained in the set of eight neighboring dyadic squares to @xmath84 ( of the same size ) .
the @xmath1324 are i.i.d .
( for all @xmath84 whose eight neighbors are contained in @xmath10 ) , and have a small probability ( say smaller than @xmath210 ) of being @xmath3 when @xmath2 is taken sufficiently large .
we now use the fact , mentioned above , that if @xmath1330 , then @xmath585 is almost surely empty ; from this we conclude easily that almost surely , for each @xmath84 ( whose incident neighbors are in @xmath10 ) every point in @xmath84 is surrounded by a loop in @xmath72 almost surely .
it follows immediately that almost surely , every point in @xmath10 is surrounded by a loop in @xmath72 .
this implies that almost surely all loops of @xmath72 belong to the same cluster ( since otherwise there would be a point on the boundary of a cluster that was not surrounded by a loop ) .
[ exponentialcrossingdecay ] if @xmath2 is sub - critical , the probability that there are @xmath355 disjoint chains of loops in @xmath85 crossing from the inside to the outside of a fixed annulus decays exponentially in @xmath355 .
we know that @xmath2 is subcritical , so that for all @xmath1366 , the probability that there exists a crossing of the annulus @xmath1367 by a chain of loops goes to @xmath45 as @xmath1368 .
hence , there exists @xmath1369 such that the probability that there is a single crossing of @xmath1370 is strictly smaller than one .
hence , it follows easily that if we consider any given annulus , @xmath1371 , there exists a positive probability that no cluster of the loop - soup crosses the annulus ( just consider the two independent events of positive probability that the loop - soup restricted to @xmath1372 contains no crossing of the annulus , and that no loop intersects both the circles of radius @xmath1373 and @xmath1374 around @xmath924 ) .
in other words , the probability that there exists a crossing of the annulus is strictly smaller than one .
the result then follows from the bk inequality for poisson point processes ( see for instance @xcite and the references therein ) . as a consequence ( letting @xmath1375 )
, we see that for each fixed annulus , the probability that it is crossed by infinitely many disjoint chains of loops is zero .
we are now ready to state and prove the final lemma of this section : [ l.simpleloops ] if @xmath2 is sub - critical , the set @xmath85 is almost surely locally finite . moreover , when @xmath2 is sub - critical , then almost surely , all elements of @xmath27 are continuous simple loops .
if the set @xmath85 is not locally finite ( i.e. if there exists some @xmath26 such that there are infinitely many clusters of diameter at least @xmath26 ) , then for some positive @xmath26 there exists a point @xmath1376 and a sequence @xmath1377 of elements of @xmath85 of size greater than @xmath26 such that @xmath1378 as @xmath212 . hence any annulus with outer radius less than @xmath1379 that surrounds @xmath40 will have infinitely many crossings by disjoint chains of loops . by lemma [ exponentialcrossingdecay ] ,
the probability that such a point exists is zero ( this result just follows from the lemma by considering a countable collection of annuli such that each point in @xmath23 is surrounded by arbitrarily small annuli from this collection ) .
note that by construction , no element of @xmath85 can have a cut point ( recall that almost surely no two loops of @xmath72 intersect at only a single point and that the elements of @xmath72 are all simple loops ) .
it therefore remains only to check that the outer boundary of a cluster is a continuous loop .
several approaches are possible to justify this .
let us first show that the outer boundary of a cluster can be viewed as the outer boundary of a single continuous ( non - simple ) loop .
let @xmath1380 be any given countable collection of simple loops in @xmath23 such that the graph whose vertices are the loops ( with two loops connected if they intersect ) is connected .
assume that there are at most finitely many loops above any given diameter , and at most finitely many disjoint crossings by chains of loops of any given annulus ( note that all this holds almost surely when @xmath1380 is the family of loops in a given sub - critical loop - soup cluster ) .
we would like to define a single continuous loop that traces exactly through all of the points on the closure of the union of the @xmath1381 .
it is not so surprising that this is possible , but requires some justification : let @xmath318 be a spanning tree of the adjacency graph on the @xmath1381 and fix a root vertex . now , relabeling appropriately if necessary ( for instance by choosing @xmath1382 to be the root loop and then inductively choose @xmath1381 to be the largest child in @xmath318 of @xmath1383 ) , assume that @xmath1384 are enumerated in such a way that for every @xmath1385 , the loop @xmath1381 has one of @xmath1386 as a parent in @xmath318 .
we can inductively define a parameterization of each of the @xmath1381 , starting and ending at a point @xmath1387 on @xmath1381 and taking @xmath1388 time , as follows .
pick a point on @xmath1382 and traverse the loop in time @xmath1389 starting and ending at an arbitrary point @xmath707 on the loop . given the parameterizations up to @xmath1390 ,
let @xmath1387 be the location of the first place that the parameterization of the parent loop of @xmath1381 hits @xmath1381 .
then choose an arbitrary parameterization of @xmath1381 starting and ending at @xmath1387 and taking @xmath1388 time .
now we can define a single loop @xmath1391 that traverses all of the @xmath1392 in a total time @xmath1393 : we define a sequence of loops @xmath1394 converging to @xmath1391 uniformly .
first @xmath1395 is the loop that traverses @xmath1382 in order except that each time it hits a point which is an @xmath1387 for some @xmath1385 it waits at that point for a time equal to the sum of all the lengths of @xmath1381 and all of its descendants .
thus @xmath1395 traverses @xmath1382 in unit time .
we define @xmath1396 the same way except that at each one of the @xmath1387 that the @xmath1395 loop paused at we actually traverse the loop @xmath1381 ( pausing at each @xmath1397 on that loop for an amount of time equal to the sum of the lengths of @xmath1398 and all of its descendants ) .
inductively we similarly define @xmath1399 , which traverses all loops @xmath1400 , and note that the finite diameter and finite number of annulus crossing conditions imply that @xmath1401 converge uniformly to a limiting continuous loop @xmath1391 whose range contains the range of each of @xmath1401 , hence each @xmath1392 .
the range of @xmath1391 is therefore exactly the closure of @xmath1402 .
now , it is easy to see that the outer boundary of a continuous loop in the plane is necessarily a continuous loop ( this is for instance explained and used the proof of theorem 1.5(ii ) in @xcite , page 1003 ) . in the case of our loop - soup clusters ,
we know that this loop is almost surely simple because the cluster has no cut - point .
proposition [ p.satisfiesconformal ] now follows from the results proved in this section : lemma [ pc ] gives the existence of @xmath586 , lemma [ l.simpleloops ] implies that the loops are simple and locally finite a.s .
( which in particular also implies the existence of outermost clusters ) , and lemma [ l.confrest ] yields the restriction property .
the main result of our markovian characterization , combined with proposition [ p.satisfiesconformal ] now implies the following : [ clecor ] if @xmath2 is sub - critical , then the set @xmath27 is a cle@xmath1 for some @xmath15 $ ] . in other words , for such @xmath2 ,
the @xmath27 is equivalent in law to the loop ensemble constructed in @xcite via branching sle(@xmath145 ) .
we will now identify @xmath6 in terms of @xmath2 .
[ p.ckappa ] for all subcritical @xmath2 , the set @xmath27 is in fact a cle@xmath1 with @xmath92 .
note that in our proof , we will not really use the description of cle@xmath1 via branching sle(@xmath1155 ) .
we will only use the description of the conformal loop - ensemble via its pinned loop measure , as described in the earlier sections : suppose that @xmath27 is a random loop ensemble that satisfies the conformal restriction axioms . consider its version in the upper half - plane @xmath1403 , and consider the loop @xmath439 of @xmath27 that surrounds @xmath438 .
let us now consider the law of @xmath1055 conditioned on the event @xmath1404 .
we have seen that this law converges when @xmath43 to some limit @xmath56 , and furthermore that for some @xmath15 $ ] , @xmath56 is equal to an sle - excursion law @xmath1405 that we will describe in the next paragraph . furthermore , when this is the case , it turns out that the entire family @xmath27 is a cle@xmath1 for this value of @xmath6 .
consider an sle@xmath1 in the upper half - plane , started from the point @xmath24 on the real axis to the point @xmath45 ( on the real axis as well ) .
such an sle path will typically be very small when @xmath42 is small .
however , one can show that the limit when @xmath43 of ( the law of ) this sle , conditioned on disconnecting @xmath438 from @xmath702 in the upper half - plane exists .
this limit ( i.e. , its law ) is what we call @xmath1405 .
conformal invariance of sle@xmath1 enables to define an analogous measure in other simply connected domains .
suppose for instance that @xmath1406 .
we can again consider the limit of the law of sle@xmath1 in @xmath49 from @xmath42 to @xmath45 , and conditioned to disconnect @xmath438 from @xmath1407 .
this limit is a probability measure @xmath1408 that can also be viewed as the image of @xmath1409 under the conformal map @xmath52 from @xmath1403 onto @xmath49 that keeps the points @xmath45 and @xmath438 invariant .
note that the same argument holds for other choices of @xmath49 , but choosing this particular one will be enough @xmath49 to identify the relationship between @xmath2 and @xmath6 .
one can use sle techniques to compare `` directly '' the laws of an sle @xmath98 from @xmath42 to @xmath45 in @xmath1403 and of an sle @xmath1410 also from @xmath42 to @xmath45 in @xmath49 . more precisely , the sle martingale derived in @xcite show that the radon - nikodym of the former with respect to the latter is a multiple of @xmath1411 on the set of loops @xmath98 that stay in @xmath49 ( recall that @xmath1412 denotes the @xmath70-mass of the set of brownian loops in @xmath10 that intersect both @xmath47 and @xmath73 ) , where @xmath1413 this absolute continuity relation is valid for all @xmath42 , and it therefore follows that it still holds after passing to the previous limit @xmath43 , i.e. , for the two probability measures @xmath1405 and @xmath1414 .
this can be viewed as a property of @xmath1405 itself because @xmath1414 is the conformal image of @xmath1405 under @xmath52 .
note that the function @xmath1415 is strictly increasing on the interval @xmath7 $ ]
. only one value of @xmath6 corresponds to each value of @xmath90 $ ] .
hence , in order to identify the value @xmath6 associated to a family @xmath27 of loops satisfying the conformal restriction axioms , it suffices to check that the probability measure @xmath56 satisfies the corresponding absolute continuity relation for the corresponding value of @xmath2 .
suppose that @xmath2 is subcritical , that @xmath72 is a loop - soup with intensity @xmath2 in @xmath1403 , and that @xmath27 is the corresponding family of disjoint loops ( i.e. , of outer boundaries of outermost clusters of loops of @xmath72 ) .
for the semi - circle @xmath49 defined above , we denote by @xmath1309 the loops of @xmath72 that stay in @xmath49 , and we denote by @xmath1416 the corresponding family of disjoint loops ( i.e. outermost boundaries of outermost clusters ) .
note that @xmath1309 is a loopsoup in @xmath49 .
let @xmath1055 denote the loop in @xmath27 that surrounds @xmath438 , and let @xmath1417 denote the loop in @xmath1416 that surrounds @xmath438 .
note that if @xmath1418 , then necessarily @xmath1419 .
furthermore , in order for @xmath439 and @xmath1417 to be equal it suffices that : * @xmath1420 contains no loop that intersects @xmath1417 ( note that the probability of this event conditionally on @xmath1417 is @xmath1421 ) .
* one did not create another disjoint cluster that goes `` around '' @xmath1417 by adding to @xmath1309 the loops of @xmath72 that do not stay in @xmath49 .
when @xmath1417 already intersects the disk of radius @xmath42 , the conditional probability that one creates such an additional cluster goes to @xmath45 as @xmath43 ( using the bk inequality for poisson point processes , as in the proof of lemma [ exponentialcrossingdecay ] ) .
if we now condition on the event that @xmath1417 intersects the disk of radius @xmath42 and let @xmath43 , it follows that under the limiting law @xmath56 satisfies the absolute continuity relation that we are after : on the set of curves @xmath98 that stay in @xmath49 , the radon - nikodym derivative of @xmath56 with respect to the measure defined directly in @xmath49 instead of @xmath1403 is @xmath1422 ) . hence , one necessarily has @xmath1423 for @xmath1424 .
this identification allows us to give a short proof of the following fact , which will be instrumental in the next section : [ p.cosubcritical ] @xmath586 is subcritical .
note that the standard arguments developed in the context of mandelbrot percolation ( see @xcite ) can be easily adapted to the present setting in order to prove that @xmath586 is subcritical , but not in the sense we have defined it .
it shows for instance easily that at @xmath586 , there exist paths and loops that intersect no loop in the loop - soup , but non - trivial additional work would then be required in order to deduce that no cluster touches the boundary of the domain .
since we have this identification via sle@xmath1 loops at our disposal , it is natural to prove this result in the following way : let @xmath1425 be the outermost cluster surrounding @xmath438 if the loop - soup ( in @xmath1403 ) has intensity @xmath2 .
if we take the usual coupled poisson point of view in which the set @xmath1426 is increasing in @xmath2 ( with loops `` appearing '' at random times up to time @xmath2 ) then we have by definition that almost surely @xmath1427 ( this is simply because , almost surely , no loop appears exactly at time @xmath586 ) .
let @xmath1428 denote the euclidean distance between @xmath1429 and the segment @xmath1430 $ ] .
clearly @xmath1431 almost surely for each @xmath1432 and @xmath1433 . by the remark after lemma [ l.firstpcdef ] , we know that in order to prove that @xmath586 is subcritical , it suffices to show that @xmath1434 almost surely ( by moebius invariance , this will imply that almost surely , no cluster touches the boundary of @xmath1403 ) . note that @xmath99 is a non - increasing function of the loop - soup configuration ( i.e. , adding more loops to a configuration can not increase the corresponding distance @xmath99 ) .
similarly , the event @xmath1435 that the outermost cluster surrounding @xmath438 does intersect the @xmath42-neighborhood of the origin is an increasing event ( i.e. , adding more loops to a configuration can only help this event to hold ) .
hence it follows that for each @xmath42 , the random variable @xmath1428 is negatively correlated with the event @xmath1435 . letting @xmath43
, we get that ( for subcritical @xmath2 ) , the law of @xmath1428 is `` bounded from below '' by the law of the distance between the curve @xmath98 ( defined under the probability measure @xmath1405 ) and @xmath1436 $ ] , in the sense that for any positive @xmath118 , @xmath1437 ) \ge x ) .\ ] ] but we also know that @xmath1438 , so that @xmath1439 .
it follows readily that for all @xmath1432 and for all @xmath118 , @xmath1440 ) \ge x ) \
\ge\ p^{i , \kappa_0 } ( d(\gamma , [ 1,2 ] ) \ge 2x).\ ] ] but we know that for any @xmath57 , the sle excursion @xmath98 stays almost surely away from @xmath1436 $ ] .
putting the pieces together , we get indeed that @xmath1441 ) \ge 2x ) \ \ge \
p^{i , \kappa_0 } ( d ( \gamma , [ 1,2 ] ) > 0 ) \ = \ 1 .\end{aligned}\ ] ]
the statements of our main results on brownian loop - soup cluster , theorems [ loopsoupsatisfiesaxioms ] and [ kappacorrespondence ] , are mostly contained in the results of the previous section : corollary [ clecor ] , proposition [ p.ckappa ] , proposition [ p.cosubcritical ] .
it remains only to prove the following statement : [ criticaltheorem ] the critical constant of lemma [ l.firstpcdef ] is @xmath1442 ( which corresponds to @xmath1443 , by proposition [ p.ckappa ] ) .
propositions [ p.satisfiesconformal ] , [ p.ckappa ] and our markovian characterization results already imply that we can not have @xmath1444 , since in this case the @xmath27 corresponding to @xmath1445 would give additional random loop collections satisfying the conformal axioms ( beyond the cle@xmath1 with @xmath0 $ ] ) , which was ruled out in the first part of the paper .
it remains only to rule out the possibility that @xmath1446 .
the proof of this fact is not straightforward , and it requires some new notation and several additional lemmas . let us first outline the very rough idea .
suppose that @xmath1447 .
this means that at @xmath586 , the loop - soup cluster boundaries are described with @xmath58-type loops for @xmath1448 .
certain estimates on sle show that sle curves for @xmath624 have a probability to get @xmath26 close to a boundary arc of the domain that decays quickly in @xmath26 , and that this fails to hold for sle@xmath107 .
in fact , we will use corollary [ boundedness ] roughly shows that the probability that the diameter of the set of loops intersecting a small ball on the boundary of @xmath1403 is large , decays quickly with the size of this ball when @xmath624 .
hence , one can intuitively guess that when @xmath624 , two big clusters will be unlikely to be very close , i.e. , in some sense , there is `` some space '' between the clusters .
therefore , adding a loop - soup with a very small intensity @xmath1291 on top of the loop - soup with intensity @xmath586 might not be sufficient to make all clusters percolate , and this would contradict the fact that @xmath586 is critical .
we find it convenient in this section to work with a loop - soup in the upper half plane @xmath1403 instead of the unit disk . to show that there are distinct clusters in such a union of two cles @xmath27 and @xmath1416 , we will start with the semi - disk @xmath1293 of radius @xmath3 centered at the origin .
we then add all the loops in @xmath27 that hit @xmath1293 , add the loops in @xmath1449 that hit those loops , add the loops in @xmath27 that hit those loops , etc . and try to show that in some sense the size of this growing set remains bounded almost surely .
the key to the proof is to find the right way to describe this `` size '' , as the usual quantities such as harmonic measure or capacity turn out not to be well - suited here .
we will now define a generalization of the usual half - plane capacity .
suppose that @xmath1450 $ ] , and that @xmath47 is a bounded closed subset of the upper half - plane @xmath1403 .
we define @xmath1451 where @xmath1452 is a brownian motion started at @xmath1453 stopped at the first time @xmath1454 that it hits @xmath1455 .
note that @xmath1456 is just the usual half - plane capacity used in the context of chordal loewner chains , whereas @xmath1457 is the harmonic measure of @xmath1458 `` viewed from infinity '' .
recall that standard properties of planar brownian motion imply that the limit in ( [ hcappsi ] ) necessarily exists , that it is finite , and that for some universal constant @xmath1459 and for any @xmath174 such that @xmath47 is a subset of the disk of radius @xmath174 centered at the origin , the limit is equal to @xmath1460 where the brownian motion @xmath317 starts at a random point @xmath1461 , where @xmath164 distributed according to the density @xmath1462 on @xmath1463 $ ] .
a _ hull _ is defined to be a bounded closed subset of @xmath1403 whose complement in @xmath1403 is simply connected .
the union of two hulls @xmath47 and @xmath73 is not necessarily a hull , but we denote by @xmath1464 the hull whose complement is the unbounded component of @xmath1465 .
when @xmath47 is a hull , let us denote by @xmath1466 the conformal map normalized at infinity by @xmath1467 .
recall that for all @xmath1468 , @xmath1469 . then , when @xmath73 is another hull , the set @xmath1470 is not necessarily a hull .
but we can still define the unbounded connected component of its complement in the upper half - plane , and take its complement .
we call it @xmath1471 ( by a slight abuse of notation ) .
it is well - known and follows immediately from the definition of half - plane capacity that for any bounded closed @xmath47 and any positive @xmath59 , @xmath1472 .
similarly , for any two @xmath47 and @xmath73 , @xmath1473 .
furthermore @xmath1474 is increasing with respect to @xmath47 , and behaves additively with repect to composition of conformal maps for hulls .
we will now collect easy generalizations of some of these four facts .
observe first that for any positive @xmath59 we have @xmath1475 similarly , we have that for any two ( bounded closed ) @xmath47 and @xmath73 , @xmath1476 this follows from the definition and the fact that for each sample of the brownian motion we have @xmath1477 applying the optional stopping time theorem to the local supermartingale @xmath1478 , we know that @xmath1479 since @xmath1480 we next claim that for any three given hulls @xmath73 , @xmath1293 and @xmath1304 , we have @xmath1481 to verify the claim , note first that for @xmath1482 , we have @xmath1483 .
recall that @xmath1484 for all @xmath1485 , so that in particular , @xmath1486 where @xmath112 is the first hitting time of @xmath1487 by the brownian motion .
we let the starting point of the brownian motion tend to infinity as before , and the claim follows
. it will be useful to compare @xmath1488 with some quantities involving dyadic squares and rectangles that @xmath47 intersects .
( this is similar in spirit to the estimates for half - plane capacity in terms of hyperbolic geometry given in @xcite . )
we will consider the usual hyperbolic tiling of @xmath1403 by squares of the form @xmath1489 \times [ 2^j , 2^{j+1}]$ ] , for integers @xmath1490 .
let @xmath1491 be the set of all such squares .
for each hull @xmath47 , we define @xmath1492 to be the set of squares in @xmath1491 that @xmath47 intersects , and we let @xmath1493 be the union of these squares , i.e. @xmath1494 and its corresponding @xmath1493,width=480 ] [ mhatmagree ] there exists a universal positive constant @xmath84 such that for any hull @xmath47 , @xmath1495 clearly , @xmath1496 by since @xmath1497 . on the other hand ,
if we stop a brownian motion at the first time it hits @xmath1493 i.e. a square @xmath190 of @xmath1498 , then it has a bounded probability of later hitting @xmath47 at a point of about the same height , up to constant factor : this can be seen , for example , by bounding below the probability that ( after this hitting time of @xmath1493 ) the brownian motion makes a loop around @xmath190 before it hits any square of @xmath1491 that is not adjacent to @xmath190 , which would in particular imply that it hits @xmath47 during that time .
this probability is universal , and the lemma follows . [ mhatsum ] there exists a universal positive constant @xmath88 such that for any hull @xmath47 , @xmath1499 the right - hand inequality is obvious by ( [ munionsubadditive ] ) and lemma [ mhatmagree ] . by lemma [ mhatmagree ] ,
it is sufficient to prove the result in the case where @xmath1500 is the union of squares in @xmath527 . for each @xmath591 in @xmath1501 , we will call @xmath1502 the set of squares that are at height between @xmath1503 and @xmath1504
. we will say that a square @xmath1505 \times [ 2^j , 2^{j+1}]$ ] in @xmath1502 is even ( respectively odd ) if @xmath59 is even ( resp .
we know that @xmath1506 to bound this expectation , we note that for each @xmath1507 , for each even square @xmath1508 , and for each @xmath1509 , the probability that a brownian motion started from @xmath40 hits the real line before hitting any other even square in @xmath1502 is bounded from below independently from @xmath40 , @xmath591 and @xmath190 . hence , the strong markov property implies that the total number of even squares in @xmath1502 hit by a brownian motion before hitting the real line is stochastically dominated by a geometric random variable with finite universal mean ( independently of the starting point of the brownian motion ) that we call @xmath1510 .
the same is true for odd squares .
note also that if the starting point @xmath40 of @xmath1511 is in @xmath1512 and if @xmath1513 , then the probability that the imaginary part of @xmath317 reaches @xmath1514 before @xmath317 hits the real line is not larger than @xmath1388 .
it follows from the strong markov property that the expected number of squares in @xmath1515 that @xmath317 visits before exiting @xmath1403 is bounded by @xmath1516 .
it follows that for a brownian motion started from @xmath40 with @xmath1517 @xmath1518 for some universal constant @xmath84 ( bear in mind that @xmath1519 and that for @xmath1520 , @xmath1521 ) . if we now apply this statement to the brownian motion @xmath1452 after its first hitting time @xmath1522 of @xmath1523 , we get that for all large @xmath1342 and for some universal positive constant @xmath88 , @xmath1524 combining this with ( [ eqsum ] ) concludes the proof . for each square @xmath1505 \times [ 2^j , 2^{j+1}]$ ] of @xmath1491
, we can define the union @xmath1525 of @xmath190 with all the squares of @xmath1526 that lie strictly _ under _
@xmath190 , i.e. @xmath1527 \times [ 0 , 2^{j+1}]$ ] .
note that that scaling shows immediately that for some universal constant @xmath1528 and for all @xmath1529 , @xmath1530 let us now use these quantities to study our random loop - ensembles .
suppose that @xmath27 is the conformal loop ensemble corresponding to any given @xmath1531 $ ] .
given a hull @xmath47 we denote by @xmath1532 the random hull whose complement is the unbounded component of the set obtained by removing from @xmath1533 all the loops of @xmath27 that intersect @xmath47 .
local finiteness implies that @xmath1534 is itself a hull almost surely .
( sketch),width=576 ] now define @xmath1535 recall that if @xmath1536 and @xmath1537 then @xmath75 defines a cle@xmath1 with @xmath624 .
we can therefore reformulate corollay [ boundedness ] in terms of @xmath2 as follows ( this corresponds intuitively to the statement that sle is unlikely to be very close to a boundary arc when @xmath624 ) : [ p.swbound ] if @xmath1538 and @xmath1537 , then there is an @xmath1539 such that , if we denote by @xmath1540 the diameter of @xmath47 , then we have @xmath1541 for all hulls
@xmath47 . throughout the remainder of this subsection
, we will suppose that @xmath1542 and @xmath1543 , and that this @xmath103 is fixed .
we then choose @xmath1544 , and we define @xmath585 and @xmath213 using this value of @xmath287 .
we will then let @xmath2 vary in @xmath1545 $ ] .
it follows from the previous proposition that for all @xmath1546 , @xmath1547 here is a more elaborate consequence of the previous proposition : [ rectanglenbound ] consider @xmath1548 and @xmath1549 fixed as above . for any hull @xmath47 and any @xmath1529 , if @xmath1550 , then @xmath1551 for some constant @xmath1552 depending only on @xmath2 and tending to zero as @xmath97 . by scaling , it suffices to consider the case where @xmath1553 \times [ 1,2]$ ] and hence @xmath1554 is the rectangle @xmath373 \times [ 0,2]$ ] .
proposition [ p.swbound ] then implies that @xmath1555 we want to prove that @xmath1556 tends to zero uniformly with respect to @xmath47 as @xmath97 .
let @xmath1557 denote the event that some loop - soup cluster ( in the loop - soup of intensity @xmath2 ) intersecting the rectangle @xmath400 has radius more than @xmath1558 .
when @xmath1557 does not hold , then standard distortion estimates yield an @xmath26 bound on the height of ( i.e. , the largest imaginary part of an element of ) @xmath1559 .
but we then also know that @xmath1560 is a subset of @xmath1561 \times [ 0,3]$ ] , so that a brownian motion started from @xmath1453 will hit @xmath1560 before hitting @xmath1455 with a probability bounded by @xmath1562 times some universal constant @xmath84 .
hence , unless @xmath1557 holds , we have @xmath1563 . summing up
, we get that @xmath1564 where @xmath1565 denotes the @xmath1566 corresponding to a larger loop - soup of intensity @xmath103 that we couple to the loop - soup of intensity @xmath2 .
but @xmath1567 as @xmath97 and @xmath1568 , so that if we take @xmath2 sufficiently small , @xmath1569 for all hulls @xmath47 .
this completes our proof .
we are now ready to prove our final lemma : [ generalnbound ] for @xmath1546 , there exists a finite constant @xmath1570 such that for all hulls @xmath47 , @xmath1571 . furthermore , we can take @xmath1572 in such a way that @xmath1572 tends to zero as @xmath97 .
putting together the estimates in lemma [ mhatsum ] and corollary [ rectanglenbound ] we have @xmath1573 and @xmath1574 when @xmath1575 , whereas @xmath88 does not depend on @xmath2 .
as we will now see , this property implies that for @xmath1576 , @xmath213 is necessarily infinite for all positive @xmath287 ( i.e. it shows that the size of clusters at the critical point can not decay too fast ) , and this will enable us to conclude the proof of proposition [ criticaltheorem ] in the manner outlined after its statement : suppose that @xmath1447 .
we choose @xmath1577 ( and @xmath1578 ) .
we take @xmath1291 to be positive but small enough so that the product of the corresponding constants @xmath1579 and @xmath1580 in lemma [ generalnbound ] is less than @xmath3 .
we will view the loop - soup @xmath72 with intensity @xmath1581 as the superposition of a loop - soup @xmath1582 with intensity @xmath586 and an independent loop - soup @xmath1309 with intensity @xmath1291 i.e. we will construct @xmath27 via the loop - soup cluster boundaries in @xmath1583 and @xmath1416 .
now let us begin with a given hull @xmath47 ( say the semi - disk of radius @xmath3 around the origin ) .
suppose that @xmath72 contains a chain of loops that join @xmath47 to the line @xmath1584 .
this implies that one can find a finite chain @xmath1585 ( chain means that two consecutive loops intersect ) of loops in @xmath1586 with @xmath1587 and @xmath1588 . since the loops in @xmath1583 ( resp .
@xmath1416 ) are disjoint , it follows that the loops @xmath1589 alternatively belong to @xmath1583 and @xmath1416 .
consider the loops of @xmath1583 that intersect @xmath47 .
let us consider @xmath1293 the hull generated by the union of @xmath47 with these loops ( this is the @xmath1534 associated to the loop - soup @xmath1582 ) .
recall that the expected value of @xmath1590 is finite because @xmath1591 . then add to @xmath1293 , the loops of @xmath1416 that intersect @xmath1293 .
this generates a hull @xmath1592 ( which is the @xmath1593 associated to the loop - soup @xmath1309 ) .
then , add to @xmath1592 the loops of @xmath1583 that intersect @xmath1592 .
note that in fact , one basically adds only the loops of @xmath1583 that intersect @xmath1594 ( the other ones were already in @xmath1293 ) in order to define a new hull @xmath1595 , and continue iteratively .
let @xmath503 be the limiting set obtained .
we can also describe this sort of exploration by writing for all @xmath715 , @xmath1596 , where @xmath1597 is alternately constructed from @xmath1598 using a loop - soup with intensity @xmath586 or @xmath1291 as @xmath295 is even or odd .
the expected value of @xmath1599 decays exponentially , which implies ( borel - cantelli ) that @xmath1599 almost surely decays eventually faster than some exponential sequence .
we note that if @xmath47 is a hull such that for all @xmath1600 , @xmath1601 , we clearly have @xmath1602 . on the other hand , we know that if @xmath47 is a hull such that there exists @xmath1600 with @xmath1603 , then @xmath1604 for some absolute constant @xmath2 .
hence , we see that almost surely , for all large enough @xmath295 , @xmath1605 , which implies that almost surely @xmath1606 .
but the half - plane capacity behaves additively under conformal iterations , so that in fact @xmath1607 .
hence , for large enough @xmath400 , the probability that @xmath503 does not intersect @xmath1608 is positive , and there is a positive probability that no chain of loops in @xmath72 joins @xmath47 to @xmath1608 .
it follows that @xmath1609 and proposition [ p.satisfiesconformal ] would then imply that @xmath1610 which is impossible .
this therefore implies that @xmath1611 .
as explained after the proposition statement , we also know that @xmath586 can not be strictly larger than @xmath3 , so that we can finally conclude that @xmath1612 . as explained at the beginning of this section , this completes the proof of theorems [ loopsoupsatisfiesaxioms ] and [ kappacorrespondence ] .
when @xmath1613 , the cle@xmath1 described in @xcite are random collections of non - simple loops
. however , the outer boundaries of the sets of points traversed by the outermost loops should be simple loops .
we expect ( though this was not established in @xcite , in part because of the cle@xmath1 construction was not shown there to be starting - point independent ) that these random loop collections will satisfy all of the axioms we used to characterize cle except that they will a.s . contain loops intersecting one another and the boundary of the domain .
are these the only random loop collections for which this is the case ?
this is an apparently difficult type of question .
we expect that the constructions of one - point and two - point pinned measures in the present paper will have analogs in this non - disjoint setting ; however the pinned loop will intersect the boundary at other points as well , and our argument for establishing the connection with sle does not appear to extend to this setting in a straightforward way .
moreover , we have no direct analog of the loop - soup construction of cles in the case where @xmath1613 . | for random collections of self - avoiding loops in two - dimensional domains , we define a simple and natural conformal restriction property that is conjecturally satisfied by the scaling limits of interfaces in models from statistical physics .
this property is basically the combination of conformal invariance and the locality of the interaction in the model . unlike the markov property that schramm used to characterize sle curves ( which involves conditioning on partially generated interfaces up to arbitrary stopping times ) , this property only involves conditioning on entire loops and thus appears at first glance to be weaker .
our first main result is that there exists exactly a one - dimensional family of random loop collections with this property
one for each @xmath0$]and that the loops are forms of sle@xmath1 .
the proof proceeds in two steps .
first , uniqueness is established by showing that every such loop ensemble can be generated by an `` exploration '' process based on sle .
second , existence is obtained using the two - dimensional brownian loop - soup , which is a poissonian random collection of loops in a planar domain . when the intensity parameter @xmath2 of the loop - soup is less than @xmath3
, we show that the outer boundaries of the loop clusters are disjoint simple loops ( when @xmath4 there is a.s . only one cluster ) that satisfy the conformal restriction axioms .
we prove various results about loop - soups , cluster sizes , and the @xmath5 phase transition .
taken together , our results imply that the following families are equivalent : 1 .
the random loop ensembles traced by branching schramm - loewner evolution ( sle@xmath1 ) curves for @xmath6 in @xmath7 $ ] .
2 . the outer - cluster - boundary ensembles of brownian loop - soups for @xmath8 $ ] .
3 . the ( only ) random loop ensembles satisfying the _ conformal restriction axioms_. |
there are two experimental pieces of evidence @xcite pressing us to extend the standard model ( sm ) : the evidence for neutrino masses and for the dark matter ( dm ) content of the universe .
the ongoing experiments at the large hadron collider ( lhc ) may provide us with new insights on new heavy particles enabling weinberg s dimension - five operator @xmath2 @xcite as an explanation of the smallness of masses of observed neutrinos .
the known tree level realizations of weinberg s operator proceed by adding to the particle content of the sm a single , gut - scale multiplet : a heavy fermion singlet for the type - i @xcite , a scalar triplet for the type - ii @xcite and a fermion triplet for the type - iii @xcite seesaw mechanism . on the other hand additional discrete symmetries are instrumental @xcite in bringing the seesaw mechanism to the tev scale accessible at the lhc .
alternative lowering of the seesaw scale , without discrete symmetries , can be achieved either by going beyond dimension - five operator or by generating the neutrino masses at the loop level .
the approach used by two recent models is to employ non - zero hypercharge seesaw mediators : the dimension - seven mechanism with fermion triplets in @xcite and the dimension - nine mechanism with fermion quintuplets in @xcite .
both of these mechanisms offer the spectacular triply - charged states verifiable at the lhc .
the approach in @xcite , which we follow here , is to link a dm candidate to the scale of radiatively generated neutrino ( r@xmath0 ) masses .
the authors of @xcite exemplify it by employing a 10 tev scale fermionic quintuplet of zero hypercharge .
let us stress that an isolated @xmath3 fermion field or an isolated septuplet @xmath4 scalar field have been singeld out as viable dm particle candidates within the so - called minimal dark matter ( mdm ) model @xcite .
the dubbed r@xmath0mdm model @xcite claims to achieve the stability of the neutral mdm quintuplet component without introducing a symmetry beyond the sm gauge symmetries , on account of an accidental @xmath1 symmetry arising from the choice of the field content .
it has been adopted in more recent paper @xcite aimed at successful leptogenesis . from the point of generating the neutrino masses
, the fermion quintuplet has to be accompanied by an appropriate scalar multiplet .
high enough scalar field multiplet @xcite avoids the terms in the scalar potential which may lead to and induced vacuum expectation value ( vev ) of the new scalar field , and there are only radiative neutrino masses .
this r@xmath0mdm is realized by accompanying the zero hypercharge quintuplets @xmath3 by a hypercharge one sextuplets , @xmath5 , proposing the neutral component @xmath6 as fermionic dm candidate . in section 2
we give a brief description of this attempt and explicate a previously omitted term in the scalar potential , which spoils the stability of @xmath6 particle . in section 3
we analyze two new seesaw models , the first in which a tentative dm particle appears as a neutral component of the scalar quintuplet @xmath7 , accompanied by a quadruplet seesaw mediator @xmath8 . in the second model
we show that originally viable mdm scalar septuplet @xmath4 candidate also becomes unstable in the context of generating radiative neutrino masses .
we summarize our results in the concluding section .
to allow a yukawa coupling of the isospin two fermion @xmath3 to the lepton doublet @xmath9 , necessary for a mass generation mechanism , @xmath10 one has to introduce a scalar field @xmath11 of a half - integer ( 3/2 or 5/2 ) isospin in addition to the sm higgs @xmath12 . in order to avoid a vev of @xmath11 which might come from quartic @xmath13 term
, the fermionic quintuplet @xmath3 has to be accompanied by a sextuplet scalar @xmath14 as in @xcite .
these authors have argued that the neutral component of the fermionic quintuplet , @xmath6 , can be a dm candidate on account of an accidental @xmath1 symmetry of the lagrangian .
we show that it is a viable dm candidate only if one forbids or finetunes a renormalizable @xmath15 term with the quartic coupling @xmath16 , omitted in @xcite . in their tensor notation ,
the omitted term reads @xmath17 and it can induce a rapid decay of a dm particle at the loop level .
for example , the loop diagram on fig .
[ decay ] generates an effective dimension - six operator @xmath18 which leads to decays of the dm candidate @xmath6 .
in particular , we estimate the decay amplitude for @xmath19 arising from it , @xmath20 at the loop level . ]
it produces the decay width into four final - state particles , @xmath21 which is given by @xmath22 the inverse of it , a lifetime of a dm candidate , should be longer then the age of the universe given by the hubble time @xmath23 . for @xmath6 to have the lifetime @xmath24
, the quartic coupling @xmath16 has to be tiny . by taking @xmath25 , @xmath26 and @xmath27 ,
the values adopted from @xcite , we obtain @xmath28 as an upper limit
. in the context of the decaying dm @xcite the bound on the @xmath29 can be nine orders of magnitude larger ( @xmath30 ) , implaying even stronger limit , @xmath31 .
this means that @xmath6 is a viable dm candidate only if the coupling in eq .
( [ dim4 ] ) has an extremely small value .
this is in contradiction with the claim in @xcite , and in order to avoid the fine - tuning one has to impose the @xmath1 symmetry forbidding the renormalizable term in eq .
( [ dim4 ] ) , by hand . without such discrete symmetry
a model more minimal than in @xcite is possible with scalar quadruplet @xmath32 replacing previous sextuplet field .
like in case of previously studied non - zero hypercharge fermion @xcite , this field @xmath11 develops an induced vev from the quartic @xmath33 term .
the yukawa terms result in a tree - level dimension - nine seesaw operator .
these and additional contributions to the neutrino masses from radiative loop - suppressed diagrams are presented in a separate paper @xcite .
in the light of the described blow to fermionic mdm candidate in the radiative neutrino masses setup , let us now reconsider the scalar mdm multiplets @xcite .
the models involving hypercharge - zero scalar field in radiatively generating neutrino masses through diagrams displayed on fig .
[ mnuloop ] have two essential features : ( i ) the scalar field is accompanied with appropriate fermion multiplet ; ( ii ) there are two scalar fields of different hypercharge in play , the one with hypercharge zero being the dm candidate . a minimal possibility to realize the diagram on fig .
[ mnuloop ] through the scalar quintuplet @xmath34 as mdm candidate is to accompany it with another scalar quintuplet @xmath35 in conjunction with non - zero hypercharge fermionic quadruplet @xmath8 as the seesaw mediator . with two scalars @xmath34 and @xmath35
, there are , up to our knowledge previously unconsidered , dimension - three @xmath1 noninvariant operators in the lagrangian @xmath36 which in tensor notation read @xmath37 these terms make the dm candidate @xmath38 unstable through loop diagrams like the one on fig .
[ decay2 ] .
we estimate the amplitude for the decay @xmath39 @xmath40 and the decay width @xmath41 to have the lifetime of a dm candidate @xmath24 , the parameters @xmath42 in eq .
( [ scalar5 ] ) have to have very small values , @xmath43 .
decays . ] the second possibility to realize the diagram on fig .
[ mnuloop ] is through the scalar septuplet @xmath44 as mdm , together with another scalar septuplet @xmath45 , now in conjunction with fermionic sextuplet @xmath46 . for a single scalar septuplet @xmath47 ,
the term @xmath48 is forbidden by bose statistics @xcite , but in presence of another septuplet @xmath49 there is the dimension - three @xmath1 noninvariant operator @xmath50 making the dm candidate @xmath38 unstable . decays . ]
these terms make the dm candidate @xmath38 unstable through loop diagrams , like the one on fig .
[ decay3 ] .
we estimate the amplitude for the decay @xmath51 @xmath52 and the decay width @xmath53 to have the lifetime of a dm candidate @xmath24 , the parameter @xmath54 in eq .
( [ dim3 ] ) is restricted to very small value , @xmath55 .
to conclude , a major blow to scalar r@xmath0mdm variants comes from super - renormalizable dimension - three @xmath56 operators . besides these renormalizable @xmath1 violating operators , there are possible effects from the planck scale considered in @xcite and references therein .
we address in a common framework the open questions of the neutrino masses and the dark matter content of the universe . in view of the immense dm possibilities
, a minimality of the dm model appears as a criterion essential for its predictivity and testability .
the minimality is satisfied by adding only one extra @xmath57 multiplet in the mdm setup . however , by imposing additional seesaw mission to a given mdm multiplet , it has to be accompanied by additional scalar field introducing new obstacles to the dm stability .
it would be appealing if the dm candidate , a neutral component of a certain higher weak - isospin multiplet , could be stabilized without introducing a symmetry beyond the sm gauge symmetries , as claimed in @xcite .
we have shown that this is not possible for already selected mdm particles , if the neutrino masses are generated through simple one - loop diagrams on fig .
[ mnuloop ] .
the obstacle for the fermionic quintuplet mdm candidate in @xcite comes from renormalizable quartic term omitted there . concerning the scalar r@xmath0mdm variants , we have shown that the stability of the scalar quintuplet and septuplet mdm candidate is violated by ubiquitous super - renormalizable terms that can be tamed only by imposing extra discrete symmetry .
p. minkowski , _
mu @xmath58 e gamma at a rate of one out of 1-billion muon decays ? _ , phys .
b * 67 * , 421 ( 1977 ) ; t. yanagida , in _ proceedings of the workshop on unified theory and the baryon number of the universe _ , eds . o. sawada and a. sugamoto ( kek , tsukuba , 1979 ) , p. 95 ; m. gell - mann , p. ramond , and r. slansky , in _ supergravity _ , edited by p. van nieuwenhuizen and d. freedman ( north holland , amsterdam , 1979 ) , p. 315 ;
glashow , in _ quarks and leptons _ , edited by m. l@xmath59vy _ et al .
_ ( plenum , new york , 1980 ) , p. 707 ;
mohapatra and g. senjanovi , _ neutrino mass and spontaneous parity nonconservation _ ,
* 44 * , 912 ( 1980 ) .
w. konetschny and w. kummer , _ nonconservation of total lepton number with scalar bosons _
b * 70 * , 433 ( 1977 ) ; m. magg and c. wetterich , _ neutrino mass problem and gauge hierarchy _ ,
b * 94 * , 61 ( 1980 ) ; j. schechter and j.w.f .
neutrino masses in su(2)@xmath60u(1 ) theories _ , phys .
d * 22 * , 2227 ( 1980 ) ; t.p .
cheng and l.f .
li , _ neutrino masses , mixings , and oscillations in su(2)@xmath60u(1 ) models of electroweak interactions _ , phys .
d * 22 * , 2860 ( 1980 ) ; g. lazarides , q. shafi , and c. wetterich , _ proton lifetime and fermion masses in an so(10 ) model _ , nucl .
b * 181 * , 287 ( 1981 ) ; r.n .
mohapatra and g. senjanovi , _ neutrino masses and mixings in gauge models with spontaneous parity violation _
, phys . rev .
d * 23 * , 165 ( 1981 ) .
m. cirelli , n. fornengo and a. strumia , _ minimal dark matter _ , nucl .
b * 753 * , 178 ( 2006 ) [ hep - ph/0512090 ] ; m. cirelli and a. strumia , _ minimal dark matter : model and results _
d * 80 * , 071702 ( 2009 ) [ 0905.2710 [ hep - ph ] ] . | we examine the stability of minimal dark matter ( mdm ) particle - candidates in the setup in which they participate in radiative neutrino ( r@xmath0 ) masses .
we first point out the existence of an additional renormalizable term in recently proposed r@xmath0mdm lagrangian , which violates the claimed accidental @xmath1 symmetry and spoils the stability of the fermionic mdm quintuplet component .
we then explore the viability of r@xmath0mdm variants based on scalar mdm multiplets .
there are ubiquitous super - renormalizable terms in the scalar potential which make these scalar multiplets unstable .
= 2 * * * * + = 0.5 kreimir kumeriki , ivica picek and branimir radovi + _ department of physics , faculty of science , university of zagreb , p.o.b .
331 , hr-10002 zagreb , croatia + _
+ _ pacs _ : 14.60.pq ; 14.60.st ; 95.35.+d ; 14.80.cp + _ keywords _ : neutrino mass ; exotic leptons ; dark matter ; non - standard scalar fields |
suppose that we have independent observations @xmath2 from a probability density @xmath3 that belongs to a parametric model @xmath4 , where @xmath5 is an unknown @xmath6-dimensional parameter and @xmath7 is the parameter space .
the random variable @xmath1 to be predicted is independently distributed according to a density @xmath8 in a parametric model @xmath9 , possibly different from @xmath4 , with the same parameter @xmath10 .
the objective is to construct a predictive density @xmath11 for @xmath1 by using @xmath12 .
the performance of @xmath13 is evaluated by the kullback
leibler divergence @xmath14 from the true density @xmath8 to the predictive density @xmath11 .
the risk function is given by @xmath15 = \iint p(x^n \mid \theta ) { \displaystyle \tilde{p}}(y \mid \theta ) \log \frac{{\displaystyle \tilde{p}}(y \mid \theta)}{\hat{p}(y;x^n ) } { \mbox{d}}y { \mbox{d}}x^n.\ ] ] it is widely recognized that plug - in densities @xmath16 constructed by replacing the unknown parameter @xmath10 by an estimate @xmath17 may not perform very well and that bayesian predictive densities @xmath18 constructed by using a prior @xmath19 perform better than plug - in densities . if the value of @xmath10 is given , there is no specific meaning of considering the conditional density of @xmath1 given @xmath20 since the obvious relation @xmath21 holds .
however , if @xmath10 is unknown , bayesian predictive densities @xmath22 constructed by introducing a prior density @xmath23 on the parameter space are useful to approximate the true density @xmath24 as discussed in @xcite and @xcite .
in fact , there exists a predictive density whose asymptotic risk is smaller than that of a plug - in density unless the mean mixture curvature of the model manifold vanishes , see @xcite and @xcite for details .
the choice of @xmath19 becomes important especially when the sample size @xmath25 is not very large .
although the jeffreys prior is a widely known default prior , it does not perform satisfactorily especially when the unknown parameter is multidimensional as jeffreys himself pointed out .
@xcite constructed a bayesian predictive density incorporating the advantage of shrinkage methods for the multivariate normal model .
see also @xcite for useful results for the normal model . in the conventional setting in which the distributions of @xmath26 , @xmath27 , and @xmath1 are
the same , asymptotic theory of prediction based on general parametric models has been studied by using the framework of information geometry , see @xcite . in information geometry ,
a parametric statistical model is regarded as a differentiable manifold , which we call the model manifold , and the parameter space is regarded as a coordinate system of the manifold , see @xcite .
the fisher rao metric is a riemannian metric based on the fisher information matrix on the model manifold .
the jeffreys prior @xmath28 corresponds to the volume element of the model manifold associated with the fisher rao metric .
when the distributions of @xmath26 , @xmath27 , and @xmath1 are the same , the asymptotic difference between the risks of @xmath29 and @xmath30 is given by @xmath31 \notag\\ = & \frac{\displaystyle \delta \left(\frac{\pi}{{\pi_\mathrm{j}}}\right)}{\displaystyle \left ( \frac{\pi}{{\pi_\mathrm{j } } } \right ) } - \frac{1}{2 } \sum_{i=1}^d \sum_{j=1}^d g^{ij } \frac{\displaystyle \partial_i \left(\frac{\pi}{{\pi_\mathrm{j}}}\right ) \partial_j \left(\frac{\pi}{{\pi_\mathrm{j}}}\right ) } { \displaystyle \left(\frac{\pi}{{\pi_\mathrm{j}}}\right)^2 } + { \mathrm{o}}(1 ) = 2 \frac{\displaystyle \delta \left(\frac{\pi}{{\pi_\mathrm{j}}}\right)^{\frac{1}{2 } } } { \displaystyle \left ( \frac{\pi}{{\pi_\mathrm{j } } } \right)^{\frac{1}{2 } } } + { \mathrm{o}}(1 ) , \label{main0}\end{aligned}\ ] ] where @xmath32 denotes @xmath33 , @xmath34 , @xmath35 denotes the @xmath36-element of the inverse of the @xmath37 matrix @xmath38 , and @xmath39 is the laplacian , see @xcite .
the laplacian @xmath39 on a riemannian manifold endowed with a metric @xmath40 is defined by @xmath41 where @xmath42 is the determinant of the @xmath37 matrix @xmath38 , @xmath43 is a smooth real function on @xmath7 , and @xmath44 denotes the covariant derivative , defined in the next section .
the indices @xmath45 run from @xmath46 to @xmath6 .
note that both the definition of the laplacian and the definition @xmath47 that differs in sign are widely adopted in the mathematics literature , although it is confusing .
because of , if there exists a non - constant positive superharmonic function @xmath43 , i.e. a non - constant positive function satisfying @xmath48 for every @xmath10 , on the model manifold , then the bayesian predictive density based on the prior density defined by @xmath49 asymptotically dominates that based on the jeffreys prior . here , the riemannian geometric structure of the model manifold based on the fisher rao metric plays a fundamental role . in practical applications , it often occurs that observed data @xmath26 , @xmath50 , and the target variable @xmath1 to be predicted have different distributions .
regression models are a typical example .
suppose that we observe @xmath51 , where @xmath52 is a given @xmath53 matrix @xmath54 , and predict @xmath55 , where @xmath56 is a given @xmath57 matrix and @xmath58 is an unknown parameter .
then , the fisher information matrices for the same parameter @xmath10 based on @xmath59 and @xmath60 are different .
similar situations also occur in nonlinear regression problems . @xcite and @xcite showed that shrinkage priors are useful for constructing bayesian predictive densities for linear regression models when the observations are normally distributed with known variance .
however , it has been difficult to construct useful priors for general models other than the normal models when @xmath0 and @xmath1 have different distributions . in the present paper ,
we study asymptotic theory for the setting in which @xmath26 , @xmath27 , and @xmath1 have different distributions .
although several asymptotic properties of predictive distributions for such a setting are studied by @xcite , the result corresponding to has not been explored .
the generalization is not straightforward because two different differential geometric structures , one for @xmath3 and the other for @xmath8 , such as the fisher rao metrics exist in the present setting .
we introduce a new metric @xmath61 , which we call the predictive metric , depending on both @xmath3 and @xmath8 .
the predictive metric @xmath61 and the volume element @xmath62 of it correspond to the fisher rao metric and the jeffreys prior in the conventional setting . in section 2
, we obtain an expansion of the difference of the risk functions of bayesian predictive densities . each term in the expansion is represented by using geometrical quantities and is invariant with respect to parameter transformations . in section 3 , we introduce the predictive metric @xmath61 and evaluate the asymptotic risk difference between a bayesian predictive density based on a prior @xmath19 and that based on the volume element prior @xmath62 of the predictive metric @xmath61 .
the asymptotic risk difference is represented by using the laplacian associated with the predictive metric @xmath61 . in section 4 ,
we consider three examples and construct superior priors by using the formula obtained in section 3 .
first , we prepare several information geometrical notations to be used . in the following ,
the quantities associated with the model @xmath63 are denoted without tilde , and those associated with the model @xmath64 are denoted with tilde .
we put @xmath65 and @xmath66 .
the fisher rao metrics on the model manifolds @xmath63 and @xmath67 are given by @xmath68 respectively .
the @xmath36-elements of the inverses of the @xmath37 matrices @xmath69 and @xmath70 are denoted by @xmath71 and @xmath72 , respectively .
we define @xmath73 and @xmath74 here , @xmath75 are the e - connection coefficients , @xmath76 are the m - connection coefficients , and @xmath77 are the riemannian connection coefficients .
the relations @xmath78 represent the duality between @xmath75 and @xmath76 with respect to the metric @xmath79 , and the duality between @xmath80 and @xmath81 with respect to the metric @xmath82 , respectively .
covariant derivatives @xmath83 , @xmath84 , and @xmath85 of a vector field @xmath86 with respect to the connection coefficients @xmath87 , @xmath88 , and @xmath89 are defined by @xmath90 , @xmath91 , and @xmath90 , respectively , where @xmath92 , @xmath93 , and @xmath94 . in the same way , the covariant derivatives @xmath95 , @xmath96 , and @xmath97 , with respect to the connection coefficients @xmath98 , @xmath99 , and @xmath100 , are defined .
theorem [ riskdiff ] below is used in the following sections .
[ riskdiff ] the difference between the risk functions of bayesian predictive densities @xmath101 and @xmath102 based on priors @xmath103 and @xmath104 , respectively , is given by @xmath105 \nonumber \\ = & \left ( \frac{1}{2 } \sum_{i , j } { \tilde{g}_{ij}^ { } } { u_\pi}^i { u_\pi}^j + \sum_{i , j , k } { \tilde{g}_{ij}^ { } } { g_{}^{jk } } { \widetilde{\nabla}_{k}^{(\mathrm{e } ) } } { u_\pi}^i \right ) \notag \\ & - \left ( \frac{1}{2 } \sum_{i , j } { \tilde{g}_{ij}^ { } } { u_{\pi'}}^i { u_{\pi'}}^j + \sum_{i , j , k } { \tilde{g}_{ij}^ { } } { g_{}^{jk } } { \widetilde{\nabla}_{k}^{(\mathrm{e } ) } } { u_{\pi'}}^i \right ) + \mathrm{o}(1 ) , \label{6 - 2}\end{aligned}\ ] ] where @xmath106 the proof of theorem [ riskdiff ] is given in the appendix .
in this section , we introduce a new metric defined by @xmath107 which we call the predictive metric . since @xmath61 is positive definite , it can be adopted as a riemannian metric on @xmath7 .
it will be shown that the predictive metric @xmath61 , the corresponding volume element @xmath108 and the laplacian @xmath109 based on @xmath61 play essential roles corresponding to those played by the fisher rao metric @xmath40 , the jeffreys prior @xmath110 , and the laplacian @xmath39 based on @xmath40 in the conventional setting where @xmath111 . here ,
@xmath112 , @xmath113 , and @xmath114 denote determinants of @xmath37 matrices @xmath115 , @xmath38 , and @xmath116 , respectively .
the @xmath36-element of the inverse of the @xmath37 matrix @xmath115 is given by @xmath117 . here
, we give an intuitive meaning of the predictive metric @xmath61 by a nonrigorous argument . in the standard estimation theory , the fisher - rao metric @xmath40 , which is the fisher information matrix , corresponds to the inverse of the asymptotic variance of the maximum likelihood estimator . in the setting we consider ,
the asymptotic variance of the maximum likelihood estimator based on @xmath20 is @xmath118 , where @xmath119 is the @xmath37 matrix @xmath38 , and the asymptotic variance of the maximum likelihood estimator based on both of @xmath20 and @xmath1 is @xmath120 , where @xmath121 is the @xmath37 matrix @xmath116 .
the inverse of the reduction of the asymptotic variance by observing @xmath1 in addition to @xmath26 @xmath122 are given by @xmath123 , as we see in example 1 in section 4 , corresponding to the predictive metric @xmath124 .
the riemannian connection coefficients with respect to the predictive metric @xmath61 are given by @xmath125 and we put @xmath126 .
then , @xmath127 in the same way , we have @xmath128 thus , @xmath129 the laplacian @xmath109 with respect to the predictive metric @xmath61 is defined by @xmath130 where @xmath131 , and @xmath43 is a real smooth function on @xmath7 . by using these quantities ,
we obtain the following theorem corresponding to in the conventional setting .
[ maintheorem ] the difference between the risk functions of bayesian predictive densities @xmath101 based on a @xmath103 and @xmath132 based on @xmath133 is given by @xmath134 \notag\\ = & \frac{\displaystyle { \mathring{\delta}}\left(\frac{\pi}{{\pi_\mathrm{p}}}\right)}{\displaystyle \left ( \frac{\pi}{{\pi_\mathrm{p } } } \right ) } - \frac{1}{2 } \sum_{i , j } { \mathring{g}_{}^{ij } } \frac{\displaystyle \partial_i \left(\frac{\pi}{{\pi_\mathrm{p}}}\right ) \partial_j \left(\frac{\pi}{{\pi_\mathrm{p}}}\right ) } { \displaystyle \left(\frac{\pi}{{\pi_\mathrm{p}}}\right)^2 } + { \mathrm{o}}(1 ) = 2 \ ; \frac{\displaystyle { \mathring{\delta}}\left(\frac{\pi}{{\pi_\mathrm{p}}}\right)^{\frac{1}{2 } } } { \displaystyle \left ( \frac{\pi}{{\pi_\mathrm{p } } } \right)^{\frac{1}{2 } } } + { \mathrm{o}}(1).\label{main}\end{aligned}\ ] ] the proof of theorem [ maintheorem ] is given in the appendix .
if there exists a positive constant @xmath135 such that @xmath136 , we identify the prior @xmath137 with @xmath19 because the posterior densities based on them are identical .
in fact , the risk difference between @xmath19 and @xmath138 coincides with that between @xmath137 and @xmath138 .
[ usefulcor ] if a positive function @xmath139 is superharmonic with respect to the predictive metric @xmath124 , i.e. @xmath140 for every @xmath141 , and the strict inequality holds at a point in @xmath7 , then the bayesian predictive density based on the prior density @xmath142 asymptotically dominates the bayesian predictive density @xmath132 based on the prior density @xmath143 .
if there exists a non - constant positive superharmonic function @xmath139 with respect to the predictive metric @xmath124 , then the bayesian predictive density based on the prior density @xmath144 @xmath145 asymptotically dominates @xmath132 . the first statement is a straightforward conclusion from theorem [ maintheorem ] .
we show the second statement .
the function @xmath146 is superharmonic because @xmath147 if @xmath139 is a positive superharmonic function .
the strict inequality holds at @xmath10 satisfying @xmath148 for any @xmath149 .
such @xmath10 exists since @xmath139 is a non - constant function .
thus , the second statement follows from the first statement . by setting @xmath150
, it follows from corollary [ usefulcor ] that the bayesian predictive density based on the prior @xmath151 asymptotically dominates the bayesian predictive density based on @xmath138 if @xmath139 is a non - constant positive superharmonic function .
note that corollary [ usefulcor ] also holds if we replace the predictive metric @xmath152 with another metric @xmath153 satisfying @xmath154 with a positive constant @xmath135 .
this is because the volume element with respect to @xmath153 is proportional to that with respect to @xmath152 and the relation @xmath155 holds , where @xmath156 is the laplacian with respect to @xmath153 .
in this section , we see three examples . we verify that the results in the previous sections are consistent with several known results in examples 1 and 2 and
obtain some new results in examples 2 and 3 .
example 1 .
normal models suppose that @xmath0 is distributed according to the @xmath6-dimensional normal distribution @xmath157 with mean vector @xmath158 and covariance matrix @xmath159 and that @xmath1 is distributed according to the @xmath6-dimensional normal distribution @xmath160 with the same mean vector @xmath158 and possibly different covariance matrix @xmath161 . here , @xmath158 is the unknown parameter and @xmath162 and @xmath163 are known .
the fisher information matrix for @xmath164 is @xmath165 and that for @xmath164 is @xmath166 , where @xmath167 and @xmath168 are inverse matrices of @xmath167 and @xmath168 , respectively .
since the coefficients of the predictive metric @xmath169 do not depend on @xmath158 , the volume element with respect to the predictive metric is @xmath170 which is the uniform distribution @xmath171 .
@xcite and @xcite considered shrinkage pri - ors for this model .
the bayesian predictive density @xmath29 dominates @xmath172 based on the uniform measure @xmath173 if @xmath174 is a superharmonic function on the euclidean space @xmath175 endowed with the metric @xmath176 , see theorem 3.2 in @xcite .
this result holds for every positive integer @xmath25 . since @xmath177^{-1 }
( n{g_{}^{}})^{\frac{1}{2 } } \\[2pt ] = & ( n{g_{}^{}})^{\frac{1}{2 } } \left [ i - i + ( n{g_{}^{}})^ { -\frac{1}{2 } } { \tilde{g}_{}^ { } } ( n{g_{}^{}})^ { -\frac{1}{2 } } + \mathrm{o}(n^{-2 } ) \right]^{-1 } ( n{g_{}^{}})^{\frac{1}{2 } } \\[2pt ] = & n^2 \ , { g_{}^ { } } \ , { \tilde{g}_{}^{-1 } } \ , { g_{}^ { } } + \mathrm{o}(n)\end{aligned}\ ] ] corresponds to the predictive metric @xmath124 , theorem [ maintheorem ] is consistent with theoretical and numerical results in @xcite and @xcite .
example 2 .
location - scale models suppose that @xmath178 and @xmath179 are probability densities on @xmath180 that are symmetric about the origin .
let @xmath181 where @xmath182 and @xmath183 are unknown parameters .
suppose that we have a set of @xmath25 independent observations @xmath184 distributed according to @xmath185 .
the variable @xmath1 to be predicted is independently distributed according to @xmath186 .
the objective is to construct a prior @xmath19 for a bayesian predictive density @xmath187 .
the fisher rao metrics on the model manifolds @xmath188 and @xmath189 are @xmath190 respectively , where @xmath191 and @xmath192 are positive constants depending on @xmath178 , and @xmath193 and @xmath194 are positive constants depending on @xmath179 .
the predictive metric is given by @xmath195 define @xmath196 by rescaling the location parameter @xmath158 .
we call this coordinate system @xmath197 the upper - half plane coordinates .
then , the predictive metric is represented by @xmath198 coinciding with the metric on the hyperbolic plane @xmath199 , which is a 2-dimensional complete manifold with constant sectional curvature @xmath200 .
thus , the model manifold endowed with the predictive metric @xmath124 is isometric to @xmath199 . the volume element with respect to the predictive metric @xmath124
is given by @xmath201 and coincides with the jeffreys priors @xmath202 for @xmath185 and @xmath203 for @xmath186 .
the laplacian on the model manifold endowed with the predictive metric @xmath124 is given by @xmath204 by corollary [ usefulcor ] , the bayesian predictive density @xmath205 based on the prior @xmath206 asymptotically dominates @xmath207 based on @xmath208 because @xmath209 by theorem [ maintheorem ] , the asymptotic risk difference is @xmath210 \notag\\ = & 2 \ ; \frac{\displaystyle { \mathring{\delta}}\left(\frac{\pi_\mathrm{r}}{{\pi_\mathrm{p}}}\right)^{\frac{1}{2 } } } { \displaystyle \left ( \frac{\pi_\mathrm{r}}{{\pi_\mathrm{p } } } \right)^{\frac{1}{2 } } } + { \mathrm{o}}(1 ) = 2 \ ; \frac{{\mathring{\delta}}\sigma^{\frac{1}{2 } } } { \sigma^{\frac{1}{2 } } } + { \mathrm{o}}(1 ) = - \frac{\tilde{b}}{2 b^2 } + { \mathrm{o}}(1 )
. \label{riskr}\end{aligned}\ ] ] in fact , it can be shown that the bayesian predictive density @xmath205 exactly dominates @xmath207 for finite @xmath25 because @xmath208 is the left invariant prior and @xmath211 is the right invariant prior with respect to the location - scale group .
the bayesian procedures based on the right invariant prior dominate those based on the left invariant prior in many problems associated with group models as shown in @xcite .
the prior @xmath211 is also derived as a reference prior , see @xcite .
furthermore , as we see below , the bayesian predictive density @xmath212 based on the prior @xmath213 defined by @xmath214 asymptotically dominates @xmath205 and thus also dominates @xmath207 . to clarify the meaning of the prior @xmath213
, we introduce another coordinate system on the model manifold .
let @xmath215 be the riemannian distance based on the predictive metric @xmath124 between a point @xmath216 and an arbitrary fixed point @xmath217 on @xmath218 .
the direction of @xmath216 from @xmath217 is represented by a point @xmath219 on the unit circle in the tangent space at @xmath217 .
then , the point @xmath216 is represented by @xmath220 and @xmath219 , see e.g. @xcite p. 152 .
this coordinate system @xmath221 is called the geodesic polar coordinates .
then , the predictive metric is given by @xmath222 the laplacian is represented by @xmath223 where @xmath224 is the laplacian on the unit circle in the tangent space at @xmath217 , see e.g. @xcite p. 158 .
when the upper - half plane coordinate system is adopted , the riemannian distance @xmath215 between @xmath197 and @xmath225 is represented by @xmath226 see e.g. @xcite p. 176 .
thus , in the original coordinate system @xmath227 , theriemannian distance @xmath215 between and @xmath228 and @xmath229 is @xmath230 thus , the ratio of prior densities is given by @xmath231 note that @xmath232 depends on @xmath227 only through @xmath233 defined by .
thus , from , , and theorem [ maintheorem ] , we have @xmath234 \notag\\ = & 2 \ ; \frac{\displaystyle { \mathring{\delta}}\left(\frac{\pi_{c,\kappa}}{{\pi_\mathrm{p}}}\right)^{\frac{1}{2 } } } { \displaystyle \left ( \frac{\pi_{c,\kappa}}{{\pi_\mathrm{p } } } \right)^{\frac{1}{2 } } } + { \mathrm{o}}(1 ) = - \frac{\tilde{b}}{b^2 } \left\ { \frac{1}{2 } + c \frac{\pi_{c,\kappa}}{{\pi_\mathrm{p } } } + \frac{3}{2 } ( 1-c^2 ) \left(\frac{\pi_{c,\kappa}}{{\pi_\mathrm{p } } } \right)^2 \right\ } + { \mathrm{o}}(1 ) \notag \\ = & - \frac{\tilde{b}}{b^2 } \left\ { \frac{1}{2 } + c \frac{1}{\cosh \rho + c } + \frac{3}{2 } ( 1 - c^2 ) \frac{1}{(\cosh
\rho + c)^2 } \right\ } + { \mathrm{o}}(1 ) , \label{riskhs}\end{aligned}\ ] ] and is smaller than when @xmath235 and @xmath236 .
the asymptotic risk difference can also be derived from and the laplacian in the original coordinate system . by corollary [ usefulcor
] , the bayesian predictive density @xmath237 @xmath238 asymptotically dominates @xmath239 since the function is superharmonic for @xmath240 .
however , @xmath237 asymptotically dominates @xmath241 only when @xmath242 .
+ { \mathrm{o}}(1 ) = -(\tilde{b}/b^2 ) \ { 1/2 + c ( \pi/{\pi_\mathrm{p } } ) + ( 3/2 ) ( 1-c^2 ) ( \pi/{\pi_\mathrm{p } } ) \}^2 $ ] for bayesian predictive densities based on @xmath138 , @xmath211 , @xmath243 , @xmath244 , and @xmath245 .
we put @xmath246 just for simplicity . ]
several properties of the function are discussed in @xcite .
as @xmath247 , the prior @xmath248 converges to the right invariant prior @xmath211 , because @xmath249 when @xmath247 . here , priors are identified up to a positive multiplicative constant .
as @xmath250 , the prior @xmath248 converges to @xmath251 because @xmath252 when @xmath250 .
the prior density with respect to the rescaled parameter @xmath197 defined by is given by @xmath253 note that the cauchy prior for @xmath254 , discussed by jeffreys and many researchers , appears in .
thus , the class @xmath248 of priors bridges the right invariant prior @xmath211 , coinciding with the reference prior , and the cauchy prior @xmath243 .
figure [ fig : location - scale - risk ] illustrates the difference between the risk functions of bayesian predictive densities based on @xmath211 , @xmath243 , @xmath255 , and @xmath256 and the risk function of @xmath239 .
the risk functions of the right invariant prior @xmath211 and the cauchy prior @xmath243 are uniformly smaller than that of @xmath138 .
the asymptotic risk of the cauchy prior @xmath243 coincides with that of @xmath211 .
furthermore , the asymptotic risks of @xmath257 and @xmath258 are smaller than that of @xmath211 for every @xmath228 . therefore , the use of @xmath259 @xmath260 is recommended .
the risk of @xmath255 is smaller than that of @xmath261 when @xmath220 is small , and vice versa .
thus , there does not exist a unique best value of @xmath135 .
the choice of the value of @xmath236 is arbitrary because it corresponds to the center of shrinkage .
finite - sample decision theoretic properties such as admissibility of bayesian predictive densities @xmath262 based on proposed priors @xmath263 @xmath264 require further research .
example 3 .
poisson models suppose that @xmath265 @xmath266 are independently distributed according to the poisson distribution @xmath267 with mean @xmath268 and that @xmath269 @xmath266 are independently distributed according to the poisson distribution @xmath270 with mean @xmath271 . here , @xmath272 are known positive constants .
the unknown parameter is @xmath273 .
the objective is to construct a predictive density for @xmath1 by using @xmath0 .
this problem in the conventional setting , in which @xmath274 , is studied in @xcite .
the laplacian @xmath281 based on the predictive metric @xmath124 is given by @xmath282 where @xmath43 is a smooth real function of @xmath283 .
define @xmath284 then , from @xmath285 and @xmath286 we have @xmath287 since @xmath288 is a non - constant positive superharmonic function of @xmath283 , the bayesian predictive density @xmath289 based on @xmath290 asymptotically dominates @xmath207 by corollary [ usefulcor ] .
the model manifold endowed with the predictive metric @xmath61 is isometric to the first orthant @xmath291 of the euclidean space @xmath292 , as we see below .
define @xmath293 then , @xmath294 thus , from , the coefficients of the metric with respect to @xmath295 are given by @xmath296 this coincides with the usual metric on @xmath297 . here , the function @xmath298 of @xmath299 is the green function of the heat equation on @xmath292 and plays an essential role in bayesian methods for model manifolds isometric to the euclidean space .
for example ,
the prior density @xmath300 for the @xmath6-dimensional normal model @xmath301 , where @xmath158 is the @xmath6-dimensional unknown mean vector and @xmath302 is the @xmath303 identity matrix , is known as the stein prior .
the bayesian predictive density based on @xmath208 is @xmath304 where @xmath305 .
the bayesian predictive density based on @xmath290 is @xmath306 we have the asymptotic risk difference @xmath307 \notag\\ = & \frac{{\mathring{\delta}}(\pi_\mathrm{s}/\pi_\mathrm{p})}{\pi_\mathrm{s}/\pi_\mathrm{p } } - \frac{1}{2 } \sum_{i , j } { \mathring{g}_{}^{ij } } \frac { \partial_i ( \pi_\mathrm{s}/\pi_\mathrm{p } ) \partial_j ( \pi_\mathrm{s}/\pi_\mathrm{p})}{(\pi_\mathrm{s}/\pi_\mathrm{p})^2 } + \mathrm{o}(1 ) \notag\\ = & - \frac{1}{2 } \left ( \frac{d}{2 } -1\right)^2 \left ( \frac{\lambda_1}{s_1 } + \dotsb + \frac { \lambda_d}{s_d } \right)^{-1 } + \mathrm{o}(1 ) \label{poissonard}\end{aligned}\ ] ] by theorem [ maintheorem ] , , , and @xmath308 the asymptotic risk difference depends on @xmath283 only through @xmath309 .
when @xmath309 is small the improvement is large , and it converges to zero as @xmath309 goes to infinity . it can be shown that @xmath290 dominates @xmath208 in the sense of infinitesimal prediction , and we can construct a bayesian predictive density dominating @xmath310 for arbitrary @xmath311 @xmath266 by modifying the prior @xmath290 .
finite sample properties of this prior will be discussed in a another paper by using an approach different from the asymptotic methods in the present paper . in examples 1 , 2 , and 3 ,
the volume element based on the predictive metric @xmath61 coincides with the jeffreys priors based on @xmath40 and @xmath312 , i.e. @xmath313 , although the three metrics are different . in general , if two metrics @xmath40 and @xmath312 satisfy the relation @xmath314 where @xmath315 is a @xmath37 regular matrix not depending on @xmath10 , then latexmath:[\[|\tilde{g}_{ij}|^{\frac{1}{2 } } = |a^l_k| |g_{ij } |^{\frac{1}{2 } } , ~~ \mbox{and } ~~ and the volume elements based on @xmath40 , @xmath312 , and @xmath61 are proportional to each other .
the relation appears in many examples as in examples 1 , 2 , and 3 .
first , we prepare a preliminary result , theorem [ risk - invariant ] , to prove theorem [ riskdiff ] . @xcite generalized these results for the setting in which @xmath26 , @xmath317 , and @xmath1 have different distributions .
the bayesian predictive density is expanded as @xmath318 \partial_k { \displaystyle \tilde{p}}(y \mid { \hat{\theta}}_{\mathrm{mle } } ) + \mathrm{o}_\mathrm{p}(n^{-1 } ) , \label{fka-1}\end{aligned}\ ] ] where @xmath319 is the maximum likelihood estimator , and @xmath320 . the estimatorminimizing the bayes risk @xmath321 \pi(\theta ) { \mbox{d}}\theta$ ]
is given by @xmath322 where @xmath323 which is a covariant vector .
the expansion of the risk function of a bayesian predictive density @xmath29 up to the order @xmath324 is given in theorem [ risk - invariant ] below .
the expansion is invariant in the sense that each term is a scalar not depending on parametrization . in theorem [ risk - invariant ] ,
we put @xmath325 here , @xmath326 and @xmath327 are vectors orthogonal to the model manifolds @xmath328 and @xmath64 , respectively .
these vectors are closely related to the curvature of the manifolds
. expansions of the risk functions corresponding to when the distributions of @xmath26 , @xmath27 , and @xmath1 are the same are obtained by @xcite for curved exponential families by using differential geometrical notions and by @xcite for general models under rigorous regularity conditions .
@xcite obtained several related results when when the distributions of @xmath26 , @xmath27 , and @xmath1 are different .
the expansion is shown by lengthy calculations parallel to those in @xcite and @xcite by using the results such as , , and obtained by @xcite .
the quantity @xmath330 is the efron curvature @xcite of the model manifold @xmath331 at @xmath10 , and @xmath332 is the mixture mean curvature discussed in @xcite of the model manifold @xmath333 at @xmath10 . from and
, we have @xmath338 let @xmath339 and @xmath340 .
then , from , @xmath341 thus , when @xmath342 , @xmath343 . from , we have @xmath344 - { { \rm e}}\big [ d ( { \displaystyle \tilde{p}}(y \mid \theta ) , { \displaystyle \tilde{p}}_{\mathrm{p}}(y \mid x^n ) \big ] \biggr ) \notag \\ = & \frac{1}{2 } \sum_{i , j } { \tilde{g}_{ij}^ { } } { u_\pi}^i { u_\pi}^j + \sum_{i , j , k } { \tilde{g}_{ij}^ { } } { g_{}^{jk } } \left(\partial_k u_{\pi}^i + \sum_l { \tilde{\gamma}_{kl}^{\,(\mathrm{e})i } } u_{\pi}^l \right ) \notag \\ & - \frac{1}{2 } \sum_{i , j } { \tilde{g}_{ij}^ { } } u_{\mathrm{p}}^i u_{\mathrm{p}}^j - \sum_{i , j , k } { \tilde{g}_{ij}^ { } } { g_{}^{jk } } \left ( \partial_k u_{\mathrm{p}}^i + \sum_l { \tilde{\gamma}_{kl}^{\,(\mathrm{e})i } } u_{\mathrm{p}}^l \right ) + { \mathrm{o}}(1 ) \notag \\ \label{riskdiff-2 } = & \frac{1}{2 } \sum_{i , j } { \tilde{g}_{ij}^ { } } ( \sum_k { g_{}^{ik } } \partial_k \log f + s^i + r^i ) ( \sum_l { g_{}^{jl } } \partial_l \log f + s^j + r^j ) \notag \\ & - \frac{1}{2 } \sum_{i , j } { \tilde{g}_{ij}^ { } } ( s^i + r^i ) ( s^j + r^j ) \notag \\ & + \sum_{i , j , k } { \tilde{g}_{ij}^ { } } { g_{}^{jk } } \left\ { \sum_l \partial_k ( { g_{}^{il } } \partial_l \log f ) + \sum_{l , m } { \tilde{\gamma}_{kl}^{\,(\mathrm{e})i } } { g_{}^{lm } } \partial_m \log f \right\ } + { \mathrm{o}}(1 ) \notag \\ = & \frac{1}{2 } \sum_{i , j } { \mathring{g}_{}^{ij } } \partial_i \log f \partial_j \log f + \sum_{i , j , k } { \tilde{g}_{ij}^ { } } { g_{}^{ik } } ( \partial_k \log f ) ( s^j + r^j ) \notag \\ & + \sum_{i , j , k , l } { \tilde{g}_{ij}^ { } } { g_{}^{ik } } \partial_k ( { g_{}^{jl } } \partial_l \log f ) + \sum_{i , j , k , l , m } { \tilde{g}_{ij}^ { } } { g_{}^{jk } } { \tilde{\gamma}_{jk}^{\,(\mathrm{e})i } } { g_{}^{lm } } \partial_m \log f + { \mathrm{o}}(1).\end{aligned}\ ] ] let @xmath345 . from , it is sufficient to show that @xmath346 is equal to @xmath347 .
since @xmath348 we have @xmath349 thus , from @xmath350 we have @xmath351 hence , because of the duality of the e - connection and the m - connection , is equal to @xmath352 berger , j. o. and bernardo , j. m. ( 1992 ) .
`` on the development of reference priors ( with discussion ) . '' in bernardo , j. m. , berger , j. o. , dawid , a. p. , and smith , a. f. m. ( eds . ) , _ bayesian statistics 4 _ , 3560 .
new york : oxford university press .
davies , e. b. ( 1989 ) . _ heat kernels and spectral theory_. cambridge : cambridgeuniversity press . | bayesian predictive densities when the observed data @xmath0 and the target variable @xmath1 to be predicted have different distributions are investigated by using the framework of information geometry .
the performance of predictive densities is evaluated by the kullback leibler divergence .
the parametric models are formulated as riemannian manifolds . in the conventional setting
in which @xmath0 and @xmath1 have the same distribution , the fisher rao metric and the jeffreys prior play essential roles . in the present
setting in which @xmath0 and @xmath1 have different distributions , a new metric , which we call the predictive metric , constructed by using the fisher information matrices of @xmath0 and @xmath1 , and the volume element based on the predictive metric play the corresponding roles .
it is shown that bayesian predictive densities based on priors constructed by using non - constant positive superharmonic functions with respect to the predictive metric asymptotically dominate those based on the volume element prior of the predictive metric . |
c + although high - mass ( @xmath6 @xmath3 ) stars are clue pieces in the universe and galaxy evolution , many questions remain open concerning their formation process .
the root of the problem is that the kelvin - helmholtz timescale for a high - mass star is much shorter than the free - fall timescale of the natal core , and thus the star reaches the main - sequence while still accreting matter .
the important amounts of uv radiation emitted by the nascent massive star should halt at some point further accretion onto the star , limiting its final mass to a certain value which is currently not well established . from a theoretical point of view , there are a number of studies showing that it is possible to form a 4050 @xmath3 star through disc - mediated accretion ( e.g. , yorke & sonnhalter 2002 ; krumholz et al .
2009 ) , and even it seems possible to form stars of up to 140 @xmath3 ( e.g. , kuiper et al . 2010 , 2011 )
. however , from an observational point of view , a clear picture of the different evolutionary stages that a massive star undergoes until reaching the main - sequence is lacking , mainly due to the fast evolution of massive stars .
beuther & shepherd ( 2005 ) and keto ( 2007 ) suggest an observational sequence for massive star formation , where the o - type stars progenitors are b - type objects embedded in large amounts of dust and gas and accreting at large mass infall rates ( 10@xmath710@xmath8 @xmath3yr@xmath9 ) . in this scenario ,
the b - type object drives a collimated molecular outflow , and while progressively moving from early - b to late - o and finally early - o , the opening angle of the outflow progressively increases .
thus , the earliest stages in the formation of an o - type star should be characterized by a central b - type object , driving a collimated outflow , and embedded in an infalling envelope massive enough to provide the b - type object with the required mass to become an o - type object .
typically , the massive young stellar objects ( ysos ) in the earliest evolutionary stages also present emission from complex organic molecules ( the so - called hot molecular cores , e.g. , cesaroni et al .
2010 ) and a lack of strong centimetre emission ( e.g. , beuther et al . 2002
; molinari et al . 1998 ; mottram et al . 2011 ) .
a possible reason for the lack of an hii region is the quenching and/or trapping of the ionized gas by the infalling dense envelope ( e.g. , walmsley 1995 ; molinari et al .
1998 ; keto 2002 , 2003 ) , or , as recently proposed by hoare & franco ( 2007 ) , an increase of stellar radius as accretion proceeds onto the star due to trapped entropy ( the so - called ` swollen ' or ` bloated ' stars ) , which produces a decrease of the stellar effective temperature and of the uv radiation emitted by the nascent star ( e.g. , yorke & bodenheimer 2008 ; hosokawa , yorke , & omukai 2010 ) .
thus , the study of outflow and accretion signatures in massive ysos is essential to progress in the understanding of massive star formation .
however , there are only a few o - type yso candidates ( i.e. , with bolometric luminosities @xmath10 @xmath5 ) associated with disc - like structures or toroids of dense gas and possibly undergoing infall ( e.g. , zhang , ho , & ohashi 1998 ; pestalozzi et al . 2004 ; sollins & ho 2005 , sollins et al .
2005a ; beltrn et al .
2006 ; beuther et al . 2007a ,
b ; beuther & walsh 2008 ; franco - hernndez et al .
2009 ; cesaroni et al .
2011 ; furuya et al .
2011 ; qiu et al . 2012
; jimnez - serra et al . 2012 ; wang et al .
2012 ) , and it is not clear if these massive ysos harbor true single o - type objects or a multiple system of early - b objects ( e.g. , naranjo - romero et al .
2012 ) . this can not be easily elucidated because most of the aforementioned massive ysos undergoing infall are not detected at optical wavelengths and their spectral type can not be assessed . on the other hand , ysos with luminosities @xmath10 @xmath5 and detected in the optical / nir
have been found associated with accreting disks through near - infrared observations ( e.g. , jiang et al .
2008 ; davies et al . 2010 ; de wit et al .
2010 ; kraus et al . 2010 ;
wheelwright et al . 2012 ) , but their number is also scarce .
thus it is extremely necessary to enlarge the sample of o - type ysos undergoing infall and actively accreting , and simultaneously detected in the optical / nir to characterize not only the infalling envelope but also the central object .
the results obtained from the joint analysis of multiwavelength data presented in this paper indicate that one new actively accreting o - type ( luminosity @xmath11 @xmath5 ) candidate is iras19520 + 2759 ( r.a.(j2000)=19@xmath1254@xmath13059 and dec.(j2000)=@xmath14280741@xmath15 ) .
the status of iras19520 + 2759 ( hereafter , ) has been controversial : it was originally suggested to be an oh / ir star based on its infrared colors ( habing & olnon 1984 ) .
however , arquilla & kwok ( 1987 ) examined the infrared astronomical satellite ( iras ) images towards this region and found the far - infrared emission to be extended .
this , together with its spectral energy distribution ( nearly flat from nir to 100 @xmath16 m ) , led these authors to conclude that this object was likely a young star embedded in an extended molecular cloud .
in addition , arquilla & kwok ( 1987 ) present a single - dish @xmath0co spectrum with broad wing emission ( fwzi @xmath17 30 km s@xmath9 ) , similar to the line wings associated with outflows in star - forming regions , and gledhill ( 2005 ) reports the detection of a polarized reflection nebula in the near - infrared .
h@xmath18o and
oh maser emission with single - peaked profiles ( i.e. , unlike double - peaked profiles in evolved stars ) lends additional credence to the identification of this source as a yso ( engels et al.1984 , lewis et al . 1985 ) .
moreover , the non detection of sio maser emission ( nakashima & deguchi 2003 ) is also consistent with the yso nature ( e.g. , zapata et al .
2009b ) . stands out in the infrared as the brightest and reddest object within a field of view of @xmath19 arcmin . a recent image from the spitzer mipsgal survey ( multiband imaging photometer for spitzer inner galactic plane survey , carey et al .
2009 ) at 24 @xmath16 m shows a very bright , compact source coincident with the iras source and surrounded by a much more extended ( @xmath20@xmath21 ) , diffuse emission tracing a large molecular cloud ( see fig . [ flargefov ] ) .
cold dust emission from this large molecular cloud is traced by bolocam observations at 1.1 mm ( rosolowsky et al .
2010 ; aguirre et al .
2011 ) , which reveal an extended ( @xmath22 arcsec ) source , of about 0.7 jy of flux density , where is embedded ( fig .
[ flargefov ] ) .
interestingly , is located at the boundary of the supernova remnant g65.1 + 0.6 in a region where the remnant is probably interacting with the surrounding interstellar medium ( fig .
[ flargefov ] ; landecker et al.1990 ; tian & leathy 2006 ) .
hrivnak et al .
( 1985 ) identified an optical counterpart candidate of , which appears as a bright , point - like source as observed with the hubble space telescope ( go 9463 , pi : r.sahai ) .
recently , snchez contreras et al .
( 2008 ) report echelle long - slit 390010900 spectroscopy of , and show that the optical spectrum is dominated by emission lines consistent with the presence of a compact , photoionized region around an early - type ( b or ealier ) star ( an in depth study of the optical spectrum is in preparation by snchez contreras et al . ) . finally , the intense h@xmath23 emission detected toward the nebular core shows broad wings ( @xmath172600 km s@xmath9 ) and a remarkable p - cygni profile with _ two _ blue - shifted absorption features , suggesting multiple stellar winds with ( projected ) expansion velocities of up to 800 km s@xmath9 . as already noted by arquilla & kwok ( 1987 ) , the galactic longitude and negative v@xmath24 of indicate that this object is very remote .
we estimate a value for its kinematic distance of @xmath25kpc , determined from the target radial velocity ( v@xmath26@[email protected] / s ) and its galactic coordinates ( @xmath28=64.8131 , @xmath29=+00.1744 ) by assuming a simple galactic rotation law and adopting a value for the a oort constant of 14.4 km s@xmath9kpc@xmath9 and a galactocentric radius of 8.5kpc ( kerr & lynden - bell , 1986 ) . as indicated by arquilla & kwok ( 1987 ) ,
this value of the distance is consistent with the non - kinematic distances , 810kpc , of two hii regions in the catalog of blitz et al .
( 1982 ) that are likely to occupy approximately the same region of the galaxy ( given their similar @xmath28 and radial velocities ) .
the range of radial velocities displayed by the hi 21cm - line emission profile , v@xmath24=[@xmath2720:@xmath2726km / s ] , at the boundary of the snr g65.1 + 0.6 , next to which and the surrounding molecular cloud are located ( fig .
1 ) , consistently yield a kinematic distance for the snr of 910kpc ( landecker et al.1990 ; tian & leahy 2006 ) . in view of these results ,
we adopt a distance of 9 kpc for .
the rest of this paper is organized as follows . in section 2
, we describe the observations , in section 3 we give the main results from the ovro 2.6 mm , and @xmath0co(10 ) , @xmath1co(10 ) and c@xmath2o(10 ) , in section 4 we derive the main physical properties of the molecular gas , outflow and central object in , and in section 5 we discuss a possible interpretation for within the massive star formation framework .
interferometric mapping of the @xmath0co , @xmath1co , and c@xmath2o @xmath30=1@xmath270 line emission at 108115ghz of was carried out using the six 10.4-m antennas millimetre array of the owens valley radio observatory ( ovro ) , which is now part of the combined array for research in millimeter - wave astronomy ( carma ) .
observations were performed in different runs between 2002 and 2004 using the compact ( c ) , equatorial ( e ) , and high ( h ) array configurations , with baselines ranging between 4.01 and 93.3 k@xmath31 in the @xmath32 plane .
the coordinates of the tracking center are r.a.(j2000)=19@xmath3354@xmath130592 and dec.(j2000)=@xmath142807415 , and the total on - source time was @xmath1718hr .
.parameters of line and 2.6 mm continuum ovro interferometric maps [ cols="<,^,^,^,^ " , ] range of values for the three next best fits .
foreground interstellar extinction .
outflow inclination with respect to the plane of the sky .
in previous sections we have shown that the ovro observations reveal a strong dust condensation , referred to as mm1 , with a total mass of around 100 @xmath3 , and which is driving a high - velocity collimated @xmath0co(10 ) outflow .
the parameters of the outflow driven by mm1 , if compared to those of outflows driven by high - mass ysos , fit reasonably well the value expected for a massive yso of @xmath34 @xmath5 ( e.g. , wu et al . 2004 ; lpez - sepulcre et al .
2009 ) , after applying a factor of @xmath35 to account for the 54% of flux filtered out by ovro ( the works reporting correlations of outflow parameters use single - dish data and thus are sensitive to the most extended emission , which has been filtered out by ovro ) . to further compare the envelope properties with previous works , we estimated the mass of the cloud associated with by using the bolocam image ( rosolowsky et al . 2010 ; aguirre et al . 2011 ,
see fig .
[ flargefov ] ) , for which we measured a flux density at 1.1 mm of 0.7 jy , corresponding to 360610 @xmath3 , for a dust temperature in the range 2030 k and using the opacity law of ossenkopf & henning ( 1994 , @xmath36 @xmath37g@xmath9 at 1.1 mm ) .
this cloud mass and the cs(10 ) linewidth that we have measured ( table [ tisotop ] ) are consistent with the correlation found by beuther et al .
( 2002 ) for a sample of high - mass protostellar objects .
similarly , the @xmath1co(10 ) linewidth that we have measured follows the correlation with the bolometric luminosity found by liu et al .
( 2010 ) for a sample of 98 massive ysos .
finally , we plotted the values of the cloud mass and the measured bolometric luminosity on the mass - luminosity plot of sridharan et al .
( 2002 ) and molinari et al .
( 2008 ) , and the values follow well the trend for massive ysos . in particular , figure 9 of molinari et al .
( 2008 ) indicates that probably is in a transition stage between actively accreting objects and the ` envelope clean - up ' phase .
this is consistent with the fact that different methods used to estimate @xmath38 yield significantly different results , suggesting a non - homogenous density structure in the envelope and possibly the presence of holes and cavities . from the millimetre continuum emission we derived , assuming an homogeneous density distribution , @xmath38@xmath39 mag ( section [ srcont ] )
this is about one order of magnitude larger than the @xmath38 estimated from the fit to the sed ( total @xmath38 is @xmath40 , section [ sased ] ) , and from the optical spectrum ( ratio of h@xmath23 to h@xmath41 yields @xmath42 mag , snchez contreras , in prep . ) , which are both along the line of sight .
overall , is probably in a phase of main accretion and simultaneous start of envelope disruption as the jet interacts with the envelope creating a cavity .
examples of massive ysos reported in the literature driving outflows and with hints of simultaneous envelope disruption are iras05506 + 2414 ( sahai et al .
2008 ) and v645cyg ( hamann & persson 1989 ; clarke et al .
2006 ; miroshnichenko et al .
2009 ) , with being more luminous , and embedded in a one order of magnitude more massive envelope .
the agreement between the position of the optical / infrared source and the millimetre source suggests that they are tracing the same object .
the position of the 2mass source is ra(j2000 ) : 19:54:05.86 , dec ( j2000 ) : + 28:07:40.6 .
the position of mm1 ( table [ tcont ] ) is offset with respect to the 2mass position by ( + 0.15 , + 0.27 ) arcsec , which falls well within the ovro ( @xmath43 arcsec ) and 2mass ( @xmath44 arcsec , skrutskie et al .
2006 ) positional uncertainties .
in addition , the preliminary analysis of the optical spectrum of presented in snchez contreras et al .
( 2008 ) points to a hot object photoionizing the circumstellar gas and driving a dense and compact wind .
thus , the properties of both the optical source and the millimetre source indicate that we are dealing with a young and massive object .
further hints on the nature of the optical source may come from the non - detection of centimetre emission : if the central star was a ( main - sequence ) o9 star with a typical ionizing flux @xmath45s@xmath9 ( e.g. , panagia 1973 ; smith , norris & crowther 2002 ) , the expected 6 cm flux density from an optically thin hii region , with an electron temperature of @xmath46 k and placed at a distance of 9 kpc , would be @xmath47 mjy , which should have been well detected with the vla observations ( section [ sovla ] ) au ) and large electron density ( @xmath48 cm@xmath8 ) , which is very unlikely because such a small size ( typical of hyper - compact hii regions , e.g. , kurtz 2005 ) would require a very dense envelope strongly infalling and quenching the ionized gas ( e.g. , walmsley 1995 ; molinari et al .
1998 ) , preventing the object to be detected in the optical ( e.g. , [email protected] : sollins & ho 2005 ; [email protected] : sollins et al . 2005 ; ngc7538-irs1 : sandell et al . 2009 ) . ] .
if the spectral type is b1 ( or later ) , the non - detection of the 6 cm emission is naturally explained , as the flux of ionizing photons in this case is smaller than for the o9 case by one two orders of magnitude . as discussed above , the centimetre properties of are most consistent with an early - b type classification for the central star ; however , the spectral type inferred from its high luminosity , @xmath11 @xmath5 , assuming it is a zero - age - main - sequence star , is significantly earlier , o6o7 ( e.g. , bernasconi & maeder 1996 ; martins et al .
the properties of can be reconciled if the object is accreting , which is strongly suggested by several observational facts : i ) the infrared source is associated with a compact and dusty envelope , mm1 , of @xmath50 @xmath3 ; ii ) mm1 is driving a high - velocity collimated outflow with a dynamical timescale of @xmath51 yr ; and iii ) the ovro c@xmath2o(10 ) emission , with a size of @xmath52 au , is elongated perpendicular to the outflow , similar to the molecular toroids found in other massive star - forming regions ( e.g. , beltrn et al .
2005 , 2011 ; beuther et al .
2007b ; zapata , tang , & leurini 2010 ; furuya et al . 2011 ) , which usually show a rotation velocity pattern .
all this suggests that has been recently undergoing active accretion .
hoare & franco ( 2007 ) have proposed , and numerical simulations have shown ( e.g. , yorke & bodenheimer 2008 ; hosokawa & omukai 2009 ; hosokawa , yorke , & omukai 2010 ) , that the main effect of accretion onto a massive star ( with mass accretion rates up to 10@xmath8 @xmath3yr@xmath9 ) , regardless of the geometry of the accretion , is an increase of the stellar radius , giving rise to the so - called ` swollen ' or ` bloated ' stars .
the ` swollen - star ' scenario naturally explains the low ionizing flux observed in , even though with a bolometric luminosity of @xmath53 @xmath5 . in this scenario ,
the stellar radius attains its maximum value when the star has accreted @xmath54 @xmath3 ( hosokawa , yorke , & omukai 2010 ) . assuming that this is the current value of the stellar mass of , and given the mass measured for the envelope of @xmath50 @xmath3 ( section 3 ) , the total mass which could still be accreted onto the star , for a star formation efficiency of about @xmath5420% , is 1020 @xmath3 , implying a final stellar mass of 2030 @xmath3 , which corresponds to an o - type star .
the ` swollen - star ' scenario has also been proposed for other massive ysos .
davies et al .
( 2011 ) require a ` swollen - star ' phase to reproduce the observed luminosity distribution of massive ysos in the sample of the red msx source survey ( rms , e.g. , urquhart et al .
2008 ; see also mottram et al .
2011 ) ; simpson et al .
( 2012 ) and bik et al .
( 2012 ) explain the low effective temperatures or offsets from main - sequence in the hertzsprung russell diagram by invoking a ` swollen - star ' phase for the massive ysos of g333.2@xmath55 or w3main complexes . finally , individual objects studied in detail within the ` swollen - star ' scenario are b275 ( ochsendorf et al .
2011 ) , m8e - ir ( linz et al .
2009 ) , crl2136 ( de wit et al .
2011 ) , and orionklirc2 ( morino et al .
1998 ; testi , tan , & palla 2010 ) . , with a bolometric luminosity around 10@xmath56 @xmath5 , would be the most luminous of the four aforementioned ` swollen - star ' candidates .
accurate spectral type classification of the central star through , e.g. , optical spectroscopy in the 40005000 range , is needed to firmly establish the ` swollen ' star hypothesis proposed for .
we present the results of ovro observations at 2.6 mm and @xmath0co(10 ) , @xmath1co(10 ) , and c@xmath2o(10 ) transitions , as well as observations of cs(10 ) with the 34 m antenna dss-54 of madrid dscc , towards iras19520 + 2759 , a very bright infrared object embedded in a massive cloud .
our main conclusions can be summarized as follows : * the 2.6 mm continuum emission is dominated by one strong millimetre source , mm1 , clearly associated with the bright infrared source , and which is only barely resolved , with a size of @xmath57 au , pa=@xmath58 , and a mass of @xmath50 @xmath3 .
about 4 arcsec to the south there is a fainter and unresolved source , mm2 , of about @xmath19 @xmath3 . *
both mm1 and mm2 are driving high - velocity @xmath0co outflows , with outflow parameters typical of massive ysos , and the outflow driven by mm1 shows a high collimation factor .
in addition , both mm1 and mm2 are associated with @xmath1co and c@xmath2o emission , whose structure , especially for c@xmath2o , is elongated perpendicularly to the outflow directions , similar to toroids seen in other massive star - forming regions . *
the cs line profile indicates that the object is embedded in dense gas with important contributions of non - thermal motions .
* we fitted the sed of mm1 using the sed fitting tool of robitaille et al .
the best fit yields a foreground extinction of 4.4 mag , an envelope mass of 140 @xmath3 , and a bolometric luminosity of @xmath11 @xmath5 . for
this bolometric luminosity one would expect that the star should emit an ionizing flux high enough to detect the free - free emission at centimetre wavelengths .
however , no source associated with iras19520 + 2759 is detected at 6 cm , with a rms noise of 0.1 mjy , which could be due to on - going accretion onto the star .
thus , we present evidence that iras19520 + 2759 is a luminous yso embedded in a massive and dense envelope , which is driving a collimated outflow and does not show strong centimetre emission .
we propose that this could be an example of a ` swollen ' or ` bloated ' star as proposed by hosokawa , yorke , & omukai ( 2010 ) , where accretion produces an increase of the stellar radius , and a decrease of effective temperature and ionizing flux .
overall , iras19520 + 2759 seems to be an excellent accreting massive yso candidate which could form an o - type star in the future .
the authors are grateful to the anonymous referee for valuable comments significantly improving the quality of the paper .
ap is grateful to tom landecker and roland kothes for kindly providing the 21 cm cgps images , and to josep miquel girart for insightful comments on the paper ; rs acknowledges useful discussions with harold yorke and takashi hosokawa .
the madrid dscc observations have been done under the host country program ; the authors acknowledge the kind support of the robledo staff during such observations .
ap is supported by the spanish micinn grant aya2008 - 06189-c03 ( co - funded with feder funds ) and by a jae - doc csic fellowship co - funded with the european social fund .
this work has been partially performed at the astrophysics department of the astrobiology center ( cab , csic / inta ) and the california institute of technology and has been partially supported by the spanish micinn through grants aya2009 - 07304 and consolider ingenio 2010 for the team ` molecular astrophysics : the herschel and alma era
astromol ' ( ref .
: csd2009 - 00038 ) . ongoing development and operations for ovro and carma are supported by the national science foundation under a cooperative agreement ( grant ast 08 - 38260 ) .
rs thanks nasa for partially funding this work by nasa ltsa and adp awards ( nos .
nmo710651/ 399 - 20 - 40 - 06 & 399 - 20 - 40 - 08 ) ; rs also received partial support for this work from hst / go awards ( nos .
go-09463.01 , 09801.01 , and 10185.01 ) from the space telescope science institute ( operated by the association of universities for research in astronomy , under nasa contract nas5 - 26555 ) .
this research has made use of the simbad database , operated at cds , strasbourg , france , the nasa s astrophysics data system , aladin , akari observations , a jaxa project with the participation of esa ; the wide - field infrared survey explorer , which is a joint project of the university of california , los angeles , and the jet propulsion laboratory / california institute of technology , funded by nasa ; the two micron all sky survey , which is a joint project of the university of massachusetts and the infrared processing and analysis center / california institute of technology , funded by nasa and the national science foundation ( nsf ) ; the midcourse space experiment , for which processing of the data was funded by the ballistic missile defense organization with additional support from nasa office of space science ; the nasa/ ipac infrared science archive , which is operated by the jet propulsion laboratory , california institute of technology , under contract with nasa ; and the bolocam galactic plane survey , made using bolocam on the caltech submillimeter observatory , operated by caltech under a contract from the nsf .
support for the development of bolocam was provided by nsf grants ast-9980846 and ast-0206158 . beuther h. , shepherd d. in _ cores to clusters : star formation with next generation telescopes _
nanda kumar , m. tafalla , p. caselli eds .
astrophysics and space science library , vol .
105 - 119 , 2005 springer new york beuther h. , zhang q. , bergin e. a. , sridharan t. k. , hunter t. r. , leurini s. , 2007a , a&a , 468 , 1045 beuther h. , zhang q. , hunter t. r. , sridharan t. k. , bergin e. a. , 2007b , a&a , 473 , 493 bik a. , et al . , 2012 , apj , 744 , 87 blitz l. , fich m. , stark a. a. , 1982 , apjs , 49 , 183 hoare m. g. , franco j. , 2007 , in diffuse matter from star forming regions to active galaxies - a volume honouring john dyson , edited by t.w .
hartquist , j. m. pittard , and s. a. e. g. falle .
series : astrophysics and space science proceedings .
springer dordrecht , 2007 , p.61 , arxiv 0711.4912 rizzo j. r. , et al . , 2012a , a&a , 542 , a63 rizzo , j. r. , gutirrez , m. , kuiper , t. b. h. , cernicharo , j. , sotuela , i. , and pedreira , a. , 2012b , in radziwill , n. m. and chiozzi , g. , eds . ,
spie 8451 , software and cyberinfrastructure for astronomy ii .
spie , washington , p. 84512d
robitaille t. p. , 2008 ,
aspc , 387 , 290 | the theory of massive star formation currently suffers from a scarce observational base of massive young stellar objects to compare with . in this paper , we present interferometric @xmath0co(10 ) , @xmath1co(10 ) , c@xmath2o(10 ) , and 2.6 mm continuum images of the infrared source iras19520 + 2759 together with complementary single - dish observations of cs(10 ) , obtained with the 34 m antenna dss-54 at the madrid deep space communications complex , as well as archive images at different wavelengths . as a result from our work ,
iras19520 + 2759 , with a controversial nature in the past , is firmly established as a massive young stellar object associated with a strong and compact millimetre source and driving a collimated outflow .
in addition , a second fainter millimetre source is discovered about 4 arcsec to the south , which is also driving an outflow .
furthermore , the two millimetre sources are associated with c@xmath2o clumps elongated perpendicularly to the outflows , which may be related to rotating toroids .
the masses of gas and dust of the millimetre sources are estimated to be around 100 and 50 @xmath3 .
mm1 , the dominant source at all wavelengths , with a total luminosity of ( 12)@xmath4 @xmath5 at 9 kpc , is however not associated with 6 cm emission down to a rms noise level of 0.1 mjy .
we propose that iras19520 + 2759 could be an example of the recent theoretical prediction of ` bloated ' or ` swollen ' star , i.e. , a massive young stellar object whose radius has increased due to effects of accretion at a high - mass accretion rate . [ firstpage ] stars : formation ism : individual objects : iras 19520 + 2759 ism : molecules |
let us consider the sequence @xmath0 of monic polynomials defined using a general @xmath1 term recurrence relation with constant coefficients @xmath2 with initial conditions @xmath3 .
define the infinite banded hessenberg matrix @xmath4 if @xmath5 , where @xmath6 means taking transpose , the recurrence relation can be viewed in matrix form as @xmath7 it is well known and easy to verify that the zeros of @xmath8 are the eigenvalues associated with the @xmath9th principal section @xmath10 of @xmath11 . for @xmath12
we define the multi - index @xmath13 where @xmath14 and @xmath15 are such that @xmath16 .
note that we have @xmath17 .
the study of the asymptotic behavior of sequences of polynomials @xmath18 satisfying higher order recurrence relations with constant or periodic coefficients and their connections with hermite - pad approximation has been a subject of major interest . to our knowledge , the first paper in this direction for four term recurrences is @xcite .
the case when @xmath19 , related with faber polynomials associated with hypocycloidal domains , was considered in @xcite and @xcite .
the situation when @xmath19 and the coefficient corresponding to @xmath20 is periodic was analyzed in @xcite and @xcite .
a situation in which all the coefficients have period @xmath21 , which extends the notion of chebyshev polynomials to the context of multiple orthogonality , was treated in @xcite ( see also @xcite ) .
these questions are linked as well with the study of vector valued continued fractions @xcite and @xcite , and the spectral analysis of banded hessenberg and toeplitz operators @xcite and @xcite . for every @xmath22 let @xmath23 denote the @xmath24 solutions of the algebraic equation
@xmath25 the rational function @xmath26 is called the symbol of the toeplitz matrix @xmath11 .
its inverse @xmath27 establishes a one to one correspondence between a @xmath24 sheeted riemann surface @xmath28 of genus zero ( see hurwitz formula @xcite ) and the extended complex plane @xmath29 .
we order the solutions of the algebraic equation by absolute value so that @xmath30 when all inequalities are strict the order is unambiguously defined .
if equalities occur , choose an arbitrary ordering so that the inequalities remain true .
set @xmath31 it is known that each @xmath32 is a disjoint union of finitely many open analytic arcs and a finite number of exceptional points with no isolated points .
all @xmath32 are unbounded except @xmath33 which is compact and connected ( see ( * ? ? ?
* proposition 3.2 ) and ( * ? ? ?
* theorems 11.9 , 11.19 ) ) .
we call attention to the fact that the labeling of the @xmath34 and @xmath32 in this paper does not match the one adopted in @xcite .
the point @xmath35 is a branch point of order @xmath36 of the algebraic function and it is also a simple zero ( so @xmath37 ) . set @xmath38 with the convention @xmath39 . then @xmath40 let @xmath41 be the unit vector in @xmath42 with @xmath43 in the @xmath44 component . define @xmath45 where @xmath46 is the resolvent operator , and @xmath47 is the standard inner product in @xmath48 . for each @xmath49
, we introduce a linear functional @xmath50 defined in the space of polynomials by @xmath51 where @xmath52 .
notice that @xmath11 defines a bounded operator and thus the functions @xmath53 admit convergent laurent expansions in a neighborhood of @xmath54 , and @xmath55 .
[ teo:1 ] we have : * @xmath56 .
more precisely , both vector functions have the same laurent expansions at @xmath54 componentwise . * for each @xmath57 ,
the polynomial @xmath58 satisfies the following orthogonality conditions @xmath59 * the vector rational function @xmath60 is a type ii hermite - pad approximant to the system @xmath61 , with respect to the multi - index @xmath62 defined in .
that is , for each @xmath49 , we have @xmath63 where @xmath64 is the @xmath65-th component of @xmath62 . * for each @xmath49 @xmath66 uniformly on each compact subset of @xmath67 .
in particular , the last assertion of the theorem implies that the zeros of the polynomials @xmath8 have @xmath33 as the set of its accumulation points .
we would like to know more about this set and the functionals @xmath68 defined above , at least under additional assumptions on the coefficients @xmath69 .
when @xmath70 , the sets @xmath32 are starlike and the functionals @xmath68 admit integral representation in terms of ( complex ) rotationally invariant measures living on @xmath33 ( see @xcite , @xcite and @xcite ) .
moreover , the measures representing the functionals @xmath68 have the structure of a nikishin system ( see below for the definition and @xcite ) .
the zeros of the polynomials @xmath8 are the eigenvalues of the @xmath9-th principal section of @xmath11 . in @xcite ,
the asymptotic distribution of these eigenvalues and other points called generalized eigenvalues is obtained by solving a vector equilibrium problem on the sets @xmath71 .
the structure of the vector equilibrium problem is the same as that appearing in finding the asympototic behavior of type ii multiple orthogonal polynomials with respect to a nikishin system of measures .
these are only a few indicators which allow to predict that there is an underlying nikishin structure in the study of and even more general recurrence relations ( see @xcite ) . in the next result we point out a new situation where this is so .
set @xmath72 the algebraic equation @xmath73 is equivalent to the characteristic equation associated with and determines the same riemann surface @xmath28 as @xmath74 . if @xmath27 denotes the inverse of @xmath26 , then @xmath75 is the inverse of @xmath76 .
notice that @xmath77 therefore , the set of critical points of @xmath78 coincides with the reciprocals of the critical points of @xmath79 .
we have @xmath80 in this case , @xmath35 is a branch point of order @xmath36 and a simple pole of the inverse of @xmath79 .
the finite branch points are given by those @xmath81 such that @xmath82 for some @xmath83 .
the set of finite branch points of the algebraic equations @xmath84 is the same .
define @xmath85 obviously @xmath86 and @xmath87 now , we restrict our attention to the case when all the zeros of @xmath88 are real and simple .
in particular , @xmath89 .
we suppose also that @xmath90 .
we will assume that @xmath91 ; otherwise , if @xmath92 then @xmath93 so that @xmath88 has real zeros and we fall in the case of the standard chebyshev polynomials of the first kind for which all is known .
in the sequel @xmath94 denotes the set of zeros of @xmath88 .
notice that @xmath95 consequently , there has to be at least one positive and one negative zero of @xmath88 .
we will assume that @xmath88 has exactly @xmath21 zeros of one sign and @xmath43 of the other .
one case reduces to the other substituting in the algebraic equation @xmath96 by @xmath97 . when there are @xmath21 negative critical points of @xmath79 we enumerate them as follows @xmath98 when there are @xmath21 positive critical points it is more convenient to index them so that @xmath99 in either cases @xmath100 denotes the collection of finite branch points of the algebraic equation @xmath73 ( or , @xmath74 ) .
define the system of absolutely continuous measures @xmath101 given by @xmath102 for each @xmath103 the product @xmath104 of two measures @xmath105 supported on @xmath106 and @xmath107 , respectively , is defined by @xmath108 where @xmath109 denotes the cauchy transform of @xmath110 .
finally , set @xmath111 we will show that in the present situation all these measures are supported on subintervals of the real line and have constant sign .
we say that @xmath112 is a ( generalized ) nikishin system generated by @xmath101 and denote this by writing @xmath113 . in standard nikishin systems the product in
is defined without the factor @xmath114 .
it has been introduced here by necessity to compensate for the denominator @xmath115 in which in turn is needed in order that @xmath116 be well defined ( note that @xmath117 is not finite ) .
that the product of measures defined above is consistent in our case is discussed during the proof of [ teo:2 ] assume that the polynomial @xmath88 has @xmath21 zeros of one sign and @xmath43 of the other . with the indexing adopted above , we have : * when the polynomial @xmath88 has @xmath21 negative zeros @xmath118 $ ] and @xmath119 $ ] if it has @xmath21 positive zeros . for @xmath120 ,
@xmath121 $ ] if @xmath122 is a local maximum of @xmath79 and @xmath123 $ ] when @xmath122 is a local minimum of @xmath79 .
* there exist a system of ( signed ) measures @xmath124 supported on the interval @xmath33 absolutely continuous with respect to the lebesgue measure such that @xmath125 * there exist real constants @xmath126 such that @xmath127 where @xmath113 is the nikishin system defined above .
* for each @xmath49 @xmath128 the @xmath9 zeros of @xmath8 are simple and lie in the interior of @xmath33 .
we divide the proof of theorem [ teo:1 ] into several parts .
* proof of theorem [ teo:1 ] i ) . *
define the operations of division and multiplication of vectors as follows @xmath129 @xmath130 these operations define what is called the generalized jacobi - perron algorithm .
applied iteratively , this algorithm allows to express a vector as a vector continued fraction . in (
* theorem 2 , remark 2 ) it was proved that the generalized jacobi - perron algorithm , applied to @xmath131 outside the spectrum of @xmath11 , gives the vector continued fraction @xmath132 @xmath133 notice that the first @xmath21 floors of this continued fraction differ , but from the floor @xmath24 on , which we call the tail , they are identical . for brevity , we refrain from writing the variable @xmath134 except where it is absolutely necessary to avoid confusion .
let us obtain an expression of the tail @xmath135 of this continued fraction in terms of @xmath136 . obviously , @xmath137 due to the way in which division and multiplication of vectors is defined we get that @xmath138 identifying the first components we obtain @xmath139 equating the rest of the components , respectively , and using it follows that @xmath140 these formulas allow to express all the components in terms of @xmath136 .
indeed , when @xmath141 we get @xmath142 substituting this in the second equality of , for @xmath143 it follows that @xmath144 applying this recursively we obtain @xmath145 combining with for @xmath146 , after some trivial manipulations we see that @xmath147 this means that @xmath148 is one of the solutions of the algebraic equation .
furthermore , using the definition of product and division of vectors , , and the first equality in , we can undo the first @xmath21 floors of the continued fraction getting @xmath149 @xmath150 @xmath151 @xmath152 since @xmath153 solves the algebraic equation the equality of the first components renders that @xmath154 because @xmath155 .
it is worth noting that this equality used in implies that @xmath156 in a neighborhood of @xmath54 .
this completes the proof . @xmath157 * proof of theorem [ teo:1 ] ii)-iv ) . *
consider the difference equation @xmath158 with initial conditions @xmath159 observe that for @xmath160 , @xmath161 is a polynomial of degree @xmath9 and @xmath162 is a polynomial of degree @xmath163 for all @xmath57 . in fact , @xmath164 since both polynomials satisfy the same difference equation with the same initial conditions .
define @xmath165 observe that @xmath166 for all @xmath167 sufficiently large . in ( * ?
* theorem 1.1 , lemma 2 ) it was proved that for each @xmath49 @xmath168 and @xmath169 where @xmath64 is the @xmath65-th component of @xmath62 . since @xmath170 , ii ) follows .
( incidentally , ( * ? ? ?
* theorem 1.1 ) also contains the vector continued expansion of @xmath171 . )
let @xmath172 be the truncation of the matrix @xmath11 to the first @xmath9 rows and columns ( @xmath173 ) .
we define @xmath174}),\qquad 1\leq j\leq p , \label{defndnj}\end{aligned}\ ] ] where @xmath175 denotes the identity matrix of size @xmath176 , and @xmath177}$ ] denotes the submatrix of @xmath178 obtained after deleting the first @xmath65 rows and columns .
set @xmath179 and take @xmath180 , @xmath181 , and @xmath182 , @xmath183 .
expanding @xmath184 by its last row , we have @xmath185 c:3 in ( * ? ? ?
* lemma 2.1 ) it was proved that the following relation holds for every @xmath57 : @xmath186 from , , , and theorem [ teo:1 ] i ) , it readily follows that @xmath187 that is , @xmath188 is a type ii hermite - pad approximation of @xmath189 with respect to @xmath190 . to complete the proof of iii ) it remains to point out that since the diagonals of the matrix @xmath11 are constant we have @xmath191 . finally , using a formula of widom ( *
* theorem 2.8 ) , in ( * ? ? ?
* proposition 5.1 ) it was shown that @xmath192 uniformly on every compact subset @xmath193 of @xmath194 , where @xmath195 is a positive constant which depends on @xmath193 .
taking convenient ratios of these polynomials the previous formula immediately gives iv ) . @xmath157
in the rest of the paper , we restrict our attention to the case when the polynomial @xmath88 has only @xmath21 zeros of a given sign ( positive or negative ) and the remaining one has the opposite sign .
as mentioned in the introduction , without loss of generality we can assume that @xmath21 zeros are negative and one positive .
-.2truein notice that @xmath196 is holomorphic in a neighborhood of @xmath54 and has a simple pole at @xmath54 . in the upper and lower half planes
there are no branch points ; therefore , @xmath197 can be extended analytically to the complement of a segment contained in the real line whose extreme points are branch points .
on @xmath198 and @xmath199 the rational function @xmath76 is strictly increasing and invertible in a neighborhood of each point .
obviously , @xmath200 because @xmath201 \cap r[x_{p+1},+\infty ) = \emptyset$ ] , since the algebraic equation defined by @xmath73 is irreducible .
define @xmath202,\qquad \tilde{\mathcal{r}}_0 = \overline{\mathbb{c } } \setminus \tilde{\gamma}_1.\ ] ] and let @xmath203 be the analytic extension of @xmath204 to @xmath205 .
notice that @xmath206 and @xmath207 are first order branch points so @xmath208 returns to its initial value over any closed cycle contained in @xmath205 .
consequently @xmath208 is single valued . on @xmath209 and @xmath210 the function @xmath79 is strictly decreasing with a local minimum at @xmath211 . by continuity , the segments @xmath212 and @xmath213 must belong to the next sheet attached to @xmath214 along @xmath215 . set @xmath216,\qquad \tilde{\mathcal{r}}_1 = \overline{\mathbb{c } } \setminus ( \tilde{\gamma}_1 \cup \tilde{\gamma}_2).\ ] ] paste @xmath205 and @xmath217 along @xmath218 in a crosswise manner as usual . through @xmath218
we can extend @xmath208 analytically to all @xmath217 and we denote the extension @xmath219 .
it is easy to see that @xmath219 returns to its initial value over any closed cycle contained in @xmath217 so it is single valued . by continuity ,
the segment @xmath220 must belong to the next sheet and @xmath221 is a local maximum of @xmath79 .
define @xmath222,\qquad \tilde{\mathcal{r}}_2 = \overline{\mathbb{c } } \setminus ( \tilde{\gamma}_2 \cup \tilde{\gamma}_3).\ ] ] paste @xmath217 and @xmath223 along @xmath224 in a crosswise manner .
we can extend @xmath219 analytically to all @xmath223 through @xmath224 and we denote the extension @xmath225 .
we can repeat these arguments until we reach the interval @xmath226 . in general terms ,
if @xmath227 , is a local maximum of @xmath79 , define @xmath228 , whereas if it is a local minimum , we take @xmath229 $ ] . set @xmath230 for each @xmath231 , we can extend @xmath232 analytically to all @xmath233 through @xmath234 and we denote the extension @xmath235 .
these extensions are all single valued and @xmath236 from the uniqueness of analytic continuation we get that for each @xmath22 , @xmath237 are a permutation of the roots @xmath238 of the algebraic equation @xmath74 . if we prove that @xmath239 then we can conclude from that @xmath240 and @xmath241 .
the rest of the proof is dedicated to verify . we do this following the scheme employed in @xcite to prove a similar result . for each @xmath242 define a measure @xmath243 on @xmath234 with density @xmath244 where the prime denotes the derivative with respect to @xmath245 , and where the @xmath246 and @xmath247 subscripts stand for the boundary values obtained from the upper or lower half of the complex plane , respectively , and @xmath248 stands for the imaginary part of a complex number @xmath249 .
the second equality in comes from the fact that @xmath250 .
notice that the density is well - defined in the interior of @xmath234 ( @xmath251 ) . from the algebraic equation it is easy to deduce that @xmath252 where @xmath253 is one of the roots of @xmath254 taken in some order depending on the branch , and
if @xmath255 is a branch point of @xmath256 @xmath257 therefore , @xmath258 and @xmath259 consequently , @xmath260 and @xmath261 in particular , - imply that the measures @xmath262 are finite .
we claim that @xmath243 is a real - valued ( possibly signed ) measure on @xmath234 with total mass @xmath263 to prove equation , using contour integration , we first derive the following relation for the cauchy transforms @xmath264 where by convention @xmath265 and @xmath266 .
we show in detail how this is done for @xmath267 .
the cases @xmath268 have a slightly different geometric configuration but are in fact simpler .
notice that @xmath269 are the branch points of @xmath256 on the boundary of @xmath233 and to fix the geometry suppose that @xmath270 .
let @xmath271 and @xmath272 , where @xmath273 is sufficiently large and @xmath274 is sufficiently small .
@xmath275 is oriented positively and @xmath276 negatively . by cauchy s integral formula
, we have @xmath277 where @xmath134 is any point lying inside the region limited by @xmath278 and @xmath279 .
in the second integral it is used that @xmath280 on @xmath281 . from - , we have @xmath282 uniformly on any compact subset of @xmath233 .
therefore , follows .
multiplying by @xmath134 , taking the limit as @xmath283 , using , and , we obtain @xmath284 ( in passing the limit under the integral sign one can use lebesgue s dominated convergence theorem due to . ) now , applying consecutively , starting with @xmath285 we obtain .
( recall that @xmath286 . ) in ( * ? ? ?
* ( 4.16 ) ) a formula similar to is obtained which eventually leads to the proof of a statement similar to that of theorem [ teo:1 ] i ) . in the situation considered in @xcite all the curves
@xmath32 and intervals @xmath287 ( denoted there @xmath288 ) are bounded and the coefficients of the recurrence relation , which depend on @xmath9 , are periodic . despite these differences , the rest of the arguments used in @xcite to prove that @xmath289 can be transposed to our situation without any change .
we limit ourselves to pointing out that aside from the measures @xmath290 which live on the intervals @xmath291 the authors also use the measures @xmath292 with density @xmath293 which live on the curves @xmath32 .
( notice that here we used the branches @xmath294 instead of @xmath295 ) .
the properties needed relative to the measures @xmath292 and the curves @xmath296 can be found in ( * ? ? ?
* theorem 2.2 ( a ) , proposition 3.2 ) .
we leave the details to the reader .
@xmath157 we wish to mention that in ( * ? ? ?
* theorem 2.3 ( b)-(c ) ) it is shown that the system of measures @xmath297 is the solution of an interesting ( nikishin type ) vector equilibrium problem on the system of curves @xmath298 ( regardless of our assumptions on the critical points ) . in conclusion
, we have @xmath299 and @xmath300 .
these facts will be used in the sequel without further notice .
perhaps the equality @xmath301 is not so obvious so we prove it .
indeed the assertion is trivial for @xmath302 . since @xmath303 is holomorphic in @xmath194 and @xmath304 @xmath305 and @xmath306 . for @xmath307
, it is easy to see that @xmath308 is holomorphic and different from zero in the simply connected region @xmath309 and @xmath310 .
therefore , @xmath311 is holomorphic in a neighborhood of @xmath32 and @xmath312 consequently , @xmath313 .
the functions @xmath314 are defined outside the spectrum of the resolvent operator @xmath315 , see . under the present assumptions
we know that they can be extended analytically to @xmath316 . for simplicity , the extension will also be denoted @xmath317 .
thus , @xmath318 among other things , we wish to show that the functions @xmath317 are cauchy transforms of measures supported on @xmath33 . for this purpose , we will use the following lemma .
[ laux ] let @xmath319 where @xmath320 is a semi - infinite interval contained in the real line .
assume that @xmath321 extends continuously to @xmath320 from the upper and lower half planes and the limiting values are denoted @xmath322 and @xmath323 , respectively .
suppose that @xmath324 and @xmath325 where @xmath326 is the finite end point of @xmath320
. then @xmath327 * proof .
* let @xmath271 and @xmath328 , where @xmath273 is sufficiently large , @xmath274 is sufficiently small , and the curves are oriented in the positive and negative directions , respectively . to fix ideas , assume that @xmath329 $ ] .
then , by cauchy s integral formula , we have @xmath330 where @xmath134 is any point lying inside the region limited by @xmath331 and @xmath320 . making @xmath332 , @xmath333 , and using the hypothesis the integral representation immediately follows .
@xmath157 now we are ready to prove assertions ii)-iii ) of theorem [ teo:2 ] .
since all the coefficients @xmath69 are real , the resolvent functions are symmetric with respect to the real line , thus @xmath334 additionally , @xmath335 therefore , using the cauchy integral theorem @xmath336 for all @xmath134 exterior to @xmath337 , where @xmath337 is a positively oriented closed jordan curve that surrounds @xmath33 . shrinking @xmath337 to @xmath33 , we find that takes place with @xmath338 thus , @xmath339 and @xmath340 is absolutely continuous with respect to the lebesgue measure .
we are done with ii ) . when @xmath302 , we get that @xmath341 this function is continuous on @xmath33 and can never equal zero in the interior of @xmath33 because if that should happen then @xmath342 , but that is impossible ( except at the end points ) because @xmath343 is one to one whereas @xmath344 and @xmath345 are distinct points on @xmath28 .
so , @xmath346 has constant sign on @xmath33 .
notice that @xmath347 , so @xmath348 transforms the lower half plane into the upper half plane ( @xmath348 only takes real values on @xmath349 ) .
therefore , in fact , @xmath346 is a positive measure . to prove iii ) we use induction .
set @xmath350 we have proved that @xmath351 is a measure with constant sign on @xmath33 and we take @xmath352 whenever possible , to reduce the notation , we refrain from writing down the independent variable , and indicate separately where the formulas take place . for @xmath231 , on @xmath33 we have @xmath353 notice that on @xmath33 @xmath354 consequently , this function extends to a holomorphic function in @xmath355 and @xmath356 we have also shown that @xmath357 in particular @xmath358 let us obtain an integral representation of the functions @xmath359 , with respect to a measure supported on @xmath360 . we know
that @xmath361 and @xmath362 from the expression of @xmath363 in , it follows that for @xmath120 @xmath364 and @xmath365 thus @xmath366 and @xmath367 from lemma [ laux ] , we obtain that @xmath368 we denote by @xmath369 the absolutely continuous measure on @xmath360 given by @xmath370 let us restrict our attention to @xmath371 . in this case , from , @xmath372 since @xmath373 and @xmath374 has constant sign on @xmath360 . indeed ,
if @xmath375 at some interior point @xmath245 of @xmath360 , then @xmath376 which is impossible because @xmath377 is one - to - one .
since @xmath33 lies to one side of @xmath360 , the integral @xmath378 also takes constant sign on @xmath33 .
putting together , , and , we obtain that @xmath379 where @xmath380 , given by @xmath381 has constant sign on @xmath33 . for each @xmath382 and @xmath383 , we define @xmath384 where @xmath385 are non negative integers , and @xmath386 then , we will show that @xmath387 and @xmath388 in particular , @xmath389 and @xmath390 is a measure with constant sign on @xmath106 . for @xmath391 , we also have @xmath392 and @xmath393 let us show that - are satisfied . using induction
we start proving that @xmath394 , @xmath395 .
we proved previously that this is true when @xmath396 .
let us assume that the statement holds up to some @xmath397 and @xmath398 and prove that it is also verified for @xmath65 and @xmath399 . by definition ( we take @xmath400 ) @xmath401 by induction @xmath402 and , obviously , @xmath403 , so we only have to verify that @xmath404 .
this is done checking that @xmath405 .
let us rewrite @xmath406 as follows ( again @xmath407 ) @xmath408 by the induction hypothesis , we have @xmath409 , in particular on @xmath410 its @xmath411 values are equal .
so we must only worry about the @xmath411 values of the remaining part of the expression .
we have @xmath412 where the first equality comes from analytic extension through @xmath410 of @xmath413 and @xmath414 , whereas the second equality is due to the symmetry of the indicated sum .
thus the statement is true . in order to derive
it is sufficient to point out that @xmath415 and use lemma [ laux ] .
formula follows directly from , the definition of @xmath416 , and the fact that @xmath417 .
then , @xmath418 has constant sign on @xmath106 because @xmath419 and @xmath420 have fixed sign on that interval . to get notice that from we have on @xmath106 @xmath421 @xmath422 making the change of variables @xmath423 and taking into consideration that @xmath424 on @xmath106
, it follows that @xmath425 since @xmath426 , according to ( substitute @xmath65 by @xmath44 ) the last double sum coincides with @xmath427 and we are done . finally , follows directly from .
now , @xmath428 has constant sign on @xmath107 , thus @xmath429 has constant sign on the interval @xmath106 which is disjoint from @xmath107 . for each @xmath103 define a measure
@xmath430 whose differential expression is @xmath431 each one of these measures has constant sign on its support .
now , set @xmath432 since @xmath433 , from the definition it readily follows that the measures @xmath434 have constant sign on @xmath33 .
obviously , @xmath435 , @xmath436 , and the proof of follows directly employing and several times .
the constant @xmath437 can be expressed in terms of the values @xmath438 .
notice that the measures @xmath439 are the measures @xmath117 defined in .
we have obtained iii ) . from what has been proved ( see , , and ) , we have for all @xmath440 @xmath441 therefore , from theorem [ teo:1 ] ii ) , it follows that @xmath442 due to and the decreasing structure of the values of the components of @xmath443 , we obtain @xmath444 this settles assertion iv ) and we conclude the proof .
let us define the measures ( see ) @xmath445 where the measures @xmath117 are given by .
let @xmath446 be a multi - index ( not necessarily staircase ) .
for each @xmath49 , consider the linear form @xmath447 ( the sum is empty when @xmath448 ) .
[ lem:5.1 ] with the notations set above , for each @xmath49 the linear form @xmath449 has in @xmath450 at most @xmath451 zeros or it is the null function . * proof . * obviously , the statement is true if @xmath448 , so we can assume that @xmath452 .
fix @xmath65 .
assume that @xmath449 is non null and has at least @xmath453 zeros in @xmath454 .
let us prove that the reduced linear form @xmath455 has at least @xmath456 zeros in @xmath457 .
the zeros of @xmath458 are symmetric with respect to the real line .
therefore , there exists a polynomial with real coefficients @xmath459 of degree @xmath460 and zeros in @xmath454 such that @xmath461 thus , for each @xmath462 @xmath463 where @xmath464 and @xmath465 let @xmath320 be a jordan curve contained in @xmath454 with winding number @xmath43 that surrounds all the zeros of @xmath459 . using cauchy s theorem and cauchy s integral formula
, we get @xmath466 @xmath467 @xmath468 this implies that @xmath469 has at least @xmath456 sign changes on @xmath107 since the measure @xmath470 has constant sign on @xmath107 . if @xmath471 then @xmath472 and consequently @xmath473 , otherwise it could not have had @xmath474 zeros in @xmath475 .
continuing the process , we conclude that @xmath476 has at least @xmath477 zeros in @xmath478 which is only possible if @xmath479 and going backward we obtain that @xmath480 , against our initial assumption
. so @xmath458 has at most @xmath474 zeros in @xmath481 .
@xmath157 in the sequel , we return to multi - indices of the form for which we write @xmath482 .
[ zeros ] for all @xmath440 the zeros of @xmath8 are simple , lie in the interior of @xmath33 , and interlace the zeros of @xmath483 . additionally , @xmath484 uniformly on compact subsets of @xmath485 , where @xmath486 denotes the conformal representation of @xmath487 onto the interior of the unit circle such that @xmath488 and @xmath489 .
* proof . * from it follows that for any system of polynomials @xmath490 , with real coefficients @xmath491 then , implies that @xmath8 has exactly @xmath9 simple zeros in the interior of @xmath33 .
indeed , assume that @xmath58 has at most @xmath492 sign changes in the interior of @xmath33 . solving a homogeneous linear system of @xmath163 equations on the @xmath9 coefficients of the polynomials
@xmath493 we can construct a non - trivial linear form with a simple zero at each of the @xmath494 points where @xmath8 changes sign and a zero of order @xmath495 at one of the extreme points of @xmath33 . from lemma [ lem:5.1 ] , the corresponding linear form can have no other zero in the interior of @xmath33 except those imposed . substituting in this linear form
we reach a contradiction .
let @xmath496 be real constants such that @xmath497 . from it follows that for any system of polynomials @xmath490 , with real coefficients @xmath498 using the property of the linear form @xmath499 it follows that @xmath500 has at least @xmath9 sign changes in the interior of @xmath33 .
therefore , all the zeros of @xmath500 are simple and lie on the real line . in order to prove that the zeros of @xmath483 and @xmath8 interlace let us show first that these polynomials do not have common zeros . to the contrary , suppose that there exists a point @xmath501 such that @xmath502 .
as we have seen , @xmath503 must be a simple zero of each one of these functions .
therefore , @xmath504 .
thus , there must exist real constants @xmath505 different from zero such that @xmath506 this means that the function @xmath507 has a double zero at @xmath503 which is impossible due to what was proved in the previous sentence .
fix @xmath508 and set @xmath509 .
that is , we take @xmath510 and @xmath511 .
let @xmath512 be two consecutive zeros of @xmath483 in @xmath513 and let @xmath514 .
then @xmath515 the function @xmath516 is real when restricted to @xmath517 .
we have @xmath518 .
let us assume that @xmath519 for some @xmath520 .
since @xmath521 for all @xmath522 we obtain that @xmath523 has a zero of order @xmath524 ( with respect to @xmath96 ) at @xmath525 which is impossible . consequently , @xmath526 takes values with constant sign for all @xmath514 . at the extreme points @xmath527 this function
can not be equal to zero because the polynomials @xmath528 do not have common zeros . by continuity ,
@xmath529 preserves the same sign on all @xmath530 $ ] ( and , consequently , on @xmath531 $ ] ) .
thus @xmath532 @xmath533 since @xmath534 we obtain that @xmath535 hence , there must be an intermediate zero of @xmath8 between @xmath536 and @xmath537 . now that we know that the zeros of @xmath8 lie in @xmath33 , from theorem [ teo:1 ] iii ) and theorem [ teo:2 ] i ) , for each @xmath49 @xmath538 using theorem [ teo:2 ] iv ) ( or the interlacing property of the zeros of the monic polynomials @xmath8 ) the family of functions @xmath539 , is uniformly bounded on each compact subset of @xmath487 . fix a compact set @xmath540 and fix @xmath541 sufficiently close to @xmath43 so that @xmath542 lies in the unbounded connected component of the complement of @xmath543
. then @xmath544 and @xmath545 where @xmath546 is a positive constant independent of @xmath9 . using the maximum principle it follows that @xmath547 taking the @xmath548 root , @xmath549 as @xmath550 , and letting @xmath551 we have .
assume that @xmath113 is the ( generalized ) nikishin system given in theorem [ teo:2 ] and @xmath552 the multi - index .
let @xmath8 be the @xmath9-th monic multiple orthogonal polynomial with respect to @xmath553 .
define recursively @xmath554 [ lem:6.1 ] for each @xmath49 @xmath555 where the limit is taken along any curve non - tangential to @xmath106 at @xmath54 .
* let @xmath88 be an arbitrary polynomial of degree @xmath556 . by theorem [ teo:2]-iv )
@xmath557 which means that @xmath558 this is for @xmath559 .
moreover , using theorem [ teo:2]-iv ) for @xmath120 and @xmath560 , we have @xmath561 @xmath562 in the third equality , the definition of @xmath563 and fubini s theorem are used .
the conditions of fubini s theorem hold because @xmath564 has compact support and @xmath565 . in the fourth equality
, we use that @xmath566 and that @xmath8 is orthogonal to all polynomials of degree @xmath567 with respect to @xmath564 .
the second integral after the last equality is @xmath568 so for each @xmath120 and @xmath569 @xmath570 assume that for some fixed @xmath571 we have and for each @xmath572 and @xmath573 @xmath574 obviously , using induction , we complete the proof if we show that then holds for @xmath44 and also prove that whenever @xmath575 we have for @xmath576 and @xmath577 @xmath578 in fact , with @xmath579 implies that for any polynomial @xmath88 of degree @xmath580 @xmath581 that is @xmath582 observe that @xmath583 ( uniformly with respect to @xmath134 non - tangential to @xmath107 ) and @xmath584 so the last integral is well defined and , therefore , @xmath585 as needed . on the other hand , with @xmath586 ( in case that @xmath587 ) and @xmath588 gives @xmath589 @xmath590
@xmath591 @xmath592 as needed , since the second integral in the last line is @xmath593 . in the third equality ,
the use of fubini s theorem is justified because @xmath594 and @xmath595 when @xmath596 and @xmath597 .
thus , we have proved the assertion and the proof is complete .
we have the following widom type formulas for the second type functions .
[ widom ] for each @xmath598 , we have @xmath599 ( the product is replaced by @xmath43 when @xmath600 ) .
moreover , @xmath601 uniformly on each compact subset of @xmath602 .
* proof . *
notice that formula can be computed for all @xmath603 where @xmath604 consists of the endpoints of the intervals @xmath605 which are the only points @xmath134 where one may have @xmath606 with @xmath607 . for @xmath608 ,
is widom s formula ( see ( * ? ? ?
* theorem 2.8 ) ) . for @xmath609
, we will prove the formula by induction on @xmath610 and for @xmath611 we know it holds .
let us assume that holds true when @xmath610 is replaced by @xmath612 and we will prove that it is also satisfied for @xmath610 . the function @xmath613 satisfies the following boundary value problem : * @xmath614 . * it verifies the jump relation @xmath615 * @xmath616 remains bounded near the endpoint(s ) of @xmath617 * @xmath618 where @xmath619 is independent of @xmath9 indeed , @xmath620 and @xmath621 follow from the definition of @xmath616 and the sokhotski - plemelj formula .
@xmath622 follows from the behavior of @xmath623 near the finite endpoint(s ) of @xmath624 and the analyticity of @xmath625 in a neighborhood of such points .
finally , @xmath626 is a consequence of .
given @xmath627 , this boundary value problem has a unique solution . to prove it suffices to show that its right hand side , which we denote @xmath628 , verifies a weaker version of the boundary value problem .
let @xmath629 be the set of finite branch points of @xmath27 .
then , * @xmath630 . * it verifies the jump relation @xmath631 * near each @xmath632 @xmath633 * @xmath634 in fact , assume that for some @xmath635 holds for @xmath636 and let us prove that it also holds for @xmath610 .
suppose that ( a)-(d ) hold . from ( a ) ,
( b ) , ( a ) , ( b ) , and the induction hypothesis it follows that @xmath637 then , ( c ) and ( c ) imply that the difference is also holomorphic at the points in @xmath604 so it is an entire function . finally , from ( d ) and ( d ) , using liouville s theorem , we have @xmath638 as needed .
let us check ( a)-(d ) .
obviously , @xmath639 .
on the other hand , it is finite except possibly where @xmath640 for @xmath641 ; that is , on points in @xmath642 , and @xmath628 is a symmetric function of @xmath643 which does not depend on @xmath644 . therefore , it is analytic in @xmath645 . regarding the jump condition
, we start our with the terms in @xmath646 @xmath647 now , @xmath648 is symmetric in @xmath649 and @xmath650 , and does nt depend on @xmath651 so its @xmath247 and @xmath246 boundary values on @xmath652 coincide . on the other hand
, @xmath653 presents the same type of symmetry . in conclusion , @xmath654 @xmath655 for the remaining term , we have @xmath656 and @xmath657 .
so @xmath658 @xmath659 @xmath660 @xmath661 again the symmetry with respect to @xmath662 is used .
having in mind , , and the induction hypothesis @xmath663 is obtained . finally , ( d ) is a consequence of the behavior of the functions @xmath664 at @xmath54 , while ( c ) follows from their behavior at the finite end points of the intervals @xmath106 .
therefore is satisfied . to prove it is sufficient to notice that on any compact subset @xmath665 we have @xmath666 and use .
@xmath157 for the case of standard nikishin systems and generating measures with compact support ,
the following statements ( and even slightly stronger ones ) are known to hold ( see ( * ? ? ?
* propositions 1 , 2 ) ) .
[ teo:3 ] for each @xmath103 , the following orthogonality relations hold @xmath667 where @xmath668 and @xmath669 .
set @xmath670 . for @xmath671
has exactly @xmath672 zeros in @xmath673 they are simple and lie in the interior of @xmath107 .
@xmath674 has no zero in @xmath675 . for @xmath676
has at most @xmath677 zeros in @xmath475 and at least @xmath678 sign changes on @xmath107 .
let @xmath679 be the monic polynomial whose roots are the zeros of @xmath680 on @xmath475 ( repeated according to multiplicity ) .
we have @xmath681 and for @xmath103 @xmath682 where @xmath683
. * proof . * for @xmath559 , formula was verified in the initial induction step during the proof of lemma [ lem:6.1 ] . for the other values of @xmath65 , is almost except a higher number of orthogonality relations .
however , due to theorem [ widom ] , now we know that @xmath684 , which allows to increase the number of orthogonality relations obtained in up to the quantity specified in . indeed , notice that @xmath685 for all @xmath686 ( see last sentence in the proof of lemma [ lem:6.1 ] ) . using for @xmath559 and @xmath120
, we obtain @xmath687 where @xmath688 and the sum is empty when @xmath689 .
using lemma [ lem:5.1 ] it readily follows that @xmath568 has at least @xmath690 sign changes in @xmath360 .
let @xmath691 be the collection of all the zeros of @xmath568 in @xmath692 ( repeated according to multiplicity ) .
let us prove that @xmath691 has exactly @xmath690 points showing thus that all the zeros of @xmath568 in @xmath692 are simple and coincide with the points where this function changes sign on @xmath360 .
assume that @xmath691 contains more than @xmath690 points .
then , there exists a monic polynomial @xmath693 with real coefficients , whose zeros are points in @xmath691 such that @xmath694 on the other hand , @xmath695 , so @xmath696 let @xmath697 .
choose a curve @xmath320 with winding number @xmath43 that surrounds @xmath33 .
using cauchy s and fubini s theorems , we have @xmath698 @xmath699 this relation is clearly impossible because @xmath700 and @xmath8 would be orthogonal to itself .
therefore , @xmath691 contains exactly @xmath690 points .
define @xmath701 as the monic polynomial of degree @xmath690 whose simple zeros are the points in @xmath691 . repeating the arguments above with @xmath701 replacing @xmath702 and taking @xmath703 , we obtain .
formulas and coincide when @xmath704 .
fix @xmath705 .
assume that @xmath680 has at most @xmath677 zeros in @xmath475 and at least @xmath706 sign changes on @xmath107 .
for @xmath559 the assumption is true .
let us prove that @xmath593 has at most @xmath707 zeros in @xmath481 and at least @xmath708 sign changes on @xmath709 .
using for the index @xmath44 , we have @xmath710 where @xmath711 and the sum is empty if @xmath712 . from lemma [ lem:5.1 ]
it follows that @xmath593 has at least @xmath708 sign changes on @xmath709 .
let @xmath713 be the collection of all the zeros of @xmath593 in @xmath481 ( counting multiplicities ) .
let us prove that @xmath713 contains at most @xmath714 points .
assume that @xmath713 contains at least @xmath715 points .
then , there exists a polynomial @xmath716 with real coefficients , whose zeros are points in @xmath713 such that @xmath717 on the other hand , @xmath718 , so @xmath719 let @xmath720 .
choose a curve @xmath320 with winding number @xmath43 that surrounds all the zeros of @xmath721 .
using cauchy s theorem and the definition of @xmath593 , we have @xmath722 if @xmath723 , from fubini s theorem and cauchy s integral formula , it follows that @xmath724 when @xmath725 , we also employ to get @xmath726 @xmath727 @xmath728 these relations imply that @xmath680 has at least @xmath729 sign changes on @xmath107 which contradicts the induction hypothesis that @xmath680 has at most @xmath677 zeros in all of @xmath730 .
therefore , the induction is complete and the assertion about the zeros of @xmath731 follows .
the absence of zeros of @xmath674 in @xmath732 is a trivial consequence of formula for @xmath600 . in order to prove ,
define @xmath733 as the monic polynomial whose roots are the points in @xmath713 and repeat the arguments employed above replacing @xmath721 with @xmath733 .
we are done .
let @xmath734 denote the toeplitz matrix associated with the symbol @xmath735 .
more precisely , if @xmath10 is the @xmath9-th principal section of @xmath11 in and @xmath736 the identity matrix of order @xmath737 , then @xmath734 is the matrix of order @xmath737 obtained from @xmath738 after eliminating its first @xmath176 rows and last @xmath176 columns . for each @xmath739 ,
set @xmath740 in particular , @xmath741 .
the set @xmath742 is called the @xmath176-th generalized spectrum of @xmath10 and its points @xmath176-th generalized eigenvalues ( so , @xmath743 is the normal spectrum ) .
these generalized spectra are intimately connected with the curves @xmath744 .
define @xmath745 as the set of all points @xmath22 such that there exists a sequence @xmath746 with @xmath747 , and @xmath748 as the set of all points @xmath22 such that there exists a subsequence @xmath749 and @xmath750 . also define @xmath751 where in the sum each @xmath134 is counted according to its multiplicity as a zero of @xmath752 . in (
* theorem 2.6 ) it was proved that @xmath753 and @xmath754 for every every bounded continuous function @xmath755 on @xmath478 , where the measures @xmath756 are those defined in . for @xmath285 ,
this result is due to schmidt and spitzer @xcite .
we point out that these results were obtained in @xcite with no restriction on the coefficients @xmath69 in . in the case we are considering ,
we have carried out numerous numerical experiments in which the generalized eigenvalues lie exactly on the corresponding segments @xmath744 . set @xmath757 in @xcite
the authors give a different representation of the polynomials @xmath752 in terms of minors of the riemann hilbert matrix @xmath758 for @xmath759 , define @xmath760 then , see ( * ? ? ?
* proposition 2.6 ) @xmath761 where @xmath762 the functions @xmath763 and @xmath764 are connected .
in particular , @xmath765 . consequently , @xmath766 due to the zeros of @xmath767 and @xmath768
coincide .
[ interlacing ] @xmath768 has exactly @xmath769 zeros , they are simple and interlace the zeros of @xmath568 on @xmath770 . * proof . * due to @xmath771 fix two consecutive zeros of @xmath568 on @xmath360 , say @xmath772 . then @xmath773 between @xmath774 an @xmath211 there is exactly one simple zero of @xmath775 , because its zeros and those of @xmath776 interlace .
the proof of this statement may be carried out in the same fashion as the interlacing property of the zeros of the polynomials @xmath8 ( see , for example , ( * ? ? ?
* theorem 2.1 ) ) . on the other hand
, @xmath8 has constant sign on @xmath360 so @xmath777 we conclude that @xmath778 has an intermediate zero .
taking into consideration the order at infinity of the functions @xmath568 from one sees that @xmath779 ; therefore , @xmath780 , its zeros are simple , and interlace those of @xmath568 on @xmath360 .
@xmath157 define @xmath781 where in the sum each @xmath134 is counted according to its multiplicity as a zero of @xmath763 in @xmath782 . using the interlacing property proved in proposition [ interlacing ] , the fact that @xmath783 is absolutely continuous with respect to the lebesgue measure on @xmath360 (
thus , it has no mass points ) , and , it is not hard to prove that holds for @xmath302 if we replace @xmath784 by @xmath785 . in particular ,
a property similar to is verified by the zeros of the functions @xmath568 on @xmath360 . for @xmath267
, it would be interesting to prove : * replacing @xmath786 by @xmath787 . * a property similar to for the zeros of the functions @xmath763 on @xmath744 . * an interlacing property between the zeros of @xmath788 and @xmath752 .
here we consider a cubic polynomial , where @xmath88 has two real negative roots and a real positive one .
a special case of this example was considered by coussement et .
al @xcite , and duits and kuijlaars @xcite .
we begin with @xmath789 in this case @xmath790 let @xmath791 .
if we set @xmath792 then @xmath88 has as roots @xmath793 and @xmath794 .
also @xmath79 is given by @xmath795 the images of the roots of @xmath88 under @xmath79 are @xmath796 , @xmath797 and @xmath798 .
we look for the root of @xmath799 that tends to zero as @xmath134 tends to infinity which is @xmath800 where @xmath801 with @xmath802 @xmath803 and @xmath804 the principal branch of the logarithm is the one used above .
it is not difficult to show that @xmath805 since @xmath806 , and @xmath807 the cuts are @xmath808 $ ] and @xmath809 $ ] .
the measure @xmath564 will be supported on @xmath33 . on this interval
we see that @xmath810 and since @xmath811 and @xmath812 , @xmath813 is also positive . furthermore , from the cardano formulas we find that @xmath814 so @xmath815 is negative on this interval .
likewise , @xmath816 is positive on @xmath33 .
examination of the phase shows that for small positive imaginary @xmath134 the imaginary part of @xmath817 is positive .
thus , we find that @xmath818 for @xmath819 , and @xmath820 this implies that on @xmath33 , @xmath821 @xmath822 a. i. aptekarev and v. a. kalyagin .
analytic properties of two - dimensional continued p - fraction expansions with periodic coefficients and their simultaneous pad - hermite approximants .
rational approximation and applications in mathematics and physics .
lecture notes in math . vol .
1237 , springer verlag , berlin , 1987 , 145160 . | the study of sequences of polynomials satisfying high order recurrence relations is connected with the asymptotic behavior of multiple orthogonal polynomials , the convergence properties of type ii hermite - pad approximation , and eigenvalue distribution of banded toeplitz matrices .
we present some results for the case of recurrences with constant coefficients which match what is known for the chebyshev polynomials of the first kind .
in particular , under appropriate assumptions , we show that the sequence of polynomials satisfies multiple orthogonality relations with respect to a nikishin system of measures .
_ keywords and phrases . high order recurrence relation , hermite - pad approximation , multiple orthogonality , nikishin system _ + _ a.m.s .
subject classification .
_ primary : 30e10 , 42c05 ; secondary : 41a20 . |
in the past three decades , many theoretical efforts have been devoted to reveal and investigate the rich and novel phenomena in the photodetachment spectra of negative ions in external fields , among which the closed - orbit theory developed around 1987@xcite supplies a general semiclassical picture for the external - field effects ( note that a similar idea can also be found in ref.@xcite ) .
other more accurate but much more complicated considerations have been made which also have achieved great success@xcite .
both the predicted oscillatory structures above threshold and the quantum tunneling effect below threshold have been observed clearly in the related experiments@xcite .
very recently , based on our previous studies on the photodetachment of negative ion in an external field , and stimulated by an experiment of the washington group @xcite where they reported the observed angle - resolved photoelectron spectra from doubly charged anions having the structure of a linear chain , @xmath1 ( with @xmath2 ) , we analyzed in detail the spatial interference structures on a screen for a theoretical model of negative - ion photodetachment near a repulsive center@xcite . in this case
, a unique closed orbit can be found on the line between the electron source and the repulsive - force center . in the present paper ,
we use a well - established h@xmath0 model to investigate the effect in the photodetachment spectra induced by this returning closed orbit .
after briefly depicting the theoretical model and the electron - wave propagation after detachment in sec .
, we first use the standard closed - orbit theory@xcite to obtain a formula for the oscillatory structure in the photodetachment spectra above the zero - field threshold
. then we extend it naturally to the energy range below the zero - field threshold by introducing a `` uniform approximation '' ( sec . ) .
to understand the oscillations of the photodetachment spectra better , a comparison with the photodetachment spectra in a homogeneous electric field is made in sec . .
a brief conclusion is given in sec .. atomic units are used throughout this work unless specified otherwise .
here , we set the detached - electron source to be the origin of coordinate frame , and the @xmath3-axis is along the direction from the repulsive - force center to the electron source ( fig .
1 ) . to distinguish the new frame from the previous one in ref.@xcite
, we use a prime to denote the coordinates ( @xmath4 , @xmath5 ) relative to the repulsive center .
the zero - field photodetachment of h@xmath0 has been well established in the literature@xcite .
here we only sketch the theoretical model briefly . initially the electron is loosely bound by a short range potential and the initial bound state can be written as @xmath6 where @xmath7 is a normalization constant and @xmath8 with the binding energy @xmath9 .
after irradiation by laser light linearly polarized along the @xmath3-direction , the generated directly - outgoing wave is given by @xmath10 where @xmath11 is the initial ejection angle of the photodetached electron , and @xmath12 with @xmath13 denoting the initial kinetic energy .
@xmath14 satisfies @xmath15 .
actually , one can find the following calculation is independent of such a surface @xmath16 .
after being detached from the loosely bound initial state , the detached - electron motion is governed by the following hamiltonian @xmath17 where @xmath18 is the distance between the wave source and the repulsive center , and @xmath19 denotes the negative - charge number at the force center .
the classical trajectory analysis for the present model has been presented in our previous work@xcite , from which one can immediately find a unique closed orbit lying on the line between electron source and the repulsive - force center ( see fig . 2 in ref .
@xcite ) .
all the photoelectron wave propagates away from the source region to infinity except a small part of the outgoing wave that returns to the source region along the closed orbit .
semiclassical propagation of the detached - electron wave along the closed orbit gives the returning wave as , @xmath20 where the maslov index @xmath21 equals one as this closed orbit touches a caustic surface just once at the turning point ; @xmath22 and @xmath23 are , respectively , the semiclassical amplitude and the accumulated phase ( action variation ) of an associated wave propagating along the closed orbit .
when the detached - electron wave comes back to the source after semiclassical propagation along closed orbit , from eqs.(23 ) and ( 26 ) in ref.@xcite , we get the following expression for the semiclassical amplitude , @xmath24 with , @xmath25 where @xmath14 is an initial spherical surface around the source , and @xmath26 is the defined scaled energy as before .
hence , we get the amplitude for the returning wave as , @xmath27 the returning wave propagating along the closed orbit can be obtained by substituting the above eqs.(2 ) , ( 7 ) and ( 10 ) into eq.(4 ) .
then , after a further connecting procedure on a bound surface @xmath14 as usual@xcite , we get the returning wave near the wave source as , @xmath33~.\ ] ] the returning wave in eq.(11 ) will overlap and interfere with the wave source , and oscillatory structures will be observed in the total photodetachment cross section described quantitatively in the next section .
in this section , we use the standard procedure of the closed - orbit theory to derive a cross section formula above the zero - field threshold , and then we extend the obtained expression to a uniform one which is applicable for the energy range very close to the zero - field threshold and even below threshold . according the closed - orbit theory@xcite , the oscillatory term in the photodetachment spectra contributed by the closed orbit is given by @xmath34 where @xmath35 denotes the absorbed photon energy ; @xmath36 and @xmath37 denote the light speed and the dipole operator , respectively . by using the dipole operator ( here , @xmath38 ) , the initial bound state in eq.(1 ) and the returning wave in eq.(11 ) , the integration in eq.(12 ) can be written out as , @xmath39\int^\infty_0dr\big[r^2e^{-k_br}\int^{1}_{-1}\tau e^{ikr\tau } d\tau\big]~,\ ] ] where , @xmath40 . and the integration in eq.(13 ) can be worked out straightforward as @xmath41=\frac{4ki}{(k_b^2+k^2)^2}~.\ ] ] from the above eqs.(12)-(14 ) , the oscillatory term induced by the closed orbit can be obtained as @xmath42 combining the smooth background without any external field@xcite , the total photodetachment cross section for the simple h@xmath0 model near a repulsive coulomb center can be written out completely as , @xmath43 where we substituted @xmath22 by eq.(7 ) . very close to the zero - field threshold ,
the cross - section expression in eq.(16 ) is not appropriate .
besides , slightly below the zero - field threshold , the loosely - bound electron can also tunnel out through the repulsive potential , for which a consistent formula is also needed .
first , we transform the formula in eq.(16 ) into a product of the zero - field cross section and a field - induced modulation function as following @xmath44 and separate the modulation function into @xmath45 where @xmath46 using the semiclassical amplitude in eq.(7 ) and the action variation in eq.(10 ) , the following limiting behavior can be found when the zero - field threshold is approached : @xmath47=1~.\ ] ] hence in the modulation function only the first part @xmath48 will survive .
as the photon energy is approaching the zero - field threshold , the closed orbit mentioned above will be shorter and shorter , and the pencil of the detached - electron wave propagating near this closed orbit will be mainly affected by the repulsive potential very close to the wave source .
therefore , given that the spatial gradient of repulsive potential is proportional to @xmath49 , a linear approximation is possible for the repulsive potential very close to the zero - field threshold , and the limiting behavior of the surviving modulation function @xmath48 should be similar with that induced by a uniform electric field . by carefully examining the photodetachment process in a uniform electric field , we find the amplitude @xmath50 and the action phase @xmath51 along closed orbit has a very simple but interesting relation , @xmath52 once substituting the above equation into eq.(18 ) , one can immediately find that the modulation function @xmath53 becomes identical to @xmath48 in this limit . on the other hand , for the photodetachment of negative ion in a uniform electric field , several quantum cross - section expressions have been obtained by fabrikant et .
al.@xcite , and an completely analytic @xmath54-source model has also been well established by bracher et .
al using a quantum - source theory@xcite , from which an exact form of the modulation function can be abstracted .
accordingly , based on the above considerations , we replace @xmath55 by an exact form of the modulation function @xmath56 in the uniform - field case but with a different action variation @xmath23 , which we call `` uniform approximation '' . by this
the modulation function in eq.(18 ) can be extended as @xmath57 where @xmath58 is a _ real variable _ defined as @xmath59 and @xmath60~.\ ] ] for the uniform modulation function in eq.(22 ) , the second part in eq.(18 ) has also been uniformized using the airy function and its derivative , similar to what we did in ref . @xcite . in fig.2
, we display the behavior of the photodetachment spectra for h@xmath0 near a repulsive center using a representative case with @xmath66 and @xmath67 , where both the oscillatory structure above the zero - field threshold and the quantum tunneling effect below threshold can be observed clearly . an exact quantum calculation based on the coulomb green s function
is also displayed in fig.2 , which confirms the accuracy of the present formula .
the exact quantum treatment will be presented in a separate paper@xcite .
finally we discuss the shift of photo - electron threshold caused by the repulsive center . according to the modulation function in eq.(22 ) , the action value in eq.(28 ) indicates that the photodetachment cross section vanishes completely once the scaled energy @xmath68 is smaller than minus one .
therefore , the present theory suggests that , as a result of the presence of the repulsive center , the photo - electron threshold is actually shifted to a lower value ( @xmath69 ) relative to the zero - field photodetachment threshold .
this theoretical result is consistent with the observations reported by the washington group in their experiments@xcite .
by comparing the cross - section formula in eq.(16 ) with that in a uniform electric field@xcite , an interesting phenomenon can be observed : _ the modulation amplitude in the photodetachment spectra induced by a nearby repulsive center is identical to that induced by an effective uniform static field strength of @xmath70 _ , though the phase is different .
the equivalence of oscillatory amplitudes in the total cross section might be a surprise at first glance , because the differential cross sections are completely different in ref.@xcite . to examine this further , we compare in fig.3 the total cross section for the photodetachment of hydrogen negative ion near a repulsive center with that in a uniform electric field .
indeed , we can observe the same magnitude of the oscillations for the two different systems , and the physical reason can be found in the following . given the generality of the derivation presented in this work , the formulas in eqs.(15 ) , ( 17 ) and ( 22 ) also represent the oscillatory behavior of the photodetachment spectra in a uniform electric field but with amplitude @xmath50 and phase @xmath51 for the returning wave along the closed orbit .
therefore , to compare the two systems , we examine the related quantities in the photodetachment spectra which are , respectively , @xmath22 and @xmath23 for the photodetachment near a repulsive center and @xmath50 and @xmath51 for the photodetachment in a homogeneous electric field .
let us first examine the phase .
for photodetachment in a uniform field @xmath71 , the modulation phase can be worked out as , @xmath72 which is compared in fig.4(a ) with the phase @xmath23 in eqs .
( 10 ) and ( 27 ) by using @xmath73 and @xmath74 for example . actually , from eq.(29 )
one can find the ratio of @xmath23 to @xmath51 depends only on the scaled energy ( fig.4(b ) ) .
in fig.4 we can find that only close enough to the threshold energy , the two phases are approximately equal ; as the energy increases from the zero - field threshold , the ratio monotonically decreases from unity . at any scaled energy above threshold ,
the phase accumulation along the closed orbit in a repulsive force field is always smaller than that in the corresponding uniform static field .
both the above two points are revealed clearly in fig.3 . for the amplitudes @xmath22 and @xmath50 , one can easily find the following relationship @xmath75 which suggests the following two points .
( 1 ) the ratio of @xmath23 to @xmath51 displayed in fig.4(b ) is also the ratio of @xmath76 to @xmath77 for the photodetachment near a repulsive center , which measures the contribution of the second part in eq.(22 ) when the photon energy is away from the zero - field threshold .
( 2 ) the amplitude @xmath22 for the photodetachment near a repulsive center is exactly equal to the amplitude @xmath50 for the photodetachment in a homogeneous electric field , which is the reason why the oscillation amplitude of the photodetachment spectra in fig.3 is identical . to see this more clearly from a dynamical view
, we show in fig .
5 a pencil of trajectories propagating alongside the closed orbit .
the paths with turning points near @xmath78 ( red online ) are for a repulsive coulomb field , while those with turning points near @xmath79 ( green online ) are for the corresponding constant electric field . when the trajectories go out and then back to the dashed curve ( blue online ) , each `` red '' curve going out from the source at angle @xmath11 crosses the `` green '' curve that went out at that same angle .
therefore the flux densities through surfaces perpendicular to the trajectories must be equal , and those surfaces coincide at the source .
based on the preceding work@xcite , we have explored the total photodetachment cross section near a repulsive center by using a well - established h@xmath0 model . using the standard procedure in the closed - orbit theory
, we derived a cross - section formula for the oscillatory structures in the photodetachment spectra above the zero - field threshold .
and then , by introducing a `` uniform approximation '' , the cross - section expression has been extended to a uniform formula which is applicable for the energy range below the zero - field threshold . by comparing with the familiar case of photodetachment in a homogeneous electric field
, we find the oscillatory amplitudes for the two different systems are identical , and the ratio of their oscillatory phase is only dependent on a scaled energy .
n. d. gibson , b. j. davies , and d. j. larson , phys .
a * 47 * , 1946 ( 1993 ) ; n. d. gibson , b. j. davies , and d. j. larson , phys .
a * 48 * , 310 ( 1993 ) .
n. d. gibson , m. d. gasda , k. a. moore , d. a. zawistowski , and c. w. walter , phys .
a * 64 * , 061403(r ) ( 2001 ) ; and @xmath67 . the uniform field strength is selected as @xmath82 which is equal to @xmath83 in atomic units .
the bottom abscissa shows the ticks for the photon energy as usual , and the corresponding scaled energy values are given at the top of the figure . ]
accumulated by the detached electron after moving along the closed orbit near a repulsive center ( solid curve with @xmath66 and @xmath67 ) is compared with that in a uniform field ( dashed line with @xmath84 ) .
note the action value in the negative scaled energy range is pure imaginary .
( b ) the ratio of @xmath23 and @xmath51 is displayed and is only dependent on the scaled energy ( see the context ) .
the dotted line denotes the location of unity .
note in subplot ( b ) the solid curve also presents the ratio between @xmath76 and @xmath77 for the photodetachment near a repulsive center . ] , @xmath74 and @xmath85 .
the range of the initial ejection angles is from zero to five degree relative to the closed orbit ( note the different scales of horizontal and vertical axes ) .
the dashed curve indicates a curve on which the two trajectories ( red vs. green online ) starting at the same ejection angle cross each other . on this curve
, the flux density through surfaces perpendicular to the trajectories must be equal . at the source ,
those surfaces coincide , and are perpendicular to the closed orbit , therefore the flux densities of returning waves for the red and green ( online ) families are equal at the closed orbit , and therefore the magnitudes of interference oscillations are equal . ] | we study the total photodetachment cross section of h@xmath0 near a repulsive center .
an analytical formula for the photodetachment cross section is obtained using the standard closed - orbit theory and extending it to the energy range below the zero - field threshold .
the formula is found to be accurate by comparing with an exact quantum calculation .
a comparison with the photodetachment cross section in an effective homogeneous electric field is made , and we discuss the similarities and differences of the two systems . |
in this contribution , we describe the environment around our 1971 paper `` new dual quark models '' @xcite and the two companion papers `` the two faces of a dual pion - quark model '' @xcite and `` quantum solitons which are su(n ) fermions '' @xcite in 1971 and 1975 , which laid the foundations of non - abelian current - algebraic internal symmetry on the string . the background for our contributions included many helpful discussions with other early workers , including h.m.chan , c. lovelace , h. ruegg and c. schmidt ( during kb s 1969 visit to cern ) as well as r. brower , s. klein , c. thorn , m. virasoro and j. weis ( with mbh at berkeley ) .
both of us also acknowledge many discussions with y. frishman , g. segre , j. shapiro and especially s. mandelstam .
later discussions with i. bars and w. siegel are also acknowledged by mbh . with apologies to many other authors then , we reference here only the work which was most influential in our early thinking : before veneziano , there had been widespread interest in the quark model of gell - mann and zweig , including the four - dimensional current algebra ( adler and dashen @xcite ) of quarks . in 1970 , the work of fubini and veneziano @xcite together with the virasoro algebra @xcite provided an _ algebraic formulation _ of open bosonic string theory which we found quite exciting .
the first ingredients of this formulation are the bosonic oscillators @xmath2= -m\,g_{\mu\,\nu}\,\delta_{m+n,0},\nonumber\\ & & m , n\in { \mathbb z},\,\ , g= { \rm diag}(1,-1,\dots,-1)\ , \qquad \mu,\nu= 0,1 , ... 25\end{aligned}\ ] ] which we have rescaled here as the modes of an abelian current algebra on the circle . the conformal - field - theoretic convention for the local currents associated to these modes are @xmath3 where @xmath4 are the fubini - veneziano fields and @xmath5 is canonically conjugate to the momenta @xmath6 .
normal - ordered exponentials of the fubini - veneziano fields are the string vertex operators , to which we shall return below . in our discussions we shall emphasize the mode algebras , but for various modern applications the complex variable @xmath7 in the local fields can be taken in many forms , including @xmath8 ( the circle or loop ) , @xmath9 ( left - movers on a closed string ) , the euclidean version @xmath10 for conformal field theory ( cft ) , or the real line for open strings .
these alternate forms led to some confusion in the early days of string theory , but in fact most of the early work reviewed in this article was in the context of the open string , using the parametrization @xmath11 on the circle . finally , the virasoro algebra of the critical open bosonic string is @xmath12&=&(m - n)\ , l(m+n)+\frac{26}{12}\ , m\ , ( m^{2 } -1)\ , \delta_{m+n , 0}\end{aligned}\ ] ] where the symbol @xmath13 is bosonic normal - ordering . under these generators
the abelian currents @xmath14 covariantly transform with conformal weight equal to one .
taken together , the algebraic statements in eqs .
( 1.1)-(1.3 ) provided the first example of modern cft ( see belavin , polyakov and zamolodchikov @xcite ) .
we have here taken an historical liberty , because the central extension of the virasoro algebra was not observed until 1974 , and even then only in a private communication from j. weis quoted in ref .
@xcite ( chodos and thorn ) . throughout this article
we adopt cft conventions , including an extra factor @xmath15 ( see eq . ( 1.1 ) ) for the mode expansions of virasoro primary fields with conformal weight @xmath16 .
the cft convention guarantees that a given virasoro primary field creates a virasoro primary state of the same conformal weight as @xmath7 goes to @xmath17 on the vacuum ( in this case the zero - momentum vacuum @xmath18 of the open bosonic string ) . on the other hand
, the extra factor @xmath15 must be removed for exact correspondence of the local fields with @xmath11 in early work on the open string . as it appeared then , the open bosonic string had two deficiencies beyond the tachyonic ground state .
in the first place lie - algebraic internal symmetries were implemented only by the multiplicative chan - paton factors @xcite , which live on the boundary of the string .
second , the bosonic string had no spin , that is , no space - time fermions . our first attempt to solve the second problem was a generalization of chan - paton factors to include spin @xcite , but this multiplicative approach was unsuccessful because it included negative - norm states .
it seemed natural then to approach the problems of spin and internal symmetry in a different way , by introducing new local `` quark '' fields on the string now called _ world - sheet fermions _
@xcite : @xmath19 whose modes obey anticommutation relations @xmath20_{+}&=&[\bar{\psi}^{i}(q),\bar{\psi}^{j}(p)]_{+ } = 0,\cr [ \psi_{i}(q),\bar{\psi}^{j}(p)]_{+}&=&\delta^{j}_{i}\,\delta_{q+p,0 } , \,\,\,\,\,\,q , p\in { \mathbb z}+\frac{1}{2}.\end{aligned}\ ] ] as in the four - dimensional quark model itself , the generic indices @xmath21 were designed so that the world - sheet fermions could be space - time spinors @xmath22 or carry internal symmetry or both @xmath23 .
the virasoro generators of our world - sheet fermions are @xmath24 with fermionic normal - ordering and central charge @xmath25 equal to the number of ( @xmath26 ) pairs . in what follows
, we often refer to our world - sheet fermions as the _
dual quarks_. under these virasoro generators , the dual quark fields @xmath27 are virasoro primary fields with conformal weight @xmath28 and trivial monodromy .
the half - integer moding of the dual quark fields guaranteed a unique vacuum ( or ground state ) @xmath29 which is annihilated by the positive modes of both fields , and the adjoint operation is defined for internal symmetry as @xmath30 ( with a @xmath31 modification for spin ) . in either case , our virasoro generators satisfy the generalized hermiticity @xmath32 and the vacuum state conditions : @xmath33 with internal symmetry only , our dual quark models then provided the first examples of _ unitary _ cft , with a positive - definite hilbert space on a non - degenerate vacuum with conformal weight @xmath34 .
in modern parlance , the half - integer moded complex bh fermions are related to anti - periodic weyl fermions on the world sheet , but it should be born in mind that fermionic world - sheet action formulations of open string theory had not been studied at that time .
were these the first world - sheet fermions ?
in fact , our work was essentially simultaneous with that of ramond @xcite : our paper was submitted in november of 1970 while ramond s paper with integer - moded majorana - weyl world - sheet fermions was submitted in january 1971 ; the two papers were published in the same issue of physical review . as we now know
the r fermions @xmath35 , with their somewhat - counterintuitive space - time vector label @xmath36 and superconformal symmetry , provided the correct covariant description of space - time fermions and in a perfect match to the r construction half - integer moded majorana - weyl world - sheet fermions @xmath35 were later introduced for the bosonic sector by neveu and schwarz @xcite .
for the historical record , the lure of our seemingly more physical space - time spinor fields @xmath22 persisted for some time , leading to many connections ( halpern @xcite , @xcite , @xcite , bardakci @xcite , halpern and thorn @xcite , @xcite and mandelstam @xcite , @xcite ) between our fields and those of the rns formulation ( we mention in particular the first example of half - integer - moded scalar fields in @xcite ) .
our original intuition was however not realized until 1982 , when a world - sheet fermion with 10-dimensional majorana - weyl spinor indices was successfully incorporated by green and schwarz in the light - cone formulation of superstring theory @xcite . for continuity with the bosonic string ,
we begin this discussion with a footnote of our 1971 paper which observed that the spatial components @xmath37 of the abelian currents ( 1.1),(1.2 ) could be constructed out of our world - sheet fermions @xcite : @xmath38 this result is now recognized as a simple but important part of bose - fermi equivalence on the open string , a subject to which we will return below .
our next construction was the independent discovery of affine lie algebra in physics @xcite , still sometimes called the `` dual quark model '' , in which the non - abelian currents are naturally constructed as dual - quark bilinears with conformal weight 1 : @xmath39 here @xmath41 can be any irreducible matrix representation of any simple lie algebra @xmath42 , with structure constants @xmath43 and killing metric @xmath44 .
then one finds that the modes of the dual quark - model currents @xmath45 are the generators of the _ affine lie algebra _
@xmath46&=&i\,{f_{a b}}^{c}\,j_{c}(m+n)+ m\,\left(k\ , \eta_{a b}\right)\,\delta_{m+n , 0},\cr a,\,b,\,c & = & 1 . .
.dim(g),\,\,\,\,\,\,\,\,m,\,n\in { \mathbb z}\end{aligned}\ ] ] where the quantity @xmath47 in the central term is called the _ level _ of the affine algebra .
( we assume in this article that the highest root - length squared of lie @xmath42 is two , in which case the level in the dual quark models is also equal to the dynkin index of irreducible representation t. ) affine lie algebra is also known as centrally- extended non - abelian current algebra on the circle . in this construction , the dual - quark fields @xmath48 transform respectively in the @xmath49 representations of lie @xmath42 the lie algebra being generated by the zero modes @xmath50 of the affine algebra .
in fact our first example of this construction used the @xmath51 cartesian gell - mann matrices @xmath52 of @xmath0 ( and the dynkin index of the fundamental of @xmath1 is 1 ) so the _ first concrete representation of affine lie algebra _ was @xmath53 , that is , level one of affine @xmath0 ( see also sec .
1.8 ) . in this case , as in the four - dimensional quark model itself , the dual - quark fields @xmath27 transform respectively in the @xmath54 representations of @xmath0 .
dual quark - model constructions of the currents of @xmath55 and @xmath56 were also included in our 1971 paper , and further studied along with those of higher lorentz algebras in the first companion paper @xcite .
the dual quark models of @xmath57 were given later in the second companion paper @xcite .
because connections between open strings and field theory were not understood at the time , our interpretation of the central terms in our string - theoretic constructions of affine lie algebras on the circle was based on the presence of such a term in the abelian mode algebra ( 1.1 ) of the open bosonic string to which our constructions reduced in the abelian case .
this understanding was supported by our fermionic construction ( see eq .
( 1.7 ) above ) of the modes of the abelian current algebra .
on the other hand , we often think now of the central term in affine lie algebra as an analogue on the circle of the equal - time schwinger term @xcite ( indeed , see eq .
( 3.10 ) in ref .
@xcite ) , so we briefly sketch the history of @xmath1 current algebra in the two - dimensional lorentz - invariant quantum field theory of massless free fermions : the equal - time schwinger term in @xmath1 current algebra on the line was first computed in 1969 ( see coleman , gross and jackiw @xcite ) , and the one - dimensional @xmath1 current algebras of the light - cone currents @xmath58 with one - dimensional schwinger terms were given later by dashen and frishman @xcite .
in fact however the momenta are continuous in this non - compactified , lorentz - invariant context , so our open - string mode forms of the fermions and affine lie algebra on the circle @xcite do not appear in these papers . for a modern introduction to affine lie algebras in cft , see the work by gepner and witten @xcite and also the representation theory for all integer levels in kac @xcite . in physics
the affine algebras have sometimes also been called kac - moody algebras , but this term is more properly reserved for the general algebraic systems including affine and hyperbolic algebras ( see sec . 1.8 ) . the _ affine - sugawara constructions _
@xcite @xcite are the simplest realization of virasoro generators as quadratic forms on the generators of affine lie algebras , @xmath59&=&(m- n)\,l_{g}(m+n)+\frac{c_{g}}{12}\,m\ , ( m^{2 } -1)\,\delta_{m+n,0},\cr c_{g}&=&\frac{2\,k\,dim(g)}{2\,k+q_{g}}. \label{suga}\end{aligned}\ ] ] the quantity @xmath60 is the quadratic casimir of lie @xmath42 , which in our present root - normalization is twice the dual coxeter number of @xmath42 . the original sugawara model @xcite , sommerfield @xcite , bardakci , frishman and halpern @xcite , bardakci and halpern @xcite , gross and halpern @xcite , coleman , gross and jackiw @xcite was in four dimensions on a different current algebra , the so - called algebra of fields .
the affine - sugawara ( or sugawara - like ) constructions in eq .
( 1.11 ) hold for all integer levels of any affine lie algebra .
the first example of the affine - sugawara construction @xcite used the @xmath53 dual quark - model currents of the previous section ( the dual coxeter number of @xmath1 is @xmath61 ) .
the affine - sugawara construction on the level - one currents of affine @xmath62 was also given in our 1971 paper , and extended in the first companion paper @xcite to the construction on the level - one currents of affine so(3,1 ) and higher lorentz algebra .
the case of @xmath57 was given later in @xcite .
we emphasize that the prefactors @xmath63 in these constructions follow from careful consideration of the normal ordering @xmath13 of the current modes .
the affine - sugawara constructions , being quartic in our dual quarks , were called `` current - current '' or `` spin - spin '' interactions in those days , and models which used the dual quarks or the affine - sugawara constructions instead of extra dimensions on the open string were known as `` additive '' models . in physics the general form ( 1.11 ) of the affine - sugawara construction
was first given by knizhnik and zamolodchikov @xcite thirteen years after our examples , and used by these authors to find the general kz equations . in the same year , the important paper by witten @xcite gave the action formulation and non - abelian bosonization of these operator systems , leading to their understanding especially at higher level as interacting quantum field theories of the wzw type @xcite .
the independent history of the affine - sugawara constructions in mathematics is noted in sec .
1.8 . in this connection
, we also sketch the history of the @xmath64 sugawara - like constructions in two - dimensional lorentz - invariant field theories of massless free fermions : using a point - splitting regularization of the products of u(n ) currents , ref .
@xcite showed the equivalence of the two - dimensional analogue @xmath65 of the original sugawara stress - tensor to the free - fermionic stress - tensor \{@xmath66 , and ( building on the abelian case in ref .
dellantonio , frishman and zwanziger @xcite ) normal - ordered light - cone analogues of the u(n ) sugawara construction were later given in ref .
@xcite ( eqs .
( 5.15)-(5.16 ) of ref .
@xcite also gave ( and solved ) light - cone analogues of the @xmath64 kz equations for the fermionic four - point function . ) .
again however , these non - compactified , lorentz - invariant results had continuous momenta , so the corresponding affine - sugawara constructions do not appear in these papers .
( see also refs .
@xcite , @xcite ) .
we begin this discussion with the results in our paper @xcite . in particular for our example on @xmath53
, we first studied the addition to the affine - sugawara construction of a term @xmath67 for all constant @xmath68 , which gave virasoro generators with _ continuous _
su(3)-symmetry breaking .
this was the first example of what is today known as inner - automorphic twists or c - fixed _ conformal deformations _ ( freericks and halpern @xcite ) .
next we looked for su(3)-symmetry breaking solutions on @xmath53 when the virasoro generators were assumed to have the more general quadratic form @xmath69 with constant coefficients @xmath70 .
we stated that solutions beyond the affine - sugawara construction for the coefficients @xmath71 existed , giving virasoro generators with _
@xmath0-symmetry breaking .
this was the first _ implicit _ mention of coset constructions , but we gave no details about these solutions noting only that they all had `` infinite - degeneracy '' problems .
this issue ( amplified in our non - additive `` spin - orbit '' constructions of the next section of this reference ) was that the solutions to the virasoro conditions come in _ commuting _
`` k - conjugate pairs '' of virasoro generators @xmath72=0 $ ] which sum to the affine - sugawara construction @xmath73 . this phenomenon was called _
k - conjugation covariance _ @xcite , @xcite , @xcite , @xcite , and the associated spectral degeneracy of each commuting k - conjugate partner , due to the other , was called k - degeneracy . the first _ explicit _
examples of non - abelian coset constructions were given in the companion paper `` the two faces of a dual pion - quark model '' @xcite , which solved the virasoro conditions for all possible ( bh quartic ) conformal spin - spin interactions among the four - dimensional dual quark bilinears t , v , a , p and s ( \{@xmath74 and so on ) .
all 31 of these conformal constructions were affine - sugawara or coset constructions , and a number of these were included explicitly in the applications of the text ( see eqs .
( 3.7)(3.9 ) and ( 3.14 ) ) .
each of these results was presented as a decomposition of an affine - sugawara construction into a commuting k - conjugate pair of conformal constructions which we would today write as @xmath75=0\ , \ ] ] where @xmath42 is the original symmetry algebra , @xmath76 is any subalgebra of @xmath42 and @xmath77 is the coset space .
we will focus here on the lie - algebraic identifications in the simplest case , given as @xmath78 in eq .
( 3.14 ) of the paper .
all three sets of virasoro generators are defined in eqs .
( 3.9 ) and ( 3.14 ) as different linear combinations of the bh quartics @xmath79 and @xmath80 , and it is not difficult to see that that @xmath81 and @xmath82 are affine - sugawara constructions respectively on @xmath55 and @xmath83 . therefore , eq . ( 3.14 ) can be read as the k - conjugate decomposition @xmath84 into the commuting k - conjugate parters on so(3,1 ) and so(4,1)/so(3,1 ) .
in fact , this result was checked explicitly with the identities given in the text between the bh dual - quark bilinears t , a and certain five - dimensional ns bilinears ( see eq . ( 2.9 ) of the text ) . using these identities ,
it was noted that each quartic bh virasoro generator collapses to an ns bilinear , the affine - sugawara constructions @xmath81 and @xmath82 on @xmath55 and @xmath83 being equal respectively to the four- and five - dimensional quadratic ns virasoro generators , while the non - abelian coset construction @xmath85 ( with central charge @xmath86 ) is equal to the quadratic virasoro generators of the fifth ns operator alone .
similar quartic - to - bilinear equivalences had been seen earlier in @xcite , @xcite .
the general form of the coset construction for all @xmath87 @xmath88 was given by goddard , kent and olive @xcite fourteen years after our examples .
the k - degeneracy of the virasoro construction on @xmath77 is now understood as the local gauge invariance associated to the h - currents of the commuting k - conjugate affine - sugawara construction @xmath89 on h .
it should be emphasized however that the coset constructions are by no means the end of cft constructions on the currents of affine lie algebras : our original method of solving for the coefficients @xmath90 in quadratic forms @xmath91 eventually led to the completely k - conjugation - covariant _ virasoro master equation _ ( halpern and kiritsis @xcite ) , which includes the affine - sugawara constructions and the coset constructions as very special cases ( with rational central charge ) of irrational conformal field theory ( halpern , kiritsis , obers and clubok @xcite ) .
there is one more important strand which belongs in this story namely the second companion paper in 1975 `` quantum solitons which are su(n ) fermions '' @xcite , which described using compactified spatial dimensions on the open string
the first vertex - operator constructions of world - sheet fermions and affine lie algebra . by 1972 ,
interest among physicists was shifting back to field theory .
we mention in particular the bose - fermi equivalence of the sine - gordon and interacting thirring model studied by coleman @xcite and mandelstam @xcite in 1975 .
( see also the free - field bosonization of the single fermion bilinears by kogut and susskind @xcite . )
one of us ( mbh ) realized that with two observations this development could be applied to our 1971 non - abelian current - algebraic constructions on the string .
the first observation was that mandelstam s normal - ordered line - integral construction of a single interacting fermion could be easily generalized to @xmath92 interacting fermions on the line by including appropriate klein transformations @xcite .
this led to bosonic realizations of centrally - extended @xmath1 current algebras on the line , and the equivalence of various generalized sine - gordon models with corresponding sets of interacting @xmath64 thirring models .
in particular , the @xmath1 currents were constructed from ( n-1 ) independent two - dimensional bosons , showing the roots of su(n ) in the off - diagonal currents .
( see also the extension of these ideas to the bosonization of two - dimensional non - abelian gauge theories @xcite and in particular eq.(3.13 ) of this reference . )
the second observation in the 1975 paper was that , for massless free fermions and bosons , there was a structural parallel between mandelstam s fermions on the line and particular string vertex operators on the circle .
indeed , a formal projective map ( from the line to the circle ) was given between each two - dimensional left- and right- mover free - field construction in the field theories and a _ pair _ of our open - string counterparts on the circle , a structure which corresponded in fact to closed strings .
these points were emphasized in the second appendix `` connections with dual models '' of that paper which also gave the vertex - operator construction of many world - sheet fermions @xcite @xmath93 where @xmath94 are the fubini - veneziano fields and @xmath95 are the klein transformations of the text .
this relation guaranteed that the world - sheet fermions were virasoro primary fields with conformal weight @xmath28 under the open - string bosonic virasoro generators , and we identified these constructions as our half - integer - moded dual quarks @xcite .
although it was not explicitly mentioned , this identification had in fact been checked from vacuum expectation values using the natural identification of the bosonic zero - momentum vacuum @xmath96 as the unique vacuum @xmath29 annihilated by the positive modes of the dual quarks .
the vertex - operator construction of @xmath57 can therefore be obtained immediately in a single step by direct substitution of the fermionic vertex - operators ( 1.14 ) into our dual quark - model construction @xmath97 of the currents .
we regret having omitted this last simple step in the appendix , thinking it implicit in the context of the paper : the off - diagonal currents @xmath98 of @xmath57 follow immediately , showing the roots of @xmath1 on ( n-1 ) independent fields , and the diagonal currents of @xmath57 are a derivative of the same ( n-1 ) fields , while the extra @xmath99 current is a derivative of the normalized orthogonal sum @xmath100 .
we state here the result only for @xmath101 @xmath102 which is in fact the only case where ( as discussed in the text ) the klein transformations can be chosen to cancel .
the charged currents @xmath103 of @xmath104 show the roots of @xmath105 , and results for the neutral current @xmath106 and the extra @xmath99 current @xmath107 are of course closely related to our earlier bose - fermi relation on the string given in eq .
( 1.7 ) above .
we concluded that we had constructed a current - algebraic @xmath105 from a single compactified spatial dimension , and also that the paper was equivalent to constructing as much as current - algebraic @xmath108 from the compactified extra spatial dimensions of the critical open bosonic string .
( quantized extra momenta were well - known in early open - string theory ) . in the following year
, banks , horn and neuberger @xcite again studied the bosonization of @xmath1 thirring models , this time with periodic boundary conditions as @xmath109 . translating the fermionic zero modes of their solution at fixed @xmath110 into vertex - operators on the string ,
one now understands that integer - moded complex versions ( cr ) of world - sheet r fermions and @xmath57 currents can also be obtained from the @xmath111 vertex operators by choosing degenerate ground states with @xmath112 for each of the @xmath61 momenta . in later work ( see for example frenkel , lepowsky and meurman @xcite and lepowsky @xcite )
, one learns that the true vacuum of the full cft is the bh vacuum @xmath113 and products of bh currents on the bh vacuum live on the root - lattice of su(n ) .
the bh theory is itself reducible , containing many copies of the integrable representations of su(n ) each copy living at different @xmath16 s on the weight lattice of su(n ) .
so ended our three contributions @xcite , @xcite , @xcite to current - algebraic internal symmetry on the string . loosely speaking
, these papers also marked the end of the first string era , as an increasing number of physicists including ourselves turned their interest to gauge theory for a decade .
we were therefore fundamentally surprised to see in 1985 that the seminal paper on the heterotic string ( gross , harvey , martinec and rohm @xcite ) used the 1980 vertex - operator ( root - lattice ) construction of @xmath114 by frenkel and kac @xcite . following this , we contacted james lepowsky and igor frenkel , who patiently exchanged information with us about these parallel developments in mathematics and physics , including explanations of the first vertex - operator construction ( twisted @xmath104 ) in mathematics ( lepowsky and wilson @xcite ) , and the earlier kac - moody algebras @xcite , @xcite which include the affine and hyperbolic algebras as special cases . in mathematics ,
our 1971 dual quark - model construction of @xmath53 is known as the _ first concrete representation of affine lie algebra _ , including the first explicit central term . in this connection
we quote first from page 6304 of a 1980 paper by frenkel @xcite : `` ten years ago two physicists , bardakci and halpern , in ref .
@xcite constructed a representation of the subalgebra @xmath115 of @xmath116 in the space @xmath117 ( see formulas 3.1 - 3.11 of ref .
@xcite ) . at that time
the theory of affine lie algebras began to take its first steps . '' a less technical form of this statement is found on page 36 of the introduction to the review article by frenkel , lepowsky and meurman @xcite : `` in their proposal of current - algebraic internal symmetry for strings , bardakci and halpern @xcite , independently of mathematicians , discovered affine lie algebras in 1971 , including the first case of a concrete representation a fermionic realization of @xmath118 ; irreducibility issues arose later in independent discoveries by mathematicians . '' the early vertex - operator constructions in physics are also well - known in mathematics .
in this connection , we find again on the same page of ref .
@xcite : `` the untwisted vertex - operator construction of @xmath119-representations from compactified spatial dimensions was implicit in @xcite , @xcite . ''
a supplementary statement is found on page 365 of ref .
@xcite : `` it turned out that vertex operators and symmetry are closely related .
instances of this had already been discovered in physics , including the works @xcite and @xcite . '' . in conclusion , one might say that for us , ignorance was bliss or , more optimistically that physicists can create mathematics . in either case , it is clear that both fields have profited handsomely from this confluence . in our historical discussion
above we have focused on the discovery of current - algebraic cft on the string worldsheet , but to complete the modern perspective we should emphasize that this is just one of many mathematical methods which have been found and developed in string theory . over the last 25 years , this domain has been the main source of new methods in theoretical physics that deeply changed the discipline and had many other applications in different areas , such as statistical mechanics and condensed matter physics .
these developments involved the latest mathematical advances in algebra , topology and geometry .
some of these results ( such as the affine algebras described here ) were first discovered in string theory and then studied systematically by mathematicians , while others were imported from mathematics and made clear and practical by the string applications . this communication and cross - fertilization has also had a large impact in mathematics , where the field theory methods are now well appreciated
( see for example the lecture notes in ref .
@xcite ) . in the 80 s and 90 s
, edward witten has played a crucial role in this interdisciplinary endeavor . in our context
we note in particular the important set of string - theory mathematical methods given by general two - dimensional conformal field theory , which began with the analysis by belavin , polyakov and zamolodchikov @xcite .
they used the mathematical results of kac @xcite and feigin and fuchs @xcite on representation theory of the virasoro algebra , and realized their relevance for exact descriptions of two - dimensional critical phenomena ( see for example the collection of important papers in @xcite ) .
deligne , p. , etingof , p. , freed , d.s .
, jeffrey , l. , kazhdan , l. , morgan , j. , morrison , d.r . and witten , e. , eds . , ( 1999 ) ,
_ quantum fields and strings : a course for mathematicians _ 2 vols . , ( american mathematical society , providence ) .
lepowsky , j. ( 2007 ) . some developments in vertex operator algebra theory , old and new , in : lie algebras , vertex operator algebras and their applications , international conference in honor of j. lepowsky and r. wilson , ed . by y .- z . huang and k.c .
misra _ _ c__ontemp .
442 , p. 355 . | we briefly recall the historical environment around our 1971 and 1975 constructions of current - algebraic internal symmetry on the open string .
these constructions included the introduction of world - sheet fermions , the independent discovery of affine lie algebra in physics ( level one of affine @xmath0 ) , the first examples of the affine - sugawara and coset constructions , and finally from compactified spatial dimensions on the string the first vertex - operator constructions of the fermions and level one of affine @xmath1 .
[ 1996/06/01 ] |
room temperature atomic vapors are a popular medium for quantum information processing .
perhaps , they offer the simplest approach to quantum repeaters @xcite , quantum memory @xcite and numerous other quantum - enhanced protocols possible with deterministic light - atom interface @xcite and long - lived ground state coherence .
room temperature vapors can be effectively utilized in two distinct ways .
atoms can be placed in a paraffin coated cell without a buffer gas which they traverse multiple times during the interaction so that their quantum state is symmetric with respect to the permutations and within the holstein - primakoff approximation the cell holds a single bosonic mode @xcite . at the opposite extreme ,
the atoms are kept in a buffer gas which should virtually immobilize them . then transverse and longitudinal patterns can be stored and interfaced to @xcite .
room temperature vapors can also be a very effective four - wave mixing medium @xcite and a source of spatially multimode squeezing @xcite with a potential for quantum imaging @xcite . in this paper
we characterize a multimode generator of temporally separated random image pairs with a raman interface .
the images are formed at the write - in and the read - out of atomic excitations , accomplished by driving a spontaneous stokes and a stimulated anti - stokes scattering respectively . at the write stage
we create a random scattered stokes field , which forms a speckle pattern in the far field .
it results from interference of distinct spatial modes . at the same time coupled distribution of excitations arises within the atomic sample @xcite .
this can be thought as a storage of the random but known image in the atomic memory . after a certain time
the excitations can be converted to the anti - stokes scattered light . in the experiments we use rubidium-87 vapor mixed with a buffer gas .
we measure the effects of diffusion and deterioration of the effective optical depth due to the collisional broadening .
we quantify the spatial multimode characteristics and a capacity of the generator . in general this task
is accomplished either by measuring field coherence @xcite or by measuring second order hanbury - brown twiss type correlations @xcite .
we have chosen to measure the latter , i.e. statistical correlation between intensity fluctuations in various scattering directions from large sets of single - shot images similarly as in @xcite .
we show that the results are consistent with a simple theoretical model @xcite . in this particular operating regime
we produce multiple correlated modes of radiation that are released at separate time instants , first in the stokes scattering process , second in the anti stokes scattering @xcite .
this is similar to recent work demonstrating entangled images @xcite , however in our generator the second image can be held in memory until the first image is processed or a suitable moment comes .
in contrast to previous work @xcite it is achieved without resorting to additional , separate quantum memory implemented in another atomic ensemble .
in this article we examine purely classical properties of the correlated images .
we focus on spatial properties of light and understanding atomic decoherence .
however , we note that our source shares a number of features with sources of non - classical light based on atomic ensembles @xcite .
our main motivation for the present study was that of possible future applications in multimode quantum information processing @xcite , particularly involving spatially multimode squeezing which has attracted large attention recently @xcite .
we also suppose that our results might be of significance for quantum imaging @xcite , especially in applications which require storage of ghost images @xcite .
f=2 transition on d2 line , while write and read pulse are @xmath1ghz detuned from f=1@xmath2f=1 and f=2@xmath2f=2 transitions on d1 line respectively .
( c ) right panel depicts the relation between the wave vectors of pumps @xmath3 , @xmath4 , photons @xmath5 , @xmath6 and a spin wave @xmath7 involved in the cycle . ] the operating cycle of the multimode delayed images generator and the optical transitions are illustrated in fig .
[ fig : general ] . the cylindrical cell containing rubidium-87 vapor in a buffer gas
is initialized by optical pumping of the atoms into f=1 states of the ground manifold .
next , a pencil - shaped region is illuminated with a strong write beam .
this induces scattering of anti - stokes photons and creation of a spin wave corresponding to coherent excitation of atoms into f=2 state .
the scattering grows exponentially due to the boson nature of the fields involved .
we stop it when tens of thousands of excitations have been produced .
after that we wait for an adjustable amount of time and illuminate the atomic sample with the read beam .
it is tuned in order to cause scattering of stokes photons at the expense of the atomic excitations .
it can be thought that the read beam probes the atomic coherence by converting it to the light . in the write process
the difference of wave vectors between the annihilated write laser photon and the created stokes one @xmath8 is mapped onto the spatial phase of an atomic excitation .
the excitation takes on a form of a coherent spin wave and acquires a wave vector @xmath9 . at the readout stage
the stored phase given by the wave vector @xmath7 adds to the read beam field wave vector @xmath4 leading to scattering of anti - stokes photons @xmath6 in the direction opposite to the stokes ones , as illustrated in fig .
[ fig : general]c . with @xmath10-axis along the write beam and @xmath3-vector and the scattering in the @xmath11@xmath10 plane the spin wave wave vector
@xmath7 equals : @xmath12 \simeq 2\pi [ \theta/(0.8\,\text{mm}\,\text{mrad}),0,1/(44\,\text{mm } ) ] \ ] ] where @xmath13 is the scattering angle and the numbers correspond to 6.8 ghz stokes shift ( f=1@xmath2f=2 scattering ) at d1 line .
suppose the active part of the atomic sample has a pencil shape of a diameter @xmath14 , set by the write and the read beam , and the length @xmath15 along @xmath10-axis . within this volume atomic excitations which wave vectors @xmath7 and
@xmath16 differ more than the inverse dimensions of the active region : @xmath17 $ ] , are distinguishable and independent .
this sets the fundamental limit on the number of independent modes which may be simultaneously stored in the atomic memory . since the scattered light couples to the respective spin wave @xmath7-vectors the distinguishability of various @xmath7 components of the atomic excitation also imprints on the scattered light .
conversely , this implies the emission into close directions has to be correlated i.e. the light in the far field exhibits a finite coherence area .
in addition the stokes scattering process populates only certain superpositions of the distinguishable spatial modes .
one of the easiest way to analyze it is to decompose the atomic excitations and likewise scattered light into a pairwise coupled basis , in which each light mode is coupled with exactly one atomic mode .
the obtained light modes are all centered around the write beam @xmath3 while the spin wave modes around the origin of the @xmath7-vector space .
the modes resemble laguerre - gaussian functions @xcite .
importantly , the spatial shape of the modes is time independent .
the width of the fundamental spin wave mode is approximately @xmath18 .
the associated fundamental light mode is a gaussian beam with a confocal parameter close to the sample length @xmath15 .
the raman growth parameters of higher order modes slowly diminish along with mode number .
typically the number of significantly excited modes is close to the fresnel number of the write beam @xmath19 if the decoherence is neglected @xcite . during each single write
pulse stokes modes are populated .
they produce a speckle - like diffraction pattern in the far field .
note that each speckle is not a single mode but an interference of many modes excited with random amplitudes and phases .
thus , every speckle has a different position , size and amplitude .
yet , an approximate average size of the speckle can still be worked out from the mode picture . given
@xmath20 similarly populated laguerre - gaussian modes we typically find single spots in their interference pattern to occupy about @xmath21 of the solid angle of the fundamental mode , because the central lobe of the @xmath22-th mode is @xmath23 times thinner than the extent of the fundamental mode .
this is also equal to the coherence area of the image .
conversely , the solid angle of the entire pattern is @xmath20 times bigger than the solid angle of a single coherence area assuming @xmath20 modes are populated .
this can be justified twofold .
firstly @xmath20 modes interfere to @xmath20 independent coherence areas with uncorrelated intensities .
secondly the size of the pattern is determined by the highest order mode , which has @xmath24 times bigger solid angle than the fundamental mode which in turn is @xmath24 times bigger than the coherence area .
note , that the above are approximate statements as the population of consecutive modes is diminishing slowly and the threshold above which the modes are counted as populated is vague .
therefore the number @xmath20 is approximate . in the experiment ,
the number of excited modes is limited by the diffusion .
this is because high order modes are composed of components with large @xmath7 which correspond to abrupt spatial variations .
they are easily washed off by diffusion of the atoms in a buffer gas during the storage .
therefore in the experiment we expect the number of retrieved modes @xmath20 to be lower than the fresnel number of the write beam @xmath25 .
in our generator we observe an image created by stokes photons and anti - stokes photons in the far field , @xmath26 and @xmath27 respectively . in the ideal case of a plane - wave write and
read beam we would expect that every point in @xmath28 should carry information about the spin wave with @xmath29 . in case of a finite write and read beam widths
@xmath14 we expect the ideal images to be blurred due to beams wave vector spread .
the true images should be close to the convolution of atomic excitations distribution in the sample with the read beam distribution in @xmath30-space .
rb vapor was illuminated by the write , the read and the pump beam .
the far field of the raman scattering was observed on the electron multiplier ccd ( em ccd ) camera and the background of the scattering beams was supressed by polarization and the filtering ( wollaston polarizer and the @xmath31rb cell ) .
interference filter if filtered the spontaneous emission background during the pumping .
the pulse sequence was monitored using photodiodes pd . ] the experimental setup is schematically presented in fig . [ fig : setup ] .
the pump laser ( distributed feedback diode laser , toptica ) was resonant with the @xmath32 transition on the d2 line to pump the @xmath33rb atoms to @xmath34 ground level .
both the write and the read lasers were external cavity diode lasers from toptica .
they were detuned from the @xmath35 transition by 1 ghz to the red and by 1 ghz from the @xmath32 transition to the blue respectively .
both were additionally cleaned up spectrally as well as spatially by fabry - perrot cavities with fsr=13.6 ghz . the pulse sequence shown in fig .
[ fig : general ] was produced by acousto - optical modulators .
the pump pulse duration 780 @xmath36s and its 4.2 mm full width at @xmath37 waist diameter ensured bleaching of the medium in the volume were the raman scattering took place .
the write pulse started immediately after the pump pulse .
the @xmath37 beam diameter was 2.16 mm .
pulse duration varied from 200 ns to 10 @xmath36s and the power up to 17 mw .
after the write pulse we inserted a variable storage delay . finally a read pulse was directed onto the cell and converted the spin waves into anti - stokes photons .
the @xmath37 beam diameter was 1.76 mm .
the read beam was tilted by 13 mrad with respect to the write beam in order to direct the scattered photons onto separate regions on the camera .
all beams crossed inside a cylindrical quartz cell ( precision glassblowing ) , 100 mm long and 25 mm in diameter , which contained isotopically pure @xmath33rb .
the cell was mounted inside a two - layer magnetic shielding and was heated with bifilar wound copper coils .
for comparison we used three cells containing different buffer gas , 1 torr or 0.5 torr krypton or 5 torr neon .
the scattered light was separated from the write and read beams on a wollaston polarizer .
the residual laser background was then absorbed in the hot @xmath31rb cell .
we decided to use this method due to its simplicity but one may alternatively use fabry - perrot interferometers @xcite .
an interference filter was used to dump the 780 nm fluorescence emitted during the initial pumping .
we checked that the laser background was of the order of a few photons per sequence and it was negligible compared to the residual fluorescence background and the camera noise .
we focused the far field of the raman light onto an electron multiplier ccd camera ( imageem , hamamatsu ) .
we calibrated the position on the camera with respect to the beam propagation angle inside the cell .
the pixel pitch corresponded to 76 @xmath36rad angle inside the cell .
we also calibrated sensitivity with a power meter .
since the overlapping write and read beams were tilted with respect to each another , the corresponding stokes and the anti - stokes scattering light fell on distinct regions of the sensor .
the camera was incapable of resolving the light temporally and we always captured both the stokes and the anti - stokes scattering integrated over the duration of write and read pulse . for each setting of the pulse lengths , powers and the storage time we captured 500 or 10000 separate images of the scattered light with a cycle rate of about 90
hz . despite the usage of an interference filter before the camera we had some background coming from spontaneous emission in the pumping process
. thus the reference set of 100 frames with pump and write beams was saved for background subtraction .
we find that 500 frames are sufficient to calculate the average intensity of the scattered stokes and anti - stokes light @xmath38 and @xmath39 respectively .
the average intensities presented below were calculated subtracting the background .
the main origin of the background was the stray fluorescence of the atoms emitted during pumping .
it was sufficiently stable to calculate it with high precision from 100 frames .
we find that the average intensity of the scattered light was symmetric around the driving beam .
therefore we averaged it with respect to the polar angle to obtain @xmath40 and @xmath41 with better signal to noise ratio .
for calculating second order moments , including intensity correlations , we collected data sets containing 10000 frames .
atom density in the main @xmath33rb cell was estimated by measuring the absorption spectrum for the write laser attenuated to a fraction of saturation intensity right before the cell .
the result was fitted with a theoretical model to estimate the gas temperature and absorption at resonance .
cell heater temperature , measured by the resistance of the windings , was adjusted so as to obtain a doppler - broadened optical depth of 130 .
the stability of the setup was verified by repeating this measurement prior to and right after the proper experimental sequence .
note that the doppler broadened optical thickness is not sensitive to the pressure broadening of the order of few natural linewidths .
it is deemed that using the above method we obtain almost the same rubidium vapor pressure in all of the cells .
[ [ decay - of - the - spin - waves ] ] decay of the spin waves + + + + + + + + + + + + + + + + + + + + + + + firstly we measured the decay rate of the spin waves vs. magnitude of their wave vector @xmath42 . it can be extracted from measurements of the average intensity of anti - stokes @xmath39 as a function of the storage time for constant write pulse parameters . to produce a widespread , significant excitation of spin waves in the cell we used short write pulse with a maximum available power of 17.5 mw .
the pulse duration was adjusted to produce comparable spin wave excitations for each buffer gas resulting in @xmath43ns for 0.5 torr krypton , @xmath44ns for 1 torr krypton and @xmath45s for 5 torr neon . after a variable storage time
@xmath46 a @xmath47s read pulse is sent down the cell to convert remaining atomic excitation to anti - stokes light .
the average intensity of anti - stokes @xmath39 is a gaussian function of angles @xmath48 .
with increasing the storage time @xmath46 the height and width of this gaussian decreases .
each direction of observation @xmath49 with respect to the read beam corresponds to a certain transverse anti - stokes photon wave vector @xmath50 .
it is approximately equal to the coupled spin wave wave vector @xmath7 to within the uncertainty given by the read beam wave vector spread corresponding to the angular spread of @xmath51mrad .
therefore the evolution of the anti - stokes intensity @xmath27 as a function of storage time @xmath46 provides indication of the spin wave decay for the best coupled spin wave wave vector . in fig .
[ fig : decays](a ) we present typical decays along with fitted exponentials . in fig .
[ fig : decays](b ) we gathered decay rates as a function of angle between the read beam and scattered photons .
all the errorbars plotted on the graphs correspond to the 1@xmath52 uncertainty calculated from the variance of the experimental data .
as a function of the storage time @xmath46 together with exponential fits @xmath53 .
( b ) the decay rates @xmath54 as a function of the scattering angle @xmath13 with a quadratic fit @xmath55 .
results are presented for three different buffer gases.,scaledwidth=100.0% ] similar to @xcite we attribute the decay of the spin wave to the diffusion of the rubidium atoms in the buffer gas .
the evolution of the atomic excitation is given by the diffusion equation , which leads to damping of spin waves .
in particular the spin wave with wave vector @xmath56 decays with rate proportional to square of its wave vector @xmath57 , with @xmath58 being the diffusion coefficient .
the parabolic fits of fig .
[ fig : decays](b ) give the diffusion rates equal 397(27 ) @xmath59/s , 257(4 ) @xmath59/s , 146(5 ) @xmath59/s for 5 torr neon , 0.5 torr krypton and 1 torr krypton respectively .
the values measured for krypton are consistent with collisions cross sections @xcite . however , the value for neon is higher than recently reported , by a factor of 4 @xcite or 40 @xcite .
we have no good explanation for this observation , however owing to large discrepancies between reported values we display the data as a reference for possible future studies .
let us note that alternatively the effect of diffusion can be described as dicke - narrowed two - photon resonance , as explained in the context of eit in @xcite .
[ [ spontaneous - scattering - and - growth - of - the - spin - wave ] ] spontaneous scattering and growth of the spin wave + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + knowing the intrinsic decay of the spin wave in the dark , we are in a position to study its growth during the write stage . within the experimental precision the growth @xmath26 as a function of write pulse length @xmath60 can be approximated as a simple exponential , as shown for typical situations in fig .
[ fig : growth](a ) .
the duration of the write pulse is limited so that all saturation effects can be neglected . as a function of the write pulse duration @xmath60 for the scattering angle @xmath61 mrad with the exponential fit @xmath62const .
the data taken for three different buffer gasses with similar atomic density .
( b ) growth rate @xmath63 as a function of the scattering angle ( points with errorbars ) .
the lines show the expected damping due to diffusion.,scaledwidth=100.0% ] in the limit of plane wave driving beam , when each spin wave evolves separately , we would expect the growth rate @xmath63 to be a sum of three components @xmath64 .
@xmath65 denotes the coherent scattering rate which is directly proportional to the intensity of the pumping light @xmath66 and atomic density .
it drops for large scattering angles due to a reduced beam overlap between light fields involved .
@xmath67 denotes the diffusion - induced spin wave decay .
finally , @xmath68 denotes the rate of the single atom spontaneous emission which depends linearly on the light power @xmath66 and has a contribution from the collisions with the buffer gas pressure broadening of the excited state . as a single atom effect @xmath68
should be independent from spin wave periodicity i.e. @xmath69 . in fig .
[ fig : growth](b ) we present measured growth rate @xmath63 as a function of angle @xmath13 between the write beam and stokes photon .
we also mark the predicted contribution of the diffusion - induced spin wave decay @xmath67 inferred from the previous measurement . as apparent in the figure , measured growth rate for high angles is faster than expected from the plane wave model .
this is explained by the fact , that the fundamental mode of atomic excitation is not a plane wave .
it has a large wave vector spread , approximately @xmath70 which corresponds to an angular spread of 2.3 mrad .
fast coherent uprise of the fundamental mode boosts the growth rate at high angles .
we also note significant overall difference in scattering rate between various buffer gases despite comparable density of the atomic sample and write laser power .
it can be attributed to the perturbation of the excited state by the collisions , studied in the case of neon @xcite .
the growth is slower at a higher pressure due to higher @xmath68 which results from the loss of phase coherence of the excited state . as seen in fig .
[ fig : decays](b ) and fig .
[ fig : growth](b ) for 0.5 an 1 torr krypton one can trade slower diffusion for faster growth .
[ [ intensity - correlations ] ] intensity correlations + + + + + + + + + + + + + + + + + + + + + + . the anti - stokes scattering is cropped for higher angles due to decoherence . ]
let us now turn our attention to the multimode spatial nature of scattered fields . a typical single shot image captured by the camera is presented in fig .
[ fig : single ] .
it was taken with the 1 torr krypton as a buffer gas .
the write and read pulse power and duration were 17.5 mw , 800 ns and 5 mw , 2 @xmath36s respectively .
the frame shows the speckle pattern of both the stokes and the anti - stokes scattering with the highest intensity around the driving beam directions .
the pattern is produced due to the boson amplification of initially random fields in many spatial modes , just as it happens in spontaneous parametric down - conversion @xcite or in bose - einsten condensate collisions @xcite .
looking carefully one can see that the anti - stokes picture is similar to the stokes rotated by @xmath71 around the center , although features at high angles are cropped . upon averaging 10000 frames
the speckle pattern disappears giving way to smooth , symmetric gaussian intensity distributions as illustrated in fig .
[ fig : avg ] .
this is because the spot position in each single shot is random .
again , the angular spread of the anti - stokes is significantly smaller .
this is a consequence of the diffusion in the buffer gas .
the spin waves with high wave vector @xmath69 are washed off before the readout .
apparently the mean values in fig .
[ fig : avg ] do not carry any information about single - shot speckle pattern , coherence area or the intensity correlations of the stokes and anti - stokes . to extract and quantify interesting single - shot properties we calculate statistical correlation of intensity fluctuations between portions of the scattered light @xmath72 .
it is defined using deviations of the actual photon number from the mean value @xmath73 , @xmath74 as follows : @xmath75 we depict @xmath76 with fixed reference direction @xmath77 in the stokes as a function of @xmath48 for both the stokes
stokes @xmath78 and anti - stokes
stokes @xmath79 type of correlations .
a typical correlations pictures obtained this way are shown in fig .
[ fig : corr ] .
the spot in the stokes self - correlation image @xmath78 ( left panel ) represents the correlation between stokes light emitted in similar directions , while the highest spot in the anti - stokes cross - correlation @xmath79 ( right panel in the figure ) corresponds to a counter part correlated anti - stokes light .
the strength of the counter part beam correlations diminishes as we increase the distance from the driving beam , thus looking at larger @xmath69 .
correlation pictures for other reference directions @xmath77 are very similar , the conjugate spot always appears on the opposite side of the center with respect to the reference . note the size of the correlation spot is proportional to the size of a speckle in the single shot image fig .
[ fig : single ] and it is larger by factor @xmath80 .
this is explained by the fact that cross correlation value of each point is the sum of cross correlations of all the independent , displaced speckles .
therefore the actual cross - correlation function will be a convolution of the speckle shape with itself which for the gaussian functions give the factor @xmath80 .
let us note that in the right , anti - stokes stokes correlation panel fig .
[ fig : corr ] one can also see a second , weaker correlation spot .
it lies on the same side of the read beam as the reference point .
this is a signature of further stokes scattering induced by the read beam .
it is possible because the read beam is detuned by approximately 5 ghz from the f=1@xmath2f=1 transition , which is apparently not enough to completely extinguish further amplification .
( left panel ) and @xmath81 ( right panel ) between a fixed direction in the stokes @xmath77 and any other direction @xmath48 in the stokes and anti - stokes respectively . the reference direction @xmath82 mrad corresponds to the point in the center of the spot in the stokes self - correlation function @xmath78 in the left panel .
the center of the best correlated spot corresponds to the conjugate direction of the counterpart beam @xmath83 .
contours were drawn at levels of 0.1 , 0.2 up to 0.9 .
results obtained in a cell with 1 torr kr buffer gas . ] fig .
[ fig : ccs ] depicts cross sections through the correlation functions . in the first two panels fig .
[ fig : ccs](a ) and fig .
[ fig : ccs](b ) we compare the anti - stokes stokes correlation functions @xmath79 for the reference direction along the write beam @xmath84 and at an angle of @xmath85 mrad to it .
initial width of the correlation function with respect to @xmath86 equals 1 mrad regardless of the reference direction @xmath87 .
it shrinks first slowly and then quite abruptly for long storage times due to diffusional damping of high-@xmath56 spin waves . in fig .
[ fig : ccs](c ) we can see correlation of the stokes field with itself @xmath78 .
it also has 1 mrad width regardless of the reference direction .
along @xmath86 .
cross - correlation @xmath79 for ( a ) @xmath88 and ( b ) @xmath89 mrad , @xmath90 calculated for the subsequent storage times @xmath910 , 3 , 5 , 8 @xmath36s .
( c ) self - correlation @xmath78 for @xmath92 mrad , @xmath90 .
points depict experimental data , lines correspond to gaussian fits.,scaledwidth=100.0% ] the simple theory predicts that the stokes field is composed of spatial modes @xcite .
the number of modes @xmath20 can be approximated as a number of speckle spots that fit into the solid angle occupied by the scattered light .
we take @xmath93 , where @xmath94 and @xmath95 are @xmath37 radius of the average intensity of the scattered light and correlation function respectively .
the numerical factor arises from the fact that the correlation function is square in the intensity .
its value is confirmed by considering a simple model , in which the registered intensity is a sum of randomly populated modes of a width @xmath96 .
it is found that the width of the correlation function is @xmath97 .
first we calculate the number of modes @xmath20 for the stokes field . from the gaussian
fits we find the angular radius of the average intensity @xmath98 mrad as depicted in fig .
[ fig : avg ] and stokes - stokes correlation function @xmath99 mrad , giving a total number of modes @xmath100 .
these number allow us to asses the fundamental mode angular spread @xmath101{n}=w_c\sqrt[4]{n/4}=1.4 $ ] mrad which is of the order predicted by a simple theory @xmath102 corresponding to 2.3 mrad . in the lossless case
the number of modes @xmath20 is expected to be the fresnel number of the write beam @xcite , @xmath103 in our case , which is consistent with the experimental results .
let us note that in experiments by boyer _
@xcite they estimated the number of modes to be equal 100 , which is also consistent with the fresnel number calculated for their experimental conditions @xmath104 .
therefore we conclude that the relation between number of modes and fresnel number of the excited portion of the sample can be used for designing future experiments .
in a similar manner we can calculate the number of retrieved modes as a function of time . for each storage time
we calculate the average width of the anti - stokes scattering and the correlation function .
the result of estimated number of retrieved spatial modes is plotted in fig .
[ fig : modes ] . as a function of storage time . ]
we can see that only for 1 torr krypton the number of retrieved modes in significantly greater than one , but still much lower than the number of stored modes we calculated .
we suspect that this sudden drop results from a doppler broadening in the atomic sample .
the spin waves are imprinted on different velocity classes with varying phases @xcite that are inconsistent between write and read stages .
in this article we characterize the performance of a multimode delayed random image generator in which correlations are mediated between the beams via a spin wave excitation in the warm rubidium vapor .
we turned our attention to the experimental characterization of the spatial behavior of the collective raman scattering .
in particular we quantified the influence of buffer gas pressure on the diffusion and pair production rate in the stokes scattering .
we measured the exponential decays of the spin waves to characterize the decoherence from atomic diffusion and we determined the diffusion constants for three different buffer gases .
we compared these results to the angular dependence of the stokes scattering growth parameters and found that it is dominated by the evolution of the fundamental most occupied mode .
we calculated the number of independent spatial modes excited in the generator .
we showed how many can be retrieved after a certain storage time .
our results concerning the mode structure of scattered fields are consistent with a simple multimode model which permits space - time separation @xcite . in future applications one may attempt to increase the number of modes .
our results suggest two ways to achieve that .
one can apply the larger driving beam diameter .
the number of modes , determined by the fresnel number , is expected to grow quadratically with a driving beam diameter if constant laser intensity is applied .
an increase of a beam diameter would also shorten the wave vectors of the excited spin waves , making them more resilient to diffusion . the other way to increase
the number of modes is to slow down the atomic diffusion keeping a collisional decoherence low .
our results indicate a need for further comparisons of various buffer gases to find an optimum . this type of experiment could be also implemented in cold atomic gases and possibly in doped crystals . in this work we studied only classical properties of the light generated .
however , based on a number of similar successful experiments @xcite we conjecture the conclusions on the spatial properties of the light would stay valid also in the low light quantum regime .
our results could therefore be of use for a number of emerging multimode quantum technologies .
in particular the quantum information processing protocols devised for warm atomic samples could benefit from multimode capacity of the storage , because a single cell operated in multimode regime is equivalent to multiple single mode cells .
we acknowledge fruitful discussions with thomas fernholz and czesaw radzewicz as well as generous support from konrad banaszek .
this work was supported by the foundation for polish science team project co - financed by the eu european regional development fund and fp7 fet project q - essence ( contract no .
was also supported by the national science centre grant no .
dec-2011/03/d / st2/01941 . | we apply collective raman scattering to create , store and retrieve spatially multimode light in warm rubidium-87 vapors .
the light is created in a spontaneous stokes scattering process .
this is accompanied by the creation of counterpart collective excitations in the atomic ensemble the spin waves . after a certain storage time
we coherently convert the spin waves into the light in deterministic anti - stokes scattering . the whole process can be regarded as a delayed four - wave mixing which produces pairs of correlated , delayed random images .
storage of higher order spatial modes up to microseconds is possible owing to usage of a buffer gas .
we study the performance of the raman scattering , storage and retrieval of collective excitations focusing on spatial effects and the influence of decoherence caused by diffusion of rubidium atoms in different buffer gases .
we quantify the number of modes created and retrieved by analyzing statistical correlations of intensity fluctuations between portions of the light scattered in the far field .
99 l .- m .
duan , m. d. lukin , j. i. cirac , and p. zoller , `` long - distance quantum communication with atomic ensembles and linear optics , '' nature * 414 * , 413418 ( 2001 ) .
j. appel , e. figueroa , d. korystov , m. lobino , and a. lvovsky , `` quantum memory for squeezed light , '' phys .
rev .
lett .
* 100 * , 093602 ( 2008 ) .
k. honda , d. akamatsu , m. arikawa , y. yokoi , k. akiba , s. nagatsuka , t. tanimura , a. furusawa , and m. kozuma , `` storage and retrieval of a squeezed vacuum , '' phys .
rev .
lett .
* 100 * , 093601 ( 2008 ) .
k. hammerer , `` quantum interface between light and atomic ensembles , '' rev .
mod .
phys . * 82 * , 10411093 ( 2010 ) .
k. jensen , w. wasilewski , h. krauter , t. fernholz , b. m. nielsen , m. owari , m. b. plenio , a. serafini , m. m. wolf , and e. s. polzik , `` quantum memory for entangled continuous - variable states , '' nature phys . * 7 * , 1316 ( 2010 ) .
j. nunn , k. reim , k. c. lee , v. o. lorenz , b. j. sussman , i. a. walmsley , and d. jaksch , `` multimode memories in atomic ensembles , '' phys .
rev .
lett .
* 101 * , 260502 ( 2008 ) .
d. v. vasilyev , i. v. sokolov , and e. s. polzik , `` quantum volume hologram , '' phys .
rev .
a * 81 * , 020302 ( 2010 ) .
q. glorieux , j. b. clark , a. m. marino , z. zhou , and p. d. lett , `` temporally multiplexed storage of images in a gradient echo memory , '' opt .
express * 20 * , 1235012358 ( 2012 ) .
d. higginbottom , b. sparkes , m. rancic , o. pinel , m. hosseini , p. lam , and b. buchler , `` spatial - mode storage in a gradient - echo memory , '' phys .
rev .
a * 86 * , 023801 ( 2012 ) .
v. boyer , a. m. marino , r. c. pooser , and p. d. lett , `` entangled images from four - wave mixing . ''
science * 321 * , 544547 ( 2008 ) .
v. boyer , a. marino , and p. lett , `` generation of spatially broadband twin beams for quantum imaging , '' phys .
rev .
lett . * 100 * , 143601 ( 2008 ) .
g. brida , m. genovese , and i. ruo berchera , `` experimental realization of sub - shot - noise quantum imaging , '' nature photon .
* 4 * , 227230 ( 2010 ) .
m. g. raymer and j. mostowski , `` stimulated raman scattering : unified treatment of spontaneous initiation and spatial propagation , '' phys .
rev .
a * 24 * , 19801993 ( 1981 ) .
j. koodyski , j. chwedeczuk , and w. wasilewski , `` eigenmode description of raman scattering in atomic vapors in the presence of decoherence , '' phys .
rev .
a * 86 * , 013818 ( 2012 ) .
p. h. s. ribeiro , c. h. monken , and g. a. barbosa , `` measurement of coherence area in parametric downconversion luminescence , '' appl . opt . * 33 * , 352355 ( 1994 ) .
a. marino , v. boyer , r. pooser , p. lett , k. lemons , and k. jones , `` delocalized correlations in twin light beams with orbital angular momentum , '' phys .
rev . lett .
* 101 * , 093602 ( 2008 ) .
f. ferri , d. magatti , a. gatti , m. bache , e. brambilla , and l. lugiato , `` high - resolution ghost image and ghost diffraction experiments with thermal light , '' phys .
rev .
lett . * 94 * , 183602 ( 2005 ) .
a. m. marino , r. c. pooser , v. boyer , and p. d. lett , `` tunable delay of einstein - podolsky - rosen entanglement , '' nature * 457 * , 859862 ( 2009 ) .
c. chou , s. polyakov , a. kuzmich , and h. kimble , `` single - photon generation from stored excitation in an atomic ensemble , '' phys .
rev .
lett . *
92 * , 14 ( 2004 ) .
m. lassen , v. delaubert , j. janousek , k. wagner , h .- a .
bachor , p. lam , n. treps , p. buchhave , c. fabre , and c. harb , `` tools for multimode quantum information : modulation , detection , and spatial quantum correlations , '' phys .
rev .
lett . * 98 * , 083602 ( 2007 ) .
k. wagner , j. janousek , v. delaubert , h. zou , c. harb , n. treps , j. f. morizur , p. k. lam , and h. a. bachor , `` entangling the spatial properties of laser beams . ''
science * 321 * , 541543 ( 2008 ) .
j. janousek , k. wagner , j .- f .
morizur , n. treps , p. k. lam , c. c. harb , and h .- a .
bachor , `` optical entanglement of co - propagating modes , '' nature photon . * 3 * , 399402 ( 2009 ) .
b. chalopin , f. scazza , c. fabre , and n. treps , `` direct generation of a multi - transverse mode non - classical state of light , '' opt .
express * 19 * , 44054410 ( 2011 ) .
e. brambilla , l. caspani , o. jedrkiewicz , l. lugiato , and a. gatti , `` high - sensitivity imaging with multi - mode twin beams , '' phys .
rev .
a * 77 * , 053807 ( 2008 ) .
y .- w .
cho , j .- e .
oh ,
and y .- h .
kim , `` storage and retrieval of ghost images in hot atomic vapor , '' opt . express * 20 * , 58095816 ( 2012 ) .
s. manz , t. fernholz , j. schmiedmayer , and j .- w .
pan , `` collisional decoherence during writing and reading quantum states , '' phys .
rev .
a * 75 * , 040101 ( 2007 ) .
k. gibble and a. gallagher , `` measurements of velocity - changing collision kernels , '' phys .
rev .
a * 43 * , 13661380 ( 1991 ) .
o. firstenberg , p. london , m. shuker , a. ron , and n. davidson , `` elimination , reversal and directional bias of optical diffraction , '' nature phys . * 5 * , 665668 ( 2009 ) .
m. shuker , o. firstenberg , r. pugatch , a. ben - kish , a. ron , and n. davidson , `` angular dependence of dicke - narrowed electromagnetically induced transparency resonances , '' phys .
rev .
a * 76 * , 023813 ( 2007 ) .
o. firstenberg , m. shuker , a. ron , and n. davidson , `` coherent diffusion of polaritons in atomic media , '' pre - print arxiv:1207.6748 ( 2012 ) .
b. m. jost , a. v. sergienko , a. f. abouraddy , b. e. a. saleh , and m. c. teich , `` spatial correlations of spontaneously down - converted photon pairs detected with a single - photon - sensitive ccd camera , '' opt . express * 3 * , 8188 ( 1998 ) .
a. perrin , h. chang , v. krachmalnicoff , m. schellekens , d. boiron , a. aspect , and c. i. westbrook , `` observation of atom pairs in spontaneous four - wave mixing of two colliding bose - einstein condensates , '' phys .
. lett .
* 99 * , 150405 ( 2007 ) .
a. gorshkov , a. andr , m. lukin , and a. srensen , `` photon storage in @xmath0-type optically dense atomic media .
iii .
effects of inhomogeneous broadening , '' phys .
rev .
a * 76 * , 033806 ( 2007 ) . |
the double - exchange ( de ) model is a minimal model which explicitly incorporates the interplay between itinerant electrons and localized magnetic moments .
the hamiltonian is given by @xmath0 where @xmath1 ( @xmath2 ) is an annihilation ( creation ) operator of an electron with spin @xmath3 at site @xmath4 , @xmath5 is the transfer integral for nearest - neighbor sites @xmath6 , @xmath7 and @xmath8 are spin operators of itinerant electron and localized spin , respectively , which are coupled by the hunds - rule coupling @xmath9 , @xmath10 is the density operator , and @xmath11 is the chemical potential .
the localized spins @xmath8 are treated as classical vectors .
the model was originally introduced by zener @xcite , and has been studied for understanding the physics of perovskite manganese oxides @xcite . in particular ,
the rediscovery of colossal magneto - resistance phenomena have stimulated extensive studies for the de systems , including some extensions of the model such as the super - exchange interaction between localized moments , orbital degeneracy of itinerant electrons , and electron - phonon couplings @xcite .
the model ( [ eq : h ] ) is known to exhibit a ferromagnetic metallic state at low temperature ( @xmath12 ) to gain the kinetic energy of electrons by aligning localized moments in parallel .
this is called the de mechanism @xcite .
there have been many efforts to estimate the ferromagnetic transition temperature @xmath13 @xcite . for the model on the three - dimensional cubic lattice ,
@xmath13 was determined precisely by a large - scale monte carlo simulation : for example , @xmath14 at the electron density @xmath15 in the limit of @xmath16 @xcite . in this contribution
, we present our numerical results for the ferromagnetic transition when the model is defined on the geometrically frustrated lattice structure . among many frustrated lattice structures
, we consider the pyrochlore lattice , which is a three - dimensional network of corner - sharing tetrahedra as shown in fig .
[ fig1](a ) .
one of the experimental motivations is found in a family of mo pyrochlore oxides _
r_@xmath17mo@xmath18o@xmath19 : when the ionic size of rare earth element _ r _ is relatively large such as _ r _ = nd and sm , the compounds become ferromagnetic metal at low @xmath12 @xcite , and it was pointed out by the first principle calculations that the de mechanism plays a key role for this behavior @xcite . as a first step toward the understanding of thermodynamic properties of the pyrochlore systems , below we will investigate the ferromagnetic transition by monte carlo calculations for the simplest case , i.e. , the model ( [ eq : h ] ) in the limit of @xmath16 .
we employ a monte carlo method to take account of large fluctuations in the frustrated system .
the method is a standard one in which configurations of classical localized spins are sampled by monte carlo procedure ; the monte carlo weight is calculated by the exact diagonalization of the fermion hamiltonian matrix for a given spin configuration .
the bottleneck of the calculations is the exact diagonalization , which usually limits the accessible system sizes to several hundreds sites .
an improved method based on the polynomial expansion of the density of states was developed by the authors @xcite , but for frustrated systems , in general , the method becomes less efficient because of singular form of the density of states : indeed , in the present pyrochlore case , the density of states exhibits a @xmath20-functional peak due to two flat bands ( @xmath21 ) as well as two van - hove singularities ( @xmath22 and @xmath23 ) in the non - interacting model as shown in fig . [ fig1](b ) .
therefore we here employ the standard method . to enable systematic analysis within the limited system sizes
, we apply a technique of averaging over the twisted boundary conditions @xcite . in this technique ,
a twisted boundary condition is imposed with replacing the transfer integral @xmath5 by @xmath24 , where @xmath25 denotes a magnetic flux and @xmath26 represents the vector connecting the nearest - neighbor sites @xmath6 .
the average is taken by the integral over @xmath27 , which is approximately calculated by the sum over @xmath28 grid points .
it has been shown that the procedure reduces finite size effects originating from the discreteness of the wave numbers .
it is shown for the model ( [ eq : h ] ) that in the limit of @xmath29 , the averaging procedure for a state of a finite size cluster with a given set of spin configurations provides results for the infinite size system which consists of a periodic array of the finite size cluster .
for example , for the perfectly ordered ferromagnetic state , the averaging procedure with @xmath29 for any finite size cluster gives the exact result in the thermodynamic limit . in order to take this advantage
, we apply the averaging procedure to each monte carlo snapshot for calculating the monte carlo weight .
this corresponds to ensemble average over the independent systems with different boundary conditions . in the following calculations ,
we take the average over the grid points with @xmath30 , where @xmath31 and @xmath32 ( @xmath33 ) . here , @xmath34 and @xmath35 is a linear dimension of the system measured in the cubic unit cell , i.e. , the total number of sites @xmath36 .
we demonstrate here the efficiency of the averaging technique by calculating the electron density @xmath37 as a function of @xmath12 .
we show the monte carlo results at @xmath38 in figs .
[ fig2 ] and [ fig3 ] as an example . in fig .
[ fig2 ] , we present @xmath28 dependence for two different system sizes . in both cases ,
the results converge onto a single curve as increasing @xmath28 .
furthermore , the necessary @xmath28 for the convergence becomes smaller as the system size @xmath39 increases .
figure [ fig3 ] shows @xmath39 dependence with and without taking averages . in the case without taking averages , the results are largely scattered for different system sizes as shown in fig .
[ fig3](a ) .
in contrast , as in fig .
[ fig3](b ) , when we take averages over a sufficient number of @xmath28 chosen for each system size , all the results converge onto a single curve and the finite size effects are sufficiently small , even for the rectangular - shaped clusters .
we have checked the efficiency in wide parameter regions of @xmath11 and @xmath12 and confirmed that the averaging procedure is efficient enough to suppress the finite size effects coming from the discreteness of the wave numbers .
we adopt @xmath28 used in fig .
[ fig3](b ) for each @xmath39 in the following calculations . dependence of the electron density @xmath37 at @xmath38 for different numbers of grid points @xmath28 .
( a ) the system size @xmath40 and ( b ) @xmath41 .
the lines are guides for the eyes . ] dependence of @xmath37 at @xmath38 for different system sizes @xmath39 ( a ) without taking average ( @xmath42 ) and ( b ) with taking average @xmath28 ( the numbers are shown in the legend ) .
the lines are guides for the eyes . ]
applying the method above , we investigate the magnetic behavior of the model ( [ eq : h ] ) in the limit of @xmath16 at the electron density @xmath15 .
the results are presented in fig .
[ fig4 ] . at @xmath43 , the square of total magnetization per site @xmath44 grows rapidly [ fig .
[ fig4](a ) ] and the uniform magnetic susceptibility @xmath45 exhibits a peak [ fig .
[ fig4](b ) ] , signaling a ferromagnetic transition .
we estimate @xmath13 from the crossing point of the binder parameter @xmath46 for different system sizes : the estimate is @xmath47 .
dependences of ( a ) square of total magnetization per site , ( b ) uniform magnetic susceptibility , and ( c ) binder parameter for the total magnetization per site .
the electron density is fixed at @xmath15 by controlling @xmath11 . ]
the value of @xmath13 is quite similar to that for the cubic lattice model , @xmath48 @xcite .
this is reasonable because the de ferromagnetism is governed by the kinetic energy of electrons , and does not strongly depend on the details of the lattice structure .
our results demonstrate that systematic and quantitative study of the phase diagram is feasible within relatively small size clusters by implementing the averaging procedure over twisted boundary conditions in the monte carlo calculations .
this gives a starting point for further study of frustrated de - based models and for understanding the physics of pyrochlore - based ferromagnets such as mo pyrochlore oxides .
studies of phase competitions in the model including the super - exchange coupling between localized moments are in progress .
this work was supported by grant - in - aid for scientific research on priority areas ( nos .
17071003 , 19052008 ) , global coe program the physical sciences frontier " and by the next generation super computing project , nanoscience program , mext , japan .
9 zener c 1951 _ phys . rev . _ * 82 * 403 | the double - exchange model , which has been extensively studied in the context of colossal magneto - resistance in perovskite manganese oxides , is known to exhibit a ferromagnetic metallic state at low temperatures because of the interplay between localized moments and itinerant electrons through the hunds - rule coupling . here
we investigate numerically the ferromagnetic transition in the double - exchange model defined on the frustrated pyrochlore lattice as a simple model for ferromagnetic pyrochlore oxides .
we demonstrate that the finite - size corrections are largely reduced by implementing averages over the twisted boundary conditions in the monte carlo simulation , which enables to estimate the ferromagnetic transition temperature in relatively small size clusters .
the estimate is compared with that for the non - frustrated cubic lattice system . |
following the suggestion of @xcite , several collaborations , notably macho and eros , began to search for gravitational microlensing towards the magellanic clouds as an indicator of compact objects in the halo of the milky way @xcite . at about the same time , the ogle collaboration began a survey in the direction of the galactic bulge @xcite .
it was soon found that a much higher event rate occurred in fields towards the galactic bulge relative to the rate towards the magellanic clouds @xcite . since 1990 , approximately 1000 such events have been detected @xcite .
several groups including planet ( probing lensing anomalies network , @xcite ) , mps ( microlensing planet search , @xcite ) and @xmath1fun ( microlensing follow - up network , @xcite ) monitor events much more intensively than the survey groups in order to identify anomalous behavior that can signal the presence of a planet associated with the lens star .
high - magnification events in particular ( those with @xmath2 ) attract the attention of follow - up groups since it is these that are most likely to give detectable planetary signals @xcite .
in addition , for high magnification events the angular size of the source star may be non - negligible in comparison to the lens - source angular separation . in these cases the lightcurves of the events can provide the possibility to determine the lens - source relative proper motion @xcite and atmospheric properties of the source @xcite . in the first years of operation ,
when microlensing alerts came primarily from the macho collaboration , detected event rates were low enough that planet could monitor almost all potentially interesting events with ease .
for the last two years ( the 2002 and 2003 bulge seasons ) , this has not been the case , due to the much improved alert rate since the advent of the ogle iii early warning system ( ews ) , http://www.astrouw.edu.pl/~ogle/ogle3/ews/ews.html @xcite . in excess of 400 events
were alerted by the ews in each of these years .
in addition , approximately 75 events were alerted in 2003 by the moa collaboration @xcite although some of these were duplicates of ews events .
we are now in an era in which a careful selection of events is necessary to optimize planet detection and exclusion productivity .
for this reason , follow - up groups require accurate predictions of eventual maximum amplications in the early days following a detection . for
the remainder of this paper i will focus exclusively on events detected by the ogle iii ews .
most microlensing events are well fitted by a point - source point - mass - lens ( pspl ) model for the magnification @xmath3 at time @xmath4 , @xmath5 the impact parameter @xmath6 is the angular separation between the source and lens measured in units of the angular einstein radius , @xmath7 where @xmath8 is the mass of the lens , @xmath9 is the observer - source distance , @xmath10 the observer - lens distance , @xmath11 the source - lens distance and @xmath12 is the impact parameter at @xmath13 , the time of maximum magnification . the einstein radius crossing time , @xmath14 where @xmath15 is the relative proper motion between the lens and source .
the lightcurve of a pspl event can thus be characterised by @xmath16 parameters , ( @xmath17 ) plus for each @xmath18 telescope + filter combinations , the unmagnified ( baseline ) magnitude of the source star @xmath19 and the blending parameter @xmath20 , where @xmath21 is the fraction of blended ( non - lensed ) light . the maximum magnification
, @xmath22 frequently replaces @xmath23 as a parameter .
the conventional method for predicting peak magnifications is to use @xmath0 minimization techniques to fit pspl models to data from ogle iii ( possibly supplemented by a follow - up group s own data ) as they become available .
such fits are continuously updated and the subsequent predictions revised as data accumulate . experience has shown that early predictions of eventual maximum magnification using these methods systematically yield overpredictions , strongly limiting the usefulness of such estimates .
in particular , very large maximum magnifications ( with large uncertainties ) are often predicted for events that turn out to be of rather low amplitude .
valuable observing time is often wasted monitoring such events in order to confirm their nature .
the reason for this overprediction of values for @xmath24 is that in using @xmath0 minimization for a predictive purpose , one implicitly assumes that all parameter values are equally likely .
however , for microlensing events this is far from being the case . from a purely geometrical perspective ,
high magnification events are exceedingly rare . in practice ,
being of high magnification , they have a higher probability of detection by a survey group .
it is the individual detection efficiency of a survey convolved with the intrinsic event rates ( both of these as a function of event parameters ) that determines the magnification probability , given that an event has been detected .
these ideas can be given a quantitative basis in a bayesian formulation of the problem .
the merits of the bayesian approach to statistical analysis have been discussed at length elsewhere and will not be reargued @xcite . here
we simply note that a bayesian formulation with appropriate priors should produce an unbiased estimate of the eventual microlensing event parameters during the rising part of a lightcurve . from bayes theorem , the probability density for a microlensing event to have a certain set of parameter values @xmath25 , given the fact , @xmath26 , that it has been detected by the ogle iii ews and that data @xmath27 have been acquired
, @xmath28 where @xmath29 . the second term on the right hand side of equation ( 6 ) , @xmath30 ( known as the prior ) , is the underlying probability density for @xmath25 given a detection , i.e. @xmath31 is the probability that @xmath25 is in the range @xmath32 $ ] .
it is this function that incorporates both the natural event occurrence probability and the particular parameter sensitivities of the detection system . the first term , @xmath33 , the likelihood function for @xmath34 given @xmath27 . data from the ogle iii survey consist of @xmath35-band magnitudes and their uncertainties @xmath36 at time @xmath37 ( i.e. @xmath38 ) .
i assume each @xmath39 to be drawn randomly from a normal distribution @xmath40 where the true value of the magnitude at time @xmath37 is @xmath41 .
this implies that @xmath42 where @xmath43 and @xmath44 is the magnitude evaluated from the model parameterised by @xmath25 .
analagous to a @xmath0 minimization , the value of @xmath34 that maximises @xmath45 ( i.e. the posterior mode ) is taken as the best estimator of @xmath46 , the true value of @xmath25 . in the absence of a prior
, this solution reduces to the minimum @xmath0 solution .
we stress here that when sufficient data are available to constrain a fit to a certain event , e.g. when the event is over , the bayesian and @xmath0 minimization techniques give the same parameter values and the choice of prior is largely irrelevant . in other words ,
the solution is not driven by prior probabilities when sufficent empirical information is available ( see @xcite , chapter 2 for a discussion of this point ) .
if the parameters @xmath25 are statistically independent quantities , @xmath47 factorizes as @xmath48 where for brevity i have omitted `` @xmath49 '' in the probability densities on the right hand side of the equation .
it is often more convenient to work in decadic logarithmic units for several of these quantities , in which case @xmath50 with @xmath51 being an arbitrary unit of time . here
i define @xmath52 to be the time to peak magnification from an initial `` alert date '' . for the remainder of this paper
i adopt @xmath53 . it is worth noting that even if each parameter in @xmath25 is independent in @xmath47 , they are not independent in the likelihood function @xmath54 and hence not independent in @xmath45 .
thus when fitting a model to a lightcurve , particularly when only early data are available , the fitted maximum magnification is affected not only by the prior on @xmath24 but also by the priors on the other other parameters .
the criterion used by the ogle iii ews is that a blending parameter is only used when it is more than 3-@xmath55 less than unity and is larger than its formal uncertainty . in this paper
i use the odds ratio test , a natural way to decide between two different models .
i define the odds ratio @xmath56 where @xmath57 indicates the set of model parameters without blending ( @xmath38 when considering only ews data ) . having no _ a priori _ indication about whether to include blending
i choose @xmath58 .
if we assume a unform prior probability density for @xmath20 in the range @xmath59 and zero outside this range , and assuming a gaussian probability density function for @xmath20 about @xmath60 , it can be shown ( see for instance @xcite ch 4 ) that equation ( [ oreqn1 ] ) reduces to @xmath61 for cases in which @xmath60 is more than several @xmath62 away from the cutoffs imposed by the prior . otherwise , for @xmath60 close to 1 , equation ( [ oreqn1 ] ) becomes @xmath63 while for @xmath60 close to 0 @xmath64 the odds ratio is then made up of two terms .
the first of these represents a relative `` goodness of fit '' between the two models while the second is the `` occam penalty '' for introducing a new parameter . only when @xmath65 and the odds ratio is less than unity is a blending parameter used in this analysis .
i have used the set of microlensing events detected by the ews in 2002 to determine the parameter priors . since
our interest is in the set of pspl events , i have removed 41 events that showed deviations from pspl behavior from this analysis .
excluded events were numbered 18 , 23 , 40 , 51 , 68 , 69 , 77 , 80 , 81 , 99 , 113 , 119 , 126 , 127 , 128 , 129 , 131 , 135 , 143 , 149 , 159 , 175 , 194 , 202 , 203 , 205 , 215 , 228 , 229 , 232 , 238 , 254 , 255 , 256 , 266 , 273 , 307 , 315 , 339 , 348 , 360 , out of the complete set of 389 alerts .
parameter values have been obtained for these events using a simplex downhill method to minimise @xmath0 ( ews estimates of @xmath24 , @xmath66 , @xmath67 , @xmath20 can also be obtained from the ews web page ) .
for @xmath68 we require an objective definition of an `` alert date '' that can be applied to all events .
i have arbitrarily chosen a working definition of an alert date as being the date at which three successive data points have been more than 1-@xmath55 brighter than @xmath67 , the baseline magnitude . in practice , @xmath67 can be determined separately from and in advance of the other parameters .
the distributions of @xmath69 , @xmath70 and @xmath71 are shown in figure [ fig1 ] .
for the purposes of obtaining bayesian prior probability densities for these quantities , the distribution functions are adequately represented by the following empirically - chosen functions , also shown in figure [ fig1 ] : @xmath72 \\ p(\lg ( t_{\rm e}/t^ { * } ) ) & = & 0.476 \exp \left [ -(\lg ( t_{\rm e}/t^ { * } ) - 1.333)^2 /0.330
\right ] \\ p(\lg ( \delta t_{\rm 0}/t^ { * } ) ) & = & 0.156 \exp \left [ -(\lg ( \delta t_{\rm 0}/t^ { * } ) - 1.432)^2 /0.458 \right].\end{aligned}\ ] ] it is also instructive to examine the distribution of @xmath12 , shown in figure [ fig2](a ) . in the absence of any selection effects
, this distribution should be uniform .
in fact , there is an enhanced sensitivity to detection of high magnification ( low @xmath12 ) events and a rapid decrease in sensitivity for @xmath73 .
figure [ fig2](a ) also shows the shape of the adopted prior on @xmath69 ( eq . 15 ) when transformed to @xmath12 .
figure [ fig2](b ) shows the same data but excluding those events where @xmath24 has a formal uncertainty greater than 50% .
this illustrates that many high amplification events have maxima that are poorly constrained from ogle data alone .
as a test of the bayesian method , i have applied a fitting procedure that maximises @xmath45 to a sample of the pspl events alerted in real time by the ogle iii ews in 2003 . these consist of events ogle-2003-bul-138 to ogle-2003-bul-462 and excluding events numbered 145 , 160 , 168 , 170 , 176 , 192 , 200 , 230 , 236 , 252 , 260 , 266 , 267 , 271 , 282 , 286 , 293 , 303 , 306 , 311 , 359 , 380 , 419 that do not appear to be due to pspl microlensing and 188 , 197 , 245 , 263 , 274 , 297 , 387 , 399 , 407 , 412 , 413 , 417 , 420 , 422 , 429 , 430 , 432 , 433 , 435 , 437 , 440 , 441 , 442 , 443 , 444 , 449 , 450 , 452 , 453 , 454 , 455 , 457 , 459 , 461 , 462 that were still ongoing at the time of writing .
events ogle-2003-bul-137 and earlier were anounced by the ews in a single email at the beginning of the 2003 bulge season and thus not alerted in real time .
ogle-2003-bul-238 ( a. gould 2004 , private communication ) and 262 @xcite are events in which the lens is known to have transited the source and ogle-2003-bul-208 and 222 may also involve finite source effects .
these events have not been excluded . for the remaining sample of 267 events ,
i have used only the ogle iii data taken before the ews alert time , defined as the reception of the alert email by the author .
for the zero point of @xmath52 for each event , i have used the definition in 2 except for cases in which this has not occured before the ews alert time in which case the latter has been used as the zero point . as mentioned in 1 , different fitting codes can produce different estimates of maximum magnifications , particularly for high - magnification events for which blending may be involved . in particular , there is a concern that a direct comparison of predictions with the ews alert predictions may suffer from such differences . in order to compare the maximum magnifications predicted by the bayesian method with those predicted using @xmath0 fitting , i have thus used very similar computer codes to make @xmath45 and @xmath0 optimisations , electing not to use the ews - fitted parameters . to avoid the problem of slightly different blending parameters resulting in large differences in derived magnifications , for each event
i compute the brightness increase , @xmath74 , where @xmath75 is the magnitude at @xmath13 .
the predicted values of @xmath76 using only the pre - alert data for an event are compared with the values determined using all the data . when all the data are available , the parameters derived using @xmath0 and @xmath77 are almost always identical .
exceptions to this are in a few cases for which there are no data over the peak to constrain the fits . the predictive performances of the bayesian and @xmath0 models at the time of ews alert are illustrated in figures [ fig3 ] [ fig5 ] .
figure [ fig3 ] shows the distributions of predicted peak magnifications for both models and compares these with the eventual values .
figure [ fig4 ] shows the same data as a function of @xmath12 . for the @xmath0 models ,
there is clearly a population of low @xmath12 events with predicted brightenings of more than 10 magnitudes that do not eventuate .
such overpredictions are not present in the bayesian fitted models . on the other hand
, there is a tendency for the bayesian models to underpredict the peak , and at alert time to fail to predict the small population of high magnification events in figure [ fig3](a ) . in figure [ fig5 ]
i compare the distribution of the differences in predicted vs actual brightenings for both models .
again , the tendency for the @xmath0 fits to overestimate the peak is obvious .
as pointed out in 4.1 , baysesian solutions to early lightcurve data often fail to indicate the nature of high magnification events
. it would be of concern if high magnification events were not observed due to this tendency . to illustrate in more detail the behavior of bayesian vs @xmath0 models , i consider here examples of low and high magnification events .
these examples show several generic aspects of how bayesian vs @xmath0 solutions evolve as data accumulate .
ogle-2003-blg-171 was a low magnification event ( @xmath78 ) . at this magnification
, the source star barely passes within one einstein radius of the lens and the event is unsuitable for detecting a planetary anomaly .
this is typical of the type of event that a follow - up program should avoid observing .
figure [ fig6 ] shows @xmath0 and bayesian fitted lightcurves at 5 day intervals as the event evolves from its alert date .
the predicted maximum magnifications corresponding to each panel are listed in table [ tbl1 ] . at alert ,
the event is predicted to be of low magnification but by jd 2452785 ( panel d in fig .
[ fig6 ] ) , the @xmath0 solution suggests a high magnification , albeit with a large uncertainty .
since the lightcurve appears to be rising rapidly , follow - up programs may well begin observing the event in order to improve on the high uncertainty in the predicted peak magnification .
as more data accumulate , the low - magnification nature of the event becomes apparent .
although the true nature of the event would be identified relatively quickly by a follow - up observing program , there is a not - insignificant overhead associated with adding the event to the program .
in contrast to the @xmath0 fit , the predicted peak magnification for the bayesian solution changes steadily with time . at no time
is a high - magnification event suggested and a follow - up strategy based on this method would ignore the event .
ogle-2003-blg-208 ( fig .
[ fig7 ] , table [ tbl2 ] ) reached a moderately high magnification ( @xmath79 , @xmath80 ) .
the projected source trajectory passed as close as 0.02 @xmath81 to the lens and thus had a high probability of intersecting a central caustic if it were present .
the alert date for this event corresponds to panel ( c ) in figure [ fig7 ] at which time the predictions of peak magnification are @xmath82 and 4.4 for the @xmath0 and bayesian solutions respectively .
the bayesian prediction of @xmath83 is sufficiently high to warrant the attention of a follow - up observing program such as planet . as data acumulate ,
the @xmath0 predicted peak magnification rises until reaching @xmath84 in panel ( g ) . the true peak magnification ( @xmath85 starts to become apparent from panel ( h ) as the event peaks .
in contrast , the bayesian predicted peak magnification rises steadily until the true peak magnification is identified from around the time of panels ( f ) ( g ) .
the behavior illustrated by these two examples is typical . for low magnification events ,
the bayesian model never indicates them as being worthy of observational follow - up . for high magnification events that should be observed ,
the peak magnification is initially underestimated but adjusts to an appropriate prediction as soon as the data indicate . in all cases
examined , this occurs relatively early in the event when the magnification , @xmath86 . for both high and low magnification events ,
the bayesian predicted peak magnification changes smoothly while the @xmath0 prediction is prone to large changes as new data points are included .
the bayesian solutions usually converge to the correct amplification earlier than the @xmath0 solutions .
high magnification events provide the best opportunity for detecting signals of planets around lens stars and for obtaining upper limits on their abundances .
intensive photometric monitoring programs are hampered currently by difficulties in identifying high magnification events well before peak .
systems that use @xmath0 minimization to fit pspl models to early data are prone to exagerated predictions of peak magnification .
such prections induce observers to spend their time monitoring events that ultimately have little statistical power .
i have shown here that a predictive system based on a bayesian formalism that takes account of the characteristics of a detection system is immune to such behavior .
although such a bayesian system tends to initially underpredict the peak for high magnification events , accurate prediction occurs as soon as sufficient data accumulate to justify the assertion . in all cases
examined , this occurs well ahead of peak in their associated lightcurves and early enough for the events to be targeted for observation .
implementation of such a system based on the ogle early warning system should result in much improved observing productivity for the 2004 season .
i am grateful to martin dominik for his comments on an earlier version of this paper .
i think the referee , andy gould , for his suggested improvements to the manuscript .
this work was supported by the marsden fund under contract uoc302 .
albrow , m.d . , et al . , 1998 , , 509 , 687 albrow , m.d . , et al . , 2001 , , 556 , 113 alcock , c. , et al . , 1993 , , 365 , 621 alcock , c. , et al . , 1995 , , 445 , 133 alcock , c. , et al . ,
1997a , , 479 , 119 alcock , c. , et al .
, 1997b , , 491 , 436 alcock , c. , et al . , 2000 , , 541 , 734 aubourg , e. , et al . , 1993 , , 365 , 623 bond , i.a . , et al , 2002 , , 331 , 19 dominik , m. , et al , 2002 , planetary and space science , 50 , 299 gaudi , b.s . , naber , r.m . , sackett , p.d . , 1998 , , 500 , 33 gaudi , b.s .
, et al . , 2002 , , 566 , 463 gould , a. , 1994 , , 421 , l71 griest , k. , safizadeh , n. , 1998 , , 500 , 37 heyrovsk , d. , 2003 , , 594 , 464 loredo , t.j . , 1990 , in p.f .
fougere , ed , maximum entropy and bayesian methods , kluwer , dordrecht , pp81 - 142 paczynski , b. , 1986 , , 304 , 1 rhie , s.h . , 1999 , , 522 , 1037 sivia , d.s . ,
1996 , data analysis .
oxford university press , oxford udalski , a. , et al .
, 1992 , acta astron . , 42 , 253 udalski , a. , et al . , 1993 , acta astron . , 43 , 289 udalski , a. , et al .
, 1994a , acta astron . , 44 , 165 udalski , a. , et al . ,
1994b , acta astron .
, 44 , 227 udalski , a. , et al . ,
2000 , acta astron . ,
50 , 1 udalski , a. , 2003 , acta astron . , 53 , 291 yoo , j. , et al . , 2004 , , in press , astro - ph/0309302 rrrrrrrrrrrr a & 1.240 & 0.670 & 1.000 & & 1.240 & & 1.053 & 0.014 & 1.000 &
& 1.053 + b & 1.050 & 0.009 & 1.000 & & 1.050 & & 1.051 & 0.009 & 1.000 & & 1.051 + c & 1.050 & 0.009 & 1.000 & & 1.050 & & 1.051 & 0.009 & 1.000 &
& 1.051 + d & 34934.748 & & 1.000 & & 34934.748 & & 1.300 & 0.314 & 1.000 & & 1.300 + e & 1.449 & 0.557 & 1.000 & & 1.449 & & 1.273 & 0.150 & 1.000 & & 1.273 + f & 1.459 & 0.330 & 1.000 & & 1.459 & & 1.367 & 0.169 & 1.000 &
& 1.367 + g & 1.539 & 0.353 & 1.000 & & 1.539 & & 1.446 & 0.199 & 1.000 & & 1.446 + h & 1.404 & 0.082 & 1.000 & & 1.404 & & 1.392 & 0.072 & 1.000 &
& 1.392 + i & 2.524 & 1.366 & 0.233 & 0.206 & 1.355 & & 1.356 & 0.019 & 1.000 & & 1.356 + j & 1.374 & 0.005 & 1.000 & & 1.374 & & 1.374 & 0.005 & 1.000 & & 1.374 + rrrrrrrrrrrr a & 63601.415 & & 1.000 & & 63601.415 & & 2.055 & 0.657 & 1.000 & & 2.055 + b & 138528.609 & & 0.209 & 0.349 & 28987.904 & & 2.150 & 0.700 & 1.000 & & 2.150 + c & 226215.441 & & 0.170 & 0.168 & 38337.850 & & 4.362 & 3.407 & 1.000 & & 4.362 + d & 224373.536 & & 0.382 & 0.324 & 85643.167 & & 5.934 & 4.681 & 1.000 & & 5.934 + e & 778431.912 & & 0.438 & 0.232 & 340873.531 & & 15.425 & 14.683 & 1.000 & & 15.425 + f & 1669005.679 & & 0.412 & 0.159 & 687516.769 & & 41.122 & 31.856 & 0.452 & 0.007 & 19.150 + g & 4966165.089 & & 0.586 & 0.138 & 2912064.088 & & 46.816 & 12.146 & 0.456 & 0.004 & 21.888 + h & 55.641 & 15.221 & 0.293 & 0.076 & 17.022 & & 49.816 & 1.378 & 0.326 & 0.002 & 16.893 + i & 53.452 & 13.513 & 0.303 & 0.073 & 16.916 & & 48.446 & 1.356 & 0.333 & 0.002 & 16.813 + j & 48.204 & 8.750 & 0.334 & 0.056 & 16.740 & & 45.814 & 1.096 & 0.350 & 0.002 & 16.661 + | gravitational microlensing events with high peak magnifications provide a much enhanced sensitivity to the detection of planets around the lens star .
however , estimates of peak magnification during the early stages of an event by means of @xmath0 minimization frequently involve an overprediction , making observing campaigns with strategies that rely on these predictions inefficient .
i show that a rudimentary bayesian formulation , incorporating the known statistical characteristics of a detection system , produces much more accurate predictions of peak magnification than @xmath0 minimisation .
implementation of this system will allow efficient follow - up observing programs that focus solely on events that contribute to planetary abundance statistics . |
visualizing set systems is a basic problem in data visualization . among the oldest and most popular set visualization tools are the venn and euler diagrams .
however , other methods are preferred when the data involves a large number of sets with complex intersection relations @xcite .
in particular , a variety of tools have been proposed for set systems where the elements are associated with location data .
many of these methods use geometric graphs to represent set membership , motivated by reducing the amount of ink used in the representation , including linesets @xcite , kelp diagrams @xcite and kelpfusion @xcite . ,
@xmath7 , and @xmath8 , drawn as outlines in different colours .
right : the corresponding ( minimum ) coloured spanning graph . refer to section [ sec : prelim ] for an explanation of colour use.,title="fig : " ] , @xmath7 , and @xmath8 , drawn as outlines in different colours .
right : the corresponding ( minimum ) coloured spanning graph . refer to section [ sec : prelim ] for an explanation of colour use.,title="fig : " ] hurtado et al .
@xcite recently proposed a method for drawing sets using outlines that minimise the total visual clutter .
the underlying combinatorial problem is to compute a _
minimum colored spanning graph _ ; see figure [ fig : example ] .
they studied the problem for @xmath0 points in a plane and two sets ( each point is a member of one or both sets ) .
the output is a graph with the minimum sum of edge lengths such that the subgraph induced by each set is connected .
they gave an algorithm that runs in @xmath9-time , and a @xmath10-approximation in @xmath11 time , where @xmath5 is the steiner ratio ( the ratio between the length of a minimum spanning tree and the length of a minimum steiner tree ) .
efficient algorithms are known in two special cases : one runs in @xmath6 time for collinear points that are already sorted @xcite ; the other runs in @xmath12 time for cocircular points , where @xmath13 is the number of points that are elements of both sets @xcite .
this problem also has applications for connecting different networks with minimum cost , provided that edges whose endpoints belong to both networks can be shared .
* results and organization .
* we study the minimum colored spanning graph problem for @xmath0 points in a plane and @xmath1 sets , @xmath2 . the formal definition and some properties of the optimal solution are in section [ sec : prelim ] . in section [ sec : hard ] ,
we show that min-@xmath1csg is np - complete for all @xmath14 , and in section [ sec : approximation ] we provide an @xmath15-approximation algorithm for @xmath4 that runs in @xmath16 time , where @xmath13 is the number of multichromatic points .
this improves the previous @xmath17-approximation from @xcite .
section [ sec : collinear ] describes an algorithm for the special case of collinear points that runs in @xmath18 time .
due to space constraints , some proofs are omitted ; they can be found in appendix [ app : proofs ] .
in this section , we define the problem and show a property of the optimal solution related to the minimum spanning trees , which is used in sections [ sec : hard ] and [ sec : approximation ] .
* definitions . * given a set of @xmath0 points in the plane @xmath19 and subsets @xmath20 , we represent set membership with a function @xmath21 , where @xmath22 iff @xmath23 for every _ primary color _ @xmath24 .
we call @xmath25 the _ color _ of point @xmath26 . a point @xmath26 is _ monochromatic _ if it is a member of a single set @xmath27 , that is , @xmath28 , and _ multi - chromatic _ if @xmath29 . for an edge @xmath30 in a graph @xmath31 , we use the shorthand notation @xmath32 for the shared primary colors of the two vertices . for every @xmath33 ,
we let @xmath34 denote the subgraph of @xmath31 induced by @xmath35 .
all figures in this paper depict only three primary colors : ` r ` , ` b ` , and ` y ` for red , blue , and yellow respectively .
multi - chromatic points and edges are shown green , orange , purple , or black if their color is @xmath36 , @xmath37 or @xmath38 , or @xmath39 , respectively .
see , for example , fig .
[ fig : example ] ( b ) .
a _ colored spanning graph _ for the pair @xmath40 , denoted csg@xmath40 , is a graph @xmath31 such that @xmath41 is connected for every primary color @xmath33 .
the _ minimum colored spanning graph _ problem ( min - csg ) , for a given pair @xmath40 , asks for the minimum cost @xmath42 of a csg@xmath40 , where @xmath43 is the euclidean length of @xmath44 .
when we wish to emphasize the number @xmath1 of primary colors , we talk about the min-@xmath1csg problem . * monochromatic edges in a minimum csg . *
the following lemma shows that we can efficiently compute some of the monochromatic edges of a minimum csg for an instance @xmath40 using the _ minimum spanning tree _
( _ mst _ ) of @xmath35 for every primary color @xmath45 .
[ lem : monocrhomatic ] let @xmath40 be an instance of min - csg and @xmath24 .
let @xmath46 be the edge set of an mst of @xmath35 , and let @xmath47 be the set of multi - chromatic points in @xmath35 .
then there exists a minimum csg that contains at least @xmath48mst@xmath49 edges of @xmath46 .
the common edges of @xmath46 and of such a minimum csg can be computed in @xmath11 time .
construct a monochromatic subset @xmath50mst@xmath51 by successively removing a longest edge from the path in mst@xmath52 between any two points in @xmath47 .
an mst@xmath52 can be computed in @xmath11 time , and @xmath53 can be obtained in @xmath6 time . the graph @xmath54 has @xmath55 components , each containing one element of @xmath47 , hence @xmath56mst@xmath49 .
let @xmath57 be a minimum csg .
while there is an edge @xmath58 , we can find an edge @xmath59 such that exchanging @xmath60 for @xmath44 yields another minimum csg . indeed , since @xmath61 is connected , the insertion of the edge @xmath44 creates a cycle @xmath62 that contains @xmath44 . consider the longest ( open or closed ) path @xmath63 that is monochromatic and contains @xmath44 .
note that at least one of the endpoints of @xmath44 is monochromatic , therefore @xmath64 contains at least two monochromatic edges . since every component of @xmath54 is a tree and contains only one multi - chromatic point
, there is a monochromatic edge @xmath65 in @xmath64 .
we have @xmath66 , because there is a cut of the complete graph on @xmath35 that contains both @xmath44 and @xmath60 , and @xmath67mst@xmath51 .
since @xmath68 , the deletion of @xmath60 can only influence the connectivity of the induced subgraph @xmath69 .
consequently , @xmath70 is a csg with equal or lower cost than @xmath57 .
by successively exchanging the edges in @xmath71 , we obtain a minimal csg containing @xmath53 .
hurtado et al .
@xcite gave an @xmath9-time algorithm for min-2csg , by a reduction to a matroid intersection problem on the set of all possible edges on @xmath72 , which has @xmath73 elements .
their algorithm for matroid intersection finds @xmath73 single source shortest paths in a bipartite graph with @xmath73 vertices and @xmath74 edges , which leads to an overall running time of @xmath9 .
we improve the runtime to @xmath16 , where @xmath13 is the number of multi - chromatic points . [
cor : m6 ] an instance @xmath40 of min-2csg can be solved in @xmath16 time , where @xmath13 is the number of multi - chromatic points in @xmath72 . by lemma [ lem : monocrhomatic ]
, we can compute two spanning forests on @xmath75 and @xmath7 , respectively , each with @xmath13 components , that are subgraphs of a minimum csg in @xmath11 time .
it remains to find edges of minimum total length that connect these components in each color , for which we can use the same matroid intersection algorithm as in @xcite , but with a ground set of size @xmath76 .
we show that the decision version of min - csg is np - complete .
we define the decision version of min - csg as follows : given an instance @xmath40 and @xmath77 , is there a csg @xmath78 such that @xmath79 ? [
lem : np ] min-@xmath1csg is in np . given a set of edges @xmath80 , we can verify if @xmath78 is a @xmath81 in @xmath82 time by testing connectivity in @xmath41 for each primary color @xmath83 , and then check whether @xmath84 in @xmath85 time .
we reduce min-3csg from planar - monotone-3sat , which is known to be np - complete @xcite . for every instance @xmath86 of planar - monotone-3sat , we construct an instance @xmath87 of min-3csg .
an instance @xmath86 consists of a plane bipartite graph between @xmath0 variable and @xmath13 clause vertices such that every clause has degree three or two , all variables lie on the @xmath88-axis and edges do not cross the @xmath88-axis .
clauses are called _ positive _ if they are in the upper half - plane or _ negative _ otherwise .
the problem asks for an assignment from the variable set to @xmath89 such that each positive ( negative ) clause is adjacent to a ` true ` ( ` false ` ) variable .
equivalent to the boolean formula @xmath90 . ]
given an instance @xmath86 of planar - monotone-3sat , we construct @xmath87 as shown in fig .
[ fig : reduction - complete ] ( a single variable gadget is shown in fig . [
fig : reduction ] in the appendix ) .
the points marked with small disks are called _ active _ and they are the only multi - chromatic points in the construction . the dashed lines in a primary color represent a chain of equidistant monochromatic points , where the gap between consecutive points is @xmath91 . a purple ( resp . ,
black ) dashed line represents a red and a blue ( resp .
, a red , a blue , and a yellow ) dashed line that run @xmath91 close to each other .
informally , the value of @xmath91 is set small enough such that every point in the interior of a dashed line is adjacent to its neighbors in any minimum csg .
the boolean assignment of @xmath86 is encoded in the edges connecting active points .
we break the construction down to gadgets and explain their behavior individually
. the long horizontal purple dashed line is called _ spine _ and the set of yellow dashed lines ( shown in fig .
[ fig : gadgets](a ) ) is called _ cage_. the rest of the construction consists of variable and clause gadgets ( shown in figs .
[ fig : gadgets](b ) and ( c ) ) .
the width of a variable gadget depends on the degree of the corresponding variable in the bipartite graph given by the instance @xmath86 . for every edge incident to the variable , we repeat the middle part of the gadget as shown in fig . [
fig : gadgets](b ) ( cf . fig .
[ fig : reduction ] , where a variable of degree-2 is shown ) .
the vertical black dashed lines are called _ ribs _ and the set of three or four active points close to an endpoint of a rib is called _
switch_. the variable gadget contains switches of two different sizes alternately from left to right .
a _ 2-switch _ ( resp . ,
_ 2@xmath92-switch _ ) is a switch in which active points are at most 2 ( resp .
, @xmath93 ) apart .
the clause gadgets are positioned as the embedding of clauses in @xmath86 ; refer to fig .
[ fig : reduction - complete ] .
each active point of a positive ( negative ) clause is assigned to a @xmath93-switch and positioned vertically above ( below ) the active point of the rib , at distance @xmath93 from it .
let @xmath94 be the set of all monochromatic edges of a minimum csg computable by lemma [ lem : monocrhomatic ] .
let @xmath95 be the number of edges in the bipartite graph of @xmath86 .
the instance @xmath87 contains @xmath96 active points , so @xmath97 contains @xmath96 connected components . by construction ,
the number of @xmath91-edges in a solution of @xmath87 between components of @xmath97 is upper bounded by @xmath98 ( one edge per color per component ) .
finally , we set @xmath99 and we choose @xmath100 and @xmath101 . this particular choice of @xmath91 and @xmath92 is justified by the proofs of corollaries [ cor:2-pair ] and [ cor : clause ] . by construction
, @xmath87 has the following property : 1 . for every partition of the components of @xmath102 into two sets @xmath103 , where @xmath104 is a primary color , let @xmath105 be the shortest edge between @xmath106 and @xmath107 .
then either @xmath108 or @xmath109 and @xmath110 are active points in the same switch .
a _ standard solution _ of min-3csg is a solution that contains @xmath94 and in which every edge longer than @xmath91 is between two active points of the same switch .
[ lem : sat2csg ] let @xmath86 be a positive instance of planar - monotone-3sat . then @xmath87 is a positive instance of min-3csg . to prove the lemma
, we construct a standard solution for @xmath87 based on the solution for @xmath86 .
this proof , and subsequent proofs , argues about all possible ways to connect the vertices in a switch of @xmath87 .
the most efficient ones are shown in fig . [
fig : cases ] ; these may appear in an optimal solution .
refer to fig .
[ fig : switch ] in the appendix for a full list .
[ lem : existance ] if @xmath87 is a positive instance of min-3csg , there exists a standard solution for this instance . before proving the other direction of the reduction , we show some properties of a standard solution .
the active points in a switch impose some local constraints .
the black and purple points attached to horizontal dashed lines determine the _ switch constraint _ : since these points have more colors than their incident dashed lines , they each are incident to at least one edge in the switch .
each rib determines a _ rib constraint _ to a pair of switches that contain its endpoints : at least one of these switches must contain an edge between its black active points or else there is no yellow path between this rib and the cage .
the following lemmas establish some bounds on the length of the edges used to satisfy local constraints of a pair of switches adjacent to a rib .
we refer to this pair as a @xmath111-pair or @xmath93-pair according to the type of the switch .
[ lem : min ] in a standard solution , the minimum length required to satisfy the local constraints of a 2-pair ( resp . , @xmath93-pair ) is @xmath112 ( resp . , @xmath113 ) .
[ cor:2-pair ] in a standard solution , every 2-pair is connected minimally .
[ lem : clause ] in a standard solution , for each clause gadget , there exists a @xmath93-pair with local cost at least @xmath114 .
[ cor : clause ] in a standard solution , for each clause gadget , there exists a @xmath93-pair connected as fig .
[ fig : cases](c ) .
all other @xmath93-pairs are connected minimally as shown in fig .
[ fig : cases](b ) .
[ lem : csg2sat ] let @xmath87 be a positive instance of min-3csg . then @xmath86 is a positive instance of planar - monotone-3sat .
the following theorem is a direct consequence of lemmata [ lem : np ] , [ lem : sat2csg ] , and [ lem : csg2sat ] .
min-@xmath1csg is np - complete for @xmath115 .
hurtado et al . @xcite gave an approximation algorithm for min-@xmath1csg that runs in @xmath11 time and achieves a ratio of @xmath116 , where @xmath5 is the steiner ratio .
the value of @xmath5 is not known and the current best upper bound is @xmath117 by chung and graham @xcite ( gilbert and pollack @xcite conjectured @xmath118 ) . for the special case @xmath4 , the previous best approximation ratio is @xmath119 .
we improve the approximation ratio to 2 , and then further to 1.81 .
our first algorithm immediately generalises to @xmath120 , and yields an @xmath121-approximation , improving on the general result by hurtado et al . ; our second algorithm also generalizes to @xmath122 , however , we do not know whether it achieves a good ratio .
suppose we are given an instance of min-3csg defined by @xmath40 where @xmath123 and the set of primary colors is @xmath124`r , b , y`@xmath125 .
we define @xmath126 where @xmath127 .
let @xmath128 be an optimal solution for min-3csg , and put @xmath129 .
algorithm a1 computes a minimum red - blue - purple graph @xmath130 in @xmath16 time , where @xmath131 by corollary [ cor : m6 ] ; then computes a minimum spanning tree @xmath132 of @xmath133 , and returns the union @xmath134 .
since @xmath128 contains a red , a blue , and a yellow spanning tree , we have @xmath135 and @xmath136 ; that is , algorithm a1 returns a solution to min-3csg whose length is at most @xmath137 .
algorithm a1 returns a 2-approximation for min-3csg ; it runs in @xmath16 time on @xmath0 points , @xmath13 of which are multi - chromatic .
algorithm a1 can be extended to @xmath1 colors by partitioning the primary colors into @xmath138 groups of at most two and computing the minimum csg for each group .
the union of these graphs is a @xmath138-approximation that can be computed in @xmath139 time .
algorithm a2 computes six solutions for a given instance of min-3csg , @xmath140 , and returns one with minimum weight .
graph @xmath141 is the union of @xmath142 and @xmath132 defined above .
graphs @xmath143 and @xmath144 are defined analogously : @xmath145 and @xmath146 , each of which can be computed in @xmath9 time by @xcite .
let @xmath147 be the set of `` black '' points that have all three colors , and let @xmath148 be an mst of @xmath149 , which can be computed in @xmath11 time .
we augment @xmath148 into a solution of min-3csg in three different ways as follows .
first , let @xmath150 be the minimum forest such that @xmath151 is a minimum red - blue - purple spanning graph on @xmath152 .
@xmath150 can be computed in @xmath16 time by the same matroid intersection algorithm as in corollary [ cor : m6 ] , by setting the weight of any edge between components containing black points to zero .
similarly , let @xmath153 be the minimum forest such that @xmath154 is a spanning tree on @xmath133 , which can be computed in @xmath11 time by prim s algorithm .
now we let @xmath155 .
similarly , let @xmath156 and @xmath157 .
algorithm a2 returns a @xmath3-approximation for min-3csg ; it runs in @xmath16 time on an input of @xmath0 points , @xmath13 of which are multi - chromatic .
consider an instance @xmath40 of min-3csg , and let @xmath158 be an optimal solution with @xmath159 .
partition @xmath160 into @xmath161 subsets : for every color @xmath162 , let @xmath163 , that is @xmath164 is the set of edges of color @xmath165 in @xmath128 . put @xmath166 .
then we have @xmath167 . without loss of generality
, we may assume @xmath168 .
first , consider @xmath169 .
the edges of @xmath128 whose colors include red or blue ( resp .
, yellow ) form a connected graph on @xmath152 ( resp .
, @xmath170 .
consequently , @xmath171 the combination of and yields @xmath172 next , consider @xmath155 .
the edges of @xmath128 whose colors include yellow contain a spanning tree on @xmath133 , hence a steiner tree on the black points @xmath149 . specifically , the black edges in @xmath173 form a black spanning forest , which is completed to a steiner tree by some of the edges of @xmath174 .
this implies @xmath175 since @xmath148 is a spanning tree on the black vertices @xmath176 , and reduce to @xmath177 the combination of and yields @xmath178 therefore , @xmath179 if we set @xmath180 , then both and give the same upper bound @xmath181 where we used the current best upper bound for the steiner ratio @xmath117 from @xcite .
in this section we consider instances of min-@xmath1csg , @xmath40 , where @xmath14 and @xmath72 consists of collinear points .
an example is shown in fig . [
fig : example-1d ] . without loss of generality ,
@xmath182 and the points @xmath183 , @xmath184 , lie on the @xmath88-axis sorted by @xmath88-coordinates .
we present a dynamic programming algorithm that solves min-@xmath1csg in @xmath18 time .
our first observation is that if the points in @xmath72 are collinear , we may assume that every edge satisfies the following property . _
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ if @xmath185 , @xmath186 , is an edge , then there is no @xmath95 , @xmath187 , such that @xmath188 . (
@xmath189 ) _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ [ lem : unique - color ] for every graph @xmath31 , there exists a graph @xmath190 of the same cost that satisfies ( @xmath189 ) and for each color @xmath191 , every component of @xmath41 is contained in some component of @xmath54 . in particular , min-@xmath1csg has a solution with property ( @xmath189 ) .
let @xmath31 be a graph , and let @xmath192 denote the set of triples @xmath193 such that @xmath194 , @xmath30 , and @xmath195 .
if @xmath196 , then @xmath197 satisfies ( @xmath189 )
. suppose @xmath198 .
for every triple @xmath199 , successively , replace the edge @xmath200 by two edges @xmath201 and @xmath202 ( i.e. , subdivide edge @xmath203 at @xmath204 ) .
note that @xmath205 , consequently @xmath183 and @xmath206 remain in the same component for each primary color @xmath207 .
each step maintains the total edge length of the graph and strictly decreases @xmath192 .
after @xmath208 subdivision steps , we obtain a graph @xmath190 as required . in the remainder of this section
we assume that every edge has property ( @xmath189 ) .
furthermore , all graphs considered in this section are defined on an interval of consecutive vertices of @xmath72 .
[ cor : unique - edge ] let @xmath31 be a graph and let @xmath209 . 1 .
[ cor : unique ] if @xmath210 is an edge between @xmath211 and @xmath212 and @xmath213 , then the endpoints of @xmath44 are uniquely determined .
specifically , if @xmath214 with @xmath215 , then @xmath216 is the largest index such that @xmath217 , and @xmath218 is the smallest index such that @xmath219 .
[ cor : overlap ] if two edges @xmath220 overlap , then @xmath221 .
suppose , to the contrary , that there is index @xmath222 , @xmath223 , such that @xmath224 .
then edge @xmath185 and point @xmath204 violate ( @xmath189 ) .
the case that there is some @xmath222 , @xmath225 , leads to the same contradiction . without loss of generality
@xmath226 and @xmath227 with @xmath228
. then both edges @xmath229 and @xmath230 are between @xmath211 and @xmath231 , contradicting part .
the basis for our dynamic programming algorithm is that min-@xmath1csg has the _ optimal substructure _ and _ overlapping substructures _ properties when the points in @xmath72 are collinear .
we introduce some notation for defining the subproblems . for indices
@xmath232 , let @xmath233=\{p_a,\ldots , p_b\}$ ] . for every graph @xmath31 and index @xmath234 ,
we partition the edge set @xmath80 into three subsets as follows : let @xmath235 be the set of edges induced by @xmath236 $ ] , @xmath237 the set of edges induced by @xmath238 $ ] , and @xmath239 the set of edges between @xmath236 $ ] and @xmath238 $ ] . with this notation , min-@xmath1csg has the following optimal substructure property .
[ lem : opt - subproblem ] let @xmath31 be a minimum csg , @xmath209 , and @xmath240 be the family of edge sets @xmath241 on @xmath236 $ ] such that @xmath242 is a csg .
then @xmath242 is a minimum csg iff @xmath243 has minimum cost .
if @xmath242 is a minimum csg , but some @xmath244 costs less than @xmath235 , then @xmath245 would be a csg that costs less than @xmath31 , contradicting the minimality of @xmath242 .
if @xmath243 has minimum cost , but @xmath31 costs less than @xmath242 , then @xmath246 would costs less than @xmath241 , contradicting the minimality of @xmath243 .
lemma [ lem : opt - subproblem ] immediately suggests a nave algorithm for min-@xmath1csg : guess the edge set @xmath247 of a minimum csg @xmath31 , and compute a minimum - cost set @xmath241 on @xmath236 $ ] such that @xmath248 is a csg .
however , all possible edge sets @xmath247 could generate @xmath249 subproblems .
we reduce the number of subproblems using the overlapping subproblem property . instead of guessing @xmath247
, it is enough to guess the information relevant for finding the minimal cost @xmath241 on @xmath236 $ ] .
first , the edges in @xmath239 can be uniquely determined by the set of their colors ( using corollary [ cor : unique - edge ] ) .
second , the only useful information from @xmath237 is to tell which points in @xmath236 $ ] are adjacent to the same component of @xmath250_c,(e_i^+)_c)$ ] , for each primary color @xmath33 .
this information can be summarized by @xmath1 equivalence relations on the sets @xmath251 .
we continue with the details .
we can encode @xmath239 by the set of its colors @xmath252 . for @xmath209 , a set of edges @xmath253 between
@xmath236 $ ] and @xmath238 $ ] is _ valid _ if there exists a csg @xmath31 such that @xmath254 .
[ lem : valid ] for @xmath209 , an edge set @xmath253 between @xmath236 $ ] and @xmath238 $ ] is _ valid _ iff for every primary color @xmath191 , there is an edge @xmath255 such that @xmath256 whenever both @xmath236_c$ ] and @xmath238_c$ ] are nonempty .
we encode the relevant information from @xmath237 using @xmath1 equivalence relations as follows .
for every @xmath257 , the components of @xmath250_c,(e_i^+)_c)$ ] define an equivalence relation on @xmath258 , which we denote by @xmath259 : two edges in @xmath258 are related iff they are incident to the same component of @xmath250_c,(e_i^+)_c)$ ] .
let @xmath260 .
the equivalence relation @xmath259 , in turn , determines a graph @xmath261_c , e(\pi_i^c))$ ] : two distinct vertices in @xmath236_c$ ] are adjacent iff they are incident to equivalent edges in @xmath258 ( that is , two distinct vertices in @xmath236_c$ ] are adjacent iff they both are adjacent to the same component of @xmath250_c,(e_i^+)_c)$ ] ) . see fig .
[ fig : collinear ] for examples of @xmath262 and @xmath263 .
the condition that @xmath248 is a csg can now be formulated in terms of @xmath239 and @xmath263 ( without using @xmath237 directly ) .
[ lem : halves ] let @xmath31 be a csg , @xmath264 , and @xmath241 an edge set on @xmath236 $ ] .
the graph @xmath248 is a csg iff the graph @xmath261_c,(x_i^-)_c\cup e(\pi_i^c))$ ] is connected for every @xmath265 .
we can now define subproblems for min-@xmath1csg . for an index @xmath209 , a valid set @xmath239 , and equivalence relations @xmath266 ,
let @xmath267 be the family of edge sets @xmath241 on @xmath236 $ ] such that for every @xmath265 , the graph @xmath261_c,(x_i^-)_c\cup e(\pi_i^c))$ ] is connected .
the subproblem * * a**@xmath268 $ ] is to find the minimum cost of an edge set @xmath269 .
note that for @xmath270 , * * a**@xmath271 $ ] is the minimum cost of a csg for an instance @xmath40 of @xmath272 .
next , we establish a recurrence relation for * * a**@xmath273 $ ] , which will allow computing * * a**@xmath271 $ ] by dynamic programming . for @xmath274
, we have * * a**@xmath275=0 $ ] for any valid @xmath276 and @xmath277 . for all @xmath278 , @xmath279
, we wish to express * * a**@xmath268 $ ] in terms of * * a**@xmath280 $ ] s for suitable @xmath281 and @xmath282 .
we say that two valid edge sets @xmath281 and
@xmath239 are _ compatible _ if there exists an @xmath269 for some @xmath263 such that @xmath283 . we can characterize compatible edge sets as follows .
[ lem : compatible ] two valid edge sets @xmath281 and @xmath239 are compatible iff every edge @xmath44 in the symmetric difference of @xmath281 and @xmath239 is incident to @xmath183 .
for two valid compatible edge sets , @xmath284 and @xmath285 , and a sequence of equivalence relations @xmath263 , we define equivalence relations @xmath286 as follows . for every primary color @xmath191 ,
let the equivalence relation @xmath287 on @xmath288 be the transitive closure of the union of four equivalence relations : two edges in @xmath288 are related if ( 1 ) they both incident to @xmath183 ; ( 2 ) they both are in @xmath258 and @xmath259-equivalent ; ( 3 ) they are both in @xmath258 and each are equivalent to some edge in @xmath258 that are @xmath259-equivalent ; ( 4 ) one is incident to @xmath183 and the other is in @xmath258 and @xmath259-equivalent to some edge in @xmath258 incident to @xmath259 .
[ lem : part ] let @xmath281 and @xmath239 be two valid compatible edge sets , and @xmath289 .
let @xmath290 be a set of edges on @xmath291 $ ] , and put @xmath292 .
then , @xmath293 has the following property : @xmath294 if and only if 1 .
@xmath295 ; and 2 . if @xmath296 and @xmath236_c\neq \{p_i\}$ ] , then @xmath183 is incident to an edge in @xmath288 or an edge in @xmath258 that is @xmath259-equivalent to some edge incident to @xmath291_c$ ] .
[ lem : colcur ] for all @xmath297 , we have the following recurrence : @xmath298=\sum_{\{p_h , p_i\}\in e^0_i}w(\{p_h , p_i\ } ) + \min_{e_{i-1}^0\text{\rm compatible } } \textbf{a}[i-1 , e_{i-1}^0 , \widehat{\pi}_{i-1 } ] .\ ] ] [ thm : collinear - alg ] for every constant @xmath299 , min-@xmath1csg can be solved in @xmath6 time when the input points are collinear .
we determine the number of subproblems . by corollary [ cor : unique - edge ]
, every valid @xmath239 contains at most @xmath300 edges .
we have @xmath301 , since @xmath302 different colors contain any primary color @xmath83 .
the number of equivalence relations of a set of size @xmath303 is known as the @xmath303-th _ bell number _ , denoted @xmath304 .
it is known @xcite that @xmath305 .
consequently , the number of possible @xmath263 is at most @xmath306 .
the total number of subproblems is @xmath307 , which is @xmath6 for any constant @xmath1 .
we solve the subproblems * * a**@xmath268 $ ] , @xmath279 , by dynamic programming , using the recursive formula .
the time required to evaluate is @xmath308 for the sum of edge weights and @xmath309 to compare all compatible subproblems * * a**@xmath310 $ ] , that is , @xmath311 time when @xmath1 is a constant .
therefore , the dynamic programming can be implemented in @xmath6 time .
we have shown that min-3csg is np - complete in general and given a @xmath6 time algorithm for min-@xmath1csg in the special case that all points are collinear and @xmath1 is a constant .
we also improved the approximation factor of a polynomial time algorithm from @xmath312 @xcite to @xmath15 when @xmath4 .
it remains open whether there exists a ptas for min-@xmath1csg , @xmath2 .
several other special cases are open for min-3csg , such as when the points in @xmath72 are on a circle or in convex position .
we can generalize min-@xmath1csg so that the edge weights need not be euclidean distances .
given an arbitrary graph @xmath313 and a coloring @xmath314 , what is the minimum set @xmath315 such that @xmath316 is a colored spanning graph ?
since the 2-approximation algorithm presented here did not rely on the geometry of the problem , it extends to the generalization ; however , this problem may be harder to approximate than its euclidean counterpart .
* acknowledgements .
* research on this paper was supported in part by the nsf awards ccf-1422311 and ccf-1423615 .
akitaya was supported by the science without borders program .
lffler was partially supported by the netherlands organisation for scientific research ( nwo ) projects 639.021.123 and 614.001.504 . 1 b. alper , n. henry riche , g. ramos , and m. czerwinski , design study of linesets , a novel set visualization technique , _ ieee trans . vis . comput
. graphics _ , 17(12):22592267 , 2011 .
b. alsallakh , l. micallef , w. aigner , h. hauser , s. miksch , and p. rodgers , visualizing sets and set - typed data : state - of - the - art and future challenges , in _ proc .
eurographics conf .
visualization ( eurovis ) _ , pp . 121 , 2014 .
d. berend and t. tassa , improved bounds on bell numbers and on moments of sums of random variables , _ prob .
_ 30(2):185205 , 2010 .
m. de berg and a. khosravi .
optimal binary space partitions for segments in the plane .
j. comput .
_ , 22(3):187206 , 2012 .
a. biniaz , p. bose , i. van duijn , a. maheshwari , and m. smid .
a faster algorithm for the minimum red - blue - purple spanning graph problem for points on a circle . in _ proc .
28th canadian conf .
( vancouver , bc ) _ , pages 140146 , 2016 . f. chung and r.
graham , a new bound for euclidean steiner minimum trees , _ ann .
_ , 440:328346 , 1986 .
k. dinkla , m. j. van kreveld , b. speckmann , and m. a. westenberg .
kelp diagrams : point set membership visualization .
_ computer graphics forum _
, 31(3):875884 , 2012 . e. gilbert and h. pollak .
steiner minimal trees .
_ siam j. appl .
_ , 16:129 , 1968 .
f. hurtado , m. korman , m. j. van kreveld , m. lffler , v. s. adinolfi , r. i. silveira , and b. speckmann .
colored spanning graphs for set visualization .
in s. wismath and a. wolff ( eds . ) , _ proc .
21st sympos .
graph drawing _ , pp . 280291 , lncs , vol . 8242 , springer , cham , 2013 .
f. hurtado , m. korman , m. van kreveld , m. lffler , v. sacristan , a. shioura , r. i. silveira , b. speckmann , and t. tokuyama , colored spanning graphs for set visualization .
preprint , arxiv:1603.00580 , 2016 .
w. meulemans , n. henry riche , b. speckmann , b. alper , and t. dwyer .
kelpfusion : a hybrid set visualization technique .
_ ieee trans .
graphics _ , 19(11):18461858 , 2013 .
we construct a standard solution for @xmath87 based on the solution for @xmath86 .
first connect all points along each dashed line using @xmath317 length . if a variable is set to ` true ` in @xmath86 ,
connect the active points of all 2-switches of the corresponding variable gadget as shown in fig .
[ fig : switch](a ) . otherwise connect them as the reflection of fig .
[ fig : switch](a ) about the @xmath88-axis . the length of these edges sum to @xmath318 . for each positive ( resp . , negative ) clause , choose an arbitrary neighbor variable assigned ` true `
` false ` ) and connect the active points in the @xmath93-switch as is shown in fig .
[ fig : switch](d ) ( resp . , reflection of fig .
[ fig : switch](d ) ) . the length of such edges sum to @xmath319 .
connect all remaining @xmath93-switches as shown in figs .
[ fig : switch](c ) ( or its reflection ) depending on the assignment of the neighbor switch . the length of such edges sum to @xmath320 .
the resulting graph is a colored spanning graph and its total weight is @xmath321 .
by lemma [ lem : monocrhomatic ] , if @xmath87 is a positive instance of min-3csg , there exists a solution containing @xmath94 .
we consider only such solutions .
suppose that there exists an edge @xmath322 such that @xmath323 , and @xmath324 or @xmath325 is not an active point .
since only active points are multi - chromatic , @xmath44 is necessarily monochromatic , therefore , its removal can only affect the connectivity of the color @xmath104 , where @xmath326 , disconnecting the induced graph into two connected components .
since this solution includes @xmath94 , by property ( i ) there exists an edge of length @xmath91 that reconnects the components or there exists an edge @xmath327 such that @xmath328 . in both cases ,
@xmath44 can be replaced by such an edge , obtaining a lighter solution .
it remains to consider the case that @xmath324 and
@xmath325 are active points , but in different switches . in that case , @xmath329 by construction . the deletion of @xmath330 may disconnect up to three graphs induced by primary colors . for each primary color @xmath104 ,
the two components are either @xmath91 apart or there exists a switch that contains active points belonging to both components , by property ( i ) .
hence , replacing @xmath330 by at most three edges , each of length at most 2 , produces a solution of equal or smaller cost . to satisfy the rib constraint ,
assume without loss of generality that the upper switch contain a black edge .
since the switch constraints of the pair still require at least two more edges , the solution must have at least three edges .
we can enumerate all possible local solutions that satisfy the switch constraints using a total of three edges ( fig .
[ fig : switch](a ) and ( b ) ) . notice that every edge in the switch is @xmath111 or @xmath331 long ( @xmath93 or @xmath332 for @xmath93-pairs ) .
therefore , all possible solutions that use four edges have a greater cost than the ones using three edges .
then , the local solution with minimum cost must be as shown in fig .
[ fig : switch](a ) .
notice that this lower bound continues to hold in the presence of an active point of a clause .
however , if an active point of a clause is part of a switch , the local solution with minimum cost is no longer unique ; fig .
[ fig : switch](c ) shows all solutions in that case . for contradiction ,
suppose that there exists a 2-pair whose local cost is at least @xmath333 ( the second minimum is depicted in fig .
[ fig : switch](b ) ) .
then , by lemma [ lem : min ] the cost of such a solution must be at least @xmath334 $ ] which is greater than @xmath321 because @xmath335 and by the choice of @xmath91 and @xmath92 .
each clause gadget requires at least a blue path between one of its active points and the spine , or else the subgraph induced by @xmath336 is disconnected .
consider the @xmath93-pair that contains an edge in such a path .
we can enumerate all local solutions that satisfy the local constraints and contain a blue path between the clause active point and the spine and that uses 4 edges in the @xmath93-pair ( fig .
[ fig : switch](d)(g ) ) . by corollary [ cor:2-pair ]
, we consider only solutions in which @xmath111-pairs are minimally connected .
notice that removing any of the edges violates some local constraint , hence , at least 4 edges are required .
also notice that every local solution that uses 5 or more edges have cost @xmath337 or greater .
the local solutions with minimum cost are shown in figs .
[ fig : switch](d ) and ( e ) and their cost is @xmath114 .
cor : clause in a standard solution , for each clause gadget , there exists a @xmath93-pair connected as fig . [
fig : switch](d ) .
all other @xmath93-pairs are connected minimally as shown in fig .
[ fig : switch](c ) . by lemma [ lem :
clause ] , the lower bound for the cost of @xmath13 of the @xmath93-pairs is @xmath319 .
for contradiction , suppose that one of the @xmath93-pairs uses more length than its lower bound .
the second minimal configuration on the switches occurs when @xmath13 distinct @xmath93-pairs are connected using @xmath319 length , @xmath338 of the remaining @xmath93-pairs are connected using @xmath339 , and the remaining @xmath93 pair uses @xmath114 .
this construction costs at least @xmath340 which is greater than @xmath321 by the choice of @xmath91 and @xmath92 .
hence , @xmath341 of the @xmath93-pairs have to be connected as shown in fig . [
fig : switch](c ) and @xmath13 of them have to be connected as in fig . [
fig : switch](d ) and ( e ) . however , if one of these @xmath93-pairs is connected as fig . [ fig : switch](e )
, then there exists no local red path between the corresponding clause gadget and the spine .
if we assume that such a path passes through a different switch , it must be part of a minimally connected @xmath93-pair , since we can only afford any @xmath93-pair with @xmath114 cost per clause .
the only option would be to use the edges shown in fig .
[ fig : conter - expamples](d ) , which is impossible because then the graph induced by @xmath342 would be disconnected ( notice that in fig . [ fig : conter - expamples](d ) there exists a red component in the bottom that is not connected to the spine ) .
hence , all @xmath13 these @xmath93-pairs must be as shown in fig .
[ fig : switch](d ) .
we say that a 2-pair is set to ` true ` if its upper switch contains a black edge and is set to ` false ` otherwise .
first we show that every @xmath111-pair of the same variable gadget is set to the same value . for contradiction
assume that two consecutive 2-pairs are set to different truth values .
[ fig : conter - expamples ] shows all possible configurations that satisfy corollaries [ cor:2-pair ] and [ cor : clause ] . as already stated in the proof of corollary [ cor : clause ] , the local configuration in fig .
[ fig : conter - expamples](d ) is excluded .
the configurations in figs .
[ fig : conter - expamples](a ) and ( c ) lead to a red induced component on the top and bottom part of the construction , respectively , hence @xmath342 would induce a disconnected graph . the configuration in fig .
[ fig : conter - expamples](b ) would also imply that @xmath342 induces a disconnected graph , since there would not be any red induced path between the corresponding clause and the spine .
we conclude that any standard solution is similar to the one described in the proof of lemma [ lem : sat2csg ] .
then , we can easily assign a truth value from @xmath343 to every variable , obtaining a solution for the planar - monotone-3sat .
lem : valid for @xmath209 , an edge set @xmath253 between @xmath236 $ ] and @xmath238 $ ] is _ valid _ iff for every primary color @xmath191 , there is an edge @xmath255 such that @xmath256 whenever both @xmath236_c$ ] and @xmath238_c$ ] are nonempty .
let @xmath31 be a csg with property ( @xmath189 ) .
then for every @xmath45 , the graph @xmath41 is connected .
if both @xmath236_c$ ] and @xmath238_c$ ] are nonempty , there is an edge @xmath344 , hence @xmath256 .
conversely , assume every edge in @xmath253 has property ( @xmath189 ) . let @xmath235 ( resp . ,
@xmath237 ) be the set of all edges on @xmath236 $ ] ( resp . , @xmath238 $ ] ) with property ( @xmath189 ) . by lemma [ lem : unique - color ] , @xmath261_c,(e_i^-)_c)$ ] and @xmath250)c,(e_i^+)_c)$ ] have the same components as the compete graph in each primary color @xmath24 .
then @xmath258 contains at least one edge between @xmath236_c$ ] and @xmath238_c$ ] if both are nonempty .
consequently , @xmath345 is a csg for @xmath72 , as required .
lem : halves let @xmath31 be a csg , @xmath264 , and @xmath241 an edge set on @xmath236 $ ] .
the graph @xmath248 is a csg iff the graph @xmath261_c,(x_i^-)_c\cup e(\pi_i^c))$ ] is connected for every @xmath265 .
let @xmath265 .
we claim that @xmath346 is connected iff @xmath261_c,(e_i^-)_c\cup e(\pi_i^c))$ ] is connected . if @xmath236_c=\emptyset$ ] , then @xmath347_c,(x_i^-)_c)$ ] . if @xmath238_c=\emptyset$ ] , then @xmath348_c,(e_i^+)_c)$ ] . in both cases ,
the claim follows .
assume that neither @xmath236_c$ ] nor @xmath238_c$ ] is empty .
since @xmath197 is a csg , then @xmath41 is connected , and so each component of @xmath250_c,(e_i^+)_c)$ ] is incident to some edge in @xmath258 that has an endpoint @xmath236_c$ ] .
consequently , each component of @xmath346 has a vertex in @xmath236_c$ ] .
therefore @xmath346 is connected iff it contains a path between any two vertices of @xmath236_c$ ] .
two vertices in @xmath236_c$ ] are connected by a path in @xmath349 iff they are adjacent along an edge in @xmath350 .
consequently , there is a path between any two vertices in @xmath236_c$ ] iff they are in the same component of @xmath261_c,(e_i^-)_c\cup e(\pi_i^c))$ ] , as required .
assume that @xmath281 is compatible with @xmath239 , witnessed by some @xmath269 .
property ( @xmath189 ) implies that the left endpoint of an edge @xmath351 is @xmath183 iff @xmath352 .
similarly , the right endpoint of an edge in @xmath281 is @xmath183 iff @xmath353 .
conversely , assume every edge @xmath44 in the symmetric difference of @xmath281 and @xmath239 is a incident to @xmath183 .
let @xmath290 be the set of all edges on @xmath354 $ ] with property ( @xmath189 ) ; and let @xmath355 .
define @xmath263 such that all members of @xmath258 are @xmath259 equivalent for every @xmath191 . by lemma [ lem : unique - color ] , the graph @xmath356_c,(e_{i-1}^-)_c)$ ]
is connected for all primary colors @xmath33 .
for every color @xmath357 , we have @xmath236_c = s[1,i-1]_c$ ] and so @xmath261_c,(x_i^-)_c\cup e(\pi_i^c))$ ] is connected .
let @xmath296 . if @xmath236_c=\{p_i\}$ ] , then @xmath261_c,\emptyset)$ ] is clearly connected .
otherwise , @xmath183 is incident to an edge in @xmath288 or @xmath258 , by lemma [ lem : valid ] .
if @xmath183 is incident to an edge in @xmath288 , then it is adjacent a vertex in @xmath291_c$ ] .
otherwise @xmath183 is incident to some @xmath358 , consequently @xmath238_c\neq\emptyset$ ] . by lemma [ lem : valid ]
, there is some @xmath359 between @xmath291_c$ ] and @xmath238_c$ ] . by construction , @xmath229 and @xmath230
are @xmath259-equivalent , and so @xmath261_c,(x_i^-)_c\cup e(\pi_i^c))$ ] is connected , as required .
lem : part let @xmath281 and @xmath239 be two valid compatible edge sets , and @xmath289 .
let @xmath290 be a set of edges on @xmath291 $ ] , and put @xmath292 .
then , @xmath293 has the following property : @xmath294 if and only if 1 .
@xmath295 ; and 2 . if @xmath296 and @xmath236_c\neq \{p_i\}$ ] , then @xmath183 is incident to an edge in @xmath288 or an edge in @xmath258 that is @xmath259-equivalent to some edge incident to @xmath291_c$ ] .
assume @xmath294 .
then the graph @xmath261_c,(e_i^-)_c\cup e(\pi_i^c))$ ] is connected for every @xmath191 , however , its induced subgraph on @xmath291_c$ ] may have two or more components . between any two such components ,
there is a path in @xmath360 , where @xmath361 is the set of edges whose right endpoint is @xmath183 .
the first and last edge of any such path is @xmath287-equivalent by the definition of @xmath362 .
consequently , @xmath295 , and ( * d1 * ) follows . for every @xmath296 , if @xmath291_c\neq \{p_{i-1}\}$ ] , then @xmath183 is incident to some edge in @xmath363 by ( * d1 * ) , which implies ( * d2 * ) .
conversely , assume ( * d1 * ) and ( * d2 * ) .
since @xmath364 , then the graph @xmath261_c,(e_{i-1}^-)_c\cup e(\hat{\pi}_{i-1}^c))$ ] is connected for every @xmath191 .
that is , there is a path in @xmath365 between any two vertices in @xmath291_c$ ] .
if we replace the edges in @xmath366 with a sequence of edges thate in @xmath367 or incident to @xmath183 ( using the definition of @xmath287 ) , there is a path in @xmath368 between any two vertices in @xmath291_c$ ] . finally , for every @xmath296 , we need to show that there is a path in @xmath368 between @xmath183 and any other vertex of @xmath236_c$ ] .
this is vacuously true if @xmath183 is the only vertex of @xmath236_c$ ] .
suppose that @xmath236_c\neq \{p_i\}$ ] . by ( * d2 * )
, @xmath183 is incident to an edge in @xmath288 or an edge in @xmath350 .
consequently , @xmath294 , as required .
lem : colcur for all @xmath297 , we have the following recurrence : @xmath369=\sum_{\{p_h , p_i\}\in e^0_i}w(\{p_h , p_i\ } ) + \min_{e_{i-1}^0\text{compatible } } \textbf{a}[i-1 , e_{i-1}^0 , \widehat{\pi}_{i-1 } ] .\ ] ] | we study a problem proposed by hurtado et al .
@xcite motivated by sparse set visualization . given @xmath0 points in the plane , each labeled with one or more primary colors , a _ colored spanning graph _ ( csg ) is a graph such that for each primary color , the vertices of that color induce a connected subgraph .
the min - csg problem asks for the minimum sum of edge lengths in a colored spanning graph .
we show that the problem is np - hard for @xmath1 primary colors when @xmath2 and provide a @xmath3-approximation algorithm for @xmath4 that runs in polynomial time , where @xmath5 is the steiner ratio .
further , we give a @xmath6 time algorithm in the special case that the input points are collinear and @xmath1 is constant . |
the quenching of high transverse momentum ( @xmath1 ) particles in central heavy - ion collisions is one of the main discoveries at bnl - rhic .
the effect is quantified using the nuclear modification factor : @xmath9 where @xmath10 is the average number of nucleon nucleon ( nn ) collisions for nucleus nucleus ( aa ) collisions in a given centrality class c. the nuclear modification factor would be equal to unity if the aa collision were a mere superposition of @xmath11 independent nucleon nucleon collisions . in central collisions at a c.m.s .
energy per nucleon
nucleon pair of @xmath6 , the phenix and star experiments have measured a suppression by a factor 45 with respect to unity in @xmath12 for @xmath13 , independent of the particle species @xcite . a similar suppression at high @xmath1
is observed also in the central - to - peripheral nuclear modification factor @xmath14 where a class p of peripheral nucleus nucleus collisions replaces the pp reference ( see e.g. @xcite ) .
the measured suppression is interpreted as being due to energy loss of the hard partons traversing the high - density qcd medium expected to be formed in high - energy heavy - ion collisions @xcite .
parton energy loss would predominantly occur via the mechanism of medium - induced gluon radiation @xcite .
preliminary results @xcite from the rhic run at @xmath15 show a @xmath12 suppression by about a factor 3 , not much smaller than that observed at @xmath6 .
this motivates the search for parton energy loss effects at even lower c.m.s .
energy , i.e. in collisions with @xmath16 at cern - sps .
possible indications of such effects were found by the wa98 collaboration @xcite , who observed a suppression of the @xmath17 @xmath0 . in addition , a recent @xmath17 @xmath12 analysis @xcite , using the wa98 pb pb data @xcite and a parameterization of a wide compilation of pp data at similar energy available in the literature , favours a scenario of significant energy loss .
we investigate the presence of energy loss effects in collisions at @xmath16 by measuring @xmath5 for @xmath2 , @xmath3 and @xmath4 particles , and for unidentified negatively charged particles , @xmath18 .
the relative behaviour of the @xmath5 patterns for @xmath2 and @xmath3 particles is also expected to be sensitive to parton coalescence effects @xcite in the hadronization dynamics .
it has been suggested that these effects occur at rhic energy ( see e.g. @xcite ) . in section [ analysis ] the experimental setup ,
the data sets and the analysis procedures are described .
the @xmath0 results are presented in section [ results ] and they are compared in section [ discussion ] to other experimental results at sps and rhic energies as well as with theoretical calculations .
the na57 apparatus , described in detail in @xcite , was designed to study the production of strange and multi - strange particles in fixed - target heavy - ion collisions by reconstructing their weak decays into final states containing charged particles only .
tracks are reconstructed in the 30 cm - long silicon telescope : an array of pixel detector planes with a cross section of @xmath19 placed inside a 1.4 t magnetic field normal to the beam direction .
the telescope is inclined in the non - bending plane by a 40 mrad angle relative to the beam line and it points to the target located 60 cm from the first detector plane .
the acceptance covers about half a unit in rapidity at central rapidity and transverse momentum larger than about @xmath20 . for the runs , the centrality trigger , based on charged multiplicity ,
was set so as to select approximately the most central 60% of the inelastic collisions .
the results presented in this letter are based on the analysis of data collected during the 1998 ( about @xmath21 events ) and 2000 ( about @xmath22 events ) runs with beam momentum of 158 @xmath23 .
.average number of participants and of nn collisions with their systematic errors , as a function of centrality .
[ cols="^,^,^",options="header " , ] the collision centrality is determined using the charged particle multiplicity @xmath24 in the pseudorapidity range @xmath25 , sampled by the microstrip silicon detectors ( msd ) as described in @xcite .
@xmath24 is related to the centrality assuming @xmath26 ( a modified wounded nucleon model ) @xcite , where @xmath27 is the number of participants , i.e. nucleons participating in the primary nucleon
nucleon collisions , estimated from the glauber model @xcite .
the @xmath24 distribution is well described by a wounded nucleon model fit with @xmath28 @xcite .
the inelastic cross section extracted from the fit @xcite , using an inelastic non - diffractive nn cross section @xmath29 @xcite , is @xmath30 b. five centrality classes are defined , with @xmath24 limits corresponding to given fractions of @xmath31 . for each class the average values of @xmath27 and @xmath11
are calculated from the glauber model , with pb nucleus woods - saxon density - profile parameters as given in @xcite . in table
[ tab : centrality ] we present the definitions of the five centrality classes along with the corresponding values of @xmath32 and @xmath33 with their systematic errors , estimated by varying the @xmath34 parameter in the range [ 1.0,1.1 ] , the woods - saxon parameters within their tabulated uncertainties and the inelastic non - diffractive nn cross section at @xmath35 within its systematic error estimated to be 1.5% @xcite values given in table [ tab : centrality ] for the most central and the most peripheral classes differ slightly from those published in @xcite due to a refined treatment of the trigger effect in the fit of the multiplicity distribution done for the present analysis . ] .
note , however , that the contribution due to the uncertainty on @xmath36 is negligible for @xmath32 , while for @xmath33 it ranges from 1.5% in the most central class to 1.1% in the most peripheral , but it cancels out in the ratio @xmath37 used for @xmath0 .
( upper row ) and @xmath38 ( lower row ) candidates in different @xmath1 ranges .
arrows indicate the nominal masses of the two particles and shaded areas indicate the ranges considered for the analysis . no centrality selection is applied . ]
strange particles are reconstructed using their decay channels into charged particles : @xmath39 , @xmath40 , and @xmath41 .
the selection procedure is described in detail in @xcite .
the main criteria are the following : ( a ) the two decay tracks are compatible with the hypothesis of having a common origin point ; ( b ) the reconstructed secondary vertex is well separated in space from the target ; ( c ) the reconstructed candidate points back to the primary vertex position . for @xmath2 ,
the two decay tracks are required to miss the interaction vertex by applying a @xmath1-dependent cut on the product of their impact parameters ( distances of closest approach to the interaction vertex in the bending plane ) .
ambiguities among @xmath2 , @xmath3 and @xmath4 are eliminated by means of cuts @xcite on the podolanski - armenteros plot @xcite .
the final invariant mass distributions for @xmath2 and @xmath3 candidates in different @xmath1 ranges are shown in fig .
[ fig : masses ] .
the shaded areas correspond to the windows used for this analysis ; the size of these windows is smaller at low @xmath1 , where the invariant mass resolution is better .
the statistics amounts to @xmath42 @xmath2 , @xmath43 @xmath3 , and @xmath44 @xmath4 .
we have estimated and subtracted the combinatorial background using the event - mixing technique on a subsample of events representative of the full statistics @xcite .
the background fraction in the selected mass window is negligible for our most peripheral class ( 4055% ) and it increases going to more central classes , where it also increases from low to high transverse momentum . in the worst case , in class 05% ,
the subtracted background fraction amounts to : @xmath45 for @xmath2 with @xmath46 ; @xmath47 for @xmath3 with @xmath46 ; @xmath48 for @xmath4 with @xmath49 .
negatively charged particles , @xmath18 , are selected from a sample of good - quality tracks ( clusters in more than 80% of the telescope planes , less than 30% of the clusters shared with other tracks ) using an impact parameter cut to ensure they come from the interaction vertex .
the residual contamination of secondary tracks ( decay products of weakly - decaying particles ) has been estimated to be of about 2% , independent of @xmath1 , for the most central class of events , and lower for the other classes .
the statistics amounts to @xmath50 particles .
we calculate @xmath5 , eq . ( [ eq : rcp ] ) , using @xmath1 distributions which are unweighted , i.e.not corrected for geometrical acceptance and reconstruction / selection efficiency . for other analyses ,
in particular for the measurement of the @xmath1-integrated strange particle production yields @xcite , we adopted a procedure , specifically developed for rarer signals like @xmath51 and @xmath52 , in which every selected particle was assigned a correction weight , calculated on the basis of a monte carlo simulation @xcite .
this correction is very time consuming and , for the more abundant signals ( @xmath18 , @xmath2 and @xmath3 ) , it was calculated only for a representative subsample of the available statistics . for the present study , we have verified that these weights do not depend on the event centrality over the full transverse momentum range covered ( see fig .
[ fig : weights ] ) . using unweighted spectra results in a negligible systematic error on @xmath5 as compared to the other contributions discussed in the following .
we use class 4055% as the reference peripheral class in the denominator of @xmath0 , see eq .
( [ eq : rcp ] ) , and vary the ` central ' class in the numerator from 05% to 3040% .
we estimated possible residual systematic effects by comparing the @xmath0 results corresponding to the year 1998 sample and to the year 2000 sample , which were also processed with different versions of the reconstruction software .
we find compatible results , with point - to - point differences smaller than 10% for @xmath18 , @xmath2 and @xmath53 .
we assign ` reconstruction ' systematic errors of 5% for all four species . for strange particles ,
a small contribution to the systematic errors has been recognized as due to the procedure used to remove fake tracks , which yield duplicated candidates , particularly in central collisions that have higher track density analysis , fake tracks were avoided by imposing tighter track - quality conditions . ] . for @xmath2 , @xmath3 and @xmath4 , the resulting error , estimated to be at most 3% at high @xmath1 for the ratio 05%/4055% and smaller for the other ratios , has been added in quadrature to the ` reconstruction ' systematic error and to the systematic error due to background subtraction , which is also at most 3% .
figure [ fig : rcp ] shows our @xmath5 results for four different ` central ' classes ( c ) .
the error bars are obtained as a quadratic sum of the statistical errors and the @xmath1-dependent systematic errors ( @xmath54 ) .
the shaded bands centered at @xmath55 represent the @xmath1-independent systematic error due to the uncertainty in the ratio @xmath37 , while the shaded bands at low @xmath1 represent the @xmath0 values corresponding to @xmath27-scaling , with the band indicating the systematic error due to the uncertainty in the ratio @xmath56 .
we first focus on the 05%/4055% ratio . at low @xmath1 ( @xmath57 ) ,
the @xmath18 , @xmath2 and @xmath3 patterns are compatible with @xmath27-scaling , while the @xmath4 points are clearly below . as @xmath1 increases , @xmath0 for the @xmath2 approaches one ; @xmath0 for the @xmath3 behaves differently from the @xmath2 above @xmath58 , reaching a value of about 1.5 , as does the @xmath4 . for @xmath1 values below @xmath59 , the @xmath0 for negative particles
is dominated by negative pions , and stays below the corresponding values for @xmath2 . at higher @xmath1
the @xmath18 @xmath0 lies between those for @xmath2 and @xmath3 ; in this region there may be significant @xmath60 and @xmath61 contributions in the sample of negative particles ( note that @xmath62 for @xmath63 in central collisions at @xmath6 @xcite ) . for less central collisions , the @xmath2
@xmath0 exhibit a small enhancement in the range @xmath64 , while within errors the @xmath0 for the other particles do not vary with respect to 05%/4055% @xmath0 . for @xmath18 , @xmath2 , @xmath3 and @xmath4 in collisions at @xmath16 .
shaded bands centered at @xmath55 represent the systematic error due to the uncertainty in the ratio of the values of @xmath33 in each class ; shaded bands at low @xmath1 represent the values expected for scaling with the number of participants , together with their systematic error . ]
the difference between the @xmath3 and the @xmath4 at low @xmath1 , already observed in the centrality dependence of the @xmath1-integrated yields in collisions @xcite , is pronounced . on the basis of the symmetry of the apparatus and of the signal extraction procedure , we are confident that the effect is not caused by an experimental bias . it may be due to a centrality - dependent absorption effect of @xmath4 in a nucleon - rich environment . in this respect , note that at sps energy the @xmath65 ratio is measured to be lower by a factor about 2 in with respect to sulfur - induced collisions @xcite and the @xmath4 yield per participant is measured to be lower by a factor @xmath66 in @xcite with respect to collisions @xcite .
for @xmath18 and @xmath2 from na57 and @xmath17 from wa98 @xcite in pb
pb collisions at @xmath16 .
right : @xmath5 for @xmath2 and @xmath38 in pb pb at @xmath16 ( na57 ) and in au au at @xmath6 ( star ) @xcite ; slightly different peripheral classes are employed for the comparison .
the bars centered at @xmath55 represent the normalization errors ; the point - by - point bars are the quadratic sum of statistical and systematic errors . ] in fig .
[ fig : cmpexp ] we compare our results to @xmath0 measurements at the sps and at rhic . in the left - hand panel ,
the wa98 @xmath17 data @xcite for the ratio 16%/2243% in collisions at @xmath16 are plotted together with the na57 @xmath18 and @xmath2 data for the same centrality classes . here
, we rename the classes in terms of percentiles of the inelastic cross section , @xmath31 .
the values used by the two experiments for the inelastic cross section are very similar : 7.26 b for na57 and 7.41 b for wa98 . ] . using these classes , the @xmath2 @xmath0 is approximately constant at 0.9 for @xmath67 and is significantly larger than that measured by the wa98 collaboration for @xmath17 ( @xmath68 ) , even when taking into account the normalization systematic errors , independent for the two experiments .
the @xmath18 data from na57 are compatible , within the systematic errors , with the @xmath17 data from wa98 for @xmath69 , where the @xmath18 sample is expected to be dominated by @xmath70 . for higher @xmath1 ,
@xmath18 have a larger @xmath0 than @xmath17 ; this may be due to increasing contributions from @xmath60 and @xmath71 in the @xmath18 sample . at top rhic energy , @xmath6 , the kaon @xmath0 ( as measured by phenix @xcite and star @xcite ) is larger than that of neutral pions ( phenix @xcite ) for @xmath72 , while they are similar for higher transverse momenta .
the observed difference in @xmath0 between kaons and pions at sps energy , and at rhic energy for low @xmath1 , is reminiscent of the ` cronin enhancement ' above @xmath11-scaling originally observed for @xmath73 in proton nucleus ( pa ) collisions at @xmath74 values up to @xmath75 @xcite .
this enhancement , commonly interpreted as due to initial - state multiple scattering ( partonic intrinsic transverse momentum broadening ) , was in fact found to increase according to the hierarchy pions kaons protons @xcite . at rhic energy , owing to the hardening of the @xmath1 distributions , the effect would be reduced compared to lower energies , as confirmed by preliminary results on the particle - species dependence of the ( similar to proton nucleus ) @xmath0 at @xmath6 @xcite .
the comparison for @xmath2 and @xmath3 at sps and rhic ( star data for at @xmath6 @xcite ) is presented in the right - hand panel of fig .
[ fig : cmpexp ] . in the @xmath1 range covered by our data , up to @xmath8 , the _ relative _ pattern for @xmath2 and @xmath3 is similar at the two energies , while absolute values are higher at sps than at rhic , where parton energy loss is believed to have a strong effect .
part of the difference between the @xmath0 values at the two energies may be due to different nuclear modification of the parton distribution functions ( pdfs ) .
in fact , in the @xmath76 range relevant for parton production at a given @xmath1 at the sps , e.g.@xmath77 for @xmath78 , nuclear pdfs are expected to be enhanced by about 1020% ( anti - shadowing ) , while almost no effect is expected for the smaller values , @xmath79 , relevant for the same @xmath1 values at rhic energy ( see e.g.@xcite ) . at rhic ,
the larger @xmath0 for @xmath3 with respect to kaons @xcite or , more generally , for baryons with respect to mesons @xcite , in the intermediate @xmath1 range , @xmath80@xmath8 , has been interpreted as due to parton coalescence @xcite in a high - density medium with partonic degrees of freedom .
our data show that a similar @xmath3@xmath81 pattern is present also at @xmath16 .
we note that such a pattern may also be explained in terms of larger cronin effect for @xmath3 with respect to kaons . in fig .
[ fig : cmpth ] we compare our @xmath2 data to predictions provided by x.n .
wang @xcite , obtained from a perturbative - qcd - based calculation @xcite , including ( thick line ) or excluding ( thin line ) in - medium parton energy loss .
the initial gluonic rapidity density of the medium , @xmath82 , was scaled down from that needed to describe rhic data , according to the decrease by about a factor 2 in the charged multiplicity per unit of rapidity from rhic to sps energy . for the 05%/4055% @xmath0 ,
the curve without energy loss shows a large enhancement , increasing with @xmath1 . in the calculation
this enhancement arises principally from initial - state partonic intrinsic transverse momentum broadening , which is assumed to be proportional to the number of scatterings that the two colliding partons suffer inside the nuclei before the hard scattering , and thus larger for central than for peripheral collisions @xcite .
the magnitude of the broadening is tuned on the basis of the original cronin effect data @xcite .
a ` cronin - like ' enhancement is observed in the central - to - peripheral nuclear modification factor for charged kaons at rhic energy @xcite , while it is not present in our nucleus
nucleus @xmath2 data .
the curve that includes energy loss , scaled down from rhic as explained above , describes the data better . moving to less central collisions the predictions with and without energy loss get closer to one another , and both
are compatible with the data . as a cross - check
, we compare the @xmath0 value predicted by x.n .
wang with energy loss to the prediction of an independent model of parton energy loss , the parton quenching model ( pqm ) , based on the bdmps formalism @xcite , that describes several energy - loss - related observables at rhic energies @xcite .
since the pqm model does not include initial - state effects , the predicted @xmath0 was rescaled using the wang baseline without energy loss ( thin line ) .
the pqm result has an uncertainty due to the fact that the medium - induced energy loss becomes of the order of the initial parton energy @xcite ; since this uncertainty is larger for low - energy partons , the calculation result is meaningful for @xmath83 only .
the pqm result , shown in fig . [ fig : cmpth ] by the hatched areas , agrees with the wang calculation with energy loss ( thick line ) .
the two models predict a similar energy loss effect at sps energy , i.e.a reduction of the 05%/4055% @xmath0 ( for @xmath83 ) by about a factor 2 , with respect to the value calculated without energy loss . for @xmath2 in collisions at @xmath16 ,
compared to predictions @xcite with and without the effect of parton energy loss ( details in the text ) . ]
central - to - peripheral nuclear modification factors for @xmath2 , @xmath3 , @xmath4 and @xmath18 in collisions at top sps energy have been measured as a function of @xmath1 up to about @xmath8 . at low @xmath1
, @xmath0 agrees with @xmath27 scaling for all the particles under consideration , except the @xmath4 , for which the yields at low @xmath1 are found to increase slower than the number of participants . for @xmath84 , @xmath2 , @xmath3 and @xmath4
show a pattern similar to that observed in collisions at top rhic energy , although the @xmath0 values are found to be larger at sps . at rhic , this pattern has been interpreted in the framework of models that combine parton energy loss with hadronization via coalescence , at intermediate @xmath1 , and via fragmentation , at higher @xmath1 .
the measured @xmath2 05%/4055% @xmath0 is not reproduced by a theoretical calculation that includes only initial - state nuclear effects .
the data can be better described by including final - state parton energy loss as predicted for sps energy on the basis of rhic data .
voloshin , nucl .
phys . a 715 ( 2003 ) 379 ; r.c .
hwa and c.b .
yang , phys .
c 67 ( 2003 ) 064902 ; r.j .
fries , b. muller , c. nonaka and s.a .
bass , phys .
c 68 ( 2003 ) 044902 ; v. greco , c.m .
ko and p. levai , phys .
c 68 ( 2003 ) 034904 . | we present central - to - peripheral nuclear modification factors , @xmath0 , for the @xmath1 distributions of @xmath2 , @xmath3 , @xmath4 , and negatively charged particles , measured at central rapidity in collisions at top sps energy .
the data cover the 55% most central fraction of the inelastic cross section .
the @xmath2 and @xmath3 @xmath5 are similar in shape to those measured at @xmath6 at rhic , though they are larger in absolute value .
we have compared our @xmath2 @xmath0 data to a theoretical calculation .
the prediction overestimates the data at @xmath7@xmath8 , unless sizeable parton energy loss is included in the calculation .
na57 collaboration , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , and nucleus nucleus collisions , strange particles 12.38.mh , 25.75.nq , 25.75.dw |
@xmath0 annihilation into hadrons is one of the most important suppliers of experimental information on the quark interactions . at high energy it serves as a test of perturbative qcd whereas at low energies it provides insight into nonperturbative effects in qcd as well as valuable input to various phenomenological models describing strongly interacting particles .
it became conventional to use the dimensionless quantity @xmath1 to characterize the total cross section of @xmath2 hadrons : [ eq : r ] r=(e^+e^-hadrons)/ ( e^+e^-^+^- ) .
@xmath1 is widely used for various calculations .
particularly , knowledge of @xmath1 with high accuracy is required for the evaluation of @xmath3 , the leading order hadronic contribution to the anomalous magnetic moment of the muon @xmath4 ( see @xcite and references therein ) or the value of the fine structure constant at the @xmath5 boson mass @xcite .
@xmath6 known today to better than 1.0 ppm is one of the best measured quantities in physics .
the recently reported measurement of @xmath6 by the e821 collaboration at bnl @xcite and a possible deviation of its result from the predictions of standard model ( sm ) @xcite has generated numerous speculations ( for a review and discussion see @xcite ) .
+ within sm , the uncertainty of the theoretical value of the leading order @xmath6 is dominated by the uncertainty of the hadronic contribution that can be calculated via the dispersion integrals .
[ eq : dint ] a_^had , lo = ( ) ^2 _ 4m_^2^ ds , where the qed kernel @xmath7 is a smooth function of energy varying from 0.63 at @xmath8 to 1 at @xmath9 .
the precision of the @xmath3 calculation depends on the approach used and varies from 1.34 ppm based on @xmath0 data only @xcite to 0.53 ppm if in addition @xmath10-lepton decay data as well as perturbative qcd and qcd sum rules are extensively used @xcite . as it is clear from eq .
[ eq : dint ] , the major contribution to its uncertainty comes from the systematic error of the @xmath1 measurement at low energies ( @xmath11 2 gev@xmath12 ) , which , in turn , is dominated by the systematic error of the measured cross section @xmath13 or pion form factor @xmath14 directly related to it . assuming conservation of the vector current ( cvc ) and isospin symmetry , the spectral function of the @xmath15 decay , where x@xmath16 is a vector hadronic state with i=1 can be related to the corresponding isovector state x@xmath17 produced in @xmath0 annihilation @xcite .
( here x can be @xmath18 ) .
the detailed measurement of the spectral functions was provided by aleph @xcite , opal @xcite and cleo - ii @xcite .
comparison of the hadronic cross sections measured at @xmath0 colliders with the spectral functions of the corresponding @xmath19 decays provides a test of cvc .
if cvc holds with high accuracy , @xmath10-lepton decay data can be also used to substantially improve the accuracy of the calculations mentioned above @xcite .
recent indications that the accuracy of cvc relations is not as good as believed for many years necessitates a careful reanalysis of such estimations @xcite .
thus , new high precision measurements of the cross section of @xmath20 and particularly of the pion form factor as well as precise determinations of the hadronic mass spectra in @xmath10 lepton decays become extremely important .
since 1974 the @xmath0 collider vepp-2 m has been successfully running in the budker institute of nuclear physics in novosibirsk in the c.m.energy range from the threshold of hadron production to 1400 mev @xcite .
its maximum luminosity reached @xmath21 @xmath22s@xmath23 at the @xmath24 meson energy . in the last series of experiments two detectors ( cmd-2 and snd ) installed at vepp-2 m collected about 30 pb@xmath23 each .
cmd-2 described in detail elsewhere @xcite is a general purpose detector . inside a superconducting solenoid with a field of 1 t
there are a drift chamber and proportional z - chamber , both also used for the trigger , and an endcap bgo electromagnetic calorimeter .
outside there is a barrel csi electromagnetic calorimeter and muon streamer tube chambers .
the main goal of cmd-2 is to perform a high precision measurement of the exclusive cross sections of various hadronic channels of @xmath0 annihilation and detailed studies of the low lying vector mesons - @xmath25 and @xmath24 .
snd described in detail elsewhere @xcite is a nonmagnetic detector with drift chambers for tracking and a three layer nai electromagnetic calorimeter .
outside it there are a muon streamer tube chamber and plastic scintillators .
the main goal of snd is to study @xmath26 , @xmath27 and @xmath24 decays as well as main hadronic channels .
both experiments possess some special features making high precision measurements feasible : * large data samples due to the high integrated luminosity and large acceptance * multiple scans of the same energy ranges to avoid possible systematic effects ; the step was 10 mev in the c.m.energy for the continuum region and 1 - 2 mev near the @xmath27 and @xmath24 peaks * the absolute calibration of the beam energy using the resonance depolarization method @xcite reduces to a negligible level a systematic error caused by an uncertainty in the energy measurement which can be significant for cross sections with strong energy dependence * good space and energy resolution lead to small background * redundancy - unstable particles are independently detected via different decay modes ( @xmath28 ; @xmath29 ) * detection efficiencies and calorimeter response are studied by using `` pure '' experimental data samples rather than monte carlo events , e.g. more than 20 million @xmath24 meson decays per detector can be used for that purpose .
new results are available on most of the hadronic channels .
the process @xmath13 is particularly important for various applications because of its large cross section at low energies .
this reaction has been extensively studied before @xcite .
the most precise pion form factor data were obtained in late 70s
early 80s by cmd and olya detectors @xcite .
their accuracy was limited by systematic errors of the experiments , varying from 2% to 15% over the vepp-2 m energy range . in the new measurement with the cmd-2 detector
more than 2 million events of the process @xmath13 were detected from 370 to 1380 mev .
below 600 mev separation of bhabha and @xmath30 events is performed by measuring their momentum . above this energy
the energy deposition of the final particles in the calorimeter has been used .
the number of events of the reaction @xmath31 was evaluated from qed which validity at these energies had been verified before .
the systematic uncertainty of less than 0.6% was achieved in the final analysis of the data set of about 114000 events collected in the energy range 610 to 960 mev in 1994 - 1995 @xcite .
table [ tab : syst ] lists the dominant sources of the systematic error .
analysis is in progress for the rest of events and the expected systematic error ranges from 1% to 3% @xcite . fig .
[ fig : pipi ] shows results of the pion form factor measurement coming from cmd-2 .
.main sources of systematic errors .
[ tab : syst ] [ cols="<,^",options="header " , ] since e821 collaboration plans to improve the accuracy , it is clear that further progress is also needed from the theoretical side . at the present time
the analysis of the @xmath32 data as well as other hadronic final states in the whole energy range accessible to cmd-2 is in progress .
independent information is also expected in close future from other experiments studying low energy @xmath0 annihilation @xcite . when all the above mentioned data are taken into account
, one can expect further improvement of the overall error of @xmath3 compared to the current one .
it is very tempting to use an independent set of the @xmath10 lepton data to improve the existing evaluations of the hadronic effects .
however , as we learned from the talk of a. hcker at this workshop @xcite , recent analysis revealed the apparent incompatibility of @xmath0 and @xmath10 lepton spectral functions @xcite . as a result ,
the @xmath10 lepton based evaluation and the @xmath0 based one are hardly consistent , so that their averaging does nt improve the accuracy .
more work is needed to clarify the reasons of this effect .
thus , new experiments in novosibirsk and beijing considerably improved the accuracy of @xmath1 in the energy ranges below 1.38 gev and between 2 and 5 gev allowing significant improvement of the uncertainty of @xmath3 .
precise tests of the relation between the @xmath0 cross sections and @xmath10 branching ratios will require better understanding of the isospin symmetry breaking effects and radiative corrections .
together with the more detailed analysis of systematic effects in both @xmath0 and @xmath10 lepton sectors , it should allow to perform new precise evaluations of the hadronic corrections .
further significant progress will become possible after new experiments planned at beijing , cornell and novosibirsk .
also promising looks a possibility to study low energy @xmath0 annihilation by the radiative return from the @xmath33 or @xmath24 mesons @xcite .
the author is grateful to a. seiden and his colleagues from the university of california , santa cruz for an opportunity to present this talk and for the excellent organization of the workshop .
special thanks are due to a.e .
bondar , a. czarnecki , m. davier , g.v .
fedotovich , a. hcker , f. jegerlehner , p.p .
krokovny , l.m .
kurdadze , w.j .
marciano , a.i .
milstein , and a.i .
vainshtein for numerous stimulating discussions . | new results on the low energy @xmath0 annihilation into hadrons from novosibirsk and beijing are described .
implications of the new measurements for the evaluation of the hadronic contribution to the muon anomalous magnetic moment are discussed . |
solar models generated with the @xcite or @xcite abundances agree quite nicely with helioseismic inferences of the sound speed , the location of the base of the convection zone , and the helium abundance in the convection zone .
however , @xcite , @xcite , and @xcite revised the solar abundances using three - dimensional hydrodynamic models of the atmosphere with improved input physics and non - local thermodynamic equilibrium effects , lowering abundances by up to 1/3 .
solar models that use the new lower abundances yield worse agreement than those that use the older abundances .
this has led to years of heated debates over whether the disagreement represents a solar model problem or a solar abundance problem .
the ongoing disagreement has inspired a re - examination of all aspects of solar models , including opacities , diffusive settling , convective overshooting , and the possible accretion of low-_z _ material late in the sun s evolution or mass loss early in the sun s evolution ( see * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* for reviews of mitigation attempts ) .
although these proposed adjustments have led to some improvement in the agreement , they do not satisfactorily resolve the issue .
further investigations into the basic physics of the solar interior are required .
nuclear reactions generate the energy that drives our sun . developing an accurate picture of the conditions that lead to nuclear reactions is essential in order to fully understand the inner workings of the sun . with that in mind , we re - examine screening effects in the solar core . in this work ,
we focus on _ p__p _ reactions which produce most of the nuclear energy generated in the sun . in this section ,
we derive an expression for calculating nuclear reaction rates following the treatment of @xcite .
the reaction rate per unit volume between particles of types @xmath0 and @xmath1 is a product of the number densities of the particles , @xmath2 and @xmath3 , and the average value of the product of the relative velocity @xmath4 times the cross section @xmath5 , @xmath6 where @xmath7 is the relative velocity distribution and @xmath8 accounts for reactions of like particles . because we will be dealing with the correction due to dynamic screening ( a ratio between the unscreened and screened reaction rates )
, we can ignore the density factor and focus on the reaction rate per pair of particles , @xmath9 where @xmath10 @xmath11 is the reduced mass of the pair , and the maxwell boltzmann distribution is used for @xmath7 .
the cross section @xmath12 can be defined as a product of three separate energy - dependent factors @xmath13 where @xmath14 , with @xmath15 and @xmath16 being the charges of the interacting ions and @xmath17 is the reduced atomic weight .
the exponential factor in this expression comes from the barrier penetration probability , the inverse energy dependence comes from the quantum mechanical interaction between the two particles , and @xmath18 contains the intrinsically nuclear parts of the probability for a nuclear reaction to occur . with this substitution for @xmath12 , equation [ eq : lambda1 ]
can be re - written as @xmath19 in the non - resonant reaction case , @xmath18 is slowly varying with @xmath20 , so we can treat it as a constant @xmath21 evaluated at the energy where @xmath22 is maximum .
then the reaction rate per pair of particles ( without screening ) can be computed as @xmath23 @xcite developed a treatment to include the effect of static electron screening on nuclear reaction rates . here
we summarize his method which we will use in section [ sect : method ] as the inspiration for our calculation of the dynamic screening correction .
we begin by writing the total interaction energy as a combination of the bare coulomb potential and a contribution from the plasma : @xmath24 then consider a case in which the classical impact parameter @xmath25 is very small compared with the charge cloud radius @xmath26 and the nuclear radius @xmath27 is much smaller than @xmath25 .
then the barrier penetration factor for @xmath28 depends only on the expression @xmath29 for distances larger than @xmath25 , the barrier penetration factor hardly depends on the potential .
@xmath30 must be small for distances greater than @xmath26 and approach a constant value @xmath31 of the order of magnitude of @xmath32 for small @xmath33 .
then , @xmath34 where @xmath35 is the relative kinetic energy for which the integrand in equation [ eq : lambda1 ] reaches a sharp maximum .
if this inequality is satisfied , @xmath30 can be replaced by the potential at the origin @xmath31 . by examining expression [ eq : edep ] , we can see that the screening potential has effectively increased the kinetic energy by a magnitude of @xmath31 , so the cross section factors for @xmath36 are equivalent to the unscreened factors with energy @xmath37 .
equation [ eq : lambda1 ] can then be replaced by @xmath38 with the change of variables @xmath39 and the approximation @xmath40 , the reaction rate per pair of particles becomes @xmath41 because the penetration factor for @xmath42 is so small , the lower limit of the integral can be set to zero without significantly changing the value of the integral .
we then see that @xmath43 illustrating that the reaction rate for the statically screened potential can be approximated by multiplying the rate for the bare coulomb potential by @xmath44 .
salpeter then derives @xmath31 by solving the poisson - boltzmann equation for electrons and ions in a plasma under the condition of weak screening ( @xmath45 ) .
he arrives at an expression for the screening energy that is equivalent to that of the debye
hckel theory of dilute solutions of electrolytes @xcite : @xmath46 the debye length , @xmath26 , is the characteristic screening length of a plasma at temperature @xmath47 with number density @xmath48 which is defined by @xmath49 for a neutral proton electron plasma , the electron number density @xmath50 and ion number density @xmath51 are both @xmath52 , so the debye length is just @xmath53 although salpeter s expression accurately describes the effect of static screening , the issue of dynamic screening in the hot , dense plasma of the solar interior remains an open question .
dynamic screening occurs when the screened interaction energy of a pair of ions depends on the relative velocity of the pair .
most of the ions in the solar plasma are much slower than the electrons and the fastest ions .
the thermal ions are therefore not able to rearrange themselves as quickly around individual faster moving ions . since nuclear reactions require energies several times the average thermal energy , the ions that are able to engage in nuclear reactions in the sun are the faster moving ions , which are not accompanied by a full static screening cloud .
salpeter s derivation uses the mean - field approach in which the many - body interactions are reduced to an average interaction that simplifies calculations .
this technique is quite useful for calculations describing the average behavior of the plasma . however
, dynamic effects for the fast - moving , interacting ions in hot , dense plasma lead to a screened potential that deviates from the average value .
therefore , the mean - field approximation is not appropriate for computing stellar nuclear reaction rates .
instead , we use the molecular - dynamics method of @xcite to model the motion of protons and electrons in a plasma under solar conditions in order to investigate dynamic screening in _ p__p _ reactions .
the advantage of the molecular - dynamics method is that it does not assume a mean field . nor
does it assume a long - time average potential for the scattering of any two charges , which is necessary in the statistical way to solve poisson s equation to obtain the mean potential in a plasma . in previous work , @xcite present simulation results for the velocity - dependent screening energy of _
p__p _ reactions in a plasma with the temperature and density of the solar core ( @xmath54 k , @xmath55 @xmath56 ) .
they demonstrate that the static screening result does not accurately represent this plasma , and they compute a screening energy that depends on the relative kinetic energy of a pair of interacting ions ( see figure [ fig : usim ] ) . in this paper
, we use their simulation results to compute a correction to the solar _ p__p _ reaction rate due to the dynamic screening they observe .
-0.4 cm .3 cm
we begin with the calculations of @xcite for a plasma of protons and electrons with the temperature and density of the solar core .
their figure 5 shows the relationship between the total interaction energy at the turning point and the relative kinetic energy of a pair of interacting protons . as in salpeter s static screening derivation
, we can split the total interaction energy into the coulomb and screening cloud contributions : @xmath57 the screening energy now includes a velocity dependence , so @xmath31 can no longer be factored out of the energy integral as was done to obtain equation [ eq : scr.lambda3 ] for the static screening case . following salpeter s calculation for static screening , we focus on the contribution to the interaction energy from the screening cloud at small @xmath33
. this value of @xmath31 is obtained from the relationship between the screening energy of a pair of protons at their turning point and their relative kinetic energy which is shown in figure [ fig : usim ] .
the dynamic screening energy curve is described by the equation @xmath58 which comes from the best - fit curve for the @xcite @xmath59 .
( note the difference in sign from @xcite , where the screening energy was defined as a negative contribution to the total energy , @xmath60 . in this paper
, we use the @xcite convention , as shown in equations [ eq : int.energy1 ] and [ eq : int.energy2 ] . )
now we return to equation [ eq : scr.lambda ] for the screened reaction rate per pair of particles . the assumptions and approximations used in the derivation of equation [ eq : scr.lambda ] for the static case
will be examined and justified for the dynamic case in section [ subsect : assumptions ] . replacing @xmath31 from the statically screened case with @xmath61 for the dynamically screened case and using definition [ eq : sigma ] for the cross section , we have @xmath62 de.\end{aligned}\ ] ] because @xmath18 is a slowly varying function of energy , we make the approximation @xmath63 and replace this function with the constant @xmath21 , as was done in the original reaction - rate calculation in equation [ eq : lambda3 ]
. we can now evaluate the reaction rate per pair of particles using equation [ eq : lambda3 ] for the unscreened case and @xmath64de \end{aligned}\ ] ] for the statically and dynamically screened cases .
equation [ eq : debye ] gives the static screening @xmath31 and equation [ eq : uofv ] gives the dynamic screening @xmath61 .
because all three cases contain the factor @xmath65 , we only need to compute the integrals in order to compare ratios of the two screened cases to the bare coulomb potential case to obtain the correction factors for the _ p__p _ reaction rate .
[ cols="<,^,^",options="header " , ] table [ tab : integrals ] shows the results of the screening corrections for solar _ p__p _ reaction rates computed from the integrals in equations [ eq : lambda3 ] and [ eq : dyn.lambda2 ] .
the statically screened correction shows a fairly large enhancement in the nuclear reaction rate .
conversely , the dynamically screened reaction rate is almost the same as the unscreened rate .
how does the dynamic screening energy seen in figure [ fig : usim ] result in a reaction rate that is so close to the unscreened reaction rate ? to answer that question , we compare the components of the the integrands from equations [ eq : lambda3 ] and [ eq : dyn.lambda2 ] . both integrands can be written in the general form @xmath66 where @xmath67 @xmath68\ ] ] and @xmath69 the three cases only differ in the screening energy @xmath70 which is shown for each case in table [ tab : integrals ] . in figure
[ fig : ge ] , we see that the dynamic @xmath71 approaches the unscreened @xmath71 for high energies .
figure [ fig : he ] shows that this leads to the dynamic @xmath72 approaching the unscreened @xmath72 for energies above @xmath732 @xmath74 , while the static @xmath72 is very different from both the dynamic and bare @xmath72 . below energies of @xmath732 @xmath74 the dynamic @xmath72 drops rapidly away from the unscreened value of 1 .
however , the factor @xmath72 is multiplied by @xmath75 in the integrand of the reaction - rate equations . as seen in figure [ fig : je ] , @xmath75 is very small to zero below energies of @xmath732 @xmath74 , damping out the region of @xmath72 in which the unscreened and dynamic results diverge .
this leads to integrands for the dynamic and unscreened cases that are nearly identical , as seen when @xmath72 and @xmath75 are multiplied together to give @xmath76 in figure [ fig : fe ] .
the gamow - peak - like factor @xmath75 acts as a weighting function to devalue the @xmath72 contribution of the slow pairs of ions that rarely participate in nuclear reactions .
the faster pairs that cause less polarization of the surrounding plasma and therefore see less screening provide the main contribution to the reaction - rate integral .
now that we have defined the integrand @xmath76 for the dynamic case , we can return to the issue of assumptions and approximations that were justified in the static screening rate derivation and adopted in the dynamic screening rate derivation . here
we assess the validity of these assumptions and approximations in the case of dynamic screening .
the first assumption is that @xmath25 is very small compared with the charge cloud radius @xmath26 and that the nuclear radius @xmath27 is much smaller than @xmath25 , leading to expression [ eq : edep ] .
although the dynamic case does not have a traditional static screening cloud , the inequality @xmath77 is still satisfied .
this can be seen in the definition of @xmath78 as the distance beyond which an appreciable fraction of the nuclear charge is screened by the polarization charge cloud , a distance that is large for the dynamic case .
next we examine the inequality in equation [ eq : ineq ] , @xmath79 . in the dynamic case , @xmath35 is the kinetic energy for which the integrand of equation [ eq : dyn.lambda2 ] is maximum .
we see from figure [ fig : fe ] and @xmath31 in figure [ fig : usim ] that the peak of the integrand @xmath76 occurs at energies for which @xmath79 .
finally , we address the approximation that @xmath30 can be replaced by the potential at the origin @xmath31 . while this is easy to see for the static case , in the case of dynamic screening we do not have the screening energy in the form @xmath80 to examine this claim .
instead , we define @xmath61 to be the screening energy computed at the turning point of the approaching protons , since this is the relevant @xmath80 for nuclear reaction - rate calculations .
in light of the contentious debate over the validity of dynamic screening , we devote this section to a discussion of arguments that have been made against dynamic screening .
the argument for dynamic screening has been damaged by several derivations of alternate screening formulae ( see , for example , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) that were subsequently shown to be incorrect . here
we discuss two examples , @xcite and @xcite .
@xcite derived a modified screening potential for fusing ions when the gamow velocity is greater than the thermal velocity .
@xcite showed that including processes of excitation or de - excitation of the plasma in an interaction with one of the fusing ions exactly cancels the dynamic modifications proposed by @xcite .
@xcite then introduced a factor of 3/2 on the screening energy .
they arrived at this result by including the interaction of the the screening cloud from each fusing ion in the total interaction potential .
@xcite showed that @xcite misinterpreted the thermodynamics and used an incorrect potential in the schrdinger equation for the system .
@xcite summarize the problems with several different alternative screening formulae .
however , finding flaws in these ( and other ) derivations of analytical expressions for screening deviations does not rule out the effect of dynamic screening .
this argument only highlights the difficulty in developing a general analytical formalism to describe dynamic screening .
for the gibbs distribution , probabilities for momenta and coordinates are independent and can not influence each other .
this leads to the argument that velocities of fusing particles can not have an effect on screening because the distribution function is not factorable .
however , it is clear when examining individual ions that their relative velocity affects how close the ions can be to each other .
this argument extends to ions in a plasma , where the configuration of the screening cloud of approaching ions depends on the relative velocity of those ions . over the whole system ,
this velocity - dependent effect averages out to the gibbs distribution , but each screening cloud is not identical to the average configuration of a screening cloud in that system .
solar nuclear reactions select a biased sample of the ions in the system .
these nuclear reactions involve mainly the fastest ions , not a random sample of all ions in the system .
therefore the velocity distribution must be multiplied by the velocity - dependent screening energy before integration instead of beginning with an average value of the distribution .
do dynamic screening results imply that higher - ordered terms are required ?
the screening energy can be expressed as a power series expansion in the plasma coupling parameter @xmath81 : @xmath82 where @xmath83 @xmath84 , and @xmath85 is given by @xcite .
the first term reduces to the debye
hckel weak screening result shown in equation [ eq : debye ] .
for the temperature and density of our simulations ( @xmath54 k , @xmath55 @xmath56 ) , @xmath86 , so higher - order terms are small .
therefore , if dynamic screening could be described by higher - order terms in the expansion , the effect should be much smaller than the first term .
however , the dynamic correction is not just a higher - order term in the expansion .
dynamic effects come from a different approach to determining screening effects . instead of deriving an expression for screening based on average properties of the system
, we examine the formation of the screening clouds themselves and do not average out the velocity - dependent nature of the clouds .
what observational evidence can confirm any screening effect in the nuclear reactions in the sun ?
many early discussions of dynamic screening were motivated by the neutrino problem which has since been resolved with neutrino oscillation theory .
in addition , including dynamic screening corrections in models with the @xcite or @xcite abundances worsened agreement with helioseismic constraints ( see , for example , * ? ? ?
however , the solar abundance problem provides renewed motivation for exploring dynamic screening in solar nuclear reactions . before 2005 ,
solar models with the latest input physics reproduced the sound - speed profile determined from helioseismic inversions to within @xmath87 and also provided good agreement with the seismically inferred convection zone depth and convection zone helium abundance .
then @xcite , @xcite and @xcite began using three - dimensional hydrodynamic models of the solar atmosphere with improved input physics and non - local thermodynamic equilibrium effects to determine solar atmospheric abundances .
these revised calculations lowered element abundances by up to 1/3 .
when the lower abundances are incorporated in solar models , the sound - speed profiles , convection zone depths , and convection zone helium abundances give worse agreement with helioseismic constraints than models with the old , higher abundances .
many attempts have been made to improve agreement by adjusting the physics or evolutionary assumptions in solar models ( see * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
although these adjustments have shown some improvement , no model using the new lower abundances agrees as well with the helioseismic constraints as the models using the older abundances . in a forthcoming paper ,
@xcite incorporate the dynamic screening correction shown here for solar _ p__p _ reaction rates into solar models with the new lower abundances .
they show that including this correction in solar models improves the sound speed discrepancy in the solar core , as shown in figure [ fig : sound_pp0 ] .
this improvement does not fully reconcile the new abundances with helioseismic constraints , but it is a step in the right direction . perhaps in combination with other changes , dynamic screening corrections could contribute to a solution to the solar abundance problem .
-0.4 cm .3 cm
we have shown that dynamic screening in solar _
p__p _ reactions does not reproduce the enhancement of reaction rates that is predicted by salpeter s static screening approximation .
in fact , the dynamic screening seen by @xcite shows essentially no correction to the unscreened reaction rate .
although _ p__p _ reactions in the core are the main source of nuclear energy generated in the sun , this reaction - rate correction is only the beginning of understanding how including dynamic effects will alter a full solar model .
the reaction - rate correction must be generalized to treat other temperatures , densities , compositions , and reactions .
in addition , the effect of dynamic screening on the equation of state must be examined . until we can meet both of these challenges ,
dynamic screening can not be incorporated completely and consistently in solar and stellar models . | the solar abundance controversy inspires renewed investigations of the basic physics used to develop solar models . here
we examine the correction to the proton - proton reaction rate due to dynamic screening effects . starting with the dynamic screening energy from the molecular - dynamics simulations of mao et al .
, we compute a reaction - rate correction for dynamic screening .
we find that , contrary to static screening theory , this dynamic screening does not significantly change the reaction rate from that of the bare coulomb potential . |
kleinian groups , which are the discrete groups of orientation preserving isometries of hyperbolic space , have been studied for a number of years , and have been of particular interest since the work of thurston in the late 1970s on the geometrization of compact @xmath0manifolds .
a kleinian group can be viewed either as an isolated , single group , or as one of a member of a family or continuum of groups . in this note , we concentrate our attention on the latter scenario , which is the deformation theory of the title , and attempt to give a description of various of the more common families of kleinian groups which are considered when doing deformation theory .
no proofs are given , though it is hoped that reasonable coverage of the current state of the subject is given , and that ample references have been given for the interested reader to venture boldly forth into the literature .
it is possible to consider the questions raised here in much more general settings , for example for kleinian groups in @xmath1dimensions for general @xmath1 , but that is beyond the scope of what is attempted here
. some material on this aspect of the question can be found in bowditch @xcite and the references contained therein .
the author would like to thank dick canary , ed taylor , and brian bowditch for useful conversations during the preparation of this work , as well as the referee for useful comments .
we begin by giving a few basic definitions of the objects considered in this note , namely kleinian groups .
we go on to define and describe the basic structure of the deformation spaces we are considering herein .
a _ kleinian group _ is a discrete subgroup of @xmath2 , which we view as acting both on the riemann sphere @xmath3 by mbius transformations and on real hyperbolic @xmath0space @xmath4 by isometries , where the two actions are linked by the poincar extension .
the action of an infinite kleinian group @xmath5 partitions @xmath3 into two sets , the _ domain of discontinuity _ @xmath6 , which is the largest open subset of @xmath7 on which @xmath5 acts discontinuously , and the _ limit set _ @xmath8 . if @xmath8 contains two or fewer points , @xmath5 is _ elementary _ , otherwise @xmath5 is _ non - elementary_. for a non - elementary kleinian group @xmath5 , the limit set @xmath8 can also be described as the smallest non - empty closed subset of @xmath3 invariant under @xmath5 .
we refer the reader to maskit @xcite or matsuzaki and taniguchi @xcite as a reference for the basics of kleinian groups .
an isomorphism @xmath9 between kleinian groups @xmath5 and @xmath10 is _ type - preserving _ if , for @xmath11 , we have that @xmath12 is parabolic if and only if @xmath13 is parabolic .
a kleinian group is _ convex cocompact _ if its convex core is compact ; recall that the _ convex core _ associated to a kleinian group @xmath5 is the minimal convex submanifold of @xmath14 whose inclusion is a homotopy equivalence .
more generally , a kleinian group is _ geometrically finite _ if it is finitely generated and if its convex core has finite volume .
this is one of several equivalent definitions of geometric finiteness ; the interested reader is referred to bowditch @xcite for a complete discussion .
a kleinian group @xmath5 is _ topologically tame _ if its corresponding quotient @xmath0manifold @xmath14 is homeomorphic to the interior of a compact @xmath0manifold .
geometrically finite kleinian groups are topologically tame .
it was conjectured by marden @xcite that all finitely generated kleinian groups are topologically tame .
a compact @xmath0manifold @xmath15 is _ hyperbolizable _ if there exists a hyperbolic @xmath0manifold @xmath16 homeomorphic to the interior of @xmath15 .
note that a hyperbolizable @xmath0manifold @xmath15 is necessarily orientable ; _ irreducible _ , in that every embedded @xmath17sphere in @xmath15 bounds a @xmath0ball in @xmath15 ; and _ atoroidal _ , in that every embedded torus @xmath18 in @xmath15 is homotopic into @xmath19 . also , since the universal cover @xmath4 of @xmath20 is contractible , the fundamental group of @xmath15 is isomorphic to @xmath5 .
for a discussion of the basic theory of @xmath0manifolds , we refer the reader to hempel @xcite and jaco @xcite . keeping to our viewpoint of a kleinian group as a member of a family of groups , throughout this survey
we view a kleinian group as the image @xmath21 of a representation @xmath22 of a group @xmath23 into @xmath24 . unless explicitly stated otherwise , we assume that @xmath23 is finitely generated , torsion - free , and non - abelian , so that in particular @xmath21 is non - elementary . the most basic of the deformation spaces is the _ representation variety _
@xmath25 which is the space of all representations of @xmath23 into @xmath27 with the following topology . given a set of generators @xmath28 for @xmath23
, we may naturally view @xmath25 as a subset of @xmath29 , where a representation @xmath30 corresponds to the @xmath31tuple @xmath32 in @xmath29 .
the defining polynomials of this variety are determined by the relations in @xmath23 .
in particular , if @xmath23 is free , then @xmath33 .
it is easy to see that @xmath25 is a closed subset of @xmath29 .
the representations in @xmath25 are _ unnormalized _ , in the sense that there is a natural free action of @xmath27 on @xmath25 by conjugation .
depending on the particular question being addressed , it is sometimes preferable to remove the ambiguity of this action and form the quotient space @xmath34 .
though a detailed description is beyond the scope of this survey , we pause here to mention work of culler and shalen @xcite , @xcite , in which a slight variant of the representation variety as defined above plays a fundamental role , and which has inspired further work of morgan and shalen @xcite , @xcite , @xcite and culler , gordon , luecke , and shalen @xcite .
the basic object here is not the space @xmath35 of all representations of @xmath23 into @xmath27 as defined above , but instead the related space @xmath36 of all representations of @xmath23 into @xmath37 , modulo the action of @xmath37 .
the introduction of this space @xmath36 does beg the question of when a representation of @xmath23 into @xmath38 can be lifted to a representation of @xmath23 into @xmath39 .
we note in passing that this question of lifting representations has been considered by a number of authors , including culler , kra , and thurston , to name but a few ; we refer the reader to the article by kra @xcite for exact statements and a review of the history , including references . by considering the global structure of the variety @xmath36 in the case that @xmath23 is the fundamental group of a compact , hyperbolizable @xmath0manifold @xmath15 , and in particular the ideal points of its compactification , culler and shalen @xcite are able to analyze the actions of @xmath23 on trees , which in turn has connections with the existence of essential incompressible surfaces in @xmath15 , finite group actions on @xmath15 , and has particular consequences in the case that @xmath15 is the complement of a knot in @xmath40 .
we refer the reader to the excellent survey article by shalen @xcite , as well as to the papers cited above .
let @xmath41 denote the subspace of @xmath25 consisting of those representations @xmath22 for which @xmath43 is parabolic if and only if @xmath44 lies in a rank two free abelian subgroup of @xmath23 .
we refer to @xmath41 as the space of _ minimally parabolic _
representations of @xmath23 . in particular ,
if @xmath23 contains no @xmath45 subgroups , then the image @xmath21 of every @xmath22 in @xmath41 is _ purely loxodromic _ , in that every non - trivial element of @xmath21 is loxodromic .
set @xmath46 .
let @xmath47 denote the subspace of @xmath25 consisting of the discrete , faithful representations of @xmath23 , that is , the injective homomorphisms of @xmath23 into @xmath27 with discrete image .
for the purposes of this note , the space @xmath47 is our universe , as it is the space of all kleinian groups isomorphic to @xmath23 .
set @xmath49 .
we note that there exists an equivalent formulation of @xmath50 in terms of manifolds . given a hyperbolic @xmath0manifold @xmath20 , let @xmath51 denote the set of all pairs @xmath52 , where @xmath53 is a hyperbolic @xmath0manifold and @xmath54 is a homotopy equivalence , modulo the equivalence relation @xmath55 if there exists an orientation preserving isometry @xmath56 so that @xmath57 is homotopic to @xmath44 .
the topology on @xmath58 is given by noting that , if we let @xmath59 be a choice of conjugacy class of the fundamental group of @xmath20 , then each element @xmath52 in @xmath58 gives rise to a discrete , faithful representation @xmath60 of @xmath5 into @xmath38 , with equivalent points in @xmath58 giving rise to conjugate representations into @xmath27 .
hence , equipping @xmath58 with this topology once again gives rise to @xmath50 with @xmath61 .
the following theorem , due to jrgensen , describes the fundamental property of @xmath47 , namely that the limit of a sequence of elements of @xmath47 is again an element of @xmath47 .
@xmath47 is a closed subset of @xmath25 .
[ deformation - space - closed ] there is one notable case in which @xmath50 is completely understood , namely in the case that @xmath23 is the fundamental group of a compact , hyperbolizable @xmath0manifold @xmath15 whose boundary is the union of a ( possibly empty ) collection of tori . in this case , the hyperbolic structure on the interior of @xmath15 is unique , by the classical rigidity theorem of mostow , for closed manifolds , and prasad , for manifolds with non - empty toroidal boundary . rephrasing this statement as a statement about deformation spaces yields the following .
suppose that @xmath23 is the fundamental group of a compact , orientable @xmath0manifold @xmath15 whose boundary is the union of a ( possibly empty ) collection of tori .
then , @xmath50 either is empty or consists of a single point . [ mostow - rigidity ] given this result , it will cause us no loss of generality to assume that henceforth all kleinian groups have infinite volume quotients . let @xmath62 denote the subset of @xmath47 consisting of those representations @xmath22 with geometrically finite , minimally parabolic image @xmath21 . in particular , if @xmath23 contains no @xmath64 subgroups , then the image @xmath21 of every representation @xmath65 is convex cocompact . set @xmath66 , and note that since @xmath38 is connected , the quotient map gives a one - to - one correspondence between the connected components of @xmath62 and those of @xmath67 .
it is an immediate consequence of the core theorem of scott @xcite and the hyperbolization theorem of thurston that if @xmath68 is non - empty , then @xmath62 is non - empty . for a discussion of the hyperbolization theorem ,
see morgan @xcite , otal and paulin @xcite , or otal @xcite for the fibered case .
we note here that , if there exists a geometrically finite , minimally parabolic representation of @xmath23 into @xmath27 , then in general there exist many geometrically finite representations which are not minimally parabolic , which can be constructed as limits of the geometrically finite , minimally parabolic representations . this construction has been explored in detail for a number of cases by maskit @xcite and ohshika @xcite . in the case that @xmath23 is itself a geometrically finite , minimally parabolic kleinian group , the structure of @xmath67 is fairly well understood , both as a subset of @xmath50 and in terms of how the components of @xmath67 are parametrized by topological data .
we spend the remainder of this section making these statements precise .
we begin with the quasiconformal stability theorem of marden @xcite .
if @xmath23 is a geometrically finite , minimally parabolic kleinian group , then @xmath67 is an open subset of @xmath35 .
[ quasiconformal - stability ] as a converse to this , we have the structural stability theorem of sullivan @xcite .
we note here that the versions of the theorems of marden and sullivan given here are not the strongest , but are adapted to the point of view taken in this paper .
the general statements holds valid in slices of @xmath50 in which a certain collection of elements of @xmath23 are required to have parabolic image , not just those which belong to @xmath45 subgroups .
let @xmath23 be a finitely generated , torsion - free , non - elementary kleinian group .
if there exists an open neighborhood of the identity representation in @xmath35 which lies in @xmath50 , then @xmath23 is geometrically finite and minimally parabolic .
[ structural - stability ] combining these , we see that @xmath67 is the interior of @xmath69 .
a natural question which arises from this is whether there are points of @xmath50 which do not lie in the closure of @xmath70 .
@xmath50 is the closure of @xmath67 .
[ density - conjecture ] this conjecture is due originally to bers in the case that @xmath23 is the fundamental group of a surface , see bers @xcite , and extended by thurston to general @xmath23 . there has been a good deal of work in the past couple of years on the global structure of @xmath67 and its closure .
we begin with an example to show that there exist groups @xmath23 for which @xmath67 is disconnected ; the example we give here , in which @xmath67 has finitely many components , comes from the discussion in anderson and canary @xcite .
let @xmath18 be a solid torus and for large @xmath31 , let @xmath71 be disjoint embedded annuli in @xmath72 whose inclusion into @xmath18 induces an isomorphism of fundamental groups . for each @xmath73 ,
let @xmath74 be a compact , orientable surface of genus @xmath75 with a single boundary component , and let @xmath76 , where @xmath77 is a closed interval .
construct a compact @xmath0manifold @xmath15 by attaching the annulus @xmath78 in @xmath79 to the annulus @xmath80 in @xmath72 .
the resulting @xmath0manifold @xmath15 is compact and hyperbolizable @xmath0manifold and has fundamental group @xmath23 .
this @xmath0manifold is an example of a _ book of i
bundles_. let @xmath22 be an element of @xmath67 for which the interior of @xmath15 is homeomorphic to @xmath81 .
let @xmath82 be a permutation of @xmath83 , and consider now the manifold @xmath84 obtained by attaching the annulus @xmath85 in @xmath79 to the annulus @xmath86 in @xmath72 . by construction , @xmath84 is compact and hyperbolizable , and has fundamental group @xmath23 ; let @xmath87 be an element of @xmath67 for which the interior of @xmath84 is homeomorphic to @xmath88 . since @xmath15 and @xmath84 have isomorphic fundamental groups , they are homotopy equivalent . however , in the case that @xmath82 is not some power of the cycle @xmath89 , then there does not exist an orientation preserving homeomorphism between @xmath15 and @xmath84 , and hence @xmath22 and @xmath87 lie in different components of @xmath67 .
in the general case that @xmath23 is finitely generated and does not split as a free product , there exists a characterization of the components of both @xmath67 and its closure @xmath90 in terms of the topology of a compact , hyperbolizable @xmath0manifold @xmath15 with fundamental group @xmath23 .
this characterization combines work of canary and mccullough @xcite and of anderson , canary , and mccullough @xcite .
we need to develop a bit of topological machinery before discussing this characterization . for a compact , oriented , hyperbolizable @xmath0manifold @xmath15 with non - empty , incompressible boundary ,
let @xmath91 denote the set of _ marked homeomorphism types of compact , oriented @xmath0manifolds homotopy equivalent to @xmath15_. explicitly , @xmath91 is the set of equivalence classes of pairs @xmath92 , where @xmath93 is a compact , oriented , irreducible @xmath0manifold and @xmath94 is a homotopy equivalence , and where two pairs @xmath95 and @xmath96 are equivalent if there exists an orientation preserving homeomorphism @xmath97 such that @xmath98 is homotopic to @xmath99 .
denote the class of @xmath92 in @xmath91 by @xmath100 $ ] .
there exists a natural map @xmath101 , defined as follows . for @xmath102 , let @xmath103 be a compact core for @xmath104 and let @xmath105 be a homotopy equivalence such that @xmath106 is equal to @xmath22 .
set @xmath107 $ ] .
it is known that the restriction of @xmath108 to @xmath109 is surjective , and that two elements @xmath110 and @xmath111 of @xmath112 lie in the same component of @xmath109 if and only if @xmath113 .
hence , @xmath108 induces a one - to - one correspondence between the components of @xmath109 and the elements of @xmath91 ; the reader is directed to canary and mccullough @xcite for complete details . given a pair @xmath114 and @xmath115 of compact , hyperbolizable @xmath0manifolds with non - empty , incompressible boundary , say that a homotopy equivalence @xmath116 is a _ primitive shuffle _ if there exists a finite collection @xmath117 of primitive solid torus components of the characteristic submanifold @xmath118 and a finite collection @xmath119 of solid torus components of @xmath120 , so that @xmath121 and so that @xmath122 restricts to an orientation preserving homeomorphism from @xmath123 to @xmath124 ; we do not define the _ characteristic submanifold _ here , but instead refer the reader to canary and mccullough @xcite , jaco and shalen @xcite , or johannson @xcite .
let @xmath125 $ ] and @xmath126 $ ] be two elements of @xmath91 .
say that @xmath126 $ ] is _ primitive shuffle equivalent _ to @xmath125 $ ] if there exists a primitive shuffle @xmath127 such that @xmath126 = [ ( m_2 , \varphi\circ h_1)]$ ] .
we note that when @xmath15 is hyperbolizable , this gives an equivalence relation on @xmath128 , where each equivalence class contains finitely many elements of @xmath91 ; let @xmath129 denote the set of equivalence classes .
by considering the composition @xmath130 of @xmath108 with the quotient map @xmath131 , we obtain the following complete enumeration of the components of @xmath132 .
let @xmath15 be a compact , hyperbolizable @xmath0manifold with non - empty , incompressible boundary , and let @xmath125 $ ] and @xmath126 $ ] be two elements of @xmath91 .
the associated components of @xmath133 have intersecting closures if and only if @xmath126 $ ] is primitive shuffle equivalent to @xmath125 $ ] . in particular , @xmath134 gives a one - to - one correspondence between the components of @xmath132 and the elements of @xmath135 . [ enumeration ] before closing this section , we highlight two consequences of the analysis involved in the proof of theorem [ enumeration ] .
the first involves the accumulation , or , more precisely , the lack thereof , of components of @xmath109 .
( anderson , canary , and mccullough @xcite)let @xmath15 be acompact , hyperbolizable @xmath0manifold with non - empty , incompressible boundary.then , the components of @xmath109 can not accumulate in @xmath136 .
in particular , the closure @xmath132 of @xmath109 is the union of the closures of the components of @xmath109 .
the second involves giving a complete characterization , in terms of the topology of @xmath15 , as to precisely when @xmath137 has infinitely many components .
recall that a compact , hyperbolizable @xmath0manifold @xmath15 with non - empty , incompressible boundary has _
double trouble _ if there exists a toroidal component @xmath18 of @xmath19 and homotopically non - trivial simple closed curves @xmath138 in @xmath18 and @xmath139 and @xmath140 in @xmath141 such that @xmath139 and @xmath140 are not homotopic in @xmath19 , but @xmath138 , @xmath139 and @xmath140 are homotopic in @xmath15 .
let @xmath15 be a compact , hyperbolizable @xmath0manifold with non - empty , incompressible boundary.then , @xmath132 has infinitely many components if and only if @xmath15 has double trouble . moreover , if @xmath15 has double trouble , then @xmath136 has infinitely many components .
[ infinitely - many - components ] in the case that @xmath23 is itself a finitely generated kleinian group , the classical deformation theory of @xmath23 consists largely of the study of the space of _ quasiconformal deformations _ of @xmath23 , which consists of those representations of @xmath23 into @xmath27 which are induced by a quasiconformal homeomorphism of the riemann sphere @xmath144 .
we do not give a precise definition here , but roughly , a _ quasiconformal homeomorphism _
@xmath145 of @xmath144 is a homeomorphism which distorts the standard complex structure on @xmath144 by a bounded amount ; the interested reader is referred to ahlfors @xcite or to lehto and virtanen @xcite for a thorough discussion of quasiconformality .
we do note that a quasiconformal homeomorphism @xmath146 is completely determined ( up to post - composition by a mbius transformation ) by the measurable function @xmath147 , and that to every measurable function @xmath148 on @xmath144 with @xmath149 there exists a quasiconformal homeomorphism @xmath145 of @xmath144 which solves the beltrami equation @xmath150 .
set @xmath142 to be the space of those representations @xmath22 of @xmath23 into @xmath27 which are induced by a quasiconformal homeomorphism of @xmath3 , so that @xmath151 if there exists a quasiconformal homeomorphism @xmath145 of @xmath144 so that @xmath152 for all @xmath153 . by definition
, we have that @xmath142 is contained in @xmath68 .
set @xmath154 .
it is known that @xmath155 is a complex manifold , and is actually the quotient of the teichmller space of the ( possibly disconnected ) quotient riemann surface @xmath156 by a properly discontinuous group of biholomorphic automorphims .
this result , in its full generality , follows from the work of a number of authors , including maskit @xcite , kra @xcite , bers @xcite , and sullivan @xcite .
we note here , in the case that @xmath23 is a geometrically finite , minimally parabolic kleinian group , that it follows from the isomorphism theorem of marden @xcite that @xmath155 is the component of @xmath67 containing the identity representation .
sullivan @xcite has shown , for a finitely generated kleinian group @xmath23 , if there exists a quasiconformal homeomorphism @xmath145 of @xmath144 which conjugates @xmath23 to a kleinian group and which is conformal on @xmath157 , then @xmath145 is necessarily a mbius transformation .
in other words , if @xmath145 conjugates @xmath23 to subgroup of @xmath27 , then @xmath147 is equal to @xmath158 on @xmath159 . in particular ,
if @xmath157 is empty , then @xmath155 consists of a single point , namely the identity representation .
this can be viewed as a generalization of theorem [ mostow - rigidity ] , as sullivan s result also holds for an infinite volume hyperbolic @xmath0manifold @xmath20 whose uniformizing kleinian group @xmath23 has limit set the whole riemann sphere .
we note here that the study of quasiconformal deformations of finitely kleinian groups is the origin of the ahlfors measure conjecture . in @xcite ,
ahlfors raises the question of whether the limit set of a finitely generated kleinian group with non - empty domain of discontinuity necessarily has zero area .
if this conjecture is true , then it would be impossible for a quasiconformal deformation of a finitely generated kleinian group @xmath23 to be supported on the limit set of @xmath23 .
the result of sullivan mentioned above implies that no such deformation exists , though without solving the measure conjecture , which has not yet been completely resolved .
it is known that the measure conjecture holds in a large number of cases , in particular it holds for all topologically tame groups . for a discussion of this connection , we refer the reader to canary @xcite and the references contained therein .
there are several classes of kleinian groups for which @xmath160 has been extensively studied , which we discuss here .
a _ schottky group _ is a finitely generated , purely loxodromic kleinian group @xmath23 which is free on @xmath44 generators and whose domain of discontinuity is non - empty ; the number of generators is sometimes referred to as the _ genus _ of the schottky group .
this is not the original definition , but is equivalent to the usual definition by a theorem of maskit @xcite . in particular ,
a schottky group is necessarily convex cocompact .
chuckrow @xcite shows that any two schottky groups of the same rank are quasiconformally conjugate , so that @xmath155 is in fact equal to the space @xmath67 of all convex cocompact representations of a group @xmath23 which is free on @xmath44 generators into @xmath27 . in the same paper @xcite
, chuckrow also engages in an analysis of the closure of @xmath155 in @xmath35 for a schottky group @xmath23 of genus @xmath44 .
in particular , she shows that every point @xmath22 in @xmath161 has the property that @xmath21 is free on @xmath44 generators , and contains no elliptic elements of infinite order . however , this in itself is not enough to show that @xmath21 is discrete , as greenberg @xcite has constructed free , purely loxodromic subgroups of @xmath27 which are not discrete .
more generally , chuckrow also shows that the limit @xmath22 of a convergent sequence @xmath162 of type - preserving faithful representations in @xmath25 is again a faithful representation of @xmath23 , and that @xmath21 contains no elliptic elements of infinite order .
jrgensen @xcite credits his desire to generalize the results of chuckrow @xcite to leading him to what is now commonly referred to as jrgensen s inequality , which states that if @xmath12 and @xmath163 are elements of @xmath27 which generate a non - elementary kleinian group , then @xmath164 ) -2| \ge 1 $ ] , where @xmath165 is the trace of a matrix representative of @xmath12 in @xmath37 .
the proof of theorem [ deformation - space - closed ] is a direct application of this inequality . for a schottky group @xmath23
, it is known that @xmath50 is not compact .
there is work of canary @xcite and otal @xcite on a conjecture of thurston which gives conditions under which sequences in @xmath155 have convergent subsequences ; we do not give details here , instead referring the interested reader to the papers cited above .
we also mention here the work of keen and series @xcite on the riley slice of the space of @xmath17generator schottky groups , in which they introduce coordinates on the riley slice and study the cusp points on the boundary of the riley slice .
a _ quasifuchsian _ group is a finitely generated kleinian group whose limit set is a jordan curve and which contains no element interchanging the two components of its domain of discontinuity .
consequently , every quasifuchsian group is isomorphic to the fundamental group of a surface .
it is known that any two isomorphic purely loxodromic quasifuchsian groups are quasiconformally conjugate , by work of maskit @xcite , and hence for a purely loxodromic quasifuchsian group @xmath23 we have that @xmath166 .
this equality does not hold for quasifuchsian groups uniformizing punctured surfaces , for several reasons .
first , the quasifuchsian groups uniformizing the three - times punctured sphere and the once - punctured torus are isomorphic , namely the free group of rank two , but can not be quasiconformally conjugate , as the surfaces are not homeomorphic .
second , as every quasifuchsian group isomorphic to the free group @xmath23 of rank two contains parabolic elements , no quasifuchsian group isomorphic to @xmath23 lies in @xmath67 .
it is known that @xmath155 is biholomorphically equivalent to the product of teichmller spaces @xmath167 , where @xmath168 is one of the components of @xmath156 and @xmath169 is its complex conjugate .
a _ bers slice _ of @xmath155 for a quasifuchsian group @xmath23 is a subspace of @xmath155 of the form @xmath170 .
the structure of the closure of @xmath171 in @xmath50 has been studied by a number of authors , including bers @xcite , kerckhoff and thurston @xcite , maskit @xcite , mcmullen @xcite , and minsky @xcite . in particular
, bers @xcite showed that the closure @xmath172 of @xmath171 is compact , and kerckhoff and thurston @xcite have shown that the compactification @xmath172 depends on the basepoint @xmath173 , and so there are actually uncountably many such compactifications . among other major results ,
minsky @xcite has shown that every punctured torus group lies in the boundary of @xmath155 , where @xmath23 is a quasifuchsian group uniformizing a punctured torus and where a _ punctured torus group _ is a kleinian group generated by two elements with parabolic commutator . in particular , this shows that the relative version of the density conjecture holds for punctured torus groups .
there are other slices of @xmath155 which have been extensively studied .
there is the extensive work of keen and series , see for instance @xcite , @xcite , and @xcite , inspired in part by unpublished work of wright @xcite , on the maskit slice of the teichmller space of a punctured torus in terms of _ pleating coordinates _ , which are natural and geometrically interesting coordinates on the teichmller space of the punctured torus which are given in terms of the geometry of the corresponding hyperbolic @xmath0manifolds . in the case
that @xmath23 is a kleinian group for which the corresponding @xmath0manifold @xmath174 is a compact , acylindrical @xmath0manifold with non - empty , incompressible boundary , then every representation in @xmath67 in fact lies in @xmath160 ; this follows from work of johannson @xcite . in addition
, thurston @xcite has shown that @xmath50 is compact for such @xmath23 ; another proof is given by morgan and shalen @xcite .
there is one last class of deformations which we need to define , before beginning our discussion of the relationships between these spaces .
we begin with a topological definition .
a compact submanifold @xmath15 of a hyperbolic 3manifold @xmath20 is a _ compact core _ if the inclusion of @xmath15 into @xmath20 is a homotopy equivalence .
the core theorem of scott @xcite implies that every hyperbolic @xmath0manifold with finitely generated fundamental group has a compact core .
marden @xcite asked whether every hyperbolic @xmath0manifold @xmath20 with finitely generated fundamental group is necessarily _ topologically tame _ , in that @xmath20 is homeomorphic to the interior of its compact core .
set @xmath175 to be the subspace of @xmath47 consisting of the representations @xmath22 with minimally parabolic , topologically tame image @xmath21.set @xmath177 .
there is a notion related to topological tameness , namely _ geometric tameness _ , first defined by thurston @xcite .
we do not discuss geometric tameness here ; the interested reader should consult thurston @xcite , bonahon @xcite , or canary @xcite .
thurston @xcite showed that geometrically tame hyperbolic @xmath0manifolds with freely indecomposible fundamental group are topologically tame and satisfy the ahlfors measure conjecture .
bonahon @xcite showed that if every non - trivial free product splitting of a finitely generated kleinian group @xmath5 has the property that there exists a parabolic element of @xmath5 not conjugate into one of the free factors , then @xmath5 is geometrically tame .
canary @xcite extended the definition of geometrically tame to all hyperbolic @xmath0manifolds , proved that topologically tame hyperbolic @xmath0manifolds are geometrically tame , and proved that topological tameness has a number of geometric and analytic consequences ; in particular , he established that the ahlfors measure conjecture holds for topologically tame kleinian groups .
there is a second notion of convergence for kleinian groups which is distinct from the topology described above , which is equally important in the study of deformations spaces .
a sequence @xmath178 of kleinian groups converges _ geometrically _ to a kleinian group @xmath179 if two conditions are met , namely that every element of @xmath179 is the limit of a sequence of elements @xmath180 and that every accumulation point of every sequence @xmath180 lies in @xmath179 .
note that , unlike the topology of algebraic convergence described above , the geometric limit of a sequence of isomorphic kleinian groups need not be isomorphic to the groups in the sequence , and indeed need not be finitely generated .
however , it is known that the geometric limit of a sequence of non - elementary , torsion - free kleinian groups is again torsion - free .
we note here that it is possible to phrase the definition of geometric convergence in terms of the quotient hyperbolic @xmath0manifolds . setting notation ,
let @xmath158 denote a choice of basepoint for @xmath4 , and let @xmath181 and @xmath182 be the covering maps .
let @xmath183 be a ball of radius @xmath184 centered at the basepoint @xmath158 .
a sequence of torsion - free kleinian groups @xmath178 converges geometrically to a torsion - free kleinian group @xmath179 if and only if there exists a sequence @xmath185 and a sequence of orientation preserving maps @xmath186 such that the following hold : 1 .
@xmath187 and @xmath188 as @xmath189 ; 2 .
the map @xmath190 is a @xmath191bilipschitz diffeomorphism onto its image , @xmath192 , and @xmath193 converges to the identity for any compact set @xmath194 ; and 3 .
@xmath190 descends to a map @xmath195 , where @xmath196 is a submanifold of @xmath197 ; moreover , @xmath198 is also an orientation preserving @xmath191bilipschitz diffeomorphism onto its image .
[ geom - convergence ] for a proof of this lemma , see theorem 3.2.9 of canary , epstein , and green @xcite , and theorem e.1.13 and remark e.1.19 of benedetti and petronio @xcite . a fundamental example of the difference between algebraic and geometric convergence of kleinian groups is given by the following explicit example of jrgensen and marden @xcite ; earlier examples are given in jrgensen @xcite .
choose @xmath199 and @xmath200 in @xmath201 which are linearly independent over @xmath202 , and for each @xmath203 set @xmath204 , @xmath205 , and @xmath206 .
consider the loxodromic elements @xmath207 .
then , as @xmath208 , @xmath209 converges to @xmath210 , and so @xmath211 converges algebraically to @xmath212
. however , note that @xmath213 converges to @xmath214 as @xmath208 .
hence , @xmath211 converges geometrically to @xmath215 .
this example of the geometric convergence of loxodromic cyclic groups to rank two parabolic groups underlies much of the algebra of the operation of _ dehn surgery _ , which we describe here .
let @xmath15 be a compact , hyperbolizable @xmath0manifold , let @xmath18 be a torus component of @xmath19 , and choose a meridian longitude system @xmath216 on @xmath18 .
let @xmath217 be a solid torus and let @xmath218 be a simple closed curve on @xmath219 bounding a disc in @xmath217 .
for each pair @xmath220 of relatively prime integers , let @xmath221 be the @xmath0manifold by attaching @xmath219 to @xmath18 by an orientation - reversing homeomorphism which identifies @xmath218 with @xmath222 ; we refer to @xmath221 as the result of @xmath220 dehn surgery along @xmath18 .
the following theorem describes the basic properties of this operation ; the version we state is due to comar @xcite .
let @xmath15 be a compact , hyperbolizable @xmath0manifold and let @xmath223 be a non - empty collection of tori in @xmath19 .
let @xmath224 be a geometrically finite hyperbolic @xmath0manifold and let @xmath225 be an orientation preserving homeomorphism .
further assume that every parabolic element of @xmath5 lies in a rank two parabolic subgroup .
let @xmath226 be a meridian longitude basis for @xmath227 .
let @xmath228 be a sequence of @xmath31tuples of pairs of relatively prime integers such that , for each @xmath229 , @xmath230 converges to @xmath231 as @xmath208 .
then , for all sufficiently large @xmath1 , there exists a representation @xmath232 with discrete image such that 1 .
@xmath233 is geometrically finite , uniformizes @xmath234 , and every parabolic element of @xmath233 lies in a rank two parabolic subgroup ; 2 . the kernel of @xmath235 is normally generated by @xmath236 ; and 3 .
@xmath237 converges to the identity representation of @xmath5 .
[ dehn - surgery ] the idea of theorem [ dehn - surgery ] is due to thurston @xcite in the case that the hyperbolic @xmath0manifold @xmath20 has finite volume , so that @xmath19 consists purely of tori . in this case , it is also known that @xmath238 for each @xmath1 , and that @xmath239 as @xmath208 . for a more detailed discussion of this phenomenon
, we refer the reader to gromov @xcite and benedetti and petronio @xcite .
the generalization to the case that @xmath20 has infinite volume is due independently to bonahon and otal @xcite and comar @xcite .
note that the @xmath233 are not isomorphic , and hence there is no notion of algebraic convergence for these groups . in the case that we have a sequence of representations in @xmath68 , the following result of jrgensen and marden
is extremely useful .
let @xmath162 be a sequence in @xmath50 converging to @xmath22 ; then , there is a subsequence of @xmath162 , again called @xmath162 , so that @xmath240 converges geometrically to a kleinian group @xmath179 containing @xmath21 .
[ alg - geom - conv ] a sequence @xmath162 in @xmath47 converges _ strongly _ to @xmath22 if @xmath162 converges algebraically to @xmath22 and if @xmath240 converges geometrically to @xmath21 .
note that we may consider @xmath47 , and @xmath50 , to be endowed with topology of strong convergence , instead of the topology of algebraic convergence .
we also refer the reader to the recent article of mcmullen @xcite , in which a variant of the notion of strong convergence is explored in a somewhat more general setting .
generalizing the behavior of the sequence of loxodromic cyclic groups described above , examples of sequences @xmath162 in @xmath47 which converge algebraically to @xmath22 and for which @xmath240 converges geometrically to a kleinian group @xmath5 properly containing @xmath21 have been constructed by a number of authors , including thurston @xcite , @xcite , kerckhoff and thurston @xcite , anderson and canary @xcite , ohshika @xcite , and brock @xcite , @xcite , among others .
jrgensen and marden @xcite carry out a very detailed study of the relationship between the algebraic limit and the geometric limit in the case when the geometric limit is assumed to be geometrically finite . in general , not much is known about the relationship between the algebraic and geometric limits of a sequence of isomorphic kleinian groups .
we spend the remainder of this section discussing this question .
a fundamental point in understanding how algebraic limits sit inside geometric limits is the following algebraic fact , which is an easy application of jrgensen s inequality .
let @xmath162 be a sequence in @xmath47 which converges to @xmath22 and for which @xmath240 converges geometrically to a kleinian group @xmath179 containing @xmath21 .
then , for each @xmath241 , the intersection @xmath242 is either trivial or parabolic cyclic .
[ intersection ] one of the first applications of this result , also in @xcite , was to show , when the algebraic limit is a maximal cusp , that the convex hull of the quotient @xmath0manifold corresponding to the algebraic limit embeds in the quotient @xmath0manifold corresponding to the geometric limit .
this was part of a more general attempt to understand the relationship between the volume and the rank of homology for a finite volume hyperbolic @xmath0manifold .
another application was given by anderson and canary @xcite . before stating the generalization , we need to give a definition . given a kleinian group @xmath5 ,
consider its associated @xmath0manifold @xmath243 , where @xmath6 is the domain of discontinuity of @xmath5 .
then , @xmath5 has connected limit set and no accidental parabolics if and only if every closed curve @xmath12 in @xmath19 which is homotopic to a curve of arbitrarily small length in the interior of @xmath15 with the hyperbolic metric , is homotopic to a curve of arbitrarily small length in @xmath19 , with its induced metric .
let @xmath23 be a finitely generated , torsion - free , non - abelian group , let @xmath162 be a sequence in @xmath47 converging to @xmath22 , and suppose that @xmath240 converges geometrically to @xmath179 .
let @xmath244 and @xmath245 , and let @xmath246 be the covering map . if @xmath21 has non - empty domain of discontinuity , connected limit set , and contains no accidental parabolics , then there exists a compact core @xmath15 for @xmath20 such that @xmath247 is an embedding restricted to @xmath15 . [ ac - embedded - core
] one can apply theorem [ ac - embedded - core ] to show that certain algebraically convergent sequences are actually strongly convergent .
this is of interest , as it is generally much more difficult to determine strong convergence of a sequence of representations than to determine algebraic convergence .
let @xmath23 be a finitely generated , torsion - free , non - abelian group and let @xmath162 be a sequence in @xmath47 converging to @xmath22 .
suppose that @xmath248 is purely loxodromic for all @xmath1 , and that @xmath21 is purely loxodromic .
if @xmath249 is non - empty , then @xmath240 converges strongly to @xmath21 .
moreover , @xmath250 converges to @xmath251 .
[ thm - ac - i ]
let @xmath23 be a finitely generated , torsion - free , non - abelian group and let @xmath162 be a sequence in @xmath47 converging to @xmath22 .
suppose that @xmath248 is purely loxodromic for all @xmath1 , that @xmath21 is purely loxodromic , and that @xmath23 is not a non - trivial free product of ( orientable ) surface groups and cyclic groups , then @xmath240 converges strongly to @xmath21 .
moreover , @xmath250 converges to @xmath251 .
[ thm - ac - ii ] both theorem [ thm - ac - i ] and theorem [ thm - ac - ii ] have been generalized by anderson and canary @xcite to kleinian groups containing parabolic elements , under the hypothesis that the sequences are type - preserving .
one reason that strong convergence is interesting is that strongly convergent sequences of isomorphic kleinian groups tend to be extremely well behaved , as one has the geometric data coming from the convergence of the quotient @xmath0manifolds as well as the algebraic data coming from the convergence of the representations . for instance , there is the following theorem of canary and minsky @xcite .
we note that a similar result is proven independently by ohshika @xcite .
let @xmath15 be a compact , irreducible @xmath0manifold and let @xmath162 be a sequence in @xmath252 converging strongly to @xmath22 , where each @xmath253 and @xmath254 are purely loxodromic .
then , @xmath254 is topologically tame ; moreover , for all sufficiently large @xmath1 , there exists a homeomorphism @xmath255 so that @xmath256 .
[ cm - strong ] by combining the results of anderson and canary @xcite and of canary and minsky @xcite stated above , one may conclude that certain algebraic limits of sequences of isomorphic topologically tame kleinian groups are again topologically tame .
there is also the following result of taylor @xcite .
let @xmath23 be a finitely generated , torsion - free , non - abelian group , and let @xmath162 be a sequence in @xmath47 converging strongly to @xmath22 , where each @xmath248 has infinite co - volume .
if @xmath21 is geometrically finite , then @xmath248 is geometrically finite for @xmath1 sufficiently large .
the guiding conjecture in the study of the relationship between algebraic and geometric limits , usually attributed to jrgensen , is stated below .
let @xmath5 be a finitely generated , torsion - free , non - elementary kleinian group , let @xmath162 be a sequence in @xmath257 converging to @xmath22 , and suppose that @xmath258 converges geometrically to @xmath179 . if @xmath22 is type - preserving , then @xmath259 .
[ no - new - parabolics ] as we have seen above , this conjecture has been shown to hold in a wide variety of cases , including the case in which the sequence @xmath162 is type - preserving and the limit group @xmath260 either has non - empty domain of discontinuity or is not a non - trivial free product of cyclic groups and the fundamental groups of orientable surfaces .
there are several numerical quantities associated to a kleinian group @xmath5 ; one is the _ hausdorff dimension
_ @xmath261 of the limit set @xmath8 of @xmath5 , another is the smallest positive eigenvalue @xmath262 of the laplacian on the corresponding hyperbolic @xmath0manifold @xmath14 .
these two functions are closely related ; namely , if @xmath5 is topologically tame , then @xmath263 when @xmath264 , and @xmath265 when @xmath266 .
the relationship between these two quantities has been studied by a number of authors , including sullivan @xcite , bishop and jones @xcite , canary @xcite , and canary , minsky , and taylor @xcite ( from which the statement given above is taken ) .
it is natural to consider how these functions behave on the spaces we have been discussing in this note .
we begin by giving a few topological definitions . a compact , hyperbolizable @xmath0manifold with incompressible boundary is a _
generalized book of @xmath77bundles _ if there exists a disjoint collection @xmath194 of essential annuli in @xmath15 so that each component of the closure of the complement of @xmath194 in @xmath15 is either a solid torus , a thickened torus , or an @xmath77bundle whose intersection with @xmath19 is the associated @xmath267bundle .
an _ incompressible core _ of a compact hyperbolizable @xmath0manifold is a compact submanifold @xmath217 , possibly disconnected , with incompressible boundary so that @xmath15 can be obtained from @xmath217 by adding @xmath268handles .
we begin with a pair of results of canary , minsky , and taylor @xcite which relates the topology of @xmath15 to the behavior of these functions on a well - defined subset of @xmath136 , and show that they are in a sense dual to one another .
let @xmath15 be a compact , hyperbolizable @xmath0manifold .
then , @xmath269 if and only if every component of the incompressible core of @xmath15 is a generalized book of @xmath77bundles ; otherwise , @xmath270 . here , the supremum is taken over all @xmath22 in @xmath136 for which @xmath271 is homeomorphic to the interior of @xmath15 .
let @xmath15 be a compact , hyperbolizable @xmath0manifold which is not a handlebody or a thickened torus .
then , @xmath272 if and only if every component of the incompressible core of @xmath15 is a geneneralized book of @xmath77bundles ; otherwise , @xmath273 . here , the infimum is taken over all @xmath22 in @xmath136 for which @xmath271 is homeomorphic to the interior of @xmath15 .
it is also possible to consider how these quantities behave under taking limits .
we note that results similar to theorem [ hdim - continuous ] have been obtained by mcmullen @xcite , who also shows that the function @xmath274 is not continuous on @xmath275 in the case that @xmath15 is a handlebody .
let @xmath15 be a compact , hyperbolizable @xmath0manifold which is not homeomorphic to a handlebody .
then @xmath276 is continuous on @xmath275 endowed with the topology of strong convergence .
[ hdim - continuous ] recently , fan and jorgenson @xcite have made use of the heat kernel to prove the continuity of small eigenvalues and small eigenfunctions of the laplacian for sequences of hyperbolic @xmath0manifolds converging to a geometrically finite limit manifold , where the convergence is the variant of strong convergence considered by mcmullen @xcite .
there are several functions on @xmath155 which have been studied by bonahon . in order to keep definitions to a minimum , we state his results for geometrically finite @xmath23 , though we note that they hold for a general finitely generated kleinian group @xmath23 . given a representation @xmath22 in @xmath155 ,
recall that the _ convex core _
@xmath277 of @xmath81 is the smallest convex submanifold of @xmath278 whose inclusion is a homotopy equivalence . by restricting the hyperbolic metric on @xmath81 to @xmath279
, we obtain a map @xmath148 from @xmath155 to the teichmller space @xmath280 of the riemann surface @xmath156 .
another function on @xmath155 studied by bonahon , by developing an analog of the schlfli formula for the volume of a polyhedron in hyperbolic space , is the function @xmath282 , which associates to @xmath283 the volume @xmath284 of the convex core @xmath277 of @xmath81 .
it is known that the hausdorff dimension of the limit set is a continuous function on @xmath285 , using estimates relating the hausdorff dimension and quasiconformal dilitations due to gehring and visl @xcite . in some cases ,
it is possible to obtain more analytic information .
let @xmath5 be a convex cocompact kleinian group whose limit set supports an expanding markov partition .
then , the hausdorff dimension of the limit set is a real analytic function on @xmath285 . [ expanding - markov ]
earlier work of bowen @xcite shows that quasifuchsian and schottky groups support such markov partitions .
the following theorem follows by combining these results of bowen and ruelle with a condition which implies the existence of an expanding markov partition , namely that there exists a fundamental polyhedron in @xmath286 for the kleinian group @xmath23 which has the _ even cornered property _ , together with the klein combination theorem .
let @xmath23 be a convex cocompactkleinian group which is isomorphic to the free product of cyclic groups and fundamental groups of @xmath17orbifolds .
then , the hausdorff dimension of the limit set is a real analytic function on @xmath155 .
[ hdim - analytic ] another function one can consider is the injectivity radius of the corresponding quotient hyperbolic @xmath0manifold . for a hyperbolic @xmath0manifold @xmath20 , the _ injectivity radius _
@xmath287 at a point @xmath288 is one - half the length of the shortest homotopically non - trivial closed curve through @xmath289 .
the following conjecture is due to mcmullen .
let @xmath23 be a finitely generated group with @xmath44 generators .
then , there exists a constant @xmath290 so that , if @xmath20 is a hyperbolic @xmath0manifold with fundamental group isomorphic to @xmath23 and if @xmath289 lies in the convex core of @xmath20 , then @xmath291 . kerckhoff and thurston @xcite show that , if @xmath15 is the product of a closed , orientable surface @xmath168 of genus at least @xmath17 with the interval , then there exists a constant @xmath292 so that if @xmath20 is a hyperbolic @xmath0manifold which is homeomorphic to the interior of @xmath15 and if @xmath20 has no cusps , then the injectivity radius on the convex core of @xmath20 is bounded above by @xmath293 .
fan @xcite generalizes this to show that , if @xmath15 is a compact , hyperbolizable @xmath0manifold which is either a book of @xmath77bundles or is acylindrical , then there exists a constant @xmath292 so that , if @xmath20 is any hyperbolic @xmath0manifold homeomorphic to the interior of @xmath15 , then the injectivity radius on the convex core of @xmath20 is bounded above by @xmath293 .
we close by mentioning recent work of basmajian and wolpert @xcite concerning the persistance of intersecting closed geodesics .
say that a kleinian group @xmath5 has the _ spd property _ if all the closed geodesics in @xmath14 are simple and pairwise disjoint . 1 .
there exists a subset @xmath295 of @xmath294 , which is the intersection of a countably many open dense sets , so that @xmath21 has the spd property for every @xmath296 , or 2 .
there exists a pair of loxodromic elements @xmath297 and @xmath298 of @xmath23 so that the closed geodesics in @xmath81 corresponding to loxodromic elements @xmath299 and @xmath300 intersect at an angle constant over all @xmath301 ; in particular , there is no element @xmath301 so that @xmath21 has the spd property . , _ on the ergodic theory at infinity of an arbitrary discrete group of hyperbolic motions _ , from : `` riemann surfaces and related topics '' , i kra and b maskit ( editors ) , annals of math .
studies 97 , princeton university press ( 1981 ) 465496 | we give a brief overview of the current state of the study of the deformation theory of kleinian groups .
the topics covered include the definition of the deformation space of a kleinian group and of several important subspaces ; a discussion of the parametrization by topological data of the components of the closure of the deformation space ; the relationship between algebraic and geometric limits of sequences of kleinian groups ; and the behavior of several geometrically and analytically interesting functions on the deformation space . |
the standard picture of cosmological structure formation suggests that any visible object forms in a gravitational potential of _ dark matter halos_. therefore , a detailed description of dark halo clustering is the most basic step toward understanding the clustering of visible objects in the universe . for this purpose , many theoretical models for halo clustering have been developed and then tested against extensive numerical simulations .
first , i will describe our most recent theoretical model for clustering of dark matter halos ( hamana , yoshida , suto & evrard 2001b ) .
in particular , we focus on their high - redshift clustering where the past light - cone effect is important
. then i will show that our model predictions are in good agreement with the result from a light - cone output of the hubble volume simulation ( evrard et al .
finally i will discuss a fundamental difficulty in relating the halo model to clusters of galaxies .
my conclusion is that we already have a reliable empirical model for the halo clustering , but that we need to understand what are the clusters of galaxies , especially at high redshifts , before attempting _
precision cosmology _ with clusters of galaxies .
as emphasized by suto et al.(1999 ) , for instance , observations of high - redshift objects are carried out only through the past light - cone defined at @xmath2 , and the corresponding theoretical modeling should properly take account of a variety of physical effects which are briefly summarized below .
assuming the cold dark matter ( cdm ) paradigm , the linear power spectrum of the mass density fluctuations is computed by solving the boltzmann equation for systems of cdm , baryons , photons and ( usually massless for simplicity ) neutrinos . the resulting spectrum @xmath3 in real space is specified by a set of cosmological parameters including the density parameter @xmath4 , the baryon density parameter @xmath5 , and the hubble constant @xmath6 in units of 100/km / s / mpc , and the cosmological constant @xmath7 .
then one can obtain its nonlinear counterpart in real space , @xmath8 by adopting a fitting formula of peacock & doods ( 1996 ) .
the most important ingredient in describing the clustering of halos is their biasing properties .
the mass - dependent halo bias model was developed by mo & white ( 1996 ) on the basis of the extended press - schechter theory . subsequently jing ( 1998 ) and
sheth & tormen ( 1999 ) improved their bias model so as to more accurately reproduce the mass - dependence of bias computed from @xmath0-body simulation results .
we construct an improved halo bias model of the two - point statistics which reproduces the scale - dependence of the taruya & suto ( 2000 ) bias correcting the mass - dependent but scale - independent bias of sheth & tormen ( 1999 ) on linear scales as follows : @xmath9^{0.15 } , \\
b_{\scriptscriptstyle\rm st}(m , z ) & = & 1 + \frac{\nu-1}{\delta_c(z ) } + \frac{0.6}{\delta_c(z)(1 + 0.9\nu^{0.3})},\end{aligned}\ ] ] for @xmath10 , where @xmath11 is the virial radius of the halo of mass @xmath12 at @xmath13 .
in order to incorporate the halo exclusion effect approximately , we set @xmath14 for @xmath15 . in the above expressions , @xmath16 is the mass variance smoothed over the top - hat radius @xmath17 , @xmath18 is the mean density , @xmath19 , @xmath20 is the linear growth rate of mass fluctuations , and @xmath21 ^ 2 $ ] . in linear theory of gravitational evolution of fluctuations ,
any density fluctuations induce the corresponding peculiar velocity field , which results in the systematic distortion of the pattern of distribution of objects in redshift space ( kaiser 1987 ) .
in addition , virialized nonlinear objects have an isotropic and large velocity dispersion . this _ finger - of - god _ effect
significantly suppresses the observed amplitude of correlation on small scales . with those effects , the nonlinear power spectrum _ in redshift space _
is given as @xmath22^{2 } d_{\rm vel}[k\mu\sigma_{\rm halo } ] , \label{nonlinear}\ ] ] where @xmath23 is the fourier transform of the pairwise peculiar velocity distribution function ( e.g. , magira et al . 2000 ) , @xmath24 is the direction cosine in @xmath25-space , @xmath26 is the one - dimensional _ pair - wise _ velocity dispersion of halos , and @xmath27 while both @xmath26 and @xmath28 depend on the halo mass @xmath12 , separation @xmath29 , and @xmath13 in reality , we neglect their scale - dependence in computing the redshift distortion , and adopt the halo number - weighted averages : @xmath30 where we adopt the halo mass function @xmath31 fitted by jenkins et al .
the value of @xmath32 , the halo center - of - mass velocity dispersion at @xmath2 , is modeled following yoshida , sheth & diaferio ( 2001 ) .
then our empirical halo bias model can be applied to the two - point correlation function of halos at @xmath13 in redshift space as @xmath33 all cosmological observations are carried out on a light - cone , the null hypersurface of an observer at @xmath2 , and not on any constant - time hypersurface . thus clustering amplitude and shape of objects should naturally evolve even _ within _ the survey volume of a given observational catalogue . unless restricting the objects at a narrow bin of @xmath13 at the expense of the statistical significance , the proper understanding of the data requires a theoretical model to take account of the average over the light cone ( matsubara , suto , & szapudi 1997 ; mataresse et al .
1997 ; moscardini et al .
1998 ; nakamura , matsubara , & suto 1998 ; yamamoto & suto 1999 ; suto et al .
according to the present prescription , the two - point correlation function of halos on the light - cone is computed by averaging over the appropriate halo number density and the comoving volume element between the survey range @xmath34 : @xmath35 where @xmath36 is the comoving volume element per unit solid angle . while the above expression assumes a mass - limited sample for simplicity , any observational selection function
can be included in the present model fairly straightforwardly ( hamana , colombi & suto 2001a ) once the the relation between the luminosity of the visible objects and the mass of the hosting dark matter halos is specified .
figure 1 compares our model predictions with the clustering of simulated halos from `` light - cone output '' of the hubble volume @xmath37cdm simulation ( evrard et al .
2001 ) with @xmath38 , @xmath39 , @xmath40 , @xmath41 and @xmath42 . for the dark matter correlation functions ,
our model reproduces the simulation data almost perfectly at @xmath43mpc ( see also hamana et al .
this scale corresponds to the mean particle separation of this particular simulation , and thus the current simulation systematically underestimates the real clustering below this scale especially at @xmath44 .
our model and simulation data also show quite good agreement for dark halos at scales larger than @xmath45mpc .
below that scale , they start to deviate slightly in a complicated fashion depending on the mass of halo and the redshift range .
this discrepancy may be ascribed to both the numerical limitations of the current simulations and our rather simplified model for the halo biasing .
nevertheless the clustering of _ clusters _ on scales below @xmath45mpc is difficult to determine observationally anyway , and our model predictions differ from the simulation data only by @xmath46 percent at most . therefore we conclude that in practice our empirical model provides a successful description of halo clustering on the light - cone .
with the successful empirical model of halo clustering , the next natural question is how to apply it in describing _ real _ galaxy clusters . in fact , in my opinion the main obstacle for that purpose is the lack of the universal definition of clusters .
let me give some examples that i can easily think of ( see fig.2 ) .
( i ) press - schechter halos : : : almost all theoretical studies adopt the definition of dark matter halos according to the nonlinear spherical model .
this is characterized by the mean overdensity of @xmath47 ( in the case of the einstein - de sitter universe .
the corresponding expressions in other cosmological models can be also derived . ) . combining this definition with the press - schechter theory ,
the mass function of the dark halos can be computed analytically .
this makes it fairly straightforward to compare the predictions in this model with observations , and therefore this definition has been widely studied in cosmology .
( ii ) halos identified from n - body simulations : : : in reality the gravitationally bound objects in the universe quite often show significant departure from the spherical symmetry .
such non - spherical effects can be directly explored with n - body simulations . even in this methodology , the identification of dark halos from the simulated particle distribution is somewhat arbitrary .
a most conventional method is the friend - of - friend algorithm . in this algorithm ,
the linking length is the only adjustable parameter to controle the resulting halo sample .
its value is usually set to be 0.2 times the mean particle separation in the whole simulation which _ qualitatively _ corresponds to the overdensity of @xmath47 as described above .
( iii ) abell clusters : : : until recently most cosmological studies on galaxy clusters have been based on the abell catalogue . while this is a really amazing set of cluster samples ,
the eye - selection criteria applied on the palomar plates are far from objective and can not be compared with the above definitions in a quantitative sense .
( iv ) x - ray clusters : : : the x - ray selection of clusters significantly improves the reliability of the resulting catalogue due to the increased signal - to - noise and moreover removes the projection contamination compared with the optical selection .
nevertheless the quantitative comparison with halos defined according to ( i ) or ( ii ) requires the knowledge of gas density profile especially in the central part which fairly dominates the total x - ray emission .
( v ) sz clusters : : : the sz cluster survey is especially important in probing the high - z universe . in this case , however , the signal is more sensitive to the temperature profile in clusters than the x - ray selection , and thus one needs additional information / models for temperature in order to compare with the x - ray / simulation results .
the above consideration raises the importance to examine the systematic comparison among the resulting _
cluster / halo _ samples selected differently . in reality , this is a difficult and time - consuming task , and one might argue that we do not have to worry about such _
details _ at this point .
such an optimistic point of view may turn out to be reasonably right after all .
nevertheless it is still important , at present , to keep in mind that this simplistic assumption of `` dark halos = galaxy clusters '' may produce a systematic effect in the detailed comparison between observational data and theoretical models .
i have presented a phenomenological model for clustering of dark matter halos on the light - cone by combining and improving several existing theoretical models ( hamana et al .
one of the most straightforward and important applications of the current model is to predict and compare the clustering of x - ray selected clusters . in doing so ,
however , the one - to - one correspondence between dark halos and observed clusters should be critically examined at some point .
this assumption is a reasonable working hypothesis , but we need more quantitative justification or modification to move on to _ precision cosmology with clusters_. i am afraid that this problem has not been considered seriously simply because the agreement between model predictions and available observations seems already _ satisfactory_. in fact , since current viable cosmological models are specified by a set of many _ adjustable
_ parameters , the agreement does not necessarily justify the underlying assumption .
thus it is dangerous to stop doubting the unjustified assumption because of the ( apparent ) success .
i hope to examine these issues in future .
i would like to thank fred lo for inviting me to this exciting and enjoyable meeting and also for great hospitality at taiwan .
the present work is based on my collaboration with t.hamana , n.yoshida , and a.e.evrard .
this research was supported in part by the grant - in - aid by the ministry of education , science , sports and culture of japan ( 07ce2002 , 12640231 ) .
evrard , a.e .
2001 , apj , submitted hamana , t. , colombi , s. , & suto , y. 2001a , a & a , 367 , 18 hamana , t. , yoshida , n. , suto , y. , & evrard , a.e .
2001b , apj(letters ) , in press jenkins , a. et al .
2001 , mnras , 321 , 372 jing , y.p .
1998 , apj , 503 , l9 kaiser , n. 1987 , mnras , 227 , 1 magira , h. , jing , y.p . , & suto , y. 2000 , apj , 528 , 30 matarrese , s. , coles , p. , lucchin , f. , & moscardini , l. 1997 , mnras , 286 , 115 matsubara , t. , suto , y. , & szapudi , i. 1997 , apj , 491 , l1 mo , h.j .
, & white , s.d.m 1996,mnras , 282 , 347 moscardini , l. , coles , p. , lucchin , & f. , matarrese , s. 1998 , mnras , 299 , 95 nakamura , t. t. , matsubara , t. , & suto , y. 1998 , apj , 494 , 13 peacock , j.a .
, & dodds , s.j .
1996 , mnras , 280 , l19 sheth , r.k . , & tormen , g. 1999 , mnras , 308 , 119 suto , y. , magira , h. , jing , y. p. , matsubara , t. , & yamamoto , k. 1999 , prog.theor.phys.suppl . , 133 , 183 taruya , a. & suto , y . 2000 , apj , 542 , 559 yamamoto , k. , & suto , y. 1999 , apj , 517 , 1 yoshida , n. , sheth , r. , & diaferio , a. 2001 , mnras , in press | a phenomenological model for the clustering of dark matter halos on the light - cone is presented . in particular ,
an empirical prescription for the scale- , mass- , and time - dependence of halo biasing is described in detail .
a comparison of the model predictions against the light - cone output from the hubble volume @xmath0-body simulation indicates that the present model is fairly accurate for scale above @xmath1mpc .
then i argue that the practical limitation in applying this model comes from the fact that we have not yet fully understood what are clusters of galaxies , especially at high redshifts .
this point of view may turn out to be too pessimistic after all , but should be kept in mind in attempting _
precision cosmology _ with clusters of galaxies .
# 1_#1 _ # 1_#1 _ = # 1 1.25 in .125 in .25 in |
diffuse x - ray emission from the solar system was clearly observed during the _ rosat _ all - sky survey ( rass ) as a background component with temporal variations on scales from a large fraction of a day to many days .
these variations were dubbed long - term enhancements ( ltes , @xcite ) , and provided a significant background particularly at @xmath6 kev .
although the origin of these ltes at the time was unknown , the ltes were associated with solar wind parameters @xcite . with the observation of x - rays from comets @xcite , emission from solar wind charge exchange ( swcx ) between solar wind ions and neutral material within the heliosphere
was demonstrated @xcite .
swcx emission subsequently was suggested as the origin for ltes @xcite , and has recently been suggested as being responsible in quiescence for a significant fraction of diffuse x - ray emission at @xmath6 kev @xcite previously attributed to the local hot bubble @xcite .
more recently , evidence for geocoronal emission ( swcx with exospheric material ) has been detected in _
observations of the dark moon @xcite .
this emission clearly originates in the near - earth environment , which is consistent with the likely production of x - rays from the terrestrial magnetosheath @xcite .
the statistics of the dark moon data are somewhat limited but they clearly show excess o vii and o viii emission .
analysis of cometary x - ray spectra suggest that solar wind charge state composition and speed affect the x - ray emission @xcite .
thus through a combination of techniques , including x - ray and neutral atom imaging among others , it may be possible to continuously monitor the solar wind from well inside earth s magnetosphere .
however , the nature of solar system x - ray emission is strongly dependent on the point of view of the observer . while the observation of swcx x - ray emission samples aspects of the solar system of interest to astronomers
, the emission can also provide a contaminating component which strongly impacts observations of extended cosmic x - ray sources .
the swcx emission spectrum is dominated by highly ionized carbon , oxygen , neon , possibly iron , and magnesium lines , which are also of great astrophysical interest .
( the oxygen lines are particularly important as they are commonly used for temperature and density diagnostics of thermal emission from diffuse plasmas . ) for objects at zero redshift ( i.e. , emission from the milky way and nearby galaxies ) where the emission is expected to fill the entire field of view of the detector , swcx emission is indistinguishable from that of the object , except to the extent that temporal variation can be detected . while there is likely to be swcx emission at some quiescent level @xcite , the strongest emission will be associated with flux enhancements of the solar wind .
these enhancements are long enough in duration that they can `` contaminate '' an entire observation , but variable enough that if there are multiple observations of a single target they can be identified , at least to some level .
this was the basis of the lte `` cleaning '' of the rass @xcite , and of _ rosat _ pointed observations which were typically distributed over the period of a few days if not weeks @xcite .
in this paper we report the detection of swcx emission during an _ xmm - newton _ observation of the _ hubble _ deep field north .
we correlate that detection with a concurrent enhancement in the solar wind density observed by advanced composition explorer ( _ ace _ ) , solar and heliospheric observatory ( _ soho _ ) , and _ wind_. in
[ sec : data ] we present the data , data reduction , and analysis , in [ sec : discussion ] we discuss the results , and in
[ sec : conclusions ] we detail our conclusions both in regards to the emission itself and the implications for x - ray observations of more distant objects .
the swcx emission discussed in this paper was observed in the fourth ( 2001 june 12 ) _ xmm - newton _ pointing of the _ hubble _ deep field north ( hdf - n , @xmath7 , a project guaranteed time observation ) .
because of satellite constraints and the total length of the observation , it was broken into four segments ( pointings ) , which were scheduled over a period of 16 days ( table [ tbl : observations ] gives the specifics of the pointings ) .
because of this multiple coverage , the length of the individual pointings , and relative freedom from other non - cosmic backgrounds ( most critically the soft proton flaring ) , the hdf - n observation was chosen as a test case for a new non - cosmic background modeling and subtraction method for _ xmm - newton _ epic data @xcite . the serendipitous detection of the swcx emission came from that testing .
outside of time intervals obviously affected by proton flaring ( which were removed for this analysis ) , the light curves for the observations were reasonably flat indicating at least a low if not zero level of residual flaring .
figure [ fig : lc ] shows the x - ray light curves for the non - flaring periods of the fourth observation , and also includes the _ ace _ data ( discussed below ) .
the light curve for the 28 kev band is relatively constant but the 0.520.75 kev band ( covering the o vii and most of the o viii lines ) shows a significant drop in the last quarter of the interval .
we used this light curve to separate the spectrum into high and low states at @xmath8 s in figure [ fig : lc ] ( 22:39:38 ut on 2001 june 1 ; @xmath9 seconds after 0 ut , 1998 january 1 , in the _ xmm - newton _ time reference system ) .
table [ tbl : observations ] also lists the average count rates in the 0.520.75 kev and 28 kev bands for the different pointings .
the 28 kev band count rate is reasonably consistent among all of the pointings , as is the average count rate for the 0.520.75 kev band , except for the time period affected by the swcx emission .
the minor , but greater - than - statistical scatter of the x - ray count rates in table [ tbl : observations ] during the non - swcx periods is likely due to the contribution of the unsubtracted particle background which is temporally variable , and/or variations in the residual swcx emission .
data from the full field of view were used in this analysis .
they were conservatively filtered by screening the events using the parameters `` flag==0 '' and `` pattern@xmath10=12 '' ( only the best events selected by position and limiting the ccd pixel pattern to legitimate quadruple and lower pixel - number events ) . the non - cosmic background was modeled using the method of @xcite , which is currently limited to the epic mos data ( i.e. , excludes data from the epic pn detector ) .
the nominal non - cosmic background , which is dominated by the quiescent high - energy particle - induced background , is modeled using closed filter data scaled by data from the unexposed corners of the detectors in the individual observation .
the scaling is energy dependent and based on the hardness and intensity of the corner data ( which are temporally variable ) .
this background modeling removes most of the non - cosmic background but unfortunately leaves untouched any backgrounds ( particle or x - ray ) originating in , or passing down the telescope tube ( i.e. , backgrounds which are blocked by the detector mask used in the closed - filter data collection ) .
most notably there are strong fluorescent al k@xmath11 and si k@xmath11 lines , and the possibility of soft protons at some quiescent level . after background subtraction ,
the spectra from the full field of view of the first three hdf - n pointings and the low state of the fourth were reasonably statistically consistent , while the high - state data from the fourth pointing showed a strong enhancement at energies less than 1.4 kev ( consistent with the light curve ) , which was dominated by line emission .
the spectra at higher energies of all four pointings ( including both high and low states of the fourth ) were consistent with each other as expected from their average count rates ( table [ tbl : observations ] ) .
for the spectral fits of the _ xmm - newton _ x - ray data we used spectral redistribution matrices ( rmfs ) and effective areas ( arfs ) produced by the project standard analysis software ( sas ) package ( version 5.4.1 ) .
xspec ( version 11.2 , @xcite ) was used to fit a model of the cosmic x - ray background ( cxb ) , swcx line emission ( a number of discrete lines ) , probable residual soft proton contamination ( represented by a power law not folded through the instrumental efficiency ) , and the two instrumental lines ( al k@xmath11 and si k@xmath11 ) to the data after subtraction of the modeled non - cosmic background . in order to constrain the cosmic contribution to the spectrum , we simultaneously fit rass data provided through the heasarc x - ray background tool , which uses data from @xcite to create xspec compatible spectra for selected directions on the sky .
the spatial distribution of the cxb emission within the observed field , which includes the hdf - n , is relatively simple ( by design there are no significantly bright point sources or extended objects in the field ) , and can be modeled using the four standard components listed in @xcite : a cooler unabsorbed thermal component with @xmath12 kev representing emission from the local hot bubble ( e.g. , @xcite ) , an absorbed cooler thermal component also with @xmath12 kev representing emission from the lower halo , an absorbed hotter thermal component with @xmath13 kev representing milky way halo or local group emission ( e.g. , @xcite ) , and an absorbed power law with a spectral index of 1.46 representing the unresolved emission of agn and other cosmological objects ( e.g. , @xcite ) .
the absorption was fixed to the galactic column density of h i ( @xmath14 @xmath15 ) .
the spectral fitting was done simultaneously for eleven spectra : mos1 and mos2 data from the four observations with two spectra ( high and low states ) for the fourth observation , plus the rass spectrum .
the cxb model was fit to all spectra with no scaling between the different instruments besides the normalization for solid angle , which was fixed .
the instrumental lines and soft - proton power law were fit only to the mos data .
the normalizations were allowed to differ between the mos1 and mos2 detectors but were constrained to be the same for all observations .
the lines representing the swcx were fit only to the mos data from the high state of the fourth observation with the fit parameters constrained to be the same for the mos1 and mos2 data .
figure [ fig : spec ] shows the results with the excess emission attributed to swcx being clearly visible .
the aluminum and silicon instrumental lines are the strong peaks at 1.49 kev and 1.74 kev , respectively .
the swcx emission was modeled by a set of gaussians whose widths were set to zero except for the component at @xmath16 kev which covers multiple lines from o viii and possible lines from fe xvii .
the fitted line strengths are listed in table [ tbl : flux ] .
two gaussian lines with energies less than 0.5 kev were included in the fits , which can be associated with emission from highly ionized carbon ( c vi at 0.37 kev and 0.46 kev ) .
they are included in table [ tbl : flux ] with the following caveats . at energies less than 0.5
kev the energy resolution and effective areas of the mos instruments fall off significantly and while the additional lines are necessary for reasonable fits ( @xmath17 when one line is removed , @xmath18 when both lines are removed ) the specific energies of the lines are not well constrained .
however , there is no degrading of the quality of the fit when the line energies are fixed at the c vi values , and once the energies are fixed fluxes can be measured at better than the 3@xmath19 level .
still , lines at other energies may fit the data as well so the attribution of the emission to c vi is not certain , it is however reasonable ( cf .
note , however , that the addition of the c v line at 0.30 kev ( also listed in @xcite ) did not improve the quality of the fit .
the power - law background component , which is assumed to represent residual soft proton contamination , was added to bring consistency between the rass and mos data . because this power law is best fit without being folded through the mos effective area , it is not due to x - rays originating externally to the satellite as they would be cut off at lower energies by the filter .
the effective flux in this component is @xmath20 ergs @xmath15 sr@xmath21 s@xmath21 while the flux of the cosmic background model is @xmath22 ergs @xmath15 sr@xmath21 s@xmath21 in the 0.25.0 kev band , with @xmath23 ergs @xmath15 sr@xmath21 s@xmath21 from the extragalactic power law .
figure [ fig : fold ] shows the mos1 spectra from the first pointing and the high state of the fourth pointing along with the various components to the fit folded through the instrumental response .
the fit to the data is rather good with @xmath24 with 1922 degrees of freedom .
the fitted values for the energies of the lines were consistent with the expected values in initial fits , and were subsequently fixed ( except for the o viii and the possible fe complex at @xmath16 kev ) . in general it is possible to move power between one spectral component and another in fits of such complex models as used here .
the fitted parameters of the cosmic background model are , for example , strongly correlated .
however , the measurement of the swcx emission is very robust ( as indicated by the quoted 90% probability statistical errors in the line fluxes listed in table [ tbl : flux ] ) as the background is so well constrained and the swcx emission is in a limited number of discrete lines . solar wind speed and flux data from four spacecraft during the period of the hdf - n observations are shown in figure [ fig : sw ] .
three of them , _ ace _ , _ wind _ , and _
sample the near - earth environment .
_ ace _ and _ soho _ have been , for the most part , located in halo orbits around the sun - earth libration point ( l1 , @xmath25 km , 235 @xmath26 from earth ) .
_ wind _ has led a more peripatetic life and at the time of these measurements was no longer in an orbit around l1 .
_ ace _ was launched on 1997 august 25 , _
soho _ was launched on 1995 december 2 , and _ wind _ was launched 1994 november 1 .
the three spacecraft function , in part , as upstream monitors of the solar wind , the hot ( @xmath27 k ) plasma continually flowing from the sun , and provide almost continual measurements of the solar wind density , speed , and magnetic field , among many other parameters . the fourth spacecraft , _ ulysses _ , follows a high - inclination ( @xmath28 ) , elliptical ( @xmath29 with a semi - major axis of @xmath30 au ) orbit around the sun .
_ ulysses _ was launched 1990 october 6 and reached its roughly polar orbit after a fly - by of jupiter . at the time of these measurements , earth was at @xmath31 au , @xmath32 ecliptic latitude , and @xmath33 ecliptic longitude .
the solar wind speed and flux data from the three near - earth spacecraft are in reasonably good agreement showing the same structure ( figure [ fig : sw ] ) .
the major enhancements in the flux at all three spacecraft are roughly contemporaneous while discrepancies in the fluxes , although typically small but sometimes as much as a factor of two , may be in part due to the sometimes large separation ( @xmath34 ) between the spacecraft perpendicular to the sun - earth line .
( note that the deviation of _ ace _ near day 9 in figure [ fig : sw ] is due to missing data . ) on the other hand _
ulysses _ , which is in a polar orbit around the sun , at this time was @xmath35 au from the sun with an ecliptic latitude of @xmath36 and an ecliptic longitude of about @xmath37 .
this location is farther from the sun than the other spacecraft and @xmath38 ahead of the earth .
thus , the ulysses flux in figure [ fig : sw ] is only qualitatively similar to that observed near earth and shifted in time .
these differences reflect the large scale structure of the interplanetary medium at this time . in addition , the _ ulysses _ direction was @xmath39 away from the hdf - n , and so is not particularly relevant for the swcx observation , but the comparison of the data provides an illustration of the variation of the solar wind from a uniform shell - like expansion .
although there exists considerable variability , solar wind speeds are typically about 450 km s@xmath21 ( about 1 kev / nucleon ) with fluxes at 1 au of about @xmath40 @xmath15 s@xmath21 @xcite . the solar wind by is about 95% protons and 4% helium nuclei by number .
high charge state heavy ions ( i.e. , other than protons and he@xmath41 ) comprise less than 1% of the solar wind by number .
these ions include carbon , dominated by charge states c@xmath42 and c@xmath5 , oxygen , dominated by charge states o@xmath5 and o@xmath4 , neon dominated by charge state ne@xmath43 , and iron with a wide range of charge states sometimes as high as fe@xmath44 @xcite .
the solar wind source ion for the c vi lines in the x - ray spectrum is c@xmath5 and the source ions for the o vii and o viii lines are o@xmath4 and o@xmath43 , respectively .
although typically the dominant charge states of solar wind oxygen are + 6 and + 7 , under certain conditions , e.g. , in the cme - related solar wind ( coronal mass ejection ) , o@xmath43 can be comparable to and even dominate the o@xmath4 @xcite . ne@xmath45 and mg@xmath46 , the solar wind source ions for ne
ix and mg xi swcx emission , have also been observed in the solar wind ( a. b. galvin , private communication ) .
o viii and ne ix lines have been observed in cometary spectra @xcite during the period of the 2001 june 1 - 2 _ xmm - newton _ hdf - n pointing , the _ ace _ level 2 data show that the o@xmath4/o@xmath5 ratio was highly variable ( figure [ fig : lc ] ) compared to the first three pointings .
the average ratio was enhanced by over a factor of two during the period when the swcx emission was detected , and dropped by nearly an order of magnitude at the end of the pointing concurrent with the reduction in solar wind flux ( table [ tbl : ace ] ) .
this large increase in the relative abundance of the higher ionization state during the swcx period might suggest that the relative abundances of the ionization states of o@xmath43 , ne@xmath45 , and mg@xmath46 may also be significantly enhanced .
thus observation of swcx emission from these species would be reasonable .
however , preliminary results from _ ace _ indicate that the o@xmath43/o@xmath4 ratio was actually relatively low at @xmath47 during the period of swcx emission ( zurbuchen & raines , private communication ; the analysis continues ) .
the picture becomes even more murky as the preliminary results also show the o@xmath43/o@xmath4 ratio increasing by a factor of 20 to 30 just as the swcx emission and solar wind flux enhancement were cutting off .
in addition , the ( 0.720.61 kev)/(0.520.61 kev ) count rate ratio , which is a measure of the relative emission of o viii and o vii , was consistent with a constant value ( @xmath48 for 22 degrees of freedom ) over the period of the swcx emission .
this is in spite of the variation in solar wind ionization state distributions .
the temporal variation of the x - ray data requires that the enhancement originates locally , at least in the solar system if not in the satellite / near - earth environment .
the presence of x - ray emission lines corresponding to the highly ionized charge states of oxygen , neon , and magnesium eliminates an origin internal to the satellite , and strongly implicates swcx as the source of the emission enhancement .
furthermore , data from the low energy neutral atom _ lena
_ imager on the _ image _ spacecraft inside the magnetosphere during this event show an enhancement which suggests a neutral solar wind resulting from swcx @xcite .
the geometry of the 2001 june 12 observation with respect to the near - earth environment is shown in figure [ fig : geometry ] .
the satellite is on the sun side of the earth with an angle of @xmath49 between the sun and look directions .
( due to satellite constraints , the look direction of _ xmm - newton _ must lie within @xmath50 of perpendicular to the earth - sun line . )
the satellite and the observation line of sight are outside of earth s magnetosphere , based on the magnetospheric model of @xcite ( using a ram pressure of 7.4 npa and a northward @xmath51 ) . however , the line of sight may possibly pass through the magnetosheath relatively near the sub - solar point .
such an observation trajectory would sample the region of strongest production of x - rays by swcx with exospheric material @xcite . on the other hand
, the trajectory may be entirely outside of the magnetosheath and therefore miss much of that emission .
because of this uncertain geometry , the swcx emission can be from either exospheric material in earth s magnetosheath or from interstellar neutrals in the heliosphere , or more likely both ( at least to some level )
. the detailed analysis of this geometry will be the subject of a following paper . on a more global scale ( with perhaps no particular relevance to this observation ) , during the time period of the 2001 june 12 observation
the location of earth in its orbit placed it almost directly upstream from the sun relative to the flow of interstellar material through the solar system .
( earth is in the nominal upstream direction about 4 june . ) from table [ tbl : flux ] and figure [ fig : spec ] it is clear that the excess emission can be well characterized by a limited number of highly significant lines with energies consistent with those expected from swcx . with certain assumptions it is possible to use the observed x - ray line ratios to derive the concurrent ion ratios .
the fitted o viii(0.65&0.81 kev)/o vii(0.56 kev ) line ratio from table [ tbl : flux ] is 1.09 .
however , from @xcite , the @xmath52 and ( at the 2% level ) o vii @xmath53 lines contaminate the o viii line at @xmath54 kev , and contribute @xmath55 @xmath15 sr@xmath21 s@xmath21 ( using a factor of 0.079 , derived from the relative intensities of the relevant lines in their table 1 , to scale from the o vii 0.56 kev flux ) . correcting for this contribution , and assuming that the 0.81 kev emission is all from o viii ( see below ) , the o viii(0.65&0.81 kev)/o vii(0.56&0.67 kev ) emission ratio is then @xmath56 ( table [ tbl : oxygen ] ) . to convert the emission ratio to an ion number ratio we use the swcx cross sections in @xcite .
the ratios between the hydrogen and helium swcx cross sections for o@xmath4 and o@xmath43 are roughly the same , so we use the o@xmath43/o@xmath4 hydrogen cross section ratio of 1.66 to scale the flux ratio ( this assumes that the scattering is optically thin and hydrogen is the dominant swcx target neutral ) .
this implies a solar wind o@xmath43/o@xmath4 abundance ratio of @xmath57 , which is somewhat higher than the more typical value of 0.35 ( e.g. , @xcite ) , but not unreasonable , as would be expected considering the enhancement of the o@xmath4/o@xmath5 ratio .
however , as noted above , the preliminary o@xmath43/o@xmath4 ratio from _ ace _ for the swcx period is @xmath47 , a significant discrepancy which we will revisit in a subsequent paper when the o@xmath43/o@xmath4 ratio data are more robust .
the energy and non - zero width for the fitted line at @xmath16 kev is consistent with the expected o viii ly@xmath58 through ly@xmath59 emission ( @xcite ) .
however , this is also the energy of possible fe xvii emission .
the ratio of the line strengths , ly@xmath11/ly(@xmath60 ) , predicted by @xcite is 2.75 ( with a @xmath61 uncertianty ) , while our measured ratio is @xmath62 .
the two values are therefore in reasonably good agreement but the result does not rule out a contribution from fe xvii .
it does suggest that any fe contribution to the observed flux is relatively limited . as measured by _
ace _ , the solar wind during the 2001 june 12 period was in its `` slow '' state but with with fairly uncharacteristic ion abundances and flux during the enhancement period . at solar maximum
, the solar wind at 1 au shows a complex structure most likely resulting from the appearance of equatorial coronal holes .
in contrast , at solar minimum , there tends to be a two state speed structure observed in the solar wind with fewer but well - formed transients such as coronal mass ejections and corotating interaction regions .
@xcite provide a recent review of the complex variation of the solar wind during the solar cycle .
solar wind fluxes during the first three _ xmm - newton _ hdf - n observations were close to nominal levels ( see figure [ fig : sw ] ) .
however , during the observation on 2001 june 1 - 2 , the solar wind flux observed by the ace spacecraft was enhanced ( as shown in figures [ fig : lc ] and [ fig : sw ] ) by a factor of five over its typical value of @xmath63 @xmath15 s@xmath21 .
the average o@xmath4/o@xmath5 ionization state ratios for the observation ( table [ tbl : ace ] ) were also significantly enhanced .
this flux enhancement was relatively localized ; it lasted 5 - 6 hours at a solar wind speed of @xmath64 km s@xmath21 implying a width of about 0.05 au in the direction toward the sun .
the x - ray flux decrease at @xmath65 ( 2001 june 1 , 22:30 ut ) and the sharp drop in both the solar wind flux and o@xmath4/o@xmath5 ratio propagated to earth from the l1 point ( a delay of @xmath66 s ) are approximately contemporaneous , which suggests the x - ray flux is responding primarily to local solar wind conditions .
however , comparison of the solar wind flux and ionization state variation and x - ray light curves raises two critical questions .
first , if the swcx emission enhancement observed by _ xmm - newton _ is due to a local phenomena , why is the large variation in solar wind flux not reflected in the x - ray intensity , or in a variation of the relative intensity of the o vii and o viii emission ?
over the course of 20 ks ( roughly six hours ) the solar wind flux increases by a factor of six while the x - ray light curve is essentially constant ( figure [ fig : lc ] ) .
if the swcx emission was due solely to interactions with material from earth s exosphere , then the x - ray light curve should be closely linked to the solar wind flux measured by _ ace _ at the l1 point .
@xcite point out that the geometry of the magnetosheath is strongly affected by the density and velocity of the solar wind . with the observation geometry considered by @xcite , i.e. , looking out through the flanks of the magnetosheath , there is some buffering of the x - ray emission which smooths out the effects of strong solar wind enhancements .
however , with the geometry of this observation where the line of sight possibly runs tangentially through their region of maximum x - ray emission near the magnetosheath sub - solar point , the situation is much more confused , and perhaps more likely to produce variations rather than to smooth out the swcx flux . given the possible difficulty in turning the large variation of the solar wind flux into a relatively constant x - ray light curve , a possible solution might be to distribute the emission along the line of sight through the heliosphere .
this would smooth out temporal variations originating with the solar wind .
however , this brings up the second question . if the emission is distributed through the heliosphere why do the solar wind and x - ray enhancements cut off at nearly the same time ?
note again the large differential variation in the solar wind flux as measured by _
ace _ and _ ulysses _ , up to a factor of 20 , during swcx observation .
figure [ fig : geometry2 ] illustrates the observation geometry on a larger scale . since the pointing direction is @xmath49 from the earth - sun line , the observation samples a relatively long path length through the solar wind enhancement with a heliocentric distance similar to that of the earth . if the solar wind enhancement is a spherically symmetric shell moving outward there should still be a strong enhancement in the swcx emission roughly concurrent with the _ ace _ measurement .
however , even introducing large variations in the density profile and arrival time of the solar wind enhancement as a function of direction from the sun will not in general smooth out the x - ray light curve significantly .
in addition , once the position along the line of sight is greater than 1 au from the sun ( the short dashed part of the line of sight in figure [ fig : geometry2 ] ) , it is sampling the solar wind enhancement after it would have passed earth , and therefore the x - ray count rate should fall off after the fall - off of the solar wind enhancement as measured at earth .
one might argue that such a fall - off could slow down and therefore not be obvious in the relatively limited lower flux part of the 0.520.75 kev light curve .
however , the absence of o viii emission during that period indicates that the swcx emission really has cut off .
the quandary is therefore :
1 ) if the x - ray emission is from the magnetosheath why is it so constant ?
2 ) if the x - ray emission is distributed through the heliosphere why does the emission fall off concurrently with the end of the solar wind enhancement , and why ( again ) is the swcx emission light curve so constant ? the episode of swcx emission reported here lasted for at least 38 ks ( @xmath67 hours ) which would have extended over seven _ rosat _ orbits .
this is sufficiently long enough to be classified as a lte in the rass processing .
however , this swcx episode has considerably more @xmath68 kev flux than ltes observed during the rass .
the fluxes listed in table [ tbl : flux ] would have produced a _
_ pspc count rate of @xmath69 counts arcmin@xmath70 s@xmath21 in the r45 ( @xmath68 kev ) band , only slightly less than twice the typical intensity of the cosmic background at high galactic latitudes .
most ltes were dominated by emission in the r12 ( @xmath6 kev ) band and the brightest @xmath68 kev band lte enhancements were only roughly half of the intensity of this detection .
the data from the _ xmm - newton _ observation of the _ hubble _ deep field north show clear evidence for a time - variable component of the x - ray background that is most reasonably attributed to charge exchange emission between the highly ionized solar wind and exospheric or interplanetary neutrals .
the swcx emission is dominated by the lines from c vi , o vii , o viii , ne ix , and mg xi , and possibly lines from highly ionized iron ( e.g. , fe xvii ) . the emission is concurrent with the passage of a strong enhancement in the solar wind observed by the _ ace _ satellite , and which is also strongly enhanced in its o@xmath4/o@xmath5 ratio , and possibly other highly ionized species ( although preliminary results for the o@xmath43/o@xmath4 ratio indicate otherwise ) .
however , the light curve of the x - ray enhancement is relatively constant while the flux of the solar wind enhancement varies considerably .
the observation of swcx emission allows the monitoring of interactions between the solar wind and solar system and/or interstellar neutrals without the need for _ in situ _
measurements , albeit with the uncertainty of just where along the line of sight the emission arises .
however , in certain circumstances the distance ambiguity can be resolved allowing the detailed study of the related phenomena .
two such situations are mentioned above : the case of swcx emission from comets and exospheric material in the _ chandra _ dark moon observation .
the ability to remotely sense the solar wind and its interactions can also be used when the emission is expected to have a predictable variation , such as an x - ray scan across the magnetopause sub - solar point @xcite . also ,
for example , _ rosat _ data from the all - sky survey suggest that the downstream helium focusing cone can be imaged in x - rays ( e.g. , @xcite ) .
the charge exchange line emission can provide a significant contaminating background to observations of more distant objects ( by any x - ray observatory with sensitivity at energies less than 1.5 kev ) which use those lines for diagnostics .
however , the correlation of the x - ray enhancement with the solar wind density enhancement suggests a diagnostic .
for those observations at risk because of such contamination , e.g. , observations of sources which cover the entire field of view and have thermal spectra , the data from solar wind monitoring observatories such as _ ace _ , _ wind _ , and _ soho _ should be used for screening .
should the swcx emission be primarily exospheric in origin ( e.g. , from the magnetosheath ) , then the specific geometry of the observation should be examined as well .
an observation where the observatory is outside of the magnetosheath and looking away from earth would clearly have significantly less exospheric swcx emission than the case of the emission reported in this paper where the observation line of sight may be passing tangentially through the magnetopause near the sub - solar point . while swcx emission is likely responsible for the ltes observed during the _ rosat _ all - sky survey , the swcx episode presented here was significantly brighter at @xmath68 kev than what was typically observed ( a factor of two brighter than the most intense @xmath68 kev rass lte ) .
however , one important issue to note is that none of the x - ray emission lines discussed here ( or discussed in @xcite ) contribute significantly to the @xmath6 kev band , where most of the lte emission was observed . also of note
is that the geometry of this observation relative to earth s geomagnetic environment is considerably different from that of the rass . in this case
we may be looking tangentially through the magnetosheath near the sub - solar point while _ rosat _ with its low circular orbit effectively looked radially outward through the flanks of the magnetosheath .
we are grateful to the _ ace _ swepam and swics / swims instrument teams and the _ ace _ science center for providing the _ ace _ data , to _
wind_/swe team for providing the _ wind _ data , to d. j. mccomas , m. lancaster , and c. tranquille for providing the _ ulysses_/swoops , and to _ soho _
team for providing the celias / mtof pm data .
we would also like to thank m. bzowski , s. christon , t. cravens , a. galvin , c. lisse , b. pilkerton , j. raines , i. robertson , d. simpson , a. szabo , b. wargelin , and t. zurbuchen for useful discussions
. we also thank b. wargelin for providing a copy of @xcite to us before submission and the referee for their comments .
finally , we would like to thank the referee for their comments .
this work is based on observations obtained with _ xmm - newton _ , an esa science mission with instruments and contributions directly funded by esa member states and the usa ( nasa ) .
the data were provided through the heasarc _ xmm - newton _ archive at nasa / gsfc .
arnaud , k. a. 2001 , asp conference proceedings , 238 , 415 chen , l .- w . ,
fabian , a. c. , & gendreau , k. c. 1997 , mnras , 285 , 449 cox , d. p. 1998
, lecture notes in physics , ( berlin : springer verlag ) , 506 , 121 collier , m. r. , hamilton , d. c. , gloeckler , g. , bochsler , p. , sheldon , & r. b. 1996 , , 23 , 1191 collier et al .
2001 , , 106 , 24893 collier , m. r. , snowden , s. , moore , t. e. , simpson , d. , pilkerton , b. , fuselier , s. , & wurz , p. 2003 , eos trans .
agu , 84(46 ) , fall meet .
suppl . , abstract sh11a-1123 cravens , t. e. 1997 , , 24 , 105 cravens , t. e. 2000 , , 532 , l153 freyberg , m. j. , 1994 , ph.d .
thesis , teschnische universitt mnchen galvin , a. b. 1997 , geophysical monograph 99 , ed .
n. crooker , j. joselyn , & j. feynman , ( agu : washington , d.c . ) , 253 gloeckler et al . 1999 ,
lett . , 26 , 157 kharchenko , v. , & dalgarno , a. , 2000 , , 105(a8 ) , 18,351 kharchenko , v. , rigazio , m. , dalgarno , a. , & krasnopolsky , v. a. 2003 , , 585 , l73 kuntz , k. , & snowden , s. l. 2000 , , 543 , 195 kuntz , k. , et al .
2004 , , in preparation lallement , r. 2004 , , in press lisse , c. m. et al .
1996 , science , 274 , 205 lisse , c. m. , christian , d. j. , dennerl , k. , meech , k. j. , petre , r. , weaver , h. a. , and wolk , s. j. 2001 , science , 292 , 1343 mccammon , d. , et al .
2002 , , 576 , 188 petrinec , s. m. , & russell , c. t. 1996 , , 101 , 137 richardson , j. d. , & paularena , k. i. 2001 , , 106 , 239 robertson , i. p. , & cravens , t. e. 2003a , geophys . res .
lett . , 30(8 ) , 1439 robertson , i. p. , & cravens , t. e. 2003b , , 108 ( a10 ) , 8031 rucinski , d. , cummings , a. c. , gloeckler , g. , lazarus , a. j. , mobius , e. , & witte , m. 1996 , space sci .
, 78 , 73 schwadron , n. a. , & cravens , t. e. 2000 , , 544 , 558 smith , e. j. et al .
2003 , science , 302 , 1165 snowden , s. l. , mccammon , d. , burrows , d. n. , & mendenhall , j. a. 1994 , , 424 , 714 snowden , s. l. , freyberg , m. j. , schmitt , j. h. m. m. , voges , w. , trmper , j. , edgar , r. j. , mccammon , d. , plucinsky , p. p. , & sanders , w. t. 1995 , , 454 , 643 snowden , s. l. , egger , r. , freyberg , m. j. , mccammon , d. , plucinsky , p. p. , sanders , w. t. , schmitt , j. h. m. m. , trmper , j. , & voges , w. 1997 , , 485 , 125 snowden , s. l. , egger , r. , finkbeiner , d. p. , freyberg , m. j. , & plucinsky , p.
, , 493 , 715 wargelin , b. j. , markevitch , m. , juda , m. , kharchenko , v. , edgar , r. j. , & dalgarno , a. 2004 , , submitted ccccccc 0111550101 & 2001 - 05 - 18t08:46:57 & 46.4 & 39.5 & @xmath71 & @xmath72 + 0111550201 & 2001 - 05 - 18t22:17:34 & 46.6 & 35.1 & @xmath73 & @xmath74 + 0111550301 & 2001 - 05 - 27t06:15:22 & 52.6 & 13.8 & @xmath75 & @xmath76 + 0111550401 & 2001 - 06 - 01t08:16:36 & 95.4 & 38.1 & @xmath77 & @xmath78 + 0111550401 & & & 17.2 & @xmath79 & @xmath80 + ccccccc c vi & 0.37 & @xmath81 & 5.3 + c vi & 0.46 & @xmath82 & 2.1 + o vii & 0.56 & @xmath83 & 6.66 + o viii & 0.65 & @xmath84 & 6.84 + o viii & 0.81 & @xmath85 & 1.96 + ne ix & 0.91 & @xmath86 & 1.24 + mg xi & 1.34 & @xmath87 & 0.85 + | we present an _
xmm - newton _ spectrum of diffuse x - ray emission from within the solar system .
the spectrum is dominated by probable c vi lines at 0.37 kev and 0.46 kev , an o vii line at 0.56 kev , o viii lines at 0.65 kev and @xmath0 kev , ne ix lines at @xmath1 kev , and mg xi lines at @xmath2 kev .
this spectrum is consistent with that expected from charge exchange emission between the highly ionized solar wind and either interstellar neutrals in the heliosphere or material from earth s exosphere .
the emission is clearly seen as a low - energy ( @xmath3 kev ) spectral enhancement in one of a series of four observations of the _ hubble _ deep field north .
the x - ray enhancement is concurrent with an enhancement in the solar wind measured by _ ace _ ,
_ wind _ , and _ soho _ spacecraft .
the solar wind enhancement reaches a flux level an order of magnitude more intense than typical fluxes at 1 au , and has a significantly enhanced o@xmath4/o@xmath5 ratio .
besides being of interest in its own right for studies of the solar system , this emission can have significant consequences for observations of cosmological objects .
it can provide emission lines at zero redshift which are of particular interest in studies of diffuse thermal emission ( e.g. , o vii and o viii ) , and which can therefore act as contamination in the spectra of objects which cover the entire detector field of view .
we propose the use of solar wind monitoring data as a diagnostic to screen for such possibilities . |
among non - equilibrium dynamical systems , living organisms are the most challenging ones that scientists can study .
a biological system constantly exchanges material , energy and information with the environment as it regulates its growth and survival .
the energy and chemical balances at the cellular level involve an intricate interplay between the microscopic dynamics and the macroscopic environment , through which life at the intermediate mesoscopic scale is maintained @xcite .
the development of a multicellular structure requires non - equilibrium dynamics , as microscopic imbalances are translated into the macroscopic gradients that control collective action and growth @xcite .
much effort is devoted to the search for basic principles of organization ( growth , communication , regulation and control ) on the cellular and multicellular levels @xcite .
our approach is to use the successful conceptual framework for pattern formation in non - living systems as a tool to unravel their significantly more complex biological counterparts . of critical importance
is the choice of starting point , i.e. the choice of which phenomena to study : it has to be simple enough to allow progress , but also well motivated by the significance of the results .
cooperative microbial behavior is well suited for these requirements , as we explain below .
we focus on two examples : the branching growth in bacterial colonies of , and the chiral growth of
. examples of these patterns are shown in fig.[fig : intro ] .
traditionally , bacterial colonies are grown on substrates with a high nutrient level and intermediate agar concentration @xcite .
such `` friendly '' conditions yield colonies of simple compact patterns , which fit well the contemporary view of bacterial colonies as a collection of independent unicellular organisms ( non - interacting `` particles '' ) . however
, bacterial colonies in nature must regularly cope with hostile environmental conditions @xcite . when hostile conditions are created in a petri dish by using a very low level of nutrients , a hard surface ( high concentration of agar ) , or both , very complex patterns are observed .
drawing on the analogy with diffusive patterning in non - living systems @xcite , we can state that complex patterns are expected .
the cellular reproduction rate that determines the growth rate of the colony is limited by the level of nutrients available for the cells .
the latter is limited by the diffusion of nutrients towards the colony ( for low nutrient substrate ) .
hence colony growth under certain conditions should be similar to diffusion limited growth in non - living systems as mentioned above @xcite .
the study of diffusive patterning in non - living systems teaches us that the diffusion field drives the system towards decorated ( on many length scales ) irregular fractal shapes @xcite .
indeed , bacterial colonies can develop patterns reminiscent of those observed during growth in non - living systems @xcite .
but , this is certainly not the end of the story .
in fact , the colonies exhibit far richer behavior .
this is ultimately a reflection of the additional levels of complexity involved @xcite .
the building blocks of the colonies are themselves living systems , each having its own autonomous self - interest and internal degrees of freedom . yet ,
efficient adaptation of the colony to adverse growth conditions requires self - organization on all levels which can only be achieved via cooperative behavior of these individual cells .
thus , pattern formation at the colony level may be viewed as the outcome of a dynamical interplay @xcite between the micro - level ( the individual cell ) and the macro - level ( the colony ) . for this interplay to work
, the effects of changes at the micro - level must make themselves felt at the macro - level .
this is why the notion of a singular perturbation , discovered to be a key for understanding pattern selection in non - living systems , will be of even more importance here .
this manuscript is mainly a review of our modeling studies . yet
we include some important new results : 1 .
explanations and modeling of chemotaxis during chiral growth .
2 . modeling and characterization of weak chirality ( global weak twist ) during branching growth .
for completeness , section [ sec : bio - bg ] includes a brief description of the necessary biological background needed to justify our models of the growth .
how should one approach the modeling of the complex bacterial patterning ?
with present computational power it is natural to use computer models as a main tool in the study of complex systems .
however , one must be careful not to be trapped in the `` reminiscence syndrome '' , described by j. d. cowan @xcite , as the tendency to devise a set of rules which will mimic some aspect of the observed phenomena and then , to quote j. d. cowan `` they say : ` look , is nt this reminiscent of a biological or physical phenomenon ! '
they jump in right away as if it s a decent model for the phenomenon , and usually of course it s just got some accidental features that make it look like something . ''
yet the reminiscence modeling approach has some indirect value .
true , doing so does not reveal ( directly ) the biological functions and behavior .
however , it does reflect understanding of geometrical and temporal features of the patterns , which indirectly might help in revealing the underlying biological principles .
another extreme is the `` realistic modeling '' approach , where one constructs a model that includes in details all the known biological facts about the system .
such an approach sets a trajectory of ever including more and more details ( vs. generalized features ) .
the model keeps evolving to include so many details that it loses any predictive power .
here we try to promote another approach the `` generic modeling '' one @xcite .
we seek to elicit , from the experimental observations and the biological knowledge , the generic features and basic principles needed to explain the biological behavior and to include these features in the model .
we will demonstrate that such modeling , with close comparison to experimental observations , can be used as a research tool to reveal new understanding of the biological systems .
generic modeling is not about using sophisticated , as it may , mathematical description to dress pre - existing understanding of complex biological behavior .
rather , it means a cooperative approach , using existing biological knowledge together with mathematical tools and synergetic point of view for complex systems to reach a new understanding ( which is reflected in the constructed model ) of the observed complex phenomena .
the generic models can yet be grouped into two main categories : 1 .
discrete models such as the communicating walkers models of ben - jacob @xcite and the bions model of kessler and levine @xcite . in this approach
, the microorganisms ( bacteria in the first model and amoebae in second ) are represented by discrete , random walking entities ( walkers and bions , respectively ) which can consume nutrients , reproduce , perform random or biased movement , and produce or respond to chemicals .
the time evolution of the chemicals is described by reaction - diffusion equations .
2 . continuous or reaction - diffusion models @xcite .
in these models the microorganisms are represented via their 2d density , and a reaction - diffusion equation of this density describes their time evolution .
this equation is coupled to the other reaction - diffusion equations for the chemical fields . in the context of branching growth ,
this idea has been pursued recently by mimura and matsushita @xcite , kawasaki @xcite , kitsunezaki @xcite and kozlovsky @xcite .
a summary and critique of this approach can be found in @xcite and @xcite .
in section [ sec : cont - models ] we present the continuous modeling of the branching growth . in section [ sec : c ] the chiral growth is modeled using the `` atomistic '' communicating spinors model , which enables us to model chemotaxis response .
it is the first time that the chemotaxis effect on chiral growth has been studied .
the actual study of chemotaxis , both in the chiral growth and the branching growth is done in section [ sec : chemo ] .
section [ sec : weak_chiral ] is devoted to the studies of weak chirality .
the phenomenon is modeled using both continuous and discrete models .
we introduce a measure for weak chirality which enables a more crucial comparison between the models results and the observed patterns .
good agreement was found .
conclusions are presented in section [ sec : conc ] .
we explain that the weak chirality phenomenon is general , and show examples of weak chirality during growth of the chiral morphotype and the vortex morphotype . in the latter it results from a different mechanism , and
indeed the twist is not linear with the radius of growth .
following the experimental observations which are explained in this manuscript , we will describe the most relevant information for the understanding and modeling of the observed colonial patterning .
we base relevancy on our previous experience and we concentrate on bacterial movement . some additional examples of the patterns exhibited by colonies of the are shown in figures [ fig:2.1 ] , [ fig:2.2 ] and [ fig:2.3 ] . for intermediate agar concentrations ( about 1.5%
@xmath0 in @xmath1 ) , at very high peptone levels ( above 10@xmath2 ) the patterns are compact ( fig .
[ fig:2.2]a ) .
at somewhat lower but still high peptone levels ( about 5 - 10@xmath2 ) the patterns exhibit quite pronounced radial symmetry and may be characterized as dense fingers ( fig .
[ fig:2.2]b ) , each finger being much wider than the distance between fingers . for intermediate peptone levels , branching patterns with lower fractal dimension ( reminiscent of electro - chemical deposition ) are observed ( fig .
[ fig:2.2]c ) .
the patterns are `` bushy '' , with branch width smaller than the distance between branches .
as the peptone level is lowered , the patterns become more ramified and fractal like .
surprisingly , at even lower peptone levels ( below 0.25@xmath2 for 2% agar concentration ) the colonies revert to organized structures : fine branches forming a well defined global envelope .
we characterize these patterns as fine radial branches ( fig .
[ fig:2.2]d ) .
for extremely low peptone levels ( below 0.1@xmath2 ) , the colonies lose the fine radial structure and again exhibit fractal patterns ( fig .
[ fig:2.3 ] ) . for high agar concentration
the branches are very thin ( fig . [
fig:2.3]b ) . at high agar concentration and very high peptone levels the colonies display a structure of concentric rings ( fig .
[ fig:2.4 ] ) . at high agar concentrations
the branches also exhibit a global twist with the same handedness ( weak chirality ) , as shown in fig .
[ fig:2.5 ] .
similar observations during growth of other bacterial strains have been reported by matsuyama @xcite .
we referred to such growth patterns as having weak chirality , as opposed to the strong chirality exhibited by the
. a closer look at an individual branch ( fig .
[ fig:2.6 ] ) reveals a phenomenon of density variations within the branches .
these 3-dimensional structures arise from accumulation of cells in layers .
the aggregates can form spots and ridges which are either scattered randomly , ordered in rows , or organized in a leaf - veins - like structure .
the aggregates are not frozen ; the cells in them are motile and the aggregates are dynamically maintained . under the microscope , cells are seen to perform a random - walk - like movement in a fluid .
this fluid is , we assume , excreted by the cells and/or drawn by the cells from the agar @xcite .
the cellular movement is confined to this fluid ; isolated cells spotted on the agar surface do not move .
the boundary of the fluid thus defines a local boundary for the branch ( fig .
[ fig:2.9 ] ) .
whenever the cells are active , the boundary propagates slowly as a result of the cellular movement pushing the envelope forward and production of additional wetting fluid .
electron microscope observations reveal that these bacteria have flagella for swimming .
the observations reveal also that the cells are active at the outer parts of the colony , while closer to the center the cells are stationary and some of them sporulate ( form spores ) ( fig .
[ fig:2.10 ] ) .
it is known that certain bacteria respond to adverse growth conditions by entering a spore stage until more favorable growth conditions return .
such spores are metabolically inert and exhibit a marked resistance to the lethal effects of heat , drying , freezing , deleterious chemicals , and radiation . at very low agar concentrations ( below 0.5% ) the bacteria swim inside the agar and not on its surface .
between 0.5% and 1% agar concentration some of the bacteria move on the surface and some inside the agar .
chiral asymmetry ( first discovered by louis pasteur ) exists in a whole range of scales , from subatomic particles through human beings to galaxies , and seems to have played an important role in the evolution of living systems @xcite . bacteria display various chiral properties .
mendelson @xcite showed that long cells of can grow in helices , in which the cells form long strings that twist around each other .
they have shown also that the chiral characteristics affect the structure of the colony .
ben - jacob @xcite have found yet another chiral property the strong chirality exhibited by the . here
, the flagella handedness acts as a microscopic perturbation which is amplified by the diffusive instability , leading to the observed macroscopic chirality .
this appears to be analogous to the manner in which crystalline anisotropy leads to the observed symmetry of snowflakes @xcite ; more about this later .
exhibits a wealth of different patterns according to the growth conditions ( fig .
[ fig:2.14 ] ) . as for ,
the patterns are generally compact at high peptone levels and become ramified ( fractal ) at low peptone levels . at very high peptone levels and high agar concentration , conceals its chiral nature and exhibits branching growth similar to that of .
below 0.5% agar concentration the exhibits compact growth with density variations .
these patterns are almost indistinguishable from those developed by the . in the range of 0.4%-0.6% agar concentration
the exhibits its most complex patterns ( fig .
[ fig:2.16 ] ) . surprisingly , these patterns are composed of chiral branches of both left and right handedness .
microscopic observations reveal that part of the growth is on top of the agar surface while in other parts the growth is in the agar .
our model of the chiral growth explains that indeed growth on top of the surface and in the agar should lead to opposite handedness .
optical microscope observations indicate that during growth of strong chirality the cells move within a well defined envelope .
the cells are long relative to those of , and the movement appears correlated in orientation ( fig .
[ fig:2.17 ] ) .
each branch tip maintains its shape , and at the same time the tips keep twisting with specific handedness while propagating .
electron microscope observations do not reveal any chiral structure on the cellular membrane @xcite .
in the course of evolution , bacteria have developed ingenious ways of moving on surfaces .
the most widely studied and perhaps the most sophisticated translocation mechanism used by bacteria is the flagellum @xcite , but other mechanisms exist as well @xcite .
swimming is a solitary movement done in liquid .
a swimming bacterium runs nearly straight runs , interrupted by short periods of tumbling .
tumbling event is a random rotation in one location .
the direction of the next run is dictated by the final orientation of the tumbling bacterium . a swimming bacterium propagates itself by rotating a bundle of flagella .
each flagellum is an helical protein filament which is hooked to a molecular engine transversing the bacterial membrane .
the engines of all flagella rotate synchronously clockwise or counterclockwise .
when the bacterium turn them counterclockwise , the flagella form an aligned bundle and push the bacterium forward .
when they turn clockwise , the flagella disjoin and the bacterium tumbles . in our experiments , the bacteria do not swim in a pre - supplied liquid , but in a layer of fluid on the surface of the agar .
the bacterial cells move individually and at random in the same manner as flagellated bacteria move in wet mounts ( i.e. , nearly straight runs separated by brief tumbling ) .
swimming takes place only in sufficiently thick surface fluid .
microscope observations reveal no organized flow - field pattern . based on microscope observations of movement and electron microscope observations of flagella
we identify the movement of and as swimming .
cells tumble about every @xmath3 depending on external conditions .
the speed of the bacterium between tumbling events is very sensitive to conditions such as the liquid viscosity , temperature and ph level .
typically , it is of the order of 1 - 10@xmath4 .
swimming can be approximated by a random walk with variable step size @xcite . at low bacterial densities
the random walk can be described by a diffusion equation with a diffusion coefficient @xmath5 .
low bacterial densities means that the mean free path between bacterial collisions @xmath6 is longer than the tumbling length @xmath7 , thus collisions between the bacteria can be neglected .
the mean free path ( or collision length ) is @xmath8 where @xmath9 is the 3d bacterial density and @xmath10 is the 2d density the projection of @xmath9 on the surface . at high densities ( @xmath11 ) , the collisions can not be neglected . in attempt to approximate the dynamics in those conditions
, one may want to consider the time of straight motion to be @xmath12 instead of @xmath13 .
hence @xmath14 depends on the bacterial density to yield @xmath15 this approximation is valid under the assumptions that a collision event is identical to a tumbling event ( abrupt uncorrelated change in direction of motion ) , that a tumbling event is independent of the collisions , and that the speed between such events is not affected by their frequency .
the assumption that a collision event is like a tumbling event poses many problems .
even if the bacteria do not activate special response to collision it is unrealistic to assume that collisions are elastic , or that the flagella adopt immediately to the new orientation which changes during collisions .
thus it is reasonable to assume strong correlation between the cell s orientation before collision and the cell s orientation after collision .
in addition , the orientation after the collision should be biased according with the average direction of motion of the surrounding bacteria , as they carry the liquid with them .
the important parameter is not the collision length @xmath6 but re - orientation time @xmath16 .
the re - orientation time is the time it takes a bacterium to loose memory of its initial orientation , i.e. the time span on which the final orientation has effectively no correlation with the initial orientation . at low densities
the re - orientation time @xmath16 is equal to the tumbling time @xmath17 . as the density rises and the collisions become more frequent , @xmath16 decrease .
@xmath18 defines the densities above which the constant diffusion coefficient @xmath19 is not a good approximation .
it is quite possible that these densities are high enough so as to make the velocity and even the type of motion dependent on bacterial density , making relation ( [ eq : hige_density_diffusion:0 ] ) irrelevant . in any case
, high cellular densities does mean an effective decrease in the diffusion coefficient related to the bacterial movement .
when swimming in an unstirred liquid , very low cellular densities also effect the movement .
the bacteria secrete various materials into the media and some of them , e.g. enzymes and other polymers , significantly change the physical properties of the liquid making it more suitable for bacterial swimming .
the secretion of these materials depend on cellular density , thus at not - too - high densities the speed of swimming rise with the cellular density .
hence the diffusion coefficient related to the bacterial movement should be a non - monotonic function of the bacterial density .
moreover , the specific functional form might depend on the specific bacterial strain .
in other conditions there is similar but more pronounced effect . on semi - solid surface
the bacteria can not swim at all inside the agar and they have to produce their own layer of liquid to swim in it . to produce such fluid the bacteria secrete lubricant ( wetting agents ) .
other bacterial species produce known extracellular lubricants ( such as surfactants , see @xcite and references therein , or the extracellular slime produced by @xcite ) .
these are various materials ( various cyclic lipopeptides were identified ) which draw water from the agar
. the composition and properties of the lubricant of is not known , but we will assume that higher concentration of lubricant is needed to extract water from a dryer agar , and that the lubricant is slowly absorbed into the agar ( or decomposes ) .
a single bacterium on the agar surface can not produce enough fluid to swim in it , thus the bacteria can not break out of the layer fluid and the branches of a or colony can be defined by this fluid .
whenever bacteria enter the shallower parts of the layer , at the edge of the branch , they become sluggish , indicating that the depth of the layer effects the bacterial movement .
it can be argued ( see section [ nonlinear ] ) that in such cases the bacterial speed is related to the bacterial density by a power law ( at least in low densities ) . not only the diffusion coefficient related to the bacterial movement is a non - monotonic function of the bacterial density ( as in a liquid agar ) , but it is also vanishes for extremely low densities . in this case
it is clear that the specific functional form depend on the specific bacterial strain ( - , for example , can not move at all under such conditions ) .
chemotaxis means changes in the movement of the bacteria in response to a gradient of certain chemical field @xcite .
the movement is biased along the gradient either in the gradient direction or in the opposite direction .
usually chemotactic response means a response to an externally produced field , like in the case of chemotaxis towards food .
however , the chemotactic response can be also to a field produced directly or indirectly by the bacterial cells
. we will refer to this case as chemotactic signaling .
the bacteria sense the local concentration @xmath20 of a chemical by membrane receptors binding the chemical s molecules @xcite .
it is crucial to note that when estimating gradients of chemicals , the bacterial cells actually measure changes in the receptors occupancy and not in the concentration itself .
when put in continuous equations @xcite , this indirect measurement translates to measuring the gradient @xmath21 where @xmath22 is a constant whose value depends on the receptors affinity , the speed in which the bacterium processes the signal from the receptor , etc .
this means that the chemical gradient times a factor @xmath23 is measured , and it is known as the `` receptor law '' @xcite . in a continuous model
, we incorporate the effect of chemotaxis by introducing a chemotactic flux @xmath24 : @xmath25 @xmath26 is the gradient sensed by the bacteria ( with @xmath27 having the units of 1 over chemical s concentration ) .
@xmath28 is usually taken to be either constant or the `` receptor law '' .
@xmath29 is the bacterial response to the sensed gradient ( having the same units as a diffusion coefficient times the units of the bacterial density @xmath30 ) .
it is positive for attractive chemotaxis and negative for repulsive chemotaxis .
ben - jacob argued @xcite that for the colonial adaptive self - organization the bacteria employ three kinds of chemotactic responses , each dominant in different regime of the morphology diagram ( the claim was made for , but the same hold for their relatives ) .
one response is the food chemotaxis mentioned above .
it is expected to be dominant for only a range of nutrient levels ( see the `` receptor law '' below ) .
the two other kinds of chemotactic responses are signaling chemotaxis .
one is long - range repulsive chemotaxis .
the repelling chemical is secreted by starved bacteria at the inner parts of the colony .
the second signal is a short - range attractant .
the length scale of each signal is determined by the diffusion constant of the chemical agent and the rate of its spontaneous decomposition .
_ amplification of diffusive instability due to nutrients chemotaxis : _ in non - living systems , more ramified patterns ( lower fractal dimension ) are observed for lower growth velocity .
based on growth velocity as function of nutrient level and based on growth dynamics , ben - jacob @xcite concluded that in the case of bacterial colonies there is a need for mechanism that can both increase the growth velocity and maintain , or even decrease , the fractal dimension .
they suggested food chemotaxis to be the required mechanism .
it provides an outward drift to the cellular movements ; thus , it should increase the rate of envelope propagation . at the same time , being a response to an external field it should also amplify the basic diffusion instability of the nutrient field .
hence , it can support faster growth velocity together with a ramified pattern of low fractal dimension . _ repulsive chemotactic signaling : _ [ sec : repulsive ] we focus now on the formation of the fine radial branching patterns at low nutrient levels . from the study of non - living systems
, it is known that in the same manner that an external diffusion field leads to the diffusion instability , an internal diffusion field will stabilize the growth .
it is natural to assume that some sort of chemotactic agent produces such a field . to regulate the organization of the branches
, it must be a long - range signal . to result in radial branches it must be a repulsive chemical produced by bacteria at the inner parts of the colony . the most probable candidates are the bacteria entering a pre - spore stage .
if nutrient is deficient for a long enough time , bacterial cells may enter a special stationary state a state of a spore which enables them to survive much longer without food . while the spores themselves do not emit any chemicals ( as they have no metabolism ) , the pre - spores ( sporulating cells ) do not move and emit a very wide range of waste materials , some of which unique to the sporulating cell . these emitted chemicals might be used by other bacteria as a signal carrying information about the conditions at the location of the pre - spores .
jacob @xcite suggested that such materials are repelling the bacteria ( repulsive chemotactic signaling ) as if they escape a dangerous location .
- , like most bacteria , reproduce by fission of the cell into two daughter cells which are practically identical to the mother cell .
the crucial step in the cell division is the replication of the genetic material and its sharing between the daughter cells .
haste replication of dna might lead to many errors most organisms limit the rate of replication to about 1000 bases per second .
thus the reproduction must take at least minimal reproduction time @xmath31 .
this reproduction time @xmath31 is about @xmath32 in bacilli . for reproduction , as well as for movement and other metabolic processes , bacteria and all other organisms need influx of energy .
any organism which does not get its energy directly from sunlight ( by photo - synthesis ) needs an external supply of food . in
the patterning experiments the bacteria eat nutrient from the agar .
as long as there is enough nutrient and no significant amount of toxic materials , food is consumed ( for cell replication and internal processes ) at maximal rate @xmath33 . to estimate @xmath33 we assume that a bacterium needs to consume an amount of food @xmath34 of about @xmath35 .
it is 3 times its weight one quanta for doubling body mass , one quanta used for movement and all other metabolic processes during the reproduction time @xmath36 , and one quanta is for the reduced entropy of making organized cell out of food .
hence @xmath37 is about 2@xmath38 ( 1@xmath39gram ) .
if nutrient is deficient for a long enough period of time , the bacterial cells may enter a special stationary state a state of a spore which enables them to survive much longer without food .
the bacterial cells employ very complex mechanisms tailored for the process of sporulation .
they stop normal activity like movement and use all their internal reserves to metamorphose from an active volatile cell to a sedentary durable seed. while the spores themselves do not emit any chemicals ( as they have no metabolism ) , the pre - spores ( sporulating cells , see fig .
[ fig:2.10 ] ) do not move and emit a very wide range of wast e materials , some of which unique to the sporulating cell .
these emitted chemicals might be used by other cells as a signal carrying information about the conditions at the location of the pre - spores .
ben - jacob _ et al .
_ @xcite suggested that such materials are repelling the bacteria ( repulsive chemotactic signaling ) as if they escape a dangerous location .
when bacteria are grown in a petri dish , nutrients are usually provided by adding peptone , a mixture including all the amino acids and sugars as source of carbon .
bacteria which are not defective in synthesis of any amino acid can grow also on a minimal agar in which a single source of carbon and no amino acids are provided .
such growth might seem to be easier to model as the growth is limited by the diffusion of a single chemical .
however , during growth on minimal agar there is usually a higher rate of waste products accumulation , introducing other complications into the model .
moreover many of our strains are auxotrophic i.e. defective in synthesis of some amino acids and need an external supply of it .
providing the bacteria with these amino acids and only a single carbon source might pose us the question as to what is the limiting factor in the growth of the bacteria .
for all those reasons we prefer to use peptone as nutrient source .
we said that if there is ample supply of food , bacteria reproduce in a maximal rate of one division in @xmath40 .
if the available amount of food is limited , bacteria consume the maximum amount of food they can . in the limit of low bacterial density ,
the available amount of food over the tumbling time @xmath13 is the food contained in the area @xmath41 , where @xmath14 and @xmath42 are the diffusion coefficients of the bacteria and the food , respectively .
hence the rate of food consumption is given by @xmath43 ( weather @xmath14 is constant or not ) . in a continuous model , reproduction of bacteria
translate to a growth term of the bacterial density which is @xmath10 times the eating rate per bacteria . in the limit of high nutrient
it is @xmath44 , and in the limit of low nutrient it is proportional to @xmath45 .
this brings to mind michaelis - menten law @xcite of @xmath46 with @xmath47 constants .
many authors take only the low nutrient limit of this expression , @xmath48 , although it is not biologically established that the bacteria in the experiments are limited by the availability of food and not by their maximal consumption rate .
the lubricating bacteria model is a reaction - diffusion model for the bacterial colonies of the @xcite .
this model includes four coupled fields .
one field describes the bacterial density @xmath49 , the second describes the height of lubrication layer in which the bacteria swim @xmath50 , a third field describes the nutrients @xmath51 and the fourth field is the stationary bacteria that `` freeze '' and begin to sporulate @xmath52 ( see section [ sec : repulsive ] ) .
we first describe the dynamics of the bacteria and of the nutrient .
the two reaction - diffusion equations governing those fields are : @xmath53 where @xmath54 is the bacterial reproduction term .
the nutrient diffusion is a simple diffusion process with a constant diffusion coefficient @xmath42 .
the bacteria consume nutrients at the rate @xmath55 which is taken to be : @xmath56 this approximate term is correct at the limit of low nutrient level and low bacterial density . the nutrient consumed by bacteria serves as an energy source and as a precursor for synthesis of macromolecules .
there is probably a minimum amount of energy necessary to maintain cell structure and integrity , called maintenance energy , and nutrients used to supply the maintenance energy are not available for cell growth .
we assume that those nutrients are required at a constant rate @xmath57 .
bacteria then can not utilize all the nutrients for reproduction : @xmath58 there is no explicit term for sporulation in ( [ eq-1 ] ) .
instead the reproduction term @xmath59 can be used also to model sporulation .
sporulation is initiated by starvation so it is complementary to reproduction .
when @xmath60 bacteria reproduce and so sporulation is excluded . in the other case when @xmath61 bacteria reduce in number .
this should represent sporulation .
note that @xmath61 when the nutrient level is low so indeed we can claim that bacteria are starved .
this simple sporulation scheme has its limitations .
its rate is set by the nutrient consumption and effects such as density dependence are neglected .
we represent the sporulating bacteria by the field @xmath62 , whose time evolution is : @xmath63 in other continuous models the bacterial reproduction and the sporulation process are modeled differently @xcite .
the bacterial reproduction is proportional to the nutrient consumption rate and sporulation is an independent process that proceeds at a rate @xmath57 which could depend on other variables such as the bacterial density or the nutrient concentration .
note that if those modifications are applied to this model , the functional form of the bacterial reaction terms ( [ eq-1 ] ) , ( [ eq - gamma ] ) will not change .
the equation for the field @xmath64 ( [ eq - s ] ) will change to : @xmath65 since the dynamics of the other variables are separated from @xmath64 , its different dynamics are not significant .
moreover , the modification of the equation for @xmath64 is minor .
the difference is only in the biological interpretation of the terms .
we now turn to the bacterial movement . in a uniform layer of liquid ,
bacterial swimming is a random walk with a variable step length and can be approximated by diffusion .
the layer of lubricant is not uniform , and its height affects the bacterial movement .
an increase in the amount of lubricant decreases the friction between the bacteria and the agar surface .
we suggest that the bacterial movement depends on the local lubricant height through a power law with the exponent @xmath66 : @xmath67 where @xmath14 is a constant with dimensions of a diffusion coefficient .
@xmath14 is related to the fluid s viscosity and the dryness of the agar might affect this viscosity . gathering the various terms
gives the partial model : @xmath68 it is possible to define dimensionless time and space variables @xmath69 and @xmath70 . in those units the parameters
@xmath42 and @xmath57 are equal to @xmath71 .
we model the dynamics of the lubricating fluid also by a reaction diffusion equation .
there are two reaction terms : production by the bacteria and absorption into the agar .
the dynamics of the field are : @xmath72 where @xmath73 is the fluid flux , @xmath74 is the fluid production term and @xmath75 is the absorption rate of the fluid into the agar .
we assume that the fluid production depends on the bacterial density .
as the production of lubricant probably demands substantial metabolic efforts , it should also depend on the nutrient s level .
we take a simple form where the production depends linearly on the concentrations of both the bacteria and the nutrient .
the exact relation should depend on the synthetic pathway of the active agents composing the lubricant .
if they are , for example , secondary metabolites , then their production does not depend on the current nutrient level , but on the prior accumulation of primary metabolites .
however , the model is not sensitive to the exact dependence of the lubricant production on the nutrient s level since the lubricant production is important at the front of the expanding colony , where the nutrient level is close to the initial level .
it is reasonable that the bacteria produce lubricant up to a height , denoted as @xmath76 , which is sufficient for their swimming motion .
we therefore take the production term to be : @xmath77 where @xmath78 is the production rate .
we turn to the flow of the lubricating fluid .
the physical problem is very complicated .
however a simplified model will suffice .
we model the lubricant flux as a non - linear diffusion process : @xmath79 where @xmath80 is a constant with dimensions of a diffusion coefficient .
the diffusion term of the fluid depends on the height of the fluid to the power @xmath81 .
the nonlinearity causes the fluid to have a sharp boundary at the front of the colony , as is observed in bacterial colonies .
the equation for the lubricant field is : @xmath82 the functional form of the terms that we introduced are simple and plausible , but they are not derived from basic physical principles . therefore we do not have quantitative relations between the parameters of those terms and the physical properties of the agar substrate .
however we can suggest some relations . in the experiments ,
the agar concentration is controlled .
higher agar concentration gives a drier and more solid substrate .
we shall try to find what are the effects on the lubricant layer .
we recall that the lubricant fluid is composed from water and active components such as surfactants .
a drier agar can increase the absorption rate @xmath75 .
alternatively it can diminish the amount of water extracted by the active components . then either the lubricant layer will be thinner or the bacteria will have to produce more of the active components .
the former case should decrease @xmath14 while the latter should decrease the production rate @xmath78 . in both cases
the composition of the lubricant fluid will change as the concentration of the active components will increase .
the lubricant fluid should become more viscous , with the effect of @xmath14 and @xmath80 decreasing .
equation ( [ eql ] ) together with equations ( [ eq3 ] ) form the lubricating bacteria model . for the initial conditions , we set @xmath83 to have a uniform distribution of level @xmath84 , @xmath85 to be zero everywhere but in the center , and the other fields to be zero everywhere .
our results show that the model can reproduce branching patterns , similar to the bacterial colonies . in the experiments
there are two control parameters : the agar concentration and the initial nutrient concentration .
first we examine the effect of changing the initial nutrient concentration @xmath84 . as fig .
[ fig - nutrient ] shows , the model produced a dense circular colony when @xmath84 was large .
the pattern became more branched and ramified as @xmath84 decreased until @xmath84 was close to @xmath71 , which is the minimal value of @xmath84 to support growth .
changing the agar concentration affects the dynamics of the lubricant fluid .
previously we demonstrated that a higher agar concentration relates to a larger absorption rate @xmath75 and to lower production rate @xmath78 and lower diffusion coefficients @xmath80 and @xmath14 . in fig .
[ fig - lambda ] we show patterns obtained with different values of the parameters @xmath78 and @xmath75 . as we expected , increasing @xmath75 or decreasing @xmath78 produced a more ramified pattern , similar to the effect of a higher agar concentration on the patterns of bacterial colonies .
similar effects are obtained by decreasing @xmath14 . under certain assumptions ,
the lubricating bacteria model can be reduced to the non - linear diffusion model of kitsunezaki @xcite and cohen @xcite .
all the additional assumptions needed are about the dynamics near zero bacterial density : + 1 ) the lubricant height @xmath86 is much smaller than @xmath87 , so that the production of the lubricant can be assumed to be independent of its height + 2 ) the production of lubricant is proportional to the bacterial density to the power @xmath88 ( in the simplest case taken above @xmath89 ) + 3 ) the absorption of the lubricant is proportional to the lubricant height to the power @xmath90 ( in the simplest case taken above @xmath91 ) .
+ 4 ) over the bacterial length scale , the two above processes are much faster than the diffusion process , so the lubricant height is proportional to the bacterial density to the power of @xmath92 .
+ 5 ) the friction is proportional to the lubricant height to the power @xmath93 . + given these assumptions , the lubricant field can be removed from the dynamics and be replaced by a density dependent diffusion coefficient .
this diffusion coefficient is proportional to the bacterial density to the power @xmath94 .
the resulting model is : @xmath95 for @xmath96 the 1d model gives rise to a front `` wall '' , with compact support ( i.e. @xmath97 outside a finite domain ) . for @xmath98
this wall has an infinite slope .
the model exhibits branching patterns for suitable parameter values and initial conditions , as depicted in fig .
[ fig : k1 ] . increasing initial levels of nutrient leads to denser colonies , similar to the observed patterns .
the communicating spinors model was developed to explain the chirality of the colonies .
our purpose is to show that the flagella handedness , while acting as a singular perturbation , leads to the observed chirality .
it does so in the same manner in which crystalline anisotropy leads to the observed symmetry of snowflakes @xcite .
it is known @xcite that flagella have specific handedness .
ben - jacob @xcite proposed that the latter is the origin of the observed chirality . in a fluid ( which is the state in most experimental setups ) , as the flagella unfold , the cell tumbles and ends up at a new random angle relative to the original one .
the situation changes for quasi 2d motion motion in a `` lubrication '' layer thinner then the cellular length .
we assume that in this case , of rotation in a plane , the tumbling has a well defined handedness of rotation .
such handedness requires , along with the chirality of the flagella , the cells ability to distinguish up from down .
the growth in an upside- down petri- dish shows the same chirality .
therefore , we think that the determination of up down is done either via the vertical gradient of the nutrient concentration or via the vertical gradient of signaling materials inside the substrate or via the friction of the cells with the surface of the agar .
the latter is the most probable alternative ; soft enough agar enables the bacteria to swim _ below _ the surface of the agar which leads to many changes in the patterns , including reversing the bias of the branches . to cause
the chirality observed on semi - solid agar , the rotation of tumbling must be , on average , less than @xmath99 and relative to a specific direction .
co - alignment ( orientational interaction ) limits the rotation .
we further assume that the rotation is relative to the local mean orientation of the surrounding cells . to test the above idea
, we included the additional assumed features in the communicating walkers model @xcite , changing it to a ` communicating spinors ' model ( as the particles in the new model have an orientation and move in quasi-1d random walk ) .
the communicating walkers model @xcite was inspired by the diffusion - transition scheme used to study solidification from supersaturated solutions @xcite .
the former is a hybridization of the `` continuous '' and `` atomistic '' approaches used in the study of non - living systems .
ben - jacob have presented in the past a version of the communicating spinors model for the chiral growth @xcite .
the model we present here is closely related to a previous model of the chiral growth , but it differs in two features .
the first is the orientation field ( see below ) , which was discontinuous piecewise constant and in this model it is continuous piecewise linear .
the second difference is the definition of a single run of a spinor ( the stretch between two tumbling events ) , which was defined as one run per one time unit ( i.e. each step is a run ) and now is defined as variable number of steps in the same direction . the representation of bacteria as spinors allows for a close relation to the bacterial properties .
the bacterial cells are represented by spinors allowing a more detailed description . at the end of the growth of a typical experiment
there are @xmath100 bacterial cells in the petri- dish .
thus it is impractical to incorporate into the model each and every cell .
instead , each of the spinors represents about @xmath101 cells , so that we work with @xmath102 spinors in one numerical `` experiment '' .
each spinor has a position @xmath103 , direction @xmath104 ( an angle ) and a metabolic state ( internal energy ) @xmath105 .
the spinors perform an off - lattice constrained random walk on a plane within an envelope representing the boundary of the wetting fluid .
this envelope is defined on the same tridiagonal lattice where the diffusion equations are solved . to incorporate the swimming of the bacteria into the model , at each time step each of the active spinors ( motile and metabolizing , as described below ) recalculate its direction @xmath106 and moves a step of size @xmath107 to this direction .
the direction in which each spinor moves is determined in two steps ; first the spinor decides whether it should continue the current run , that is to continue in the same direction @xmath108 . in the basic version of the model ( see sec .
[ sec : cchemo ] for extension of the model ) the decision is random with a specific probability @xmath109 to continue the run .
the resulting runs have geometric distribution of lengths , with mean run length of @xmath110 .
once a spinor decides to change direction , the new direction @xmath106 is derived from the spinor s previous direction by @xmath111 @xmath112 and @xmath113 represent the new features of rotation due to tumbling .
@xmath112 is a fixed part of the rotation and @xmath113 is a stochastic part , chosen uniformly from an interval @xmath114 $ ] ( @xmath115 constant ) .
@xmath116 is an orientation term that takes , with equal probabilities , one of the values 0 ( forward direction ) or @xmath117 ( backward direction ) .
this orientation term gives the spinors their name , as it make their re - orientation invariant to forward or backward direction .
@xmath118 is the local mean orientation in the neighborhood of @xmath103 .
@xmath119 is a projection function that represents the orientational interaction which acts on each spinor to orient @xmath104 along the direction @xmath118 .
@xmath119 is defined by @xmath120 once oriented , the spinor advances a step @xmath107 in the direction @xmath106 , and the new location @xmath121 is given by : @xmath122 the movement is confined within an envelope which is defined on the tridiagonal lattice .
the step is not performed if @xmath121 is outside the envelope . whenever this is the case , a counter on the appropriate segment of the envelope is increased by one .
when a segment counter reaches @xmath123 , the envelope advances one lattice step and a new lattice cell is occupied .
note that the spinor s direction is not reset upon hitting the envelope , thus it might `` bang its head '' against the envelope time and time again .
the requirement of @xmath124 hits represent the colony propagation through wetting of unoccupied areas by the bacteria .
note that @xmath124 is related to the agar dryness , as more wetting fluid must be produced ( more `` collisions '' are needed ) to push the envelope on a harder substrate .
next we specify the mean orientation field @xmath125 . to do so
, we assume that each lattice cell ( hexagonal unit area ) is assigned one value of @xmath126 , representing the average orientation of the spinors in the local neighborhood of the center of the cell .
the value of @xmath125 is set when a new lattice cell is first occupied by the advancement of the envelope , and then remains constant .
we set the value of @xmath127 to be equal to the average over the orientations of the @xmath128 attempted steps that led to the occupation of the new lattice cell .
the value of @xmath125 in any given point inside the colony is found by linear interpolation between the three neighboring centers of cells . clearly , the model described above is a simplified picture of the bacterial movement .
for example , a more realistic model will include an equation describing the time evolution of @xmath129 .
however , the simplified model is sufficient to demonstrate the formation of chiral patterns .
a more elaborate model will be presented elsewhere @xcite .
motivated by the presence of a maximal growth rate of the bacteria even for optimal conditions , each spinor in the model consumes food at a constant rate @xmath130 if sufficient food is available .
we represent the metabolic state of the @xmath131-th spinor by an internal energy @xmath132 .
the rate of change of the internal energy is given by @xmath133 where @xmath134 is a conversion factor from food to internal energy ( @xmath135 ) and @xmath136 represent the total energy loss for all processes over the reproduction time @xmath40 , excluding energy loss for cell division .
@xmath137 is @xmath138 where @xmath139 is the maximal rate of food consumption as limited by the locally available food ( sec .
[ sec : bio - bg ] ) . when sufficient food is available , @xmath140 increases until it reaches a threshold energy . upon reaching this threshold ,
the spinor divides into two .
when a spinor is starved for long interval of time , @xmath140 drops to zero and the spinor `` freezes '' .
this `` freezing '' represents entering a pre - spore state ( starting the process of sporulation , see section [ sec : cchemo ] ) .
we represent the diffusion of nutrients by solving the diffusion equation for a single agent whose concentration is denoted by @xmath51 : @xmath141 where the last term includes the consumption of food by the spinors ( @xmath142 is their density ) .
the equation is solved on the same tridiagonal lattice on which the envelope is defined .
the length constant of the lattice @xmath143 must be larger than the size of the spinors step @xmath107 .
the simulations are started with inoculum of spinors at the center and a uniform distribution of the nutrient .
both @xmath129 and the spinors at the inoculum are given uniformly distributed random directions .
results of the numerical simulations of the model are shown in fig .
[ fig : c - results ] .
these results do capture some important features of the observed patterns : the microscopic twist @xmath112 leads to a chiral morphology on the macroscopic level .
the growth is via stable tips , all of which twist with the same handedness and emit side - branches .
the dynamics of the side - branches emission in the time evolution of the model is similar to the observed dynamics . for large noise strength
@xmath144 the chiral nature of the pattern gives way to a branching pattern ( fig .
[ fig : c - t ] ) .
this provides a plausible explanation for the branching patterns produced by grown on high peptone levels , as the cells are shorter when grown on a rich substrate .
the orientation interaction is weaker for shorter cells , hence the noise is stronger .
so far , we saw that the models can reproduce many aspects of the microscopic dynamics and the patterns in some range of nutrient level and agar concentration , but at least for the -like growth , other models can do the same ( * ? ? ? * and reference there in ) . we will now extend the non - linear diffusion model and the communicating spinors model to test for their success in describing other aspects of the bacterial colonies involving chemotaxis and chemotactic signaling ( which are believed to by used by the bacteria @xcite ) . as we mentioned in section
[ sec : bg : chemotaxis ] , in a continuous model we incorporate the effect of chemotaxis by introducing a chemotactic flux @xmath24 : @xmath145 @xmath26 is the gradient sensed by the bacteria ( with @xmath27 having the units of 1 over chemical s concentration ) .
@xmath28 is usually taken to be either constant or the `` receptor law '' .
@xmath146 is the bacterial response to the sensed gradient ( having the same units as a diffusion coefficient times the units of the bacterial density @xmath85 ) . in the non - linear diffusion model
the bacterial diffusion is @xmath147 , and the bacterial response to chemotaxis is @xmath148 .
@xmath149 is a constant , positive for attractive chemotaxis and negative for repulsive chemotaxis .
we claimed that the fine radial branching patterns at low nutrient levels result from repulsive chemotactic signaling .
the equation describing the dynamics of the chemorepellent contains terms for diffusion , production by pre - spores , decomposition by active bacteria and spontaneous decomposition : @xmath150 where @xmath151 is a diffusion coefficient of the chemorepellent , @xmath152 is an emission rate of repellent by pre - spores , @xmath153 is a decomposition rate of the repellent by active bacteria , and @xmath154 is a rate of self decomposition of the repellent .
[ fig : k - food - rep ] demonstrates the effect of repulsive chemotactic signaling . in the presence of repulsive chemotaxis
the pattern becomes much denser with a smooth circular envelope , while the branches are thinner and radially oriented .
the colonial patterns of morphotype ( e.g. fig .
[ fig : c - results ] ) are rarely as ordered as the simulated patterns of the communicating spinors model .
for example , the branches of the observed colonies usually have varying curvature . in the simulations of morphotype shown in fig . [
fig : c - results ] all the branches have a uniform curvature .
one of the reasons for this difference is the simplifications taken during the model s development .
a more elaborate model that we will present @xcite will be a better description of the colony .
however , some of the observed features can be explained in the context of the communicating spinors model . in some of the observed patterns ( fig .
[ fig:2.16]b ) , the curvature of the branches has a distinct relation to the branch s radial orientation ( the orientation relative to the radial direction ) : the curvature is smaller when the branch is in the radial orientation and larger when the branch is orthogonal to that orientation .
this brings to mind the radial organization of branches in the morphotype , and indeed we where able to explain the chiral pattern with the aid of the same concept repulsive chemotaxis .
chemotaxis was introduced in previous versions of the communicating walkers model by varying , according to the chemical s gradient , the probability of moving in different directions @xcite . modulating
the directional probability is not the way bacteria implement chemotaxis
they modulate the length of runs . however , the growth of morphotype is insensitive to the details of the movement . modulating
the directional probability is as good an implementation of chemotaxis as many other implementations ( it was chosen for computational convenience ) .
the pattern of the is based on amplification of microscopic effects ( singular perturbation ) such as the left bias in the bacterial tumbling .
small differences in the microscopic dynamics of chemotaxis might affect the global pattern .
indeed we found that modulating the directional probability yield unrealistic results in the simulations of .
we had to resort to the bacterial implementation of chemotaxis modulating the length of runs according to the chemical s gradients .
when modulating the length of runs of walkers or spinors one must be careful not to change the particles speed .
such change is not observed in experiments @xcite and it has far reaching effects on the dynamics . changing the particles speed is like changing the diffusion coefficient of the bacterial density field , a change that can have undesirable effects on the pattern .
modulating the length of spinors runs without changing their speed can be done by modulating the number of steps that compose a single run ( that was our motivation for dividing the runs into steps ) . since the mean number of steps in a run
is determined by the reorientation probability @xmath109 , chemotaxis should modulate this probability .
for chemotaxis , the probability of changing direction by the @xmath131-spinor in one time step is ( for a repellent @xmath20 ) : @xmath155 where @xmath20 is measured at the spinor position @xmath156 , @xmath157 is the same as in the continuous model ( either constant or the `` receptor law '' ) and @xmath158 is directional derivative in the spinor s direction @xmath159 .
@xmath160 is truncated to within the range @xmath161 $ ] as it is a probability .
the length of the resulting runs will depend on the runs direction , where a spinor moving up the gradient of the repellent will have shorter _
mean _ run length than the same spinor moving down the gradient .
the production and dynamics of the repulsive chemotactic signaling in the communicating spinors model is the same as in the non - linear diffusion model , see eq .
( [ r - eqn ] ) ( with @xmath64 representing the density of spinors that `` freezed '' ) .
the patterns resulting from including repulsive chemotaxis in the model have indeed branches with variable curvature , as can be seen in fig .
[ fig : cmodel : chemo ] .
the curvature is smaller for branches in the radial direction .
food chemotaxis also varies the branches curvature , but in a less ordered manner , not similar to the observed bacterial patterns . under different growth conditions
the can produce very different patterns .
as mention above , if the agar is soft enough the bacteria can move inside it . in such case ,
the bias in the bacterial movement might change or even reverse , and it is manifested in the curvature of the branches .
widely changing curvature of the branches can be seen in fig .
[ fig:2.16]a .
the agar hardness was tuned such that in the beginning of the growth the bacteria could swim inside the agar , but they are forced to swim _ on _ the agar by the end of the growth due to the marginal water evaporation during the growth . in fig .
[ fig : cmodel : long ] we demonstrate the models ability to explain such patterns by changing the spinors bias @xmath112 during the simulation .
@xmath112 is set to be a continuous random function of the colonial size , which is constrained only at the beginning and end of growth to have certain values .
the function for @xmath112 is the same in all the images of figs .
[ fig : cmodel : long ] , only that in various types of chemotaxis are used . as can be seen , repulsive chemotactic signaling is needed to explain the observed bacterial patterns .
colonies of grown on hard substrate ( above 2.0% agar concentration ) exhibit branching patterns with a global twist with the same handedness , as shown in figs .
[ fig:2.5 ] and [ fig : weak ] .
similar observations during growth of other bacterial strains have been reported by matsuyama @xcite .
we refer to such growth patterns as having weak chirality , as opposed to the strong chirality exhibited by the .
in @xcite , ben - jacob proposed that , in the case of , it is the high viscosity of the `` lubrication '' fluid during growth on a hard surface that replaces the cell - cell co - alignment of the that limit the rotation of tumbling .
they further assumed that the rotation should be relative to a specified direction .
they used gradient of a chemotaxis signaling field ( specifically , the long - range repellent chemotaxis ) as a specific direction , rather than the local mean orientation field which is used in the case of .
it was shown in @xcite that inclusion of the above features in the communicating walkers model indeed leads to a weak chirality which is highly reminiscent of the observed one .
the idea above also provides a plausible explanation to the observations of weak chirality by matsuyama @xcite in strains defective in production of `` lubrication '' fluid . in the reaction - diffusion model , weak chirality
can obtained by modifying the chemotactic mechanism and causing it to twist : we alter the expression for the chemotactic flux @xmath162 ( eq . [ j_chem ] ) so that it is not oriented with the chemical gradient ( @xmath163 ) anymore
. instead it is oriented with a rotated vector @xmath164 , where @xmath165 is the two - dimensional rotation operator and @xmath166 is the rotation angle .
the chemotactic flux is thus written : @xmath167 the effect of rotating the repulsive chemotaxis , as depicted in fig .
[ fig : k - weak ] , is to make the pattern chiral , with the degree of chirality determined by the rotation angle @xmath166 .
one must note that adding a similar rotation to the food chemotaxis does not have the same effect , because the nutrient gradients do not posses the long - range , radial nature of the chemorepellent gradients .
as was demonstrated in section [ sec : c ] , the communicating spinors model is robust enough to reproduce patterns of , as well as patterns of .
here we use it to model patterns of weak chirality .
two simulated -like colonies are shown in fig .
[ fig : cmodel : weak ] . fig .
[ fig : cmodel : weak]a shows colony with radial branches while fig .
[ fig : cmodel : weak]b shows colony with weak chirality and thinner branches . in the two simulations the spinors have exactly the same response to chemotaxis and the same bias @xmath168 .
the two runs differ in the freedom of rotation @xmath144 ; in [ fig : cmodel : weak]a the spinors can rotate freely ( @xmath169 ) while in [ fig : cmodel : weak]b the spinors rotation is somewhat limited ( @xmath170 , while for the colony of strong chirality @xmath171 ) .
it seems that both models the non - linear diffusion model and the communicating spinors model can capture the essential features of the observed weak chirality . yet a closer examination reveals that the description of the two models is incompatible . in the non - linear diffusion model
the bias from the direction of the gradient is through the chemotaxis process .
the spinors , like the bacteria , can not modulate their runs as a function of the difference between their direction and that of the gradient ; they do not know what is the direction of the gradient , only the directional derivative along their path .
as was demonstrated in figure [ fig : cmodel : weak ] , one of the key features for the weak chirality in the spinors model is the correlation in orientation of the spinors ( through @xmath129 ) .
in fact in this model the twist of the branches stems from the deviation of the runs direction from the orientation of neighboring spinors .
the twist of the branches is related only indirectly , through the neighbors orientation , to the chemorepellent s gradient .
a continuous model of such process should include information about the mean orientation of the bacterial cells .
it should include chemotaxis without rotation , anisotropic diffusion ( smaller diffusion coefficient in orientations orthogonal to the mean orientation of neighbors ) and a rotation on the diffusion operator .
such a model will be presented elsewhere @xcite .
the discrete spinors model allows for a detailed representation of the bacterial properties .
the macroscopic dynamics and resulting patterns , however , are similar in both models apparently the growth does not amplify the difference in the microscopic dynamics
. thus the ` unrealistic ' microscopic description of the non - linear diffusion model does not rule it out as an approximation to the growth dynamics of bacterial colonies with weak chirality .
all through this manuscript we referred to chirality as a two - valued property either the pattern is chiral or it is not .
there are various attempts in the literature to quantify chirality with a continuous measure .
see for example the method of avnir @xcite , who applied their method also to large disordered objects . while this method is general and
can quantify with a single number the measure of asymmetry of any given object , it is not quite satisfactory for our purpose .
we would like to know the time evolution of the chirality of a colony , and not just `` mean '' chirality given by a single number .
we sacrifice the generality of the measure to that end .
since the growth velocity of the colonies ( both experimental and simulated ) is constant , we measure the chirality as a function of radius instead of function of time .
thus we can work on chirality of an image , not of a process .
the image can be a scanned picture of the real colony or a result of a computer simulation .
we look for a mapping of the image to a new one , which in some sense does not distinguish left from right ( the ambiguity stems from the fact that a large random object will not have , in general , reflection symmetry , thus there is no trivial definition for chirality of such objects ) .
the mapping is defined by : @xmath172 where each point in the image is described by the polar coordinates @xmath173 , measured from the center of the colony .
thus , each point is rotated by an increment @xmath174 which depends on the radius @xmath175 ( i.e. the distance from the center ) . working on many experimental patterns , as well as simulated patterns ,
we have learned that in most cases a linear dependence of @xmath176 on @xmath175 is sufficient to give quite satisfactory results , that is , to transform a chiral pattern into a `` normal '' branching pattern .
the rotating angle is thus written : @xmath177 where @xmath178 is the radius of the colony , and @xmath179 is the rotation angle at that radius .
the fact that this linear angular mapping suffices to `` de - chiral '' the simulated patterns may not be of much importance ( in the case of the continuous model , at least , this is almost a direct result of the way in which we introduce the weak chirality ) .
the same transformation works for images of real colonies of , but _ does not _
work for chiral colonies of other bacteria ( see sec . [
sec : conc ] ) .
this strengthens our belief in the models .
we first briefly reviewed experimental observations of colonial patterns formed by bacteria of the species .
we described both the tip - splitting growth of the and the chiral , twisted - branches growth of .
both colonial patterns and optical microscope observations of the bacteria dynamics were presented . in this manuscript we presented observations of various forms of chiral patterns in bacterial colonies .
our goal was to explain the various aspects of chirality .
we used two types of models : continuous reaction - diffusion models which deal with bacterial density , and a hybrid semi - discrete model which deals with properties of the individual bacterium . from a comparison of the models simulation and experimental observations
we conclude that chemotactic signaling plays an important part in the development of colonies of the two types .
we also estimate how sensitive the growth is to the details of the microscopic dynamics , demonstrating that the more complex the pattern is , the more sensitive the growth is to the small details .
we would like to note that the is not the only bacteria whose colonies exhibit chirality .
ben - jacob discussed in @xcite the formation of colonies of , where each branch is produced by a leading droplet and emits side branches , each with its own leading droplet .
each leading droplet consists of hundreds to millions of bacterial cells that circle a common center ( a vortex ) at a cellular speed of about @xmath180 m/s . in fig .
[ fig : v ] we show a colonial pattern of these bacteria .
the chirality we termed ` weak chirality ' is evident in this figure . in this case
the chirality is not related to the handedness of the flagella , but to the rotation of the vortices .
when `` pushed '' by repulsive chemotaxis , magnus force acts on the vortices and drive them side - ways from the radial direction of the chemorepellent s gradient .
this difference in mechanisms is expressed in the global pattern : the colonial patterns of can not be `` de - chiraled '' by the transformation ( linear angular mapping ) that `` de - chiral '' the .
the fact that the models for weak chirality match in this respect the the weak chirality of and not the ` weak chirality ' of is another support for their success in describing the bacterial colonies .
we hope we have convinced the reader that chirality in patterns of bacterial colonies gives important clues about the underlying dynamics .
the processes leading to such patterns are more complex than those leading to non - twisted branching patterns .
the chiral patterns are more sensitive to the underlying dynamics and as such they require more accurate models .
this reflects on the success of the models we presented as being a good description of the colonies .
identifications of the and genetic studies are carried in collaboration with the group of d. gutnick .
presented studies are supported in part by a grant from the israeli academy of sciences grant no . 593/95 , by the israeli - us binational science foundation bsf grant no .
00410 - 95 and by a grnat from imk almene fond .
two of us , e. ben - jacob and i. golding , thank the i m a for hospitality during part of this project .
one of us , i. cohen , thanks the colton scholarships for their support .
10 j. adler .
chemoreceptors in bacteria . , 166:15881597 , 1969 .
g. albrecht - buehler . in defense of `` nonmolecular '' cell biology .
, 120:191241 , 1990 .
v. a. avetisov , v. i. goldanskii , and v. v. kuzmin .
handedness , origin of life and evolution . , 44(7):3341 , 1991 .
m. y. azbel .
survival - extinction transition in bacteria growth .
22(4):311316 , 1993 . e. ben - jacob .
from snowflake formation to the growth of bacterial colonies .
part i : diffusive patterning in non - living systems . , 34:247273 , 1993 .
e. ben - jacob .
from snowflake formation to the growth of bacterial colonies .
part ii : cooperative formation of complex colonial patterns .
, 38:205241 , 1997 .
e. ben - jacob and i. cohen .
cooperative formation of bacterial patterns . in j.
a. shapiro and m. dworkin , editors , _ bacteria as multicellular organisms_. oxford university press , new - york , 1997 .
e. ben - jacob , i. cohen , and a. czirk .
smart bacterial colonies . in _ physics of biological systems : from molecules to species _ , lecture notes in physics , pages 307324 .
springer - verlag , berlin , 1997 .
e. ben - jacob , i. cohen , a. czirk , t. vicsek , and d. l. gutnick .
chemomodulation of cellular movement and collective formation of vortices by swarming bacteria and colonial development .
, 238:181197 , 1997 .
e. ben - jacob , i. cohen , and d. gutnick .
cooperative organization of bacterial colonies : from genotype to morphotype . , 1998 .
( in press )
. e. ben - jacob , i. cohen , o. shochet , i. aronson , h. levine , and l. tsimering .
complex bacterial patterns . , 373:566567 , 1995 .
e. ben - jacob , i. cohen , o. shochet , a. czirk , and t. vicsek .
cooperative formation of chiral patterns during growth of bacterial colonies .
, 75(15):28992902 , 1995 .
e. ben - jacob and p. garik .
the formation of patterns in non - equilibrium growth .
, 343:523530 , 1990 .
e. ben - jacob , h. shmueli , o. shochet , and a. tenenbaum .
adaptive self - organization during growth of bacterial colonies .
, 187:378424 , 1992 .
e. ben - jacob , o. shochet , and a. tenenbaum .
bakterien schlieen sich zu bizarren formationen zusammen . in a.
deutsch , editor , _ muster des ledendigen : faszination inher entstehung und simulation_. verlag vieweg , 1994 .
e. ben - jacob , o. shochet , a. tenenbaum , and o. avidan .
evolution of complexity during growth of bacterial colonies . in p.
e. cladis and p. palffy - muhoray , editors , _ spatio - temporal patterns in nonequilibrium complex systems _ , santa - fe institute studies in the sciences of complexity , pages 619634 .
addison - weseley publishing company , 1995 .
e. ben - jacob , o. shochet , a. tenenbaum , i. cohen , a. czirk , and t. vicsek .
communication , regulation and control during complex patterning of bacterial colonies .
, 2(1):1544 , 1994 .
e. ben - jacob , o. shochet , a. tenenbaum , i. cohen , a. czirk , and t. vicsek .
generic modeling of cooperative growth patterns in bacterial colonies . , 368:4649 , 1994 .
e. ben - jacob , a. tenenbaum , o. shochet , and o. avidan .
holotransformations of bacterial colonies and genome cybernetics . , 202:147 , 1994 .
h. c. berg . .
princeton university press , princeton , n.j . ,
h. c. berg and e. m. purcell .
physics of chemoreception .
, 20:193219 , 1977 .
i. cohen . mathematical modeling and analysis of pattern formation and colonial organization in bacterial colonies , 1997 .
thesis , tel - aviv university , israel .
i. cohen , a. czirk , and e. ben - jacob .
chemotactic - based adaptive self organization during colonial development .
, 233:678698 , 1996 .
i. cohen , i. golding , and e. ben - jacob .
models of chiral bacterial growth , 1999 .
( in preperation ) .
a. czirk , e. ben - jacob , i. cohen , and t. vicsek .
cooperative formation of vortices . submitted to _ phys .
e _ , 1996 .
j. d. desai and i. m. banat .
microbial production of surfactants and their commercial potential .
61:4764 , 1997 . m. eisenbach .
functions of the flagellar modes of rotation in bacterial motility and chemotaxis .
, 4(2):161167 , 1990 .
j. feder . .
plenum , new york , 1988 .
h. fujikawa and m. matsushita .
fractal growth of on agar plates .
, 58:38753878 , 1989 .
h. fujikawa and m. matsushita .
bacterial fractal growth in the concentration field of nutrient . , 60:8894 , 1991 .
i. golding , y. kozlovsky , i. cohen , and e. ben - jacob .
studies of bacterial branching growth using reaction - diffusion models of colonial development .
, 260(3 - 4):510554 , 1998 .
h. haken . .
springer - verlag , berlin , 1988 .
r. a. hegstrom and d. k. kondepudi .
the handedness of the universe . , 262:108115 , 1990 .
j. henrichsen .
bacterial surface translocation : a survey and a classification . , 36:478503 , 1972
. t. h. henrici . .
d. c. heath & company , 3rd edition , 1948 .
j. horgan .
from complexity to perplexity .
, june:7479 , 1995 .
o. katzenelso , h. z. hel - or , and d. avnir .
chirality of large random supramolecular structures .
, 2:174181 , 1996 .
k. kawasaki , a. mochizuki , m. matsushita , t. umeda , and n. shigesada .
modeling spatio - temporal patterns created by bacillus - subtilis .
, 188:177185 , 1997 .
d. a. kessler , j. koplik , and h. levine .
pattern selection in fingered growth phenomena .
, 37:255 , 1988 .
d. a. kessler and h. levine .
pattern formation in _ dictyostelium _ via the dynamics of cooperative biological entities .
, 48:48014804 , 1993 .
d. a. kessler , h. levine , and l. tsimring .
computational modeling of mound development in _
dictyostelium_. , 106(3 - 4):375388 , 1997 . j. o. kessler .
co - operative and concentrative phenomena of swimming micro - organisms .
, 26:147166 , 1985 . j. o. kessler and m. f. wojciechowski .
collective behavior and dynamics of swimming bacteria . in j.
a. shapiro and m. dworkin , editors , _ bacteria as mullticellular organisms _ , pages 417450 .
oxford university press inc .
, new york , 1997 .
s. kitsunezaki .
interface dynamics for bacterial colony formation .
, 66(5):15441550 , 1997 .
y. kozlovsky , i. cohen , i. golding , and e. ben - jacob .
lubricating bacteria model for branching colonial patterns , 1998 .
( in press ) .
j. m. lackiie , editor . .
cambridge univ .
press , 1986 .
dendrites , viscous fingering , and the theory of pattern formation . , 243:11501154 , 1989 .
s. a. mackay .
computer simulation of aggregation in dictyostelium discoideum .
, 33:116 , 1978 .
mandelbrot . .
freeman , san francisco , 1977 .
mandelbrot . .
freeman , san francisco , 1977 .
m. a. marahiel , m. m. nakano , and p. zuber .
regulation of peptide antibiotic production in _
bacillus_. , 7:631636 , 1993 .
m. matsushita and h. fujikawa .
diffusion - limited growth in bacterial colony formation .
, 168:498506 , 1990 .
m. matsushita , j. wakita , h. itoh , i. rafols , t. matsuyama , h. sakaguchi , and m. mimura .
interface growth and pattern formation in bacterial colonies .
, 249:517524 , 1998 .
m. matsushita , j .-
wakita , and t. matsuyama . growth and morphological changes of bacteria colonies . in p.
e. cladis and p. palffy - muhoray , editors , _ spatio - temporal patterns in nonequilibrium complex systems _ , santa - fe institute studies in the sciences of complexity , pages 609618 .
addison - weseley publishing company , 1995 .
t. matsuyama , r. m. harshey , and m. matsushita .
self - similar colony morphogenesis by bacteria as the experimental model of fractal growth by a cell population . , 1(3):302311 , 1993 .
t. matsuyama and m. matsushita .
fractal morphogenesis by a bacterial cell population .
, 19:117135 , 1993 .
t. matsuyama and y. nakagawa .
bacterial wetting agents working in colonization of bacteria on surface environment .
, 7:207214 , nov 1 1996 . n. h. mendelson .
helical macrofibers : morphogenesis of a bacterial multicellular macroorganism . , 75(5):24782482 , 1978 .
n. h. mendelson .
bacterial macrofibres : the morphogenesis of complex multicellular bacterial forms . , 74:425441 , 1990 .
n. h. mendelson and s. l. keener .
clockwise and counterclockwise pinwheel colony morphologies of are correlated with the helix hand of the strain .
, 151(1):455457 , 1982 .
n. h. mendelson and j. j. thwaites .
cell wall mechanical properties as measured with bacterial thread made from .
, 171(2):10551062 , 1989 .
m. mimura , h. sakaguchi , and m. matsushita .
a reaction - diffusion approach to bacterial colony formation .
, 1997 . j. d. murray . .
springer - verlag , berlin , 1989 .
g. nicolis and i. prigogine . .
w. h. freeman and company , new - york , 1989 .
h. parnas and l. segel .
a computer simulation of pulsatile aggregation in _
dictyostelium discoideum_. , 71:185207 , 1978 .
pelce and a. pocheau .
geometrical approach to the morphogenesis of unicellular algae .
, 156:197214 , 1992 .
m. doudoroff r. y. stainer and e. a. adelberg . .
prentice - hall and inc . , n. j. , 1957 .
i. rafols . .
thesis , chuo university , japan , 1998 .
e. rosenberg , editor . .
springer series in molecular biology .
springer - verlag , 1984 .
fractal growth processes . , 322:789793 , 1986 .
j. e. segall , s. m. block , and h. c. berg .
temporal comparisons in bacterial chemotaxis . , 83:89878991 , 1986 .
j. a. shapiro .
bacteria as multicellular organisms . , 258(6):6269 , 1988 .
j. a. shapiro and d. trubatch .
sequential events in bacterial colony morphogenesis . , 49:214223 , 1991 .
c. h. shaw .
swimming against the tide : chemotaxis in _
agrobacterium_. , 13(1):2529 , 1991 .
o. shochet . .
phd thesis , tel - aviv university , 1995 .
o. shochet , k. kassner , e. ben - jacob , s.g .
lipson , and h. mller - krumbhaar .
morphology transition during non - equilibrium growth : i. study of equilibrium shapes and properties .
, 181:136155 , 1992 .
o. shochet , k. kassner , e. ben - jacob , s.g .
lipson , and h. mller - krumbhaar .
morphology transition during non - equilibrium growth : ii .
morphology diagram and characterization of the transition .
, 187:87111 , 1992 .
s. j. stahl , k. r. stewart , and f. d. williams .
extracellular slime associated with during swarming .
, 154(2):930937 , 1983 .
j. b. stock , a. m. stock , and m. mottonen .
signal transduction in bacteria .
, 344:395400 , 1990 .
matsuyama t , kaneda k , nakagawa y , isa k , h. hara - hotta , and yano i. a novel extracellular cyclic lipopeptide which promotes flagellum - dependent and -independent spreading growth of _ serratia marcescens_. , 174:17691776 , 1992 .
t. vicsek . .
world scientific , new york , 1989 .
n. wiener . .
wiley , new - york , 1948 .
h. zabrodsky and d. avnir .
continuous symmetry measures .4 .
, 117:462473 , 1995 .
1 . [ fig : intro ] ( a ) typical example of branching growth of the ( referred to as @xcite)for 0.5 g / l peptone and 1.75 % agar concentration .
( b ) chiral growth of the ( referred to as @xcite ) for 2.5 g / l peptone and 1.25 % agar concentration .
[ fig:2.1 ] patterns exhibited by the morphotype as function of peptone level ( increasing from left to right ) and agar concentration ( 1.5% bottom row , 2% middle row , 2.5% top row ) .
[ fig:2.2 ] examples of typical patterns of for intermediate agar concentration .
( a ) compact growth for 12@xmath2 peptone level and 1.75% agar concentration .
( b ) dense fingers for @xmath181 peptone and 2% agar .
( c ) branching fractal pattern for @xmath182 peptone and 1.75% agar .
( d ) a pattern of fine radial branches for @xmath183 peptone and 1.75% agar .
[ fig:2.3 ] colonial patterns of .
( a ) fractal pattern for @xmath184 peptone level and 1.75% agar concentration .
( b ) dense branching pattern for @xmath185 peptone and 2.5% agar . note that the branches are much thinner than those in fig .
[ fig:2.2]b , i.e. the branches are thinner for higher higher agar concentrations . 5 . [ fig:2.4 ] ( a ) pattern of concentric rings superimposed on a branched colony for 2.5@xmath2 peptone level and 2.5% agar concentration .
( b ) concentric rings in a compact growth for @xmath186 peptone level and 2.25% agar concentration . 6 .
[ fig:2.5 ] weak chirality exhibited by the during growth on 4@xmath2 peptone and 2.5% agar concentration
[ fig:2.6 ] density variations within branches of a colony of .
optical microscope observations @xmath187 .
[ fig:2.9 ] closer look on branches of a colony .
a ) @xmath188 magnification shows the sharp boundaries of the branches .
the width of the boundary is in the order of micron .
b ) numarsky ( polarized light ) microscopy shows the hight of the branches and their envelope .
what is actually seen is the layer of lubrication fluid , not the bacteria .
c ) @xmath187 magnification shows the bacteria inside a branch .
each bar is a single bacterium .
there are no bacteria outside the branch .
[ fig:2.10 ] electron microscope observation of bacteria . round or oval shapes with bright center are spores .
elongated shapes are living cells .
the cells engulfing oval shapes are pre - spores .
[ fig:2.14 ] patterns exhibited by the for different growth conditions .
a ) thin disordered twisted branches at 0.5@xmath2 peptone level and 1.5% agar concentration .
b ) thin branches , all twisted with the same handedness . at 2@xmath2 peptone level and 1.25% agar concentration .
c ) pattern similar to ( b ) but on softer agar : 1.4 @xmath2 peptone level and 0.75% agar concentration .
d ) four inocula on the same plate , conditions of 1@xmath2 peptone level and 1.25% agar concentration . 11 .
[ fig:2.16 ] in agar soft enough for the bacteria to swim in it , the branches lose the one - side handedness they have on harder agar .
the two colonies of ( a ) and ( b ) are of 5@xmath2 peptone level and 0.6% agar concentration .
the two patterns are of two stains of the , strains whose patterns are indistinguishable on harder agar .
c ) closer look ( @xmath189 magnification ) on a colony grown at 8@xmath2 peptone level and 0.6% agar concentration .
[ fig:2.17 ] optical microscope observations of branches of colony .
a ) @xmath188 magnification of a colony at 1.6@xmath2 peptone level and 0.75% agar concentration , the anti - clockwise twist of the thin branches is apparent .
the curvature of the branches is almost constant throughout the growth .
b ) @xmath190 magnification of a colony at 4@xmath2 peptone level and 0.6% agar concentration at branches are not thin , but have a feathery structure .
the curvature of the branches varies , but it seems that at any given stage of growth the curvature is similar in all branches .
that is , the curvature is a function of colonial growth .
c ) @xmath191 magnification of a colony at 1.6@xmath2 peptone level and 0.75% agar concentration .
each line is a bacterium .
the bacteria are long ( 5 - 50@xmath192 ) and mostly ordered . 13 .
effect of varying the initial nutrient concentration @xmath84 on colony pattern .
@xmath84 increases from left to right : 1.2 ( a ) , 1.4 ( b ) , 1.7 ( c ) , 2 ( d ) , 3 ( e ) , 6 ( f ) . the minimal value of @xmath84 to support growth is @xmath71 .
[ fig - nutrient ] 14 .
effect of varying @xmath75 , the fluid absorption rate , on colony pattern .
the fluid production rate @xmath78 is @xmath71 in the upper row and @xmath193 in the lower row . in both rows
@xmath75 increases from left to right : @xmath194 ( left ) , @xmath195 ( center ) , @xmath196 ( right ) the patterns become more ramified as @xmath75 increases . decreasing
@xmath78 also produces a more ramified pattern .
the other parameters are : @xmath197 [ fig - lambda ] 15 .
[ fig : k1 ] growth patterns of the non - linear diffusion model , for different values of initial nutrient level @xmath84 .
parameters are : @xmath198 .
the apparent 6-fold symmetry is due to the underlying tridiagonal lattice .
[ fig : c - results ] a morphology diagram of the communicating spinors model for various values of @xmath199 and initial @xmath83 concentration @xmath84 .
@xmath200 , @xmath201 , @xmath202 , @xmath203 .
[ fig : c - t ] when the noise @xmath144 is increases to @xmath169 the tumbling of the spinors becomes unrestricted .
their movement becomes like that of the bacteria and accordingly the simulated colonial pattern is like that of . on the left @xmath201 , on the right @xmath169 .
[ fig : k - food - rep ] growth patterns of the non - linear diffusion model with food chemotaxis ( left , see section [ sec : bio - bg ] ) and repulsive chemotactic signaling ( right ) included .
other parameters are the same as in figure [ fig : k1 ] .
the apparent 6-fold symmetry is due to the underlying tridiagonal lattice .
[ fig : cmodel : chemo ] the effect of repulsive chemotactic signaling on the communicating spinors model .
a ) without chemotaxis .
b ) with repulsive chemotaxis .
the spinors are repelled from the inner parts of the colony .
the resulting curvature of the branches is reduces when they are in the radial direction . in spite
the reversed handedness , the pattern resemble fig .
[ fig:2.16]b . 20 .
[ fig : cmodel : long ] the snake - like branches observed in fig .
[ fig:2.16]a can be reproduced by the communicating spinors model .
@xmath112 is a continuous function of the colony s radius ( the same function in a , b , and c ) .
maximal value of @xmath112 is @xmath205 , minimal value is @xmath206 .
( a ) with repulsive chemotactic signaling .
( b ) without chemotaxis .
( c ) with food chemotaxis .
the best resemblance to the observed colony is obtained with repulsive chemotactic signaling .
[ fig : weak ] weak chirality ( global twist of the branches ) exhibited by the for a peptone level of 0.25@xmath2 peptone level and agar concentration of 1.75 % .
[ fig : k - weak ] growth patterns of the non - linear diffusion model with a `` squinting '' repulsive chemotactic signaling , leading to weak chirality .
parameters are as in the previous picture , @xmath207 . 23 . [ fig : cmodel : weak ] weak chirality of the is modeled by the communicating spinors model .
both simulations are with repulsive chemotactic signaling and with bias in the walkers rotation .
a ) with free rotation ( @xmath169 ) the pattern is branched , without apparent chirality . expressed .
b ) with constrained rotation ( @xmath170 ) weak chirality is expressed . 24 .
[ fig : v ] a colony of on 10 @xmath2 peptone level and 2% agar concentration .
the dots at the tips of the branches are bacterial vortices each is composed of up to millions of bacterial cells rotating around a common center . the twist of the branches results from a magnus force induced by repulsive chemotactic signaling . | in nature , microorganisms must often cope with hostile environmental conditions . to do
so they have developed sophisticated cooperative behavior and intricate communication capabilities , such as : direct cell - cell physical interactions via extra - membrane polymers , collective production of extracellular `` wetting '' fluid for movement on hard surfaces , long range chemical signaling such as quorum sensing and chemotactic ( bias of movement according to gradient of chemical agent ) signaling , collective activation and deactivation of genes and even exchange of genetic material . utilizing these capabilities , the colonies develop complex spatio - temporal patterns in response to adverse growth conditions .
we present a wealth of branching and chiral patterns formed during colonial development of lubricating bacteria ( bacteria which produce a wetting layer of fluid for their movement ) . invoking ideas from pattern formation in non - living systems and using `` generic '' modeling we are able to reveal novel survival strategies which account for the salient features of the evolved patterns . using the models ,
we demonstrate how communication leads to self - organization via cooperative behavior of the cells . in this
regard , pattern formation in microorganisms can be viewed as the result of the exchange of information between the micro - level ( the individual cells ) and the macro - level ( the colony ) .
we mainly review known results , but include a new model of chiral growth , which enables us to study the effect of chemotactic signaling on the chiral growth .
we also introduce a measure for weak chirality and use this measure to compare the results of model simulations with experimental observations .
# 1#1_paenibacillus dendritiformis _ -#1_p .
dendritiformis _ # 1 # 1#1 -#1 # 1 # 1#1 -#1 # 1 # 1#1_paenibacillus vortex _ -#1_p .
vortex _ # 1 # 1#1_bacillus subtilis _ -#1_b .
subtilis _ # 1 # 1#1bacillus # 1#1 _ circulans _ -#1 _ circulans _ # 1 # 1#1_escherichia coli _ -#1_e .
coli _
# 1 # 1#1_salmonella typhimurium _ -#1_s .
typhimurium _ # 1 # 1#1_myxococcus xanthus _ -#1_m .
xanthus _ # 1 # 1morphotype#1 # 1morphotype#1 # 1morphotype#1 # 1morphotype#1 # 1#2#1#2 # 1 = 4.48 truein |
the undeniable influence of type ia supernovae explosions in , among others , the chemical evolution of the galaxy and cosmology makes the quest of solid theoretical models as an urgent task .
guided by the insight gained from two decades of modelization of spherically symmetric models , modern multidimensional calculations try to make a reliable picture of the explosion on physical basis , minimizing the number of free parameters .
nevertheless , even supossing that the exploding object is a co white dwarf near to the chandrasekhar - mass limit the subject is not firmly resolved yet . in particular , there remain uncertainties in the immediate pre - explosive stage and in the mechanism / s by which the thermonuclear flame is spreaded through the white dwarf . in this regard a difficult point
is related with the acceleration of the combustion .
we know that the explosion is triggered by a thermonuclear flame whose initial velocity is clearly subsonic .
we also know , from parametrized calculations , that after several sound - crossing times the effective velocity of the deflagration must reach a significant fraction of of the local sound speed in order to synthesize the adequate amount of nuclear species to explain both light curve and spectra .
this translates into an increase factor of thousands over the pure laminar value at maximum effective velocity .
the physical mechanism behind this huge factor is currently attributed to flame surface corrugation by hydrodynamical instabilities ( mainly rayleigh - taylor and turbulence ) .
a branch in this scenario could be achieved if the already fast deflagration turns into a steady detonation at late times , when flame density is below @xmath3 g.@xmath4 .
this is the so called delayed - detonation scenario which , in calculations carried out in spherical symmetry , have provided some of the best models for snia in the past @xcite . in this short paper we present the results of a calculation concerning the delayed detonation model carried out in three dimensions by using a sph code .
we describe the main features of the evolution of the model with more emphasis in the nucleoshynthetical side .
crrrrrrr t(s ) & @xmath5 ( g.@xmath6&@xmath7 ( g.@xmath6 & @xmath8 @xmath9&@xmath10 & @xmath11 & @xmath12 + 0&@xmath13&@xmath14&@xmath15&@xmath16&@xmath17 & @xmath18 + 1.54&@xmath19&@xmath20&@xmath21&@xmath22 & @xmath23&@xmath24 + 7.92&@xmath25&@xmath26&@xmath27&@xmath28 & @xmath29 & @xmath30 + first column is the elapsed time since the beginning of the sph simulation .
previously , the initial stages of the explosion were calculated with a lagrangian one - dimensional code .
the initial model is a white dwarf of @xmath31 m@xmath0 in hydrostatic equilibrium .
the first stages of the explosion were followed by using a 1d hydrocode until the central density declined to @xmath32 g.@xmath33 . from here on the model was mapped into a 3d distribution of 250,000 particules and its evolution followed with the sph . in addition , the velocity field around the flame front was perturbed at the beginning in order to seed the hydrodynamical instabilities .
this allows us to describe the large scale deformation of the flame in a self - consistent way although a lot of surface is lost owing to the limited resolution of the code . to take into account these hidden lengthscales
a subgrid model has to be set .
we assume that the main effect of the smallest lengthscales is to increase the laminar velocity of the flame , @xmath34 , until a first effective value @xmath35 .
afterwards the final effective velocity @xmath36 is directly computed from the hydrocode @xcite . during the progression of the calculation
the geometrical features of the flame front were tracked by calculating the main scaling parameters of the surface such as @xmath37 , @xmath38 and flame fractal dimension @xmath39 by using the method described in @xcite . as discussed in @xcite the local evolution of @xmath39 can be used as a practical criteria to turn the deflagration into a detonation .
in particular , we have allowed such transition in those regions of the flame with fractal dimension higher than 2.5 once flame mean density is @xmath40 g.@xmath4 .
the relevant features of the evolution of the model are summarized in table 1 .
basically the explosion proceeded in two phases . from t=0
s to t=1.54 s the combustion propagated subsonically as a deflagration .
however , a considerable acceleration of the flame was seen because the effective flame velocity passed from @xmath41 for @xmath42 to @xmath43 at @xmath44 . at the end of this phase
the total energy was already positive , although too low to represent a normal supernovae explosion . at t=1.54
s a detonation was artificially induced in those regions of the flame surface which had a large fractal dimension . in consequence
multiple detonations emerged from unconnected points spreaded on a very irregular surface .
information about the nucleosynthetic yields is provided in table 1 and figures 1,2 .
the ejected amount of @xmath45 is enough to power the light curve of a typical type ia supernovae and the amount of silicon and other intermediate - mass elements is consistent with the spectrum near maximum light .
a clear difference with respect 1d calculations is that the abundance distribution in velocity space show a lot of dispersion in comparison .
the isotopic composition of the ejecta , calculated by postprocessing the dynamical model , is depicted in figure 2 .
although we see a large production of vanadium there is not overproduction of problematic isotopes such as @xmath1 or @xmath2 .
this is a feature of multidimensional models where electron captures proceed , on average , at lower density owing to the buoyancy of high entropy ashes .
the ejected amount of carbon and oxygen is somewhat high , being distributed in isolated pockets within the star but not close to the center .
our results suggest that the three - dimensional version of the delayed detonation model is also able to reproduce the main observational features of type ia supernovae . 9 p. hflich and a. khokhlov , apj 457 ( 1996 ) 500 .
d. garca - senz and e. bravo , in proceedings of the eso / mpa / mpe workshop `` from twilight to highlight : the physics of supernovae '' , ( 2003 ) , in press
. d. garca - senz , e. bravo and n. serichol , apjss 115 ( 1998 ) 119 | the explosion of a white dwarf of mass 1.36 m@xmath0 has been simulated in three dimensions with the aid of a sph code .
the explosion follows the delayed detonation paradigma . in this case
the deflagration - detonation transition is induced by the large corrugation of the flame front resulting from rayleigh - taylor instability and turbulence .
the nucleosynthetic yields have been calculated , showing that some neutronized isotopes such as @xmath1 or @xmath2 are not overproduced with respect to the solar system ratios .
the distribution of intermediate - mass elements is also compatible with the spectra of normal snia .
the excepcion is , however , the abundance of carbon and oxygen , which are overproduced . |
secure communication achieved by exploiting the wireless physical layer to provide secrecy in data transmission , has drawn significant recent research attention ( see e.g. @xcite and references therein ) .
the performance of this type of secure communication can be measured in terms of the secrecy throughput , which is the capacity of conveying information to the intended users while keeping it confidential from eavesdroppers @xcite . on the other hand ,
energy efficiency ( ee ) has emerged as another important figure - of - merit in assessing the performance of communication systems @xcite . for most systems ,
both security and energy efficiency are of interest , and thus it is of interest to combine these two metrics into a single performance index called the secrecy ee ( see ) , which can be expressed in terms of secrecy bits per joule .
transmit beamforming can be used to enhance the two conflicting targets for optimizing see in multiple - user multiple - input multiple - output ( mu - mimo ) communications : mitigating mu interference to maximize the users information throughput , and jamming eavesdroppers to control the leakage of information . however , the current approach to treat both ee @xcite and see @xcite is based on costly zero - forcing beamformers , which completely cancel the mu interference and signals received at the eavesdroppers .
the ee / see objective is in the form of a ratio of a concave function and a convex function , which can be optimized by using dinkelbach s algorithm @xcite .
each dinkelbach s iteration still requires a log - det function optimization , which is convex but computationally quite complex .
moreover , zero - forcing beamformers are mostly suitable for low code rate applications and are applicable to specific mimo systems only .
the computational complexity of see for single - user mimo / siso communications as considered in @xcite and @xcite is also high as each iteration still involves a difficult nonconvex optimization problem .
this letter aims to design transmit beamformers to optimize see subject to per - user secrecy quality - of - service ( qos ) and transmit power constraints .
the specific contributions are detailed in the following dot - points . * a path - following computational procedure , which invokes a simple convex quadratic program at each iteration and converges to at least a locally optimal solution ,
is proposed .
the mu interference and eavesdropped signals are effectively suppressed for optimizing the see .
in contrast to zero - forcing beamformers , higher code rates not only result in transmitting more concurrent data streams but also lead to much better see performance in our proposed beamformer design .
* as a by - product , other important problems in secure and energy - efficient communications , such as ee maximization subject to the secrecy level or sum secrecy throughput maximization , which are still quite open for research , can be effectively addressed by the proposed procedure .
_ notation .
_ all variables are written in boldface .
for illustrative purpose , @xmath0 is a mapping of variable @xmath1 while @xmath2 is the output of mapping @xmath3 corresponding to a particular input @xmath4 .
@xmath5 denotes the identity matrix of size @xmath6 .
the notation @xmath7 stands for the hermitian transpose , @xmath8 denotes the determinant of a square matrix @xmath9 , and @xmath10 denotes its trace while @xmath11 .
the inner product @xmath12 is defined as @xmath13 and therefore the frobenius squared norm of a matrix @xmath14 is @xmath15 .
the notation @xmath16 ( @xmath17 , respectively ) means that @xmath18 is a positive semidefinite ( definite , respectively ) matrix .
@xmath19 $ ] denotes expectation and @xmath20 denotes the real part of a complex number .
@xmath21 denotes a circularly - symmetric complex gaussian random variable with mean zero and variance @xmath22 .
consider a mimo system consisting of @xmath23 transmitters and @xmath23 users indexed by @xmath24 .
each transmitter @xmath25 is equipped with @xmath26 antennas to transmit information to its intended user @xmath25 equipped with @xmath27 antennas .
there is an eavesdropper equipped with @xmath28 antennas , which is part of the legitimate network @xcite .
the channel matrices @xmath29 and @xmath30 from transmitter @xmath31 to user @xmath25 and to the eavesdropper , respectively , are known at the transmitters by using the channel reciprocity , feedback and learning mechanisms @xcite .
a complex - valued vector @xmath32 contains the information transmitter @xmath25 intends to convey to user @xmath25 , where @xmath33=i_{d_1}$ ] , and @xmath34 is the number of concurrent data streams .
denote by @xmath35 the complex - valued beamformer matrix for user @xmath25 .
the ratio @xmath36 is called the code rate of @xmath37 . for notational convenience ,
define @xmath38 and @xmath39_{j\in{{\cal d}}}$ ] .
the received signal at user @xmath25 and the signal received at the eavesdropper are @xmath40 where @xmath41 and @xmath42 are additive noises . by ( [ fd1 ] ) , the rate of information @xmath43 leaked from user @xmath25 ( in nats ) is @xmath44 where @xmath45 and @xmath46 .
+ on the other hand , the wiretapped throughput for user @xmath25 at the eavesdropper is @xmath47 where @xmath48 and @xmath49 .
the secrecy throughput in transmitting information @xmath50 to user @xmath25 while keeping it confidential from the eavesdropper is defined as @xcite @xmath51 following @xcite , the consumed power for signal transmission is modelled by @xmath52 , where @xmath53 is the total transmit power of the transmitters and @xmath54 and @xmath55 are the reciprocal of the drain efficiency of the power amplifier and the circuit power , respectively .
consider the following secure beamformer design to optimize the system s energy efficiency : [ fd5 ] @xmath56 where the constraints ( [ fd5b ] ) limit the transmit power , while ( [ fd5c ] ) are the secrecy qos constraints .
it can be seen from their definitions ( [ fd3 ] ) and ( [ fd4 ] ) that both throughput @xmath57 and wiretapped throughput @xmath58 are very complicated functions of the beamformer variable @xmath1 .
the approach of @xcite and @xcite ( to ee ) and @xcite and @xcite ( to see ) seeks @xmath1 in the class of zero - forcing beamformers @xmath59 , @xmath60 and @xmath61 to cancel completely all the mu interference and wiretapped signals .
each throughput @xmath57 becomes a log - det function of only @xmath62 .
dinkelbach s algorithm is then applied to compute a zero - forcing solution of ( [ fd5 ] ) , which requires a log - det function optimization for each iteration .
such optimization is still computationally difficult with no available polynomial - time solvers .
note that the feasibility of the zero - forcing constraints imposes @xmath63 and @xmath64 @xcite .
thus , there is not much freedom for optimizing zero - forcing beamformers whenever @xmath26 is not large . in the next section
, we will provide a completely new computational approach to ( [ fd5 ] ) by effectively enhancing its difficult objective and constraints .
by introducing a variable @xmath65 satisfying the convex quadratic constraint @xmath66 the optimization problem ( [ fd5 ] ) can be equivalently expressed as @xmath67 in what follows , a function @xmath68 is said to be a _
( _ majorant _ , resp . ) of a function @xmath3 at a point @xmath69 in the definition domain @xmath70 of @xmath3 iff @xmath71 and @xmath72 @xmath73 ( @xmath74 @xmath73 , resp . )
@xcite . by @xcite , a _
concave quadratic minorant _ of the throughput function @xmath75 at @xmath76_{j\in{{\cal d}}}$ ] , which is feasible for ( [ fd5b])-([fd5c ] ) is @xmath77 where @xmath78 , @xmath79 , @xmath80 and @xmath81 to provide a minorant of the secrecy throughput @xmath82 ( see ( [ d2 ] ) ) at @xmath83 , the next step is to find a _
majorant _ of the eavesdropper throughput function @xmath84 at @xmath83 .
reexpressing @xmath58 by @xmath85 for @xmath86 , and applying theorem [ baseth ] in the appendix for upper bounding the first term and lower bounding the second term in ( [ fd4.1 ] ) yields the following _ convex quadratic majorant _ of @xmath58 at @xmath83 : @xmath87 where @xmath88 , and @xmath89 a _ concave quadratic minorant _ of the secrecy throughput function @xmath82 at @xmath83 is then @xmath90 here , @xmath91 , @xmath92 , and @xmath93 .
therefore , the nonconvex secrecy qos constraints ( [ fd5c ] ) can be innerly approximated by the following convex quadratic constraints in the sense that the feasibility of the former is guaranteed by the feasibility of the latter : @xmath94 for good approximation , the following trust region is imposed : @xmath95 by using the inequality @xmath96 we obtain @xmath97 , for @xmath98 where @xmath99 , @xmath100 , which is a concave function @xcite . with regard to @xmath101
we define a concave function @xmath102 as follows : * if @xmath103 , define @xmath104 , which is a concave function ; * if @xmath105 , define @xmath106 , which is a linear minorant of the convex function @xmath101 at @xmath107 .
a _ concave minorant _ of @xmath108 , which is also a minorant of @xmath109 at @xmath110 , is thus @xmath111 we now solve the nonconvex optimization problem ( [ fd5 ] ) by generating the next feasible point @xmath112 as the optimal solution of the following convex quadratic program ( qp ) , which is an inner approximation @xcite of the nonconvex optimization problem ( [ fd5.1 ] ) : @xmath113 note that ( [ kappa1 ] ) involves @xmath114 scalar real variables and @xmath115 quadratic constraints so its computational complexity is @xmath116 .
+ it can be seen that @xmath117 as long as @xmath118 , i.e. @xmath119 is better than @xmath110 .
this means that , once initialized from a feasible point @xmath120 for ( [ fd5.1 ] ) , the @xmath121-th qp iteration ( [ kappa1 ] ) generates a sequence @xmath122 of feasible and improved points toward the nonconvex optimization problem ( [ fd5.1 ] ) , which converges at least to a locally optimal solution of ( [ fd5 ] ) @xcite . under the stopping criterion @xmath123 for a given tolerance @xmath124 , the qp iterations will terminate after finitely many iterations .
_ initialization : _ set @xmath125 , and choose a feasible point @xmath126 for ( [ fd5.1 ] ) .
_ @xmath121-th iteration : _ solve ( [ kappa1 ] ) for an optimal solution @xmath127 and set @xmath128 , @xmath129 and calculate @xmath130 . stop if @xmath131 @xmath132 . the proposed path - following procedure for computational solution of the nonconvex optimization problem ( [ fd5 ] )
is summarized in algorithm [ alg1 ] .
we note that a feasible initial point @xmath126 for ( [ fd5.1 ] ) can be found by solving @xmath133 by the iterations @xmath134 , which terminate upon reaching @xmath135 to satisfy ( [ fd5b])-([fd5c ] )
. the following problem of ee optimization under users throughput qos constraints and secrecy levels : @xmath136 where @xmath137 is set small enough to keep the users information confidential from the eavesdropper , is simpler than ( [ fd5 ] ) .
it can be addressed by a similar path - following procedure , which solves the following qp at the @xmath138th iteration instead of ( [ kappa1 ] ) : [ cee ] @xmath139 where @xmath140 , @xmath141 and @xmath142 and @xmath143 are defined from ( [ thetajk ] ) . a feasible initial point @xmath126 for ( [ tst1 ] ) can be found by solving @xmath144 by the iterations @xmath145 which terminate upon reaching @xmath146 , @xmath147 @xmath148 , to satisfy ( [ fd5b ] ) , ( [ tst1 ] ) .
lastly , the problem of sum secrecy throughput maximization @xmath149 is also simpler than the see optimization problem ( [ fd5 ] ) , which can be addressed by a similar path - following procedure with the qp @xmath150 solved at the @xmath138th iteration instead of ( [ kappa1 ] ) .
the fixed parameters are : @xmath151 , @xmath152 , @xmath153 , @xmath154 , @xmath155 , @xmath156 , @xmath157 bits / s / hz , @xmath158 and @xmath159 db . the secrecy level @xmath160 is set in solving ( [ tst1 ] ) .
the channels are rayleigh fading so their coefficients are generated as @xmath161 .
for the first numerical example , the number of data streams @xmath162 is set , so the code rate is @xmath163 .
each @xmath62 is of size @xmath164 .
figure [ fig1 ] shows the see performance of our proposed beamformer and the zero - forcing beamformer @xcite .
one can see that the former outperforms the latter substantially .
apparently , the latter is not quite suitable for both ee and see .
the see performance achieved by the formulation ( [ fd5 ] ) is better than that achieved by the formulation ( [ tst1 ] ) because the secrecy level is enhanced with the users s throughput in the former instead of being constrained beforehand in the latter .
when the transmit power @xmath165 is small , the denominator of the see objective in ( [ fd5 ] ) and ( [ tst1 ] ) is dominated by the constant circuit power @xmath55 . as a result ,
the see is maximized by maximizing its numerator , which is the system sum secrecy throughput . on the other hand ,
the see objective is likely maximized by minimizing the transmitted power @xmath165 in its denominator when the latter is dominated by @xmath165 .
that is why the see saturates once @xmath165 is beyond a threshold according to figure [ fig1 ] . for @xmath162 .
, scaledwidth=90.0% ] we increase the number @xmath166 of data streams to @xmath167 in the second numerical example .
the code rate is thus @xmath168 . for this higher - code - rate case ,
the zero - forcing beamformers @xcite are infeasible . comparing figure
[ fig1 ] and figure [ fig2 ] reveals that higher code - rate beamforming is also much better in terms of see because it leads to greater freedom in designing @xmath62 of size @xmath169 for maximizing the see . in other words ,
the effect of code rate diversity on the see is observed . for @xmath170 .
, scaledwidth=90.0% ]
we have proposed a path - following computational procedure for the beamformer design to maximize the energy efficiency of a secure mu mimo wireless communication system and have also showed its potential in solving other important optimization problems in secure and energy - efficient communications .
simulation results have confirmed the superior performance of the proposed method over the exiting techniques
. * acknowledgement . *
the authors thank dr . h.
h. kha for providing the computational code from @xcite .
[ baseth ] for a given @xmath171 , consider a function @xmath172 in @xmath173 . then for any @xmath174 , it is true that @xmath175 with the _ concave _ quadratic function @xmath176 and the _ convex _ quadratic function @xmath177 where @xmath178 , @xmath179 , and @xmath180 , @xmath181 . both functions @xmath68 and @xmath182 agree with @xmath3 at @xmath183 . _
proof . _ due to space limitations , we provide only a sketch of the proof .
rewrite @xmath184 , which is convex as a function in @xmath185 @xcite .
then @xmath186 defined by ( [ b2 ] ) actually is the first order approximation of this function at @xmath187 , which is its minorant at @xmath187 @xcite , proving the first inequality in ( [ b1 ] ) .
+ on the other hand , considering @xmath3 as a concave function in @xmath188 , @xmath189 defined by ( [ b3 ] ) is seen as its first order approximation at @xmath190 and thus is its majorant at @xmath190 @xcite , proving the second inequality in ( [ b1 ] ) .
t. m. hoang , t. q. duong , h. a. suraweera , c. tellambura , and h. v. poor , `` cooperative beamforming and user selection for improving the security of relay - aided systems , '' _ ieee trans .
_ , vol .
63 , no . 12 , pp .
50395050 , dec . 2015 .
r. l. g. cavalcante , s. stanczak , m. schubert , a. eisenlatter , and u. turke , `` toward energy - efficienct 5 g wireless communications technologies , '' _ ieee signal process . mag .
_ , vol . 13 , no . 11 , pp .
2434 , nov .
2014 .
n. zhao , f. r. yu , and h. sun , `` adaptive energy - efficient power allocation in green interference - alignment - based wireless networks , '' _ ieee trans
_ , vol .
64 , no . 9 , pp . 42684281 , sept .
2015 .
a. kalantari , s. maleki , s. chatzinotas , and b. ottersten , `` secrecy energy efficiency optimization for miso and siso communication networks , '' in _ proc .
ieee 16th inter .
workshop on signal process .
advances in wireless commun.(spawc ) _ , 2015 , pp . 2125 .
a. zappone , p .- h .
lin , and e. jorswieck , `` energy efficiency of confidential multi - antenna systems with artificial noise and statistical csi , '' _ ieee j. selec .
topics signal process . _ ,
10 , no . 8 , pp . 14621477 , aug .
2016 .
h. h. m. tam , h. d. tuan , and d. t. ngo , `` successive convex quadratic programming for quality - of - service management in full - duplex mu - mimo multicell networks , '' _ ieee trans .
_ , vol .
64 , pp . 23402353 , jun . 2016 . | considering a multiple - user multiple - input multiple - output ( mimo ) channel with an eavesdropper , this letter develops a beamformer design to optimize the energy efficiency in terms of secrecy bits per joule under secrecy quality - of - service constraints .
this is a very difficult design problem with no available exact solution techniques .
a path - following procedure , which iteratively improves its feasible points by using a simple quadratic program of moderate dimension , is proposed . under any fixed computational tolerance
the procedure terminates after finitely many iterations , yielding at least a locally optimal solution .
simulation results show the superior performance of the obtained algorithm over other existing methods .
mimo beamforming , secure communication , energy efficiency . |
high resolution imaging at optical , near - infrared , and millimeter - wave wavelengths have conclusively demonstrated that circumstellar disks are ubiquitous around young solar - mass stars @xcite .
the long - held notion that planets may frequently form in these disks has been dramatically confirmed with the increasing number of planets detected around nearby main - sequence stars @xcite . while a consensus has not yet been reached on how planets form in circumstellar disks , current theories propose that gas - giant planets form either as gravitational instabilities in an accretion disk @xcite , or from the more gradual build up of planetesimals to form a rocky core followed by accretion of gaseous material @xcite . in both class of models
, the accretion disk lifetime sets an important time scale on the planet formation process by constraining the time available for a gravitational instability to develop or the rocky core to form before the gas and dust are dissipated .
circumstellar accretion disks around young stars are commonly inferred based on the presence of infrared emission in excess of the stellar photosphere . in many stars the excess infrared emission from 1 - 100 can be modeled with a geometrically thin , optically thick accretion disk @xcite , although inner disk holes , flaring , and other refinements to this basic model are often needed to reproduce the observations in detail @xcite .
detailed spectral energy distributions from the near- to far - infrared have been compiled for only a few nearby star forming regions ( e.g. strom et al .
1989 ; wilking , lada , & young 1989 ) , but @xmath11- , @xmath12- , @xmath13- , and @xmath14-band imaging studies provide an efficient means to search for near - infrared excesses in hundreds of stars ( often in clusters ) of various ages .
recent results indicate that at least 50% of solar mass stars at an age of @xmath31 myr exhibit a near - infrared excess , but that the percentage decreases to @xmath1510% for ages 3 - 10 myr @xcite .
these results are consistent with the notion that the lifetime of accretion disks around most solar mass stars is @xmath33 - 10 myr .
strictly speaking , a near - infrared excess is diagnostic of only hot dust with temperatures of @xmath31000 k. for a solar mass star , these temperatures are found only within @xmath30.1 au of the star .
most of the dust mass , and presumably the majority of planet formation , will be located at larger radii and have temperatures too cool to radiate in the near - infrared .
further , some disks are thought to have inner holes such that stars without near - infrared excesses may still contain substantial disk masses .
( although 10 observations suggest that the transition from an optically thick to an optically thin disk occurs on time scales of @xmath30.3 myr , in which case near - infrared observations may in fact provide an indirect probe of the larger scale disk ; skrutskie et al .
1990 , see also haisch , lada , & lada 2001a ) .
an additional limitation of near - infrared studies is that the emission is optically thick and provides little meaningful constraints on the disk mass ( see , e.g. , wood et al .
longer wavelength observations are needed to establish the disk lifetime at large orbital radii and to provide measures on how disk masses evolve with time .
millimeter and submillimeter continuum emission are the best available tracers of dust in the cool , outer disk .
the most comprehensive continuum surveys to date have been toward young stars in taurus @xcite , @xmath16 oph @xcite , lupus @xcite , chamaeleon i @xcite , and serpens @xcite .
these observations have shown that at an age of @xmath31 myr , @xmath320 - 30% of the stars possess a circumstellar disk with a mass greater than the minimum mass of the solar nebula ( @xmath30.01m@xmath4 ; weidenschilling 1977 , hayashi 1981 ) , and that the median disk mass is @xmath150.004m@xmath4 .
further insights into disk evolution can be obtained by using rich clusters as a tool to empirically measure the mass of disks as a function of stellar mass and age in much the same manner near - infrared studies have used clusters to establish the frequency of inner accretion disks @xcite .
continuum surveys of clusters are also important since an appreciable fraction of stars in molecular clouds are found in rich clusters @xcite , and the high ultraviolet radiation fields produced by o - stars in clusters may influence disk evolution @xcite .
the main observational challenge to studying clusters in the millimeter continuum is that high angular resolution is required to resolve the stars and distinguish disk emission from the more extended molecular cloud .
the only rich cluster surveyed to date by interferometric techniques has been the orion nebula cluster @xcite .
however , these observations were 10 - 50 times less sensitive to dust mass than the observations of taurus and @xmath16 oph and did not place stringent limits on the amount of circumstellar mass around individual stars .
given the limitations in inferring disk evolution from near - infrared excesses , it is important to conduct millimeter continuum observations of additional star forming regions in order to understand the evolution of disk masses at young stellar ages when planets are in their formative stages .
to further measure the circumstellar disk masses in different star forming environments , i obtained @xmath173 mm continuum observations of the young cluster ic 348 with the ovro millimeter wave interferometer .
based on statistical analysis of near - infrared star counts , the ic 348 cluster contains @xmath3400 stars distributed over a @xmath3@xmath18 region ( @xmath19 at the assumed distance of 320 pc ; de zeeuw et al .
1999 ; herbig 1998 ) , with about half the stars located within the central @xmath20 region @xcite . to date , @xmath3200 stars and brown dwarfs have been individually identified as likely cluster members based on their spectroscopic and photometric properties @xcite , and 115 - 173 likely members have been detected with x - rays @xcite .
a cluster age of @xmath32 myr ( luhman 1999 ; see also herbig 1998 ) has been inferred by placing the low mass stars on the hr diagram and using theoretical pre - main - sequence evolutionary tracks from @xcite .
near - infrared imaging surveys of the cluster show that @xmath365% of the stars contain a @xmath21 excess characteristic of an optically thick accretion disk @xcite . by mosaicking the central @xmath0 region ( @xmath22 ) of the ic 348 cluster in the millimeter continuum , i investigate the circumstellar disk masses in 95 known cluster members .
the ovro observations of the ic 348 cluster are described in section [ obs ] .
the stellar properties of the cluster members within the ovro mosaic boundaries are reviewed in section [ cluster ] , and constraints on the circumstellar disk masses are derived in section [ constraints ] . section [ discussion ] compares these results with disk masses in taurus and the orion nebula cluster , and discuses the implications for planet formation .
the conclusions are summarized in section [ summary ] .
a @xmath0 region toward the ic 348 cluster was mosaicked in the @xmath173 mm continuum using the ovro millimeter - wave interferometer between september 2001 and january 2002 .
continuum data were recorded simultaneously in four , 1 ghz wide continuum channels centered at 96.48 ghz , 97.98 ghz , 100.98 ghz , and 102.48 ghz .
the phase and amplitude calibrator was j0336 + 323 ( @xmath23,@xmath24 = 03:36:30.10,+32:18:29.3 j2000 ) , which is located 1.7 from the cluster center .
the data were calibrated using the ovro mma data reduction package @xcite . the adopted continuum flux for j0336 + 323 , calibrated by observing neptune and/or uranus , was 1.60 jy .
the rms dispersion in the measured flux for j0336 + 323 is 0.10 jy as computed from observations on 12 separate nights in which both the calibrator and the planets were observed .
the mosaic consists of 64 pointing centers arranged in a @xmath25 grid as shown in figure [ fig : gain ] .
the pointing centers along a row are separated by 40@xmath26 ( compared to the primary fwhm beam size of 72@xmath26 ) .
adjacent rows are also separated by 40@xmath26 but with pointing centers shifted by 20@xmath26 in right ascension . the mosaicking sequence consisted of obtaining a 3 minute integration on the phase calibrator followed by observing 5 positions in the mosaic for 3 minutes each .
this sequence was repeated until all 64 positions in the mosaic were observed .
the mosaic was observed in the ovro compact , low , equatorial , and high configurations both to increase the uv sampling and to decrease the rms noise .
the top few rows in the mosaic were often observed at high airmass with lower sensitivity .
additional observations of these rows were obtained to provide more uniform sensitivity across the mosaic .
the 64 individual points were formed into an image using the mosaicking routines in miriad .
the data were averaged using natural weighting to maximize sensitivity , resulting in a beam size of @xmath27 .
unit gain was achieved over a @xmath3@xmath0 area as shown by the dotted curve in figure [ fig : gain ] .
the average rms noise within the unit gain region is 0.75 mjybeam@xmath2 and is spatially uniform over the mosaic to within 10% .
before using the ovro observations to constrain the circumstellar disk masses in the ic 348 cluster , i review the stellar properties of the ic 348 cluster members located within the mosaic boundaries . the stellar and substellar membership list of the ic 348 cluster was compiled from the spectroscopic observations by @xcite , @xcite , and @xcite , and the narrow band imaging survey by @xcite .
to ensure a reliable list of cluster members , only sources brighter than @xmath28=14.5 were considered cluster members for this study , as the field star contamination becomes appreciable at fainter magnitudes @xcite . in total , 95 cluster members brighter than @xmath28=14.5 have been identified within the ovro mosaic .
astrometry for these sources was adopted from 2mass . the astrometric uncertainty ( @xmath150.2@xmath26 )
is 20 times less than the synthesized beam size .
stellar masses and ages for 92 of the 95 cluster members have been estimated using the @xcite theoretical pre - main - sequence evolutionary tracks and published spectroscopic and photometric data as described in @xcite .
three sources do not have mass estimates since optical photometry is not available , or in one case , the spectral type could not be determined since the source is possibly a class i object @xcite .
figures [ fig : mass ] and [ fig : age ] show histograms of the inferred stellar masses and ages respectively .
( also shown in these figures are the stellar mass and age histograms for a subsample of the taurus population as discussed in section [ discussion ] . ) the stellar masses range from 0.03m@xmath4 to 4.2m@xmath4 with a broad peak in the distribution at about 0.2 - 0.5m@xmath4 .
the mean age of the stellar population is @xmath32 myr .
the properties of this subsample agree well with the properties of the entire ic 348 population identified so far @xcite . to investigate the frequency distribution of inner accretion disks in this subsample of the ic 348 cluster , figure [ fig : ccd ] shows the @xmath29 vs. @xmath30 and @xmath29 vs. @xmath31 color - color diagrams .
the @xmath11 , @xmath12 , and @xmath28 photometry are from 2mass , and the @xmath14-band photometry is from @xcite . only sources with no processing flags in the 2mass database are shown . in figure
[ fig : ccd ] , the solid curves represent the locus of main - sequence and giant stars @xcite and the dashed line is the interstellar reddening vector @xcite , where the @xmath29 and @xmath32 colors have been transformed into the 2mass photometric system @xcite . of the 82 sources in the @xmath29 vs. @xmath30 diagram , 4 have near - infrared colors to the right of the reddening vectors that are consistent with an infrared excess .
however , for each of these four stars , the magnitude of the apparent @xmath30 excess is less than the 1@xmath5 photometric uncertainties and therefore may be attributed to photometric noise .
the fraction of stars with an infrared excess is higher in the @xmath29 vs. @xmath31 diagram since @xmath14-band shows larger contrast between the disk and photospheric emission . of the 47 stars with available @xmath11 ,
@xmath12 , @xmath28 , and @xmath14-band photometry , 18 ( 38% ) contain an apparent @xmath31 excess .
@xcite do not report @xmath14-band photometric uncertainties for individual sources .
based on the distribution of stars blueward of the un - reddened main - sequence in the @xmath29 vs. @xmath31 diagram ( a region which can not be populated except from photometric noise and photometric variability between the time of the 2mass and @xmath14-band observations ) , i estimate that @xmath34 sources may have an apparent @xmath31 excess due to noise .
therefore , a minimum of @xmath3 15% of the 95 ic 348 members within the ovro mosaic have near - infrared colors indicative of an optically thick , inner accretion disk . by comparison ,
@xcite estimate that the @xmath365% of the stars in the ic 348 cluster brighter than @xmath13=12 have a @xmath21 excess .
figure [ fig : map ] shows gray scale and contour maps of the ovro mosaic . in the contour map ,
the dotted curve shows the unit gain boundary of the mosaic , and the open circles indicate the positions of the 95 known members of the ic 348 cluster within the unit gain boundary .
the contours begin at 3@xmath5 above the rms noise of 0.75mjybeam@xmath2 with increments of 1@xmath5 .
visual inspection of the contour map shows that none of the known ic 348 cluster members have been detected in the @xmath173 mm continuum at the 3@xmath5 level or greater . over the entire mosaic ,
the brightest observed flux is only 3.9@xmath5 above the noise , and the frequency distribution of fluxes are consistent with gaussian noise as shown by the solid circles in figure [ fig : flux ] .
therefore there are no definitive continuum detections in the mosaic .
circumstellar disks in taurus typically have diameters of @xmath15300 au as measured in the @xmath173 mm continuum @xcite . assuming that the disks in ic 348 are similar , the disks can be considered point sources at the resolution of these observations ( 1600 au@xmath331300 au ) when computing upper limits to the continuum flux .
a histogram of the observed fluxes toward the 95 cluster members is shown in figure [ fig : flux ] .
the dotted line through the histogram shows the expected flux distribution for gaussian noise given the observed rms noise of 0.75mjybeam@xmath2 .
the observed mean flux toward the cluster members is @xmath34mjy , where the uncertainty represents the standard deviation of the mean of the 95 cluster members .
these results indicate that while the flux distribution is consistent with gaussian noise , there is a positive bias to the average flux at the 2.8@xmath5 confidence level .
the ovro continuum observations can be used to place constraints on the circumstellar disk masses . in reality
, most disks will contain a range of dust temperatures with decreasing temperatures with increasing radial distance from the star .
the data obtained here do not permit sophisticated modeling of the disk structure , however , and a single dust temperature was assumed .
following @xcite , the disk mass ( i.e. sum of the dust and gas components ) can then be estimated as @xmath35 , where @xmath36 is the mass opacity coefficient , @xmath37 parameterizes the frequency dependence of @xmath38 , @xmath39 is the observed flux , @xmath40 is the dust temperature , and @xmath41 is the planck function . for consistency with prior studies ,
i assumed @xmath37=1.0 and @xmath42 = 0.02@xmath8 g@xmath2 at 1300 @xcite .
the value of @xmath42 , and consequently the disk masses , is uncertain by at least a factor of three @xcite .
the power law index @xmath37 may be as low as @xmath30.5 on average in circumstellar disks @xcite .
if this lower value of @xmath37 is more appropriate for circumstellar disks , the assumptions adopted here will overestimate the disk masses by 50% .
the appropriate dust temperature for calculating the disk masses was estimated using the circumstellar disk models from @xcite and @xcite .
these studies fitted radial power - law distributions for the dust temperature and disk mass surface density to the observed infrared and millimeter spectral energy distribution for well - studied stars in taurus - auriga . from the disk mass derived by these model
fits , i computed a single component dust temperature for each star that reproduces the disk mass given the observed @xmath171.3 mm flux .
a histogram of these ` effective' dust temperatures is shown in figure [ fig : tdust ] .
( four stars with effective temperatures greater than 100 k are not shown in this figure . ) of the 44 stars in taurus - auriga with model disk masses from @xcite and @xcite , the median dust temperature derived in this manner is @xmath320 k. if this dust temperature is then assumed for all of the taurus stars , the model - derived dust masses can be reproduced to within a factor of 2 for 60% of the stars , and within a factor of 3 for 70% of the stars .
( for sources with effective dust temperatures substantially greater than 20 k , the single component dust model will overestimate the disk mass since @xmath43 ) . assuming that the disk temperatures in the ic 348 cluster are not vastly different than in taurus , adopting a dust temperature of 20 k , while simplistic , should reproduce the disk masses within a factor of a few for the majority of the stars .
for these assumptions , the 3@xmath5 upper limit to the circumstellar disk mass around an individual star in the ic 348 cluster is 0.025m@xmath4 . treating the 95 cluster members as an ensemble
, the observed average flux corresponds to an average disk mass of @xmath10m@xmath4 .
the limits on the disk masses in the ic 348 cluster can be compared to disk masses in other star forming regions as a step toward determining the evolution of disk masses as a function of time and environment ( e.g. `` clustered '' vs. `` isolated '' star formation ) .
the taurus star forming region provides a meaningful comparison since it has been thoroughly studied in the millimeter continuum ( beckwith et al .
1990 ; osterloh & beckwith 1995 ; see also skinner , brown , & walter 1991 and henning et al .
1998 ) , and the stellar population has an age similar to ic 348 as shown below . since the ic 348 sample has been identified largely based on spectroscopy , only stars in taurus with spectral types and sufficient photometric information to place the stars on the hr diagram were used for this comparison . from the database of spectral types and photometry compiled from the literature by @xcite ,
117 stars were identified in taurus that met these criteria and have a published @xmath171.3 mm continuum flux measurement .
the main source of incompleteness in the taurus sample are deeply embedded iras sources which are not detected at optical wavelengths @xcite , and recently identified low mass stars and brown dwarfs @xcite that have not yet been surveyed in the millimeter continuum .
the open histograms in the top panels of figures [ fig : mass ] and [ fig : age ] show the mass and age distributions respectively for the taurus comparison sample .
( the bottom panels show the analogous distributions for the ic 348 sample as already discussed . )
the mean stellar mass of the taurus sample is log@xmath44 ( mass / m@xmath4 ) = @xmath45 with a dispersion of 0.27 . by comparison , the ic 348 sample
has a mean log@xmath44 stellar mass of @xmath46 with a dispersion of 0.44 .
therefore , the stellar mass distribution in ic 348 is skewed toward smaller stellar masses compared to the taurus sample . similarly , the mean stellar age of the taurus sample is log@xmath44 ( age / yr ) = 6.0 with dispersion of 0.6 , while the mean log@xmath44 age and dispersion for ic 348 are 6.3 and 0.6 respectively .
thus the stellar ages for ic 348 are skewed toward older values than stars in taurus .
however , the uncertainty in the distances to the star forming regions , which is at least @xmath310% , @xcite , will produce a @xmath30.09 dex uncertainty in the mean age @xcite .
at best then the difference in the mean ages between taurus and ic 348 are significant at the @xmath32@xmath5 confidence level .
disk masses for stars in taurus were re - computed from published @xmath171.3 mm fluxes and using the isothermal dust model adopted here for consistency .
the hatched histograms in figures [ fig : mass ] and [ fig : age ] show the distribution of stellar masses and ages for stars in taurus that have disk masses in excess of 0.025m@xmath4 , which is the 3@xmath5 upper limit to the disk masses toward individual stars in ic 348 .
figure [ fig : mass ] shows that in this taurus sample , such disks masses are found only around stars more massive than 0.27m@xmath4 . while the lack of such massive disks around lower mass stars may simply be due to the relative lack of sources , only stars more massive than 0.27m@xmath4 are used in the following comparison between taurus and ic 348 . in the taurus sample , 14 of the 99 stars with stellar masses greater than 0.27m@xmath4
have disk masses in excess of 0.025m@xmath4 . in ic 348 ,
none of the 48 stars in the same stellar mass range have such disk masses .
using the two - tailed fisher exact test , the probability that the frequency distribution of massive disks in ic 348 and taurus have been drawn from the same parent population is 0.0028 .
this probability is independent of the assumed values of @xmath42 and @xmath47 used to compute the disk masses to the extent that the disk properties in the two regions are similar . however ,
if the value of @xmath37 is as low as 0.5 on average @xcite as opposed to the assumed value of @xmath37=1 , then the disk masses in ic 348 will be systematically overestimated compared to taurus , and the difference in the frequency of massive disks will be enhanced .
thus despite the similar stellar ages of the star forming regions , the ic 348 cluster lacks the massive disks found in the taurus at the @xmath33@xmath5 confidence level .
the cause of this apparent difference can not be due to an photoevaporation of the disks in ic 348 from a high ultraviolet radiation field ( e.g. johnstone , hollenbach , & bally 1998 ) since the spectral type of the most massive cluster member in ic 348 is b5 @xcite .
whether or not the difference in disk masses can be attributed to the affects of a `` cluster '' vs. `` isolated '' environment remains speculative given the limited sample of star forming regions , although continuum surveys of additional clusters are possible with current interferometers .
the ic 348 results can also be compared to the disk masses in orion nebula cluster using the 86 ghz continuum survey by @xcite . again using a 20 k dust temperature for consistency with the assumptions adopted here , the 3@xmath5 upper limit to individual disk masses in orion
is 0.17m@xmath4 . @xcite
derived a 95% upper limit to the characteristic flux at 86 ghz of 1.1mjy , which corresponds to a disk mass of 0.03m@xmath4 .
thus the results from @xcite exclude the presence of massive disks in orion , but do not set as stringent limits as the ic 348 observations for disk masses in cluster environments .
similarly , @xcite derived a 3@xmath5 upper limit of 0.015m@xmath4 to the disk mass around 5 proplyds in the orion nebula cluster for an assumed dust temperature of 50 k. if a 20 k dust temperature is instead adopted , the corresponding upper limit is 0.047m@xmath4 . finally , the ic 348 results can be contrasted with the minimum disk mass needed to form the solar system . summing the mass contained in the planets and assuming an interstellar gas to dust ratio , the minimum mass of the primitive solar nebula is @xmath30.01m@xmath4 @xcite .
this minimum mass reflects primarily the mass needed to form a jupiter - mass planet .
the actual mass of the primitive solar nebula may of course be higher depending on the efficiency in converting dust and gas in the primitive solar nebula into planets .
the upper limit to the typical disk mass in ic 348 is comparable then to the minimum mass of the solar nebula , and the average mass is @xmath35 times lower than minimum mass of the solar nebula . given that the disk masses are uncertain by at least a factor of 3 due the mass opacity coefficient ( @xmath42 ) , these results suggest either that the formation of planets with masses greater than a few jupiter masses is relatively rare around 0.2 - 0.5m@xmath4 stars in ic 348 , or that the formation of gas - giant planets has significantly depleted the disk of small dust grains on a time scale of @xmath152 myr , the mean age of the ic 348 cluster .
i present ovro observations of a @xmath0 region toward the ic 348 stellar cluster that has been imaged in the 98 ghz continuum to a rms noise level of 0.75mjybeam@xmath2 at @xmath1 resolution .
a total of 95 known members of the ic 348 cluster brighter than @xmath28=14.5 are located within the mosaicked region .
the stellar masses range from 0.03m@xmath4 to 4.2m@xmath4 with a peak in the distribution between 0.2m@xmath4 and 0.5m@xmath4 .
at least 15% of these stars are surrounded by an optically thick , inner accretion disk as evidenced by a @xmath31 excess ( see haisch , lada , & lada 2001a ) . the ovro observations are used to place constraints on the circumstellar disk masses and provide a snapshot of disk evolution in a cluster environment at an age of @xmath32 myr @xcite .
none of the 95 ic 348 members within the ovro mosaic were detected at the 3@xmath5 level or greater .
the mean flux toward the 95 stars is @xmath6mjy , indicating the observed fluxes are consistent with random noise although with a slight bias toward positive fluxes .
the millimeter - wave continuum fluxes were converted into disk masses ( the sum of the gas and dust components ) assuming a dust temperature of 20 k , a mass opacity coefficient of 0.02@xmath8 g@xmath2 at 1300 , and a power law index for the particle emissivity of @xmath37=1 .
for any individual source , the 3@xmath5 upper limit to the circumstellar disk mass is 0.025m@xmath4 .
for the 95 sources as an ensemble , the average disk mass is @xmath10m@xmath4 .
the constraints placed on the disk masses in the ic 348 cluster were compared to the disk masses around stars in the taurus molecular cloud .
a subsample of the taurus population was identified that has available optical spectroscopy and photometry as well as published millimeter continuum observations .
the stellar mass range was restricted to stars with masses in excess of 0.27m@xmath4 since that is the lowest mass star in the taurus sample that has an observed disk mass in excess of 0.025m@xmath4 . in the taurus sample , 14 of the 99 stars with stellar masses
@xmath480.27m@xmath4 have disk masses in excess of 0.025m@xmath4 , while in ic 348 , none of the 48 stars in the same stellar mass range have such disk masses . the probability that the frequency distribution of massive disks in ic 348 and taurus have been drawn from the same parent population
is 0.0028 , and are thus different at the @xmath33@xmath5 confidence level .
these results suggest that the ic 348 stellar population lacks the massive disks that characterizes @xmath314% of the taurus population despite the fact that the two regions have similar ages @xcite .
whether or not the differences between the frequency of massive disks between taurus and ic 348 is related to `` isolated '' versus `` clustered '' star forming environments awaits observations of additional star forming regions .
the implications of the ic 348 observations for planet formation can be placed in context with the minimum mass of the primitive solar nebula ( @xmath30.01m@xmath4 ; weidenschilling 1977 , hayashi 1981 ) needed to form our solar system .
the ic 348 results suggest that if an optically thick accretion disk is the necessary seed to form gas - giant planets , then either few stars in ic 348 with stellar masses between 0.2m@xmath4 and 0.5m@xmath4 will form planets more massive than a few jupiter masses , or the process to form such massive planets has significantly depleted the disk of small dust grains by the age of the ic 348 cluster , or @xmath32 myr .
jmc acknowledges support from long term space astrophysics grant nag5 - 8217 and the owens valley radio observatory , which is supported by the national science foundation through grant ast-9981546 .
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 .
2mass science data and information services were provided by the infrared science archive ( irsa ) at ipac .
adams , f. c. , shu , f. h , & lada , c. j. 1988 , , 326 , 865 andr , p. & montmerle , t. 1994 , , 420 , 837 bally , j. , testi , l. , sargent , a. , & carlstrom , j. 1998 , , 116 , 854 beckwith , s. v. w. , sargent , a. i. , chini , r. s. , & gsten , r. 1990 , , 99 , 924 beckwith , s. v. w. , & sargent , a. i. 1991 , , 381 , 250 bessell , m. s. , & brett , j. m. 1988 , , 100 , 1134 boss , a. p. 2001
, , 563 , 367 briceo , c. , calvet , n. , kenyon , s. , & hartmann , l. 1999 , , 118 , 1354 briceo , c. , hartmann , l. , stauffer , j. , & martn , e. 1998 , , 115 , 2074 calvet , n. , patio , a. , magris , g. c. , & dalessio , p. 1991
, , 380 , 617 calvet , n. , magris , g. c. , patio , a. , & dalessio , p. 1992 , rev .
astrofis . , 24 , 27 carpenter , j. m. 2000 , , 120 , 3139 2001 , aj , 121 , 2851 chiang , e.,i . , & goldreich , p. 1997
, , 490 , 368 cohen , j. g. , grogel , j. a. , perrson , s. e. , & elias , j. h. 1981 , , 249 , 481 dantona , f. , & mazzitelli , i. 1997 , mem .
soc . astron .
italiana , 68 , 807 dantona , f. , & mazzitelli , i. 1998 , asp conf .
134 , `` brown dwarfs and extrasolar planets '' , eds .
r. rebolo , e. l. martin , and m. r. zapatero - osorio ( san franciso : asp ) , 442 de zeeuw , p. t. , hoogerwerf , r. , de bruijne , j. h. j. , brown , a. g. a. , & blaauw , a. 1999 , , 117 , 354 dutrey , a. , guilloteau , s. , duvert , g. , prato , l. , simon , m. , schuster , k. , & mnard , f. 1996 , , 309 , 493 haisch , k. e. jr . , lada , e. a. , & lada , c. j. 2001a , , 553 , l153 .
2001b , , 121 , 2065 hartmann , l. 2001 , , 121 , 1030 hayashi , c. 1981 , prog .
phys . supp .
1981 , 70 , 35 henning , t. , burkert , a. , launhardt , r. , leinert , c. , & stecklum , b. 1998 , , 336 , 565 henning , t. , pfau , w. , zinnecker , h. , & prusti , t. 1993 , , 276 , 129 herbig , g. h. 1998 , , 497 , 736 hildebrand , r. h. 1983 , , 24 , 267 hillenbrand , l. a. , meyer , m. r. , & carpenter , j. m. 2002 , in preparation kenyon , s. , j. , dobrzycka , d. , & hartmann , l. 1994 , , 108 , 1872 kenyon , s. , j. , & hartmann , l. 1995 , , 101 , 117 johnstone , d. , hollenbach , d. , & bally , j 1998 , , 499 , 758 lada , e. a. , depoy , d. l. , evans , n. j. ii , & gatley , i. 1991 , , 371 , 171 lada , e. a. , & lada , c. j. 1995 , , 109 , 1682 luhman , k. l. 1999 , , 525 , 466 luhman , k. l. , & rieke , g. h. 1998 , , 497 , 354 luhman , k. l. , rieke , g. h. , lada , c. j. , & lada , e. a. 1998 , , 508 , 347 lynden - bell , d. , & pringle , j. e. 1974 , , 168 , 603 mannings , v. , & emerson , j. p. , 267 , 361 marcy , g. w. , cochran , w. d. , & mayor , m. 2000 , protostars and planets iv , eds . v. mannings , a. p. boss , and s. s. russell , ( tucson : university of arizona press ) , 1285 motte , f. , & andr , p. 2001
, , 365 , 440 motte , f. , andr , p. , & neri , r. 1998 , , 336 , 150 mundy , l. g. , looney , l. w. , & lada , e. a. 1995 , , 452 , l137 nrnberger , d. , brandner , w. , yorke , h. w. , & zinnecker , h. 1998 , , 330 , 549 nrnberger , d. , chini , r. , & zinnecker , h. 1997 , , 324 , 1036 najita , j. r. , tiede , g. p. , & carr , j. s. 2000 , , 541 , 977 odell , c. r. , & wong , s. k. 1996 , , 111 , 846 osterloh , m. , & beckwith , s. v. w. 1995 , , 439 , 288 padgett , d. l. , brandner , w. , stapelfeldt , k. r. , strom , s. e. , terebey , s. , & koerner , d. 1999 , , 117 , 1490 pollack , j. b. , hubickyj , o. , bodenheimer , p. , lissauer , j. j. , podolak , m. , & greenzweig , y. 1996 , icarus , 124 , 62 preibisch , t. , & zinnecker , h. 2001 , , 122 , 866 scally , a. , & clarke , c. 2001 , , 325 , 449 scoville , n. z. , carlstrom , j. e. , chandler , c. j. , phillips , j. a. , scott , s. l. , tilanus , r. p. j. , & wang , z. 1993 , , 105 , 1482 skinner , s. l. , brown , a. , & walter , f. m. 1991 , , 102 , 1742 skrutskie , m. f. , dutkevitch , d. , strom , s. e. , & strom , k. m. 1990 , , 99 , 1187 strzer , h. & hollenbach , d. 1999 , , 515 , 669 strom , k. e. , strom , s. e. , edwards , s. , cabrit , s. , & skrutskie , m. f. 1989 , , 97 , 1451 testi , l. , & sargent , a. i. 1998 , , 116 , 854 weidenschilling , s. j. 1977 , ap&ss , 51 , 153 wilking , b. a. , lada , c. j. , & young , e. t. 1989 , , 340 , 823 wood , k. , lada , c. j. , bjorkman , j. e. , kenyon , s. j. , whitney , b. , & wolff , m. j. 2002 , , in press | a @xmath0 region toward the young cluster ic 348 has been imaged in the millimeter continuum at @xmath1 resolution with the ovro interferometer to a rms noise level of 0.75mjybeam@xmath2 at 98 ghz .
the data are used to constrain the circumstellar disk masses in a cluster environment at an age of @xmath32 myr .
the mosaic encompasses 95 known members of the ic 348 cluster with a stellar mass distribution that peaks at 0.2 - 0.5m@xmath4 .
none of the stars are detected in the millimeter continuum at an intensity level of 3@xmath5 or greater .
the mean observed flux for the ensemble of 95 stars is @xmath6mjy . assuming a dust temperature of 20 k , a mass opacity coefficient of @xmath7 0.02@xmath8 g@xmath2 at 1300 , and a power law index of @xmath9 for the particle emissivity , these observations imply that the 3@xmath5 upper limit to the disk mass around any individual star is 0.025m@xmath4 , and that the average disk mass is @xmath10m@xmath4 .
the absence of disks with masses in excess of 0.025m@xmath4 in ic 348 is different at the @xmath33@xmath5 confidence level from taurus , where @xmath314% of the stars in an optically selected sample have such disk masses .
compared with the minimum mass needed to form the planets in our solar system ( @xmath30.01m@xmath4 ) , the lack of massive disks and the low mean disk mass in ic 348 suggest either that planets more massive than a few jupiter masses will form infrequently around 0.2 - 0.5m@xmath4 stars in ic 348 , or that the process to form such planets has significantly depleted the disk of small dust grains on time scales less than the cluster age of @xmath32 myr . |
in this problem , we attempt to reconstruct the _ conductivity _ @xmath1 in a steady state heat equation of the cooling fin on a cpu .
the heat is dissipated both by conduction along the fin and by convection with the air , which gives rise to our equation ( with @xmath2 for convection , @xmath1 for conductivity , @xmath3 for thickness and @xmath4 for temperature ) :
@xmath5 the cpu is connected to the cooling fin along the bottom half of the left edge of the fin .
we use the robin boundary conditions ( detailed in @xcite ) : @xmath6 our data in this problem is the set of boundary points of the solution to ( [ eq : heatpde ] ) , which we compute using a standard finite difference scheme for an @xmath7 mesh ( here @xmath8 or @xmath9 ) .
we denote the correct value of @xmath1 by @xmath10 and the data by @xmath11 . in order to reconstruct @xmath10
, we will take a guess @xmath12 , solve the forward problem using @xmath12 and compare those boundary points to @xmath11 by implementing the metropolis - hastings markov chain monte carlo algorithm ( or mhmcmc ) .
priors will need to be established to aid in the reconstruction , as comparing the boundary points alone is insufficient .
markov chains produce a probability distribution of possible solutions ( in this case conductivities ) that are most likely given the observed data ( the probablility of reaching the next step in the chain is entirely determined by the current step ) .
the algorithm is as follows ( @xcite ) .
given @xmath13 , @xmath14 can be found using the following : 1 .
generate a candidate state @xmath12 from @xmath13 with some distribution @xmath15 .
we can pick any @xmath15 so long as it satisfies 1 .
@xmath16 2 .
@xmath15 is the transition matrix of markov chain on the state space containing @xmath17 .
2 . with probability @xmath18 set @xmath19 , otherwise set @xmath20 ( ie . accept or reject @xmath12 ) .
proceed to the next iteration .
using the probability distributions of our example , ( [ eq : alpha ] ) becomes @xmath21 } \right\}\ ] ] ( where @xmath22 and @xmath23 denote the set of measured boundary points using @xmath12 and @xmath13 respectively , and @xmath24 ) + to simplify ( [ eq : alpha2 ] ) , collect the constants and seperate the terms relating to @xmath12 and @xmath13 : @xmath25}\\ & = & \frac{-1}{2}\sum_{i , j=1}^{n , m}{\left [ \left ( \frac{d_{ij}-d_{ij}'}{\sigma } \right)^2 - \left ( \frac{d_{ij}-d_{n_{ij}}}{\sigma } \right)^2 \right]}\\ & = & \frac{-1}{2}\left [ d ' - d_n \right ] = f_n - f'\end{aligned}\ ] ] now , ( [ eq : alpha2 ] ) reads @xmath26 we now examine the means by which we generate a guess @xmath12 . if the problem consists of reconstructing a constant conductivity , we can implement a uniform change , for every iteration we take a random number @xmath27 between @xmath28 and @xmath29 and add it to every entry in @xmath13 to obtain @xmath12 ( we initialize @xmath30 to a matrix of @xmath31s ) .
the algorithm is highly efficient , and the reconstructed value will consistently converge to that of the solution to within @xmath27 .
+ in order to approximate a nonconstant @xmath10 , the obvious choice is a pointwise change , at each iteration we add @xmath27 to a random entry of @xmath13 , thus genrating @xmath12 .
unfortunately , systematic errors occur at the boundary points of our reconstruction ( they tend to rarely change from their initial position ) . + in order to sidestep this , we use a gridwise change ; change a square of the mesh ( chosen at random as well ) by adding @xmath27 to the four corners of said square . while this fixes the boundary problem , another major issue which arises from a non - uniform change is that the reconstruction will be marred with spikes " , which we must iron out .
to aid in ironing out the wrinkles in the reconstruction we use priors " .
priors generally require some knowledge of the quantity we wish to find , and will add a term to ( [ eq : alpha3 ] ) . naturally , the more unassuming the prior , the more applicable the algorithm .
this applicability will be tested as often as possible throughout these tests .
the first prior compares the sum of the differences between adjacent points of @xmath10 to those of @xmath12 ( keping the spikes in check ) , and is given by @xmath32 @xmath33 and modifying ( [ eq : alpha3 ] ) , we obtain @xmath34 so the guess is most likely if @xmath13 and @xmath12 are similarily smooth ( ie , @xmath35 ) , if an iteration gives a @xmath12 that is noticeably less smooth than the last accepted iteration , we are less likely to accept it .
+ as an initial test for the smoothness prior developed above , we attempt the gridwise change on a constant conductivity ( @xmath36 , using @xmath37 ) . while we can still see the problem at the boundary points , they are limited to being a noticeable nuisance as oppposed to adamantly ruining an otherwise accurate reconstruction ( whose mean comes to within @xmath38 of @xmath39 ) .
the next step is therefore to test the algorithm on a non - constant conductivity . as a simple nonconstant trial
, we look at a tilted plane with constant slope , given by @xmath40 once again , we take @xmath30 to be a matrix of all @xmath31s and @xmath37 .
the boundary points again have trouble increasing from @xmath31 to the desired values , and in so doing lower the mean value of the reconstruction ; though we still consistentily get to within about @xmath41 of the solution ( in @xmath42 iterations )
. tilted plane we wish to reconstruct , and a reconstruction using the smoothness prior.,height=188 ] we now attempt to reconstruct a more complicated conductivity : a gaussian well .
+ the gaussian well is the first real challenge that the algorithm will face , and will be the main focus of the rest of the paper as it contains different regions which require different priors .
it is given by the following equation @xmath43}{0.2}}}\right)}}\ ] ] this conductivity represents a much more significant challenge , with both flat regions , and regions with steep slopes .
after several trials , the optimal @xmath44s were found to be between @xmath31 and @xmath45 , though obataining a specific value for which the reconstruction is best is impossible due to the high innaccuracy of the algorithm when faced with this well .
gaussian well we wish to reconstruct , and a reconstruction using the smoothness prior.,height=188 ] there is an evident need , at this point , for much more precision .
we turn once again to priors , this time developing one that will look at the slopes of the reconstruction .
one of the main concerns in implementing a new prior is the generality mentioned earlier . in theory , one could use a prior that only accepts gaussian wells of the form we have here , but that code would not be very versatile .
we therefore try to keep our slope prior as general as possible . in keeping with this , we look at the ratios of adjacent slopes , both in the @xmath46 and @xmath47 directions , as follows : @xmath48 @xmath49 and define @xmath50 @xmath51 ( where @xmath52 is @xmath53 ) + the generality of these prior terms comes from the fact that they go to @xmath54 so long as the conductivity does nt change its mind .
it is equally happy " with a constant slope as it is with slopes that , say , double , at each grid point . it should be noted that the formulas above break down for regions where we have very small slopes adjacent to large ones , where one ratio goes to @xmath54 while the other grows very large .
nevertheless , we now set @xmath55 with this new prior , we define @xmath56 and use that in the acceptance step of the mhmcmc algorithm . again , as a first test of the algorithm , we test it on the tilted plane .
the reconstructions reach the same precision in @xmath42 iterations as we had with only the smoothness prior , so we have not yet implemented anything that is too problem - specific to the gaussian well .
+ the initial result of the test on the well is arguably substantially better , but still rather imprecise . in an attempt to see more clearly
, we make the mesh finer ( @xmath57 ) . in addition , we set @xmath30 to be a matrix of all @xmath0s
. the results of the combined slope and smoothness priors are below . and @xmath58 gaussian wells with parameters : @xmath59 and @xmath60 , respectively.,height=188 ] as we can see
, a substantial improvement has been made over the attempt in section ( 3.1 ) .
we now consistently obtain somewhat of a bowl shape . in comparing the solution we wish to achieve and the reconstruction we have ,
one notices that the major problem areas are the outer regions , where the conductivity is nearly constant .
as previously stated , equations ( [ eq : px ] ) and ( [ eq : py ] ) break down when the slopes are vanishing , so it is reasonable to assume that with this alone , the reconstruction will not improve as substantially as we need it to . as before ,
we implement another prior to aid us .
to help reconstruct the outermost regions of the well , we need a prior that will go to @xmath54 for regions that have vanishing slope .
the most obvious choice is therefore to use what we computed for the smoothness prior @xmath32 and set @xmath61 using @xmath62 in the mhmcmc algorithm .
again , the worry that adding a new prior would undermine the generality of the algorithm can be eased by noting that we are simply accounting for a problematic case not treated by the slope prior , though we still test this prior on the tilted plane .
the @xmath58 tilted plane is given by @xmath63 running the mhmcmc algorithm with all three priors yields fairly accurate reconstructions , that miss the solution by @xmath64 .
one should again note the presence of the familiar ( though no less troublesome ) boundary points .
tilted plane , and a reconstruction using all three priors.,height=188 ] we now try once again to reconstruct the gaussian well .
the results of the added prior are apparent , and regions with vanishing slope are treated much more accurately than before .
perhaps the most successful reconstruction thus far is the following ( though many more possible combinations of @xmath44 , @xmath65 and w must be explored ) . gaussian well using all three priors , at @xmath45 and @xmath66 million interations.,height=188 ] an obvious flaw in these reconstructions happens to be the width of the well , the algorithm is still capable of reconstructing the center of the well , and its depth , but it is often much narrower than in the actual solution .
it would seem the algorithm has trouble starting to drop off from the vanishing slope region into the varying one .
this exposes an inherent problem with the patchwork we have taken thus far : getting the seams to match up nicely .
as we have seen , reconstructions of the heat conductivity greatly benefit from added priors .
there is certainly much work left to be done , and a very careful analysis of the seams at which the various priors trade off is in order .
however , we believe that in testing the algorithm against other complex nonconstant conductivities , which is the next step we plan to take , it is possible to complete the aforementioned analysis of the seams and reconstruct complex quantities via this patchwork method .
i was introduced to this problem at a national science foundation reu program at george mason university , and would like to thank both of those institutions for the oppurtunity that gave me .
i would also like to thank professors timothy sauer and harbir lamba at gmu , who got me started on this project while i was there and helped me decipher the mhmcmc algorithm . | we consider the nonlinear inverse problem of reconstructing the heat conductivity of a cooling fin , modeled by a @xmath0-dimensional steady - state equation with robin boundary conditions .
the metropolis hastings markov chain monte carlo algorithm is studied and implemented , as well as the notion of priors . by analyzing the results using certain trial conductivities
, we formulate several distinct priors to aid in obtaining the solution .
these priors are associated with different identifiable parts of the reconstruction , such as areas with vanishing , constant , or varying slopes .
although more research is required for some non - constant conductivities , we believe that using several priors simultaneously could help in solving the problem .
inverse problems , heat diffusion , monte carlo , prior .
a. zambelli [email protected] |
the preceding paper @xcite studied the class of stochastic maps which break entanglement . for a given map
@xmath1 this means that @xmath2 is separable for any density matrix @xmath3 on a tensor product space .
it was observed that a map is entanglement breaking if and only if it can be written in either of the following equivalent forms [ eq : holv ] ( ) & = & _ k r_k f_k + [ eq : rank1 ] & = & _ k |_k _ k| _ k , _ k where each @xmath4 is a density matrix and @xmath5 a positive semi - definite operator . the map @xmath1 is also trace - preserving if and only if @xmath6 , in which case the set @xmath7 form a povm .
henceforth we will only consider trace - preserving maps and use the abbreviations cpt for those which are also completely positive and ebt for those which are also entanglement breaking .
an ebt map is called classical - quantum ( cq ) if each @xmath8 is a one - dimensional projection ; it is quantum - classical ( qc ) if each density matrix @xmath9 is a one - dimensional projection .
maps which break entanglement can always be simulated using a classical channel ; thus , one is primarily interested in those which preserve entanglement .
nevertheless , it is important to understand the distinction . in this paper
we restrict attention to ebt maps on qubits , for which one can obtain a number of results which do not hold for general ebt maps .
the main new result , which does not hold in higher dimensions , is that every qubit ebt map can be written as a convex combination of maps in the subclass of cq maps defined above . before proving this result in section [
sect : convcq ] , we review parameterizations and complete positivity conditions for qubit maps .
we also give a number of more specialized results which use the canonical parameterization and/or the fact that positivity of the partial transpose suffices to test entanglement for states on pairs of qubits .
recall that any cpt map @xmath1 on qubits can be represented by a matrix in the canonical basis of @xmath10 .
when @xmath11 $ ] , then @xmath12 $ ] where @xmath13 is the vector with elements @xmath14 and @xmath15 is a @xmath16 matrix , i.e. , @xmath17 .
moreover , it was shown in @xcite that we can assume without loss of generality ( i.e. , after suitable change of bases ) that @xmath15 is diagonal so that @xmath18 has the canonical form [ eq : tcanon ] = .
the conditions for complete positivity in this representation were obtained in @xcite and are summarized in section [ sect : cp ] . in the case of qubits , theorem 4 of @xcite can be extended to give several other equivalent characterizations .
[ thm : tfaeq ] for trace - preserving qubit maps , the following are equivalent * @xmath1 has the holevo form ( [ eq : holv ] ) with @xmath19 a povm .
* @xmath1 is entanglement breaking .
* @xmath20 is completely positive , where @xmath21 is the transpose .
* @xmath1 has the `` sign - change '' property that changing any @xmath22 in the canonical form ( [ eq : tcanon ] ) yields another completely positive map .
* @xmath1 is in the convex hull of cq maps .
conditions ( c ) thru ( e ) are special to qubits . conditions ( c ) and ( d ) use the fact @xcite that the ppt ( positive partial transpose ) condition for separability is also sufficient in the case of qubits .
in this section , we prove theorem [ thm : tfaeq ] and provide some results using the canonical parameters .
this gives another characterization of qubit ebt maps in the special case of cpt maps which are also unital .
the equivalence @xmath23 was proved in @xcite where it was also shown that both are equivalent to the condition that @xmath24 is cpt for all @xmath25 in a set of entanglement witnesses and that @xmath26 is cpt if and only if @xmath24 is . in the case of qubits , it is well - known that it suffices to let @xmath25 be the transpose , which proves the equivalence with ( c )
. furthermore , changing @xmath27 is equivalent to changing @xmath28 in the representation ( [ eq : tcanon ] ) , and is unitarily equivalent ( via conjugation with a pauli matrix ) to changing the sign of any other @xmath29 which yields @xmath30 .
that @xmath31 follows immediately from the facts that cq maps are a special type of entanglement - breaking maps and the set of entanglement - breaking maps is convex by theorem 2 of @xcite .
the proof that shows @xmath32 will be given in section [ sect : convcq ] . .
the proof that ( b ) @xmath33 ( a ) given in @xcite relied on the fact that there is a one - to - one correspondence @xcite ( but not a unitary equivalence ) between maps @xmath1 and states [ eq : choi ] _
= ( i ) ( | | ) where @xmath34 is one of the maximally entangled bell states .
moreover , a map is ebt if and only if @xmath35 is separable since that was shown to be equivalent to writing it in the form ( [ eq : rank1 ] ) .
one could then apply the reduction criterion for separability @xcite to @xmath35 .
this condition states that a necessary condition for separability of @xmath36 is that @xmath37 for all maximally entangled states . in the case of qubits ,
this criterion is equivalent to the ppt condition , and hence sufficient , and equivalent to @xmath38 , which gives the following result .
a qubit cpt map is ebt if and only if @xmath39 with @xmath35 as in _ ( [ eq : choi ] ) . _
we now consider entanglement breaking conditions which involve only the parameters @xmath29 .
[ thm : sumleq1 ] if @xmath1 is an entanglement breaking qubit map written in the form ( [ eq : tcanon ] ) , then @xmath40 .
it is shown in @xcite that a necessary condition for complete positivity is [ eq : cp.t0 ] ( _ 1 _ 2)^2 ( 1 _ 3)^2 . when combined with the sign change condition ( d ) , this yields the requirement @xmath41 . for unital qubit channels ,
the condition in theorem [ thm : sumleq1 ] is also sufficient for entanglement breaking . for unital maps @xmath42 and ,
as observed in @xcite , the conditions in ( [ eq : cp.t0 ] ) are also sufficient for complete positivity . since @xmath40 implies that ( [ eq : cp.t0 ] ) holds for any choice of sign in @xmath43 , it follows that any unital cpt map satisfying this condition is also ebt .
[ thm : unital.iff ] a unital qubit channel is entanglement breaking if and only if @xmath40 _ [ after reduction to the form ( [ eq : tcanon])]_. moreover , as will be discussed in section [ sect : geom ] the extreme points of the set of unital entanglement breaking maps are those for which two @xmath44 . hence these channels are in the convex hull of cq maps . for non - unital maps these conditions
need not be sufficient .
consider the so - called amplitude damping channel for which @xmath45 , and @xmath46 .
for this map equality holds in the necessary and sufficient conditions [ eq : cp.t3ne0 ] ( _ 1 _ 2)^2 ( 1 _ 3)^2 - t_3 ^ 2 . since the inequalities would be violated if the sign of one @xmath29 is changed , the amplitude damping maps are never entanglement breaking except for the limiting case @xmath47 .
thus there are maps for which @xmath48 can be made arbitrarily small ( by taking @xmath49 ) , but are _ not _ entanglement - breaking .
we begin by considering the representation of maps in the basis @xmath10 .
let @xmath1 have the form ( [ eq : holv ] ) and write @xmath50 $ ] and @xmath51 $ ] .
let @xmath52 be the @xmath53 matrices whose rows are @xmath54 and @xmath55 respectively , i.e. , @xmath56 . let @xmath57 be the matrix @xmath58 .
note that the requirement that @xmath59 is a povm is precisely that the first row of @xmath18 is @xmath60 .
the matrix @xmath61 is the representative of @xmath1 in the form ( [ eq : tcanon ] ) ( albeit not necessarily diagonal ) .
we can summarize this discussion in the following theorem .
[ thm.holv.canon ] a qubit channel is entanglement breaking if and only if it can be represented in the form _ ( [ eq : tcanon ] ) _ with @xmath61 where @xmath62 and @xmath63 are @xmath64 matrices as above , i.e. , the rows satisfy @xmath65 and @xmath66 for all @xmath67 .
we can use this representation to give alternate proofs of two results of the previous section . to show that @xmath68 observe that changing the sign of the j - th column of @xmath63 @xmath69 is equivalent to replacing @xmath70 by the povm with @xmath71 .
the effect on @xmath18 is simply to multiply the @xmath72-th column by @xmath73 .
the critical property about qubits is that the condition @xmath74 is equivalent to @xmath75 which is unaffected by the replacement @xmath71 .
next , we give an alternate proof of theorem [ thm : sumleq1 ] which is of interest because it may be extendable to higher dimensions .
let @xmath52 be as in section [ sect : qreps ] . then _
j = 1 ^ 3 |_j| & = & _ j = 1 ^ 3 | _ k = 1^n w_j^k u_j^k
| + & & _ j = 1 ^ 3 _ k = 1^n |w_j^k u_j^k | = _ k = 1^n _ j = 1 ^ 3 |w_j^k u_j^k & & _ k = 1^n ( _ j = 1 ^ 3 |w_j^k|^2 ) ^1/2 ( _ j = 1 ^ 3 |u_j^k|^2 ) ^1/2 + & & _ k = 1^n 1 u_0^k = 1 where we have used the fact that @xmath76 and @xmath77 . that @xmath78 is a consequence of the fact that the @xmath7 form a povm .
we now consider the decomposition @xmath79 for the special cases of cq , qc and point channels .
if @xmath1 is a cq channel , we can assume without loss of generality that @xmath80 .
now write @xmath81 . then = w^t u = . by acting on the left with a unitary matrix of the form @xmath82 where @xmath83 is a rotation
whose third row is a multiple of @xmath84 , this can be reduced to the form ( [ eq : tcanon ] ) with @xmath85 , @xmath86 , and @xmath87 [ since @xmath88 .
] indeed , it suffices to choose [ eq : cqmatw ] w = .
note that the requirement @xmath89 only implies @xmath90 ; however , the requirement @xmath76 implies that @xmath91 must hold with both signs and this is equivalent to the stronger condition [ eq : nasc.lin ] t_1 ^ 2 + t_2 ^ 2 + ( |t_3| + |_3|)^2 1 which is necessary and sufficient for a cpt map to reduce the bloch sphere to a line .
if @xmath1 is a qc channel , we can assume without loss of generality that @xmath92 and @xmath93 , from which one easily finds that the second and third rows of @xmath61 are identically zero and the fourth row is @xmath94 .
one then easily verifies that multiplication on the right by a matrix as above with @xmath83 a rotation whose third column is a multiple of @xmath95 reduces @xmath61 to the canonical form ( [ eq : tcanon ] ) with @xmath85 , @xmath96 , and @xmath97 .
( note that @xmath98 with equality if and only if the image reaches the bloch sphere . )
it is interesting to note that for qubits channels , every qc channel is unitarily equivalent to a cq channel .
indeed , a channel which , after reduction to canonical form has non - zero elements @xmath99 and @xmath100 with @xmath101 and @xmath102 can be written as either a qc channel with w = u = or as a cq channel with w = u = . for point channels
@xmath103 and @xmath104 .
we conclude this section with an example of map of the form ( [ eq : holv ] ) with an extreme povm , for which the corresponding map @xmath1 is _ not _ extreme .
let @xmath105 $ ] with @xmath106 .
then , irrespective of the choice of @xmath4 , the third column of @xmath61 is identically zero , which implies that , after reduction to canonical form , one of the parameters @xmath44 .
however , it is easy to find density matrices , e.g. , @xmath107 $ ] , for which the resulting map @xmath1 is not cq or point .
but by theorem [ thm : cqqubit ] , @xmath1 is a convex combination of cq maps and hence , not extreme .
not only is the set of cpt maps convex , in a fixed basis corresponding to the canonical form ( [ eq : tcanon ] ) the set of @xmath29 corresponding to any fixed choice of @xmath108 is also a convex set which we denote @xmath109
. we will also be interested in the convex subset @xmath110 of the @xmath111-@xmath112 plane for fixed @xmath113 , and in the convex set @xmath114 of points @xmath115 corresponding to fixed @xmath116 . although stated somewhat differently , the following result was proved in @xcite .
[ thm : rsw ] let @xmath13 and @xmath99 be fixed with @xmath117 . then the convex set @xmath110 consists of the points @xmath118 for which @xmath119 _ ( or , equivalently @xmath120 ) _ is positive semi - definite , where [ eq : cont.mtrx ] r _ = .
similarly , @xmath114 also consists of the points @xmath115 for which @xmath121 .
moreover , the extreme points of @xmath122 are those for which @xmath123 . although this result is stated in a form in which @xmath100 and @xmath99 play a special role and does not appear to be symmetric with respect to interchange of indices , the conditions which result are , in fact , invariant under permutations of @xmath124 .
theorem [ thm : rsw ] follows from choi s theorem @xcite that @xmath1 is completely positive if and only if @xmath35 , given by ( [ eq : choi ] ) , is positive semi - definite .
as noted in @xcite , this implies that it can be written in the form _ = where @xmath125 is a contraction .
( note , however , that the expression for @xmath125 given in ( [ eq : cont.mtrx ] ) was obtained by applying this result to the adjoint @xmath126 , i.e , to @xmath127 .
conversely , given a cpt map @xmath1 and _ any _ contraction @xmath63 on @xmath128 , one can define a @xmath129 matrix in block form , [ eq : genu ] m = it then follows that there is another cpt map which ( with a slight abuse of notation ) we denote @xmath130 for which @xmath131 .
however , ( [ eq : genu ] ) need not , in general , correspond to a map @xmath130 which has the canonical form ( [ eq : tcanon ] ) since that requires @xmath132 . for @xmath63 an arbitrary unitary or contraction , we can only conclude that ( _ x ) & = u + u^ & _ k=0 ^ 3 t_1k _ k + ( _ y ) & = u - u^ & _ k=0 ^ 3 t_2k _ k so that the map @xmath130 corresponds to a matrix of the form @xmath133 with @xmath134 real . in order to study the general case of non - zero @xmath135 , it is convenient to rewrite ( [ eq : cont.mtrx ] ) in the following form ( using notation similar to that introduced in @xcite ) .
r _ = where @xmath136 , @xmath137 , and @xmath138 , e.g. , @xmath139 .
then i - r_^ r _ m = with m_11 & = & 1 - - [ eq : diag1 ] + & & + m_22 & = & 1 - - [ eq : diag2 ] + & & + m_12 = _ 21 & = & + [ eq : offdiag ] note that the denominators , although somewhat messy , are essentially constants depending only on @xmath100 and @xmath99 . considering @xmath140 as also a fixed constant it suffices to rotate ( and dilate ) the @xmath111-@xmath112 plane by @xmath141 and work instead with the variables @xmath142 .
the diagonal conditions @xmath143 and @xmath144 define a rectangle in the @xmath145-@xmath146 plane , namely |_-|^2 & & c_ c_-+ - ||^2 = ( 1- _ 3)^2 - t_3 ^ 2 - ||^2 [ eq : m11>0 ] + & & + [ eq : m22>0 ] |_+|^2 & & c_++ c_+- - ||^2 = ( 1 + _ 3)^2 - t_3 ^ 2 - ||^2 these diagonal conditions imply the necessary conditions [ eq : cp.tau0 ] |_|^2 ( 1 _ 3)^2 - t_3 ^ 2 for complete positivity , which also become sufficient when @xmath147 .
the determinant condition @xmath148 is more complicated , but basically has the form [ eq.detcond ] e _ + ^2 + f _
-^2 + g _ + _ - in particular , we would like to know if the values of @xmath149 satisfying ( [ eq.detcond ] ) necessarily lie within the rectangle defined by ( [ eq : m11>0 ] ) and ( [ eq : m22>0 ] ) . extending the lines bounding this rectangle ,
i.e. , @xmath150 and @xmath151 one sees that the @xmath145-@xmath146 plane is divided into 9 regions , as shown in figure [ fig : regions ] and described below . * the rectangle in the center which we denote @xmath152 , * four ( 4 ) outer corners which we denote @xmath153 since both @xmath154 and @xmath155 , * the four ( 4 ) remaining regions ( directly above , below and to the left and right of the center rectangle ) which we denote as @xmath156 or @xmath157 according to the signs of @xmath158 and @xmath159 .
we know that the determinant condition ( [ eq.detcond ] ) is _ never _ satisfied in the @xmath156 or @xmath157 regions since @xmath160 when @xmath158 and @xmath159 have opposite signs .
this implies that equality in ( [ eq.detcond ] ) defines a curve which bounds a convex region lying entirely within the @xmath152 rectangle .
although ( [ eq.detcond ] ) also has solutions in the @xmath153 regions as shown in figure [ fig : regions ] , one expects that these will typically lie _ outside _ the region for which @xmath161 , i.e. , the rectangle bounded by the line segments satisfying @xmath162 and @xmath163
. however , john cortese @xcite has shown that this need not necessarily be the case .
nevertheless , one need only check one of the two conditions @xmath164 , @xmath165 , and might substitute a weaker condition , such as @xmath166 , to exclude points in the @xmath153 regions .
for example , one could substitute for the diagonal conditions , @xmath167 which is equivalent to [ eq : diag.alt ] ( _ 1 ^ 2 + _ 2 ^ 2 ) ( 1 + t_3 ) + _ 3 ^ 2 ( 1 - t_3 ) ( 1 + t_3 ) ( 1 - ||)^2 + 2 _ 1 _ 2 _ 3 .
thus , strict inequality in both ( [ eq.detcond ] ) and ( [ eq : diag.alt ] ) suffice to ensure complete positivity . in general , when @xmath168 , the convex set @xmath169 is determined by ( [ eq.detcond ] ) , i.e , by the closed curve for which equality holds and its interior .
since changing the sign of @xmath111 or @xmath112 is equivalent to changing @xmath170 , the corresponding set of entanglement breaking maps is given by the intersection of this region with the corresponding one with @xmath145 and @xmath171 switched , as shown in figure [ fig : swap ] .
if , instead of looking at @xmath119 , we had considered @xmath172 , the matrix @xmath173 would change slightly and the conditions ( [ eq : m11>0 ] ) or ( [ eq : m22>0 ] ) would be modified accordingly .
( in fact , the only change would be to replace @xmath174 by @xmath175 in the fraction multiplying @xmath176 . )
however , the determinant condition ( [ eq.detcond ] ) would _ not _ change .
since @xmath177 and @xmath178 , are unitarily equivalent , @xmath179 = \det \big ( u[i - r_{\phi } r_{\phi}^{\dg } ] u^{\dg } \big ) = \det[i - r_{\phi } r_{\phi}^{\dg}].\ ] ] it is worth noting that whether or not @xmath180 is a contraction is _ not _ affected by the signs of the @xmath135 .
( in particular , changing @xmath181 takes @xmath182 , changing @xmath183 takes @xmath184 , and changing @xmath185 takes @xmath186 . ) therefore , one can change the sign of any _ one _ of the @xmath135 without affecting completely positivity .
by contrast , one can _ not _ , in general , change @xmath187 without affecting the complete positivity conditions .
( note , however , that one can always change the signs of any _ two _ of the @xmath29 since this is equivalent to conjugation with a pauli matrix on either the domain or range . the latter will also change the signs of two of the @xmath135 . ) changing
the sign of @xmath112 is equivalent to composing @xmath1 with the transpose , so that changing the sign of one of the @xmath29 is equivalent to composing @xmath1 with the transpose and conjugation with one of the pauli matrices .
furthermore , if changing the sign of one particular @xmath29 does not affect complete positivity , then one can change the sign of _ any _ of the @xmath29 without affecting complete positivity . in view of the role of the sign change condition it is worth summarizing these remarks .
let @xmath1 be a cpt map in canonical form ( [ eq : tcanon ] ) and let @xmath21 denote the transpose . then * @xmath188 is also completely positive , i.e. , changing @xmath189 does not affect complete positivity .
* @xmath190 is completely positive if and only if changing any @xmath187 does not affect complete positivity .
* @xmath190 is completely positive if and only @xmath191 is .
the only difference between @xmath190 and @xmath191 is that the former changes the sign of @xmath112 while the latter changes the signs of both @xmath192 and @xmath112 .
* image of the bloch sphere * we first consider the geometry of entanglement breaking channels in terms of their effect on the bloch sphere .
it follows from the equivalence with the sign change condition in theorem [ thm : tfaeq ] that any cpt map with some @xmath44 is entanglement breaking .
we call such channels _ planar _ since the image lies in a plane within the bloch sphere .
similarly , we call a channel with two @xmath44 _ linear_. if all three @xmath44 , the bloch sphere is mapped into a point . note that the subsets of channels whose images lie within points , lines , and planes respectively are _
not _ convex . however , they are well - defined and useful classes to consider .
* points : * a channel which maps the bloch sphere to a point has the holevo form ( [ eq : holv ] ) in which the sum reduces to a single term with @xmath193 $ ] and @xmath194 .
then @xmath195 and @xmath196 .
when @xmath197 , @xmath198 is a pure state and the map is extreme .
it is also a special case of the so - called amplitude damping channels , and ( as noted at the end of section [ sect : qubit.thms ] ) these are the only amplitude damping channels which break entanglement .
* lines : * when two of the @xmath44 so that the image of the bloch sphere is a line , the conditions for complete positivity reduce to a single inequality , which becomes ( [ eq : nasc.lin ] ) in the case @xmath199 . moreover , it is straightforward to verify that any such channel can be realized as a cq channel .
indeed , it suffices to choose @xmath62 as in ( [ eq : cqmatw ] ) . * planar channels : * the image of a map with exactly one @xmath44 lies in a plane .
when this is @xmath99 , the condition @xmath121 becomes 0 .
where @xmath200 , and the condition on the diagonal becomes [ eq : nasc.planar ] ( |_1| + | _ 2|)^2 + ||^2 1 .
now , if either diagonal element is identically zero , then one must have @xmath201 . thus , if both @xmath202 and equality holds in the necessary condition ( [ eq : nasc.planar ] ) , one must have @xmath203 , in which case it reduces to @xmath204 .
this implies that a truly planar channel can not touch the bloch sphere , unless it reduces to a point or a line .
* geometry of @xmath29 space * we now consider , instead of the geometry of the images of entanglement - breaking maps , the geometry of the allowed set of maps in @xmath29 space . after reduction to the canonical form ( [ eq : tcanon ] ) it is often useful to look at the subset of @xmath205 $ ] which correspond to a particular class of maps .
we first consider maps for which @xmath206 .
[ thm : octa ] in a fixed ( diagonal ) basis , the set of unital entanglement breaking maps on qubits corresponds to the octahedron whose extreme points correspond to the channels for which @xmath205 $ ] is a permutation of @xmath207 $ ] .
since this octahedron is precisely the subset with @xmath40 the result follows immediately from theorem [ thm : unital.iff ] .
alternatively , one could use theorem [ thm : cqqubit ] and the fact that the unital cq maps must have the form above
. * remarks : * 1 .
the channels corresponding to a permutation of @xmath207 $ ] belong to the subclass known as cq channels .
hence , the set of unital entanglement breaking maps is the convex hull of unital cq maps . 2 .
this octahedron in theorem [ thm : octa ] is precisely the intersection of the tetrahedron with corners @xmath208,~[1,-1,-1],~[-1,1,-1 ] , ~[-1,-1,1]$ ] with its inversion through the origin , as shown in figure [ fig : octa ] .
( a similar picture arises in studies of entanglement and bell inequalities .
see , e.g. , figure 3 in @xcite or fig . 2 in @xcite ) .
the tetrahedron of unital maps is precisely the intersection of the four planes of the form @xmath209 = 1 $ ] with @xmath210 $ ] and an odd number of negative signs , i.e. , @xmath211 .
the octahedron of unital ebt maps is precisely the intersection of all eight planes of this form . 4 .
if the octahedron of unital entanglement breaking maps is removed from the tetrahedron of unital maps , one is left with four disjoint tetrahedrons whose sides are half the length of the original .
each of these defines a region of entanglement - preserving " unital channels with fixed sign .
for example , the tetrahedron with corners , @xmath208,~[1,0,0],~[0,10 ] , ~[1,0,0]$ ] ; this is the interior of the intersection of the plane @xmath212 \cdot [ \lambda_1 , \lambda_2 , \lambda_3 ] = -1 $ ] and the three planes of the form @xmath209 = 1 $ ] with @xmath213,~ [ 1,-1,1 ] , [ -1,1,1]$ ] . for many purposes , e.g. , consideration of additivity questions , it suffices to confine attention to one of these four corner tetrahedrons .
indeed , conjugation with one of the pauli matrices , transforms the corner above into one of the other four .
we next consider non - unital maps , for which one finds the following analogue of theorem [ thm : octa ] .
[ thm : sign ] let @xmath214 be a fixed vector in @xmath215 and let @xmath216 denote the convex subset of @xmath215 corresponding to the vectors @xmath205 $ ] for which the canonical map with these parameters is completely positive .
then the intersection of @xmath216 with its inversion through the origin ( i.e. , @xmath217 ) is the subset of ebt maps with translation @xmath13 .
the effect of changing the sign of @xmath112 is @xmath218 and of changing the sign of @xmath111 is @xmath219 . in either case , the effect on the determinant condition ( [ eq.detcond ] ) is simply to switch @xmath218 , i.e , to reflect the boundary across the @xmath220 line .
thus , the intersection of these two regions will correspond to entanglement breaking channels .
the remainder will , typically , consist of 4 disjoint ( non - convex ) regions , corresponding to the four corners remaining after the `` rounded octahedron '' of theorem [ thm : sign ] is removed from the `` rounded tetrahedron '' .
in @xcite we found it useful to generalize the extreme points of the set of cpt maps @xmath221 to include all maps for which @xmath180 is unitary , which is equivalent to the statement that _ both _ singular values of @xmath180 are @xmath222 .
in addition to true extreme points , this includes `` quasi - extreme '' points which correspond to the edges of the tetrahedron of unital maps
. some of these quasi - extreme points are true extreme points for the set of entanglement - breaking maps
. however , there are no extreme points of the latter which are not generalized extreme points of @xmath221 .
this will allow us to conclude the following .
first we note that the following was shown in @xcite .
after reduction to canonical form ( [ eq : tcanon ] ) , for any map which is a generalized extreme point , the parameters @xmath29 must satisfy ( up to permutation ) @xmath223 .
this is compatible with the sign change condition if and only if at least two of the @xmath44 , which implies that @xmath1 be a cq map .
we now wish to examine in more detail those maps for which @xmath180 is not unitary .
we can assume , without loss of generality , that the singular values of @xmath180 can be written as @xmath224 and @xmath225 , that @xmath226 , and that @xmath227 .
recall that we showed in lemma 15 of @xcite that one can use the singular value decomposition of @xmath180 to write r _
= v w^ = u_+ + u_- where @xmath228 . and @xmath229 are unitary .
thus , @xmath1 is the midpoint of a line segment in @xmath221 and can be written as [ eq : convcb2 ] = _
u+ + _ u- with @xmath230 defined as in ( [ eq : genu ] ) .
although @xmath230 need not have the canonical form ( [ eq : tcanon ] ) , they are related so that their sum does .
we now use the singular value decomposition of @xmath180 to decompose it into unitary maps in another way .
r _ & = & [ eq : rsingval ] v w^ + & = & v w^ + [ eq : preconvcomb ] & = & vw^ + v _ z w^ .
moreover , it follows from ( [ eq : preconvcomb ] ) that [ eq : convcb3 ] = _
vw^ + _ v _ z w^ + ( 1 - _ 1 ) _ 0 where @xmath231 is the qc map corresponding to @xmath232 .
since we have assumed that we do _ not _ have @xmath233 , equation ( [ eq : convcb3 ] ) represents @xmath1 as a non - trivial convex combination of at least _ two _ distinct cpt maps , the first two of which are generalized extreme points .
( unless @xmath234 or @xmath235 , we will have three distinct points , and can already conclude that @xmath1 lies in the interior of a segment of a plane within @xmath221 . ) now , the assumption that @xmath236 suffices to show that the decompositions ( [ eq : convcb3 ] ) and ( [ eq : convcb2 ] ) involve different sets of extreme points and , hence , that @xmath1 can be written as a point on two distinct line segments in @xmath221 .
therefore , there is a segment of a plane in @xmath221 which contains @xmath1 and for which @xmath1 does not lie on the boundary of the plane ( although the plane might be on the boundary of @xmath221 ) .
thus we have proved the following .
now let @xmath237 denote the set of maps for which @xmath238 or , equivalently @xmath239 , is in @xmath221 . since @xmath237 is a convex set isomorphic to @xmath221 , its elements can also be broken into two classes as above .
the set of entanglement breaking maps is precisely @xmath240
. we can now prove theorem [ sect : convcq ] by showing that the convex hull of cq maps is @xmath240 .
let @xmath1 be in @xmath240 which is also a convex set .
if @xmath1 is a generalized extreme point of either @xmath221 or @xmath237 , then the only possibility consistent with @xmath1 being entanglement - breaking is that it is cq .
thus we suppose that @xmath1 belongs to class ii for both @xmath241 and @xmath242
. then @xmath1 lies within a plane in @xmath241 and within a plane in @xmath242 .
the intersection of these two planes is non - empty ( since it contains @xmath1 ) and its intersection must contain a line segment in @xmath240 which contains @xmath1 and for which @xmath1 is _ not _ an endpoint . therefore , @xmath1 is _ not _ an extreme point of @xmath240 .
thus all possible extreme points of @xmath240 must be generalized extreme points of @xmath221 or @xmath237 , in which case they are cq .
although this shows that all extreme points of @xmath240 are cq maps , this need not hold for the various convex subsets , corresponding to allowed values of @xmath243 in a fixed basis , discussed at the start of section [ sect : cp ] .
the following remark shows that `` most '' points in the convex subset @xmath110 of the @xmath111-@xmath112 plane can , in fact , be written as a convex combination of cq maps in canonical form in the same basis .
it also shows why it is necessary to go outside this region for those points close to the boundary .
* first consider the set of entanglement - breaking maps with @xmath244 , which is the convex set @xmath245 .
every extreme point must be an extreme point of the convex set @xmath246 for some @xmath100 . by theorem [ thm : rsw ] , these are the maps for which @xmath247 is unitary , which implies that either * * @xmath203 and @xmath248 which implies that either @xmath249 or @xmath250 with @xmath251 for @xmath252 or @xmath253 , _ or _ * * @xmath199 and @xmath254 . + the first type of extreme point is obviously a cq map ; the second is a `` point '' channel which , as noted before , is a special case of a cq map . thus any map in @xmath246 can be written as a convex combination of cq maps in @xmath246 . +
similar results hold if @xmath249 or @xmath250 .
therefore , any entanglement breaking channel with some @xmath44 , can be written as a convex combination of cq channels with at most one non - zero @xmath29 _ in the same basis_. thus any planar channel can be written as a convex combination of cq channels in the same plane .
* next consider entanglement - breaking maps with at most one non - zero @xmath135 .
we can assume , without loss of generality , that @xmath203 in which case the conditions for complete positivity reduce to ( [ eq : cp.tau0 ] ) .
combining this with the sign change condition yields ( |_1| + | _ 2|)^2 ( 1 - | _ 3|)^2 - t_3 ^ 2 .
it follows that for each fixed value of @xmath99 the set of allowable @xmath255 form a square with corners @xmath256 where @xmath257 .
thus , the extreme points of @xmath258 are planar channels which , by part(a ) are in the convex hull of cq channels . in particular , a map with @xmath259 ,
can be written as a convex combination of cq maps with either @xmath250 or @xmath244 .
however , these maps need not necessarily lie in @xmath258 ; we can only be sure that @xmath249 and @xmath260 , but not that @xmath261 .
thus we can only state that @xmath258 is in the convex hull of those cq maps with @xmath262 and @xmath263 for either @xmath252 or @xmath253 .
although it may be necessary to enlarge the set @xmath258 in order to ensure that it is in the convex hull of some subset of cq maps , these cq maps will have the canonical form _ in the same basis _ , and the same value for @xmath99 in that basis .
* now consider the convex subset @xmath264 of the @xmath111-@xmath112 plane corresponding to entanglement breaking maps with @xmath265 fixed .
these two regions intersect when either @xmath249 or @xmath250 ( or , equivalently , @xmath266 where @xmath136 ) .
one can again use part ( a ) to see that these intersection points can be written as convex combinations of cq maps in canonical form in the _
same basis_. since their convex hull has the same property , the resulting parallelogram , as shown in figure [ fig : convhull ] , is also a convex combination of cq maps of the same type .
only for those points in the strip between the parallelogram and the boundary might one need to make a change of basis in order to write the maps as a convex combination of cq maps .
* now suppose @xmath267 . since any two signs can be changed by conjugation with a pauli matrix , @xmath1 is unitarily equivalent to a map with @xmath268 .
one can then conjugate with another unitary matrix ( corresponding to a rotation on the bloch sphere ) to conclude that @xmath1 is unitarily equivalent to a channel @xmath269 with @xmath270 , and @xmath271 .
it then follows from part ( b ) that @xmath269 , and thus also @xmath1 , can be written as a convex combination of cq channels which have the form described above in the _ rotated basis_. however , these maps need not necessarily have the canonical form in the original basis .
consider the region @xmath169 with @xmath272 and @xmath273 .
the maps with @xmath274 lie on the boundary of this region ( in fact , at the intersection of the boundary with the @xmath142 axes , as shown in figure [ fig : convhulloct ] ) . since these maps have the form considered in part ( d ) they can be written as a convex combination of cq maps
; however , those cq maps need not have the canonical form in the original basis .
nevertheless , every point in the octagon formed from the convex hull of the intersection points of the lines @xmath275 , @xmath276 , and @xmath277 with the boundary , as shown in figure [ fig : convhulloct ] ) , can be written as a convex combination of cq maps as described above .
as another example , consider the set of entanglement breaking maps with @xmath278 fixed .
for any fixed @xmath279 , the set @xmath280 is a convex subset of the @xmath112-@xmath99 plane .
let @xmath281 be a point in this subset that lies between the boundary and a parallelogram as described in ( c ) above .
by considering the associated map as a point in the set @xmath264 instead , one can be sure that it can be written as a convex combination of cq maps since this subset of the @xmath111-@xmath112 plane is of the type described in ( b ) .
moreover , these boundary points can be added to the convex hull of cq maps without need for a change of basis .
one might expect that additional boundary points could be added in various ways with additional ingenuity and bases changes . that this is always true ,
is the essence of theorem [ thm : cqqubit ] . only for points near the boundary with two @xmath135 non - zero
is it necessary to actually make the change of basis used in the proof of this theorem . in other cases , the necessary convex combinations ( which are not unique )
can be formed using the strategies outlined above .
* acknowledgment : * this paper is the outgrowth of discussions with a number of people .
i owe a particularly large debt to peter shor for stimulating my interest in this topic by communicating the results of @xcite prior to publication .
both m. horodecki and p. shor communicated several other results which have now been incorporated into @xcite .
it is also a pleasure to thank elisabeth werner for helpful discussions , particularly regarding section [ sect : convcq ] ; chris king for many helpful discussions ; john cortese for making stimulating comments , communicating a counter - example to the sufficiency of the determinant condition , and providing many plots , including four of the figures in this paper , which helped to clarify my thinking .
part of this work was done while visiting the mathematics department at the amherst campus of the university of massachusetts ; i am grateful to professor donald st .
mary for arranging this visit and providing a very hospitable working environment .
finally , i would like to thank heide narnhofer for permission to use the postscript files produced by k. durstberger for figure [ fig : octa ] .
a. fujiwara and p. algoet , affine parameterization of completely positive maps on a matrix algebra " , preprint ; subsequently published as `` one -to - one parameterization of quantum channels '' _ phys .
a _ * 59 * , 3290 3294 , 1999 . m. hordecki , p. hordecki and r. hordecki , `` mixed - state entanglement and quantum communication '' in _ quantum information : an introduction to basic theoretical concepts and experiments _ ( springer tracts in modern physics * 173 * , 2001 ) .
quant - ph/0109124 .
-@xmath146 plane showing the regions described by the diagonal conditions ( dotted lines ) and the curves corresponding to @xmath284 for @xmath285 and @xmath286 . the closed curve and its interior describes the parameters for which the corresponding map is completely positive.,width=566,height=566 ] -@xmath146 plane showing the region determined by determinant condition when @xmath287 and @xmath288 and the corresponding region with @xmath145 and@xmath146 interchanged .
their intersection corresponds to the entanglement breaking maps with the indicated parameters.,width=302,height=302 ] -@xmath146 plane corresponding to entanglement breaking maps with @xmath287 and @xmath289 .
the dotted lines show the convex hull of the intersection points , which are planar maps.,width=264,height=264 ] -@xmath146 plane corresponding to entanglement breaking maps with @xmath290 and @xmath291 , because the intersections of the axes with the boundary ( at @xmath292 , for which all @xmath293 correspond to maps known to be in the convex hull of cq maps , one can enlarge the convex hull of such maps from the dotted line to the octagon shown by the dashed line.,width=340,height=340 ] | this paper continues the study of stochastic maps , or channels , for which @xmath0 is always separable in the case of qubits .
we give a detailed description of entanglement - breaking qubit channels , and show that such maps are precisely the convex hull of those known as classical - quantum channels .
we also review the complete positivity conditions in a canonical parameterization and show how they lead to entanglement - breaking conditions . |
in general , under some appropriate conditions , @xmath0-estimators enjoy some nice properties , for example , the asymptotic efficiency of ( quasi-)mle and oracle properties of regularized estimators .
however , it may suffer from heavy computation load since it is batch estimation in the sense that we have to optimize the appropriate objective function .
it would be of great help to be able to carry out recursive estimation , where we update the estimator by doing some fine tuning of the previous one . in this way
, it enables us to process enormous data effectively .
recursive estimation is established as an application of stochastic approximation theory , which is mainly used in control system and computer science ( see e.g. , robbins and monro @xcite , nevelson and khasminskii @xcite , dvoretzky @xcite , borkar @xcite ) .
the case of independent data was developed by , among others , fabian @xcite and @xcite .
we refer to sharia @xcite , @xcite , @xcite and @xcite for asymptotics of time - series model recursive estimator .
see kutoyants ( * ? ? ?
* section 2.6.6 ) and lazrieva et al .
@xcite as well as the references therein for the case of continuous - time data from a diffusion model .
the model we consider through this paper is the scaled wiener process with drift : @xmath1 where @xmath2 is a random and time - varying nuisance process .
suppose that we observe only discrete - time sample @xmath3 for @xmath4 with @xmath5 , where @xmath6 .
we know that the quasi - mle of a true value @xmath7 with regarding @xmath8 is @xmath9 where @xmath10 , and @xmath11 has the asymptotic normality and the asymptotic efficiency .
the aim of this paper is to propose an update estimation method for the diffusion parameter @xmath12 under the nuisance drift element @xmath13 .
specifically , we wish to construct a sequence of estimates @xmath14 , computed in a recursive manner , in such a way that for any @xmath15 , the difference @xmath16 is a function described by @xmath17 , @xmath18 , @xmath19 , and some ( not all ) data , and @xmath20 exhibits suitable convergence property for an arbitrary initial value @xmath21 .
usually this estimator does not require any numerically hard optimization , while its asymptotic behavior does require careful investigation .
note the difference from the recursive ( online ) estimation for sample @xmath22 , where each successive estimator is obtained by using data observed one after another .
we will bring about good asymptotic property of @xmath20 as efficient as the quasi - mle , i.e we ensure the asymptotic equivalence of @xmath20 to @xmath23 : @xmath24 this paper is organized as follows . in section [ sec : objective ] , we propose an update formula and give a main theorem with the proof .
we consider an example and present the simulation result to illustrate the theory in section [ sec : sim ] .
we consider the scaled wiener process with drift , where @xmath2 is a random and time - varying nuisance process and @xmath25 .
we denote by @xmath26 the family of distributions of @xmath27 associated with @xmath12 : @xmath28 , @xmath29 .
let @xmath30 , the underlying filtration .
suppose that we observe only discrete - time sample @xmath31 for @xmath32 with @xmath5 , where @xmath33 .
we wish to estimate the true value @xmath34 based on @xmath35 .
as is well known , the quasi - likelihood function , denote by @xmath36 , with regarding @xmath8 is @xmath37 therefore , the quasi - mle of @xmath12 is , and @xmath11 has the asymptotic normality @xmath38 and the asymptotic efficiency .
the aim of this paper is to propose an update estimation method for the diffusion parameter @xmath12 having the form @xmath39 where @xmath40 is an appropriate fine tuning function .
the upper index @xmath41 implies that we have the data set @xmath35 . under the nuisance drift element @xmath13
, we ensure the asymptotic equivalence of @xmath20 to @xmath23 .
therefore , the asymptotic normality and the asymptotic efficiency of @xmath20 hold .
we now define some notations : denote @xmath42 and @xmath43 by the convergence rate and the asymptotic variance of @xmath23 , respectively , i.e. , @xmath44 and @xmath45 .
we also define the quasi - likelihood function as @xmath46 . then , we propose the following update formula : @xmath47 note that , in finite sample , this formula is more stable than the one which is derived by direct using of the newton - raphson method .
we have @xmath48 hence the update formula is @xmath49 note that right - hand side of is positive a.s . for any initial value @xmath50 .
then , for any @xmath51 and @xmath15 we define @xmath52 note that @xmath53 for @xmath15 .
now , we set two assumptions : there exists @xmath54 such that @xmath55 .
[ ass : hn - rate2 ] @xmath56\xrightarrow{p } 0 ; \label{ass : m1}\end{aligned}\ ] ] @xmath57<\infty .
\label{ass : m2}\end{aligned}\ ] ] [ ass : moment ] the notation @xmath58 in assumption [ ass : hn - rate2 ] means that @xmath59 .
the following theorem [ thm : mainthm ] is a main result in this paper .
consider the model .
assume that assumptions [ ass : hn - rate2 ] and [ ass : moment ] hold .
then , for any @xmath50 the update formula generates an estimator @xmath20 , which is asymptotic equivalent to @xmath23 : @xmath60 therefore , it has the asymptotic normality : @xmath61 and the asymptotic efficiency .
[ thm : mainthm ] to prove the theorem , we use the result of sharia ( * ? ?
* lemma 1 ) .
we change some notations from the original ones of sharia @xcite in terms of a triangular array of random variables .
we drop the index @xmath41 to simplify some notations .
we can rewrite to @xmath62 where @xmath63 moreover , we define the following notations : @xmath64 ; \nonumber\end{aligned}\ ] ] @xmath65 @xmath66 where @xmath67 and @xmath68 , and introduce the conditions of sharia ( * ? ?
* lemma 1 ) applied to our model setting : 1 .
@xmath69 + w.r.t .
@xmath70 , where @xmath71 is a random variable with @xmath72 ; 2 .
@xmath73 + in probability @xmath70 ; 3 .
@xmath74 + in probability @xmath70 .
now , let us show the above conditions and derive the _ local asymptotic linearity _ of @xmath20 : @xmath75 where @xmath76 is a linear statistic .
clearly , ( i ) holds since we have @xmath77 . to check ( ii ) , we calculate @xmath78 \nn \\ & = -\frac{1}{2\hat{\beta}_{j-1}}+\frac{\beta_{0}}{2\hat{\beta}_{j-1}^{2}}\left(1 + 2\sqrt{\frac{h_{n}}{\beta_{0}}}e_{\beta_{0}}\left[y_{j}\overline{\mu}_{j}\big|\mathcal{f}_{t_{j-1}}\right]+\frac{h_{n}}{\beta_{0}}e_{\beta_{0}}\left[|\overline{\mu}_{j}|^{2}\big|\mathcal{f}_{t_{j-1}}\right]\right ) \nn \\ & = -\frac{d_{j-1}}{2\hat{\beta}_{j-1}^{2}}+\frac{\sqrt{h_{n}\beta_{0}}}{\hat{\beta}_{j-1}^{2}}e_{\beta_{0}}\left[y_{j}\overline{\mu}_{j}\big|\mathcal{f}_{t_{j-1}}\right]+\frac{h_{n}}{2\hat{\beta}_{j-1}^{2}}e_{\beta_{0}}\left[|\overline{\mu}_{j}|^{2}\big|\mathcal{f}_{t_{j-1}}\right ] , \nonumber\end{aligned}\ ] ] @xmath79+\frac{h_{n}}{2\hat{\beta}_{j-1}^{2}}e_{\beta_{0}}\left[|\overline{\mu}_{j}|^{2}\big|\mathcal{f}_{t_{j-1}}\right]\right\ } \nn \\ & = \frac{1}{\beta_{0}^{2}}\left(-\frac{d_{j-1}}{2}+\sqrt{h_{n}\beta_{0}}e_{\beta_{0}}\left[y_{j}\overline{\mu}_{j}\big|\mathcal{f}_{t_{j-1}}\right]+\frac{h_{n}}{2}e_{\beta_{0}}\left[|\overline{\mu}_{j}|^{2}\big|\mathcal{f}_{t_{j-1}}\right]\right ) .
\nonumber\end{aligned}\ ] ] hence , we have @xmath80+\frac{h_{n}}{2}e_{\beta_{0}}\left[|\overline{\mu}_{j}|^{2}\big|\mathcal{f}_{t_{j-1}}\right]\right ) \nn \\ & = \frac{1}{\beta_{0}^{2}}\left(\sqrt{h_{n}\beta_{0}}e_{\beta_{0}}\left[y_{j}\overline{\mu}_{j}\big|\mathcal{f}_{t_{j-1}}\right]+\frac{h_{n}}{2}e_{\beta_{0}}\left[|\overline{\mu}_{j}|^{2}\big|\mathcal{f}_{t_{j-1}}\right]\right ) , \nonumber\end{aligned}\ ] ] and therefore @xmath81+\frac{h_{n}}{\sqrt{n}}\sum_{j=1}^{n}e_{\beta_{0}}\left[|\overline{\mu}_{j}|^{2}\big|\mathcal{f}_{t_{j-1}}\right ] . \label{eq:(ii)}\end{aligned}\ ] ] the first term of converges to @xmath82 in probability @xmath70 since we assume .
let us show @xmath83\rightarrow 0 \nonumber \ ] ] in probability @xmath70 .
then , by making use of markov s inequality , assumption [ ass : hn - rate2 ] and , we obtain for any @xmath84 @xmath85\lesssim h_{n}\sqrt{n}\rightarrow 0 , \nonumber\end{aligned}\ ] ] hence ( ii ) . to check the condition ( iii )
, we also calculate @xmath86-\frac{h_{n}}{2\hat{\beta}_{j-1}^{2}}e_{\beta_{0}}\left[\overline{\mu}_{j}^{2}\big|\mathcal{f}_{t_{j-1}}\right]\right\}-\left\{-\frac{1}{2\beta_{0}}+\frac{\beta_{0}}{2\beta_{0}^{2}}\left(y_{j}+\overline{\mu}_{j}\sqrt{\frac{h_{n}}{\beta_{0}}}\right)^{2}\right\ } \nn \\ & = \frac{1}{\beta_{0}^{2}}\left\{-\frac{\hat{\beta}_{j-1}}{2}+\frac{\beta_{0}}{2}\left(y_{j}+\overline{\mu}_{j}\sqrt{\frac{h_{n}}{\beta_{0}}}\right)^{2}+\frac{d_{j-1}}{2}-\sqrt{h_{n}\beta_{0}}e_{\beta_{0}}\left[y_{j}\overline{\mu}_{j}\big|\mathcal{f}_{t_{j-1}}\right]-\frac{h_{n}}{2}e_{\beta_{0}}\left[|\overline{\mu}_{j}|^{2}\big|\mathcal{f}_{t_{j-1}}\right]\right\ } \nn \\ & \ \ \ + \frac{1}{2\beta_{0}}-\frac{1}{2\beta_{0}}\left(y_{j}+\overline{\mu}_{j}\sqrt{\frac{h_{n}}{\beta_{0}}}\right)^{2 } \nn \\ & = \frac{1}{\beta_{0}^{2}}\left(-\sqrt{h_{n}\beta_{0}}e_{\beta_{0}}\left[y_{j}\overline{\mu}_{j}\big|\mathcal{f}_{t_{j-1}}\right]-\frac{h_{n}}{2}e_{\beta_{0}}\left[|\overline{\mu}_{j}|^{2}\big|\mathcal{f}_{t_{j-1}}\right]\right ) . \nonumber\end{aligned}\ ] ] assumptions [ ass : hn - rate2 ] and [ ass : moment ] also ensure the condition ( iii ) : @xmath87 in probability @xmath70 .
consequently , we derive the local asymptotic linearity from sharia ( * ? ? ? * lemma 1 ) . in this case , we obtain @xmath88 therefore we can conclude the claim .
see e.g. , jacod and shiryaev @xcite for more detailed discussion on the asymptotic properties of an asymptotically linear estimator .
we performed a simulation study to validate our proposed update formula .
we consider the model @xmath90 where we know the true value @xmath91 .
note that this means @xmath92 , hence assumption [ ass : moment ] holds .
we set @xmath93 , data size @xmath94 and the monte carlo trial number @xmath95 .
we also set @xmath96 to satisfy assumption [ ass : hn - rate2 ] .
we generate @xmath97 through the following steps : we repeat step 1 and 2 @xmath105 times .
the following figures preset the simulation results . figures [ fig : sam ] and [ fig : beta - trace ] are traces of data @xmath106 and @xmath101 at @xmath107 , respectively ( @xmath108 denote each steps of monte carlo trials ) .
@xmath109-axis has update times @xmath110 . the dotted line in figure [ fig : beta - trace ] denotes the true value @xmath111 . in figure [
fig : average - trace ] we plot @xmath112 for @xmath110 , where @xmath113 and @xmath114 denotes the value of @xmath101 at @xmath115 . we give a qq - plot for @xmath116 in figure [ fig : qqplot ] . from the figure [ fig : average - trace ]
, we expect that @xmath117 as @xmath118 .
actually , whenever we repeat this algorithm , @xmath119 nearly equal to @xmath7 and standard deviations are not so large ( about @xmath120 ) .
we also expect that @xmath121 as @xmath118 from the qq - plot in figure [ fig : qqplot ] since the plots almost lay on the 45 degree line .
n. lazrieva , t. sharia , and t. toronjadze .
semimartingale stochastic approximation procedure and recursive estimation .
, 153(3):211261 , 2008 .
nevelson , r.z .
providence , ri : american mathematical society , 1973 . | we propose an update estimation method for a diffusion parameter from high - frequency dependent data under a nuisance drift element .
we ensure the asymptotic equivalence of the estimator to the corresponding quasi - mle , which has the asymptotic normality and the asymptotic efficiency .
we give a simulation example to illustrate the theory . |
the simplest and most interesting generalization of the friedmann - lemaitre - robertson - walker ( flrw ) cosmology is the bianchi ix model , whose geometry has the homogeneity constraint but the dynamics makes allowance of anisotropic evolution of different ( linearly independent ) spatial directions .
belinski , kalatnikov and lifshitz ( bkl ) at the end of sixties @xcite derived the oscillatory regime characterizing the asymptotic evolution near the cosmological singularity , describing via a discrete map the resulting chaos . + this mixmaster @xcite dynamics , whose geometry is invariant under the @xmath6 group , allows the line element to be decomposed as a flrw model plus a gravitational waves packet ( @xcite , @xcite ) sufficiently far from the singularity .
+ the various approaches in terms of continuous variables ( i.e. construction of an invariant measure for the system @xcite and study of the system covariance @xcite-@xcite ) have shown how this chaotic feature is invariant with respect to any choice of the temporal gauge and how in this sense it is intrinsic .
+ such deep nature of the mixmaster deterministic chaos and its very early appearance in the universe evolution lead to believe in the existence of a relation with the quantum behavior the system performs during the planckian era .
+ the aim of this paper is to give a precise meaning to this relation by constructing the semiclassical limit of a schedinger approach to the canonical quantization of the arnowitt - deser - misner ( adm ) dynamics whose corresponding probability distribution coincides with the ( deterministic ) microcanonical one .
+ in section [ billiard ] are outlined the stochastic properties of the mixmaster dynamics , being isomorphic to a billiard on a lobachevsky plane and is given a stationary statistical distribution within the framework of the microcanonical ensemble .
+ in section [ limit ] we show the existence of a direct correspondence between the classical and quantum dynamics outlined by the common form of the continuity equation for the statistical distribution and the one for the first order approximation in the semiclassical expansion . + as a remarkable feature , both formalisms ( the classical and the quantum one ) are constructed in a generic temporal gauge .
as well known ( see @xcite ) , the dynamics of the bianchi type ix model , in terms of generic misner - chitr like ( mcl ) variables @xmath7 and corresponding coniugate momenta , is summarized by the variational principle @xmath8 where @xmath9 and @xmath10 denotes a potential term which , asymptotically to the singularity , is modeled by the potential wall @xmath11 being @xmath12 ( @xmath13 ) the anisotropy parameters @xcite , such that the motion of the point universe is restricted in the domain @xmath14 ( see figure [ fig : domain ] ) .
such a dynamical scheme has the relevant feature to let free the choice of the lapse function unless explicitly chosen the form of the function @xmath15 .
the bounces against the potential walls and the instability of the geodesic flow on the lobachevsky plane let the dynamics acquire a stochastic feature .
moreover in @xmath14 the adm hamiltonian , where the anisotropy parameters @xmath12 are simultaneously greater than 0 .
[ fig : domain],width=302 ] becomes ( asymptotically ) an `` energy - like '' integral of motion @xmath16 , say @xmath17 .
as shown in @xcite , the jacobi metric associated with the principle ( [ qnew ] ) describes a point
universe moving over a lobachevsky plane reduced by the potential walls ( [ aa ] ) to a billiard .
the statistics of the mixmaster stochasticity is reformulated following the lines presented in @xcite , @xcite and @xcite to derive the invariant measure .
+ this system is well - described by a _ microcanonical ensemble _ , whose liouville invariant measure @xmath18 can be obtained as a solution of the continuity equation @xmath19 over the reduced phase space denotes the @xmath20-circle of the momentum space variable . ]
@xmath21 , @xmath18 explicitly reads @xmath22 using generic mcl variables , the above invariant measure is independent of the choice of the temporal gauge , i.e. of the lapse function . in order to restrict the invariant measure to the two - dimensional space @xmath23
, we stress how the original hamiltonian equations in the asymptotic limit for which @xmath24 get in @xmath14 the free geodesic motion @xcite @xmath25 by this system , we easily rewrite the stationary continuity equation in the form @xmath26 when assuming the independence of the distribution function @xmath27 from @xmath20 , then the normalization condition leads to the restricted function @xmath28 thus , we get the following continuity equation for @xmath29 @xmath30 where @xmath20 plays the role of a parameter and the corresponding _ microcanonical _ solution on the whole configuration space @xmath31 reads @xmath32 we conclude observing how the hamilton equations retain the same form even when the potential walls have a finite ( non - zero ) value .
what above outlined is a mixmaster intrinsic feature and not an effect induced by a particular class of references : the whole mcl formalism and its consequences has been developed in a framework free from the choice of a specific time gauge .
+ since chaos appears close enough to the big bang , we infer that it has some relations with the indeterministic quantum dynamics the model performs in the _ planckian era_. this relation between quantum and deterministic chaos is searched in the sense of a semiclassical limit for a canonical quantization of the model .
the asymptotical principle ( [ qnew ] ) describes a two dimensional hamiltonian system , which can be quantized by a natural schredinger approach @xmath33 being @xmath34 the wave function for the point - universe and , implementing @xmath35 ( see ( [ n2 ] ) ) to an operator = i\hbar \ , , \qquad \left [ \hat{\theta},\hat{p_{\theta}}\right ] = i\hbar \ , .\ ] ] ] , i.e. @xmath36 the equation ( [ sch ] ) rewrites explicitly , in the asymptotic limit @xmath37 , @xmath38^{1/2 } \psi \ , , \end{aligned}\ ] ] where we took an appropriate symmetric normal ordering prescription and we left @xmath39 to stress that the potential can not be neglected on the entire configuration space @xmath31 .
being @xmath39 equal to infinity out of @xmath14 , @xmath40 requires as boundary condition to vanish outside the potential walls , say @xmath41 the _ quantum _ equation ( [ sch1 ] ) is equivalent to the wheeler - dewitt one for the same bianchi model , once separated the positive and negative frequencies solutions @xcite , with the advantage that now @xmath42 is a real time variable .
since the potential walls @xmath39 are time independent , a solution of this equation has the form @xmath43 where @xmath44 are constant coefficients and we assumed a discrete `` energy '' spectrum because the quantum point - universe is restricted in the finite region @xmath14 ; the position ( [ sol ] ) in ( [ sch1 ] ) leads to the eigenvalue problem @xmath45 \varphi_n = \nonumber \\ = { \left(\frac{{e_n}^2-u_{\infty}}{\hbar^2}\right ) } \varphi_n \equiv \frac{{e_{\infty}}^2_n}{\hbar^2 } \varphi_n \ , .
\end{aligned}\ ] ] in what follows we search the semiclassical solution of this equation regarding the eigenvalue @xmath46 as a finite constant ( i.e. we consider the potential walls as finite ) and only at the end of the procedure we will take the limit for @xmath39 .
we infer that , in the semiclassical limit when @xmath47 and the `` occupation number '' @xmath48 tends to infinity ( but @xmath49 approaches a finite value ) , the wave function @xmath50 approaches a function @xmath51 as @xmath52 the expression for @xmath51 is taken as a semiclassical expansion up to the first order , i.e. @xmath53 where @xmath54 and @xmath55 are functions to be determined . + substituting ( [ expa ] ) in ( [ eig ] ) and separating the real from the complex part , we get two independent equations , i.e. @xmath56\sqrt{r } \ , , \end{aligned}\ ] ] where we multiplied both sides by @xmath57 and , respectively , @xmath58 in the limit @xmath47 the second term of ( [ 1eq ] ) is negligible meanwhile the first one reduces to the hamilton - jacobi equation @xmath59 the solution of ( [ hamb ] ) can be easily checked to be is due to the model adopted and does not affect the probability distribution . ]
@xmath60 we observe that ( [ hamb ] ) , through the identifications @xmath61 is reduced to the algebraic relation @xmath62 the constraint ( [ hamr ] ) is nothing more than the asymptotic one @xmath63 and can be solved by setting @xmath64 where @xmath65 is a momentum - function related to @xmath66 and @xmath67 by the dynamics . on the other hand , by ( [ esse ] )
we get @xmath68 to verify the compatibility of these expressions with ( [ sethm ] ) we use the equations of motion ( [ ham ] ) @xmath69 the required compatibility comes from the identification @xmath70 . since @xmath71 we see by ( [ esse ] ) that the solution @xmath72 vanishes , as due in presence of infinite potential walls , outside @xmath14 .
+ the substitution in ( [ 2eq ] ) of the positions ( [ sethm ] ) ( when the potential walls become infinite , because of chaotic dynamics , @xmath20 behaves like an independent variable and plays here the role of a free parameter ) leads to the new equation @xmath73 we emphasize how this equation coincides with ( [ cont2 ] ) , provided the identification @xmath74 ; it is just this correspondence between the statistical and the semiclassical quantum analysis to ensure that the indeterminism of the quantum dynamics for the bianchi ix model approaches the deterministic chaos in the considered limit . + any constant function is a solution of ( [ cont3 ] ) , but the normalization condition requires @xmath75 and therefore we finally get @xmath76 say the limit for the quantum probability distribution as @xmath77 and @xmath47 associated to the wave function @xmath78 coincides with the classical statistical distribution on the microcanonical ensemble .
+ though this formalism of correspondence remains valid for all bianchi models , only the types viii and ix admit a normalizable wave function @xmath79 , being confined in @xmath14 , and a continuity equation ( [ cont2 ] ) which has a real statistical meaning .
+ since referred to stationary states @xmath80 , the considered semiclassical limit has to be intended in view of a `` macroscopic '' one and is not related to the temporal evolution of the model @xcite . | within a cosmological framework , we provide a hamiltonian analysis of the mixmaster universe dynamics on the base of a standard arnowitt - deser - misner approach , showing how the chaotic behavior characterizing the evolution of the system near the cosmological singularity can be obtained as the semiclassical limit of the canonical quantization of the model in the same dynamical representation .
+ the relation between this intrinsic chaotic behavior and the indeterministic quantum dynamics is inferred through the coincidence between the microcanonical probability distribution and the semiclassical quantum one .
* mixmaster chaoticity as semiclassical limit of the canonical quantum dynamics * * giovanni imponente@xmath0 + giovanni montani@xmath1 * @xmath2 dipartimento di fisica universit di napoli `` federico ii '' + @xmath3 dipartimento di fisica universit di roma `` la sapienza '' + @xmath4 infn napoli + @xmath5 icra international center for relativistic astrophysics + c / o dip .
fisica ( g9 ) , piazzale aldo moro 5 , 00185 roma , italy + e - mail : [email protected] , [email protected] pacs : 04.20.jb , 98.80.dr , 83c |
the process of tidal capture of a body into a bound orbit through the excitation of oscillatory internal modes is thought to be of general importance in astrophysics .
it is likely to play a role in binary formation in globular clusters ( eg .
press & teukolsky , 1977 , hereafter pt and references therein ) as well as in tidal interactions of stars within galactic centres . when there are repeated close encounters as for a highly eccentric orbit
, circularisation may result , a process of potential importance for extrasolar planets ( eg .
ivanov & papaloizou 2004 , hereafter ip ) .
until now , only the oscillation modes associated with spherical non rotating stars have been considered with possibly rotation being treated as a perturbation ( eg .
pt , ip ) . but
angular momentum is transfered in tidal encounters and when , as is often the case , the internal inertia is much less than that of the orbit , the tidally excited object would be expected to rotate at a balanced rate or undergo pseudo synchronisation .
then the rotation frequency is matched to the most important ones in a fourier decomposition of the tidal forcing .
it is then natural to expect that inertial modes controlled by the rotation ( see eg .
papaloizou & pringle 1981 , hereafter pp ) become important for the tidal response .
this is especially the case for a barotropic configuration has the same functional dependence on the density @xmath2 : @xmath3 , for the unperturbed configuration and during perturbation . ] , with no stratification and hence no low frequency @xmath4 modes .
we would expect inertial modes , which have periods that scale with the rotation frequency , to be particularly important for large periastron distances as the fundamental spherical modes suffer increasing frequency mismatch as this distance increases .
it is the purpose of this paper to calculate the response of a rotating body taking the inertial modes into account .
in particular we focus on modes with @xmath0 which dominate the angular momentum exchange and a full polytrope with index @xmath5 we show that the inertial modes indeed dominate the response at larger periastron distances of interest in many cases . in section 2 we give the basic equations governing the linear response of a uniformly rotating barotropic star to the tidal forcing due to a close encounter . in section 3
we give a self - adjoint representation of the linear problem and discuss how it can be used to provide a formal solution in terms of a spectral decomposition as can be done for a non rotating star . in section 4
we develop expressions for the energy and angular momentum exchange . in section 5
we give results for a polytrope with index @xmath6 , showing that the @xmath0 inertial wave response is contained within just a few global modes and that inertial modes dominate the tidal response for large periastron passages .
accordingly these must be considered when discussing tidal circularisation from large eccentricity for objects such as extrasolar planets .
we assume that the star is rotating with uniform angular velocity @xmath7 directed perpendicular to the orbital plane . in this case
the hydrodynamic equations for the perturbed quantities take the simplest form in the rotating frame . for a convective star , approximated as barotropic
, we have @xmath8 here @xmath9 is the lagrangian displacement vector , and @xmath10 where @xmath11 is the density , @xmath12 is the density perturbation , @xmath13 is the adiabatic sound speed , @xmath14 is the stellar gravitational potential arising from the perturbations and @xmath15 is the external forcing tidal potential .
note that the centrifugal term is absent in equation @xmath16 being formally incorporated into the potential governing the static equilibrium of the unperturbed star .
the convective derivative @xmath17 as there is no unperturbed motion in the rotating frame .
the linearised continuity and poisson equations are @xmath18 we use a cylindrical coordinate system @xmath19 with origin at the centre of the star and assume that the tidal potential and the resulting perturbation to some quantity , say @xmath20 , is represented in terms of a fourier transform over the angle @xmath21 and time @xmath22 .
thus @xmath23 , where the sum is over @xmath24 and @xmath25 and @xmath26 denotes the complex conjugate .
the reality of @xmath20 implies that the fourier transform , indicated by tilde satisfies @xmath27 to solve the forcing problem we obtain equations for the fourier transforms of the perturbations using @xmath28 we can then express @xmath29 in terms of @xmath30 with help of @xmath16 , the density perturbation in terms of @xmath31 , @xmath32 and @xmath30 with help of ( [ eq 2 ] ) , and substitute the results into the continuity equation ( [ eq 3 ] ) to get @xmath33 where @xmath34 , and @xmath35 @xmath36 it is very important to note that the operators @xmath37 , @xmath38 and @xmath39 are self - adjoint when the inner product @xmath40 with @xmath41 denoting the volume of the star . also when @xmath42 is not a constant @xmath43 and @xmath38 are positive definite and @xmath39 is non negative definite .
equations ( [ eq 2 ] ) , ( [ eq 4 ] ) and the poisson equation ( [ eq 3 ] ) form a complete set . when @xmath44 equation ( [ eq 4 ] ) describes the free oscillations of a rotating star .
these may be classified as relatively high frequency @xmath45 and @xmath46 modes and inertial modes with eigen frequencies @xmath47 .
the @xmath45 and @xmath46 modes exist in non rotating stars and can be treated in a framework of perturbation theory ( eg . ip ) . here
we focus on the response of inertial waves to tidal forcing , for slow rotation .
we seek a solution as a series expansion in the small parameters @xmath48 and @xmath49 , where @xmath50 , with @xmath51 and @xmath52 being the mass and the radius of the star , respectively .
the small parameters will be assumed to be of the same order . to zeroth order ( when @xmath53 ) the tidal potential induces a quasi - static bulge in which tidal forces are balanced by pressure and self gravity and @xmath54 with @xmath55 . from equation ( [ eq 4 ] )
we conclude that @xmath42 is smaller than the potential perturbations by a factor of order @xmath56 ( see pp for details ) .
accordingly , setting @xmath57 in equation ( [ eq 2 ] ) and using equation ( [ eq 3 ] ) we obtain @xmath58 in the lowest order approximation we find @xmath30 using equation ( [ eq 4 ] ) neglecting the higher order term proportional to @xmath59 on the right hand side .
we thus have @xmath60 where the source term @xmath61 is completely determined by the external forcing via equation ( [ eq 8 ] ) .
finally , the perturbation of the internal gravitational potential associated with @xmath30 , @xmath62 follows from ( [ eq 2 ] ) , ( [ eq 3 ] ) and ( [ eq 8 ] ) @xmath63
equation ( [ eq 9 ] ) is of a general type that governs the linear response of rotating bodies in a context that is much wider than the specific one considered here ( see lynden bell & ostriker 1967 ) .
one recurring difficulty has been that the way in which the frequency @xmath64 occurs does not readily allow application of the general spectral theory of self - adjoint operators to the response calculation even though @xmath43 , @xmath38 and @xmath39 are themselves self - adjoint ( eg .
dyson & schutz 1979 ) .
we here provide a formulation in which the problem can be dealt with in the same way as a standard self - adjoint one such as occurs for a non rotating star .
we begin by remarking that the self - adjoint and non negative character of @xmath43 , @xmath38 and @xmath39 allows the spectral theorem to be used to specify their square roots , eg .
@xmath65 , defined by condition @xmath66 .
the positiveness of @xmath43 allows definition of the inverse of @xmath65 , @xmath67 .
technically , this is done through using the spectral decomposition @xmath68 .. etc .
, where @xmath69 and @xmath70 are the eigenfunctions and the eigenvalues of the corresponding operators and the last equation defines a general function of the operator @xmath43 . the sum may formally include integration with an appropriate measure in the case of a continuous spectrum of the corresponding operators .
now let us consider a generalised two dimensional vector @xmath71 with components such that @xmath72 and the straightforward generalisation of the inner product given by equation ( [ eq 7 ] ) .
it is easy to see that equation ( [ eq 9 ] ) can be derived from the matrix form @xmath73 where @xmath74 and the source vector @xmath75 has the components @xmath76 it then follows from ( [ eq 12 ] ) that we may choose @xmath77 since the off diagonal elements in the matrix ( [ eq 13 ] ) are adjoint of each other and the diagonal elements are self adjoint , that the operator @xmath78 is self - adjoint is manifest .
now we can use the spectral theory of self - adjoint operators to look for a solution to ( [ eq 9 ] ) in the form @xmath79 , where @xmath80 are the real eigenfunctions of @xmath78 which satisfy @xmath81 the associated necessarily real eigen frequencies being @xmath82 substituting ( [ eq 14 ] ) into ( [ eq 12 ] ) we obtain @xmath83 where , following the landau prescription , we have added an infinitesimal imaginary part @xmath84 to the frequency @xmath85 the inner product induced by @xmath78 turns out to be @xmath86 , where @xmath87 . using ( [ eq 10 ] ) and ( [ eq 15 ] )
we explicitly obtain @xmath88 where @xmath89 is the norm .
the time derivative @xmath90 of the wave energy @xmath91 associated with the rotating frame and the time derivative @xmath92 of the wave angular momentum @xmath93 are given by usual expressions @xmath94 the wave energy associated with the inertial frame , @xmath95 is given in terms of @xmath91 and @xmath93 ( eg .
friedman @xmath96 schutz 1978 , hereafter fs ) through @xmath97 expressing @xmath98 with help of ( [ eq 2 ] ) in terms of @xmath42 and discarding the total derivatives which can not contribute to the total energy transfer @xmath99 , we obtain @xmath100 contributions @xmath101 to @xmath102 arising from different values of @xmath103 may be calculated separately .
substituting the corresponding fourier transforms into ( [ eq 19 ] ) , and integrating over time , we obtain @xmath104 where @xmath105 . note that for the problem on hand @xmath106 and @xmath107 are real ( see below ) .
now we substitute the spectral decomposition @xmath108 , where @xmath109 , together with ( [ eq 16 ] ) into ( [ eq 20 ] ) .
the singularities in the integral over @xmath64 corresponding to @xmath110 and determining all non - zero contributions to the integral are dealt with using the landau prescription .
then the energy transfer is a sum of contributions associated with each mode , being given by @xmath111 where @xmath112 , and both integrals are evaluated for @xmath110 . using the symmetry properties of the operator @xmath113
one can readily see that the integrals are , in fact , the negatives of each other , thus we finally obtain @xmath114 where @xmath115 .
the expression for the angular momentum transfer follows immediately from ( [ eq 23 ] ) and the general relation between wave energy and angular momentum corresponding to a mode with particular value of @xmath103 : @xmath116 ( eg .
we have @xmath117 and @xmath118 the expressions ( [ eq 23 ] ) , ( [ eq 24 ] ) can be evaluated when the star is in a parabolic orbit , being the limit of a highly eccentric orbit , using the expression for @xmath119 given in the quadrupole approximation by pt . from this
we obtain @xmath120 where @xmath121 , @xmath122 .
@xmath123 is the ratio of @xmath51 to the mass of the star exerting the tides , @xmath124 : @xmath125 .
@xmath126 is a typical frequency of periastron passage and @xmath127 is the periastron distance .
the functions @xmath128 specify the dependence of @xmath31 on @xmath64 , they are described by pt .
y=@xmath129 , where we use dimensionless frequencies @xmath130 and @xmath131 .
@xmath132 is the associated legendre function , and @xmath133 are the spherical coordinates .
now we express the density perturbation entering in ( [ eq 23 ] ) , ( [ eq 24 ] ) as @xmath134 here we introduce a factor @xmath135 to account for self - gravity of the star that can be obtained from the solution of ( [ eq 8 ] ) in the frequently used cowling approximation . ] .
we substitute this expression and ( [ eq 25 ] ) in ( [ eq 23 ] ) , ( [ eq 24 ] ) and express all quantities in natural units .
these are such that the spatial coordinates , density and sound speed are expressed in units of @xmath52 , the averaged density @xmath136 , and @xmath137 , respectively .
we then have @xmath138 @xmath139 where @xmath140 , @xmath141 , and @xmath142 , @xmath143 and @xmath144 .
the overlap integrals @xmath145 have the form : @xmath146 where @xmath147 .
the eigen frequencies @xmath148 and the integrals @xmath145 in equations ( [ eq 26]-[eq 27 ] ) have to be calculated numerically , for a given model of the star
. therefore , equation ( [ eq 14 ] ) describing free oscillations of the star must be solved . for our numerical work we model the star as @xmath6 polytrope and use the method of solution of ( [ eq 14 ] ) proposed by pp .
our results are in a good agreement with results obtained by lockitch @xmath96 friedman ( 1999 ) and dintrans @xmath96 ouyed ( 2001 ) who used different methods . ]
namely , we reduce ( [ eq 14 ] ) to an ordinary matrix equation using an appropriate set of trial functions , and solve the corresponding eigenvalue problem . the details of our approach will be given in a separate publication .
here we note that we use different numbers @xmath149 of the trial functions to study convergence of our numerical scheme with the largest number being equal to @xmath150 .
also we adopt a spherical model of the star assuming that the distortion of the stellar structure due to rotation does not influence significantly our results .
since the modes corresponding to @xmath0 fully determine the transfer of angular momentum in the quadrupole approximation and apparently dominate the energy transfer , which can only be increased if @xmath24 modes are included , we consider only them . [ theor3 ] the absolute values of the eigen frequencies @xmath151 are always smaller than or equal to @xmath152 .
for complete polytropes it is expected that they ultimately yield an everywhere dense discrete spectrum ( pp ) . in our numerical approach
their number is equal to @xmath153 , and therefore the number of the terms in the series in ( [ eq 26 ] ) and ( [ eq 27 ] ) depends on @xmath149 .
however , only the eigenfunctions with some definite values of @xmath151 give a significant contribution to the series .
these terms correspond to global eigen modes with a large scale distribution of @xmath154 over the star .
therefore , the overlap integrals corresponding to the global modes are not averaged to small values after integration over the volume of the star .
the number of the global modes does not depend on @xmath149 . in figure 1
we show the overlap integrals as functions of the eigen frequencies .
there is a sharp rise of @xmath145 near @xmath155 and very close to @xmath156 , and several peaks corresponding to different values of @xmath151 .
the rise at @xmath157 , @xmath25 is due to numerical inaccuracy of our method , but the corresponding modes practically do not influence our results and therefore the numerical inaccuracy is not significant for our purposes . note that the feature at @xmath157 can be removed by an appropriate regularisation of @xmath39 .
three peaks with @xmath158 and @xmath159 , @xmath160 and @xmath161 , @xmath162 and @xmath163 correspond to the global modes .
the spatial distribution of the corresponding eigenfunctions is shown in figure 2 .
for comparison we also show a non global mode with @xmath164 and a very small value of @xmath165 .
it is clear from this figure that the global modes have a large scale distribution of @xmath154 .
it follows from ( [ eq 26 ] ) and ( [ eq 27 ] ) that for a given model of the star , the quantities @xmath166 depend only on @xmath167 . in figure 3
we show the dependence of @xmath168 ( the solid black curve ) and @xmath169 ( the solid red curve ) on @xmath170 .
the solid green curve shows the dependence of the energy transfer in the inertial frame , @xmath171 .
note that the energy transfer in the inertial frame is negative when @xmath172 .
the dotted curves show the dependencies of the corresponding quantities calculated with only two global modes , the retrograde mode with @xmath158 and the prograde mode with @xmath160 .
figure 3 shows that the dotted and the solid curves are very close to each other .
therefore , only two modes are essential for our problem .
the angular momentum transfer is equal to zero when @xmath173 , where @xmath174 .
the condition @xmath175 may be easily realised in an astrophysical situation where the inertial modes dominate the tidal response and the moment of inertia of the star is sufficiently small . in this case
the system quickly relaxes to the so - called state of pseudo synchronisation where @xmath175 . in figure 4
we show the dependence of @xmath176 on @xmath177 for this case .
the black curve shows the contribution of the inertial waves , @xmath178 the red curve shows the contribution of the @xmath45 mode to the energy transfer calculated for a non - rotating star ( pt ) and the green curve shows the same quantity but calculated when @xmath179 ( ip ) .. this has no influence on the conclusions of the paper . ]
it is clear from this figure that when @xmath180 the inertial waves dominate the tidal response . for a planet with jupiter mass and radius circularising from large eccentricity to attain a final period of @xmath181 days , @xmath182 ( ip ) , such that inertial modes dominate by a wide margin .
in this paper we formulate a new self - adjoint approach to the problem of oscillations of uniformly rotating fully convective star and apply this approach to the tidal excitation of the inertial waves and tidal capture .
the approach can be used in any other problem where oscillations with typical frequencies of the order of the angular velocity are important .
it can also be extended to the case of convectively stable stars .
the oscillations with higher frequencies can also be treated in framework of an extension of our formalism .
we show that the tidal response due to inertial waves can be represented as a spectral decomposition over normal modes .
this formalism is general enough to be applied even when the spectrum is not regular and discrete . for full polytropes , only a few global modes with a large scale distribution of the perturbed quantities over the star contribute significantly to the response .
we find general expressions for the energy and angular momentum transfer between the orbit and inertial waves .
it is shown that when the inertial waves dominate the tidal response and a state of pseudo synchronisation is reached the energy transfer is determined by simple equation ( [ eq 29 ] ) . as follows from that equation the energy transfer does not decrease exponentially with the orbital periastron and therefore at large periastra the energy transfer due to inertial waves dominates over the energy transfer due to excitation of the fundamental mode .
this can be very important for the problem of tidal circularisation of the extra solar planets ( see above , ip and references therein ) , and can be of interest for other problems , such as eg .
the problem of tidal capture of convective stars in globular clusters .
the application of our approach to extra solar planets will be discussed in a separate publication .
pbi has been supported in part by rfbr grant 04 - 02 - 17444 . | we consider the excitation of the inertial modes of a uniformly rotating fully convective body due to a close encounter with another object .
this could lead to a tidal capture or orbital circularisation depending on whether the initial orbit is unbound or highly eccentric .
we develop a general self - adjoint formalism for the response problem and thus solve it taking into account the inertial modes with @xmath0 for a full polytrope with @xmath1 we are accordingly able to show in this case that the excitation of inertial modes dominates the response for large impact parameters and thus can not be neglected in calculations of tidal energy and angular momentum exchange or orbital circularisation from large eccentricity .
[ firstpage ] hydrodynamics ; stars : oscillations , binaries , rotation ; planetary systems |
the nel state is the simplest state with antiferromagnetic ordering , differing though from the true ground state of the antiferromagnetic xxx spin-@xmath0 chain , which is more complicated .
one may ask how close the nel state is to the true eigenstate of the spin - chain hamiltonian .
the answer to this question is actually known , as the overlap of the nel state with any given eigenstate can be explicitly calculated . as pointed out in @xcite ,
the overlap is related to the partition function of the six - vertex model on a rectangular lattice with reflecting boundary conditions .
a determinant representation for this overlap was obtained by tsuchiya @xcite .
restricting tsuchiya s expression to bethe eigenstates requires an extra step , and leads to a simpler , more compact determinant expression @xcite .
applications of these results range from condensed - matter physics @xcite to string theory @xcite and algebraic combinatorics @xcite . in this note
, we generalize tsuchiya s result to the case of partition functions of six - vertex model configurations with domain - wall boundary conditions that are only partially reflecting .
these partially reflecting domain - wall boundary conditions are related to tsuchiya s @xcite in the same way that the partial domain - wall boundary conditions and partition function in @xcite are related to korepin s domain - wall boundary conditions @xcite , and izergin s determinant expression for the corresponding partition function @xcite . the on - shell version of the partition function that we study describes overlaps of the bethe states with partial , or generalized nel states @xcite .
our derivations closely follow those in @xcite . in section
* [ section.02 ] * , we recall basic definitions related to the xxx spin-@xmath0 chain and the rational six - vertex model . in * [ section.03 ] * , we recall the definition of tsuchiya s reflecting - boundary boundary conditions , and the corresponding partition function , then introduce partial versions thereof , which we subsequently calculate . in * [ section.04 ] * , we compute the overlap of a partial nel state and a parity - invariant highest - weight on - shell bethe state . in * [ section.05 ] * , we collect a number of remarks . and in appendix * a *
, we explain the reduction of the @xmath1 $ ] determinant partition function to an @xmath2 $ ] one .
we restrict our attention to the xxx spin-@xmath0 chain of length @xmath3 @xcite .
each site carries an up - spin @xmath4 , or a down - spin @xmath5 .
the hamiltonian @xmath6 acts on @xmath7 as @xmath8 where @xmath9 permutes the spins on two adjacent sites labeled @xmath10 and @xmath11 .
-matrix and the boundary state .
_ , width=377 ] the key object in the bethe ansatz solution of the xxx spin-@xmath0 chain is the @xmath12-matrix @xcite , represented in figure * [ rk]*. the @xmath12-matrix , @xmath13 , acts on the tensor product of two spins labelled @xmath14 and @xmath15 , and depends on a complex spectral parameter @xmath16 , @xmath17 where @xmath18 is the permutation operator .
most importantly , the @xmath12-matrix satisfies the yang - baxter equation , @xmath19 represented in figure * [ ybfig]*. we will use the shorthand notation @xmath20 the @xmath12-matrix acts from south to north or , for the vertical crossing , from south - east to north - west , as represented in figure * [ rk]*. if each line carries a rapidity variable , the argument of the @xmath12-matrix is the difference of the two rapidity variables .
the states of the xxx spin-@xmath0 chain are generated by the @xmath21-operators .
these operators are constructed by multiplying the @xmath12-matrices along the spin - chain threaded with a single auxiliary space , @xmath22 the variables @xmath23 , @xmath24 , are the quantum - space or inhomogeneity variables .
they play an important rle in the intermediate steps of the derivations , but they are not necessary for diagonalizing the hamiltonian ( [ hham ] )
. the inhomogeneity variables are common to all @xmath21-operators used to construct the state . in this note , we do not consider the most general inhomogeneity variables as in ( [ b.operator.unpaired ] ) , but focus on parity - invariant bethe states such that the inhomogeneity variables are paired , as in @xmath25 the restriction to paired inhomogeneity variables will be important later on , when we consider overlaps of bethe states and the boundary states introduced in section * [ boundary.state]*. the bethe states of the xxx spin-@xmath0 chain are constructed by applying the @xmath21-operators on the ferromagnetic vacuum , @xmath26 , of the spin - chain , @xmath27 where @xmath28 , that is , the @xmath21-operator with all @xmath29-variables set to zero . for a bethe state to be an eigenstate of the hamiltonian ,
the rapidity variables @xmath30 must satisfy the bethe equations , @xmath31 as usual , we call bethe states with rapidity variables subject to the bethe equations on - shell states . generic bethe states , that are not eigenstates of the hamiltonian , are referred to as off - shell states . given the definition of the nel state , on an xxx spin-@xmath0 chain of length @xmath3 , @xmath32 and an integer @xmath33 , such that @xmath34 , we define an @xmath33-partial nel state as @xcite @xmath35 for @xmath36 , the state has an equal number of up- and down - spins , and we recover the original nel state , @xmath37 .
the matrix product state , _ mps _ , is defined as @xmath38 where @xmath39 and @xmath40 .
following @xcite , all partial nel states can be obtained from _ mps _ by projecting on subspaces with a definite number of up- and down - spins . denoting the projector on the state with @xmath33 down - spins by @xmath41 , we have , @xmath42 where @xmath43 is the spin - lowering operator .
the last term does not contribute to the overlap with a bethe state that is annihilated by @xmath44 , such as the highest - weight on - shell bethe states of the homogeneous xxx spin-@xmath0 chain .
as noted in @xcite , the nel state can be constructed from the boundary state associated with the diagonal reflection matrix .
the overlap of a nel state on a one - dimensional lattice of length @xmath3 and a bethe state that is not necessarily on - shell , characterised by @xmath45 rapidity variables , is equal to the partition function of the six - vertex model on a rectangular lattice that has @xmath45 horizontal lines and @xmath46 vertical lines , with reflecting - boundary domain - wall boundary conditions .
following @xcite , the latter is an @xmath47 $ ] determinant @xcite .
we extend this construction by effectively allowing for non - diagonal scattering off the boundary .
the latter boundary state reduces to a partial nel state ( [ generalized - neel ] ) for appropriate values of the variables .
the boundary state is defined as @xmath48 although the boundary state is associated with two lattice lines , it depends on a single rapidity @xmath16 .
the alternating inhomogeneity variables in ( [ b - op ] ) lead to one independent inhomogeneity variable for each boundary state .
the variables @xmath49 and @xmath50 are arbitrary , but fixed complex numbers .
the boundary state is the cross - channel representation of the reflection matrix @xcite .
the most commonly used boundary state is the neutral one with @xmath51 , which corresponds to the diagonal reflection @xcite .
it is this diagonal reflection matrix that was used in the derivation of @xcite .
we extend this result by adding the last , two - spins - up term , , which would correspond to the most general rational solution of the reflection equation @xcite .
it would be interesting to generalize the computation of the overlaps and the partition functions defined below to this case as well . ] allowed by the consistency conditions for integrable boundary scattering @xcite .
when representing the boundary state in terms of a diagram , we assume that it can only connect spins whose rapidity variables add to @xmath52 , as shown in figure * [ rk]*. the boundary state obeys the reflection equation @xcite , see figure * [ reflectionfig ] * , which , in our notation , takes the form @xmath53 the object of our interest is the overlap @xmath54 it depends on two sets of rapidity variables @xmath55 , and @xmath56 .
we will not put further constraints on these variables at this point .
the spectral parameters of the boundary state , as in figure * [ rk ] * , are correlated with the arguments of the @xmath12-matrices in ( [ b - op ] ) .
this is the reason for pairing them , instead of keeping them arbitrary .
the overlap of the on - shell bethe states of the homogeneous spin - chain with the partial nel states ( [ generalized - neel ] ) can be obtained by setting @xmath57 , @xmath58 , and @xmath49 to @xmath59 , and imposing the bethe equations on the rapidity variables @xmath30 , @xmath60 this follows from the structure of the boundary state ( [ boundary - state ] ) .
setting @xmath61 leaves only two terms in the boundary state ( [ boundary - state ] ) whose @xmath45-th tensor power then generates the sum of all partial nel states .
since any given bethe state has exactly @xmath33 down - spins , only the @xmath33-th nel state can have a non - zero overlap with it .
to evaluate the overlaps ( [ gentsoverlap ] ) and ( [ six - vertex - to - neel ] ) , we proceed along the same lines as @xcite , introducing along the way modifications necessary to account for spin - non - preserving term in the boundary state , or equivalently , a non - diagonal term in the reflection matrix .
the key step is to reformulate the problem in terms of a partition function of the rational six - vertex model on a rectangular lattice with a modified , or partial version of tsuchiya s reflecting - boundary domain - wall boundary conditions .
it should also be possible to use the recursion relation that was derived using the algebraic bethe ansatz in @xcite .
the tsuchiya determinant is a solution of this recursion relation .
the overlap ( [ gentsoverlap ] ) can be represented as a partition function of the rational six - vertex model on the @xmath62 $ ] rectangular lattice , as in figure * [ zxy ] * , where the spin - chain sites are associated with the vertical lines and the horizontal lines represent the auxiliary spaces of the @xmath21-operators .
the weights of the bulk vertices are the matrix elements of the @xmath12-matrix , while the weights of the boundary vertices are the coefficients of the boundary state ( [ boundary - state ] ) , as in figure * [ weights]*. the spins in the bulk are conserved in the sense that each vertex has two in- and two out - bound arrows , as in the left and middle columns of figure * [ weights]*. the spins on the boundary , with reflecting boundary conditions at @xmath63 are also conserved , as in the right column of figure * [ weights]*. in this case , spin conservation implies that @xmath64 .
the partition function on the resulting @xmath65 $ ] lattice admits a determinant representation , as shown by tsuchiya @xcite . in this note , we are interested in the more general case where the boundary conditions are partially reflecting , and the upper boundary can absorb an excess spin , thus allowing @xmath33 to be smaller than @xmath45 .
the resulting statistical mechanical system can be regarded as a degenerate version of tsuchiya s .
the extra horizontal lines can be systematically eliminated by taking @xmath66 auxiliary - space rapidity variables , in the original system , to infinity , and renormalizing the partition function appropriately to obtain a finite result .
the procedure is described in detail in @xcite , where the partition function of the six - vertex model with partial domain - wall boundary conditions was calculated by degenerating the partition function of the system with the domain - wall boundary conditions @xcite .
we do not follow this route here . instead , we follow the derivation of @xcite , while taking the more general structure of the boundary reflection matrix into account .
the derivation of @xcite relies on a recursion relation which relates the partition functions on lattices of different sizes .
the recursion relation is derived by f parts of the configurations using suitable choices of some of the free variables , and the observation that the type-@xmath14 or type-@xmath15 weights in figure * [ weights ] * vanish if the vertical and horizontal rapidity variables differ by @xmath59 . in this note
, we generalize this recursion relation to accommodate the partially - reflecting boundary conditions at @xmath67 .
the partition function of the statistical mechanical system that we are interested in is completely specified by the following four conditions .
_ , width=340 ] @xmath68 is symmetric in @xmath69 , and separately in @xmath70 .
this follows from repeated application of the yang - baxter and reflection equations to the partition function .
the symmetry in @xmath69 follows from commutativity of the @xmath21-operators , which is a consequence of the yang - baxter equation .
the symmetry in @xmath70 can be proven by standard manipulations with the reflection equation , which we reproduce here for completeness . multiplying both sides of the reflection equation ( [ reflection - eq ] ) by
@xmath71 and using the yang - baxter equation twice we get the equality depicted in figure * [ r1bar ] * , [ reflection - eq - bis ] @xmath72 the process can be iterated to add an arbitrary number of horizontal lines .
the resulting equality is an identity of two vectors in @xmath73 .
the next step is to project these vectors on the ground state @xmath74 , in other words to specify all arrows at the bottom of the diagram in figure * [ rfr]*. by spin conservation , all vertices below the freezing line are of type @xmath15 , and consequently produce a common scalar factor @xmath75 on both sides of the equation .
the remaining diagrams above the freezing line differ by the order of the vertical rapidity variables @xmath16 and @xmath76 .
the same procedure can be applied without much change to swap any two vertical rapidity variables on the @xmath62 $ ] lattice .
@xmath68 is a polynomial of degree @xmath77 in @xmath30 .
the variable @xmath30 appears in the boltzmann weights associated with the @xmath46 vertices on the @xmath78-th row of the partition function .
each type-@xmath14 and type-@xmath15 vertex contributes one power of @xmath30 , but does not flip the horizontal spin , while a type-@xmath79 vertex flips the spin but does not depend on @xmath30 . since the spin flips at least once , each horizontal line has to contain at least one type-@xmath79 vertex .
configurations with one @xmath79-vertex and @xmath77 vertices of @xmath14-type and @xmath15-type on the same line give a non - zero contribution to the statistical ensemble and , consequently , the partition function is a polynomial of degree @xmath77 in @xmath30 .
@xmath68 satisfies a recursion relation that equates the partition function on an @xmath80 $ ] lattice to a partition function on a smaller @xmath81 $ ] lattice , as soon as one of the horizontal rapidity variables is set to @xmath82 .
the recursion relation is derived by the freezing trick illustrated in figure * [ freetrick]*. first , using the yang - baxter and reflection equations , the @xmath78-th horizontal rapidity and the @xmath14-th vertical rapidity can be moved to the bottom - left corner of the lattice .
spin conservation leaves two possibilities for the bottom - left vertex , it is either type-@xmath15 or type-@xmath79 , as follows from figure * [ weights]*. if @xmath83 , the type-@xmath15 vertex has zero weight and the bottom - left corner of the partition function is then unambiguously determined .
once the corner weight is fixed , the vertices on the two left - most columns and the lower row are recursively determined by spin conservation .
this freezes the lower and left edges of the partition function as in figure * [ fig : subfig1]*. similarly , one can freeze the bottom - right corner of the partition function by setting @xmath84 .
the freezing trick results in two recursion relations : @xmath85 where @xmath86 and @xmath87 means that the variable @xmath30 is omitted . at @xmath88 , [ condition.4 ]
@xmath89 the partition function with no horizontal lines is just a product of @xmath45 non - reflective boundary weights .
the partition function vanishes for @xmath91 . for @xmath92 ,
it is completely determined by conditions * 1 * , * 2 * , * 3 * and * 4 * , for any @xmath33 and @xmath45 , since a polynomial of degree @xmath77 is completely fixed by its values at @xmath46 distinct points .
condition * 2 * , therefore , determines @xmath68 as a function of @xmath30 . eliminating @xmath93 s one by one
, we are left with @xmath94 , specified by condition * 4*. the solution of the recursion relation that satisfies all four conditions can be represented in the determinant form @xmath95 where @xmath96 is an @xmath47 $ ] matrix , @xmath97 checking the conditions * 1 * , * 2 * , * 3 * and * 4 * is straightforward .
symmetry in @xmath30 and @xmath29 is obvious .
the poles of the prefactor at @xmath98 , as well as at @xmath99 , are cancelled by the zeros of the determinant .
hence , the partition function is a polynomial in each of the @xmath30 s .
it is perhaps not immediately obvious why the degree of this polynomial is exactly @xmath77 , but one can check that the expansion of @xmath100 at @xmath101 starts with @xmath102 , because the lower - order terms are linear combinations of the first @xmath66 rows of the matrix @xmath103 . checking the recursion relations
is also easy , as the @xmath104 element of @xmath103 develops a pole at @xmath105 , which eliminates its @xmath106-th row and @xmath14-th column .
the determinant representation ( [ detts ] ) generalizes tsuchiya formula @xcite , to which this expression reduces when @xmath36 .
it represents the overlap as an @xmath107 $ ] determinant , where @xmath45 is half of the length of the spin chain . in the sequel
, we derive a more compact representation in terms of an @xmath1 $ ] determinant , where @xmath33 is the number of magnons , which is general is smaller than @xmath45 .
we also study the homogeneous limit when we set all the vertical rapidity variables to zero .
we should stress that the expression ( [ detts ] ) is valid off - shell , for any values of vertical and horizontal rapidity variables .
further , we study the on - shell limit of the partition function when the horizontal rapidity variables satisfy the bethe equations and the state @xmath108 is an eigenstate of the heisenberg hamiltonian . observing that the determinant in ( [ detts ] ) scales as @xmath109 , and using @xmath110 where @xmath111 means equality up to the leading order in @xmath112 , as well as the expansion @xmath113
we find that @xmath114
to compute the overlap of a partial nel state with a bethe state , according to ( [ six - vertex - to - neel ] ) , we * 1 . *
put the inhomogeneous @xmath47 $ ] determinant of section * [ inhomogeneous.n.n ] * in a smaller , @xmath1 $ ] , still inhomogeneous form , * 2 . *
take the homogeneous limit of the @xmath1 $ ] determinant , * 3 . *
put the auxiliary - space rapidity variables on - shell by imposing the bethe equations , then * 4 . *
reduce the size of the determinant to @xmath2 $ ] .
the normalized overlap is given by the resulting formula divided by the gaudin norm of the bethe vector .
the @xmath116 $ ] determinant ( [ detts ] ) admits two , different but equivalent @xmath1 $ ] representations , @xmath117 the derivation of this result involves standard manipulations of rational sums @xcite , the details of which are presented in appendix * [ adr]*. in this form , the determinant has almost no dependence on @xmath45 .
we reproduce here , for completeness , the derivation in @xcite of the on - shell overlap for the homogeneous spin chain , but all the next steps are mathematically the same as in the case of @xmath118 considered in @xcite .
taking the homogeneous limit in the @xmath1 $ ] determinant representation is straightforward and yields , @xmath120 where @xmath21 is an @xmath1 $ ] matrix with matrix elements , @xmath121 following @xcite , the mps ( [ mps - state ] ) is an eigenstate of the degree-3 conserved - charge operator @xmath122 , with an @xmath122-eigenvalue zero .
any on - shell bethe state , @xmath123 , must be an eigenstate of @xmath122 .
since the eigenvalue of @xmath122 is conserved , the overlap of the mps with @xmath123 , can be non - zero only if the @xmath122-eigenvalue of @xmath123 is equal to that of the mps , that is , also zero .
for @xmath123 to have an @xmath122-eigenvalue zero , it must be parity - even , that is , the set of bethe roots , @xmath124 , of @xmath123 must be invariant , as a set , under the parity transformation @xmath125 . since the partial nel states are components of the mps , the same reasoning applies to the overlap of a partial nel state and @xmath123.*. the reason is that the latter is not an eigenstate of @xmath126 , and the @xmath126 selection rule does not apply to its overlap with mps and consequently with any of the nel states . ] aside from the above reasoning , the technical reason for the vanishing of the overlap , in either ( [ off - shell ] ) or ( [ homogenb ] ) , is not immediately obvious .
however , it can be proven in the latter representation by noting that the matrix @xmath21 in ( [ bmatrix ] ) has a zero eigenvalue if the variables @xmath30 are bethe roots .
this can been seen as follows .
setting the rapidity variables @xmath124 on - shell , the matrix @xmath21 becomes , @xmath127 the fact that this matrix is degenerate follows from the identity @xmath128 where the contour of integration encircles the poles of the integrand at @xmath129 counterclockwise , and the last equality is obtained by evaluating the residue at infinity . as a consequence of ( [ identity.01 ] ) ,
the vector with components @xmath130 can be seen to be a zero eigenvector of @xmath21 , @xmath131 this implies that a partial nel state has a zero overlap with a generic highest - weight on - shell bethe state , where _ generic _ here means a state which is _ not _ invariant under parity transformation , @xmath132 .
if the set of rapidity variables @xmath133 of the on - shell bethe state @xmath123 is invariant , as a set , under @xmath132 , then @xmath123 is parity - invariant , its @xmath122-eigenvalue is zero , and the argument of subsection * [ overlap.vanishes ] * fails .
more concretely , the rapidity variables form pairs of equal magnitudes but opposite signs , @xmath134 which leads to a pole in the prefactor of ( [ homogenb ] ) .
this pole cancels the zero in the determinant which lead to the vanishing of the overlap in subsection * [ overlap.vanishes]*. to compute the overlap , we resolve the @xmath135 ambiguity , due to the zero and the pole , by shifting the rapidity variables slightly away from their parity - invariant values , by defining @xmath136 calculate the overlap for small but finite @xmath137 , then take the limit @xmath138 .
some of the matrix elements of @xmath21 in ( [ bmatrix ] ) diverge as @xmath138 , but the resulting matrix is degenerate and we need to consider also the subleading , @xmath139 term in the determinant . in the following ,
it is convenient to factor out the diagonal matrix @xmath140 , with elements @xmath141 where @xmath142 , and @xmath143 . defining @xmath144 and taking into account that @xmath145 , we find after some calculation , @xmath146 where we have used the bethe equations , but only after expanding in @xmath147 .
the singular part of the matrix @xmath148 is proportional to the @xmath149$]-projector , and has a zero determinant , which why we need to keep the next @xmath139 term . to leading order in @xmath147
, the matrix @xmath148 has @xmath150 large eigenvalues , with eigenvectors proportional to @xmath151^t$ ] and @xmath150 small eigenvalues , with eigenvectors proportional to @xmath152^t$ ] . denoting the projections on these subspaces as @xmath153 where the large and small components
are now @xmath2$]-matrices , and taking into account that @xmath154 and @xmath155 , we have , @xmath156 the large component is @xmath157 to write the small component in compact form , we introduce the following notations , @xmath158 and @xmath159 the small component is @xmath160 hence , @xmath161 the @xmath162 singularity of @xmath163 cancels the zero in the denominator of ( [ homogenb ] ) , and from ( [ six - vertex - to - neel ] ) and ( [ homogenb ] ) , we get , @xmath164 following @xcite , the determinant expression for the gaudin norm of a parity - invariant on - shell bethe state @xcite , can be factorized in the form @xmath165 the normalized overlap is given by @xmath166 this formula was obtained in @xcite for the nel state with @xmath167 .
the derivation for arbitrary @xmath33 follows from a symmetry argument and can be found in @xcite . here , we rederive it by inspecting the partition function of the six - vertex model with partially reflecting domain - wall boundary conditions .
in @xcite , kuperberg lists eight classes of domain - wall - type boundary conditions and partition functions . these include the original boundary conditions and partition function of korepin and izergin @xcite as well as tsuchiya s @xcite .
it is clear that the remaining six classes admit partial versions in parallel with those discussed in @xcite and in this note .
the overlap studied in this note was used in @xcite to compute a class of one - point functions in a four - dimensional conformally - invariant supersymmetric gauge theory , in the presence of a defect .
the formulation of the latter problem in terms of the six - vertex model with particular boundary conditions , may be useful in the sense that the six - vertex boundary states may have a direct physical meaning in the gauge theory .
the boundary state in ( [ boundary - state ] ) is a building block of the generalized nel states , and consequently of the mps , which naturally appears in the weak - coupling gauge - theory calculcations @xcite . identifying a similar building block in the d - brane boundary state that is related , at strong - coupling , to the defect _ via _ the ads / cft correspondence , would help in finding a fully non - perturbative construction , valid at any coupling .
o f wishes to thank m wheeler for collaboration on @xcite .
k z wishes to thank the university of melbourne and the australian national university , where the work in this note was initiated , for kind hospitality .
o f is supported by the australian research council .
k z is supported by the marie curie network gatis of the european union s fp7 programme under rea grant agreement no 317089 , the erc advanced grant no 341222 , the swedish research council ( vr ) grant 2013 - 4329 , and rfbr grant 15 - 01 - 99504 .
to derive ( [ alterts ] ) from ( [ detts ] ) , we introduce two @xmath47 $ ] matrices , these matrices have the structure similar to ( [ matrixm ] ) , and while @xmath169 are not exactly inverse to @xmath96 , the product @xmath170 is a rather simple matrix with a trivial determinant , as we shall see in the moment .
we denote the product of @xmath96 and @xmath169 by @xmath171 : the indices of @xmath173 naturally decompose in two sets , @xmath174 and @xmath175 with @xmath176 , as in ( [ matrixm ] ) and ( [ matrixn ] ) .
the matrix @xmath177 therefore consists of four blocks , @xmath173 , @xmath178 , @xmath179 , and @xmath180 , where the indices take values @xmath181 and @xmath182 .
we use the shorthand notation for the @xmath106-th index of @xmath177 by simply omitting @xmath183 in the label .
the key observation is that @xmath184 is zero . indeed , from ( [ matrixm ] ) , ( [ matrixn ] ) , where the contour of integration encircles the set of points @xmath186 counterclockwise . but
the integrand has no singularities outside the contour of integration .
in particular the pole infinity vanishes because the integral behaves as @xmath187 , and @xmath188 is always bigger than one .
therefore , @xmath189 and consequently the matrix @xmath171 , in the block form , is lower triangular , where the last equality is obtained by inflating the contour of integration and computing the residues at @xmath192 and @xmath193 .
the latter residue vanishes unless @xmath194 . for the @xmath195 components , we get 100 b. pozsgay , _ overlaps between eigenstates of the xxz spin-1/2 chain and a class of simple product states _ , ` http://arxiv.org/abs/1309.4593 ` o. tsuchiya , _ determinant formula for the six - vertex model with reflecting end _ , j. math .
* 39 * ( 1998 ) 135 , ` http://arxiv.org/abs/solv-int/9804010 ` m. brockmann , j. denardis , b. wouters and j .- s .
caux , _ a gaudin - like determinant for overlaps of nel and xxz bethe states _ , j. phys . * a 47 * ( 2014 ) 145003 , ` http://arxiv.org/abs/1401.2877 ` m. brockmann , j. de nardis , b. wouters and j .- s .
caux , _ nel - xxz state overlaps : odd particle numbers and lieb - liniger scaling limit _ , j. phys . * a 47 * ( 2014 ) 345003 , ` http://arxiv.org/abs/1403.7469 ` m. brockmann , _ overlaps of @xmath204-raised nel states with xxz bethe states and their relation to the lieb - liniger bose gas _ , j. stat . mech . * 5 * ( 2014 ) p05006 , ` http://arxiv.org/abs/1402.1471 ` m. brockmann , b. wouters , d. fioretto , j. de nardis , r. vlijm and j .- s .
caux , _ quench action approach for releasing the nel state into the spin-1/2 xxz chain _ , j. stat . mech .
* 12 * ( 2014 ) 9 , ` http://arxiv.org/abs/1408.5075 ` p. mazza , j. stphan , e. canovi , v. alba , m. brockmann and m. haque , _ overlap distributions for quantum quenches in the anisotropic heisenberg chain _
, ` http://arxiv.org/abs/1509.04666 ` m. de leeuw , c. kristjansen and k. zarembo , _ one - point functions in defect cft and integrability _ , jhep * 1508 * ( 2015 ) 098 , ` http://arxiv.org/abs/1506.06958 ` g. kuperberg , _ symmetry classes of alternating - sign matrices under one roof _ ,
ann . math .
* 156 * ( 2002 ) 835 , ` http://arxiv.org/abs/math/0008184 ` o. foda and m. wheeler , _ partial domain wall partition functions _ ,
* 1207 * ( 2012 ) 186 , ` http://arxiv.org/abs/1205.4400 ` v. korepin , _ calculation of norms of bethe wave functions _ , commun . math . phys .
* 86 * ( 1982 ) 391 .
a. izergin , _ partition function of the six - vertex model in a finite volume _ ,
* 32 * ( 1987 ) 878 .
r. j. baxter , _ partition function of the eight vertex lattice model
_ , annals phys .
* 70 * ( 1972 ) 193 .
v. e. korepin , n. m. bogolyubov and a. g. izergin , _ quantum inverse scattering method and correlation functions _ , cambridge univversity press ( 1993 ) .
i. v. cherednik , _ factorizing particles on a half line and root systems _ , theor .
* 61 * ( 1984 ) 977 .
e. k. sklyanin , _ boundary conditions for integrable quantum systems _ , j. phys .
* a21 * ( 1988 ) 2375 .
h. j. de vega and a. gonzalez - ruiz , _ boundary k matrices for the xyz , xxz and xxx spin chains _ , j. phys . * a27 * ( 1994 ) 6129 , ` http://arxiv.org/abs/hep-th/9306089 ` l. piroli and p. calabrese , _ recursive formulas for the overlaps between bethe states and product states in xxz heisenberg chains _ , j. phys . * a 47 * ( 2014 ) 385003 , ` http://arxiv.org/abs/1407.2242 ` a. izergin , n. kitanine , j. maillet and v. terras , _ spontaneous magnetization of the xxz heisenberg spin-1/2 chain _ , nucl
* b 554 * ( 1999 ) 679 , ` http://arxiv.org/abs/solv-int/9812021 ` n. kitanine , k. k. kozlowski , j. m. maillet , n. a. slavnov and v. terras , _ algebraic bethe ansatz approach to the asymptotic behavior of correlation functions _ , j. stat .
mech . * 0904 * ( 2009 ) p04003 , ` http://arxiv.org/abs/0808.0227 ` k. k. kozlowski and b. pozsgay , _ surface free energy of the open xxz spin-1/2 chain _ , j. stat .
* 1205 * ( 2012 ) 5021 , ` http://arxiv.org/abs/1201.5884 ` m. gaudin , _
diagonalisation dune classe dhamiltoniens de spin _ , j. phys .
france * 37 * ( 1976 ) 1087 .
i. buhl - mortensen , m. de leeuw , c. kristjansen and k. zarembo , _ one - point functions in ads / dcft from matrix product states _ , ` http://arxiv.org/abs/1512.02532 ` | partial nel states are generalizations of the ordinary nel ( classical anti - ferromagnet ) state that can have arbitrary integer spin
. we study overlaps of these states with bethe states .
we first identify this overlap with a partial version of reflecting - boundary domain - wall partition function , and then derive various determinant representations for off - shell and on - shell bethe states . `
nordita-2015 - 133 ` + ` uuitp-27/15 ` |
let @xmath0 be a @xmath1 random vector and @xmath2 be a random variable .
then , @xmath0 and @xmath3 are said to follow a _ normal - gamma distribution _
( ng distribution ) , if their joint probability density function is given by @xmath4 where @xmath5 denotes a multivariate normal density with mean @xmath6 and covariance @xmath7 and @xmath8 denotes a gamma density with shape @xmath9 and rate @xmath10 . in full , the density function is given by ( koch , 2007 , p. 55 ) @xmath11 \cdot \frac{{b}^{a}}{\gamma(a ) } \ , y^{a-1 } \ , \exp[-b y ] \ ; .\ ] ] the _ kullback - leibler divergence _
( kl divergence ) is a non - symmetric distance measure for two probability distributions @xmath12 and @xmath13 and is defined as @xmath14 = \sum_{i \in \omega } p(i ) \ , \ln \frac{p(i)}{q(i ) } \ ; .\ ] ] for continuous probability distribtions @xmath12 and @xmath13 with probability density functions @xmath15 and @xmath16 on the same domain @xmath17 , it is given by ( bishop , 2006 , p. 55 ) @xmath18 = \int_{x } p(x ) \
, \ln \frac{p(x)}{q(x ) } \ , \mathrm{d}x \ ; .\ ] ] the kl divergence becomes important in information theory and statistical inference .
here , we derive the kl divergence for two ng distributions with vector - valued @xmath0 and real - positive @xmath3 and provide two examples of its application .
first , consider two multivariate normal distributions over the @xmath1 vector @xmath0 specified by @xmath19 according to equation ( [ eq : cont - kl ] ) , the kl divergence of @xmath12 from @xmath13 is defined as @xmath20 = \int_{\mathbb{r}^k } \mathrm{n}(x ; \mu_1 , \sigma_1 ) \ , \ln \frac{\mathrm{n}(x ; \mu_1 , \sigma_1)}{\mathrm{n}(x ; \mu_2 , \sigma_2 ) } \ , \mathrm{d}x \ ; .\ ] ] using the multivariate normal density function @xmath21 \ ; , \ ] ] it evaluates to ( duchi , 2014 ) @xmath22 = \frac{1}{2 } \left [ ( \mu_2 - \mu_1)^t \sigma_2^{-1 } ( \mu_2 - \mu_1 ) + \mathrm{tr}(\sigma_2^{-1 } \sigma_1 ) - \ln \frac{|\sigma_1|}{|\sigma_2| } - k \right ] \ ; .\ ] ] next , consider two univariate gamma distributions over the real - positive @xmath3 specified by @xmath23 according to equation ( [ eq : cont - kl ] ) , the kl divergence of @xmath12 from @xmath13 is defined as @xmath20 = \int_{0}^{\infty } \mathrm{gam}(y ; a_1 , b_1 ) \ , \ln \frac{\mathrm{gam}(y ; a_1 , b_1)}{\mathrm{gam}(y ; a_2 , b_2 ) } \ , \mathrm{d}y \ ; .\ ] ] using the univariate gamma density function @xmath24 \quad \text{for } \quad y > 0 \ ; , \ ] ] it evaluates to ( penny , 2001 ) @xmath25 = a_2 \ , \ln \frac{b_1}{b_2 } - \ln \frac{\gamma(a_1)}{\gamma(a_2 ) } + ( a_1 - a_2 ) \ ,
\psi(a_1 ) - ( b_1 - b_2 ) \ , \frac{a_1}{b_1}\ ] ] where @xmath26 is the digamma function . now , consider two normal - gamma distributions over @xmath0 and @xmath3 specified by @xmath27 according to equation ( [ eq : cont - kl ] ) , the kl divergence of @xmath12 from @xmath13 is defined as @xmath28 = \int_{0}^{\infty } \int_{\mathbb{r}^k } p(x , y ) \
, \ln \frac{p(x , y)}{q(x , y ) } \ , \mathrm{d}x \ , \mathrm{d}y \ ; .\ ] ] using the law of conditional probability , it can be evaluated as follows : @xmath29 & = \int_{0}^{\infty } \int_{\mathbb{r}^k } p(x|y ) p(y ) \ , \ln \frac{p(x|y ) p(y)}{q(x|y ) q(y ) } \ , \mathrm{d}x \ , \mathrm{d}y \\ & = \int_{0}^{\infty } p(y ) \int_{\mathbb{r}^k } p(x|y ) \ , \ln \frac{p(x|y)}{q(x|y ) } \ , \mathrm{d}x \ , \mathrm{d}y \\ & + \int_{0}^{\infty } p(y ) \ , \ln \frac{p(y)}{q(y ) } \int_{\mathbb{r}^k } p(x|y ) \ , \mathrm{d}x \ , \mathrm{d}y \\ & = \left\langle \mathrm{kl}[p(x|y)||q(x|y ) ] \right\rangle_{p(y ) } + \mathrm{kl}[p(y)||q(y ) ] \end{split}\ ] ] in other words , the kl divergence for two normal - gamma distributions over @xmath0 and @xmath3 is equal to the sum of a multivariate normal kl divergence regarding @xmath0 conditional on @xmath3 , expected over @xmath3 , and a univariate gamma kl divergence regarding @xmath3 .
together with equation ( [ eq : mvn - kl ] ) , the first term becomes @xmath30 \right\rangle_{p(y ) } \\ & = \left\langle \frac{1}{2 } \left [ ( \mu_2 - \mu_1)^t ( y \lambda_2 ) ( \mu_2 - \mu_1 ) + \mathrm{tr}\left ( ( y \lambda_2 ) ( y \lambda_1)^{-1 } \right ) - \ln \frac{|(y \lambda_1)^{-1}|}{|(y \lambda_2)^{-1}| } - k \right ] \right\rangle_{p(y ) } \\ & = \left\langle \frac{y}{2 } ( \mu_2 - \mu_1)^t \lambda_2 ( \mu_2 - \mu_1 ) + \frac{1}{2 } \ , \mathrm{tr}(\lambda_2 \lambda_1^{-1 } ) - \frac{1}{2 } \ln \frac{|\lambda_2|}{|\lambda_1| } - \frac{k}{2 } \right\rangle_{p(y ) } \ ; .
\end{split}\ ] ] using the relation @xmath31 , we have @xmath32 \right\rangle_{p(y ) } = \frac{1}{2 } \frac{a_1}{b_1 } ( \mu_2 - \mu_1)^t \lambda_2 ( \mu_2 - \mu_1 ) + \frac{1}{2 } \
, \mathrm{tr}(\lambda_2 \lambda_1^{-1 } ) - \frac{1}{2 } \ln \frac{|\lambda_2|}{|\lambda_1| } - \frac{k}{2 } \ ; . \end{split}\ ] ] thus , from ( [ eq : exp - mvn - kl ] ) and ( [ eq : gam - kl ] ) , the kl divergence in ( [ eq : ng - kl1 ] ) becomes @xmath33 & = \frac{1}{2 } \frac{a_1}{b_1 } \left [ ( \mu_2 - \mu_1)^t \lambda_2 ( \mu_2 - \mu_1 ) \right ] + \frac{1}{2 } \ , \mathrm{tr}(\lambda_2 \lambda_1^{-1 } ) - \frac{1}{2 } \ln \frac{|\lambda_2|}{|\lambda_1| } - \frac{k}{2 } \\ & + a_2 \ , \ln \frac{b_1}{b_2 } - \ln \frac{\gamma(a_1)}{\gamma(a_2 ) } + ( a_1 - a_2 ) \ ,
\psi(a_1 ) - ( b_1 - b_2 ) \ , \frac{a_1}{b_1 } \ ; . \end{split}\ ] ] consider bayesian inference on data @xmath3 using model @xmath34 with parameters @xmath35 . in this case ,
bayes theorem is a statement about the posterior density : @xmath36 the denominator @xmath37 acts as a normalization constant on the posterior density @xmath38 and according to the law of marginal probability is given by @xmath39 this is the probability of the data given only the model , regardless of any particular parameter values .
it is also called `` marginal likelihood '' or `` model evidence '' and can act as a model quality criterion in bayesian inference , because parameters are integrated out of the likelihood . for computational reasons ,
only the logarithmized or log model evidence ( lme ) @xmath40 is of interest in most cases . by rearranging equation ( [ eq : bt ] )
, the model evidence can be represented as @xmath41 logarithmizing both sides of the equation and taking the expectation with respect to the posterior density over model parameters @xmath35 gives the lme @xmath42 using this reformulation , the lme as a model quality measure can be naturally decomposed into an accuracy term , the posterior expected likelihood , and a complexity term , the kl divergence between the posterior and the prior distribution : @xmath43 \end{split}\ ] ] intuitively , the accuracy acts increasing and the complexity acts decreasing on the log model evidence .
this reflects the capability of the lme to select models that achieve the best balance between accuracy and complexity , i.e. models that explain the observations sufficiently well ( high accuracy ) without employing too many principles ( low complexity ) .
the fact that the complexity term is a kl divergence between posterior and prior means that models with prior assumptions that are close to the posterior evidence receive a low complexity penalty , because one is not surprised very much when accepting such a model which renders the bayesian complexity a measure of surprise . consider multiple linear regression using the univariate general linear model ( glm ) @xmath44 where @xmath3 is an @xmath45 vector of measured data , @xmath17 is an @xmath46 matrix called the design matrix , @xmath47 is a @xmath48 vector of weight parameters called regression coefficients and @xmath49 is an @xmath45 vector of errors or noise .
these residuals are assumed to follow a multivariate normal distribution whose covariance matrix is the product of a variance factor @xmath50 and an @xmath51 correlation matrix @xmath52 .
usually , @xmath17 and @xmath52 are known while @xmath47 and @xmath53 are unknown parameters to be inferred via model estimation . for mathematical convenience ,
we rewrite @xmath54 and @xmath55 so that equation ( [ eq : glm ] ) implies the following likelihood function : @xmath56 the conjugate prior relative to this likelihood function is a normal - gamma distribution on the model parameters @xmath47 and @xmath53 ( koch , 2007 , ch .
2.6.3 ) : @xmath57 due to the conjugacy of ( [ eq : glm - ng - prior ] ) to ( [ eq : glm - lf ] ) , the posterior is also a normal - gamma distribution @xmath58 where the posterior parameters in ( [ eq : glm - ng - post ] ) are given by ( koch , 2007 , ch .
4.3.2 ) @xmath59 from ( [ eq : lme2 ] ) , the complexity for the model defined by ( [ eq : glm - lf ] ) and ( [ eq : glm - ng - prior ] ) is given by @xmath60 \ ; .\ ] ] in other words , the complexity penalty for a general linear model with normal - gamma priors ( glm - ng ) is identical to a kl divergence between two ng distributions and using ( [ eq : ng - kl2 ] ) can be written in terms of the prior and posterior parameters as @xmath61 + \frac{1}{2 } \
, \mathrm{tr}(\lambda_0 \lambda_n^{-1 } ) - \frac{1}{2 } \ln \frac{|\lambda_0|}{|\lambda_n| } - \frac{p}{2 } \\ & + a_0 \ , \ln \frac{b_n}{b_0 } - \ln \frac{\gamma(a_n)}{\gamma(a_0 ) } + ( a_n - a_0 ) \ , \psi(a_n ) - ( b_n - b_0 ) \
, \frac{a_n}{b_n } \ ; . \end{split}\ ] ]
consider a linear model with polynomial basis functions ( bishop , 2006 , p. 5 ) given by @xmath62 essentially , this model assumes that @xmath3 is an additive mixture of polynomial terms @xmath63 weighted with the coefficients @xmath64 with @xmath65 where the natural number @xmath66 is called the model order .
this means that @xmath67 corresponds to a constant value ( plus noise @xmath49 ) ; @xmath68 corresponds to a linear function ; @xmath69 corresponds to a quadratic pattern ; @xmath70 corresponds to a 3rd degree polynomial etc .
given that @xmath0 is an @xmath45 vector of real numbers between @xmath71 and @xmath72 , this model can be rewritten as a glm given in equation ( [ eq : glm ] ) with @xmath73 based on this reformulation , we simulate polynomial data .
we perform @xmath74 simulations with @xmath75 data points in each simulation .
we generate simulated data based on a true model order @xmath76 and analyze these data using a set of models ranging from @xmath77 to @xmath78 .
the predictor @xmath0 is equally spaced between @xmath71 and @xmath72 and design matrices @xmath79 are created according to equation ( [ eq : pbf2 ] ) . in each simulation , six regression coefficients @xmath80 are drawn independently from the standard normal distribution @xmath81 .
then , gaussian noise @xmath49 is sampled from the multivariate normal distribution @xmath82 with a residual variance of @xmath83 .
finally , simulated data is generated as as @xmath84 .
then , for each @xmath85 , bayesian model estimation is performed using the design matrix @xmath79 , a correlation matrix @xmath86 and the prior distributions ( [ eq : glm - ng - prior ] ) with the prior parameters @xmath87 , @xmath88 invoking a standard multivariate normal distribution and @xmath89 , @xmath90 invoking a relatively flat gamma prior .
posterior parameters are calculated using equation ( [ eq : glm - ng - post - par ] ) and give rise to the model complexity via ( [ eq : glm - ng - com2 ] ) as well as model accuracy and the log model evidence via ( [ eq : lme2 ] ) .
average lmes , accuracies and complexities are shown in figure 1 .
one can see that the true model order is correctly identified by the maximal log model evidence .
this is achieved by an increasing complexity penalty which outweighs the saturating accuracy gain for models with @xmath91 .
this demonstrates that the kl divergence for the ng distribution can be used to select polynomial basis functions when basis sets can not be separated based on model accuracy alone .
[ fig : figure_pbf ] * figure 1 . * bayesian model selection for polynomial basis functions .
all displays have model order on the x - axis and average model quality measures ( across simulations ) on the y - axis .
intuitively , the model accuracy ( middle panel ) increases with model order , but saturates at around @xmath92 with no major increase after @xmath93 .
moreover , the model complexity ( lower panel ) which is the kl divergence between prior and posterior distribution also grows with model order , but switches to a linear increase at around @xmath93 reaching a value of @xmath94 at @xmath95 .
together , this has the consequence that the log model evidence ( upper panel ) is maximal for @xmath93 ( black cross ) where exact values are : @xmath96 . in neuroimaging , especially functional magnetic resonance imaging ( fmri ) , glms as given by equation ( [ eq : glm ] )
are applied to time series of neural data @xmath3 ( friston et al . , 1995 ) .
the design matrix @xmath17 is specified by the temporal occurrence of experimental conditions and the covariance structure @xmath52 is estimated from residual auto - correlations .
model estimation and statistical inference are performed `` voxel - wise '' , i.e. separately for each measurement location in the brain , usually referred to as the `` mass - univariate glm '' . here
, we analyze data from a study on orientation pop - out processing ( bogler et al . , 2013 ) .
during the experimental paradigm , the screen showed a @xmath97 array of homogeneous bars oriented either 0 , 45 , 90 or 135 relative to the vertical axis .
this background stimulation changed every second and was interrupted by trials in which one target bar on the left and one target bar on the right were independently rotated either 0 , 30 , 60 or 90 relative to the rest of the stimulus display .
those trials of orientation contrast ( oc ) lasted 4 seconds and were alternated with inter - trial intervals of 7 , 10 or 13 seconds .
each combination of oc on the left side and oc on the right side was presented three times resulting in 48 trials in each of the 5 sessions lasting 672 seconds .
after fmri data preprocessing ( slice - timing , realignment , normalization , smoothing ) , two different models of hemodynamic activation were applied to the fmri data . the first model ( glm i ) considers the experiment a factorial design with two factors ( left vs. right oc ) having four levels ( 0 , 30 , 60 , 90 ) .
this results in @xmath98 possible combinations or experimental conditions modelled by @xmath99 onset regressors convolved with the canonical hemodynamic response function ( hrf ) . the second model ( glm ii )
puts all trials from all conditions into one hrf - convolved regressor and encodes orientation contrast using a parametric modulator ( pm ) that is given as @xmath100 with @xmath101 , resulting in @xmath102 , such that the parametric modulator is proportional to orientation contrast .
there was one pm for each factor of the design , i.e. one pm for left oc and one pm for right
note that both models encode the same information and that every signal that can be identified using glm ii can also be characterized using glm i , but not vice versa , because the first model allows for a greater flexibility of activation patterns across experimental conditions than the second .
for these two models , we performed bayesian model estimation . to overcome the challenge of having to specify prior distributions on the model parameters , we apply cross - validation across fmri sessions .
this gives rise to a cross - validated log model evidence ( cvlme ) as well as cross - validated accuracies and complexities for each model in each subject .
we then performed a paired t - test to find voxels where glm ii has a significantly higher cvlme than glm i. due to the specific assumptions in glm ii and the higher flexibility of glm i , we assumed that these differences might be primarily based on a complexity advantage of glm ii over glm i. we focus on visual area 4 ( v4 ) that is known to be sensitive to orientation contrast . within left v4 , specified by a mask from a separate localizer paradigm ( bogler et al . ,
2013 ) , we identified the peak voxel ( @xmath103 = [ -15 , -73 , -5]$ ] mm ) defined by the maximal t - value ( @xmath104 ) and extracted log model evidence as well as model accuracy and model complexity from this voxel for each subject .
differences in lme , accuracy and complexity are shown in figure 2 .
again , the model complexity enables a model selection that would not be possible based on the model accuracy alone .
[ fig : figure_nms ] * figure 2 . * bayesian model selection for orientation pop - out processing .
all displays have subject on the x - axis and difference in model qualities ( @xmath105 = glm i , @xmath106 = glm ii ) on the y - axis .
interestingly , there is a slight disadvantage for glm ii regarding only the model accuracy ( middle panel ) , its mean difference across subjects being smaller than zero .
however , model complexity ( lower panel ) measured as the kl divergence between prior and posterior distribution is consistently higher for glm i. together , this has the consequence that the log model evidence ( upper panel ) most often favors glm ii .
average values are : @xmath107 .
we have derived the kullback - leibler divergence of two normal - gamma distributions using earlier results on the kl divergence for multivariate normal and univariate gamma distributions . moreover , we have shown that the kl divergence for the ng distribution occurs as the complexity term in the univariate general linear model when using conjugate priors .
analysis of simulated and empirical data demonstrates that the complexity penalty has the desired theoretical features , namely to quantify the relative informational content of two generative models and to detect model differences that can not be detected by just relying on model accuracy , e.g. given by the maximum log - likelihood ( as in information criteria like aic or bic ) or the posterior log - likelihood ( as in the bayesian log model evidence ) .
9 friston kj , holmes ap , worsley kj , poline jp , frith cd , frackowiak rsj ( 1995 ) : `` statistical parametric maps in functional imaging : a general linear approach '' .
_ human brain mapping _ , vol . 2 , iss . 4 , pp . 189 - 210 | we derive the kullback - leibler divergence for the normal - gamma distribution and show that it is identical to the bayesian complexity penalty for the univariate general linear model with conjugate priors . based on this finding
, we provide two applications of the kl divergence , one in simulated and one in empirical data .
kullback - leibler divergence + for the normal - gamma distribution + joram soch^1,3,^ & carsten allefeld^1,2^ ^1^ bernstein center for computational neuroscience , berlin , germany + ^2^ berlin center for advanced neuroimaging , berlin , germany + ^3^ department of psychology , humboldt - universitt zu berlin , germany + corresponding author : [email protected] . |
in the last decade , there have been a number of studies of systems in which the states of individuals and the connections between them coevolve , see @xcite .
the systems considered include evolutionary games @xcite@xcite and epidemics @xcite@xcite , but here we will concentrate on the spread of opinions @xcite@xcite .
different from the models of cascades @xcite@xcite which are also widely used in the study of opinion spread , the evolving voter model we study here allows an agent to switch between different opinions and the network topology to change accordingly , yet we assume that agents impose equal influence over each other ( cf . , multi - state complex contagions @xcite@xcite ) .
this model provides building blocks to quantitatively study collective behaviors in various social systems , e.g. , segregation of a population into two or more communities with different political opinions , religious beliefs , cultural traits , etc .
we are particularly interested here in systems that generalize the model proposed by holme and newman @xcite . in their model
there is a network of @xmath5 vertices and @xmath6 edges .
the individual at vertex @xmath7 has an opinion @xmath8 from a set of @xmath9 possible opinions and the number of people per opinion @xmath10 stays bounded as @xmath5 gets large . on each step of the process
, a vertex @xmath0 is picked at random .
if its degree @xmath11 equals @xmath12 , nothing happens .
if @xmath13 , ( i ) then with probability @xmath14 a random neighbor @xmath1 of @xmath0 is selected and we set @xmath15 ; ( ii ) otherwise ( i.e. , with probability @xmath3 ) an edge attached to vertex @xmath0 is selected and the other end of that edge is moved to a vertex chosen at random from those with opinion @xmath16 . this process continues until the ` consensus time ' @xmath17 , at which there are no longer any discordant edges that is , there are no edges connecting individuals with different opinions . for @xmath18 ,
only rewiring steps occur , so once all of the @xmath6 edges have been touched , the graph has been disconnected into @xmath9 components , each consisting of individuals who share the same opinion . since
none of the opinions have changed , the components of the final graph are all small ( i.e. , their sizes are poisson with mean @xmath19 ) . by classical results for the coupon collector s problem
, this requires @xmath20 updates , see e.g. , page 57 in @xcite . in the case of sparse graphs we consider here @xmath21 ( i.e. , @xmath22 ) so the number of steps is @xmath23 , i.e. , when @xmath5 is large it will be @xmath24 .
in contrast , for @xmath25 this system reduces to the voter model on a static graph .
if we suppose that the initial graph is an erds - rnyi random graph in which each vertex has average degree @xmath26 , then ( see e.g. , chapter 2 of @xcite ) there is a `` giant component '' that contains a positive fraction , @xmath27 , of the vertices and the second largest component is small having only @xmath28 vertices
. the voter model on the giant component will reach consensus in @xmath29 steps ( see , e.g. , section 6.9 of @xcite ) , so the end result is that one opinion has @xmath30 followers while all of the other groups are small . using simulation and finite size scaling , holme and
newman showed that there is a critical value @xmath31 so that for @xmath32 all of the opinions have a small number of followers at the end of the process , while for @xmath33 `` a giant community of like - minded individuals forms . ''
when the average degree @xmath34 and the number of individuals per opinion @xmath35 , this transition occurs at @xmath36 .
see @xcite@xcite for recent work on this model . in @xcite
, we studied a two - opinion version of this model in which on each step an edge is chosen at random and is given a random orientation , @xmath37 .
if the individuals at the two ends have the same opinion nothing happens . if they differ , then ( i ) with probability @xmath14 we set @xmath15 ; ( ii ) otherwise ( i.e. , with probability @xmath3 ) @xmath0 breaks its edge to @xmath1 and reconnects to ( a ) a vertex chosen at random from those with opinion @xmath16 , a process we label ` rewire - to - same ' , or ( b ) at random from the graph , a process we label ` rewire - to - random ' . here
, we will concentrate on the second rewiring option , rewire - to - random . while this process may be less intuitive than the rewire - to - same version , it has a more interesting phrase - transition , as documented in @xcite .
the remainder of this paper is organized as follows . in section [ sec:2 ]
, we recall the main results from @xcite that provide essential context for our observations of the multiple - opinion case , which we begin to explore in section [ sec : multi ] .
we then continue in section [ sec : quant ] with further quantitative details about the phase transitions and their underlying quasi - stationary distributions , before concluding comments in section [ sec : conclusion ] .
suppose , for concreteness , that the initial social network is an erds - rnyi random graph in which each individual has average degree @xmath38 , and that vertices are assigned opinions 1 and 0 independently with probabilities @xmath39 and @xmath40 .
simulations suggest that the system has the following to help understand the last statement , the reader should consult the picture in figure [ fig : rewire_to_random_p1 ] .
if the initial fraction of 1 s @xmath41 then as @xmath3 decreases from 1 , the ending density @xmath42 stays constant at 1/2 until @xmath43 and then decreases to a value close to 0 at @xmath25 . for convenience
, we call the graph of @xmath42 for @xmath44 , the _
universal curve_. if the initial density is @xmath45 , then the ending density @xmath46 stays constant at @xmath39 until the flat line @xmath47 hits the universal curve and then @xmath48 for @xmath49 .
the main aim of @xcite was to use simulations , heuristic arguments , and approximate models to explain the presence and properties of this universal curve describing the consensus states that result from the slow - consensus process . to make it easier to compare the results here with the previous paper , we rescale time so that times between updating steps are exponential with rate @xmath6 , where @xmath6 is the total number of edges .
the final fractions of the minority below phase transitions follow an universal curve independent of the initial fractions.,scaledwidth=45.0% ] to further explain the phrase `` quasi - stationary distributions '' in this context , we refer the reader to figure [ fig : n1n010v3 ] .
let @xmath50 be the number of vertices in state 1 at time @xmath51 , @xmath52 be the number of @xmath12-@xmath53 edges ( that is , the number of edges connecting nodes @xmath0 and @xmath1 with @xmath54 , @xmath55 ) .
similarly , let @xmath56 be the number of connected triples @xmath0-@xmath1-@xmath57 with @xmath58 , @xmath59 , and @xmath60 .
the top panel of figure [ fig : n1n010v3 ] plots @xmath61 versus @xmath62 for five different simulations ( with different initial densities , @xmath39 ) for @xmath63 . note that in each case the simulation rapidly approaches a curve @xmath64 and then diffuses along the curve until consensus is reached ( @xmath65 ) . at both of the possible consensus points on the curve ,
the fraction of the minority opinion is @xmath66 , in accordance with the simulation in figure [ fig : rewire_to_random_p1 ] . the bottom panel of figure [ fig : n1n010v3 ] similarly plots @xmath67 versus @xmath62 for @xmath63 and @xmath41 .
again the simulation rapidly approaches a curve ( approximately cubic ) and diffuses along it until consensus is reached .
since @xmath68 if @xmath65 , and it is very unlikely that all @xmath12-@xmath53 s only occur in @xmath12-@xmath53-@xmath53 triples , the zeros of the cubic curve for @xmath12-@xmath53-@xmath12 and quadratic curve for @xmath12-@xmath53 coincide .
-@xmath53 edges , @xmath61 , versus the population of opinions @xmath62 when @xmath63 for the rewire - to - random dynamic .
five simulations starting from @xmath39=0.2 , 0.35 , 0.5 , 0.65 , and 0.8 are plotted in different colors .
each simulation starts from an erds - rnyi graph with n=100,000 nodes and average degree @xmath69 .
after initial transients , the fraction of discordant edges behaves as a function of the population of opinions .
( bottom ) similarly , the number of @xmath12-@xmath53-@xmath12 connected triples behaves as a function of @xmath70 after an initial transient ( one simulation).,title="fig:",scaledwidth=50.0% ] -@xmath53 edges , @xmath61 , versus the population of opinions @xmath62 when @xmath63 for the rewire - to - random dynamic .
five simulations starting from @xmath39=0.2 , 0.35 , 0.5 , 0.65 , and 0.8 are plotted in different colors .
each simulation starts from an erds - rnyi graph with n=100,000 nodes and average degree @xmath69 .
after initial transients , the fraction of discordant edges behaves as a function of the population of opinions .
( bottom ) similarly , the number of @xmath12-@xmath53-@xmath12 connected triples behaves as a function of @xmath70 after an initial transient ( one simulation).,title="fig:",scaledwidth=50.0% ] one can repeat the simulations in figure [ fig : n1n010v3 ] for other network measurements , with the result that their values are similarly determined by the density @xmath71 .
this is somewhat analogous to a stationary distribution from equilibrium statistical mechanics e.g . , the maxwell - boltzmann distribution associating the velocity distribution with the temperature .
we call our distributions quasi - stationary because our system is a finite state markov chain , which will eventually reach one of its many absorbing states @xmath65 , and hence there is no true stationary distribution .
nevertheless , an improved understanding of the system is obtained from these observations , displaying a fast dynamics rapidly converging to a family of neutrally - stable quasi - stationary distributions followed by slow , diffusive dynamics through the space local to the quasi - stationary distributions until consensus is reached . to begin to explain the behavior of @xmath72 given in ( [ wf ] ) , note that when an edge is picked with two endpoints that differ , a rewiring will not change the number of 1 s , while a voting event , which occurs with probability @xmath2 , will result in an increase or decrease of the number of 1 s with equal probability . when @xmath73 the rate at which @xmath12-@xmath53 edges are chosen is equal to the expected fraction of @xmath12-@xmath53 edges under @xmath74 , which is @xmath75 .
as shown in @xcite , the behaviors for the rewire - to - same model in terms of quasi - stationary distributions are very similar , but with small differences from the rewire - to - random model that yield fundamentally different consensus states . in rewire - to -
same , there are quasi - stationary distributions @xmath76 under which the expected fraction of @xmath12-@xmath53 edges is @xmath77 .
again the simulation comes rapidly to this curve and diffuses along it until consensus is reached .
that is , unlike figure [ fig : n1n010v3 ] ( top ) , the arches of quasi - stationary @xmath78 values versus @xmath70 maintain their zeros at @xmath79 .
thus , for @xmath80 , the minority fraction obtained at the consensus time is always @xmath81 for rewire - to - same .
bhme and gross @xcite have studied the three - opinion version of the evolving voter model with rewire - to - same dynamics . in this case , the limiting behavior is complicated
one may have partial fragmentation ( 1 s split off rapidly from the 2 s and 3 s ) in addition to full fragmentation and coexistence of the three opinions .
see their figures 35 . as we will see in the present section ,
the behavior of the multi - opinion rewire - to - random model is much simpler because small groups of individuals with the same opinion will be drawn back into the giant component .
we thus aim to extend the understanding of the two - opinion model behavior to larger numbers of opinions .
consider now the @xmath82-opinion model in which voters are assigned independent initial opinions that are equal to @xmath83 with probability @xmath84 .
let @xmath85 and let @xmath86 be the number of edges at which the endpoint opinions differ .
when @xmath87 , frequencies of the three types must lie in the triangle of possible values @xmath88 . to preserve symmetry , we draw @xmath89 as an equilateral triangle in barycentric coordinates by mapping @xmath90 .
the top panel in figure [ fig : levelset2 ] plots @xmath91 as a function of the opinion densities as the system evolves , generalizing the one - dimensional arch observed for @xmath92 to a two - dimensional cap for @xmath87 .
generalizing the parabolic form of the arch for @xmath92 , we conjecture e_u n_/m = ( 1- _ i=1^k u_i^2 ) - c_0 ( ) .
[ quadcap ] as in the two opinion case , the simulated values come quickly to the surface and then diffuse along it . in some situations ,
one opinion is lost before consensus occurs and the evolution reduces to that for the two opinion case .
however , in one of the simulations shown , the realization ending with @xmath93 , all three opinions persist until the end . .
multiple simulations corresponding to different initial densities are shown while each one starts from an erds - rnyi graph with n=10,000 nodes and average degree @xmath69 .
similar to the two - opinion case , the simulations quickly converge to a parabolic cap of quasi - stationary distributions .
bottom : top view of the parabolic caps of quasi - stationary distributions for @xmath3=0.1,0.2, ... ,0.8 .
we fit the parabolic cap eq .
( [ quadcap ] ) to simulation data at various @xmath3 s and then plot the level sets @xmath94 , which are the intersections of the parabolic caps with the @xmath95 plane , as the large circles with colors indicating values of @xmath3.,title="fig:",scaledwidth=50.0% ] .
multiple simulations corresponding to different initial densities are shown while each one starts from an erds - rnyi graph with n=10,000 nodes and average degree @xmath69 .
similar to the two - opinion case , the simulations quickly converge to a parabolic cap of quasi - stationary distributions .
bottom : top view of the parabolic caps of quasi - stationary distributions for @xmath3=0.1,0.2, ... ,0.8 .
we fit the parabolic cap eq .
( [ quadcap ] ) to simulation data at various @xmath3 s and then plot the level sets @xmath94 , which are the intersections of the parabolic caps with the @xmath95 plane , as the large circles with colors indicating values of @xmath3.,title="fig:",scaledwidth=50.0% ] the picture is somewhat easier to understand if we look at the cap from a top view , where the @xmath96 level sets for different @xmath3 are observed to be circles . in the bottom panel of figure [ fig : levelset2 ]
we plot the @xmath96 circles for different @xmath3 s fitted from simulation data using eq .
( [ quadcap ] ) as well as the consensus opinion frequencies from the simulations ( indicated by small circle data points ) .
the two agree with each other up to small stochastic fluctuations .
the size of the @xmath96 level set then dictates different consensus state properties .
for example , the circle corresponding to @xmath63 intersects @xmath89 in three disconnected arcs . as @xmath3 increases , the radius of the @xmath96 level set decreases .
when @xmath97 , the critical value of the two opinion model , the circle @xmath96 falls fully inside the triangle , so an initial condition including all three opinions will continue to demonstrate all three opinions at consensus .
for example , the small circles around the innermost circle give the ending frequencies for several simulations for @xmath98 .
if the initial frequencies fall within the @xmath96 circle , then the model will quickly relax to the quasi - stationary distributions above the circle and then diffuse along the cap until consensus is reached at some @xmath96 point . if instead the initial frequencies @xmath39 fall outside the @xmath96 circle that is , for @xmath3 above the phase transition point @xmath99the consensus time jumps from @xmath29 to @xmath100 , similar to @xmath101 for the two - opinion model , with the final opinion frequencies essentially the same as the initial @xmath39 .
what is new in this case is that when starting with three opinions and @xmath102 , the system always ends up with three distinct opinions . for @xmath103 ,
our simulation results indicate the same type of behavior as the system evolves .
we define @xmath104 to be the largest @xmath3 for which consensus takes @xmath29 updates when we start with @xmath82 opinions with density @xmath105 for each opinion . then as @xmath106 the multi - opinion model has infinitely many phase transitions .
when @xmath107 , consensus occurs after @xmath100 steps if we start with @xmath82 opinions , while if we start with @xmath108 equally likely opinions the system quickly converges to a quasi - stationary distribution and diffuses until consensus occurs after @xmath29 updates and there will always be @xmath108 opinions present at the end .
the associated picture is the natural dimensional extension of the relationship between the @xmath92 and @xmath87 models : just as @xmath109 corresponds to the point at which the @xmath96 circle for @xmath87 is the inscribed circle within the @xmath89 triangle , @xmath110 corresponds to the point at which the @xmath96 circle reaches zero radius that is , the point at which the @xmath96 sphere for @xmath111 has become the inscribed sphere within the corresponding barycentric tetrahedron .
for each @xmath82 we simulate our multi - opinion rewire - to - random model starting from @xmath82 opinions with each opinion taking @xmath105 fraction of nodes at random for a wide range of @xmath3 s . generalizing the picture of the one - dimensional arch for @xmath92 and the two - dimensional cap for @xmath87 , the number of discordant edges as a function of frequencies conjectured in eq .
( [ quadcap ] ) is a co - dimension 1 hypersurface characterizing the quasi - stationary states , and the behavior of the equal - initial - populations case will allow us to describe this surface , thereby characterizing behaviors for general initial populations .
first the critical @xmath104 s are identified when the slow diffusion of @xmath86 can not be observed for the first time as @xmath3 increases from @xmath12 to @xmath53
. then we fit @xmath91 to @xmath112 ( @xmath113 ) using eq .
( [ quadcap ] ) at every @xmath3 up to @xmath104 , and plot the fitted coefficients @xmath114 and @xmath115 against @xmath116 in figure [ fig : c_2 ] . remarkably , the coefficients in ( [ quadcap ] ) appear to be well approximated by linear functions of @xmath4 .
the graphs shows some curvature near @xmath117 , which may be caused by the fact that @xmath117 ( @xmath25 ) corresponds to a voter model without evolution of the underlying network . in the rest of the paper
, we will work with @xmath118 for simplicity .
naturally , critical points @xmath104 translate to @xmath119 .
the fitted coefficients from the 2-opinion model deviate slightly from those fitted from higher - order models , which implies that eq .
( [ quadcap ] ) is not universal for the multi - opinion model and higher - order terms are possible . however , while the discrepancy between the fitted coefficients of the 2-opinion model and those of the 3-opinion one is apparent , difference between fitted coefficients of higher - order models is negligible , which implies that the inclusion of higher - order terms beyond the 3rd would not make significant changes to the equation . to probe the effect of higher - order terms
we introduce terms up to @xmath82th order for @xmath82 opinions . noting @xmath120 , eq .
( [ quadcap ] ) is equivalent to : e_u n_/m = - c_0 ( ) + c_2()_i , j=1;i >
j^k u_i u_j .
given the symmetry of the system in @xmath84 s , the only possible choice in degree - k polynomials is : @xmath121 where @xmath122 is the collection of all @xmath83-element subsets of @xmath123 . using the same simulation data as above , we refit @xmath91 to @xmath124 s ( @xmath113 ) according to the generalized formula eq .
( [ kthcap ] ) and plot the fitted coefficients @xmath114 and @xmath115 against @xmath118 in figure [ fig : c_2_high ] .
fitting diagnostics suggest that higher - order terms are significant from zero ( with @xmath125 ) and it can be seen that those terms explain the inconsistency between fitted coefficients of different models in figures [ fig : c_2 ] . however , the difference between the two fitted functions of eq .
( [ quadcap ] ) and eq .
( [ kthcap ] ) is actually small ( @xmath126 in @xmath127-norm ) and thus higher - order terms are small corrections to the hyper - surface eq .
( [ quadcap ] ) .
+ values of the coefficients @xmath128 for the three opinion model near its critical value @xmath129 show some scatter , but this is to be expected since the surface is very small at this point .
values for the four opinion model appear to become more difficult to fit prior to @xmath130 since @xmath131 is a three - dimensional hyper - surface in four - dimensional space , so much more data is required to get reliable estimates of coefficients .
as is visually apparent in figure [ fig : c_2_high ] , the coefficients @xmath114 and @xmath115 for the first two terms in eq .
( [ kthcap ] ) are well approximated by linear functions , with best fits @xmath132 and @xmath133 , while coefficients for higher - order terms are not linear in @xmath118 ( e.g. , see figure [ fig : c_3 ] for @xmath134 ) . for comparison ,
the best fits for @xmath114 and @xmath115 in eq .
( [ quadcap ] ) ( as in figure [ fig : c_2 ] ) are c_2 ( ) 1.3 + 0.5 , c_0 ( ) 0.25 .
[ coeff ] since eq .
( [ quadcap ] ) well approximate the higher - order hyper - surface eq .
( [ kthcap ] ) , its simple form can be used to estimate the critical points for phase transitions . combining ( [ quadcap ] ) and ( [ coeff ] ) , and then solving @xmath135 gives @xmath136 which agrees with the critical @xmath137 s identified when
the slow diffusion of @xmath86 can not be observed in simulations as @xmath118 increases . in eq .
( [ kthcap ] ) for models with multiple opinions .
each value of @xmath134 is obtained by fitting eq .
( [ kthcap ] ) to the same data in figure [ fig : c_2].,title="fig:",scaledwidth=50.0% ] +
our multi - opinion voter model has infinitely many phase transitions .
when @xmath138 , consensus occurs rapidly when we start with @xmath82 opinions , while if we start with @xmath108 equally likely opinions there will always be @xmath108 opinions present at the end . to a good approximation @xmath139 , but the departures from linearity in the plots of @xmath140 and @xmath141 suggest that this result is not exact .
however , formulas for various quantities associated with this model are close to polynomials , so an exact solution may be possible .
more complicated rewiring rules might also be considered , particularly if they maintained high clustering or other global macroscopic properties .
an even more complete understanding of the present rewiring system would help motivate similar investigations for other rewiring rules . | we consider an idealized model in which individuals changing opinions and their social network coevolve , with disagreements between neighbors in the network resolved either through one imitating the opinion of the other or by reassignment of the discordant edge .
specifically , an interaction between @xmath0 and one of its neighbors @xmath1 leads to @xmath0 imitating @xmath1 with probability @xmath2 and otherwise ( i.e. , with probability @xmath3 ) @xmath0 cutting its tie to @xmath1 in order to instead connect to a randomly chosen individual . building on previous work about the two - opinion case ,
we study the multiple - opinion situation , finding that the model has infinitely many phase transitions . moreover , the formulas describing the end states of these processes are remarkably simple when expressed as a function of @xmath4 . |
at low energies , the main features of quantum chromodynamics ( qcd ) are confinement and the spontaneous breaking of chiral symmetry .
the former implies that we only observe hadrons ( rather than quarks and gluons ) , while the latter is believed to govern the ( low - lying ) hadron - mass spectrum .
lattice - qcd calculations predict a phase transition from nuclear / hadronic matter to a deconfined , chirally symmetric state @xcite at temperatures @xmath5@xmath6150 - 200 mev , dictating a major reshaping of the hadronic spectrum in terms of degenerate chiral partners .
the observation of such medium modifications is therefore an important objective in relativistic heavy - ion collision experiments .
large theoretical efforts have been devoted to evaluate in - medium properties of vector mesons which are accessible experimentally through dilepton invariant - mass spectra @xcite . in most of these studies ,
baryon - driven effects are essential to account for the dilepton enhancement observed in @xmath7-@xmath3 collisions at the sps below the free @xmath8 mass @xcite .
thus , changes in the baryon properties themselves deserve further investigation . in addition , recent measurements of @xmath2 invariant - mass spectra in nuclear collisions @xcite may open a more direct window on modifications of the @xmath0(1232 ) . to date ,
in - medium properties of the @xmath0 have mostly been assessed in cold nuclear matter @xcite , with few exceptions @xcite . in this article
we will discuss properties of the nucleon and the @xmath0(1232 ) in a hot and dense medium @xcite , employing a finite - temperature field theory framework based on hadronic interactions .
both direct interactions of the @xmath0 with thermal pions as well as modifications of its free @xmath2 self - energy ( incuding vertex corrections ) will be accounted for .
the article is organised as follows . in sec .
[ sect.2 ] we introduce the hadronic lagrangian and outline how its parameters are determined using scattering and decay data in vacuum . in sec .
[ sect.3 ] we compute in - medium self - energies for nucleon and @xmath0 . in sec .
[ sect.4 ] we first check our model against photoabsorption cross sections on the nucleon and nuclei , followed by a discussion of the spectral functions under conditions expected to occur in high - energy heavy - ion collisions .
we close with a summary and outlook .
the basic element of our analysis are 3-point interaction vertices involving a pion and two baryons , @xmath9 .
baryon fields are treated using relativistic kinematics , @xmath10 , but neglecting anti - particle contributions and restricting rarita - schwinger spinors to their non - relativistic spin-3/2 components .
pions are treated fully relativisticly ( @xmath11 ) .
the resulting interaction lagrangians are thus of the usual nonrelativistic form involving ( iso- ) spin-1/2 pauli matrices , 1/2 to 3/2 transition operators , as well as spin-3/2 matrices @xcite , see ref .
@xcite for explicit expressions . to simulate finite - size effects we employed hadronic form factors with a uniform cutoff parameter @xmath12 mev ( except for @xmath13 and @xmath14 vertices ) .
the imaginary part of the vacuum self - energy for the @xmath0@xmath15@xmath16 decay takes the form @xmath17 with @xmath18 the center - of - mass decay momentum ( an extra factor @xmath19 has been introduced in eq .
( [ 3 ] ) to restore lorentz - invariance ) , and the real part is determined via a dispersion relation . with a bare mass of @xmath20=1302 mev ,
a form - factor cutoff @xmath21=290 mev and a coupling constant @xmath22 we obtain a satisfactory fit to the experimental @xmath23-phase shift @xcite . to account for resonant interactions of the @xmath0 with pions we identified the relevant excited baryons via their decay branchings @xmath24@xmath25@xmath1 .
the pertinent coupling constants have been determined assuming the lowest partial wave to be dominant ( unless otherwise specified ) @xcite , using pole masses and ( total ) widths of the resonances .
the same procedure has been adopted for resonant @xmath2 interactions ( which are used to evaluate finite - temperature effects on the nucleon ) .
the resonances included are @xmath26(1440 ) , @xmath26(1520 ) , @xmath26(1535 ) , @xmath0(1600 ) , @xmath0(1620 ) and @xmath0(1700 ) .
the total widths figuring into the resonance propagators have been obtained by scaling up the partial @xmath2 and @xmath1 channels , and vacuum renormalizations of the masses have been neglected .
finally , the evaluation of the photoabsorption cross section requires a @xmath27 vertex for which we employ the magnetic coupling @xcite @xmath28
the interactions described in the previous section are now used to evaluate the @xmath0 self - energy in hot hadronic matter .
the first modification concerns the @xmath2 loop which we obtain within the matsubara formalism as @xmath29 { a}_{\pi}(k ) g_n(l ) -f^n(l_0 ) a_n(l ) g_{\pi}(k ) \rbrace \ , \nonumber \end{aligned}\ ] ] where @xmath30 is the pion 4-momentum .
the thermal distributions are defined by @xmath31=@xmath32 and @xmath33=@xmath34 , with @xmath35 and @xmath36 the fermi and bose functions , respectively . for simplicity , finite pion - chemical potentials , @xmath37@xmath380 ,
are treated in the boltzmann limit to avoid bose singularities in the presence of broad pion spectral functions ( a more detailed discussion of this point will be given elsewhere ) .
( [ sgdnpi ] ) positive energies @xmath39@xmath380 correspond to outgoing pions , i.e. , @xmath0@xmath15@xmath2 decays , while @xmath39@xmath400 accounts for scattering with ( incoming ) pions from the heat bath . the key quantities in eq .
( [ sgdnpi ] ) are the in - medium pion and nucleon propagators , @xmath41 and @xmath42 , and pertinent spectral functions @xmath43=@xmath44 and @xmath45=@xmath46 .
the modifications of the pion propagator are implemetend via a self - energy , arising from two parts : ( i ) interactions with thermal pions modeled by a four - point interaction in second order ( sunset diagram " ) @xcite , with a coupling constant adjusted to qualitatively reproduce the results of more elaborate @xmath47 interactions in @xmath48 , @xmath49 , and @xmath50-wave @xcite ; ( ii ) interactions with baryons via @xmath49-wave nucleon- and @xmath0-hole excitations at finite temperature , described by standard lindhard functions , supplemented by short - range correlations encoded in migdal parameters @xcite ( our default values are @xmath51=0.8 , @xmath52=@xmath53=0.33 ) . these excitations induce a softening of the pion - dispersion relation which can even lead to a ( near ) vanishing of the pion group velocity at finite momentum , inducing an artificial threshold enhancement in the @xmath0 self - energy @xcite .
this feature is remedied by accounting for appropriate vertex corrections , which in the case of @xmath8@xmath25@xmath47 decays are required to maintain a conserved vector current in the medium @xcite .
here we apply the same technique to the @xmath14 vertex , cf .
[ fig.vcorr ] . ] ] the nucleon self - energy is calculated in terms of resonant interactions with thermal pions , at the same level of approximation as the pion lindhard functions ( i.e. , neglecting off energy - shell dependencies in the spectral functions of the excited baryons ) . the second contribution to the in - medium @xmath0 self - energy consists of resonant @xmath54 interactions , corresponding to the finite - temperature part of @xmath55 loops .
the resulting self - energy expressions are similar to eq .
( [ sgdnpi ] ) but with only the scattering part ( @xmath56 ) retained ( note that this is consistent with our description of the @xmath0(1232 ) in vacuum where @xmath55 loops are not included ) .
valuable constraints on the @xmath0 spectral function in cold nuclear matter can be obtained from photoabsorption cross sections on nuclei . to leading order in @xmath57
, the latter can be related to the photon self - energy ( electromagnetic current correlator ) , @xmath58 , by @xcite @xmath59 where @xmath60 denotes the photon energy ( momentum ) . with the vertex of eq .
( [ 4 ] ) we evaluate the @xmath61-induced @xmath0-hole loop using our full @xmath0 propagator .
the cross section for a nucleon target follows from the low - density limit ( @xmath62 ) of eq .
( [ sig_gam ] ) involving the free @xmath0-spectral function , which we use to fix the coupling constant and form - factor cut - off of the @xmath63 vertex at @xmath64=0.653 and @xmath65=400 mev , respectively , cf . left panel of fig .
[ fig_photo ] ( we have also included an estimate of the nonresonant background of @xmath66b @xcite ) . = 180 mev , @xmath67=0.68@xmath68 ; dashed line : @xmath5=100 mev , @xmath67=0.12 @xmath68 ) .
right panel : pion spectral function for cold nuclear matter ( dashed line : @xmath5=0 , @xmath67=0.68@xmath68 ) and at rhic ( solid line : @xmath5=180 mev , @xmath67=0.68@xmath68 ) ; the dash - dotted line corresponds to switching off baryonic effects leaving only the 4-point interactions with thermal pions . ]
= 180 mev , @xmath67=0.68@xmath68 ; dashed line : @xmath5=100 mev , @xmath67=0.12 @xmath68 ) .
right panel : pion spectral function for cold nuclear matter ( dashed line : @xmath5=0 , @xmath67=0.68@xmath68 ) and at rhic ( solid line : @xmath5=180 mev , @xmath67=0.68@xmath68 ) ; the dash - dotted line corresponds to switching off baryonic effects leaving only the 4-point interactions with thermal pions . ]
taking an average nuclear density of @xmath69 , our prediction for nuclei follows without further assumption , cf .
right panel of fig .
[ fig_photo ] .
the sensitivity to changes in the migdal parameters or the nuclear density is very moderate . given our rather simple approach for the cross section , the agreement with data is fair .
the discrepancies at low energy ( which seem to be present already for the nucleon ) could be due to interference with the background , collective effects involving direct @xmath26@xmath70-excitations , or transverse contributions with in - medium @xmath8 mesons in the vertex corrections of the @xmath0 decay . at higher energies ,
further resonances in the photon self - energy need to be included .
let us finally turn to the results for hot hadronic matter . in heavy - ion collisions
one expects a hierarchy of chemical freeze - out ( determining the ratios of stable hadrons ) and thermal freeze - out ( where elastic rescattering ceases ) .
the former is characterized by a temperature @xmath71 and a common baryon chemical potential @xmath72 .
thermal freezeout occurs at lower @xmath73@xmath6100 mev , which requires the build - up of additional chemical potentials for pions , kaons , etc .
@xcite , to conserve the observed hadron ratios , including relative chemical equilibrium for elastic processes , e.g. @xmath2@xmath74@xmath0 implying @xmath75=@xmath76 . under rhic conditions
the nucleon spectral function exhibits an appreciable broadening and a moderate downward mass shift ( left panel of fig .
[ fig_nuc - pi ] ) due to resonant scattering off thermal pions .
the pion spectral function ( right panel of fig .
[ fig_nuc - pi ] ) is strongly broadened mostly due to scattering off baryons , with little mass shift .
thermal motion completely washes out the multi - level structure visible at zero temperature ( dashed line ) . also for the @xmath0 spectral function ( left panel in fig .
[ fig_del ] ) the main effect is a broadening with a slight repulsive mass shift .
half of the increase of the in - medium width is due to baryon - resonance excitations ( slightly enhanced due to in - medium pion propagators ) , adding to the contribution of the @xmath2 loop . in the real part ,
however , the predominantly repulsive contributions from baryon resonances are counterbalanced by net attraction in the @xmath2 loop ( mostly due to the pion - bose factor ) . at thermal freeze - out
we find a peak position at about @[email protected] gev and a width @xmath78@xmath6177 mev , to be compared to the corresponding vacuum values of @[email protected] gev and @xmath78@xmath6110 mev , in qualitative agreement with preliminary data from star @xcite . for more conclusive comparison
a detailed treatment of the freeze - out dynamics is mandatory . in the vicinity of @xmath79 ,
the @xmath0 width increases substantially .
we expect this trend to be further magnified when including transverse parts in the vertex corrections , especially in combination with in - medium @xmath8-mesons @xcite . in the right panel of fig .
[ fig_del ] we show the @xmath0-spectral function in a net - baryon rich medium , representative for the future gsi facility .
whereas in dilute matter the line shape is only little affected , the resonance structure has essentially melted close to @xmath79 , mostly due to a strong renormalization of the pion propagator at high density .
based on hadronic interaction lagrangians employed within a finite - temperature many - body approach we have evaluated medium effects on pions , nucleons and deltas .
the resulting @xmath0-spectral functions in cold nuclear matter provide fair agreement with photoabsorption data on nuclei . in hot hadronic matter , we found a significant broadening and a slight upward peak shift of the @xmath0 resonance , qualitatively in line with preliminary measurements of @xmath2 invariant - mass spectra at rhic .
( 1232 ) spectral functions in heavy - ion collisions compared to free space ( dash - dotted lines ) ; left panel : rhic ; dashed line : @xmath5=100 mev , @[email protected]@xmath68 ( @xmath81=531 mev ) , @xmath37@xmath8096 mev ; solid line : @xmath5@xmath80180 mev , @[email protected]@xmath68 ( @xmath81=333 mev ) , @xmath37@xmath800 .
right : future gsi facility ; dashed line : @xmath5@xmath8070 mev , @[email protected]@xmath68 ( @xmath81@xmath80727 mev ) , @xmath37@xmath80105 mev ; solid line : @xmath5@xmath80160 mev , @[email protected]@xmath68 ( @xmath81@xmath80593 mev ) , @xmath37@xmath800 . ]
( 1232 ) spectral functions in heavy - ion collisions compared to free space ( dash - dotted lines ) ; left panel : rhic ; dashed line : @xmath5=100 mev , @[email protected]@xmath68 ( @xmath81=531 mev ) , @xmath37@xmath8096 mev ; solid line : @xmath5@xmath80180 mev , @[email protected]@xmath68 ( @xmath81=333 mev ) , @xmath37@xmath800 .
right : future gsi facility ; dashed line : @xmath5@xmath8070 mev , @[email protected]@xmath68 ( @xmath81@xmath80727 mev ) , @xmath37@xmath80105 mev ; solid line : @xmath5@xmath80160 mev , @[email protected]@xmath68 ( @xmath81@xmath80593 mev ) , @xmath37@xmath800 . ]
future improvements of the @xmath82 system in vacuum include @xmath83-channel exchange diagrams as well as spin-3/2-@xmath84 excitations which we expect to increase the rather low form - factor cut - off used so far .
we further plan to implement in - medium baryon propagators into the description of axial-/vector mesons within a chiral framework to arrive at a more consistent picture of the equation of state of hadronic matter under extreme conditions @xcite and the chiral phase transition .
another interesting ramification @xcite concerns the role of the medium - modified @xmath0 spectral functions in the soft photon enhancement as recently observed at the sps @xcite .
one of us ( hvh ) acknowledges support from the alexander - von - humboldt foundation as a feodor - lynen fellow .
10 url # 1`#1`urlprefix[2][]#2 | we investigate properties of the @xmath0(1232 ) and nucleon spectral functions at finite temperature and baryon density within a hadronic model .
the medium modifications of the @xmath0 consist of a renormalization of its pion - nucleon cloud and resonant @xmath1 scattering .
underlying coupling constants and form factors are determined by the elastic @xmath2 scattering phase shift in the isobar channel , as well as empirical partial decay widths of excited baryon resonances . for cold nuclear matter the model provides reasonable agreement with photoabsorption data on nuclei in the @xmath0-resonance region . in hot hadronic matter typical for late stages of central @xmath3-@xmath3 collisions at rhic
we find the @xmath0-spectral function to be broadened by @xmath465 mev together with a slight upward mass shift of 5 - 10 mev , in qualitative agreement with preliminary data from the star collaboration . |
the interplay between the electron s spin degree of freedom and the inter - electron interaction has been of interest in the condensed matter physics , in particular in the jellium model where the neutralizing background is taken to be rigid and uniform .
the relative strength of the interaction is conventionally measured through the parameter @xmath3 which is the interparticle distance measured in units of the bohr radius @xmath4 , @xmath5 being the dielectric constant and @xmath6 the band mass of electron . for electrons confined to two dimensions , which will be our focus in this paper , we have @xmath7 where @xmath8 is the two - dimensional density of electrons .
the interaction strength is enhanced relative to the kinetic energy as the system becomes more dilute , i.e. , as @xmath3 increases .
it has been predicted that as @xmath3 is increased , the electron liquid eventually becomes spontaneously spin polarized to gain in the exchange energy , ultimately going into a wigner crystal at @xmath9 @xcite .
the two - dimensional electron systems , obtained experimentally at the interface of two semiconductors , constitute an almost ideal realization of the jellium model for several reasons .
samples with mobility in access of 10 million @xmath10/vs are available @xcite , minizing the effect of disorder .
furthermore , the density of electrons , which controls the strength of the interaction relative to the kinetic energy , can be varied by a factor of 20 @xcite .
we will consider electrons in the presence of a magnetic field , specifically , at filling factor @xmath0 , which is a particularly clean test case for the kind of physics in which we are interested .
there are three relevant energy scales here : @xmath11 is the cyclotron energy , @xmath12 is the typical coulomb energy , @xmath13 being the magnetic length , and @xmath14 is the zeeman splitting energy which is the zeeman energy cost necessary for a single spin flip .
( to an extent , the zeeman and the cyclotron energies can be varied independently by application of the magnetic field at an angle ; while the former depends on the total magnetic field , the latter is determined by the normal component only . )
the ground state is known in two limits .
when @xmath15 dominates , the ground state is a spin singlet , with 0@xmath16 and 0@xmath17 landau levels fully occupied . on the other hand , when @xmath14 is the largest energy , the ground state is fully polarized ; when the coulomb interaction is not strong enough to cause substantial landau level mixing ( i.e. , in the limit of @xmath18 ) , the ground state has 0@xmath16 and 1@xmath16 landau levels occupied .
when the ground state is described in terms of filled landau levels , we will denote it by @xmath19@xmath20@xmath21 , where @xmath22@xmath16 and @xmath22@xmath17 are the numbers of occupied landau levels for up and down spin electrons .
the possible filled landau level states at @xmath0 are then @xmath23 and @xmath24 , the unpolarized and the fully polarized states , respectively .
our interest will be in situations when @xmath25 becomes comparable to or greater than the cyclotron energy .
this again corresponds to @xmath26 , where at @xmath0 , @xmath3 can be seen to be given by @xmath27 and is clearly a measure of the strength of the interaction relative to the cyclotron energy .
here , landau level mixing becomes crucial and may destabilize the above states
. our principal goal will be to determine the phase diagram of the fully polarized and the unpolarized states @xmath24 and @xmath23 . our work has been motivated by the recent experiments of eriksson _ et al . _
@xcite where they investigate by inelastic light scattering both the spin and the charge density collective modes at @xmath0 for samples with @xmath3 as large as 6 ,
corresponding to densities as low as @xmath28 @xmath29 .
they find a qualitative change in the number and the character of collective modes at approximately @xmath30 .
this was interpreted in a landau fermi liquid approach , where the magnetic field was treated as a perturbation on the zero field fermi liquid .
however , the fermi liquid approach , which is suitable at small magnetic field , is not an obviously valid starting point for the problem at hand , and other approaches are desirable . a comparison between the ground state energies of the unpolarized and the fully polarized hartree fock state shows that a transition between them takes place at @xmath31 for @xmath32 , as we will see in .
this raises the question : is the ground state at large @xmath3 fully polarized ?
if true , this would indeed be an interesting example of an interaction driven ferromagnetism .
if it is indeed fully polarized , is the @xmath24 state a reasonable starting point for its study ?
besides being fully polarized , @xmath24 incorporates the effect of landau level mixing , albeit in a very special manner , through promoting all electrons in 0@xmath17 to 1@xmath16 . a reliable treatment of landau level mixing lies at the crux of the problem .
we shall incorporate landau level mixing in a perturbative time - dependent hartree - fock scheme , i.e. , by incorporating vertex corrections through ladder diagrams in the random phase approximation ( rpa ) .
the most crucial approximation in our calculations will be a restriction to the subspace of a single - particle hole pair ; within this subspace , however , the landau level mixing is treated accurately .
clearly , this approach is not quantitatively valid except at small @xmath3 , but we believe that it gives an insight into the physics even when @xmath3 is not small . for a better quantitative description at large @xmath3 , it would be important to deal with screening by more than one particle - hole pair , but that is outside the scope of the present paper .
as mentioned above , there certainly are parameters for which the fully polarized @xmath24 state describes the actual ground state ; in particular , it may occur even when @xmath3 is large provided that the zeeman energy is sufficiently strong .
our approach here will be to take it as the starting point and investigate the regime of its stability by calculating the dispersion of the charge and the spin density collective modes .
the collective modes have a simple interpretation when the landau level mixing is negligible ( @xmath33 ) .
the lowest charge density excitation mode corresponds to the excitation of one electron from 1@xmath16 to 2@xmath16 landau level .
the kinetic energy change of any collective mode is well defined when the landau level mixing is weak , and will be used to label the various modes ( this notation will be used even when the landau level mixing is significant , by looking at the evolution of the modes from small to large @xmath3 ) .
the 1@xmath16@xmath342@xmath16 mode will be referred to as the @xmath35 mode . for spin - density excitation
, there are three modes of primary interest , corresponding to excitations 0@xmath16@xmath340@xmath17 , 1@xmath16@xmath341@xmath17 , and 1@xmath16@xmath340@xmath17 which are depicted schematically in ( a ) , ( b ) , and ( c ) in fig.[sdediagram ] respectively .
the first two are @xmath36 modes and the last one is @xmath37 mode .
these are of course coupled , and a diagonalization of the problem produces the usual spin - wave excitation mode , with the energy approaching the zeeman splitting in the long wave length limit , in accordance with the goldstone theorem , as well as two massive spin - density modes .
our principal result is that while the charge density collective mode shows no instability , the spin - density collective mode develops a deep roton minimum in the presence of substantial landau level mixing and becomes soft in certain parameter regimes .
( indeed , as @xmath3 is increased , there will eventually also be an instability in the charge density channel , indicating the formation of a wigner crystal , but we have not explored this question since our perturbation theory is not valid at very large @xmath3 . ) both the existence of the roton minimum and its softening as the zeeman energy is reduced are experimentally testable predictions of our theory .
the phase diagram thus obtained later is shown in fig.[phase ] .
it is instructive to compare it with the phase diagram in the absence of interactions , which consists of the fully polarized state @xmath24 for @xmath38 and the unpolarized state @xmath23 for @xmath39 with a transition taking place at precisely @xmath40 . at small @xmath3
, interactions make the @xmath24 state more stable , indicated by the fact that it survives even for @xmath41 .
however , at large @xmath3 , @xmath38 is required to stabilize the fully polarized state .
it is noteworthy that at small @xmath14 , the fully polarized state is found to be unstable at arbitrary @xmath3 .
we also consider the collective modes of @xmath23 , expected to be valid at small @xmath14 and @xmath3 .
it becomes unstable as @xmath3 is increased , consistent with an earlier conclusion of macdonald @xcite .
what about the state at large @xmath3 but small zeeman energy ?
as discovered in our study , here the @xmath0 state is not described by either of the two aforementioned hartree fock states .
the finite wave vector spin - wave instability of the fully polarized state suggests that it is some kind of spin density wave state .
it is not possible to be more definitive about it based on our present study .
of course , at extremely large @xmath3 , when @xmath25 is much larger than the cyclotron energy , our calculation is unreliable , but indicates that even if a fully polarized state occurs , as expected based on the zero field result , it will most likely not be described by the @xmath24 hartree fock state .
there have been many theoretical studies of the collective excitations of integer quantum hall effect ( iqhe ) states in the past , but , to our knowledge , the spin - density wave excitations of the fully polarized @xmath24 state have not been considered previously . for other collective modes ,
our results reduce to the earlier results in appropriate limits . if possible transitions of the electron and hole between different landau levels are ignored and if self - energy corrections are omitted , the problem of collective excitation is reduced to determining the binding energy of two oppositely charged particles strictly confined in their respective landau levels@xcite . in this case
the wave function for the bound state is independent of interaction potential , and is uniquely determined by the wave vector @xmath42 .
the transition of the electron and hole or the recombination of the particle - hole pair has been considered in the random phase approximation(rpa)@xcite . later the rpa was incorporated with the self - energy correction and the binding energy term by kallin and halperin @xcite where a number of interesting collective excitations from the unpolarized and the partially polarized ground state were considered in the absence of landau level mixing ( valid when @xmath43 ) .
the landau level mixing was considered by macdonald in , treating the mixing matrix elements between various modes as small parameters and applying a second order perturbation theory .
our calculation will be formulated in terms of diagrams , following kallin and halperin in , and will be performed with a full treatment of the mixing matrix elements within the subspace of a single particle - hole pair at any given instant .
the diagrammatic formulation of the problem is presented in detail in sec.[sec - diagram ] below . in sec.[sec - fullypol ] we describe the diagrammatic formalism used to compute dispersion curves of the collective excitation from the fully polarized ground state at @xmath0 .
we will concentrate on the spin density excitation which will be responsible for an instability of the iqhe state in the parameter regime under consideration .
similarly , is devoted to the collective excitations of the unpolarized ground state at @xmath0 .
dispersion curves of the spin density excitation is computed for various values of @xmath3 and zeeman splitting energy , @xmath14 . the phase diagram as a function of @xmath3 and @xmath14
is obtained in by determining the critical @xmath14 at which the energy of the spin density excitation vanishes . from the phase diagram we will learn that for large @xmath3 and small @xmath14 neither the fully polarized nor the unpolarized state is stable against a spin - density wave state . the paper is concluded in , where we also discuss the implications of our study for the fractional quantum hall effect ( fqhe ) .
even though the choice of gauge does not affect physical quantities , we find it convenient to use the symmetric gauge
. we will start by establishing the basic algebra in the symmetric gauge closely following .
the hamiltonian for an electron moving in a two dimensional space under a perpendicular magnetic field is given by @xmath44 where the kinetic momentum is written down as @xmath45 when the magnetic field is uniform , we can use a convenient algebraic method analogous to the solution by the ladder operator of the one - dimensional harmonic oscillator . in order to construct the algebraic formalism
we first note that the @xmath46 and @xmath47 components of the kinetic momentum are canonically conjugate coordinates : @xmath48 = \frac{-i \hbar e}{c } \hat{z } \cdot ( \vec{\nabla } \times \vec{a } ) = \frac{-i { \hbar}^2 } { { l_0}^2}.\ ] ] from the commutation relationship between @xmath49 and @xmath50 we can define a ladder operator so that the ladder operator and its hermitian conjugate satisfy the same commutation relation as those of the one - dimensional harmonic oscillator .
that is , @xmath51 = 1 , \ ] ] where @xmath52 the hamiltonian can now be written in the form of a one - dimensional harmonic oscillator .
@xmath53 therefore the eigenvalues are @xmath54 where @xmath55 is a non - negative integer which is known as the landau level index .
the eigenstates , however , can not be fully determined by the landau level index alone because the energy does not depend on the coordinates of the cyclotron orbit center , indicating a degeneracy of the landau level .
let us define @xmath56 which is convetionally known as the guiding - center operator .
the @xmath46 and @xmath47 components of the guiding - center operator are canonically conjugate coordinates similar to those of the kinetic momentum .
@xmath57 = i { l_0}^2\ ] ] therefore we can define another ladder operator by @xmath58 which satisfies @xmath59 = 1\ ] ] and @xmath60 = [ a^{\dagger } , b ] = [ h , b ] = 0 .\ ] ] the fact that the ladder operator @xmath61 commutes with the hamiltonian shows that the degeneracy of the lowest landau level is actually related to the positioning of the guiding - center coordinate . since we identify two independent sets of ladder operators in a two - dimensional space , the full set of eigenstates can be generated by repeatedly applying raising operators on the ground state : @xmath62 direction , the vector potential in the symmetric gauge is given by @xmath63 and the ladder operators may be written in terms of @xmath64 as follows .
@xmath65 @xmath66 @xmath67 @xmath68 with the hamiltonian given by @xmath69 here and in the rest of the paper , we use the convention that @xmath70 , as well as the area@xmath71 .
in particular , this implies that the total number of flux quanta piercing the system , @xmath72 .
up to this point we have concentrated on the single particle hamiltonian . in order to compute concrete physical quantities with
the interaction treated perturbatively , we will need to use certain matrix elements and their various properties , which we now list @xcite .
@xmath73 _ plane wave matrix elements _ the matrix element of a plane wave @xmath74 is given by @xmath75 where @xmath76 , @xmath77 and @xmath78 where , for @xmath79 , @xmath80 is the associated laguerre polynomial , defined as : @xmath81 for @xmath82 we define : @xmath83 can be evaluated first by rewriting @xmath84 , expressing @xmath85 and @xmath86 in terms of the ladder operators , and moving all annihilation operators to the right using @xmath87}$ ] .
@xmath73 _ matrix products _ one of the most important properties of the matrix @xmath88 is its product @xmath89 this can be derived by using the definition in eq .
( [ eq : g ] ) and the completeness of the ladder operator eigenstates .
@xmath73 _ eigenfunctions in the symmetric gauge _
similar to duality of the position - space and momentum - space in the one - dimensional harmonic oscillator , we have the orbital wave function closely related to the plane - wave matrix element .
@xmath90 to derive this , first note that where @xmath91 is given by @xmath92\ ] ] which is annihilated by both @xmath93 and @xmath61 .
the eigenfunction for a general @xmath55 and @xmath6 is therefore given by @xmath94\ ] ] now write @xmath95 and use @xmath96 ( \frac{\bar{z}}{2}-2 \frac{\partial}{\partial z})^m ( \frac{z}{2}-2 \frac{\partial}{\partial \bar{z}})^n \exp[\frac{1}{4 } z\bar{z}]=(-2)^{m+n } ( \frac{\partial}{\partial \bar{z}})^n ( \frac{\partial}{\partial z})^m\ ] ] defining @xmath97 one gets @xmath98 which reduces to eq .
( [ eq : phi ] ) with the standard definition of the associated laguerre polynomial .
@xmath73 _ trace _ the trace of @xmath88 is a constant because the charge density of a completely filled landau level is a constant .
@xmath99 where @xmath100 stands for the area of system and therefore @xmath101 is @xmath102 . using the definition of @xmath88 , , and properties of the laguerre polynomial we find the transpose and complex conjugate as follows .
@xmath73 _ transpose _
@xmath103 @xmath73 _ complex conjugate _
@xmath104 @xmath73 _ orthogonality _ the final property we will use is the orthogonality between the plane - wave matrix elements which will be useful in computing the self - energy : @xmath105 the response function is a very important quantity from which we can deduce many physical observables assuming linear response . in particular , the collective excitations correspond to the poles of the response function . in order to calculate the response function we use the zero temperature limit of the matsubara formalism@xcite , using the standard analytic continuation ( @xmath106 ) in order to get the retarded response function .
we will describe below in detail only the charge - density response function , since the spin - density response function can be obtained with a straightforward modification .
the charge - density response function is related to the density - density correlation function as follows : @xmath107 where the density operator , @xmath108 , is given in the symmetric gauge by @xmath109 the actual computation of the density response function is performed in the perturbation theory .
assuming at any given time there is a single particle - hole pair in the system , we write the response function which is depicted by the feynman diagram in .
@xmath110 where the vertex function @xmath111 satisfies the following equation : @xmath112 \nonumber \\ & \times & d_{{n_{\alpha}}'{n_{\beta}}'}(i\omega ) \gamma^{m_{{\alpha } } ' { m_{\beta}}'}_{{n_{\alpha } } ' { n_{\beta}}'}(k , i\omega ) .
\label{eq : gamma1}\end{aligned}\ ] ] where @xmath113 is a potential and @xmath114 is its fourier transform
. the feynman diagrams corresponding to each term on the right - hand side of the are shown in : the first term describes just the bare vertex and the second and third terms depict the rpa correction ( bubble diagram ) and the binding energy term ( ladder diagram ) , respectively . in the eq .
( [ eq : chi1 ] ) and eq .
( [ eq : gamma1 ] ) @xmath115 is defined as the frequency integral of the product of two green functions : @xmath116 and the green function is given by @xmath117 where @xmath55 is the landau level index of the electron or hole and @xmath118 is the chemical potential which is halfway between the highest occupied landau level and the lowest unoccupied landau level .
we note that the green function is fully dressed , containing the self energy ( @xmath119 ) correction , and is also independent of the @xmath6 quantum number . in general the self energy
@xmath119 is a very complicated function of @xmath42 and @xmath120 , but it turns out that in this system the self energy is real and depends on just the landau level index but not @xmath42 and @xmath120 .
we will use this result below , postponing the explicit derivation to the next section . plugging the explicit form of green function into eq.([eq : d1 ] )
gives @xmath121 where @xmath122 and it is assumed that the self energy does not modify the fermi level significantly but moves the position of the pole in the denominator of . in order to solve the vertex equation , , we first eliminate the angular momentum index by defining a new vertex function @xmath123 : @xmath124 eq.([eq : msum2 ] ) is obtained from eq.([eq : msum1 ] ) through the property of the plane - wave matrix product . noting that the bare vertex is just a plane - wave matrix element defined in the eq.([eq : planewave ] ) and utilizing the properties of the plane - wave matrix elements , we succeed in summing over all the angular momentum indices to get a new vertex equation .
@xmath125 d_{{n_{\alpha}}'{n_{\beta}}'}(i\omega ) \gamma_{{n_{\alpha } } ' { n_{\beta}}'}(k , i\omega ) \label{eq : gammasum}\end{aligned}\ ] ] where @xmath126 similarly , the density response function can be written in terms of the new vertex function , @xmath127 .
@xmath128 the third term on the right hand side of eq.([eq : gammasum ] ) contains a formal expression for the coulomb - potential matrix elements between the wave function @xmath129 and @xmath130 which prove to be the eigenfunctions of the two - body hamiltonian for oppositely charged particles confined in the landau levels @xmath131 and @xmath132 , respectively @xcite .
therefore , following kallin and halperin , we will call it the binding energy . the second term in eq.([eq : gammasum ] )
has the physical interpretation of the exchange energy of particle - hole pair which is the only term in the usual rpa . for convenience
let us define @xmath133 , and denote the binding energy and the exchange energy as @xmath134 and @xmath135 respectively : @xmath136 @xmath137 then eq .
( [ eq : gammasum ] ) becomes @xmath138
d_{{n_{\alpha}}'{n_{\beta}}'}(i\omega ) \gamma_{{n_{\alpha } } ' { n_{\beta}}'}(k , i\omega).\end{aligned}\ ] ] the green function @xmath139 contains a factor @xmath140 which vanishes for certain choices of @xmath141 . however , these choices co not contribute to @xmath142 in eq .
( [ chi ] ) . therefore ,
restricting to only those values of @xmath141 for which @xmath139 is non - zero , we transform the equation into an associated set of linear equations : @xmath143 d_{{n_{\alpha}}'{n_{\beta}}'}(i\omega ) \gamma_{{n_{\alpha } } ' { n_{\beta}}'}(k , i\omega ) \nonumber \\ & = & \frac{{i}^{n_{\alpha\beta}}}{2\pi } e^{-k^2 /2 } { \bar{g}}_{n_{\alpha}n_{\beta}}(-\bar{\kappa } ) , \label{eq : gammauv}\end{aligned}\ ] ] where @xmath144 ( note that here @xmath145 is the inverse of the matrix element , not the matrix element of the inverse . ) eq.([eq : gammauv ] ) can be written in the form of a matrix equation by defining a matrix @xmath146 as follows : @xmath147 @xmath148 can be viewed as a matrix element of @xmath146 if we consider the set of indices @xmath149 to be a collective index .
therefore we write eq.([eq : gammauv ] ) in the following form : @xmath150 inverting the matrix @xmath146 amounts to solving the vertex equation .
that is , @xmath151 where @xmath152 is the inverse matrix of @xmath146 .
finally , we obtain the density response function in terms of the matrix @xmath146 . @xmath153 since the summation of the landau level indices should be performed over all the possible states , the number of terms is actually infinite .
however , it is possible to obtain an accurate estimate keeping a reasonably small number of terms .
the collective excitations are the poles of the response function , which , from eq.([eq : chi ] ) , correspond to energies for which the inverse of @xmath146 becomes singular , that is to say , when @xmath154 = 0 .
\label{eq : detm}\ ] ] we carry out detailed computations of the binding energy , rpa energy and self energy in order to explicitly evaluate the dispersion curve of the collective excitation .
we start with the binding energy .
@xmath73 binding energy ( ladder diagram contribution ) the explicit form of @xmath155 is obtained from eq.([eq : phi ] ) by utilizing the plane - wave matrix product formula , eq.([eq : product ] ) : @xmath156 following kallin and halperin , the wave function @xmath157 can be shown to be an eigenstate of the following hamiltonian where the particles 1 and 2 are projected onto the landau levels with indices @xmath158 and @xmath159 respectively .
@xmath160 \hat{p}_{n_\alpha , n_\beta } \nonumber \\ & = & \hbar\omega_c ( n_\alpha + n_\beta + 1 ) + \hat{p}_{n_\alpha , n_\beta}v_{arb}(|\vec{r_1}-\vec{r_2}| ) \hat{p}_{n_\alpha , n_\beta}\end{aligned}\ ] ] where @xmath161 @xmath162 is the kinetic momentum for the particle 1 ( 2 ) , @xmath163 is an arbitrary ( attractive ) potential and @xmath164 is the operator projecting the particles onto the landau levels with indices @xmath158 and @xmath159 .
one can prove that @xmath157 is the eigenfunction of @xmath165 by computing the overlap between @xmath166 and any arbitrary basis state in the projected hilbert space , for example @xmath167 .
then we note that it is proportional to the overlap between @xmath157 and @xmath167 .
of course , the proportionality constant is the eigenvalue which is equal to the binding energy previously defined in eq.([eq : v ] ) in the case of @xmath168 and @xmath169 .
now let us get the explicit formula for the binding energy .
@xmath170 where @xmath171 is an associated laguerre polynomial . by using the fact
@xmath172 we perform the angle integrations to get the final formula : @xmath173 where @xmath174 , @xmath175 and @xmath176 is a bessel function and @xmath177 is the angle of @xmath42 measured from the @xmath46-axis .
@xmath73 random phase approximation energy ( bubble diagram contribution ) the exchange energy from the rpa is rather straightforward to calculate since there is no integration involved .
the explicit form is given by @xmath178 incidentally , a comparison between eq.([eq : explicitv ] ) and eq.([eq : explicitu ] ) reveals that @xmath135 and @xmath134 have the same phase factor , @xmath179 .
therefore the phase factor can be eliminated in a consistent way , which is expected because the system is uniform and isotropic .
@xmath73 self energy the last diagram in the fig.[vertex ] is the self - energy contribution to the `` full '' green function .
as before , a single particle - hole pair is assumed , which is reflected in the feynman diagram through the unscreened coulomb line .
the corresponding dyson equation is solved in the conventional way : @xmath180 where @xmath181 and @xmath182 \nonumber \\ & \times & \int \frac{d \omega}{2\pi } g^{(0)}_{n_{\beta}}(i\omega ) \nonumber \\ & = & -\sum_{n_{\beta}m_{\beta } } \bigg [ \int d^2 \vec{r_1 } \int d^2 \vec{r_2 } { \bar{\phi}}_{n_{\alpha}m_{\alpha}}(\vec{r_1 } ) \phi_{n_{\alpha}m_{\alpha}}(\vec{r_2 } ) v(\vec{r_1}-\vec{r_2 } ) \phi_{n_{\beta}m_{\beta}}(\vec{r_1 } ) { \bar{\phi}}_{n_{\beta}m_{\beta}}(\vec{r_2 } ) \bigg ] \nonumber \\ & \times & \theta ( n_0 + 1/2 -n_{\beta } ) \nonumber \\ & = & -\sum^{n_0}_{n_{\beta}=0 } \bigg [ \int d^2 \vec{r_1 } \int d^2 \vec{r_2 } { \bar{\phi}}_{n_{\alpha}m_{\alpha}}(\vec{r_1 } ) \phi_{n_{\alpha}m_{\alpha}}(\vec{r_2 } ) v(\vec{r_1}-\vec{r_2 } ) \sum_{m_{\beta } } \phi_{n_{\beta}m_{\beta}}(\vec{r_1 } ) { \bar{\phi}}_{n_{\beta}m_{\beta}}(\vec{r_2 } ) \bigg ] \nonumber \\ & = & -\int \frac{d^2 \vec{r_1}}{2\pi } \int \frac{d^2 \vec{r_2}}{2\pi } e^{-({r_1}^2+{r_2}^2 -z_2 \bar{z_1})/2 } v(\vec{r_1}-\vec{r_2 } ) { \bar{g}}_{m_{\alpha}n_{\alpha}}(i\bar{z_1 } ) g_{m_{\alpha}n_{\alpha}}(i\bar{z_2 } ) \sum^{n_0}_{n_{\beta}=0 } g_{n_{\beta}n_{\beta}}(i(\bar{z_1}-\bar{z_2}))\end{aligned}\ ] ] in the above equation the explicit form of the single particle eigenstate , eq.([eq : phi ] ) , and the plane - wave matrix product formula , eq.([eq : product ] ) , have been used .
it is convenient for the computation of integrals to change the variables from @xmath183 and @xmath184 to the center of mass coordinate @xmath185 and the relative coordinate @xmath186 . in the form of a complex number @xmath187 and @xmath188 .
@xmath189 the angular momentum index @xmath190 is naturally eliminated after the integration over the center - of - mass coordinate by using eq.([eq : orthogonality ] ) .
the plane - wave matrix product formula has also been used when we proceed from the third step to the fourth step . using @xmath191
one can finally write down the self energy as follows : @xmath192 in order to see what modifications need to be made in the above analysis to include the spin degree of freedom , it is instructive to recall the physical meaning of the three parts of the collective exciton energy .
the binding energy due to the ladder diagram is the _ direct _ interaction between the excited electron and the hole .
therefore it will not be affected by the presence of the spin degree of freedom .
on the other hand , the rpa energy is the _ exchange _ energy between the excited electron and the hole .
so it will vanish if the excited electron has a different spin than the hole , as in the case of the spin density excitation .
in other words the electron - hole pair with the same spin can not recombine through the coulomb potential . in the case of the charge density excitation ,
however , the rpa energy depends on the polarization of the ground state .
we have computed the rpa energy in the previous section assuming that all the electrons have the same spin , which corresponds to the fully polarized state .
if the ground state is unpolarized , the rpa energy will be twice as large as that of the fully polarized state simply because the particle - hole pair can be created and annihilated as either a spin - up or spin - down pair .
formally speaking , the vertex equation , , will have two identical rpa terms for a given set of indices .
the self energy term is due to the exchange energy between a given electron and the rest of electrons in the system while there is no direct term because we assume a neutralizing positive charge background .
since it is an exchange term , we have to include only the interaction between the electrons with the same spin .
therefore is generalized to include the spin degree of freedom as follows : @xmath193 where @xmath194 and @xmath195 is the index of the highest landau level occupied by the electron with spin @xmath196 .
using this new self energy one can rewrite as follows : @xmath197 where the chemical potential @xmath198 was defined earlier , and the zeeman coupling is included through the term @xmath199 .
computing the dispersion curve of collective excitations for a general @xmath3 requires solving the equation for the pole of the response function , . in the limit @xmath200 , when there is no landau level mixing , the matrix has a block diagonal form since collective modes with different kinetic energies do not couple , and each block , which has a finite dimension , can be diagonalized separately to obtain the collective mode energies @xcite . at non - zero values of @xmath3 , however , the full matrix must be diagonalized .
strictly speaking , the matrix @xmath146 in is of infinite dimension , but in practice , we work with a finite size matrix , keeping a sufficient number of landau levels to ensure a convergence of the collective mode energy . for the lowest energy collective modes , which are our primary concern , and for @xmath201
, we find that it is adequate to work with @xmath146 of dimension of up to 20 .
we also find it useful to convert the eq .
( [ eq : detm ] ) into an eigenvalue equation , the solution of which can be obtained using standard linear algebraic methods . in this formulation
it is natural to define an effective hamiltonian to be the matrix in the eigenvalue problem .
the details of the procedure are discussed in the remainder of this section . for convenience
we write down here after the analytic continuation .
@xmath202 for an arbitrary set of indices @xmath131 where @xmath203 is greater than @xmath204 , there is a reversed set @xmath205 whose kinetic energy cost is negative .
the mode described by the reversed set of landau level indices was called a negative - energy mode by macdonald because of its negative kinetic energy cost .
macdonald considered mixing between positive and negative energy modes as well as between the positive energy modes in second - order perturbation theory in order to compute the collective excitations at general @xmath3 in the spin unpolarized ground state . in the present work
we directly solve the pole equation , , instead of approximating it to the second - order .
in computing the lowest lying mode it is especially important to consider mixing with the negative - energy modes since the lowest positive - energy mode is energetically closest to the negative - energy modes . in order to make our discussion concrete and transparent ,
let us explicitly write down the matrix elements in the case of the full spin polarization . if @xmath206 , @xmath207 and @xmath208 , is written , after the analytic continuation , as follows : @xmath209 if the order of the indices is reversed , we get @xmath210 therefore the matrix @xmath146 defined in and its counterpart with reversed indices are @xmath211 and @xmath212 where we set @xmath213 without loss of generality and therefore @xmath214 . the sign in front of the second term in can be interpreted so that the negative - energy mode has the angle of wave vector equal to @xmath215 : for the negative - energy mode @xmath216 in the phase factor of @xmath217 and @xmath218 .
according to the negative - energy mode has a negative frequency and the opposite direction of wave vector relative to the positive - energy mode .
therefore when the positive - energy mode is viewed as a plane wave @xmath219 , the negative - energy mode is the complex conjugate plane wave @xmath220 . in this interpretation
the requirement of negative - energy mode is natural since an arbitrary plane wave with @xmath42 is written as a linear combination of @xmath219 and @xmath220 .
therefore in general a collective excitation with @xmath42 should be described by both the positive - energy and negative - energy modes .
incidentally , we mention that the mode describing an excitation within the same landau level , for example the spin - wave goldstone mode , does not have the negative - energy counterpart because there should not be any double counting in .
in any case we realize that the pole equation , , is not an eigenvalue equation as it stands because of the sign change in @xmath120 for the negative - energy mode .
but it can be transformed to an eigenvalue equation as follows .
let us denote @xmath146 in terms of sub - matrices .
@xmath221\ ] ] where @xmath222 is associated with mixing between the positive - energy ( negative - energy ) modes and @xmath223 is between the positive and negative energy modes .
thanks to a property of determinant we can obtain the solution of @xmath224 = 0 $ ] by solving the following equation .
@xmath225 = 0 , \label{eq : detmtilde}\ ] ] where @xmath226\ ] ] now one can define _ an effective hamiltonian matrix _
@xmath227 using @xmath228 .
@xmath229 therefore amounts to the eigenvalue equation of the effective hamiltonian matrix @xmath230 , which , however , is not a hermitian matrix because of the sign change . solving the eigenvalue equation of a non - hermitian matrix
is complicated by the fact that the eigenvalues of a non - hermitian matrix can be highly sensitive to small changes in the matrix elements @xcite .
the sensitivity of eigenvalues to rounding errors during the execution of some algorithms can be reduced by the procedure of _ balancing_. the idea of balancing is to use similarity transformations to make corresponding rows and columns of a matrix have comparable norms , thus reducing the overall norm of the matrix while leaving the eigenvalues unchanged .
then the general strategy for finding the eigenvalues of a matrix is to reduce the matrix to a simpler form , and perform an iterative procedure on the simplified matrix .
the simpler structure we use is called the hessenberg form .
an upper hessenberg matrix has zeros everywhere below the diagonal except for the first subdiagonal row .
then one can find the eigenvalues by applying the _ qr algorithm _ repeatedly to the hessenberg form until convergence is reached .
finally , we mention that even though the above formalism has been developed for a general situation , we will concentrate on the physics at @xmath0 in the following sections taking the coulomb potential as the interaction .
when the finite thickness effect of the 2d system is of interest , one can replace the coulomb potential by an effective potential such as the stern - howard potential @xcite .
equipped with the explicit formulas for the binding energy , the rpa energy and the self energy , we compute the dispersion curves of the collective excitation from the fully polarized ground state .
first we study the large b field limit , i.e. the small @xmath3 limit , where the ( time - dependent ) hatree - fock approximation is valid . in the small @xmath3 limit the effective hamiltonian @xmath230 defined in is already block - diagonalized so that only the matrix elements within the hilbert subspace of the same kinetic energy survive .
the off - diagonal terms due to the interaction energy become negligible compared to the kinetic energy .
therefore the pole equation is simple to solve in this case . for an arbitrary value of @xmath3
the hamiltonian is generally complicated and needs to be diagonalized over whole hilbert space .
when we consider the small @xmath3 limit , the integrals we encounter in eq.([eq : explicitv ] ) , eq.([eq : explicitu ] ) and eq.([eq : sigma ] ) can be expressed in terms of @xmath231 and @xmath232 , defined as follows : @xmath233 and @xmath234 where @xmath55 is an integer .
the explicit functional forms of @xmath231 and @xmath232 are categorized into ones with even @xmath55 or with odd @xmath55 : @xmath235 where @xmath236 and @xmath237 is a modified bessel function .
similarly , @xmath238 where @xmath239 since the lowest mode in the energy spectrum is most relevant , we consider the mode where an electron is taken from the @xmath240 landau level and promoted to the @xmath241 landau level without flipping the spin in order to obtain a charge density excitation .
since its kinetic energy is @xmath242 , we will call the mode the @xmath35 mode . as discussed before ,
this mode is not mixed with other modes in the limit of small @xmath3 , i.e. , the matrix @xmath146 is already diagonal . for the lowest charge density excitation
we have the following equation : @xmath243 in the above equation we defined dimensionless matrix elements so that @xmath244 let us denote the solution of the pole equation as @xmath245 from now on .
then the dispersion curve of the lowest charge density excitation is given by @xmath246_{\alpha=1/2,\beta = k } \nonumber \\ & + & \frac{e^2}{\epsilon l_0 } e^{-k^2/2}k { \bigg ( 1-\frac{k^2}{4 } \bigg)}^2\end{aligned}\ ] ] as explained earlier , landau level mixing in the non - zero @xmath3 regime is included by diagonalizing the effective hamiltonian defined in eq.([eq : effh ] ) . using the term @xmath35 mode for the sake of convenience to indicate the lowest charge density excitation in the fully polarized state at a general @xmath3 , we plot their dispersion curves in fig.[polcde ] which shows that the charge density excitation modes do not exhibit any sign of an instability in the parameter range of @xmath3 considered here
. a non - trivial check of our calculations is to make sure that the collective excitations computed within our approximation satisfy the exact kohn s theorem which states that the @xmath35 mode energy must approach @xmath247 as @xmath248 @xcite .
fig.[polcde ] shows that kohn s theorem is satisfied not only for the pure mode but also the mode with landau level mixing in general @xmath3 .
incidentally , the zeeman coupling will not affect the dispersion curves because the spin is not flipped .
following the convention used in the previous sections , we indicate the excitation modes in terms of @xmath6 , the kinetic energy of the mode in units of @xmath15 in the limit of small @xmath3 .
we shall see that , unlike the charge density excitation , the lowest lying spin excitation can be either @xmath37 mode or the lower one of the two @xmath36 modes , depending on the value of @xmath3 .
the @xmath37 mode describes the process whereby an electron in the @xmath249 landau level is demoted to the @xmath250 landau level with its spin reversed , whereas the @xmath36 mode has an electron with its spin flipped in the same landau level . in the latter case , there are two possible modes : spin density excitation within @xmath249 landau level or @xmath250 landau level . at small @xmath3 the @xmath37 mode
the lowest excitation while for large @xmath3 the lowest spin density excitation is a @xmath36 mode .
also , without the zeeman coupling , the spin reversed mode always causes an instability of the fully polarized ( 2:0 ) state .
a determination of the zeeman splitting energy @xmath14 required to make the ( 2:0 ) state stable for general @xmath3 will be one of the main goals when we try to obtain the phase diagram of the spin polarization as a function of @xmath3 and @xmath14 .
the dispersion curves of the pure spin density excitation , however , can be computed in the limit of small @xmath3 without recourse to the actual value of @xmath14 .
we assume that it is large enough to stablize the excitation because the zeeman energy is just a constant shift in this limit .
@xmath73 @xmath251 mode as in the case of the charge density excitation , we first solve the pole equation in the small @xmath3 limit to get the pure mode without any landau level mixing . with matrix elements
@xmath252 = 0\ ] ] the solution is @xmath253\ ] ] where @xmath254_{\alpha=1/2 } = \frac{5}{4 } \sqrt { \frac{\pi}{2}}\end{aligned}\ ] ] and @xmath255_{\alpha=1/2,\beta = k } \nonumber \\ & = & -\frac{1}{2 } \sqrt { \frac{\pi}{2 } } e^{-k^2 /4 } \bigg [ ( 1+\frac{k^2}{2})i_0(\frac{k^2}{4 } ) -\frac{k^2}{2}i_1(\frac{k^2}{4 } ) \bigg ] \end{aligned}\ ] ] the dispersion curve of this @xmath37 mode is plotted in fig.[puresde ] .
@xmath73 @xmath36 modes and the spin wave mode since there are two possible @xmath36 modes , the pole equation becomes a matrix equation even in the small @xmath3 limit : @xmath256 the matrix elements are given by @xmath257 \nonumber \\ & = & \omega- \frac{e^2}{\epsilon l_0 } \big [ 2g_0(\alpha)-\frac{1}{2}g_2(\alpha ) -f_0(\alpha,\beta ) \big]_{\alpha=1/2,\beta = k } \nonumber \\ & = & \omega - \frac{e^2}{\epsilon l_0 } \sqrt{\frac{\pi}{2}}\bigg [ \frac{3}{2 } - e^{-k^2 /4 } i_0(\frac{k^2}{4 } ) \bigg ] , \\
m^{0,0}_{1,1}(k ) & = & \frac{e^2}{\epsilon l_0 } \tilde{v}^{0,0}_{1,1}(k ) \nonumber \\ & = & \frac{e^2}{\epsilon l_0 } \big [ \frac{1}{2 } f_2(\alpha,\beta ) \big]_{\alpha=1/2,\beta = k } \nonumber \\ & = & \frac{e^2}{\epsilon l_0 } \frac{1}{2 } \sqrt{\frac{\pi}{2 } } e^{-k^2 /4 } \bigg [ ( 1-\frac{k^2}{2})i_0(\frac{k^2}{4})+\frac{k^2}{2}i_1(\frac{k^2}{4 } ) \bigg ] \nonumber \\ & = & m^{1,1}_{0,0}(k ) , \\
\mbox{and } \nonumber \\
m^{1,1}_{1,1}(k,\omega ) & = & \omega -\frac{e^2}{\epsilon l_0 } \big [ -\tilde{\sigma}^{1}_{n=1 } -\tilde{v}^{1,1}_{0,0}(k ) \big ] \nonumber \\ & = & \omega -\frac{e^2}{\epsilon l_0 } \big [ 2g_0(\alpha ) -\frac{3}{2 } g_2(\alpha ) + \frac{1}{4}g_4(\alpha ) -f_0(\alpha,\beta)+f_2(\alpha,\beta ) -\frac{1}{4}f_4(\alpha,\beta ) \big]_{\alpha=1/2,\beta = k } \nonumber \\ & = & \omega -\frac{e^2}{\epsilon l_0 } \frac{5}{4}\sqrt { \frac{\pi}{2 } } \nonumber \\ & + & \frac{e^2}{\epsilon l_0}\frac{1}{4 } \sqrt { \frac{\pi}{2 } } e^{-k^2 /4 } \bigg [ ( 3-k^2+\frac{3}{8}k^4)i_0(\frac{k^2}{4 } ) + k^2(1-\frac{k^2}{2})i_1(\frac{k^2}{4 } ) + \frac{k^4}{8}i_2(\frac{k^2}{4 } ) \bigg]\end{aligned}\ ] ] since spontaneous symmetry breaking occurs in the fully polarized state , there must be a spin - wave goldstone mode whose energy approaches the unshifted zeeman splitting energy in the long wavelength limit .
similar to the case of the charge density excitation it is important to check if the spin density excitations computed within our approximation satisfy this exact theorem .
when we take the @xmath248 limit of , we have the following equation : @xmath258 the solutions are @xmath259 this confirms the existence of a goldstone ( spin - wave ) mode .
furthermore it can be shown that the massless spin - wave mode exists with an arbitrary potential and landau level mixing within our approximation .
the dispersion curves of the two @xmath36 modes are also plotted in fig.[puresde ] along with that of @xmath37 mode . as before
, the dispersion curves for general @xmath3 and @xmath14 are obtained from diagonalization of the effective hamiltonian .
we take the dispersion curves at @xmath260 1.0 and 3.0 as examples and plot them in fig.[polsde ] to illustrate the qualitative difference between the small and large @xmath3 regimes .
we will sometimes use the term goldstone mode in order to indicate spin - wave excitation mode for general @xmath3 since its energy approaches @xmath14 as @xmath261 goes to zero .
we will also use the term @xmath37 mode for the lowest excitation for small @xmath3 since it has the lowest energy in the limit of vanishing @xmath3 .
the physical implications and the corresponding phase diagram will be discussed in more details in the later section .
the collective excitations in the spin unpolarized state are computed in this section in complete analogy with the previous section .
first , let us consider the small @xmath3 limit with zero @xmath14 .
the energy of the lowest charge density excitation is @xmath262 -\frac{e^2}{\epsilon l_0 } \tilde{v}^{1,0}_{1,0}(k ) + \frac{e^2}{\epsilon l_0 } 2 \tilde{u}^{1,0}_{1,0}(k ) \nonumber \\ & = & \omega_c + \frac{e^2}{\epsilon l_0}k e^{-k^2 /4 } + \frac{e^2}{\epsilon l_0 } \bigg [ g_0(\alpha ) -\frac{1}{2}g_2(\alpha ) -f_0(\alpha,\beta ) + \frac{1}{2}f_2(\alpha,\beta ) \bigg]_{\alpha=1/2,\beta = k } .\end{aligned}\ ] ] and the energy of the lowest spin density exciation is @xmath262 -\frac{e^2}{\epsilon l_0 } \tilde{v}^{1,0}_{1,0}(k ) \nonumber \\ & = & \omega_c + \frac{e^2}{\epsilon l_0 } \bigg [ g_0(\alpha ) -\frac{1}{2}g_2(\alpha ) -f_0(\alpha,\beta ) + \frac{1}{2}f_2(\alpha,\beta ) \bigg]_{\alpha=1/2,\beta = k}. \end{aligned}\ ] ] the dispersion curves of the above pure modes are plotted in fig.[pureun ] .
we can use kohn s theorem to check if our approximation is reasonable .
fig.[pureun ] shows that the energy of charge density excitation approach @xmath247 as @xmath248 . since the charge density excitation does not show any instability , we will concentrate only on the spin density excitation to obtain the phase boundary which is the critical zeeman splitting energy needed for the stable excitation .
shows the dispersion curves of the lowest spin density excitation for @xmath263 and various values of @xmath14 .
the phase diagram of the state at @xmath0 can be obtained at two levels of sophistication .
the simplest approximation is that of non - interacting electrons . in this case , the phase boundary is given simply by @xmath264 , as shown in .
as we shall see , this is sensible only in the limit of @xmath200 ; interactions modify the phase diagram substantially elsewhere . in the simplest approximation , interaction can be incorporated by comparing the energies of the fully polarized and unpolarized ground states in the hatree - fock approximation , that is , by assuming that the ground state contains either 0@xmath16 and 0@xmath17 landau levels fully occupied or 0@xmath16 and 1@xmath16 .
the contributions of the kinetic energy and the zeeman coupling to the ground state energy are straightforward .
the exchange interaction energy in the hartree - fock approximation can be evaluated in terms of the self - energy defined in the previous section .
the self - energy @xmath265 is the exchange interaction between an electron in the landau level with index @xmath55 and all the other electrons with the same spin .
the exchange interaction energy per particle is then the sum of the self - energies for all electrons divided by two times the number of electrons where the factor of two prevents a double counting .
that is to say , @xmath266 therefore the ground state energy per particle is @xmath267.\ ] ] the average energy of the fully polarized ground state is @xmath268 similarly the average energy of the unpolarized ground state is @xmath269 the phase boundary is given by the solution of the following equation : @xmath270 therefore the critical @xmath271 as a function of @xmath3 is @xmath272 shown in .
the phase diagram obtained by a comparison of the energies of the hartree fock states is not fully reliable for two reasons .
first , it neglects landau level mixing , which is crucial for the issue of interest here .
secondly , it does not allow for the possibility of other states in the phase diagram .
therefore , it is more appropriate to look for instabilities of the two states by asking when one of the collective modes becomes soft .
indeed , a first order phase transition may ( and likely will ) occur even before the collective mode energy approaches zero , but we believe that the phase diagram obtained by considering instabilities ought to be reliable qualitatively and even semi - quantitatively . as noted previously , there is no instability in the charge density wave collective mode in the parameter range considered here . we have determined the onset of the spin density collective mode instability both in the fully polarized and the unpolarized states by varying @xmath3 and @xmath14 , as shown in and .
shows the phase diagram thus obtained .
the following features are noteworthy .
\(i ) the nature of instability is different depending on whether @xmath3 is small or large . at small @xmath3 , where the interactions are negligible and
the physics is dictated by the zeeman energy , the lowest energy spin - density excitation is clearly the @xmath37 mode .
this continues to be the case for @xmath273 ; here the @xmath37 spin - density mode is responsible for the instability of the ( 2:0 ) state , as shown in .
however , for @xmath274 , interactions are sufficiently strong that the spin - wave mode becomes the lowest energy mode and causes the instability . one way to understand why the @xmath36 spin wave mode has lower energy at large @xmath3 than the @xmath37 mode is because whereas the former has an energy equal to @xmath14 in the long wave length limit no matter what @xmath3 , as guaranteed by the goldstone theorem , the energy of the latter is determined by the interactions .
\(ii ) at small @xmath3 , the interactions make the fully polarized ( 2:0 ) state more stable as compared to the non - interacting problem , as evidenced by the fact that the transition out of it takes place at a zeeman energy
_ smaller _ than @xmath275 .
this is precisely as expected from the exchange physics , as also captured by the hartree - fock phase diagram .
\(iii ) the instability in the spin - wave mode occurs through the development of a roton minimum , the energy of which vanishes at certain zeeman energy .
the roton minimum in turn is caused by the landau level mixing , underscoring the important role of landau level mixing at large @xmath3 . without landau level mixing the spin - wave mode
does not show any instability as shown in .
\(iv ) for the unpolarized state , the lowest spin density excitation is the @xmath35 mode at all @xmath3 , the long wavelength limit of the excitation energy of which is fixed to be @xmath276 independent of @xmath3 . here also , the instability occurs through a roton , which becomes deeper as the @xmath3 is increased ( and interactions become stronger ) . since a spin flip is favored by exchange , the energy of the @xmath35 mode decreases with increasing @xmath3 , consistent with the feature that the critical @xmath14 in this case is monotonically decreasing as a function of @xmath3 , as is shown in .
\(v ) at small @xmath3 there is a small region in where the two phases coexist .
this clearly is an artifact of various approximations involved in our calculation , and implies that the actual locations of the phase boundaries can not be taken too seriously . as mentioned earlier , a first order transition is likely to occur _ before _ the roton energy vanishes .
\(vi ) at large @xmath3 and small @xmath14 , the ground state is derived neither from ( 2:0 ) nor from ( 1:1 ) . since we find a _ finite _ wave vector instability in the spin density wave excitation , it is natural to expect that the state here has a spin - density wave state .
further work will be required to establish the nature of this state in more detail .
the principal outcome of our calculations is the phase diagram in which shows the regions where the fully polarized and the unpolarized hatree fock states ( 2:0 ) and ( 1:1 ) are valid .
we believe that it should be possible to investigate the roton minimum in the spin - wave excitation of the fully polarized state as well as its instability in inelastic light scattering experiments .
another situation where similar physics may apply is in the case of composite fermions @xcite at effective filling factor @xmath277 , which corresponds to the electron filling factor @xmath278 . it is easier in this case to see a transition between the fully polarized and the unpolarized states because the effective cyclotron energy for composite fermions is substantially small compared to the cyclotron energy of electrons , which makes it possible to obtain @xmath14 comparable to or larger than the effective cyclotron energy in tilted field experiments .
the critical zeeman energy at the transition was calculated by park and jain by comparing the ground state energies @xcite , in reasonable agreement with the experiments of kukushkin , von klitzing , and eberl @xcite .
park and jain also estimated a mass for composite fermions , the polarization mass " by equating the critical zeeman energy to the effective cyclotron energy .
there is one subtlety though . in the case of composite fermions both the effective cyclotron energy and
the effective interactions derive from the same underlying energy , namely the coulomb interaction between the electrons , and therefore neither the interactions between composite fermions nor a mixing between _ composite fermion _
landau levels can , in principle , be neglected in any realistic limit .
these would provide a correction to the mass obtained in .
however , we note that the mass was reliable to no more than 20 - 30% , and the corrections may be negligible compared to that . interestingly , there is experimental evidence @xcite that the transition between the fully polarized and the unpolarized composite fermion states does not occur directly but through an intermediate state with a partial spin polarization .
murthy @xcite has proposed that this state is a hofstadter lattice of composite fermions , and has half the maximum possible polarization .
it would be interesting to see if similar physics obtains for @xmath0 as well .
in particular , the phase diagram of predicts that for @xmath279 , the transition from the fully polarized state to the unpolarized state as a function of the zeeman energy is not direct but through another , not yet fully identified state .
( we suspect that this may be true at any arbitrary @xmath3 , although not captured by our calculated phase diagram . ) | we consider the effect of interactions on electrons confined to two dimensions at landau level filling @xmath0 , with the specific aim to determine the range of parameters where the fully polarized state is stable .
we calculate the charge and the spin density collective modes in random phase approximation ( rpa ) including vertex corrections ( also known as time dependent hartree fock ) , and treating the landau level mixing accurately within the subspace of a single particle hole pair .
it is found that the spin wave excitation mode of the fully polarized state has a roton minimum which deepens as a result of the interaction induced landau level mixing , and the energy of the roton vanishes at a critical zeeman energy signaling an instability of the fully polarized state at still lower zeeman energies .
the feasibility of the experimental observation of the roton minimum in the spin wave mode and its softening will be discussed .
the spin and charge density collective modes of the unpolarized state are also considered , and a phase diagram for the @xmath1 state as a function of @xmath2 and the zeeman energy is obtained . |
the results of atmospheric , solar , accelerator and reactor neutrino experiments @xcite accumulated in the last decades show that flavor mixing occurs not only in the hadronic sector , as it has been known for long , but in the leptonic sector as well .
this outstanding result has been obtained exploiting a quantum interference phenomenon speculated by pontecorvo in 1957 and named - in analogy with the oscillation of neutral k mesons - `` neutrino oscillation '' @xcite . in its modern formulation ,
neutrino oscillations occur because flavor states generated by weak interactions , i.e. what we call `` electron neutrinos '' ( ) , `` muon neutrinos '' ( ) and `` tau neutrinos '' ( ) , are not stationary states but , indeed , a coherent superposition of states with rest mass @xmath5 , @xmath6 and @xmath7 . as a consequence ,
flavor evolves with time and neutrinos of a given flavor @xmath8 produced from a source at a distance @xmath9 from the detector , might be measured to have a different flavor @xmath10 .
pontecorvo noted in 1967 that , assuming only two flavors ( at that time the tau lepton and the were still unknown ) , the probability of observing a from a pure source of is given by : @xmath11 ) l([km ] ) } { e([gev ] ) } \ , \label{eq : pontecorvo}\ ] ] being @xmath12 the angle that parametrizes a rotation between the stationary state @xmath13 and @xmath14 and the flavor states and , while @xmath15 represents the squared mass difference between @xmath5 and @xmath6 .
neutrino oscillations occur if neutrinos are massive particles , if their masses are non degenerate ( @xmath16 ) and if @xmath17 .
nature has fulfilled for us all these conditions and neutrino oscillations are actually observables in many systems : in neutrinos produced from the sun , in neutrinos generated by cosmic ray interactions with the earth atmosphere and , more recently , in artificial neutrinos produced by nuclear reactors and accelerators .
all experimental results gathered so far are mutually consistent if we assume just three active neutrinos - the ones observed in a direct manner in 1956 ( @xcite ) , 1962 ( @xcite ) , and 2001 ( @xcite ) - and if we describe such flavor states as superposition of three mass eigenstates labeled @xmath5 , @xmath6 and @xmath7 , respectively @xcite .
as for the case of quarks , mass eigenstates and flavor eigenstates do not coincide and the unitary matrix that transform the former to the latter is non - diagonal . in the quark case , this matrix
is called the cabibbo - kobayashi - maskawa ( ckm @xcite ) matrix and can be parametrized by three euler angles @xmath18 and @xmath0 and one `` complex phase '' @xmath19 is actually a real number that describes the phase factor @xmath20 appearing in the mixing matrix . ] , the first angle @xmath21 being the `` cabibbo angle '' @xmath22 . in the quark sector we know for sure that @xmath19 is different from zero : it causes cp violation in hadrons as observed e.g. in the neutral kaon decays since 1964 @xcite or in the neutral b decays since 2001 @xcite . for leptons , the corresponding matrix is sometimes referred as the pontecorvo - maki - nakagawa - sakata ( @xmath23 ) mixing matrix @xcite . if neutrinos are dirac particle as the quarks - i.e. neutrinos and antineutrinos are different particles - @xmath23 is , again , a @xmath24 unitary matrix and its degrees of freedom are the same as for the ckm ( three euler angles bounded between 0 and @xmath25 and one complex cp violating phase @xmath19 ranging from 0 to @xmath26 ) .
two additional complex phases appear if neutrinos are majorana particles , i.e. if neutrinos and antineutrinos are the same particle ; these `` majorana phases '' are , however , unobservable in oscillations @xcite and will not be considered hereafter .
moreover , @xmath23 is not necessarily unitary : in future - and in a way similar to the ckm - direct unitarity tests will likely become an active subject of empirical investigation .
neutrino oscillations are a powerful tool to determine the squared mass differences of the rest masses of neutrinos and the four parameters @xmath18 , @xmath0 and @xmath19 .
the experimental results obtained so far point to two very distinct mass differences , @xmath27 ev@xmath28 and @xmath29 ev@xmath28 @xcite . in jargon , @xmath30 is called the `` solar mass scale '' because it drives oscillation of solar neutrinos , but , of course , if the energy of the neutrino and the source - to - detector distance are properly tuned , it can be observed also using man - made neutrinos , e.g. reactor neutrinos located about 100 km from the detector ( as it was the case for kamland @xcite ) . similarly , the standard framework predicts that atmospheric neutrinos mainly oscillate at a frequency that depends on @xmath31(``atmospheric scale '' ) .
accelerator neutrinos can see ( actually , saw in k2k @xcite and minos @xcite ) the same effect using neutrinos of energy around 1 gev and source - to - detector distances ( `` baselines '' ) of a few hundreds of km ( 250 km in k2k and 730 km in minos ) .
only two out of the four parameters of the leptonic mixing matrix are known : @xmath32 degrees and @xmath33 degrees @xcite .
surprisingly , these values are much larger than for the ckm ( @xmath34 and @xmath35 ) .
the other two parameters , @xmath0 and @xmath19 , are still unknown : for the mixing angle @xmath0 direct searches at reactors ( the chooz @xcite and palo verde @xcite experiments ) combined with solar neutrino data give the upper bound @xmath36 at 90% cl , whereas for the leptonic cp - violating phase @xmath19 we have no information whatsoever . in the standard framework ,
atmospheric neutrinos mainly oscillate from @xmath37 to @xmath38 and vice - versa with the @xmath31-driven frequency .
all other transitions are sub - dominant and depend on the parameter @xmath0 .
the probability of @xmath39transitions at the atmospheric scale is , at leading order , @xmath40 $ ] , so its smallness is a consequence of the smallness of @xmath0 in @xmath23 .
all these considerations hold in the standard framework , i.e. under the assumption that atmospheric neutrinos disappear due to @xmath41 oscillation .
this is a well - grounded assumption , which is however being tested in a very straightforward manner ( observation of charged - current interactions ) at the cern - to - gran sasso ( cngs ) neutrino beam @xcite .
reactor antineutrinos are a classical tool to probe the size of @xmath0since they have an energy of a few mev and the ratio @xmath42 matches the atmospheric scale , i.e. @xmath43 , at baselines of about 1 km . if they oscillate toward tau or muon antineutrinos , an apparent flux reduction in the detector will be observed .
currently , the best direct experimental limit on @xmath0comes from the chooz reactor experiment .
a world limit can be derived @xcite by a full @xmath44 analysis of all the neutrino oscillation experiments , see tab .
[ tab : th13 ] .
such world limit provides a slightly looser value than the chooz limit .
it is an indication that the best fit for @xmath0might be different from zero @xcite , although with small statistical significance ( @xmath45 ) .
@xmath46 a preliminary analysis of the minos experiment @xcite shows a @xmath47 excess of -like events in the far detector , that could be interpreted as a manifestation of non - zero value of @xmath0 .
very recently , however , the super - kamiokande collaboration @xcite presented an updated analysis of its full atmospheric neutrino sample : their data do not support the above hint for @xmath0@xmath48 . a non - zero value of the mixing angle @xmath0plays an essential role in neutrino oscillation physics .
since late 90 s , it has been realized that the oscillation @xmath39at the atmospheric scale has a very rich structure .
its leading term depends on @xmath49 but other subdominant terms suppressed by power of the ratio @xmath30/@xmath31 , contain explicit dependence on the cp phase @xmath19 .
moreover , neutrino propagation from the source to the detector through the earth is affected by matter effects similar to the ones perturbing the free streaming of the solar neutrinos @xcite .
it is a remarkable fact that such perturbations are sensitive not only to the overall size of @xmath31 but also to its sign , i.e. to the relative ordering of @xmath50 and @xmath7 . in the most used parametrization of the leptonic mixing @xcite @xmath51 by definition and @xmath7
can be either larger ( `` normal hierarchy '' ) or smaller ( `` inverted hierarchy '' ) than @xmath6 , giving @xmath52 and @xmath53 respectively .
again , this information might be retrieved from high precision experiments measuring both @xmath39 and @xmath54transitions .
the unique role of @xmath39 and its cp conjugate @xmath54 at the atmospheric scale is a consequence of the fact that the ratio @xmath30/@xmath31 is not too small ( about 1/30 ) and three - family interference effects are sizable .
hence , subdominant three - family interference amplitudes are observables with neutrinos in analogy to the @xmath55 system for quarks .
again , this occurrence is fortunate and purely accidental . before 2001 , several other values , much smaller than @xmath56 ev@xmath28 were still experimentally allowed .
the demonstration that @xmath30 is in the ballpark of @xmath57 ev@xmath28 is considered the real inception of the precision era of neutrino physics , since it opened up the possibility to measure the whole of @xmath23 with artificial sources , save , once more , a finite value for @xmath0 . to give a feeling of the complexity of the task of studying subdominant @xmath39transitions we report below the approximate formula that holds when matter effects are not negligible but matter density is constant along the neutrino path . in this case
the transition probability @xmath58 ( @xmath59 ) can be written as @xcite : @xmath60 where @xmath61 refers to neutrinos and antineutrinos , respectively and @xmath62=\pm 2\sqrt{2}g_fn_ee_\nu=7.6 \times 10^{-5}\rho[g / cm^3]e_\nu[{\rm gev}]$ ] is the electron density in the material crossed by neutrinos .
the coefficients of the two equations are : @xmath63 , + @xmath64 , + @xmath65 .
the @xmath39 transitions are driven by the @xmath0term which is proportional to @xmath66 and @xmath67 could be strongly influenced by the unknown value of @xmath68 and @xmath69 .
given the complexity of the @xmath39 transition formula it will be very difficult for pioneering experiments to extract all the unknown parameters unambiguously . correlations are present between @xmath0and @xmath19 @xcite .
moreover , in absence of information about the sign of @xmath70 @xcite and the approximate @xmath71 $ ] symmetry for the atmospheric angle @xcite , additional clone solutions rise up . in general , the measurement of @xmath72 and @xmath73 will result in eight allowed regions of the parameter space , the so - called eightfold - degeneracy @xcite .
it can be solved only combining several measurements at different energies or baseline , or exploiting channels different from @xmath39 , as @xmath74 .
this is the reason why it is commonly believed that the next generation of experiments will mainly focus on the very evidence of appearance at the atmospheric scale , while the task of precision measurement , including cp violation and determination of the mass hierarchy , will take a further round of facilities @xcite . at present
, there is only one accelerator experiment in data taking - actually in the startup phase - that has been optimized for the search of non - null @xmath0 .
it is the t2k @xcite experiment in japan , which is expected to improve by an order of magnitude the limit from chooz exploiting a new beam from jparc to the pre - existing far detector of superkamiokande ( fig .
[ fig : t2k ] ) .
the relevance of @xmath0search to ground the future of the field is such that it boosted a large number of proposals to improve chooz with novel detectors at reactors . among the @xmath75
original proposals , three of them survived and are either in commissioning or in construction phase .
the most advanced is double - chooz @xcite , in the original chooz site in france , which makes use of a larger far detector and of a new near detector identical to the far detector for flux normalization ( a near detector was not available at the time of chooz ) .
double - chooz will start data taking in a few months with the far detector only ; the near detector will be available after 2011 .
the other reactor experiments under construction are daya - bay @xcite in china and reno @xcite in korea . in particular , the basic experimental layout of daya bay consists of three underground experimental halls , one far and two near , linked by horizontal tunnels .
each near hall will host two 20 t gadolinium doped liquid scintillator detectors , while the far hall will host four such detectors .
it is the most aggressive layout to push systematics below 0.4% .
other experiments currently in data taking exhibit some sensitivity to @xmath0 .
they are minos , whose measurement has already been mentioned and opera @xcite at the gran sasso laboratories ( lngs ) of the italian institute for nuclear research ( infn ) . in the original plans minos was supposed to improve by a factor of two the limit of chooz . however , considering the achieved precision and the fact that minos has swapped to antineutrinos , it is unlikely that it will be able to improve significantly the current chooz limit .
also opera , which exploits the above - mentioned cngs beam , is expected to improve the chooz limit @xcite , although with a timescale that is not competitive with t2k .
on the other hand , in the occurrence of large @xmath0(see below ) the measurement from cngs would be of great value , being affected by very different systematics with respect to t2k and double - chooz .
[ fig : t2k ] the observation of the extraordinarily large neutrino mixings when compared to naive cabibbo - like expectations from quarks , the possibility of majorana - like couplings and the extraordinary small mass spectrum indicate that neutrinos might have unique features with respect to other elementary fermions .
this is why non - standard scenarios have been deeply investigated in literatures , often fostered by experimental anomalies as the one of lsnd that dates back to 1995 @xcite or , more recently , the `` low energy anomaly '' of miniboone @xcite .
perspectives outside the standard three - family framework are discussed in sec [ sec : nonstandard ] .
it is of great practical interest to consider the expected sensitivities of accelerator and reactor experiments in the near future .
it is , clearly , a quite approximate exercise since most of these experiments are not in regular data taking and therefore , delays or re - scheduling might be possible .
this holds particularly for t2k , whose startup will be at very low power ( 0.1 mw ) compared with the design one of 0.75 mw and a progressive ramp - up is foreseen .
[ fig : th13vstime ] @xcite shows the evolution of the @xmath0sensitivities as a function of time gathering all the information we have presently in our disposal . even with such caveat , from the plot one can easily derive that in the next 5 years or so the @xmath0parameter will be probed with a sensitivity about 25 times better than the present limit .
the figure includes also the startup of nova @xcite , a new facility in the us based on the existing numi beam from fermilab to minnesota and its upgrade ( see sec .
[ sec : new_facility ] ) .
in fact , the @xmath0sensitivity of t2k depends on the unknown @xmath19 parameter and on the mass hierarchy , i.e. @xmath69 ; a more detailed comparison of the time evolution of the t2k and reactors sensitivities has to take into account that t2k will actually provide a band of excluded values of @xmath0and not just a single point .
this is shown in fig .
[ fig : time - evolution ] @xcite . from the plot
some considerations can be taken : * the double - chooz reactor experiment is very competitive in the first years of operation , when , however , the information coming from its near detector will likely not be available .
* t2k will be dominating our knowledge of @xmath0for long time and the evolution of its beam power is the most critical issue in the extrapolation of fig . [
fig : th13vstime ] .
it is impossible to state now how the the jparc neutrino beam line evolves and when it reaches the top value of 0.75 mw , since the facility is based on an entirely new accelerator complex .
therefore , the sensitivity shown here is no more than an educated guess . *
scheduling is critical for daya bay , too .
the achievement of tiny systematic errors claimed by the experiment will likely be the most relevant issue to be clarified .
this discussion is based on the limits that can be achieved by each experiment . in case of @xmath0 in the reach of those experiments , their information will be truly complementary to measure the actual value of the parameter : for a discussion under this hypothesis see @xcite .
as already mentioned , there is consensus @xcite on the fact that none of currently running facilities - including the experiments in startup phase as t2k , double - chooz or daya - bay - will be able to extract all missing parameters of the @xmath23 even in the most optimistic case of @xmath0close to current limits . on the other hand , what has prompted a rather heated debate is the definition of a `` minimal '' facility that might be able to address the challenge of extracting such parameters by a detailed study of subdominant transitions at the atmospheric scale . defining the `` optimal facility '' without knowing - at least approximately - the size of @xmath0is a hopeless exercise but
several non - trivial considerations can be done assuming ( or excluding ) that an evidence of subdominant @xmath39transitions is within the reach of t2k .
such considerations have also driven important r&d efforts toward novel sources of neutrinos in the gev range and , correspondingly , toward new detectors for their observation at the far location .
the technologies for neutrino sources that are considered promising for the next generation of facilities can be grouped in three main streams @xcite called , in jargon , `` superbeams '' , `` beta beams '' and `` neutrino factories '' @xcite . coordinated r&d efforts have been made around the world to raise these technologies to their adulthood but it is clear that a positive result in the quest of @xmath0within the next 2 - 5 years will prompt a tremendous boost toward their realization .
the status of the art on the eve of the t2k run is summarized below , with special emphasis to their implementation in the european framework of infrastructures . since the 1962 experiment of lederman et al .
@xcite , accelerator neutrinos in the gev ballpark have only been produced from the decay in flight of pions .
this 40-year old technology @xcite has progressed together with high intensity proton accelerators and , in the last decade , allowed for the construction of long - baseline @xmath77 beams to explore the leading @xmath41 transition at the atmospheric scale ( fig .
[ fig : longbaseline2 ] ) .
long baseline experiments have also been possible thanks to the large size of the atmospheric mixing ( @xmath78 ) , which partially compensates the loss in intensity due to the large source - detector distance . at cern , it allowed for the re - use of the sps , formerly employed for short - baseline experiments at the wanf ( chorus @xcite and nomad @xcite ) to feed a 17 gev beam from cern to gran sasso . moving from leading to subleading oscillations ( @xmath39transitions ) ,
the bonus of the large mixing angle disappears ( @xmath0@xmath79 @xmath80 ) and the intensity is no more sufficient to explore these transitions at a level better than chooz .
such considerations hold also for other long - baseline beams in the world that played a role in the current round of experiments , as the already mentioned numi beam from fermilab to minnesota and the k2k beam from kek to superkamiokande ( see fig . [
fig : longbaseline2 ] ) . in japan
, it would have been impossible to re - use the very low intensity k2k beam in this new phase . as a result ,
the construction of a new facility has been pursued since 2003 , leveraging the enormous investment ( about 1 billion ) made for jparc and aimed to a much broader community than neutrino physicists .
it brought the t2k project into existence without the need of building either a dedicated proton accelerator or a new far detector : the former leverages the 50 gev general purpose synchrotron of jparc complemented by a dedicated neutrino beamline , the latter the existing superkamiokande detector ; although the additional investment for the neutrino beamline was demanding ( about 200 m$ ) , it comes to no surprise that the t2k project is by far the most advanced among the experiments conceived to improve the current limits of @xmath0 .
its running schedule has already been discussed in sec .
[ sec : intro ] .
both us and europe have neutrino facilities much more powerful than the k2k beamline .
the numi beam might be operated up to 0.4 mw , i.e. at 40% of the maximum power achievable by t2k .
unfortunately , there is no far detector in the us that can exploit such power since minos is too dense to perform precision measurements of appearance .
the construction of a dedicated 15 kton detector ( nova @xcite ) , started on may 2009 and located 810 km from fermilab ( see fig.[fig : nova ] ) brings the timeline of the first physics results a few years after t2k and without significant improvements on the knowledge of @xmath0but with some marginal sensitivity to the mass hierarchy due to the large baseline ( 810 versus 292 km ) .
however , the existence of nova together with a positive result from t2k could ground a second generation facility , based on a new neutrino beam fed by a 2 mw , 8 gev , proton accelerator ( `` project x '' @xcite ) .
in fact , as a part of the nova project , fermilab plans an upgrade of numi based on the exploitation of the fermilab `` recycler ring '' after the shut - down of the tevatron .
the recycler ring will be used as a pre - injector of the fermilab `` main injector '' reducing the cycle time ( fig.[fig : fermilab ] ) .
it needs , however , the construction of a dedicated transfer line and the upgrade of the radio - frequency system in both the recycler and in the main injector .
this upgrade @xcite could finally bring numi to an peak power comparable to t2k : 0.7 mw and @xmath81 protons - on - target per year ( pot / y ) at 120 gev . in europe ,
the perspectives for a high intensity neutrino experiment based on superbeams are entangled with the evolution of the cern acceleration complex and , in particular , of the injection system of the lhc .
the cngs at nominal intensity can be operated to accumulate @xmath82 pot / y at an energy `` nova '' pot / y .
] of 400 gev . in the last few years , particularly in the framework of the cern paf ( `` proton accelerators for future '' ) working group , it has been investigated @xcite the possibility of increasing the intensity of the cngs both using present facilities and , on a longer timescale , exploiting an upgrade of the acceleration complex ( fig.[fig : cern ] ) . the ultimate cngs performance
are actually limited by the injection from the 50-year - old proton synchrotron ( ps ) . in this scenario ( cngs as the only user of the sps at cern beyond the lhc )
, the facility could deliver up to @xmath83 pot / y ( @xmath84 nova pot / y ) . at a longer timescale ( @xmath852016 ) , the replacement of the ps with a new 50 gev synchrotron ( ps2 @xcite ) might surpass these limitations , provided an appropriate upgrade of the sps radio - frequency system .
it would bring cngs to a maximum intensity ( cngs as only user of the sps beyond the lhc ) of @xmath86 pot / y ( @xmath87 nova pot / y ) .
it is , therefore , a situation that closely resembles the one of the numi upgrade but unfortunately , its realization can not be anticipated before 2017 due to the priority to the running of the lhc and the corresponding funding limitations from cern for the construction of the new injectors . at that timescale
, it would be more interesting to consider a high - flier option , specifically suited for the exploration of cp violation and the mass hierarchy in case of positive result from t2k - along the line of the us `` project x '' .
the rationale of this approach is that a multi - mw proton source can , in principle , serve for a variety of needs and feed a rich physics programme ranging from neutrino physics to muon and kaon physics and to physics with radioactive ion beams @xcite .
this option has been considered at cern in the framework of the construction of a pre - injector for ps2 : the superconducting proton linac ( spl ) .
the pre - injector , feeding ps2 with 2 - 4 gev protons , has been conceived in year 2000 as a 2.2 gev accelerator for a future neutrino factory based at cern and its design was based on the recycling of the radio - frequencies formerly employed at lep @xcite . over the time
, the design has been updated and currently points toward a 3.5 - 4 gev , 4 mw system with dedicated equipment based on bulk niobium cavities @xcite .
this proton energy might feed a superbeam producing neutrinos at a rather low energy ( about 300 mev ) that could be exploited by a very massive detector located not far from cern ( cern - to - frejus @xcite ) .
the technology of the superbeam is considered the most straightforward way to explore leptonic mixing in case of positive result from t2k and/or double - chooz and , therefore , it is viewed as a solid option in japan ( tokai to hyperkamiokande - see below ) and us ( fermilab / project x to dusel ) @xcite .
unfortunately , it also shows evident limitations : * it is not a `` pure '' source of neutrinos of a given flavor , being plagued by the @xmath88 produced by the decay - in - flight of the kaons and of the muons .
when seeking for subdominant @xmath39transitions , the systematics on the knowledge of the @xmath88 contamination will likely be the main limitation for a precise determination of cp violation in the leptonic sector @xcite ; * it produces mainly @xmath77 and , therefore , the leptonic mixing is studied through @xmath39and its cp conjugate @xmath54 , i.e. looking for _ electrons _ in the final state instead of muons .
it thus requires huge , low density detectors to be hosted in underground sites ( see below ) ; * in a superbeam , cp violation appears as an asymmetry between the probability of transition @xmath39and its cp conjugate @xmath54 . at low proton energies ,
the production yield of @xmath89 is suppressed with respect to @xmath90 .
this suppression , together with the suppression due to the cross section ( @xmath91 ) , makes the antineutrino run much more time - consuming that the neutrino run .
a multi - mw proton accelerator and the corresponding neutrino beamline is considered feasible in less than a decade @xcite but it requires a substantial effort in accelerators and targetry , a major resource investment ( of the order of 500 m ) and feedback from the operation of neutrino facilities in the sub - mw range ( firstly t2k ) . * far detectors for the superbeams * most of the information gathered at the atmospheric scale on leptonic mixing come from the study of muon neutrino disappearance in natural ( atmospheric ) and artificial ( accelerator ) sources . the use of superbeam forces experimentalists to build detectors that are specifically designed for the identification of electron neutrinos .
@xmath92 appearance is extremely challenging compared with @xmath37 disappearance because prompt electromagnetic showers must be distinguished from neutral - currents events where electromagnetic showers occur through the production of @xmath93 . a low density , high granularity target is mandatory to allow showers to develop and to perform a precise kinematic reconstruction in order to separate effectively neutral - current ( nc ) from charged - current ( cc ) events .
unfortunately , only a few of the techniques currently applied for electromagnetic calorimetry can be scaled to a size suitable for neutrino oscillation physics ( at least tens of ktons ) .
actually , due to the current constraints to the size of @xmath0 , it is believed that even in the occurrence of a clear evidence for @xmath39oscillations at t2k , the size of the detectors should be upgraded by nearly an order of magnitude with respect to the current generation of experiments .
thanks to the experience of superkamiokande , japan maintains a leadership in water cherenkov detectors and the proposal of building a new very massive detector of about 500 kton fiducial mass is considered the most plausible upgrade of t2k ( the `` megaton - size '' hyper - kamiokande detector @xcite ) , possibly complemented with a second detector in korea @xcite .
similar options are considered in us for the new dusel underground laboratory in south dakota , and in europe for a possible major extension of the frejus labs .
the only two other viable technologies that are considered scalable to such size @xcite are the ones of the liquid argon detectors and of liquid / plastic scintillators . c. rubbia proposed the use of liquid argon time projection chambers ( lar tpc ) since 1977
@xcite and infn maintains a world leadership in the lar technology thanks to the results of the icarus collaboration that built such detectors up to a size of 600 tons @xcite .
due to its finer granularity and superior quality in reconstructing and identifying particles and rejecting nc events , lar can achieve sensitivities to @xmath88 appearance comparable or better than superkamiokande with smaller masses ( in the 10 kton range ) . to catch up with the megaton - size water cherenkov detectors , lar tpc s should reach fiducial volumes of about 100 kton : an extrapolation of about two order of magnitudes with respect to the t600 , which necessarily requires a very intensive r&d .
this phase might be substantially reduced following a modular approach , where the 10 - 20 kton initial mass can be successively enlarged as needed . * synergies with non - accelerator astroparticle physics * bringing liquid detectors to this immense size requires experimental halls that do not exist in any part of the world .
if the technology of superbeams is chosen as a reference for the post - t2k phase , the gran sasso laboratories will not be able to host any far detector . in this framework , an intermediate size ( 20 kton ) detector in italy
could only be hosted by a new shallow - depth laboratory , as proposed in @xcite .
on the other hand , large liquid detectors with a fiducial mass significantly bigger than superkamiokande come with a bonus of a physics programme much broader than neutrino oscillations @xcite and , therefore , are at the focus of important r&d programmes worldwide .
if located underground or - at least for lar - even at a relatively shallow depths ( @xmath94 m.w.e .
@xcite ) , they could improve significantly the current measurements on proton decay , supernovae , solar and atmospheric neutrinos and , hence , they would represent a major step forward in experimental astroparticle physics . in 2007
the european union has approved a design study ( laguna @xcite ) aimed at evaluating extensions of the present deep underground laboratories in europe ( boulby , canfranc and frejus ) in order to host such facilities ; it also considers the creation of new laboratories in the following regions : caso ( italy ) , pyhasalmi ( finland ) , sierozsowice ( poland ) and slanic ( romania ) .
the laguna ds is mainly oriented to infrastructure studies for the underground sites since these new halls pose major engineering problems . in parallel , the physics case and the technological challenges for the detectors to be hosted are pursued by most european countries following three baseline technologies : water cherenkov ( the 500 kton mass memphis @xcite detector ) , liquid scintillator ( the 50 kton lena @xcite detector ) and liquid argon tpc s ( the 100 kton glacier @xcite lar tpc ) .
clearly , the choice of an european laboratory as a possible hosting site would boost enormously the superbeam option although ( see below ) the beta beams could profit of this unique opportunity , too .
production , acceleration and stacking of high intensity muon beams for muon colliders have been envisaged since the 60 s and it has been noted very early that their decays might produce useful beams of @xmath37 and @xmath96 ( exploiting @xmath97 decays into @xmath98 ) or @xmath99 and @xmath92 ( @xmath100 decays into @xmath101 ) . however , realistic layouts to get intense neutrino sources have become available only in recent times @xcite . in the modern formulation of the `` neutrino factory '' concept ,
muons are created from an intense pion source at low energies , their phase space compressed to produce a bright beam , which is then accelerated to the desired energy and injected into a storage ring with long straight sections pointing in the desired direction . in 1997 s. geer @xcite noted that this source could be ideal to study @xmath102 oscillations at the atmospheric scale , i.e. the t - conjugate of the channel observed in superbeams ( @xmath39 ) . since @xmath100 decay into @xmath103 , it is possible to investigate @xmath102 oscillations seeking for the appearance of @xmath97 from @xmath37 cc events ( `` wrong sign muons '' ) , provided that we are able to separate these events from the bulk of @xmath100 ( `` right sign muons '' ) coming from unoscillated @xmath104 .
a. de rujula et al .
@xcite underlined that the simultaneous exploitation of @xmath97 and @xmath100 decays would be an ideal tool to address cp violation in the leptonic sector , with outstanding performance compared with pion - based sources .
the only condition to be fulfilled was a large ratio @xmath30/@xmath31(see sec . [
sec : intro ] ) and , of course , the finite size of @xmath0 . the results from kamland and sno @xcite confirming @xmath105 boosted enormously geer s proposal , together with the above - mentioned proposals for superbeams . moreover , geer s ideas resonated with the needs of the muon collider accelerator community , who appreciated the possibility of a strong physics - motivated intermediate step before facing the enterprise of the muon collider itself .
actually , motivations were so strong that the early ideas of building the neutrino factory ( nf ) as a front - end of the muon collider have been dropped and current designs are all optimized for neutrino physics .
it mainly implied a relaxation of the ultra - challenging constraints on muon cooling needed for the construction of the collider @xcite . in spite of this , the realization of the neutrino factory still represents a major accelerator challenge compared with superbeams .
it is met through a world - wide r&d programme @xcite ; in europe this programme is especially fostered by uk . among the nf - oriented projects we recall mice @xcite at the rutherford appleton laboratories ( ionization cooling ) , harp @xcite at cern ( hadroproduction for the front - end proton accelerator ) , merit @xcite at cern ( targetry ) , emma @xcite at daresbury ( fixed - field alternating - gradient accelerators ) and the mucool r&d @xcite at fermilab ( radio - frequency and absorbers ) .
moreover , the nf has to be seeded by a very powerful low - energy proton accelerator ( 4 mw ) ; its realization requires similar r&d as for the superbeams , although its optimal energy lays in the few - gev range ( e.g. the aforementioned spl ) .
current designs aim at 10@xmath106 muon decays per year running with a muon energy of 50 gev , although more realistic scenarios suggest decay rates in the ballpark of @xmath107 decay / y and with energies in the 20 - 50 gev range ( corresponding to neutrinos in the 10 - 30 gev range ) @xcite . after the work of the international scoping study ( iss ) @xcite , there is a rather widespread consensus on the fact that the neutrino factory can be considered the most performing facility for the determination of @xmath0 , cp violation and the mass hierarchy . with respect to superbeams , they profit of much smaller systematics in the knowledge of the source and much higher energies ( i.e. statistics , due to the linear rise of the deep - inelastic @xmath77 cross section with energy ) .
in fact , the energy is so high that for any realistic baseline ( @xmath108 km ) the ratio l / e will be off the peak of the oscillation maximum ) when @xmath109 .
this situation holds also at cngs , where opera and icarus are located at 730 km but the neutrino energy can not be lowered below o(10 ) gev to stay beyond the kinematic threshold for tau production . in this case , the oscillation peak is at @xmath110 .
similarly , at a neutrino factory with a detector located 3000 km far from the source , the peak is at 6 gev .
] at the atmospheric scale .
this condition is the main cause of the occurrence of multiple solutions when the mixing parameters are extracted from the physics observables , i.e. the rates of appearance of wrong sign muons . in jargon , this issue is dubbed the `` degeneracy problem '' @xcite and it has already been mentioned in sec . [
sec : intro ] .
it also affects other facilities than nf but it is particularly severe for experiments running off the peak of the oscillation probability .
the iss suggests as an ideal solution the positioning of two detectors at baseline around 3000 and 7000 km .
an alternative to the second 7000 km detector could be the detection of @xmath74 at baseline around 1000 km ( `` silver channel '' ) @xcite ( see below ) .
the exploitation of the silver channel , moreover , is useful to investigate the occurrence of non - standard interactions in the neutrino sector @xcite .
although the superior physics reach of the neutrino factory is nearly undisputed and no evident showstoppers have been identified , the r&d needed to build this facility remains impressive . in turn ,
the time schedule for its realization and the cost estimate are vague ( @xmath111 after an investment of 1 - 2 billion$ ) . on the other hand , a clear indication on the size of @xmath0will enormously boost the interest of particle physics on this technology .
neutrino factories are virtually capable of performing real precision physics on the leptonic mixing in a way that resembles the former physics potential of the b - factories on quark mixing . * far detectors for the neutrino factories * at the neutrino factory , experimentalists search for the appearance of multi - gev @xmath77 , a signal with a very clear topology even in high density , low granularity detectors . as a result , the baseline technology for nf far detectors are magnetized iron calorimeters with masses of the order of several tens of ktons , although other low - density options have also been investigated .
magnetized iron calorimetry is a classic technique in neutrino physics , brought to the multi - kton size by the minos collaboration .
similar techniques based on resistive plate chambers instead of plastic scintillators have been proposed in the past ( monolith @xcite ) for a second generation atmospheric neutrino experiment in the hall c of lngs - the same hall that presently hosts opera and borexino @xcite - and , more recently , in a new underground lab close to bangalore in india ( ino @xcite ) .
they have been proposed to measure the sinusoidal patter of the neutrino oscillations with multi - gev @xmath77 and @xmath104 produced in the earth atmosphere , improve the knowledge of the leading atmospheric parameters @xmath112 and @xmath113 and , at least for large values of @xmath114 ( @xmath115 @xcite ) , determine the sign of @xmath31 through matter effects .
the main challenge with respect to minos is represented by the compelling request on the muon charge identification ( @xmath116 of misidentification probability ) in order to suppress the bulk of `` right sign muons '' .
current r&d are focused on the optimization of the granularity and on the technology choice for the active detector to achieve optimal charge identification and pion rejection at the lowest possible visible energy @xcite .
clearly , hall c of lngs is an ideal place to host a 30 - 40 kton detector .
a neutrino factory based at the rutherford labs ( ral ) , along the line pursued by uk , could be suited for a lngs far - detector , although the ral - lngs distance is about 1500 km ( i.e. a factor of two smaller than the optimal one suggested by the iss ) . in the context of a cern - based nf ,
lngs is too close for an optimal sensitivity to cp violation or mass hierarchy . on the other hand ,
the lngs location is interesting to study the `` silver channel '' ( @xmath74 ) .
the silver channel represents a unique opportunity to exploit technologies mastered within infn @xcite : the appearance of the can be observed by nuclear emulsion based detectors in a way similar to the one currently employed by opera at cngs . in the neutrino factory framework , the detectors should be suited for the identification of the tau lepton , provided that they are able to measure the wrong - sign muons .
an opera - like detector @xcite in the 5 kton mass ballpark or a lar tpc with comparable mass complemented by a muon spectrometer are considered as viable options .
the only drawback might be due to the early construction of the indian laboratory , which is located at about 7000 km both from ral and from cern .
if this detector is built , the use of the `` silver channel '' as a way to solve the degeneracies in a european nf will be marginal compared with the baseline iss scenario ( two iron detectors at 3000 and 7000 km , respectively ) .
the enormous progress in the technology of radioactive ion beams has led p. zucchelli @xcite to the proposal of a neutrino facility based on the decay in flight of @xmath117-unstable ions ( for a full review see @xcite ) . unlike the nf , these `` betabeams '' ( bb ) are pure sources of or , in the occurrence of @xmath118 decays , of . hence , they are ideal tools to study @xmath102transitions and their cp - conjugate .
they share with nf the nearly complete absence of systematics in the knowledge of the source with the bonus of no `` right sign muon '' background ( no @xmath37 in the initial state ) . on the other hand , due to the very different mass - to - charge ratio between muons and @xmath117-unstable ions ,
the energy of the neutrinos are typically much smaller than what can be obtained at the nf .
the original proposal of @xcite was tuned to leverage at most the present facilities of cern - the ps and the sps - and it was based on @xmath119 and @xmath120 as and sources respectively .
it goes without saying that the betabeam triggered the interest of nuclear physics community , which was offered a stimulating synergy with the neutrino programme at cern . as a result ,
such proposal @xcite was studied in a systematic manner within the framework of the eurisol design study ( task 12 : beta beam aspects ) .
the study aimed at @xmath121 antineutrinos per year from @xmath119 and @xmath122 neutrinos per year from @xmath120 .
the outcome was extremely encouraging , except for the production of @xmath120 , which can not attain the needed rate using standard methods and medium - intensity proton accelerators ( 200 kw ) .
along this line , the most straightforward alternative would be direct production on mgo based on a 2 mw , few mev , proton accelerators , which are quite similar to the linacs that have to be built for the international fusion materials irradiation facility @xcite . in this case
, the bb would partially miss the advantage of a low - power front - end compared with the multi - mw accelerators needed for the superbeams and for the nf , although a few tens of mev mw accelerator is anyway a much simpler machine than a few gev mw linac . from the point of view of the physics performance ,
an additional weakness stems from the fact that the sps is able to accelerate these ions up to relatively low energies , so that the corresponding emitted neutrinos are just in the sub - gev range .
this impacts on the choice of the detectors ( see below ) and on the cross sections , which are highly depleted . to improve the performance of the eurisol beta beam several alternatives to the sps have been considered : a high energy sps ( `` supersps '' @xcite ) accelerating protons up to 1 tev ( a machine originally envisaged for the energy and luminosity upgrade of the lhc ) or even the lhc itself @xcite .
these configurations improve the sensitivities to cp violation and the mass hierarchy at the expense of a large increase of costs : large investments are needed especially to the construction of the decay ring since the length of the ring depends from the magnetic rigidity of the circulating ions , which is proportional to their lorentz @xmath123 factor , and for the compensation of potential flux reduction due to the longer lifetime of the ion in the laboratory frame . in 2006 , c. rubbia et al .
@xcite proposed the use of 8 and 8 as neutrino sources noting that these isotopes could be produced in a multiturn passage of a low - energy ion beam through a low - z target . in this case
, ionization cooling techniques could increase the circulating beam lifetime and thus enhance the ion production to a level suitable for the betabeam .
this option has the advantage of employing isotopes with higher q - value than @xmath120 and @xmath119 , increasing correspondingly the neutrino energy ( from @xmath124 to @xmath125 gev for the sps - based bb ) .
this alternative approach will be at focus in the framework of the euro@xmath126 design study @xcite is a fp7 design study which started on sept 2008 and will run for 4 years .
the primary aims are to study three possible future neutrino oscillation facilities for europe ( a superbeam from cern - to - frejus , a ral or cern based nf and high - q bb ) and do a cost and performance comparison . ] .
a drawback with respect to the use of low - q ions is that the flux at the far location is smaller due to the larger beam divergence and a larger amount of ions stacked in the decay ring is needed . although the bb optimization is a complex task @xcite , some simple scaling laws can be used as reference . for a given number of decays per year @xmath127 ,
if we label with @xmath123 the lorentz factor of the ion ( which depends on the machine employed to accelerate the ion ) , with @xmath128 the q - value of the isotope and with @xmath9 the source - detector distance , the events at the far location are proportional to the convolution of the flux ( @xmath129 ) , of the cross section ( @xmath130 ) and of the oscillation probability , times @xmath127 . if the facility is operated at the maximum of the oscillation probability , then @xmath131
; therefore , @xmath132 . as a result
the number of events are proportional to @xmath133 .
note , however , that also @xmath127 has a dependence on @xmath134 due to the increase of the ion lifetime in the lab frame at larger @xmath123 .
summarizing , a high - q bb needs a smaller @xmath123 for the same neutrino energy , with the advantage that an accelerator of larger energy than the sps would not be needed and that the length of the decay ring could be shortened . on the other hand , given the beta beam kinematics , for the same baseline l an high - q bb needs an order of magnitude more ions at the source to match the performance of a high-@xmath123 bb .
in general , the clarification of the issue of the ion production yield is considered a crucial milestone for the betabeam .
given an appropriate yield , the acceleration and stacking is viewed as less demanding than what is needed for a nf both from the point of view of r&d and cost . clearly , the possibility of employing existing facilities ( e.g. the cern ps - sps complex or its upgrades ) might substantially strengthen this option .
* detectors and experimental challenges for the betabeams * a far detector within a bb facility seeks for the appearance of @xmath37 ( or during the run with @xmath119 and 8) in a bulk of unoscillated . therefore , their detection can be even simpler than for the nf , charge identification being immaterial in this case . in the sps - based bb option , the energy of the outcoming neutrinos lays in the sub - gev range . here
, the range of the muons is comparable with the one of the pions and high density detectors can not perform a clear nc / cc separation .
atmospheric neutrinos in such range are typically studied with water cherenkov detectors or with lar tpc s .
therefore , the ideal technology for such `` low - energy '' bb turns out to be identical to the ideal technology for the superbeams .
moreover , even for nominal ion fluxes , the smallness of the cross section requires very large detector masses , again in the range of the superbeam far detectors ( e.g. the above - mentioned hyper - kamiokande detector ) .
clearly , with respect to a superbeam , the bb offers a nearly complete control of the source systematics at the expenses of a riskier technology and larger costs . unlike the nf ,
atmospheric neutrinos are a significant background since - lowering the energy -its cross - section weighted rate increases roughly as @xmath135 .
suppression of atmospheric @xmath77 requires the exploitation of time - correlation between the ion bunch injected in the decay ring and the time of detection at the far location . in turn , bunches shorter than @xmath136 ns are needed , which also represent an accelerator challenge .
betabeams producing neutrinos in the multi - gev range ( advanced beta beams ) would open , once more , the possibility of employing high density detectors and offer lngs the opportunity to host a bb far detector .
this option ( iron calorimeter in hall c ) has been addressed explicitly in @xcite for a supersps bb with @xmath120 and @xmath119 .
it is less appealing than for the above - mentioned case of the nf : the somewhat limited mass can not compensate for the lower event rate of the bb with respect to the nf .
moreover , the neutrino energy is still quite small ( about 1.5 gev ) and , therefore , the neutral - current contamination is significant .
larger neutrino energies ( e.g. from 8 and 8) could further enhance its physics case , although the cern - lngs baseline would not be any more at the oscillation maximum . if the ion yield were appropriate , a natural solution - of noteworthy strategic interest for infn and cern - would be the use of the sps with 8 and 8 pointing to an iron calorimeter located at lngs .
physics performance are expected to be comparable with the ones of @xcite .
depending on the value of @xmath0 , the experimental approach toward a full measurement of the @xmath23 will be substantially different .
we discuss below three different ranges of @xmath0values : very large values , those suggested at 90% cl by the global fits of fogli et al .
@xcite and experimentally accessible in a couple of years , large values , corresponding to the ultimate sensitivity of the next generation of dedicated experiments , say t2k and daya bay , accessible in 5 - 7 years , and small values of @xmath0 , i.e. smaller than the combined sensitivity of the next experiments . to discriminate among the options , we use as a reference the sensitivity plots computed within the aforementioned ( see sec . [
sec : nf ] ) international scoping study ( iss ) @xcite .
[ fig : iss - th13],[fig : iss - sigdm],[fig : iss - lcpv ] display the expected sensitivity of the facilities discussed in sec .
[ sec : new_facility ] - super beams , beta beams and neutrino factories - in measuring @xmath0 , @xmath69 and @xmath19 as a function of @xmath0 .
since the performance of the facilities have large uncertainties , they are displayed as wide areas under different assumptions on the final layout and on the limiting systematics .
note that in some cases , the width of the area can be even larger than one order of magnitude .
it is worth stressing that so far no realistic estimate of costs and timescales of neutrino factories and beta beams exists : it is one of the prime tasks of the european network euro@xmath126 ( see footnote 7 ) to reach firm conclusions on this item . at 3@xmath137 level for different facilities .
the discovery limits are shown as a function of the fraction of all possible values of the true value of the cp phase @xmath19 ( ` fraction of @xmath138 ' ) and the true value of @xmath139 .
the right - hand edges of the bands correspond to more conservative setups while the left - hand edges correspond to fully optimized setups .
the discovery reach of the spl super beam from cern to frejus is shown as the orange band , that of hyper - kamiokande ( t2hk ) as the yellow band , and that of fermilab to dusel `` wide band '' beam @xcite as the green band .
the discovery reach of the beta beam is shown as the light green band and the neutrino factory discovery reach is shown as the blue band . from @xcite . , width=453 ] level for different facilities .
the discovery limits are shown as a function of the fraction of all possible values of the true value of the cp phase @xmath19 ( ` fraction of @xmath140 ' ) and the true value of @xmath141 .
the right - hand edges of the bands correspond to the conservative setups while the left - hand edges correspond to the optimized setups .
the discovery reach of the spl super beam is shown as the orange band , that of t2hk as the yellow band , and that of the wide - band beam experiment as the green band .
the discovery reach of the beta beam is shown as the light green band and the neutrino factory discovery reach is shown as the blue band . from @xcite . , width=453 ] level for different facilities .
the discovery limits are shown as a function of the fraction of all possible values of the true value of the cp phase @xmath19 ( ` fraction of @xmath140 ' ) and the true value of @xmath141 .
the right - hand edges of the bands correspond to the conservative setups while the left - hand edges correspond to the optimized setups .
the discovery reach of the spl super beam is shown as the orange band , that of t2hk as the yellow band , and that of the wide - band beam experiment as the green band .
the discovery reach of the beta beam is shown as the light green band and the neutrino factory discovery reach is shown as the blue band . from @xcite .
, width=453 ] if @xmath0is just below the chooz limit , it will be measured in the first runs of t2k and double - chooz , hopefully within 2010 . for so large @xmath0 , first results on leptonic cp violation might be reached by upgrading the beam intensity of the existing facilities ( t2k and no@xmath126a ) with no new gigantic far detectors @xcite .
it is easily predictable that such upgrades will have a very high priority in japan and usa and in the meantime a megaton water cerenkov detector ( or 100 kton liquid argon detectors ) could likely be funded at kamioka and/or dusel .
in such a scenario a major investment in europe for a new infrastructure would be fully justified . in italy
, it might bring to the construction of a new shallow - depth laboratory to host lar tpc s of a size not accessible at lngs .
this option has been discussed in the framework of the modular proposal @xcite , where a 20 kton detector would be installed slightly off the main axis of cngs , in a way that resembles the location of nova along the fermilab - to - minnesota neutrino beam ( see fig.[fig : nova ] ) . in general
, the use of the imaging capability of the lar - tpc ensures a higher discovery potential than in the case of scintillator ( or water ) detectors , i.e. a comparable sensitivity may be achieved with a smaller sensitive mass . from the point of view of lngs , large values of @xmath0will
bring measurable signals in opera
. the opera result would be very interesting : it is truly complementary to the t2k and nova results since the facility is operated off the peak of the oscillation maximum and with completely different systematics @xcite . as a consequence
it would be very recommendable a prolongation of the opera run , possibly with an intensity upgrade of the cngs . in this range of values
a non - zero @xmath0value will be established at 3@xmath137 in the full run of t2k and daya bay ( around 2015 ) .
for these values no simple upgrades of t2k and nova will have sensitivity on cp violation and mass hierarchy ( we include modular in this class of experiments ) , and detectors of the class of hyper - kamiokande will be needed .
this is a favorable scenario for the superbeams : they represent a relatively low risk technology and can get a clearly accessible physics case for medium values of @xmath0 . in europe
the spl project , with a megaton class detector installed in a new cavern at the frejus laboratories , could compete with these projects , having a better sensitivity on @xmath0and cp violation and a worse sensitivity on mass hierarchy @xcite . in this race
, europe suffers from some disadvantages .
it is likely that only one megaton - size detector will be funded world - wide .
japan is a leader on this technology and us would likely exploit the discovery of @xmath0to boost this opportunity at dusel .
moreover , at that time ( 2016 ) cern will be engaged with the replacement of the ps and , probably , with the luminosity upgrades of the lhc , so funding for an ultimate facility for astroparticle physics might be difficult to be gathered to successfully face a competition with us and japan .
as noted above , however , europe could pursue better suited strategies than the standard superbeam .
if the feasibility of a high - q sps - based betabeam can be demonstrated in a few years from now , cern could potentially host a novel facility for neutrino oscillations exploiting either smaller detectors ( high density calorimeters in hall c ) or with significantly better performance if combined with a 100 kton size lar tpc . according to @xcite , in this range of values the optimal facility (
the one that could measure @xmath0 , @xmath69and @xmath19 with just one detector ) would be a high-@xmath123 ( or even an sps based ) beta beam pointing to a megaton water cerenkov detector or a detector with equivalent performance ( e.g. a 100 kton lar tpc ) .
there is however a more conservative approach that might be extremely interesting for uk and italy .
the guarantee of the physics case due to the finite size of @xmath0can be used to strongly boost the realization of a neutrino factory based at the rutherford labs . as a consequence ,
cp violation and matter effect could be studied in the experimental halls of lngs building a iron magnetized detector ( see sec .
[ sec : nf ] ) located about 1500 km from the source .
ultimate precision for @xmath0and for the resolution of the degeneracies will be achieved exploiting the second ral - to - india baseline ( 7000 km ) .
if no signal of @xmath0will be detected by the next generation of accelerator and reactor experiments , a rather problematic scenario will disclose .
the discovery potential of super beam experiments like hyper - kamiokande and dusel would be quite limited ( they have @xmath148 cp violation sensitivity for @xmath0values as small as about @xmath149 ) and it seems very unlikely that those facilities will be built under these circumstances .
advanced beta beams and neutrino factories can explore @xmath0values two or three orders of magnitude smaller than the t2k sensitivity , but again it seems unlikely that facilities of such cost will be strongly supported having as a physics case just the exploration of the @xmath0value and a search for cp violation without any guaranteed signal . on the other hand it seems unlikely that neutrino oscillation experiments stop without mapping the values of three fundamental parameters of the standard model : @xmath0 , @xmath69 and @xmath19 , also considering that in the framework of leptogenesis @xmath19 is the single most important unknown ingredient to address matter - antimatter asymmetry in the universe @xcite . in this case , however , any progress in oscillation neutrino physics will rely on advances in beta beams , neutrino factory or further novel sources .
among elementary fermions , neutrinos have a unique status due to charge neutrality and the extremely small value of their mass .
most likely , they are the only majorana particles up to the electroweak scale and , differently from other oscillating particles , their coherence length can extend up to cosmic scales @xcite .
it comes to no surprise that these particle have been speculated to have unusual properties . within several models , their free streaming in vacuum and matter
is perturbed by non - standard interactions , which can be constrained mainly by oscillation experiments .
even lorenz - invariance and cpt conservation @xcite have been challenged for neutrinos . as a matter of fact none of these hypotheses
are empirically grounded and the constraints coming from the oscillation experiments and from cosmology are extremely tight .
however , a further stimulus to pursue these searches has come from the persistent lsnd anomaly , which mainly brought to the consideration of additional sterile neutrinos .
lsnd indicated @xcite a @xmath150 oscillation with a third , very distinct , neutrino mass difference : @xmath151 ev@xmath28 . ;
what is quoted here is the region not completely excluded by other experiments . ] already in 1999 , fogli et al .
@xcite showed that the standard three neutrino framework is inconsistent with the ensemble of all experimental data , included lsnd . to explain the whole set of data at least four different light neutrino species would have been needed ( referred as `` 3 + 1 '' models ) .
sterile neutrinos are particles with the same quantum number of the vacuum and , therefore , are `` sterile '' with respect to electroweak interactions . in the standard three - family framework ,
massive sterile neutrinos are the ordinary right - handed components of the dirac fermions and they do not mix with active neutrinos . beyond this model , light steriles could mix with active @xmath126 and , therefore , change the transition probability through a @xmath152 leptonic mixing matrix with @xmath153 @xcite .
these model predicted of course several other oscillation processes in excess to the @xmath154 transitions detected by lsnd , the largest part already excluded by other short baseline experiments .
as a consequence , these @xmath155 models provided anyway poor overall fits to the existing experimental data @xcite . the miniboone experiment at fermilab was designed to directly check the lsnd claim , performing a search for @xmath156 appearance with a baseline of 540 m and a mean neutrino energy of about 700 mev , i.e. with energy and baseline larger than lsnd but with the same ratio @xmath42 . in 2007
miniboone published a no - evidence analysis for @xmath39oscillation @xcite .
while this analysis was not able to rule - out the whole parameter space of the lsnd signal @xcite , it represented a further suppression of the four - neutrino interpretation of the lsnd anomaly . in 2009
miniboone @xcite published also results for @xmath54 oscillation showing , again , no appearance effects , although at a smaller significance with respect to the neutrino data .
miniboone neutrino data reported also an unexplained excess for low energy neutrinos ( mineboone anomaly in the following ) , that can not be reconciled with any sterile model .
this data excess was not present in the antineutrino run . at present , electron ( anti)neutrino appearance results , including the lsnd and the miniboone anomalies , can be reconciled in a `` 3 + 2 '' model @xcite .
however results from muon neutrino disappearence experiments at short baselines are in disagreement with this model @xcite so that the world dataset are clearly inconsistent with both 3 + 1 and 3 + 2 model predictions and even more baroque models should be advocated .
it is interesting to analyze whether the cngs might be able to test the lsnd anomaly , in spite of the much larger @xmath42 . naively , if the lsnd anomaly comes from a two family oscillation source of @xmath157 parametrized as @xmath158 with @xmath159 ev@xmath28 could play the role of @xmath30 or @xmath31 .
although oversimplified , this formula is still used to express the sensitivity coverage of karmen @xcite and miniboone with respect to the lsnd claim . ] , at cngs the probability of @xmath39 oscillation would be very high ( i.e. @xmath160 ) since the oscillating term averages out to 1/2 .
moreover , thanks to the increase of the cross section at 17 gev , the cngs would perform significantly better than minos or other long - baseline experiments . unfortunately
, if any realistic model ( e.g. 3 + 1 ) is used to compute the actual rate at lngs , the effect is rather small , since models must evade the already stringent bounds from atmospheric neutrinos .
a detailed calculation of the performance of the cngs to test the lsnd anomaly has been done in @xcite . in general
, we can expect that the cngs will improve marginally the bounds on the sterile parameters and , surprisingly , much more from the @xmath41 analysis than from the @xmath39 . on the other hand , @xmath41 oscillations are particularly suited to test unconstrained parameters for the non - standard interactions .
it has been shown , for instance , that operators affecting the @xmath41transitions and compatible with current constraints from atmospheric might change substantially the number of observed tau events in opera @xcite .
it is , however , in the framework of the neutrino factories that a superior test of non - standard interactions ( at 1 - 0.1 % level ) can be carried out either with or without a detector for the silver channel @xcite . given the present experimental situation , the most straightforward way to test ( again ) lsnd is to perform a similar experiment at the same l and e but with significantly larger statistics .
an experiment based on large fluxes of pions decaying at rest and in flight has been recently proposed at the new us neutron spallation source ( sns ) @xcite .
another straightforward way would be to incorporate a close detector in the miniboone setup , greatly reducing the impact of systematic errors in the experimental analysis .
the recent boone letter of intents @xcite showed that building a second miniboone detector at ( or moving the existing miniboone detector to ) a distance of @xmath161 m from the booster neutrino beam ( bnb ) production target , the sensitivity of the setup would result greatly enhanced .
similarly , the miniboone low - energy anomaly - which is located at a different e and l / e with respect to lsnd - might be studied directly using a high granularity detector positioned along the same beam of miniboone ( boone beamline ) at fermilab ( `` microboone '' @xcite ) .
this detector , operated both at boone and at numi could simultaneously test the above - mentioned anomaly and perform high precision cross section measurement in the critical region around 1 gev .
this region is of great practical interest since most of the superbeams proposed so far are operated in such energy range ; it is explored by a dedicated cross - section experiment called sciboone @xcite and , along numi , by minerva @xcite .
the lar technology is ideal to perform these measurements and , therefore , it has been proposed as the basic technology option for microboone up to mass of 170 t ( 70 t fiducial mass ) . as discussed in @xcite , an even larger detector of 500 t active mass
could be built at cern restoring the low - energy neutrino beamline driven by the ps and originally used for bebc - ps180 @xcite .
the detector would be positioned 850 m far from the source and complemented by a smaller near detector at 127 m ( 150 t active mass ) .
an identical setup , with the notable difference of a totally active scintillator detector instead of liquid argon , was proposed to cern already in 1997 @xcite .
with respect to microboone , the presence of an identical near detector strengthen significantly the reliability of the oscillation search and the overall mass allows for a test based on a much larger statistics .
on the other hand , this setup requires the construction of a dedicated beamline and proper sharing of the protons of the ps , which also feed the rest of the cern physics programme ( sps fixed target and the lhc ) . the cern option ( 1.1 gev at 850 m ) works , once more , at the same l / e as lsnd . the values of l and e , however , are different both with respect to the original lsnd apparatus and also with respect to miniboone ( 0.7 gev at 550 m ) .
this situation is ideal if the source of anomaly has the usual l / e pattern but could be non optimal if the lsnd and/or the miniboone excess has a different origin .
in this paper , we investigated the perspectives for accelerator neutrino physics in europe , with special emphasis to the line of research where infn infrastructures ( firstly lngs ) and expertise could be most profitably exploited .
if all further data coming from the current and next generation of experiments fit the standard oscillation framework , i.e. the assumption of three active neutrinos with standard model couplings , then the strategy for the next decade will be solely driven by the actual size of @xmath0 .
in particular : * if @xmath0is close to the current limits ( @xmath162 ) and therefore can be observed very early by t2k or double - chooz , an independent measurement done in parallel by the cngs experiments would be of high value : it could be obtained by high granularity detectors with completely different systematics with respect to t2k and , moreover , running off the peak of the oscillation probability , with no dependence on matter effect and with a dependence of the cosine of the cp phase instead of the usual @xmath163 term . therefore , a prolongation of the cngs , possibly running in dedicated mode or profiting from an intensity upgrade should definitely be pursued .
moreover , in this special case medium size detectors ( t2k , nova ) with upgraded beams could play a relevant role for precision measurement in the leptonic sector . a medium size ( 20 kton ) lar detector - having performance comparable or better than nova - could complement the t2k measurements , especially with the aim of determining the mass hierarchy of neutrinos through the observation of matter effects .
clearly , it is an ideal scenario for cern and infn , who would fully profit of the investments done for cngs even beyond the primary aim of the experiment , i.e. the observation of @xmath41oscillations .
* for lower values of @xmath0 , several options can be envisaged both inside and outside the gran sasso laboratories . in a timescale shorter than the neutrino factory , a sps - based betabeam at cern accelerating
high - q ions might become available . in this case
, an experimental hall of the gran sasso labs would be enough to host a high density detector ( e.g. iron calorimeter ) operating in appearance mode .
a high - q betabeam requires a significant r&d which , however , is synergical with the activities of nuclear physics community and it surely exploits in a clever manner european infrastructures , as cern , the infn laboratories of legnaro , crc at louvain and several other labs working on the development of radioactive ion beams .
+ on the other hand , in this range of @xmath0 , the construction of the neutrino factory is considered as the ultimate and most precise facility for the study of lepton mixing .
lngs could host the far detector of a uk - based neutrino factory although the distance from the rutherford labs to lngs is shorter than the optimal one ( 1500 versus 3000 km ) .
this option offers the best physics performance at a price of a longer timescale since it is unrealistic that a neutrino factory comes into operation before 2020 .
+ in the context of a cern - based or ral - based neutrino factory ( @xmath164 ) , an option highly synergical with infn infrastructures and technologies is the study of @xmath74 transitions at lngs using detectors able to identify the appearance of tau leptons , as it was the case of the cngs .
this `` silver channel '' can be exploited at lngs by a @xmath1655 kton opera - like or icarus - like detector equipped by a magnetic spectrometer for @xmath95 charge measurement and detecting neutrinos from cern or ral .
this channel can be used to solve the degeneracies in the mixing parameters when measured with an ultimate facility as the neutrino factory itself .
+ without betabeams or neutrino factories , there is no way to use the halls of lngs profitably over this range of @xmath0 . as an alternative ( or in parallel , if resources and interest from other countries can be gathered ) , the european neutrino community could pursue an aggressive r&d to boost the lar technology to the 100 kton size : an ultimate facility for astroparticle physics and for the technology of the superbeam . * in case of no evidence of appearance after t2k , it is unlikely that accelerator neutrino physics will gather major investments .
r&d toward the neutrino factory will proceed at a lower rate and will be done synergically with a possible muon collider while a megaton - size water cherenkov detector or a 100 kton lar tpc will be mainly motivated by astroparticle physics and proton decay searches .
outside the standard framework , we still see some opportunities especially for `` low '' mass liquid argon detectors in the quest for the clarification of the low energy excess of miniboone .
the most natural application is in the framework of microboone @xcite : an experiment based on a @xmath166 ton lar tpc and located at fermilab , along the boone and numi beamline . alternatively ,
as pointed out in @xcite , the ps neutrino beamline at cern could be restored to perform a similar test with better significance , and with the added value of the continuity in the development and exploitation of the lar technology within europe .
note , however , that in a few year a further test @xcite of the original lsnd anomaly might also be done at the sns , the us high intensity neutron source already in operation at the oak ridge national labs . as a final consideration
, we wish to stress that these years - which have witnessed the completion and the startup of the cngs and , more recently , of double - chooz - are unique times to look to the future of neutrino physics with artificial sources in europe @xcite .
clearly , the optimal strategy will be finally driven by the outcomes of the @xmath0-oriented experiments , but a systematic exploration of the different alternatives as a function of these outcomes was missing in literature , especially when focusing to the existing european infrastructures and to technologies where infn has a world leadership : we hope that this paper fills such gap at best of our present knowledge of lepton mixing .
this work originated within the `` commissione scientifica nazionale ii '' ( astroparticle physics ) of infn and euro@xmath126 .
we thank our colleagues of commissione ii and the director of the lngs labs ( lucia votano ) for many enlightening discussions during the drafting of the paper .
we also want to express our gratitude to carlo rubbia who provided us detailed information on the double - lar project @xcite at cern - ps . finally , we gratefully acknowledge d. gibin , a. guglielmi , f. pietropaolo , c. rubbia and p. sala for critical reading of the manuscript and many interesting suggestions .
g. danby , j. m. gaillard , k. goulianos , l. m. lederman , n. b. mistry , m. schwartz and j. steinberger , phys .
* 9 * ( 1962 ) 36 . for early proposals of accelerator neutrino experiments
see also m. a. markov `` early development of weak interactions in the ussr '' nauka publisher , central depart . of oriental literature , moscow ( 1985 ) and
b. pontecorvo , sov .
jetp * 10 * ( 1960 ) 1236 [ zh .
* 37 * ( 1959 ) 1751 ] .
z. maki , m. nakagawa and s. sakata , prog .
* 28 * ( 1962 ) 870 ; y. katayama , k. matunoto , s. tanaka and e. yamada , progr .
* 28 * ( 1962 ) 675 ; b. pontecorvo , sov . phys .
jetp * 26 * ( 1968 ) 984 [ zh .
teor . fiz .
* 53 * ( 1967 ) 1717 ] ; v. n. gribov and b. pontecorvo , phys .
b * 28 * ( 1969 ) 493 .
a. b. balantekin and d. yilmaz , j. phys .
g * 35 * ( 2008 ) 075007 ; g. l. fogli , e. lisi , a. marrone , a. palazzo and a. m. rotunno , phys . rev .
* 101 * ( 2008 ) 141801 ; m. maltoni and t. schwetz , arxiv:0812.3161 [ hep - ph ] .
current and future facilities in europe and perspectives in us and japan have been discussed in the `` european strategy for future neutrino physics '' workshop , cern , 1 - 3 october 2009 .
information are available at ` http://indico.cern.ch/conferencedisplay.py?confid=59378 ` .
m. guler _ et al . _
[ opera collaboration ] , cern - spsc-2000 - 028 , cern - spsc - p-318 , lngs - p25 - 00 , 2000 ; + r. acquafredda _ et al . _ [ opera collaboration ] , new j. phys .
* 8 * ( 2006 ) 303 ; + r. acquafredda _ et al .
_ [ opera collaboration ] , jinst * 4 * ( 2009 ) p04018 . c. athanassopoulos
_ et al . _
[ lsnd collaboration ] , phys .
* 75 * ( 1995 ) 2650 ; c. athanassopoulos _ et al .
_ [ lsnd collaboration ] , phys .
c * 54 * ( 1996 ) 2685 ; c. athanassopoulos _ et al .
_ [ lsnd collaboration ] , phys .
c * 58 * ( 1998 ) 2489 ; a. aguilar _ et al .
_ [ lsnd collaboration ] , phys . rev .
d * 64 * ( 2001 ) 112007 .
m. mezzetto , `` next challenge in neutrino physics : the theta(13 ) angle , '' proceedings of `` thirteen international workshop on neutrino telescopes '' , venezia , march 10 - 13 , 2009 , arxiv:0905.2842 [ hep - ph ] .
other venues have also been explored : they include the exploitation of mossbauer neutrinos ( r. s. raghavan , arxiv : hep - ph/0601079 ; h. minakata , h. nunokawa , s. j. parke and r. zukanovich funchal , phys .
d * 76 * ( 2007 ) 053004 ) , high intensity radioactive sources ( y. giomataris and j. d. vergados , nucl . instrum .
a * 530 * ( 2004 ) 330 ) and laser - driven sources ( s. v. bulanov , t. esirkepov , p. migliozzi , f. pegoraro , t. tajima and f. terranova , nucl .
instrum .
a * 540 * ( 2005 ) 25 ) .
w. marciano , talk at the 2nd workshop on physics with an intense proton source , fermilab , il , jan 25 - 28 2008 .
available at ` www.fnal.gov/directorate/longrange/steering_public/workshop-physics-2nd.html ` m. koshiba , phys . rep . * 220 * ( 1992 ) 229 ; k. nakamura , talk at int .
workshop on next generation nucleon decay and neutrino detector , 1999 , suny at stony brook ; k. nakamura , neutrino oscillations and their origin , ( universal academy press , tokyo , 2000 ) , p. 359 .
l. oberauer , f. von feilitzsch and w. potzel , nucl .
b ( proc . suppl . )
* 138 * ( 2005 ) 108 ; t. marrodan undagoitia _
_ , j. phys .
* 120 * ( 2008 ) 052018 ; j. peltoniemi , arxiv:0911.4876 [ hep - ex ] ; j. peltoniemi , arxiv:0911.5234 [ hep - ph ] .
kirk et al . , `` the merit high - power target experiment at the cern ps '' , epac08-wepp169 , fermilab - conf-08 - 224-apc , presented at the european particle accelerator conference ( epac 08 ) , genova , italy , 23 - 27 jun 2008 .
the `` emulsion cloud chamber '' detector concept dates back to 1956 [ j. nishimura , soryusiron kenkyu , japan * 12 * ( 1956 ) 24 ] .
in the 80 s , this technology has been revived by the progress in automatic scanning techniques achieved in japan [ s. aoki , et al . , nuclear tracks * 12 * ( 1986 ) 249 ; s. aoki , et al .
instr . and meth .
b * 51 * ( 1990 ) 466 ] and applied in large size experiments as donut @xcite , chorus @xcite and opera @xcite . j. burguet - castell , d. casper , j. j. gomez - cadenas , p. hernandez and f. sanchez , nucl . phys .
b * 695 * ( 2004 ) 217 ; f. terranova , a. marotta , p. migliozzi and m. spinetti , eur .
j. c * 38 * ( 2004 ) 69 ; s. k. agarwalla , a. raychaudhuri and a. samanta , phys .
b * 629 * ( 2005 ) 33 .
a. donini , e. fernandez - martinez , p. migliozzi , s. rigolin , l. scotto lavina , t. tabarelli de fatis and f. terranova , eur .
j. c * 48 * ( 2006 ) 787 ; a. donini _ et al .
_ , eur .
j. c * 53 * ( 2008 ) 599 .
n. armenise _
, `` letter of intent : search for @xmath39oscillation at the cern ps '' , cern - spsc/97 - 21 , spsc / i216 ( 1997 ) ; m. guler _
et al . _ , `` proposal : search for @xmath39oscillation at the cern ps '' , cern - spsc/99 - 26 , spsc / p311 ( 1999 ) . | very soon a new generation of reactor and accelerator neutrino oscillation experiments - double chooz , daya bay , reno and t2k - will seek for oscillation signals generated by the mixing parameter @xmath0 .
the knowledge of this angle is a fundamental milestone to optimize further experiments aimed at detecting cp violation in the neutrino sector .
leptonic cp violation is a key phenomenon that has profound implications in particle physics and cosmology but it is clearly out of reach for the aforementioned experiments . since late 90 s ,
a world - wide activity is in progress to design facilities that can access cp violation in neutrino oscillation and perform high precision measurements of the lepton counterpart of the cabibbo - kobayashi - maskawa matrix . in this paper
the status of these studies will be summarized , focusing on the options that are best suited to exploit existing european facilities ( firstly cern and the infn gran sasso laboratories ) or technologies where europe has a world leadership .
similar considerations will be developed in more exotic scenarios - beyond the standard framework of flavor oscillation among three active neutrinos - that might appear plausible in the occurrence of anomalous results from post - miniboone experiments or the cngs .
+ r. battiston@xmath1 , m. mezzetto@xmath2 , p. migliozzi@xmath3 , f. terranova@xmath4 + @xmath1 dep . of physics ,
univ . of perugia and i.n.f.n .
, sezione di perugia , perugia , italy + @xmath2 i.n.f.n .
, sezione di padova , padova , italy + @xmath3 i.n.f.n .
, sezione di napoli , naples , italy + @xmath4 i.n.f.n .
, laboratori nazionali di frascati , frascati ( rome ) , italy + |
the main aim of relativistic heavy - ion collisions is to unravel the basic properties of qcd associated with , e.g. , its phase diagram .
two of the main features characterizing the ground state of the theory , confinement and the spontaneous breaking of chiral symmetry , are expected to cease in a possibly common phase transition at finite temperature and/or density @xcite .
the dynamical breaking of chiral symmetry is associated with the condensation of quark - antiquark pairs in the qcd vacuum .
as the temperature and/or baryon density increase , the qcd vacuum undergoes a phase transition to the chirally symmetric phase where the quark - antiquark condensate , the order parameter of the chiral phase transition , vanishes .
the fluctuation of this order parameter is related to the associated susceptibilities @xcite .
the quantity of interest here is the chiral / scalar density susceptibility which measures the response of the chiral condensate / scalar density to the variation of the current quark mass .
the chiral susceptibility was recently measured in lattice qcd @xcite with two light flavors .
the chiral susceptibility has also been studied in chiral perturbation theory @xcite , in the multiflavor schwinger model for a small nonzero quark mass @xcite , in the njl model @xcite , and using the dyson - schwinger equation @xcite . in this paper
, we will investigate the chiral / scalar density susceptibility following the same line as in the case of the quark number susceptibility @xcite and the free energy of the quark - gluon plasma @xcite within the hard thermal loop ( htl ) approximation @xcite , which selectively resums higher order corrections corresponding to medium effects such as screening , quasiparticles masses and landau damping beyond the usual perturbation theory .
let @xmath0 be a heisenberg operator . in a static and uniform external field @xmath1
, the ( induced ) expectation value of the operator @xmath2 is written @xcite as @xmath3 } { { \rm tr}\left [ e^{-\beta \left ( { \cal h}+{\cal h}_{ex}\right ) } \right ] } = \frac{1}{v}\int { \rm d}^{3}x\ , \left\langle { \cal o } _ { \alpha } \left ( 0,\vec{x}\right ) \right\rangle \ : , \label{eq1}\ ] ] where translational invariance is assumed and @xmath4 is given by @xmath5 the ( static ) susceptibility @xmath6 is defined as @xmath7 assuming no broken symmetry @xmath8 .
the quantity @xmath9 is the two point correlation function with operators evaluated at equal times .
the chiral susceptibility measures the response of the chiral condensate to the infinitesimal change of the current quark mass @xmath10 .
here , @xmath11 corresponds to the current quark mass and the operator @xmath12 to @xmath13 . then the _ static
_ chiral susceptibility can be obtained @xcite from @xmath14 where @xmath15 is the static correlator of the scalar channel .
the chiral condensate is given as @xmath16 } { { \rm{tr}}\left [ e^{-\beta { \cal h}}\right ] } \
= \frac{\partial\omega } { \partial m } \ . \label{eq9}\ ] ] here , @xmath17 is the dynamical hamiltonian of the system containing the quark mass , @xmath18 is the thermodynamic potential , and @xmath19 is the partition function of a quark - antiquark gas . taking the fourier transform of the static correlator @xmath20 , it can be shown that @xmath21 where @xmath22 is the fourier transformed correlator . applying the fluctuation - dissipation theorem @xcite
, one finds the static limit of the _ dynamic _ chiral susceptibility @xmath23 where @xmath24 is the scalar self - energy . using @xmath25 , ( [ eq10 ] )
can also be written as @xmath26
the chiral condensate @xmath27 at finite temperature can be obtained from the tadpole diagram in fig.1 as @xmath28 \ , , \label{eq11}\end{aligned}\ ] ] where , @xmath29 and @xmath30 are , respectively , the numbers of quark flavors and colors .
@xmath31 is the free quark propagator given as @xmath32 summing over the matsubara frequencies @xmath33 , we obtain @xmath34 \ , \label{eq11b}\end{aligned}\ ] ] where @xmath35 is the fermi distribution function and @xmath36 .
( 300,100)(0,0 ) ( 150,50)(40,270,90 ) ( 150,50)(40,90,270 ) ( 50,10)(150,10)5 ( 150,10)(230,10)5 ( 150,103)[]@xmath37 fig.1 : tadpole diagram determining the free static chiral susceptibility .
alternatively , the chiral condensate can be derived from the lowest order thermodynamic potential of the quark - antiquark system @xcite @xmath38 + \log \left [ 1 + e^{-\beta(e_k+\mu_q ) } \right ] \right \ } \ . \label{eq12}\ ] ] using ( [ eq9 ] ) the condensate ( [ eq11b ] ) is reproduced .
now , the chiral susceptibility in the free case follows from ( [ eq8 ] ) and ( [ eq11b ] ) as @xmath39 + 2\beta \frac { m^2}{e_k^2}n_f(e_k)\big[1-n_f(e_k)\big ] \right \}\right |_{m=0 } \nonumber \\ & = & 2n_f n_c \int \frac{{\rm d}^3k}{(2\pi)^3 } \frac{1}{k } \big[1 - 2n_f(k)\big ] \ , .
\label{eq14 } \end{aligned}\ ] ] note that , in contrast to the quark number susceptibility @xcite where the zero temperature contribution vanishes due to the transversality properties of the vector channel @xcite , the static chiral susceptibility contains a quadratic ultraviolet divergence coming from the zero temperature contribution .
using dimensional regularization the temperature independent term in ( [ eq14 ] ) disappears , as there is no scale associated with it , leading to @xmath40 the next order contribution follows from the two - loop thermodynamic potential given in ref .
however , the second derivative of this expression with respect to @xmath41 diverges at @xmath42 ( see also ref .
the reason for this divergence can be seen in the following way : the second derivative with respect to the quark mass @xmath41 of the two - loop thermodynamic potential , which is a quark loop with an internal gluon line , introduces two additional quark propagators .
hence the originating diagram looks like the polarization tensor containing an internal quark line .
this diagram is known to be logarithmically infrared divergent in the case of a vanishing bare quark mass ( see e.g. @xcite ) .
hence , the static chiral susceptibility can not be calculated consistently in usual perturbation theory beyond leading order but requires the htl resummation .
next , we calculate the static limit of the dynamical chiral susceptibility from the lowest order self - energy diagram shown in fig.2 : @xmath43 \ , .
\label{eq15}\ ] ] summing over @xmath33 , we obtain @xmath44 \big [ \delta(\omega -e_k - e_q ) - \delta ( \omega+e_k+e_q ) \big ] \right .
\nonumber \\ & + & \left .
\left ( 1+\frac{{\vec k}\cdot { \vec q}+ m^2}{e_ke_q } \right ) \big [ n_f(e_k)-n_f(e_q ) \big ] \big [ \delta(\omega -e_k+e_q ) - \delta ( \omega+e_k - e_q ) \big ] \right \ } \label{eq16 } \end{aligned}\ ] ] ( 300,100)(0,0 ) ( 150,50)(40,180,0 ) ( 150,50)(40,180,0 ) ( 50,50)(110,50)5 ( 190,50)(250,50)5 ( 75,40)[]@xmath45 ( 225,40)[]@xmath45 ( 150,105)[]@xmath37 ( 150,-5)[]@xmath46 fig.2 : self - energy diagram determining the free dynamic chiral susceptibility . combining ( [ eq10 ] ) and ( [ eq16 ] ) yields @xmath47 + 2\beta \frac { m^2}{e_k^2}n_f(e_k)\big [ 1-n_f(e_k)\big ] \right \}\right |_{m=0 } \nonumber \\ & = & 2n_f n_c \beta \int \frac{{\rm d}^3k}{(2\pi)^3 } \big [ 1 - 2n_f(k)+2n_f^2(k)\big ] \ .
\label{eq17 } \end{aligned}\ ] ] obviously ( [ eq14 ] ) and ( [ eq17 ] ) do not agree . for , in general , the static susceptibility , given in ( [ eq14 ] ) as the second derivative of the thermodynamical potential , does not coincide with the static limit of the dynamical susceptibility in ( [ eq17 ] ) which satisfy the fluctuation - dissipation theorem .
there is an exception : if the static susceptibility were associated with a conserved quantity such as the ( quark ) number density , the conservation law would ensure the equality of the two different quantities @xcite .
thus , as shown in refs .
@xcite , the quark number susceptibility can be derived in two different ways as the imaginary part of the self - energy is proportional to @xmath48 implying a conserved number density .
however , the static chiral susceptibility is given by the derivative of the scalar / chiral density in ( [ eq14 ] ) which obeys no conservation law , so it does not coincide with the static limit of the dynamic susceptibility in ( [ eq17 ] ) as represented by the fluctuation - dissipation theorem since the imaginary part of the self energy in the static limit is not proportional to @xmath48 .
note that , in contrast to the static chiral susceptibility , there is no zero temperature contribution as there is an overall factor @xmath49 .
the term containing no distribution function in ( [ eq17 ] ) shows a cubic ultraviolet divergence .
here , we want to calculate the chiral susceptibility beyond the free quark approximation by taking into account the in - medium properties of quarks in a qgp . in the weak coupling limit ( @xmath50 ) ,
a consistent method is to use the htl - resummed quark propagator and htl quark - meson vertex if the quark momentum is soft ( @xmath51 ) .
however , in the following we will use the htl - resummed quark propagator also for hard momenta , in order to consider , at least to some extent , non - perturbative features of the qgp such as effective quark masses and landau damping .
this strategy is in the same spirit as in the case of the quark number susceptibility @xcite and the meson correlators @xcite , although this approach might not be consistent in the weak coupling limit @xcite .
however , since we do not want to restrict to the weak coupling limit , but consider temperatures close to @xmath52 , we do not aim at a complete leading order calculation , using here the htl green functions only for including medium effects in a gauge invariant way .
another , maybe more convincing , reason for studying the chiral susceptibility in the htl approximation is that this approach has also been used for the free energy of the qgp , where different implementations of the htl method and their validity have been discussed .
the chiral susceptibility may serve as another test for these questions . in the rest frame of the medium
the most general expression for the fermion self energy in the chiral limit of vanishing current mass has the form @xcite @xmath53 where the scalar quantities @xmath54 and @xmath55 are functions of energy @xmath33 and the magnitude @xmath56 of the three momentum
. the htl - resummed quark propagator , @xmath57 $ ] , can be expressed using the helicity representation in the massless case as @xcite @xmath58 with @xmath59 \
, , \label{2.7}\end{aligned}\ ] ] where the thermal quark mass is given by @xmath60 , and @xmath61 is the legendre function of second kind .
the zeros @xmath62 of @xmath63 describe the two dispersion relations of collective quark modes in a thermal medium @xcite .
furthermore , the htl - resummed quark propagator acquires a cut contribution below the light cone ( @xmath64 ) as the htl quark self - energy , contained in @xmath63 , has a non - vanishing imaginary part , which can be related to landau damping for spacelike quark momenta resulting from interactions of the quarks with gluons of the thermal medium . for the coupling constant
entering @xmath65 , @xmath66 , we choose the two - loop result of the temperature dependent running coupling @xcite @xmath67 \ , , \label{coupling}\ ] ] where @xmath68 , @xmath69 , @xmath70 and @xmath71 is the renormalization scale for the running coupling constant .
it should also be noted that the htl - resummed propagator is chirally symmetric in spite of the appearance of an effective quark mass @xcite . the htl - spectral function of @xmath72 , @xmath73 , can be written as @xcite @xmath74 + \beta_\pm(k_0,k )
\theta(k^2-k_0 ^ 2 ) \label{2.10}\ ] ] with @xmath75 ^ 2 + \biggl [ \frac{\pi}{2 } m_q^2 \frac{\pm k_0 -k}{k } \biggr ] ^2 } \ , \ , \ , , \label{2.11}\ ] ] where the first term describing the quasiparticle dispersion in ( [ 2.10 ] ) is due to the poles of the propagator and the second term containing landau damping corresponds to the cut contribution for spacelike quark momenta .
now , we compute the chiral condensate @xmath76 from the tadpole diagram in fig.1 , where we use the effective htl quark propagator , yielding @xmath77 . \label{2.18}\end{aligned}\ ] ] a similar calculation has been done in the case of the quark contribution to the free energy by andersen et al .
@xcite . according to ( [ eq8 ] ) ,
but in contrast to the htl approximation discussed above , we have to take into account a non - vanishing current quark mass . in the weak coupling limit ( htl approximation )
there are two possibilities : 1 ) @xmath78 , i.e. , @xmath79 and 2 ) @xmath80 .
in the first case , @xmath81 can be neglected in the effective quark propagator compared to the hard scale @xmath41 , implying @xmath82 , which leads to the free susceptibility . in the second case ,
@xmath41 can be neglected in the calculation of the htl quark self energy , requiring hard loop momenta , i.e. , @xmath83 .
then , the effective htl quark propagator becomes @xmath84 with @xmath85 where @xmath63 is given by the massless htl - approximation in ( [ 2.7 ] ) .
note that the current quark mass , @xmath80 , is of the same order as the htl contributions @xmath54 and @xmath55 and can not be neglected therefore .
summing over @xmath33 , we obtain @xmath86 \ , \label{2.22}\ ] ] where the spectral function corresponding to the massive htl propagator is given by @xmath87 following ( [ eq8 ] ) , the static chiral susceptibility in htl - approximation is found as @xmath88 + \frac { \omega^2_-(k)-k^2 } { 2m_q^2 } { \rm re}d_+^{-1}(\omega_-,k)\left [ 1 - 2n_f(\omega_-)\right ] \nonumber \\ & + & \int_{0}^k { \rm d}x \ > [ \beta_+(x , k){\rm re}d_-^{-1}(x , k)+ \beta_-(x , k){\rm re}d_+^{-1}(x , k ) ] \left [ 1 - 2n_f(x)\right ] \biggr\ } \ , \label{2.24}\end{aligned}\ ] ] where @xmath89 ^ 2 + \biggl [ \frac{\pi}{2 } m_q^2 \frac{\pm k_0 -k}{k } \theta ( k^2-k_0 ^ 2)\biggr ] ^2 } \ , \ , \ , .
\label{2.24a}\ ] ] in the infinite temperature limit , the first term in ( [ 2.24 ] ) reduces to the free static susceptibility given in ( [ eq14 ] ) whereas the second and third terms vanish .
the integrals without the distribution functions in the first and third term in ( [ 2.24 ] ) are ultraviolet divergent ( quadratically and logarithmically ) because of the asymptotic behavior of the quasiparticle dispersion relation @xmath90 , @xmath91 , and @xmath92 . however , the second term is convergent since the plasmino dispersion relation , @xmath93 , approaches @xmath56 exponentially at large @xmath56 .
the quadratic divergence is temperature independent and can be removed by subtracting the zero temperature part , coinciding with dimensional regularization , or by subtracting the free chiral susceptibility ( [ eq14 ] ) as discussed below .
the remaining logarithmic divergence turns out to be temperature dependent as in the case of the gluonic part of the free energy @xcite .
this might be an artifact of the approximation used here and might vanish if one goes beyond the one - loop htl approximation . here
we will use dimensional regularization to remove this divergence following ref .
@xcite .
first we subtract the free susceptibility given in ( [ eq14 ] ) from the htl one in ( [ 2.24 ] ) : @xmath94 \nonumber \\ & & + \frac{4n_fn_c}{\pi^2 } \int_{0}^\infty { \rm d}k \
, k^2 \ , \int_{0}^k dx \ , \left [ \beta_+(x , k ) \ , { \rm { re } } d_-^{-1 } ( x , k ) \ , n_f(x ) + \ , \beta_-(x , k ) \ , { \rm { re } } d_+^{-1 } ( x , k ) \ , n_f(x ) \right ] \nonumber \\ & & - \frac{2n_fn_c}{\pi^2 } \int_{0}^\infty { \rm d}k \ , k^2 \ , \int_{0}^k dx \ , \left [ \ , \beta_+(x , k ) \ , { \rm { re } } d_-^{-1 } ( x , k ) \ , + \ , \beta_-(x , k ) \ , { \rm { re } } d_+^{-1 } ( x , k)\right ] .
\label{fsub } \end{aligned}\ ] ] the integrals in 4@xmath95 and 6@xmath95 terms are , respectively , logarithmically ultraviolet divergent . using a momentum cutoff @xmath96 and energy cutoff @xmath97 , the 4@xmath95 and 6@xmath95 terms can be seen to be proportional to @xmath98 .
this is in contrast to the quark contribution to the free energy but similar to the case of the gluonic part of the free energy where these terms do not cancel @xcite .
dimensional regularization will then replace the remaining logarithmic divergences by poles in @xmath99 , which finally will require a counter term for cancellation @xcite . as pointed out in ref .
@xcite it is important to use also @xmath100-dimensional expressions for the htl propagator . here we will follow ref.@xcite and expand the chiral susceptibility up to order @xmath101 . the angular average leading to the legendre functions in ( [ 2.7 ] ) can be generalised to @xmath102 spatial dimensions in minkowski space as @xmath103 where the weight function @xmath104 is @xmath105 eq .
( [ 1 ] ) can also be written as @xcite @xmath106 now , it is convenient to define the functions @xmath107 and @xmath108 as @xmath109 which are related to the htl - resummed quark propagator ( [ 2.7 ] ) by @xmath110 using ( [ eq8 ] ) , ( [ 2.18 ] ) , and ( [ 2.20 ] ) the chiral susceptibility in one - loop htl order reads @xmath111 where @xcite @xmath112 with the arbitrary renormalization scale @xmath113 of the ( @xmath114 ) renormalization scheme and euler s constant @xmath115 .
expanding @xmath116 in a power series of @xmath117 , and keeping terms up to second order , we obtain @xmath118 .
\label{8}\end{aligned}\ ] ] using the sum - integrals evaluated in appendix a , we end up with @xmath119 \label{9}\end{aligned}\ ] ] the chiral susceptibility is logarithmic divergent at order @xmath120 , but finite at order @xmath101 .
we choose a counter term within the minimal subtraction renormalization scheme just to cancel the pole in @xmath121 : @xmath122 the complete expression for chiral susceptibility is then given by @xmath123 .
\label{11}\end{aligned}\ ] ] fig 3 : the htl static susceptibility ( free susceptibility subtracted ) as a function of temperature for @xmath124 mev , and @xmath125 with the choices of the renormalization scale @xmath126 and @xmath127 for up to order @xmath120 ( left ) and up to order @xmath101 ( right ) . note that the @xmath120-term is the first finite @xmath128-correction to the free chiral susceptibility , since the two - loop contribution in usual perturbative qcd diverges for @xmath42 .
the quantity @xmath129 as a function of @xmath130 is shown in fig.3 for two different choices of @xmath131 , where we have also chosen @xmath132 in @xmath128 entering @xmath133 .
obviously the @xmath101 contribution is only a small correction .
whereas in the @xmath120 contribution only the effective quark mass ( pole contribution ) enters , there is a landau damping contribution coming from @xmath134 in the @xmath101 term ( see ( [ 8 ] ) ) . to understand the behavior of the susceptibility in fig.3
we ascertain that the quantity @xmath135 , plotted in fig.3 , depends on the temperature only via the temperature dependent running coupling constant , i.e. @xmath136 $ ] .
this can be seen by scaling the variables @xmath56 and @xmath137 in the integrals of ( [ 2.24 ] ) by @xmath133 .
it holds also after dimensional regularization , if a renormalization scale proportional to the temperature is adopted in the latter case ( see e.g. ( [ 11 ] ) ) .
this dependence has also been verified numerically , where a constant , temperature independent chiral susceptibility has been observed , if a constant , temperature independent coupling has been employed . in other words , the decrease of the susceptibility with temperature in fig.3
is entirely caused by the temperature dependence of the coupling constant .
due to asymptotic freedom , @xmath138 , the htl susceptibility reduces to the free one in the infinite temperature limit , i.e. @xmath139=0 $ ] . in lattice simulations
a peak around the critical temperature associated with chiral symmetry restoration has been observed .
in contrast to lattice qcd the htl approximation , although taking into account medium effects in the plasma due to interactions , does not contain any physics related to the chiral restoration , which is a truly non - perturbative effect . in fig.3
we observe a strong increase of the magnitude of the htl chiral susceptibility towards low temperatures , similar as in lattice simulations above @xmath52 .
however , this increase is caused only by the fact that for our choice of the renormalization constant @xmath131 the htl result is above the free one , @xmath140 , and that the temperature dependent coupling increases at smaller temperatures .
if we choose a small enough value for @xmath131 , @xmath141 holds according to ( [ 11 ] ) , and the htl susceptibility @xmath142 approaches the free susceptibility @xmath143 from below at large temperatures .
this indicates the strong sensitivity to the choice of the regularization procedure and the renormalization constant similar as in the case of the free energy @xcite .
hence we understand the difference between the lattice and the htl approach qualitatively but the different ultraviolet behavior of the htl and the lattice result , which does not contain a temperature dependent logarithmic divergence @xcite , renders a direct , quantitative comparison of the both approaches difficult . here , we need to calculate the imaginary part of the self energy diagram given in fig.2 , where both bare quark propagators are replaced by effective htl quark propagators .
it is sufficient to restrict to the massless case , i.e. , ( [ 2.6 ] ) .
apart from constant factors the dynamic chiral susceptibility agrees with the scalar meson correlator at @xmath144 [ compare ( 2.3 ) of ref.@xcite with ( [ eq10a ] ) ] .
hence , we will not repeat the derivation here but just present the final results .
as in the case of the meson correlator @xcite , the dynamic susceptibility can be decomposed into pole - pole , pole - cut and cut - cut contributions , @xmath145 the pole - pole contribution reads @xmath146 ^ 2}{4m_q^4 } \big [ 1 -2n_f(\omega_+ ) + 2n_f^2(\omega_+)\big ] \nonumber \\ & & + \frac
{ [ \omega_-^2(k ) -k^2]^2}{4m_q^4 } \big [ 1 -2n_f(\omega_- ) + 2n_f^2(\omega_-)\big ] \nonumber \\ & & + \frac { [ \omega_+^2(k ) -k^2 ] [ \omega_-^2(k ) -k^2 ] } { 2m_q^4 } \big [ n_f(\omega_+)[1-n_f(\omega_-)]+ n_f(\omega_-)[1-n_f(\omega_+ ] \big ] \biggr\ } \ ; . \label{2.15 } \end{aligned}\ ] ] in the infinite temperature limit the plasmino mode decouples from the medium , causing the second and third terms to vanish , whereas the first term reduces to the free susceptibility given in ( [ eq17 ] ) .
the pole - cut contribution is given as @xmath147 n_f(\omega-\omega_+ ) n_f(\omega_+ ) \beta_+(\omega-\omega_+,k ) ( \omega_+^2-k^2 ) \nonumber\\ & ~&+ \theta[k^2-(\omega-\omega_-)^2]n_f(\omega-\omega_-)n_f(\omega_- ) \beta_-(\omega-\omega_-,k ) ( \omega_-^2-k^2 ) \nonumber\\ & ~&+\theta [ k^2-(\omega+\omega_-)^2 ] n_f(\omega+\omega_- ) [ 1-n_f(\omega_- ) ] \beta_+(\omega+\omega_-,k ) ( \omega_-^2-k^2 ) \nonumber\\ & ~&+\theta [ k^2-(\omega+\omega_+)^2 ]
n_f(\omega+\omega_+ ) [ 1-n_f(\omega_+ ) ] \beta_-(\omega+\omega_+,k ) ( \omega_+^2-k^2 ) \bigr\}\ ; , \label{2.16}\end{aligned}\ ] ] and the cut - cut contribution as @xmath148 \ ; . \label{2.17}\end{aligned}\ ] ] the dynamic chiral susceptibility in the 1-loop htl approximation , considered here , has a cubic ultraviolet divergence , which can be removed , for example , by subtracting the free result ( [ eq17 ] ) .
we will not discuss the dynamic susceptibility further on as it is not directly related to qcd lattice results @xcite .
we will just mention here that the ( ultraviolet divergent ) dynamic susceptibility , ( [ 2.15 ] ) to ( [ 2.17 ] ) , normalized to the free case ( [ eq17 ] ) is equal to unity , as can be seen from the scalar meson correlator at @xmath144 discussed in ref.@xcite .
in the present paper we have considered the chiral susceptibility of the quark - gluon plasma within lowest order perturbation theory ( free case ) and the htl approximation .
we have shown that there are two possible definitions of the chiral susceptibility leading to different results due to a missing conservation law associated with the chiral susceptibility in contrast to the quark number susceptibility @xcite . using the definition based on the derivative of the chiral condensate with respect to the current quark mass , we have computed this static chiral susceptibility within perturbative qcd and using an effective htl resummed quark propagator . in the case of the usual perturbation theory we found that the chiral susceptibility can not be computed beyond the leading order ( free case ) as it diverges due to setting the bare quark mass equal to zero at the end .
similar as in the case of the free energy @xcite but in contrast to the quark number susceptibility @xcite , the chiral susceptibility using the htl approach is ultraviolet divergent . even after subtracting the free susceptibility or the zero temperature part , a logarithmic temperature dependent singularity remains as in the case of the gluonic part of the free energy @xcite .
applying dimensional regularization the htl chiral susceptibility can be larger or smaller than the free one depending on the choice of the renormalization constant @xmath131 .
here we expanded the chiral susceptibility up to order @xmath101 .
the @xmath120 contribution is the lowest order correction to the free susceptibility , which has no @xmath42 divergence .
the temperature dependence of the htl chiral susceptibility is entirely determined by the temperature dependence of the running coupling constant that enters through the medium effects ( quasiparticle quark mass , landau damping ) contained in the htl approximation .
no indication of a peak at the critical temperature is found , as seen in lattice qcd simulations @xcite .
this indicates that this peak is due to truly non - perturbative effects such as chiral restoration , not included in the htl approach .
we would like to thank t. kunihiro , f. karsch and , in particular , m. strickland for useful correspondence .
is also thankful to e. braaten , c. greiner , k. gallmeister , s. juchem , and a. peshier for various discussion .
m.h.t . was supported by the dlr ( bmbf ) under grant no . 50wm0038 .
m.g.m . gratefully acknowledges the support from avh foundation and from gsi ( darmstadt ) .
following the same strategy , the remaining sum - integral in ( [ 8 ] ) involving @xmath160 can be written as @xmath161 where @xmath162 has been set to zero , as this integral is finite .
the result is given as @xmath163 \ \ .
\label{a11}\ ] ] 1 s. digal , e. laermann , and h. satz , euro .
c * 18 * , 583 ( 2001 ) .
t. hatsuda and t. kunihiro , phys . rep . *
247 * , 221 ( 1994 ) ; t. kunihiro , phys .
b*271 * , 395 ( 1991 ) .
f. karsch and e. laermann , phys .
d * 50 * , 6954 ( 1994 ) .
a. smigla and j.j.m .
verbaarschot , phys .
d * 54 * , 1087 ( 1996 ) .
p. zhuang , j. hfner , and s.p .
klevansky , nucl .
a * 576 * , 525 ( 1994 ) .
d. blaschke , a. hoell , c.d .
roberts , and s.m .
schmidt , phys .
c * 58 * , 1758 ( 1998 ) .
p. chakraborty , m.g .
mustafa , and m.h .
thoma , euro .
j. c * 23 * , 591 ( 2002 ) .
andersen , e. braaten , and m. strickland , phys .
lett . * 83 * , 2139 ( 1999 ) .
andersen , e , braaten , and m. strickland , phys .
d * 61 * , 014017 ( 1999 ) ; _ ibid _ 074016 ( 2000 ) .
e. braaten and r.d .
pisarski , nucl .
b * 337 * , 569 ( 1990 ) .
kapusta , _ finite - temperature field theory _
( cambridge university press , cambridge , 1989 ) . l. mclerran , phys .
d * 36 * , 3291 ( 1987 ) .
k. mukherjee , _ hep - ph/0106236_. j.i .
lapusta , p. lichard , and d. seibert , phys .
d * 44 * , 2774 ( 1991 ) .
callen and t.a .
welton , phys .
rev . * 122 * , 34 ( 1961 ) ; r. kubo , j. phys .
japan * 12 * , 570 ( 1957 ) ; d. forster , _ hydrodynamic fluctuation , broken symmetry and correlation functions _
( benjamin / cummings , melno park , 1975 )
. f. karsch , m.g .
mustafa , and m.h .
thoma , phys .
b * 497 * , 259 ( 2001 ) .
blaizot , e. iancu , and a. rebhan , _ hep - ph/0206280_. h. a. weldon , phys .
d * 26 * , 2789 ( 1982 ) .
e. braaten , r.d .
pisarski , and t.c .
yuan , phys .
lett . * 64 * , 2241 ( 1990 ) . j .-
blaizot , e. iancu , a. rebhan , phys .
d * 63 * , 065003 ( 2001 ) .
andersen and m. strickland , phys .
d * 66 * , 105001 ( 2002 ) .
andersen , e. braaten , e. petitgirard , and m. strickland , phys .
d * 66 * , 085016 ( 2002 ) .
f. karsch , private communications . | the static and dynamic chiral susceptibilities in the quark - gluon plasma are calculated within the lowest order perturbative qcd at finite temperature and the hard thermal loop resummation technique using an effective quark propagator .
after regularization of ultraviolet divergences , the hard thermal loop results are compared to qcd lattice simulations . |
Subsets and Splits
No saved queries yet
Save your SQL queries to embed, download, and access them later. Queries will appear here once saved.