article
stringlengths 0
745k
| abstract
stringlengths 0
92.7k
|
|---|---|
in a khler manifold which is calabi - yau or , more generally , khler - einstein , the mean curvature vector of a lagrangian submanifold is an infinitesimal symplectic motion and therefore preserves the lagrangian constraint .
this implies that the problem of minimizing volume among lagrangian submanifolds in a khler - einstein manifold is formally possible .
it also implies that mean curvature flow in a khler - einstein manifold preserves the lagrangian constraint for as long as the flow exists .
this property inspired thomas and yau to conjecture @xcite that a stable embedded lagrangian submanifold in a calabi - yau manifold with vanishing maslov class will , under classical mean curvature flow , converge to an embedded lagrangian submanifold with mean curvature zero .
such a limiting submanifold would then be a special lagrangian submanifold calibrated by a parallel section of the canonical line bundle , hence minimizing .
the precise statement of thomas - yau s conjecture allows for degeneration resulting from connect sums and other simple singularities , but the spirit of the conjecture is that classical mean curvature flow leads to special lagrangian submanifolds .
a. neves @xcite has shown that there are embedded lagrangian submanifolds in a calabi - yau manifolds with vanishing maslov class that , under classical mean curvature flow , develop singularities in finite time .
these examples give counterexamples to the thomas - yau conjecture if the assumption of stability is removed . on the other hand ,
mean curvature of any codimension for singular submanifolds ( rectifiable varifolds ) has been defined and studied by brakke @xcite .
it is natural to try to extend brakke s work to the lagrangian setting .
however this is not possible . by work of the second author
@xcite there are lagrangian surfaces in a calabi - yau surface with non - zero mean curvature which minimize area among lagrangian representing a homology class . under brakke s flow , indeed under any generalized mean curvature flow , the area of such a surface will decrease .
but since the surface minimizes among lagrangians , the flow can not then preserve the lagrangian constraint .
taken together these two observations seem to imply that neither classical nor weak mean curvature flow is useful in the construction of special lagrangians .
+ in this paper we introduce and study a _
lagrangian varifold flow_. this is a flow of certain lagrangian rectifiable varifolds in euclidean space , or , more generally , in a calabi - yau manifold , which we call _ maslov index zero rectifiable varifolds_. we give the precise definition below , but note here that an embedded lagrangian submanifold with zero maslov index in the usual sense satisfies our definition and an embedded lagrangian submanifold with non - vanishing maslov class in the usual sense does not . our first result concerns a lagrangian rectifiable varifold in euclidean space @xmath0 .
let @xmath1 be a compactly - supported maslov index zero varifold with @xmath2 in @xmath0 .
there is a one - parameter family @xmath3 , @xmath4 , of zero - maslov varifolds with @xmath2 with : 1 .
2 . @xmath6 is nonincreasing and lower - semicontinuous in @xmath7 .
3 . on spacetime regions where @xmath3 is a @xmath8-submanifold , @xmath3 moves by lagrangian mean curvature flow .
we next consider a lagrangian homology class @xmath9 in a closed calabi - yau manifold .
let @xmath10 be a rectifiable lagrangian cycle representing @xmath11 whose associated varifold is a maslov index zero has @xmath2 .
there is a one - parameter family @xmath12 of rectifiable cycles with : 1 .
2 . @xmath14 is nonincreasing and lower - semicontinuous in @xmath7 .
moreover , @xmath12 converges sequentially , as @xmath15 , to @xmath16 , which can be written as a sum of finitely many special lagrangian cycles ( possibly with different phases ) . as a consequence of our results
, we have : [ cor : thomas - yau ] let @xmath17 be a closed calabi - yau manifold .
if an integral lagrangian homology class @xmath18 can be represented by an imbedded lagrangian submanifold with zero maslov index then @xmath19 where each @xmath20 is a lagrangian homology class that can be represented by a special lagrangian current .
the phases of the calibrating @xmath21-forms may be different for each @xmath22 . the special lagrangian varifolds are volume minimizing in their homology class and therefore by the work of almgren @xcite have the regularity of volume minimizers , namely , they are regular except on a set of hausdorff codimension two .
we say they are _ special lagrangian varieties_. let @xmath17 be a closed calabi - yau manifold of complex dimension @xmath21 .
a class in @xmath23 is called _ a lagrangian homology class _ if it can be represented by a simplex consisting of simplices with all @xmath21-simplices lagrangian .
denote the subspace of lagrangian homology classes by @xmath24 .
let @xmath25 be the subspace of the lagrangian homology that is generated by the special lagrangian cycles .
let @xmath17 be a closed calabi - yau manifold of complex dimension @xmath21 .
if an integral lagrangian homology class @xmath18 can be represented by the image of a smooth map @xmath26 , where @xmath27 is a simply connected closed @xmath21-manifold then @xmath28 . in particular , the image of the hurewicz homomorphism @xmath29 in the lagrangian homology lies in @xmath30 .
let @xmath17 be a simply connected closed calabi - yau manifold of complex dimension @xmath21 . then for @xmath31 all lagrangian homology classes
lie in @xmath30 .
a central idea is the definition of a lagrangian integer - rectifiable varifold with maslov index zero .
we do not define the maslov class on such a varifold , only the more restricted notion of maslov index zero .
we exploit the geometry of the ambient calabi - yau manifold to make the definition . on any symplectic @xmath21-manifold
@xmath17 consider the fiber bundle whose fiber at @xmath32 consists of the grassmann bundle @xmath33 of oriented lagrangian @xmath21planes at @xmath34 . on a calabi - yau manifold
there is a complex @xmath35-form @xmath36 that is a parallel section of the canonical line bundle . evaluating this form on a lagrangian @xmath21-plane defines a map : @xmath37 that we call the @xmath38-valued lagrangian angle .
it determines a map : @xmath39 where @xmath40 is defined mod @xmath41 .
we call this the lagrangian angle .
a rectifiable @xmath21-varifold @xmath1 has an approximate tangent @xmath21-plane almost everywhere and therefore @xmath40 is well defined mod @xmath41 almost everywhere on @xmath1 .
if @xmath1 is a @xmath42 embedded lagrangian submanifold then @xmath40 is well defined mod @xmath43 everywhere , and the classical notion of maslov index zero is equivalent to the statement that @xmath40 has a continuous lift to a @xmath44-valued function . on a varifold , we can not require that @xmath40 has a continuous lift to a scalar - valued function .
we will require that @xmath40 has a lift to a scalar valued function in @xmath45 .
abusing notation we will denote this lift @xmath40 as well .
this condition is not itself adequate , since on a @xmath42 embedded lagrangian submanifold selecting such a lift is always possible by allowing @xmath40 to have jump discontinuities across codimension one sets . to eliminate the possibility of such jump discontinuities we introduce a notion of the weak derivative and require that the weak derivative @xmath46 of @xmath47 lies in @xmath48 with values @xmath49 . in sum ,
a lagrangian rectifiable varifold has maslov index zero if the @xmath38 lagrangian angle admits a lift to a scalar function @xmath47 and @xmath40 has a weak derivative in @xmath48 .
if @xmath1 is a lagrangian rectifiable varifold with maslov index zero then on @xmath1 we have a function @xmath40 .
we would like to use @xmath40 as a hamiltonian potential function to generate a flow ; however , @xmath40 is merely a function on @xmath1 . for each @xmath50
, we smooth @xmath40 to @xmath51 , a smooth function on @xmath17 given by mollification with respect to @xmath1 . using @xmath51 a hamiltonian potential to define a hamiltonian flow on @xmath17 and flowing for time @xmath52 yields a new lagrangian rectifiable varifold with maslov index zero . iterating and letting @xmath53 we construct a one - parameter family of lagrangian varifolds for each @xmath54 .
then we let @xmath55 to construct the `` hamiltonian flow '' referred to above .
this `` hamiltonian flow '' consists of integral lagrangian rectifiable varifolds with maslov index zero .
the mass of the varifolds is non - increasing and decreasing if the lagrangian angle is not a constant .
we are indebted to a number of excellent works on varifolds .
we mention the work of brakke @xcite on generalized mean curvature flow , the work of allard @xcite on varifolds , the recent work of menne @xcite on rectifiable varifolds with locally bounded first variation and finally the book of simon @xcite on the whole subject of geometric measure theory .
let @xmath56 be a khler manifold of complex dimension @xmath21 and let @xmath57 be a real @xmath21-dimension imbedded ( or immersed ) submanifold .
we say @xmath58 is _ lagrangian _ if @xmath59 .
let @xmath60 be the mean curvature vector field along @xmath58 and @xmath61 denote the ricci two - form of @xmath27 .
a simple geometric computation yields @xmath62 in particular if the khler metric is khler - einstein so that @xmath63 then @xmath64 . from cartan s formula
it follows that @xmath65 thus , @xmath60 is an infinitesimal symplectic motion .
it then follows that : if @xmath56 is a khler - einstein manifold and @xmath57 is an immersed lagrangian submanifold then classical mean curvature flow exists for some time interval @xmath66 and for all @xmath67 the submanifold @xmath68 is lagrangian .
classical mean curvature flow can not be defined when the flow develops a singularity and so , in general , classical mean curvature flows exist only finite intervals .
brakke @xcite was able to define a `` generalized mean curvature flow '' for a class of singular submanifolds known as rectifiable varifolds .
rectifiable varifolds have a tangent plane almost everywhere but are otherwise highly singular . to define a lagrangian rectifiable varifold we require the tangent plane to be a lagrangian plane .
then it is reasonable to expect that the brakke flow preserves the lagrangian condition if the ambient manifold if khler - einstein .
however , this is false due to a result of the second author . in general the brakke flow does not preserve the lagrangian constraint . to explain this result we must find a lagrangian rectifiable varifold in a khler - einstein manifold for which the brakke flow exists but does not preserve the lagrangian constraint .
we consider a lagrangian homology class represented by an immersed two - sphere in a k3 surface .
use the minimization procedure of @xcite to construct a lagrangian two - sphere that minimizes area in its homology class . by careful choice of the lagrangian homology class
it can be shown that the class can not be represented by a holomorphic curve for any complex structure compatible with the metric .
in particular the lagrangian minimizer is not a holomorphic curve for any compatible complex structure .
applying the regularity theory of @xcite , the minimizer is regular ( smooth ) except at finitely many points which are ( i ) branch points , or ( ii ) sw singularities .
the sw singularities are modeled by a countable family of non - trivial lagrangian cones that can be written explicitly . around the vertex of each cone there
is non - zero maslov index .
if there are no sw singularities then the minimizer is a classical ( branched ) minimal surface that is lagrangian . for a compatible complex structure
such a surface is a holomorphic curve .
but this has been ruled out and therefore there must be sw singularities and the minimizer can not be a classical minimal surface .
in particular the lagrangian minimizer is a lagrangian rectificable varifold with non - vanishing generalized mean curvature .
the brakke flow decreases the area of the lagrangian minimizer and preserves the homology class .
but since the lagrangian minimizer minimizes area among lagrangians the brakke flow can not preserve the lagrangian constraint .
the sw singularities lie at the core of these examples .
we observe that the link of any sw singularity has non - zero maslov index . because our definition of maslov index zero precludes even local maslov index , the present paper is a demonstration that it is the non - zero _ local _ maslov index which obstructs the existence of any weak lagrangian mass - decreasing flow .
let @xmath56 be a khler @xmath21-manifold . at each point @xmath32
a real dimension @xmath21 subspace @xmath69 is called _ lagrangian _ if @xmath70 . at each point
@xmath32 we consider the grassmann manifold of lagrangian @xmath21-planes @xmath71 .
denote this manifold @xmath72 and note that @xmath73 can be identified with the homogeneous space @xmath74 .
these manifolds are the fibers of a bundle over @xmath17 that we call the lagrangian grassmann bundle and denote @xmath75 .
( we can instead consider the grassmann manifold of oriented lagrangian @xmath21-planes @xmath71 . in this case
the relevant grassmann manifold is identified with the homogeneous space @xmath76 . )
suppose next that @xmath56 has a calabi - yau metric .
this implies that the canonical line bundle is geometrically trivial , in particular , that it has a non - zero parallel section .
such a section is a nowhere vanishing closed @xmath35 form that we denote @xmath77 .
restricting @xmath77 to a lagrangian @xmath21-plane determines a unit complex number and therefore @xmath77 defines a smooth map : @xmath78 we call this map the lagrangian angle .
we will often write @xmath79 , where the function @xmath40 is well - defined mod @xmath41 .
( in the oriented case , @xmath40 is well - defined mod @xmath43 . ) along a lagrangian submanifold @xmath80 , tangent plane @xmath81 is lagrangian and we can write @xmath82 . on @xmath80
the function @xmath40 is well - defined mod @xmath41 .
( in the oriented case , @xmath40 is well - defined mod @xmath43 ) . in both cases the tangential gradient of @xmath40 along @xmath80 , @xmath83 , is well defined .
if @xmath84 is a lagrangian immersion , we can consider the assignment @xmath85 , which gives a mod-@xmath43 smooth map @xmath86 ( which we continue to denote by @xmath87 ) .
it is immediate that @xmath88 is a well - defined closed one - form so @xmath89\in h^1(\sigma;{\mathbb r})$ ] , called the _ maslov class _ of the immersion @xmath58 .
a simple computation ( see e.g. @xcite ) shows that , if @xmath60 is the mean curvature of @xmath58 , then @xmath90 the _ maslov index _ around a one - cycle @xmath11 in @xmath80 is given by @xmath91 which computes the winding of @xmath92 as @xmath34 traverses @xmath11 .
in particular , @xmath93=[d\beta]$ ] lies in the @xmath94 lattice of @xmath95 .
[ thm : maslov - zero ] if @xmath96 for all one - cycles @xmath11 then the lagrangian angle @xmath40 admits a smooth lift to a smooth scalar valued function ; conversely if the lagrangian angle @xmath40 admits a smooth lift to a smooth scalar valued function then @xmath96 for all one - cycles @xmath11 . our approach below will be to define a flow using the lagrangian angle @xmath87 . as
this function is central to our construction , we need to understand how @xmath87 behaves under smooth deformations which preserve the lagrangian condition . we will use the following convenient way to compute the lagrangian angle . given a lagrangian @xmath21-plane @xmath97 at @xmath98 with oriented orthonormal basis @xmath99,the
associated orienting form is @xmath100 .
the associated complexified orienting form is @xmath101 it is elementary that the lagrangian condition implies @xmath102 has the properties : @xmath103 note that the space of @xmath21-vectors on @xmath104 satisfying ( [ eqn.n0form ] ) is of complex dimension 1 .
the @xmath21 vector @xmath105 dual to the parallel section @xmath106 , when restricted to @xmath107 , also satisfies ( [ eqn.n0form ] ) , so it must be a unit complex multiple of @xmath102 . in fact@xmath108
let @xmath109 be an infinitesimal symplectic motion .
then the lagrangian angle @xmath87 of the plane @xmath97 has first variation @xmath110 where @xmath111 is the levi - civita connection and @xmath112 is the almost - complex structure coming from the calabi - yau manifold @xmath113 .
let @xmath114 be a one - parameter family of diffeomorphisms @xmath115 , generated by the vector field @xmath109 ( which we assume is @xmath116 ) . by assumption @xmath114 preserves the lagrangian condition , so that @xmath117 induces a diffeomorphism @xmath118 on @xmath119 which covers @xmath117 .
then for any @xmath98 , there are coordinates for @xmath17 at @xmath34 in which , for small @xmath7 , @xmath120 if @xmath97 is a lagrangian plane with orthonormal frame @xmath121 , we compute @xmath122 here we have used the fact that @xmath97 is lagrangian to decompose @xmath123 into normal and tangential components , and write everything in terms of @xmath121 . to compute the variation of the lagrangian angle , we will consider the function @xmath124 .
then we have @xmath125 where @xmath126 is the levi - civita connection coming from the bundle metric on @xmath75 induced by the calabi - yau metric on @xmath17 and the homogeneous metric on each fiber .
to compute @xmath127 , note that the tangent plane @xmath128 splits as @xmath129 . in this splitting , @xmath130 . because @xmath36 is parallel
, so is @xmath131 and we have @xmath132 .
if we choose an orthonormal frame @xmath133 for a lagrangian plane @xmath97 and identify @xmath97 with the orienting form @xmath100 , we can give a basis for @xmath134 : @xmath135 for @xmath136 . to compute the @xmath137 component of @xmath138 ,
let @xmath139 be a path of planes with @xmath140 and @xmath141 .
then we compute @xmath142 so we have the formula @xmath143 applying ( [ eqn.grassgradbeta ] ) to ( [ eqn.deltabeta ] ) , we have @xmath144 since @xmath87 is only ambiguous up to addition of a multiple of @xmath43 , @xmath145 ( and therefore @xmath146 ) is well - defined .
( [ eqn.deltabeta2 ] ) yields @xmath147 to carry out this computation , we identified @xmath97 with an orienting form @xmath148 .
had we chosen the opposite orientation , we would have computed @xmath149 using @xmath150 , which would result in an angle of @xmath151 .
the formula is not affected by the addition of this constant .
we will be concerned mostly with one - parameter families of integer - multiplicity rectifiable varifolds . for details on the theory of rectifiable varifolds ,
we refer the reader to leon simon s book @xcite ; we briefly recall the salient points here as formulated in @xcite , @xcite , and @xcite for general dimension and codimension . following simon s notation ,
if @xmath27 is a countably @xmath21-rectifiable subset of @xmath152 with multiplicity @xmath153 , we write @xmath154 . to each such @xmath154
we may associate a radon measure on @xmath155 , @xmath156 , given by @xmath157 in particular , if @xmath158 is @xmath159-measurable , then @xmath160 the rectifiable set @xmath27 has an approximate tangent @xmath21-plane with respect to @xmath161 for @xmath159 a.e .
@xmath34 , denoted @xmath162 .
we then define the approximate tangent plane to the varifold @xmath1 by : @xmath163 for @xmath159 a.e .
@xmath34 .
if @xmath109 is an ambient @xmath42 vector field the divergence of @xmath109 along @xmath1 , written @xmath164 , is given by @xmath165 the first variation formula for the rectifiable varifold @xmath1 with respect to the variation @xmath109 is : @xmath166 we will also use a more general notion , following allard @xcite , simply called a _ @xmath21-varifold _ , which is a radon measure on the grassmann bundle @xmath167 of @xmath21 planes in @xmath152 .
it is elementary that to each integer - rectifiable @xmath21-varifold @xmath1 there is an associated @xmath21-varifold @xmath168 given by @xmath169 where @xmath170 is the projection map .
the first variation formula for varifolds in allard s sense is : @xmath171 we also note that each @xmath21-varifold induces a radon measure @xmath172 on @xmath152 , and a probability measure @xmath173 on each fiber @xmath174 , so that @xmath175 we say that @xmath1 has _ locally bounded first variation in u _ if for each @xmath176 with @xmath177 there is a constant @xmath178 such that for all @xmath42 vector fields @xmath109 in @xmath17 with @xmath179 : @xmath180 define the total variation measure of @xmath1 on @xmath181 by : @xmath182 for any open @xmath176 with @xmath177
. then @xmath183 being a radon measure on @xmath181 is equivalent to @xmath1 having _ locally bounded first variation in u_. integer - rectifiable varifolds with locally bounded first variation satisfy a compactness result .
for the proof see @xcite or @xcite .
[ thm : compactnessvarifolds ] suppose @xmath184 is a sequence of integer - rectifiable varifolds in @xmath181 that have locally bounded first variation in @xmath181 and suppose @xmath185 is bounded for all @xmath176 with @xmath177 . then there is a subsequence @xmath186 and an integer - rectifiable varifold @xmath1 of locally of bounded first variation in @xmath181 such that @xmath187 in the sense of radon measures on @xmath167 . moreover , for all @xmath176 with @xmath177 , @xmath188 let @xmath1 be an integer - rectifiable varifold with locally bounded first variation .
write @xmath189 where @xmath190 is @xmath183-measurable with @xmath191 @xmath183 a.e . using the differentiation theory of radon measures ,
the function @xmath192 exists @xmath193 almost everywhere and is @xmath193 measurable . for any borel set @xmath194 , @xmath195 where @xmath196 where @xmath197 is a borel set of @xmath193-measure zero .
set @xmath198 and call @xmath199 the _ generalized mean curvature _ of @xmath1 .
@xmath197 is called the _ generalized boundary _ of @xmath1 .
see @xcite for more details .
it follows that we can write : @xmath200 continuing to assume that @xmath1 is an integer - rectifiable varifold with locally bounded first variation following allard @xcite we impose an additional condition .
let @xmath201 and @xmath202 be its hlder conjugate .
let @xmath203 be a constant .
@xmath204 the following proposition is essentially due to allard @xcite , 8.1 .
[ prop : no - sing ] if @xmath1 is an integer - rectifiable varifold with locally bounded first variation satisfying ( [ equ : allard - bound ] ) for @xmath201 then @xmath205 , @xmath206 with @xmath207 and for any compactly - supported @xmath42 test vector field @xmath109 , @xmath208 recall ( [ equ : handsing ] ) @xmath196 where @xmath197 is a borel set of @xmath193-measure zero .
let @xmath209 be an open set with @xmath210 and @xmath211 .
let @xmath212 be the function that is identically one on @xmath213 and zero on @xmath209 .
set @xmath214 .
then @xmath215 and @xmath216 as @xmath55 .
we have : @xmath217 as @xmath55 and @xmath218 for all @xmath54 .
therefore , letting @xmath55 , we have @xmath219 it follows from ( [ equ : allard - ineq ] ) that @xmath206 and @xmath220
. + for any @xmath221 there is @xmath222 , a smooth vector field with compact support , so that @xmath223 and @xmath224 . for each @xmath225 ,
let @xmath226 be a smooth function so that @xmath227 , @xmath228 on @xmath229 , and @xmath230 .
set @xmath231 .
then we have @xmath232 so that @xmath233and @xmath234 on the other hand , @xmath235 therefore @xmath205 .
the result follows . for @xmath1
is an integer - rectifiable varifold with locally bounded first variation , menne @xcite defines condition @xmath236 by : @xmath237 proposition [ prop : no - sing ] shows for @xmath238 , allard s condition ( [ equ : allard - bound ] ) implies @xmath236 ; the converse implication is immediate .
the advantage of allard s condition is the following compactness result .
we will refer to the following result as allard s compactness theorem .
[ thm : compact - hp - varifolds ] suppose @xmath184 is a sequence of integer - rectifiable varifolds in @xmath181 that have locally bounded first variation in @xmath181 and satisfy @xmath236 with @xmath201 and @xmath239 for some @xmath240 independent of @xmath241 .
suppose @xmath185 is bounded for all @xmath176 with @xmath177 .
then there is a subsequence of @xmath186 and an integer - rectifiable varifold @xmath1 of locally of bounded first variation in @xmath181 also satisfying @xmath236 such that @xmath187 in the sense of radon measures on @xmath167 .
using theorem [ thm : compactnessvarifolds ] there is a subsequence @xmath186 and an integer - rectifiable varifold @xmath1 of locally of bounded first variation in @xmath181 such that @xmath187 in the sense of radon measures on @xmath167 .
given any smooth test vector field @xmath109 , observe that @xmath242 and @xmath243thus condition holds for @xmath1 . by proposition [ prop : no - sing ] , the theorem follows .
the integer - rectifiable varifolds that are locally of bounded first variation and that satisfy @xmath236 with @xmath244 are of particular importance in this paper , in part because of the previous compactness result and in part because of the following partial regularity results proved by menne @xcite .
[ thm : weak - regularity - m2 ] suppose @xmath181 is an open subset of @xmath245 and @xmath1 is an integer rectifiable @xmath21-varifold with locally bounded first variation ( equivalently , with @xmath183 a radon measure )
. then there exists a countable collection @xmath246 of @xmath21-dimensional @xmath8 submanifolds of @xmath245 such that @xmath247 and each member @xmath248 satisfies : @xmath249 and @xmath250 for @xmath156 almost all @xmath251 .
[ thm : menne - tilt - excess ] suppose @xmath181 is an open set in @xmath245 , @xmath252 and @xmath1 is an integer - rectifiable @xmath21-varifold satisfying @xmath236 .
let @xmath253 be the orthogonal projection to @xmath254 . if @xmath255 and @xmath201 or @xmath256 and @xmath257 then for @xmath258 a.e .
@xmath259 : @xmath260 in particular the theorem applies to @xmath244 ; we will henceforth make use almost exclusively of the @xmath244 case
. we will refer to @xmath259 satisfying the conclusions of theorems [ thm : weak - regularity - m2 ] and [ thm : menne - tilt - excess ] as _ points of @xmath8 rectifiability_. theorems [ thm : weak - regularity - m2 ] and [ thm : menne - tilt - excess ] depend in turn on the following tilt - excess decay estimate proved by brakke @xcite and menne @xcite : [ thm : tilt - excess - b - m ] suppose @xmath181 is an open subset of @xmath245 and @xmath1 is an integer - rectifiable @xmath21-varifold with locally bounded first variation ( equivalently , with @xmath183 a radon measure ) . for @xmath156-a.e .
@xmath261 , the approximate tangent space @xmath262 exists and @xmath263 in fact we will use these notions as they apply to lagrangian subspaces .
hence we assume @xmath264 .
we say that the integer rectifiable @xmath21-varifold @xmath1 is _ a lagrangian integer rectifiable varifold _ if for @xmath159 a.e . @xmath34
the approximate tangent plane @xmath265 is a lagrangian @xmath21-plane .
lagrangian varifold _ is an @xmath21-varifold which has support contained in the grassmann bundle @xmath266 of lagrangian @xmath21-planes . because the grassmannian of lagrangian planes at @xmath34 , @xmath267 , is closed as a subset of all the grassmannian of all @xmath21-planes @xmath174 , and @xmath75 is closed in @xmath167 , the compactness theorem [ thm : compactnessvarifolds ] yields a compactness theorem for lagrangian integer - rectifiable varifolds as well .
we remark also that we may replace @xmath268 with a calabi - yau manifold @xmath113 in a straightforward way to obtain versions of all theorems in this section which apply to lagrangian integer - rectifiable varifolds and lagrangian varifolds in @xmath113 , though for exposition we will remain mostly in the euclidean context . for reference
, we include the following table of terminology for the various classes of varifolds we will consider in the remainder of this paper , ordered by inclusion : * _ lagrangian varifold _ , a radon measure on the bundle @xmath75 .
* _ lagrangian integer varifold _ or _
lagrangian integer - rectifiable varifold _ ,
an integer - rectifiable varifold whose tangent planes are almost all lagrangian . *
_ varifold with maslov index zero _ , a lagrangian integer varifold with a weakly - differentiable lift of the lagrangian angle , as described in
[ sec : maslovindexzero ] . * _ varifold with maslov index zero and mean curvature in @xmath269 _ , a varifold with maslov index zero which in addition satisfies the hypotheses of theorem [ thm : menne - tilt - excess ] .
we will choose @xmath244 for definiteness . because we will be interested in applications to lagrangian homology classes in a closed calabi - yau manifold
, we will henceforth assume that all varifolds have finite mass and compact support .
in what follows , inspired by brakke s construction , we use mollification , but rather than mollify the varifold @xmath1 as brakke does , we will mollify functions with respect to the varifold measure @xmath156 .
our exposition follows that of evans @xcite . for exposition
we state results in this section for varifolds in @xmath268 ; the adaptation to the manifold case is straightforward and discussed in [ sec : manifoldcase ] . + consider the function @xmath270 given by @xmath271 observe that @xmath272 for all @xmath7 . for any @xmath50 ,
the mollifier at scale @xmath273 is @xmath274 given by @xmath275where @xmath246 is chosen so that @xmath276 .
observe that the support of @xmath277 is @xmath278 .
if @xmath279 is @xmath156-integrable , its @xmath273-mollification with respect to @xmath1 , @xmath280 , is @xmath281 we choose the mollifying function @xmath282 to have compact support because this choice is easily adapted to the manifold case . to ensure that pointwise @xmath283 as @xmath55 we need to normalize the mollification . in the case of mollification with respect to lebesgue measure on @xmath284 ,
the normalization would be @xmath285 , but for us , this quantity must depend on @xmath34 .
the most natural choice of normalization is @xmath286 .
however because @xmath282 is compactly supported , the normalization vanishes for @xmath34 sufficiently far from @xmath287 and this may introduce singularities into @xmath280 .
we use the additional term @xmath288 to ensure that the denominator in @xmath280 does not vanish , and thus that @xmath289 is smooth for all @xmath290 .
mollification can be considered as a linear map @xmath291 ; that is , for @xmath292 define @xmath293 then @xmath294 is a smoothing operator . the theory of mollification with respect to a varifold is largely similar to the standard theory of mollification with respect to lebesgue measure on @xmath284 .
for completeness we state the following basic facts about mollification .
our exposition follows that of evans @xcite , mutatis mutandis .
@xmath295 .
clear . for classical mollification with respect to lebesgue measure ( see for example @xcite ) , because @xmath296 , trivially @xmath297 in our case we have the weaker statement : [ lem : dphiepsilonbounded ] for @xmath34 in the support of @xmath156 where @xmath1 has an approximate tangent plane , @xmath298 is bounded independent of @xmath273 .
we compute @xmath299 : @xmath300 therefore setting @xmath301 , @xmath302 as @xmath303 , @xmath304 ; the integration converges to integration on @xmath305 .
moreover , the integrand is uniformly bounded .
we remark that this lemma does not assert that @xmath306 is uniformly bounded in @xmath307 .
[ lem : mollifiertomultiplicity ] if @xmath34 is a point where @xmath1 has an approximate tangent plane , @xmath308 as @xmath303 , where @xmath309 is the multiplicity of @xmath1 at @xmath34 .
@xmath310 where we have used the fact that @xmath311 .
we have the following explicit bounds on the derivatives of @xmath280 , which degenerate as @xmath303 but will nonetheless be useful in [ sec : epsilonflows ] .
there is a universal constant @xmath246 so that for any @xmath312 , @xmath313 for each @xmath314 .
first observe that @xmath315 observe that because @xmath316 we have @xmath317 , where @xmath318 is the maximum of @xmath319 on @xmath320 . now
compute @xmath321 : @xmath322 [ lem : derivativesofbetaepsilon ] for any @xmath323 , there is a universal constant @xmath324 so that for any @xmath312 , we have the estimate @xmath325 for each @xmath314 .
the proof for the higher derivatives of @xmath280 is similar to the proof for the first derivative and is left to the reader .
[ thm : strong - conv - lp ] for @xmath252 , if @xmath326 then @xmath327 strongly in @xmath328 .
[ lem : continuousmollifier ] if @xmath279 is continuous , then @xmath329 for @xmath156-a.e .
@xmath34 .
let @xmath330 .
consider @xmath331 .
we have @xmath332 since @xmath279 is continuous for any @xmath333 there is @xmath273 so small that @xmath334 on @xmath335 .
for such @xmath273 , is estimated by @xmath336 suppose @xmath1 has compact support .
if @xmath279 is continuous , then @xmath327 strongly in @xmath328 for each @xmath337 .
since @xmath279 is continuous and @xmath1 has compact support , @xmath338 .
so by , @xmath339 is dominated by @xmath340 , so the dominated convergence theorem together with lemma [ lem : continuousmollifier ] gives @xmath327 in @xmath269 .
recall that @xmath341 is the mollification operator , @xmath342 .
[ prop : bound - on - moll - op ] consider @xmath343 then there is a constant @xmath246 , independent of @xmath273 , @xmath344 , and @xmath279 so that @xmath345 consider some @xmath346 .
we have @xmath347 therefore @xmath348 observe that as @xmath303 , the measure @xmath349 tends to @xmath350 ; thus the right - hand side of tends to @xmath351 . on the other hand at @xmath352 the quantity @xmath353 is finite and independent of both @xmath279 and @xmath344 .
the claim follows . in the case
@xmath354 , given any @xmath355 , since @xmath156 is a radon measure , there is a sequence @xmath356 of continuous functions with @xmath357 in @xmath269 ( see , e.g. , @xcite ) .
we have by proposition [ prop : bound - on - moll - op ] @xmath358 now for any @xmath359 , since @xmath357 in @xmath328 , the first and third terms can each be made smaller than @xmath360 ; since each @xmath356 is continuous , lemma [ lem : continuousmollifier ] allows us to make the second term smaller than @xmath360 . for the case @xmath361 , we may as well assume the support of each function involved is compact , apply the @xmath269 case , and let @xmath344 tend to @xmath362 .
for use later we observe that mollification with respect to @xmath1 gives a proof of the following proposition .
[ prop : mollific ] let @xmath355 .
then there is a sequence @xmath363 with @xmath357 strongly in @xmath328 .
set @xmath364 .
we wish to extend the notion of vanishing maslov class to varifolds .
we do this by observing that in the smooth case , the vanishing of the maslov class is equivalent to the existence of a continuous lift of the lagrangian angle . in the varifold
setting , the idea of a ` continuous lift ' of the lagrangian angle is not sensible , because the varifold itself is not a continuous object .
we therefore consider varifolds which admit a weakly differentiable lift of the lagrangian angle . in this section
, we will make precise a notion of weak derivative for functions on an integer - rectifiable varifold and use it to define a class of lagrangian integer rectifiable varifolds we call _ maslov index zero_. we use mollification as described in
[ sec : mollification ] and menne s second - order rectifiability result to establish a formula relating the weak derivative of the lagrangian angle to the generalized mean curvature .
finally we show show that the class of lagrangian varifolds with maslov index zero is compact in a reasonable sense .
+ to compute the weak derivative of a function on a varifold we will use the first variation formula . formally the computations are possible on , for example , varifolds with bounded first variation .
however the generalized boundary or , equivalently , the singular part of the variation measure @xmath365 will come into these computations . to simplify the computations
we will formulate the weak derivative on varifolds satisfying the condition @xmath366 , @xmath367 .
this condition also satisfies a compactness result , theorem [ thm : compact - hp - varifolds ] , that is essential for the applications of this paper .
+ throughout , let @xmath1 be a rectifiable @xmath21-varifold in the riemannian @xmath368-manifold @xmath17 , with induced radon measure @xmath156 that satisfies @xmath366 , @xmath367 . if @xmath369 is a lipschitz function in a neighborhood of the support of @xmath156 then the tangential component of the derivative of @xmath369 is well - defined @xmath156-a.e . and will be denoted @xmath370 .
suppose @xmath1 is a rectifiable @xmath21-varifold satisfying @xmath236 for @xmath238 .
let @xmath279 be a @xmath156-integrable function .
we say that @xmath279 has a _ weak derivative _
@xmath371 , where @xmath371 has values in @xmath49 , if @xmath371 is @xmath258-integrable and for any @xmath372 : @xmath373 weak derivatives , if they exist , are unique . given @xmath374 weak derivatives for @xmath279 , we have for any test function @xmath375 , @xmath376 and since this holds for all test functions @xmath375 , we have @xmath377 @xmath156-almost everywhere . to illustrate the notion of weak derivative we consider the special case in which @xmath378 is open , @xmath379 and @xmath380 is a lipschitz function on @xmath181 such that on @xmath381 , @xmath382 ( i.e. , @xmath279 can be extended to a lipschitz function on @xmath181 ) . [
prop : weak - derivative - lipschitz ] let @xmath1 be a rectifiable @xmath21-varifold with induced radon measure @xmath156 that satisfies @xmath366 , @xmath367 .
on @xmath381 , @xmath383 is well - defined .
moreover , @xmath279 has a weak derivative @xmath371 on @xmath381 , given by @xmath384 , where @xmath60 is the mean curvature of @xmath1 .
let @xmath369 be a compactly supported smooth test function with @xmath385 contained in an open coordinate neighborhood @xmath176 .
let @xmath386 be a covariant constant frame in @xmath176 .
observe that @xmath387 is an admissible variation , so we can apply the first variation formula .
we compute @xmath388 where we have used the leibniz rule for the divergence operator and the fact that @xmath389 is covariant constant .
let @xmath80 be a smooth embedded submanifold without boundary , and @xmath181 an open set so that @xmath390 and @xmath391 have nonempty interior in @xmath80 .
let @xmath1 be the varifold arising from @xmath80 .
then the characteristic function @xmath392 does not have a weak derivative on @xmath1 .
the fact that there is no weak derivative for such @xmath279 is central to our purposes .
we use the lack of a weak derivative to detect jump discontinuities " .
we have only defined the notion of weak derivative for varifolds satisfying @xmath236 for @xmath238 .
in particular the generalized boundary vanishes in this case . for varifolds with nonvanishing generalized boundary ,
the formula for a weak derivative would necessarily involve a generalized boundary term .
menne @xcite @xcite has studied function theory on varifolds in a general setting ; these results are potentially important for generalizations of the present paper .
the quantity @xmath393 also appears in the sobolev inequality proved by michael - simon ( @xcite , theorem 2.1 ) and allard ( @xcite , theorem 7.3 ) : [ thm : ms ] let @xmath1 be an integer - rectifiable @xmath394-varifold in @xmath152 that satisfies @xmath366 , @xmath367 . then there is a universal constant @xmath395 so that for any @xmath396 , @xmath397 using proposition [ prop : weak - derivative - lipschitz ] , we may rephrase this result
so that it looks more like the classical sobolev inequality : [ thm : msrestate ] let @xmath1 be an integer - rectifiable @xmath394-varifold in @xmath152 that satisfies @xmath366 , @xmath367 .
then there is a universal constant @xmath395 so that for any @xmath396 , @xmath398 where @xmath181 is the weak derivative of @xmath399 with respect to @xmath1 .
[ divergenceformula ] @xmath371 is a weak derivative for @xmath279 if and only if for each compactly - supported smooth test vector field @xmath109 we have @xmath400 given a test vector field @xmath109 , we can write @xmath401 where @xmath402 are a covariant constant ambient orthonormal frame . we have : @xmath403 because the @xmath404 are covariant constant .
now if @xmath371 is a weak derivative for @xmath279 , we can rewrite the latter integral as @xmath405 where again we have used the fact that each @xmath404 is covariant constant .
conversely , given a test function @xmath375 , apply the formula @xmath406 to the vector field @xmath407 for some covariant constant vector field @xmath408 to obtain @xmath409 .
recall menne s decomposition theorem for varifolds stated in theorem [ thm : weak - regularity - m2 ] : if @xmath183 is a radon measure then there is a countable collection @xmath410 such that : @xmath411 where @xmath412 has @xmath258-measure zero and for each @xmath413 there is a @xmath8 submanifold @xmath414 with @xmath415 .
moreover , the generalized mean curvature @xmath60 of @xmath1 satisfies for a.e .
@xmath416 @xmath417 where @xmath418 is the mean curvature along @xmath414 . given an ambient @xmath8 vector field @xmath109 we can decompose @xmath109 into tangential @xmath419 and normal @xmath420 components along @xmath414 .
both @xmath419 and @xmath420 can be extended to @xmath42 ambient vector fields and therefore both @xmath421 and @xmath422 are well - defined integrable functions .
we define the decomposition of @xmath109 into tangential @xmath419 and normal @xmath420 components along @xmath423 using the decomposition along @xmath414 and define @xmath424 @xmath425 we have , thus , defined the decomposition of @xmath109 into tangential @xmath419 and normal @xmath420 components along @xmath1 @xmath258 a.e . and defined the integrable functions @xmath426 and @xmath427 @xmath258 a.e .
the next result gives a partial characterization of the weak derivative along a varifold that satisfies @xmath236 for @xmath367 .
[ thm : weak - derivative - menne ] suppose @xmath1 is a @xmath21-varifold that satisfies @xmath236 for @xmath367 .
suppose @xmath428 with weak derivative @xmath429 .
then @xmath371 admits a decomposition @xmath430 where @xmath431 satisfies @xmath432 for @xmath258 a.e .
moreover , @xmath433 first we claim that if @xmath434 then @xmath435 to see this we compute the first variation of @xmath156 with respect to the vector field @xmath436 , using menne to decompose @xmath437 into @xmath42 vector fields on the @xmath8 pieces of some menne decomposition and using the fact that pointwise ( as in the @xmath438 case ) @xmath439 , @xmath440 on the other hand , by the definition of @xmath60 , we have @xmath441 . next , suppose that @xmath442 is a sequence of @xmath443 functions with @xmath357 in @xmath444 , constructed say using proposition [ prop : mollific ] .
for any @xmath445 , @xmath446 which converges ( using proposition [ prop : weak - derivative - lipschitz ] ) to @xmath447 since @xmath109 is arbitrary , this means @xmath448 in @xmath328 .
moreover , @xmath60 is @xmath156-almost everywhere normal and @xmath449 are everywhere tangential , so @xmath450 .
the claimed integration - by - parts formula for @xmath451 follows from the fact that for each @xmath452 , @xmath453 [ powerrule ] let @xmath1be an @xmath21-varifold that satisfies @xmath236 for @xmath367 .
suppose @xmath454 has a weak derivative @xmath455 , and @xmath456 for some @xmath457 .
then @xmath380 has weak derivative @xmath458 using proposition [ prop : mollific ] , let @xmath356 be a sequence of @xmath459 functions with @xmath357 in @xmath45 .
each @xmath356 satisfies the claimed formula by direct computation , that is , @xmath460 has weak derivative @xmath461 we have @xmath462 ; as in the proof of the theorem this goes to @xmath463 .
clearly @xmath464 .
so @xmath465 on the other hand , we have for any smooth test function @xmath466 @xmath467and taking the limit in @xmath452 we obtain that @xmath468 is a weak dervative for @xmath380 . more generally , we have the following chain rule : let @xmath1be an @xmath21-varifold that satisfies @xmath236 for @xmath201 .
let @xmath469 .
if @xmath454 has a weak derivative @xmath455 , and @xmath470 , then @xmath380 has weak derivative @xmath471 as above , we observe that the claimed formula holds for smooth @xmath279 . for general @xmath338 , we let @xmath356 be a smooth approximating sequence with @xmath357 in @xmath45 .
consider @xmath472 and its weak derivative @xmath473 since @xmath338 , we may assume that each @xmath474 ; thus because @xmath475 we see that @xmath476 and @xmath477 are uniformly bounded in @xmath45 .
moreover @xmath478 and @xmath479 in @xmath45 . thus because @xmath480 , @xmath481 also @xmath482 , so we have @xmath483 so that @xmath484 as in the previous corollary this shows that @xmath485 is a weak derivative for @xmath486 .
another interesting corollary is that , on varifolds satisfy @xmath487 , the michael - simon inequality extends from ambient @xmath42 functions to functions which are weakly - differentiable with respect to @xmath156 .
let @xmath488 .
let @xmath1 be a rectifiable @xmath394-varifold in @xmath152 that satisfies @xmath487 .
there is a universal constant @xmath395 so that for any @xmath489 which has a weak derivative @xmath490 , @xmath491 let @xmath492 have @xmath357 in @xmath493 .
each @xmath356 has weak derivative @xmath494 with @xmath495 in @xmath307 and as in the proof of theorem [ thm : weak - derivative - menne ] @xmath496 in @xmath48 .
now consider the smooth test vector field @xmath497 , and obtain @xmath498 as @xmath499 . passing to a subsequence
, we have @xmath500 on the other hand by orthogonality , @xmath501 .
so @xmath502 converges weakly and in norm to @xmath451 , hence @xmath503 in @xmath48 , hence @xmath504 in @xmath307 .
thus @xmath505 in @xmath307 . using the restated michael - simon inequality ( theorem [ thm : msrestate ] )
, each pair @xmath506 has @xmath507 the claimed inequality follows .
it is standard to go from the inequality above to inequalities involving other exponents .
observe that to do so involves changing the assumed bound on @xmath60 as well .
here we assume @xmath17 is a calabi - yau manifold and @xmath1 is a integer rectifiable lagrangian @xmath21-varifold in @xmath17 that satisfies @xmath366 , @xmath367 . at each point
@xmath34 in the support of @xmath1 with approximate tangent plane @xmath305 , since @xmath305 is a lagrangian @xmath21-plane in @xmath17 , there is an associated lagrangian angle @xmath508 .
let @xmath1 be a lagrangian integer - rectifiable varifold .
we say @xmath1 has @xmath509-maslov index zero if there is a real valued lift of the @xmath38 lagrangian angle , denoted @xmath510 , such that @xmath40 has a weak derivative @xmath511 , where @xmath512 .
[ prop : embeddedzm ] if the varifold @xmath1 is given by a smoothly embedded lagrangian submanifold @xmath513 then the condition that @xmath1 has @xmath509-maslov index zero is equivalent to the classical notion of the maslov class vanishing on the smooth lagrangian submanifold @xmath58 in the case that @xmath514 .
if the maslov class vanishes , there is a smooth lift of the @xmath38 lagrangian angle .
conversely , if there is a scalar - valued lift @xmath510 of the @xmath38 lagrangian angle with weak derivative @xmath515 then @xmath516 . hence by the sobolev embedding theorem @xmath517 .
this implies that the maslov class vanishes . for the purposes of this paper we confine ourselves to the special case with @xmath518 and @xmath519 .
therefore we define : we say @xmath1 has maslov index zero if there is a real valued lift of the @xmath38 lagrangian angle , denoted @xmath47 , such that @xmath40 has a weak derivative @xmath520 .
[ prop : immersedzm ] if @xmath1 is given by a smoothly immersed lagrangian submanifold @xmath521 with vanishing maslov class , then @xmath1 has maslov index zero . by hypothesis , the maslov class @xmath93=[d\beta]=0 $ ] in @xmath95
, so there is a smooth function @xmath522 with , for all @xmath523 , @xmath524
. if @xmath58 is not an embedding , the assignment @xmath525 may not be well - defined , but because @xmath58 is in immersion , it is only so on a set of @xmath156-measure zero , so we may think of @xmath526 . then @xmath87 is a lift of the lagrangian angle .
similarly , @xmath527 is defined @xmath156-a .
e. we claim that @xmath528 is a weak derivative for @xmath87 .
let @xmath109 be a test vector field .
considering the bundle @xmath529 , observe that @xmath80 is an embedded submanifold of @xmath530 , and we can equip @xmath530 with a riemannian metric which agrees with the pullback metric along @xmath80 , so that the local submanifold geometry of @xmath80 in @xmath530 is the same as that of @xmath531 in @xmath17 .
in particular , if we let @xmath532 , @xmath533 , and @xmath534 denote the volume form , divergence operator , and mean curvature respectively of @xmath80 in @xmath530 , we have @xmath535 the following example illustrates the essential global nature of the zero maslov index property and its relationship to the notion of vanishing maslov class for an immersed lagrangian submanifold . in particular , it is possible to decompose a varifold with maslov index zero into a sum of varifolds , each summand of which does not have maslov index zero .
consider the figure - eight curve in figure [ fig : fig8trans ] as a 1-varifold @xmath1 in @xmath536 .
there is an immersion @xmath246 of @xmath38 into @xmath536 whose image is @xmath1 ; thus @xmath1 is a lagrangian varifold satisfying @xmath537 .
the lagrangian angle @xmath87 in this case is nothing more than the angle @xmath153 between the tangent line to the immersion and some fixed line , say the @xmath34-axis . because the winding number of the immersion is zero ,
its maslov class vanishes and hence the varifold @xmath1 has maslov index zero . , decomposed . ]
decompose @xmath1 into @xmath538 and @xmath539 as indicated in figure [ fig : fig8trans ] .
each @xmath540 comes from a smooth immersion @xmath541 .
computing classically we see that the change in @xmath153 along @xmath538 and @xmath539 is @xmath542 .
both @xmath538 and @xmath539 are lagrangian varifolds whose supports overlap in a set of @xmath156-measure zero , and @xmath543 as measures .
however , observe that each @xmath540 has nonvanishing generalized boundary , hence does not satisfy condition @xmath537 .
thus neither @xmath538 nor @xmath539 has maslov index zero in our sense . +
now consider the figure - eight curve in figure [ fig : fig8tan ] as a 1-varifold @xmath176 in @xmath536 .
as above , @xmath176 has maslov index zero in our sense .
if we let @xmath544 be the left lobe of @xmath176 and @xmath545 the right lobe , then again we have a decomposition @xmath546 as varifolds ; moreover @xmath544 and @xmath545 themselves arise from smooth immersions @xmath547 , @xmath548 of @xmath38 , hence satisfy @xmath537 .
, decomposed .
] however , @xmath544 and @xmath545 do not have maslov index zero in our sense .
classically , the immersions @xmath547 , @xmath548 have nonzero winding numbers which sum to zero .
if @xmath1 is a lagrangian varifold coming from a special lagrangian variety , we say @xmath1 is a _ special lagrangian varifold_. if @xmath1 is a special lagrangian varifold , then @xmath1 has maslov index zero . if @xmath1 is special lagrangian then there is a constant lift of the lagrangian angle .
we claim that this constant lift has weak derivative zero .
that is , we must show that for any @xmath549 : @xmath550 let @xmath551 be a fixed frame . using the first variation formula and the fact that @xmath1 is stationary we have : @xmath552 the result follows .
we will now use mollification to give a formula for the weak derivative of @xmath87 on a varifold with maslov index zero that satisfies @xmath537 .
we can think of @xmath87 as a composition @xmath553 , where @xmath554 is the mod-@xmath41 angle function on the lagrangian grassmannian and @xmath555 is the tangent plane map , which by menne s theorem [ thm : menne - tilt - excess ] has a @xmath156-approximate differential @xmath556 for @xmath156-almost every @xmath34 .
observe that @xmath557 and following menne , we think of @xmath558 for each @xmath290 , @xmath559 .
in fact we will first prove a more general result for any function @xmath355 obtained as @xmath560 for some smooth @xmath561 . at points @xmath34 where @xmath562 has a @xmath156-approximate differential , so does @xmath279 , and
we have following formula for it : we define the tangential gradient of @xmath279 at a point @xmath34 of @xmath8 rectifiability as follows : for any @xmath563 , @xmath564 where @xmath126 is the gradient on @xmath119 with respect to the invariant metric and @xmath565 is the inner product on @xmath566
. we will write @xmath567 when we need to omit the test vector @xmath197 .
[ thm : mollificationnabla ] @xmath568 pointwise @xmath156-almost everywhere .
in particular , at any point @xmath34 of @xmath8 rectifiability , we have @xmath569 as @xmath303 .
we first use taylor s theorem applied to @xmath570 , which is a smooth function . to apply menne s theorem [ thm : menne - tilt - excess ]
, we identify @xmath305 with orthogonal projection onto @xmath305 , @xmath571 .
we have @xmath572 recall the computation of @xmath321 .
@xmath573 consider the numerator of the first term of : @xmath574d\phi_{\varepsilon}(x - y)d\lvert v\rvert_y
\\ = & f(x)\int d\phi_{\varepsilon}(x - y)d\lvert v\rvert_y + \int \langle{\mathbb d}{\mathbb f}(t_xv),\nabla_{y - x } t(x)\rangle d\phi_{\varepsilon}(x - y)d\lvert v\rvert_y \\ & + \int \langle{\mathbb d}{\mathbb f}(t_xv),(r(y)-r(x ) ) -\nabla_{y - x } t(x)\rangle d\phi_{\varepsilon}(x - y)d\lvert v\rvert_y \\ & + \int o(\lvert r(y ) - r(x ) \rvert^2)d\phi_{\varepsilon}(x - y)d\lvert v\rvert_y\\ = & f(x)\int d\phi_{\varepsilon}(x - y)d\lvert v\rvert_y + \int \nabla f(x)\cdot ( y - x ) d\phi_{\varepsilon}(x - y)d\lvert v\rvert_y\\ & + \int \langle { \mathbb d}{\mathbb f}(t_xv),(r(y)-r(x ) -\nabla_{y - x}t\rangle d\phi_{\varepsilon}(x - y)d\lvert v\rvert_y \\ & + \int o(\lvert r(y ) - r(x ) \rvert^2)d\phi_{\varepsilon}(x - y)d\lvert v\rvert_y \end{aligned}\ ] ] so that @xmath575\frac{\int d\phi_{\varepsilon}(x - y)d\lvert v\rvert_y}{\int \phi_{\varepsilon}(x - z)d\lvert v\rvert_z + { \varepsilon}\lvert v\rvert}\\ & + \frac{\int \nabla f(x)\cdot ( y - x ) d\phi_{\varepsilon}(x - y)d\lvert v\rvert_y}{\int \phi_{\varepsilon}(x - z)d\lvert v\rvert_z + { \varepsilon}\lvert v\rvert}\\ & + \frac{\int \langle { \mathbb d}{\mathbb f}(t_xv),(r(y)-r(x ) -\nabla_{y - x}t\rangle d\phi_{\varepsilon}(x - y)d\lvert v\rvert_y } { \int \phi_{\varepsilon}(x - z)d\lvert v\rvert_z + { \varepsilon}\lvert v\rvert}\\ & + \frac{\int o(\lvert r(y ) - r(x ) \rvert^2)d\phi_{\varepsilon}(x - y)d\lvert v\rvert_y}{\int \phi_{\varepsilon}(x - z)d\lvert v\rvert_z + { \varepsilon}\lvert v\rvert } \end{aligned}\ ] ] by virtue of lemma [ lem : dphiepsilonbounded ] and theorem [ thm : strong - conv - lp ] , the first term tends to zero .
we will now handle the remaining terms of .
observe that by lemma [ lem : mollifiertomultiplicity ] , each denominator tends to @xmath309 .
consider the numerator of the second term of .
@xmath576 as @xmath303 , this converges to @xmath577 and observing that on a plane , @xmath578 , we may integrate by parts to obtain @xmath579 our choice of normalization is that the integral of @xmath580 on @xmath284 is 1 , so @xmath581 the corresponding denominator @xmath582 , so the corresponding term in the limit of is @xmath583 .
now consider the third term of . recall that by our choice of @xmath277 , all integrals are over @xmath335 .
@xmath584 where the constant @xmath246 absorbs @xmath585 and @xmath586 and the penultimate step is an application of hlder s inequality .
menne s theorem [ thm : menne - tilt - excess ] says precisely the last line of this inequality tends to @xmath587 as @xmath303 . for the fourth and final term of
, we estimate : @xmath588 which tends to zero by brakke - menne s quadratic tilt - excess decay estimate theorem [ thm : tilt - excess - b - m ] .
thus we have @xmath569 as @xmath303 , as claimed .
[ prop : nablabeta ] at a point @xmath34 of @xmath8 rectifiability , @xmath589 .
because @xmath34 is a point of @xmath8 rectifiability , we have @xmath34 in some @xmath8 submanifold @xmath27 so that @xmath590 and @xmath591 .
moreover , it follows from the proof of menne s theorem [ thm : menne - tilt - excess ] ( see 3.8 of @xcite ) that the @xmath156-approximate differential of @xmath562 computed at @xmath34 is the same as that coming from @xmath27 .
so we compute @xmath592 with respect to @xmath27 , observing that the approximate differential of the tangent plane to a @xmath8 submanifold is its second fundamental form @xmath593 .
@xmath594 where we have used the computation of @xmath595 as in
[ sec : variationofbeta ] . in particular , @xmath596 is independent of which such @xmath8 submanifold @xmath27 we chose . observe that this formula for @xmath597 is the same as in the @xmath8 case , see @xcite .
with respect to @xmath87 , for varifolds satisfying @xmath537 , we can thus interpret theorem [ thm : mollificationnabla ] as saying that _ almost everywhere , in the limit @xmath303 , mollification and ( tangential ) differentiation commute_. for curvature varifolds in the sense of hutchinson @xcite and mantegazza @xcite , similar results to proposition [ prop : nablabeta ] hold for other functions @xmath570 , whose differentials @xmath585 we think of as algebraic operations on the second fundamental form other than trace .
[ thm : harvey - lawson - varifold ] suppose @xmath1 is a lagrangian @xmath21-varifold with maslov index zero satisfying @xmath537 .
suppose that the lagrangian angle @xmath47 has weak derivative @xmath46 .
suppose @xmath181 is an open subset of @xmath0 .
then for @xmath156-almost all @xmath598 : @xmath599 let @xmath600 be a sequence with @xmath601 .
we have for any ambient test vector field @xmath109 @xmath602 now as in the proof of theorem [ thm : weak - derivative - menne ] , splitting @xmath437 , we obtain the equation @xmath603 i. e. , @xmath604 in @xmath48 .
now let @xmath605 and consider the test vector field @xmath606 .
we have @xmath607 as @xmath608 , the left - hand side converges to @xmath609 .
we may take a diagonal sequence on the right - hand side to conclude that as @xmath610 , @xmath611 thus @xmath612 converges weakly and in norm to @xmath613 , hence @xmath614 in @xmath48 . on the other hand ,
the almost - everywhere pointwise limit of @xmath612 , by theorem [ thm : mollificationnabla ] , is @xmath615 .
therefore we have @xmath616 almost everywhere . by proposition
[ prop : nablabeta ] , @xmath617 . by the statement of theorem
[ thm : weak - derivative - menne ] , @xmath618 almost everywhere .
observe that the formula is formally the same as if @xmath87 were an ambient lipschitz function .
we have chosen @xmath244 in theorem [ thm : harvey - lawson - varifold ] for convenience ; for other @xmath619 we would obtain the same formula .
observe that from ( [ equ : compute - weak - derivative ] ) and the assumption @xmath526 , it follows that @xmath620 if and only if @xmath621 .
as in the classical case , our notion of weak derivative has good compactness properties .
we note these in the following lemmas , which in turn lead to a weak compactness result for maslov zero index varifolds .
let @xmath1 be an integer - rectifiable varifold satisfying @xmath236 for @xmath367 and let @xmath622 be a sequence of functions with weak derivatives @xmath623 .
suppose there is a constant @xmath246 such that @xmath624 then there is a subsequence ( which we still denote @xmath442 ) and a function @xmath454 with weak derivative @xmath625 such that @xmath626 , @xmath627 and for any @xmath628 : @xmath629 the existence of @xmath279 and @xmath371 and ( [ equ : weak - limits - f - f ] ) follow from the precompactness of @xmath45 and @xmath328 for @xmath201 in the weak topology . to show that @xmath371 is the weak derivative of @xmath279 we must verify that for any @xmath628 : @xmath630 since @xmath631 in @xmath45 we have @xmath632 , for @xmath1 a.e .
@xmath34 . using @xmath633 by dominated convergence
we have that as @xmath634 : @xmath635 then ( [ equ : weak - deriv .- in - the - limit ] ) follows from ( [ equ : weak - limits - f - f ] ) and for any @xmath628 and every @xmath452 : @xmath636 the result follows .
our compactness result will concern a sequence @xmath637 of maslov - index - zero varifolds , each with an associated lagrangian angle @xmath638 and weak derivative @xmath639 .
we will therefore need to examine the convergence of the triple @xmath640 .
we will use the following result about radon measures : [ lem : radon - measures - p ] let @xmath641 , and let @xmath642 , @xmath643 be sequences of radon measures on @xmath268 with @xmath644 and @xmath645 in the weak topology .
suppose that that there exists @xmath246 , independent of @xmath452 , so that for each @xmath646-measurable @xmath647 , @xmath158 is @xmath648-measurable and @xmath649 then @xmath650 is absolutely continuous with respect to @xmath190 and for any @xmath190-measurable set @xmath158 , @xmath158 is @xmath650 measurable with @xmath651 first we recall ( see e. g. @xcite ) that the convergence @xmath652 , @xmath645 implies that for any open set @xmath181 and any compact set @xmath653 , we have @xmath654 therefore , for any bounded set @xmath158 , @xmath655 where @xmath656 and @xmath657 are the interior and closure of @xmath158 , respectively .
let @xmath653 be any compact set , and let @xmath658 be the open @xmath659-neighborhood of @xmath653 .
then we have @xmath660 hence @xmath661 let @xmath46 be any borel set ; we have that @xmath46 is both @xmath190- and @xmath650-measurable and @xmath662 now if @xmath158 is a set with @xmath663 , then for any @xmath664 there is an open set @xmath665 with @xmath666 and @xmath667 . then @xmath668 , @xmath181 is borel , and for all @xmath664 , @xmath669 .
so @xmath181 is a nullset for @xmath650 , hence @xmath670 .
thus we have shown that @xmath650 is absolutely continuous with respect to @xmath190 .
+ by the radon - nikodym theorem , we have that any @xmath190-measurable set is @xmath650-measurable , and the inequality extends for all @xmath190-measurable @xmath158 .
[ lem : lpweak - conv ] let @xmath671 be a sequence of integer - rectifiable varifolds which converge weakly to the integer - rectifiable varifold @xmath1 .
suppose @xmath671 and @xmath1 satisfy @xmath236 for @xmath367 .
let @xmath672 be a sequence of functions with weak derivatives @xmath673 .
suppose there is a constant @xmath246 such that @xmath674 then there is a subsequence ( which we still denote @xmath442 ) and a function @xmath454 with weak derivative @xmath429 such that @xmath626 , @xmath675 and for any @xmath628 : @xmath676 for each @xmath452 write : @xmath677 @xmath678 then @xmath679 for a @xmath671-measurable set @xmath680 define : @xmath681 then @xmath682 and @xmath683 are radon measures . @xmath682 and @xmath683 are both absolutely continuous with respect to @xmath684 .
moreover , @xmath685 choosing subsequences we can suppose that there are radon measures @xmath686 and @xmath687 such that as @xmath634 we have : @xmath688 where the convergence is weak convergence of radon measures . by the @xmath689 case of lemma [ lem : radon - measures - p ] we have for any @xmath156-measurable set that @xmath690 let @xmath691 , @xmath692 .
we have for any @xmath693 , as @xmath634 : @xmath694 setting @xmath695 , it follows from the @xmath689 case of lemma [ lem : radon - measures - p ] that @xmath696 , hence @xmath697 and we have as @xmath634 , @xmath698 next we consider the sequence @xmath699 .
let @xmath700 be an orthonormal frame and set @xmath701 for @xmath702 .
define for each @xmath452 and @xmath241 : @xmath703 and similarly define @xmath704 . for a @xmath671-measurable set @xmath680 define : @xmath705 then @xmath706 and
@xmath707 are radon measures . both are both absolutely continuous with respect to @xmath684 . moreover , using the hlder inequality , @xmath708 applying lemma [ lem : radon - measures - p ] with @xmath202 chosen so @xmath709 , we obtain limit measures @xmath710 which are absolutely continuous with respect to @xmath156 .
let @xmath711 be the radon - nikodym derivative of @xmath710 with respect to @xmath156 .
observe that for any @xmath712 , @xmath713 setting @xmath714 , we obtain @xmath371 so that for any smooth compactly supported test vector field @xmath109 , @xmath715 in particular , consider @xmath109 with @xmath716 .
then @xmath717 hence @xmath718 .
+ we wish to show that @xmath371 is the weak derivative of @xmath279 .
we have verified that : @xmath719 next we wish to show that : @xmath720 observe that ( perhaps passing to a subsequence ) may assume that the radon measures @xmath721 on @xmath167 converge to @xmath722 with @xmath723 .
then we have : @xmath724 similarly : @xmath725 clearly , ( [ equ : nabla - convergence ] ) follows .
but since @xmath494 is the weak derivative of @xmath356 we have : @xmath726 taking the limit as @xmath634 shows that @xmath371 is the weak derivative of @xmath279 .
[ lem : l - infy - conv ] let @xmath671 be a sequence of integer - rectifiable varifolds of uniformly bounded mass which converge weakly to the integer - rectifiable varifold @xmath1 .
let @xmath727 be a sequence of functions .
suppose there is a constant @xmath246 independent of @xmath452 such that @xmath728 then there is a subsequence ( which we still denote @xmath442 ) and a function @xmath729 such that @xmath730 and for any @xmath628 : @xmath731 to begin note that by proposition [ lem : lpweak - conv ] there is a subsequence of @xmath442 that we continue to denote @xmath442 and a function @xmath729 such that for any @xmath628 : @xmath732 we want to show that for each integer @xmath733 and any @xmath628 we have as @xmath634 : @xmath734 to prove ( [ equ : induction - l ] ) we use induction and suppose we have for any @xmath628 as @xmath634 : @xmath735 we use mollification as follows : mollify @xmath356 with respect to the measure @xmath671 .
that is , @xmath736 note that for all @xmath452 , @xmath737 and @xmath738 in addition for every @xmath50 ; @xmath739 we have by the inductive assumption and dominated convergence that for any @xmath54 : @xmath740 hence , letting @xmath55 @xmath741 this proves the inductive step , verifying ( [ equ : induction - l ] ) for each integer @xmath77 .
the result follows using the taylor expansion for sine and cosine .
[ thm : convergence - beta ] let @xmath671 be a sequence of varifolds with maslov index zero satisfying @xmath537 , each with finite mass , that converge , in the sense of varifold convergence , to the lagrangian integer - rectifiable varifold @xmath1 also satisfying @xmath537 .
suppose that for each @xmath452 the varifold @xmath671 admits a lift @xmath742 of the @xmath38 lagrangian angle with weak derivative @xmath639 so that there are constants @xmath743 independent of @xmath452 with : @xmath744 and @xmath745 then @xmath1 admits a lift @xmath40 of the @xmath38 lagrangian angle with weak derivative @xmath46 satisfying : @xmath746 apply proposition [ lem : lpweak - conv ] to the sequences @xmath747 and @xmath748 to conclude that there exist @xmath749 with weak derivative @xmath520 with @xmath750 and @xmath751 such that for subsequences of @xmath747 and @xmath748 ( which we continue to denote @xmath747 and @xmath748 ) we have for any @xmath628 as @xmath634 @xmath752 apply lemma [ lem : l - infy - conv ] to the sequence @xmath747 to conclude that there is a subsequence ( which we continue to denote @xmath747 ) such that for any @xmath628 as @xmath634 : @xmath753 we need to verify that @xmath754 is a lift of the lagrangian angle .
to show this we lift the lagrangian integer - rectifiable varifolds @xmath671 and @xmath1 to lagrangian varifolds @xmath755 and @xmath168 , i.e. radon measures on @xmath75 .
on @xmath75 the lagrangian angle @xmath77 is a global smooth function . by assumption for each @xmath452
we have , @xmath756 let @xmath757 .
then , as @xmath634 , @xmath758 on the other hand , @xmath759 if we take @xmath369 to be independent of @xmath97 , i.e. @xmath760 , and apply ( [ equ : version2-beta ] ) , we have for any @xmath761 as @xmath634 : @xmath762 using the uniqueness of the weak limit we conclude that @xmath763 a.e .
the result follows . combining allard s compactness theorem [ thm : compact - hp - varifolds ] and theorem [ thm : convergence - beta ]
, we obtain the following compactness result for varifolds with maslov index zero : let @xmath671 be a sequence of lagrangian varifolds satisfying @xmath537 with maslov index zero .
suppose there are @xmath764 so that each lift @xmath638 of the lagrangian angle of @xmath671 and its weak derivative @xmath639 satisfy : @xmath765 then there is a subsequence of the @xmath671 which converges to a lagrangian varifold @xmath1 satisfying @xmath537 with maslov index zero , for which the lift @xmath87 of the lagrangian angle of @xmath1 and its weak derivative @xmath46 satisfy @xmath766 in virtue of theorem [ thm : harvey - lawson - varifold ] , we observe that in the presence of a uniform bound on @xmath767 , a uniform bound on @xmath768 implies a bound on @xmath769 and vice versa .
let @xmath770 be a symplectic diffeomorphism .
if @xmath1 is a varifold satisfying @xmath537 with maslov index zero , then so is @xmath771 .
[ sec : epsilon - flows ] now let @xmath5 be a varifold with maslov index zero satisfying @xmath537 . in this section , we will show that for each @xmath50 , the flow equation @xmath772 has a solution starting from initial data @xmath5 , where the mollification is carried out with respect to @xmath773
. each @xmath773 will be a varifold with maslov index zero satisfying @xmath537 . throughout
, @xmath774 denotes the @xmath775 norm of a continuous function , and the supremum of the operator norm of a tensor .
+ we will produce the one - parameter family @xmath3 of varifolds with maslov index zero solving [ eqn : epsilonflowequation ] by a sort of euler s method .
we wish to define , for any mollification parameter @xmath273 , and any @xmath664 , a sequence of varifolds with maslov index zero .
each varifold in the sequence will be given by flowing by the hamiltonian diffeomorphism generated by @xmath776 for time @xmath777 ; then we compute @xmath776 for the new varifold and repeat . again for
exposition we state all results and proofs for varifolds in @xmath268 .
+ for now we fix @xmath50 . given @xmath778 , define the diffeomorphisms @xmath779 and lagrangian integer - rectifiable varifolds @xmath780 by :
@xmath781 we point out for clarity that the mollification in @xmath782 is with respect to the varifold @xmath780 .
each @xmath783 is the hamiltonian - diffeomorphic image of @xmath780 , hence of @xmath784 .
therefore the @xmath780 all have maslov index zero . our goal is to use the @xmath780 to approximate a one - parameter family of varifolds with maslov index zero moving according to ( [ eqn : epsilonflowequation ] ) ; we will use the fact that the dyadic rationals are dense in @xmath44 . to that end
, we adopt the notation : given a dyadic rational @xmath7 , define for each @xmath664 @xmath785 where @xmath786 .
we wish to roughly estimate how the lagrangian angle @xmath87 changes along each sequence @xmath780 .
[ lem : betaepsilonexponential ] @xmath787 , where @xmath246 and @xmath202 are universal constants coming from lemma [ lem : derivativesofbetaepsilon ] . for any @xmath788
, we have using the first variation of @xmath87 ( [ eqn.deltabeta3 ] ) that @xmath789 thus @xmath790 . applying lemma [ lem : derivativesofbetaepsilon ] ,
we obtain @xmath791 iterating this estimate gives the result .
fixing @xmath7 and noting that @xmath792 for @xmath793 , we obtain : [ cor : betaexponentialtime ] @xmath794 .
in particular , @xmath795 is bounded uniformly in @xmath323 and uniformly for @xmath7 in a compact interval . moreover
, @xmath796 is bounded uniformly in @xmath323 and uniformly for @xmath7 in a compact interval .
differentiating the taylor expansion @xmath797 , we have @xmath798 and we can estimate @xmath799 and apply lemma [ lem : derivativesofbetaepsilon ] and corollary [ cor : betaexponentialtime ] to get @xmath800 applying this estimate to each factor in @xmath801 and using lemma [ lem : betaepsilonexponential ] , we have @xmath802 where the first term converges to @xmath803 and the second term goes to zero as @xmath610 .
similar estimates hold for @xmath804 , uniform in @xmath323 and @xmath7 in a compact interval , for any @xmath659 .
observe that @xmath805 is uniformly bounded , independent of @xmath323 .
thus @xmath806 is lipschitz with lipschitz constant @xmath807 , which is uniform in @xmath323 and uniform in @xmath7 in compact intervals .
therefore we obtain , for each dyadic rational @xmath786 , a sequence of @xmath780 of varifolds with maslov index zero arising as the pushforward of @xmath1 under uniformly - bounded diffeomorphisms @xmath808 ; we can therefore extract a subsequential limit @xmath773 .
since the dyadic rationals are countable , we can apply a diagonal argument so that the same indices @xmath323 give convergence to @xmath773 for each @xmath7 .
moreover , if we define at each dyadic rational @xmath786 , @xmath809 , we have uniform similar uniform bounds allowing us to extract @xmath810 , which is the lipschitz limit in @xmath7 and the smooth limit spatially .
+ we would like to say that these @xmath773 can be extended into a one - parameter varifolds which satisfies the flow equation ; in fact we will see that the @xmath773 are generated by a one - parameter family of hamiltonian diffeomorphisms .
given an interval @xmath811 $ ] , let @xmath812 denote a choice of @xmath813 subsequential limit of the @xmath806 for each dyadic rational @xmath814 $ ] .
we may extend @xmath815 to all @xmath814 $ ] so that @xmath816 is smooth in @xmath7 .
moreover , if we define @xmath817 , then the @xmath815 are hamiltonian , generated by @xmath818 : @xmath819 first observe that each @xmath820 can be expressed in terms of the derivatives of @xmath821 .
by lemma [ lem : derivativesofbetaepsilon ] , these are bounded depending on @xmath659 , @xmath273 , and @xmath822 , which by lemma [ lem : betaepsilonexponential ] can be bounded independently of @xmath323 .
consider @xmath823 , dyadic rationals .
we compute @xmath824 where @xmath825 and the sum is over all @xmath77 with @xmath826 .
now everything in sight that depends on @xmath323 converges uniformly spatially and in @xmath7 , so we have @xmath827 this extends to all @xmath828 , @xmath829 by continuity , which in turn implies the claimed evolution of @xmath830 .
that @xmath831 follows from the convergence @xmath832 .
we define the @xmath273-approximate flow by @xmath817 .
consider the @xmath273-flow of varifolds with maslov index zero @xmath773 constructed in section [ sec : epsilon - flows ] and let @xmath833 denote the single - valued lift of the @xmath38 lagrangian angle .
the main theorem of this section is : [ thm : boundonbeta ] there is a constant @xmath158 , depending on @xmath834 but independent of @xmath273 and @xmath835 $ ] such that for @xmath836 , @xmath837 a key observation in the proof of this result is along each @xmath273-flow , the lagrangian angle @xmath838 satisfies @xmath839 apply the variation formula for @xmath87 and the fact that @xmath773 evolves by @xmath840 .
while is similar to the parabolic equation @xmath841 which has a maximum principle because we have mollified , is not a partial differentiation equation .
rather it is a pseudo - differential evolution equation . to obtain estimates
we regularize this equation in two distinct ways .
first we mollify the varifolds @xmath773 to produce an initial value problem on an open set in @xmath0 .
next we regularize the operator @xmath842 to make it into a uniformly elliptic operator .
the standard elliptic estimates can then be exploited to prove the theorem .
this argument is similar , in spirit , to the method of vanishing viscosity .
our treatment of the pseudo - differential evolution equation is motivated by treatment of a similar equation in @xcite .
[ lem : unique - linfty - solution ] fix @xmath50 . for any @xmath843
the initial value problem @xmath844 has a unique solution @xmath845 in @xmath846 for @xmath847 .
we regard the equation as an ordinary differential equation on @xmath848 and use the method of picard . for any @xmath849 , use @xmath850 to denote @xmath851;l^\infty(v^{\varepsilon}(t)))$ ] equipped with the norm @xmath852}\exp(-2c(\eta)t)\lvert \psi(t)\rvert_{l^\infty(v^{\varepsilon}(t))}\ ] ] where @xmath853 .
note that as @xmath854 , @xmath855 and therefore the norm goes to zero .
now ( cf . lemma [ lem : betaepsilonexponential ] ) define @xmath856 by @xmath857 where @xmath858 is the one - parameter family of diffeomorphisms generated by @xmath859 . recall that @xmath858 takes the varifold @xmath860 to @xmath861 .
we compute @xmath862 } \exp(-2c(\eta)t ) c(\eta)\int_0^t \lvert \psi_2(s)-\psi_1(s)\rvert_{l^\infty(v^{\varepsilon}(s))}\ ds\\ & \leq \sup_{t\in [ 0,t ] } \exp(-2c(\eta)t ) c(\eta)\int_0^t \lvert \psi_2-\psi_1\rvert \exp(2c(\eta)s)\ ds\\ & = \frac{1}{2}\left\lvert \psi_2-\psi_1\right\rvert \sup_{t\in [ 0,t ] } \left(1-\exp(-2c(\eta)t)\right)\leq \frac{1}{2}\left\lvert \psi_2-\psi_1\right\rvert \end{aligned}\ ] ] where the first inequality is obtained by differentiating on the mollifier @xmath863 and using the definition @xmath853 .
the second inequality follows from the definition ( [ defn - of norm ] ) .
thus @xmath856 is a contraction . in the standard way
we conclude that the sequence @xmath864 defined by @xmath865 converges to a unique solution of , defined on @xmath811 $ ] .
moreover none of the choices involved depended on @xmath562 , so we can extend the solution to all @xmath866 . for any lagrangian varifold @xmath1 define the mollified lagrangian varifold @xmath867 by setting , for any @xmath868 , @xmath869 for each @xmath290 , define the mollified divergence operator @xmath870 by @xmath871 this operator is the integral over the entire fiber of the grassmann bundle of the divergence of @xmath109 computed on each lagrangian plane @xmath97 .
+ we can estimate the derivatives of @xmath872 as follows : @xmath873 [ lem : unique - smooth - solution ] the initial value problem @xmath874 for @xmath875 has a unique solution in @xmath876 .
we will show that the equation has a unique solution in @xmath877 for each @xmath878 , @xmath879 ; then uniqueness will give existence and uniqueness in @xmath813 .
for fixed @xmath323 , we regard the equation as an ordinary differential equation on @xmath880 and use the method of picard . for fixed @xmath849 , use @xmath881 to denote @xmath882;c^{k,\alpha}({\mathbb r}^{2n}))$ ] equipped with the norm @xmath883}\exp(-2c_k(\eta,\sigma)t)\lvert
\psi(t)\rvert_{c^{k,\alpha}({\mathbb r}^{2n})}\ ] ] where @xmath884 .
now define @xmath885 by @xmath886 then we compute @xmath887 } \exp(-2c_k(\eta , \sigma)t ) c_k(\eta , \sigma)\int_0^t \lvert \psi_2(s)-\psi_1(s)\rvert_{c^k({\mathbb r}^{2n})}\ ds\\ & \leq \sup_{t\in [ 0,t ] } \exp(-2c_k(\eta)t ) c_k(\eta)\int_0^t \lvert \psi_2-\psi_1\rvert \exp(2c_k(\eta,\sigma)s)\ ds\\ & = \frac{1}{2}\left\lvert \psi_2-\psi_1\right\rvert \sup_{t\in [ 0,t ] } \left(1-\exp(-2c_k(\eta,\sigma)t)\right)<\frac{1}{2}\left\lvert \psi_2-\psi_1\right\rvert \end{aligned}\ ] ] so that @xmath888 is a contraction mapping , hence has a unique fixed point in @xmath881 , which is the solution of ( [ smoothivp ] ) .
consider a subset @xmath889 . then : @xmath890 define the set @xmath891 where @xmath112 is an orthogonal complex structure on @xmath0 by : @xmath892 let @xmath1 be a varifold on @xmath893
. then @xmath158 is @xmath1-measurable if and only if @xmath891 is @xmath1-measurable .
we define the varifold @xmath894 by : @xmath895 we introduce the notation , for each @xmath896 , the differential operator @xmath897 for each @xmath898 we define the differential operator @xmath899 the symbol of @xmath900 is @xmath901 where @xmath902 denotes the projection of @xmath903 onto the lagrangian @xmath21-plane @xmath97 .
the symbol of @xmath904 is @xmath905 it follows that for @xmath906 and @xmath907 open that on @xmath908 the differential operator @xmath904 is strictly elliptic .
the garding inequality gives : @xmath909 where the constant @xmath910 .
the @xmath911 inner product is defined using the lebesgue measure .
hence , @xmath912 therefore , for all @xmath906 , @xmath913 hence , letting @xmath914 , @xmath915 apply the inequality ( [ equ : garding - ineq-2 ] ) to the function @xmath916 where @xmath344 is a positive even integer .
then : @xmath917 for any @xmath918 , any @xmath344 , we have @xmath919 we compute : @xmath920 applying the lemma to ( [ equ : garding - ineq-3 ] ) for @xmath921 we derive : @xmath922 hence @xmath923 recall the operator @xmath924 for @xmath925 defined by : @xmath926 then @xmath924 is a pseudo - differential operator .
we can , of course , consider the differential operator @xmath900 as a pseudo - differential operator and therefore the composition @xmath927 is a pseudo - differential operator . to make this precise we follow hormander [ ho ] and define the space of symbols @xmath928 where @xmath908 is an open set in @xmath0 and @xmath929 :
@xmath930 a pseudo - differential operator with symbol in @xmath931 is lies in the space @xmath932 .
then @xmath900 lies in @xmath933 and @xmath924 lies in @xmath934 . by proposition [ prop : bound - on - moll - op ]
the operators @xmath924 as operators on @xmath935 are uniformly bounded for @xmath936 and therefore from ( [ equ : garding - ineq-4 ] ) for @xmath936 there is a constant independent of @xmath333 such that : @xmath937 we rewrite the initial value problem ( [ smoothivp ] ) using this notation . @xmath938
the unique smooth solution @xmath939 of ( [ smoothivp-2 ] ) satisfies for every positive even integer @xmath344 : @xmath940 where the constant is independent of @xmath344 .
hence , @xmath941 , x \in \r^{2n } } |u(t , x)| \leq c \sup _ { x \in \r^{2n } } |\phi_\sigma \star u_0(x)|.\ ] ] for even @xmath921 , we compute .
the inequality uses ( [ equ : garding - ineq-5 ] ) @xmath942 from this inequality we have : @xmath943 applying gronwall s inequality [ e ] to ( [ equ : lp - space - inequality ] ) yields for each @xmath344 : @xmath944 the result follows . we have shown : for each @xmath945 the unique smooth solutions of ( [ smoothivp-2 ] ) satisfy the estimate : @xmath946 where the constant is independent of @xmath333 .
[ thm : sup - bound - for - u ] there is a constant @xmath947 such that for each @xmath54 and @xmath843 the unique solution @xmath399 of ( [ ivp ] ) satisfies the estimate : @xmath948 fix @xmath50 and @xmath843 .
consider a sequence @xmath949 .
let @xmath950 be solutions of ( [ smoothivp-2 ] ) ) for @xmath951 .
choose a subsequence ( that we will continue to denote @xmath950 ) that converges in @xmath952 to a function @xmath399 .
then @xmath399 satisfies the bound ( [ equ : sup - bound ] ) .
we claim that @xmath399 is a solution of ( [ ivp ] ) .
therefore , the unique solution of ( [ ivp ] ) satisfies the bound ( [ equ : sup - bound ] ) . for each @xmath241 the smooth function @xmath953 satisfies the equation : @xmath954 hence for any test function @xmath466 we have : @xmath955 as @xmath949 by the dominated convergence theorem and brakke s orthogonality of @xmath60 we have : @xmath956 therefore letting @xmath949 we have : @xmath957 since this is true for any test function @xmath466 it follows that @xmath399 is a solution of ( [ ivp ] ) .
the result follows . there is a constant @xmath958 such that for any @xmath959 : @xmath960 } \sup _ { v^{\varepsilon}(t ) } |{\beta}^{\varepsilon}| < a\ ] ] on the @xmath273-flow apply theorem [ thm : sup - bound - for - u ] with initial condition @xmath961 and @xmath962 .
in section [ sec : epsilon - flows ] we have , for each fixed @xmath54 , produced a flow @xmath773 of maslov index zero , integral rectifiable lagrangian varifold satisfying @xmath537 .
we would like to extract a limit flow @xmath3 by taking a limit as @xmath55 for each @xmath7 of the varifolds @xmath773 .
moreover we wish extract a limit that inherits the properities of the @xmath773
. however the estimates derived in section [ sec : epsilon - flows ] blow - up as @xmath55 . in section [ sec : sup - bounds ] we showed that the lagrangian angles satisfy sup bounds . in this section we exploit the weak derivatives of the @xmath963 to derive decay estimates for both the volumes @xmath964 and the lagrangian angles @xmath963 along the @xmath273-flows .
let @xmath1 be a integer lagrangian varifold with maslov index zero .
let @xmath47 denote the lift of the lagrangian angle with weak derivative @xmath520 .
the mollified lagrangian angle is : @xmath965 we have noted above that , for fixed @xmath273 , @xmath966 is globally well - defined and smooth . [
thm : errorestimate ] suppose @xmath1 is an integral lagrangian @xmath21-varifold with maslov index zero satisfying @xmath537 , @xmath967 .
suppose that the lagrangian angle @xmath47 has weak derivative @xmath520 .
let @xmath60 denote the mean curvature on @xmath1 .
then if @xmath51 is the mollified lagrangian angle : @xmath968 where @xmath969 and @xmath970 where @xmath971 is a constant depending on @xmath972 .
we begin by computing : @xmath973 first observe : @xmath974 fix @xmath975 and let @xmath408 be an arbitrary fixed vector in @xmath0 . using the first variation formula we have : @xmath976 thus , @xmath977 we define : @xmath978 using ( [ equ : compute - b ] ) , ( [ equ : compute - h ] ) and ( [ equ : define - e ] ) in ( [ equ : rough - formula - nabla - beta ] ) yields ( [ equ : final - formula ] ) .
it remains to estimate the error function @xmath979 .
we consider the @xmath21-plane @xmath980 as the orthogonal projection @xmath981 and write : @xmath982 note that on @xmath983 : @xmath984 therefore there is a constant @xmath985 such that : @xmath986 using the hlder inequality , @xmath987 thus , @xmath988 where the constant @xmath246 depends on @xmath989 . by a similar argument : @xmath990 for a constant @xmath246 .
since @xmath991 we conclude from ( [ equ : define - e ] ) : latexmath:[\[\label{equ : estimate - e } where the constant @xmath246 depends on @xmath989 .
let @xmath993 be an @xmath273-flow . recall that for each @xmath4
, @xmath773 is an integral lagrangian @xmath21-varifold with maslov index zero satisfying @xmath537 .
we will denote mollification using the notation : @xmath994 note that mollification is independent of scale in the measure so that @xmath995 for any scalar @xmath996 .
set @xmath997 the total mass of the measure @xmath773 .
observe that since each @xmath773 is generated by a one - parameter family of diffeomorphisms , we have that @xmath998 for each @xmath999 .
normalize the measure @xmath1000 to the probability measure @xmath1001 .
set : @xmath1002 we will need the following lemmas to estimate the error term ( [ equ : define - e ] ) . [
lem : sequence - allard ] given a sequence @xmath1003 with @xmath1004 suppose @xmath1005 exists and @xmath1006 , possibly @xmath1007 , then the rescaled varifolds @xmath1008 have : 1 .
uniformily bounded mass , + 2 .
satisfy @xmath537 and the bound : @xmath1009 therefore there is a subsequence , that we continue to denote @xmath1003 , such that , @xmath1010 as varifolds , @xmath1011 is a lagrangian varifold satisfying @xmath537 .
the result follows from theorem [ thm : compact - hp - varifolds ] .
observe that the error term @xmath979 is independent of the rescalings of the varifold @xmath1012 .
[ lem : dom - conv - to - e ] given a sequence @xmath1003 with @xmath1004 such that @xmath1005 exists and @xmath1006 ( possibly @xmath1007 ) then there is a subsequence , which we continue to denote @xmath1003 , such that : @xmath1013 introduce the function : @xmath1014 the function @xmath1015 is also independent of the rescalings of the varifold @xmath1012 .
observe that : @xmath1016 the estimate ( [ equ : error - estimate ] ) becomes : @xmath1017 for @xmath1018 and @xmath967 we have : @xmath1019 thus , @xmath1020 for @xmath1021 and @xmath1022 , set : @xmath1023 fix @xmath273 .
then @xmath1024 is continuous for @xmath1025 and @xmath1026 $ ] and for each @xmath1022 : @xmath1027 thus there exists a constant @xmath1028 such that : @xmath1029 } f(x , r , { \varepsilon } ) < { \lambda}\ ] ] it follows that @xmath1030 } | e(x , r , { \varepsilon } ) |^2 \leq c_1 { \lambda}.\ ] ] on the other hand by theorem [ thm : tilt - excess - b - m ] pointwise for a.e .
point @xmath34 we have : @xmath1031 therefore by the dominated convergence theorem for each @xmath54 : @xmath1032 in particular , given a sequence @xmath1033 with @xmath1034 we have for each @xmath273 : @xmath1035 given a sequence @xmath1003 with @xmath1004 such that lemma [ lem : sequence - allard ] applies to give @xmath1036 we can use the previous reasoning to conclude : @xmath1037 and @xmath1038 where @xmath1039 represents the error computed with respect to the varifold @xmath1040 , that is , @xmath1041 therefore , @xmath1042 taking a diagonal sequence and setting @xmath1043 gives the result .
[ lem : sequence - allard2 ] given a sequence @xmath1003 with @xmath1004 suppose @xmath1044 then the rescaled varifolds @xmath1045 have : 1 .
uniformily bounded mass , + 2 .
satisfy @xmath537 and the bound : @xmath1046 therefore there is a subsequence , that we continue to denote @xmath1003 , such that , @xmath1047 as varifolds , @xmath1048 is a lagrangian varifold with vanishing mean curvature .
same as the proof of lemma [ lem : sequence - allard ] .
[ lem : dom - conv - to - e2 ] given a sequence @xmath1003 with @xmath1004 such that @xmath1049 then there is a subsequence , which we continue to denote @xmath1003 , such that : @xmath1050 same as the proof of lemma [ lem : dom - conv - to - e ] . using the results and estimates of section [ sec : epsilon - flows ] the geometric quantities along the @xmath273-flow are lipschitz in time on @xmath811 $ ] and continuous in @xmath273 for @xmath54 .
for a fixed @xmath1051 $ ] we consider three cases : 1 .
@xmath1052 , possibly @xmath1053 .
@xmath1054 and @xmath1055 + 3 .
@xmath1054 and @xmath1056 + the main theorem of this section is : [ thm : volumedecrease ] let @xmath1057 be an @xmath273-flow .
for each @xmath1058 , @xmath773 is an integral lagrangian @xmath21-varifold with maslov index zero satisfying @xmath537 .
suppose that the lagrangian angle @xmath1059 has weak derivative @xmath1060 .
fix @xmath1051 $ ] . in cases ( 1 ) and ( 2 )
there is a sequence @xmath1003 with @xmath1004 such that : @xmath1061 in case ( 3 ) there is a sequence @xmath1003 with @xmath1004 such that : @xmath1062 in all cases there is a sequence , that we continue to denote @xmath1003 , such that for each @xmath1051 $ ] and any @xmath1063 there is an integer @xmath10 such that for @xmath1064 : @xmath1065 moreover @xmath1066 as @xmath1067 .
the following lemmas are essential to the proof of theorem [ thm : volumedecrease ] .
[ thm : mollified - mean - curv1 ] given a sequence @xmath1003 with @xmath1004 there is a subsequence , that we will continue to denote @xmath1003 , such that : @xmath1068 where @xmath1 is a radon measure and @xmath1069 is vector - valued . *
( i ) * the set of radon measures @xmath1070 is bounded and hence there is a sequence @xmath1071 with @xmath1072 such that @xmath1073 converges weakly to a radon measure @xmath1 . *
( ii ) * since @xmath1074 , it follows that for @xmath1075 there is a constant @xmath1076 with @xmath1077 as @xmath854 such that : @xmath1078 note that if @xmath158 is a set with @xmath1079 then @xmath1080 for @xmath1081 .
that is , the vector - valued function @xmath1082 is supported on the support of the measure @xmath1083 .
hence for any @xmath925 : @xmath1084 therefore the vector - valued functions @xmath1085 and satisfy @xmath1086 .
hence there is a sequence @xmath1071 with @xmath1004 such that @xmath1087 converges weakly in @xmath48 to a vector - valued function @xmath1088 . in particular , the vector - valued functions @xmath1089 are bounded in @xmath48 .
therefore , there is a sequence @xmath1071 with @xmath1004 such that @xmath1090 converges weakly in @xmath48 to a vector - valued function @xmath1069 . *
( iii ) * combining ( i ) and ( ii ) there are sequences @xmath1091 and @xmath1092 such that for any @xmath843 : @xmath1093 and @xmath1094 taking a diagonal sequence we have : @xmath1095 and @xmath1096 * ( iv ) * using ( ii ) and ( iii ) above we have that there are sequences @xmath1071 and @xmath1097 such that : @xmath1098 taking a diagonal sequence we have : @xmath1099 * ( v ) * from ( ii ) we have that @xmath1088 and that there is a constant @xmath1076 with @xmath1077 as @xmath854 such that @xmath1100 . hence , for small @xmath333
, there is a constant @xmath246 independent of @xmath333 such that @xmath1101 . then there is a sequence @xmath1102 and a vector - valued function @xmath1103 such that @xmath1104 converges weakly in @xmath48 to @xmath1105 as @xmath1106 .
since @xmath1107 goes to @xmath1108 strongly in @xmath48 as @xmath854 , it follows that : @xmath1109 * ( vi ) * from ( iii ) there is a sequence @xmath1097 such that : @xmath1110 taking the diagonal sequence @xmath1111 we have that @xmath1112 hence @xmath1113 .
we conclude that : @xmath1114 * ( vii ) * we have : @xmath1115 hence , @xmath1116 .
this completes the proof of ( [ equ : mollified - mean - curv1 ] ) .
[ cor : mollified - mean - curv1 ] given a sequence @xmath1003 with @xmath1004 there is a subsequence , that we will continue to denote @xmath1003 , such that : @xmath1117 where @xmath112 is the complex structure on @xmath0 .
[ thm : mollified - mean - curv2 ] given a sequence @xmath1003 with @xmath1004 there is a subsequence , that we will continue to denote @xmath1003 , such that : @xmath1118 where @xmath1 and @xmath1069 are the same measure and vector - valued function as in theorem [ thm : mollified - mean - curv1 ] and @xmath1119 .
the proof is similar to the proof of lemma [ thm : mollified - mean - curv1 ] except that the smooth function @xmath1120 must be taken into account .
observe that @xmath1121 for all @xmath273 and therefore there is a sequence @xmath1122 such that @xmath1123 converges in @xmath45 to @xmath1124 with @xmath1125 .
using the proof of lemma [ thm : mollified - mean - curv1 ] there is a sequence @xmath1122 and vector - valued functions @xmath1126 such that : @xmath1127 and @xmath1128 weakly in @xmath48 .
therefore there are sequences @xmath1003 , @xmath1129 and @xmath1097 such that : @xmath1130 taking a diagonal sequence we have : @xmath1131 using @xmath1132 , the result follows .
[ thm : mollified - mean - curv3 ] given a sequence @xmath1003 with @xmath1004 there is a subsequence , that we will continue to denote @xmath1003 , such that : @xmath1133 where @xmath1 and @xmath1069 are the same measure and vector - valued function as in theorem [ thm : mollified - mean - curv1 ] and @xmath1134 .
using theorem [ thm : boundonbeta ] the set @xmath1135 of radon measures is bounded and therefore there is a sequence @xmath1091 such that @xmath1136 converges weakly to the vector - valued signed measure @xmath1137 .
we have already shown , in the proof of lemma [ thm : mollified - mean - curv1 ] , that there is a sequence @xmath1003 and a vector - valued functions @xmath1138 such that : @xmath1127 weakly in @xmath48 .
therefore the sequence of the radon measures @xmath1139 converges weakly as radon measures to the measure @xmath1140 .
the measure @xmath1137 is absolutely continuous with respect to the measure @xmath650 and therefore there is an integrable function @xmath1141 such that : @xmath1142 observe that @xmath1143 for all @xmath273 and therefore @xmath1144 . also from the proof of lemma [ thm : mollified - mean - curv1 ] there is a a sequence @xmath1003 and vector - valued functions @xmath1145 such that : @xmath1146 weakly in @xmath48 .
therefore , as in the previous proofs , we conclude that : @xmath1147 using @xmath1132 , the result follows .
_ proof of theorem [ thm : volumedecrease ] : _
denote the @xmath273-flow beginning with @xmath1 by @xmath1148 , where @xmath773 is the varifold along the @xmath273-flow at time @xmath7 .
the flow vector field at time @xmath7 is @xmath1149 , where @xmath40 is the lagrangian angle along @xmath861 and @xmath51 is its mollification . denote the mean curvature vector field along the @xmath273-flow by @xmath1150 .
we have by the first variation , @xmath1151 the final equality follows using theorem [ thm : harvey - lawson - varifold ] . clearly , if @xmath1152 , or equivalently @xmath1153 , then @xmath1154 fix @xmath7 .
in case ( 1 ) multiply ( [ equ : deriv - of - volume ] ) by @xmath1155 to get : @xmath1156 choose a sequence @xmath1003 such that @xmath1157 . to estimate the term : @xmath1158 use the hlder inequality to give : @xmath1159 then @xmath1160 for a suitable subsequence of @xmath1003 this term goes to zero as @xmath1161 by lemma [ lem : dom - conv - to - e ] .
applying lemma [ cor : mollified - mean - curv1 ] , lemma [ thm : mollified - mean - curv2 ] and lemma [ thm : mollified - mean - curv3 ] and choosing a subsequence ( that we continue to denote @xmath1003 ) as @xmath1161 the right hand side goes to @xmath1162 .
therefore since @xmath1163 we conclude that : @xmath1164 fix @xmath7 . in case
( 2 ) we rewrite ( [ equ : deriv - of - volume ] ) as @xmath1165 choose a sequence @xmath1003 such that @xmath1166 ( or equivalently , @xmath1167 ) and such that @xmath1168 from this it follows that @xmath1169 @xmath1170 the remaining term can be estimated using the holder inequality and ( [ lem : dom - conv - to - e2 ] ) .
we have : @xmath1171 this term goes to zero since both @xmath1172 and @xmath1173 go to zero as @xmath1004 . since @xmath1168 it follows that @xmath1174 fix @xmath7 . in case
( 3 ) choose a sequence @xmath1003 such that @xmath1166 ( or equivalently , @xmath1167 ) .
repeat the analysis of case ( 2 ) to conclude : @xmath1175 fix @xmath7 .
all three cases show that there is a sequence @xmath1003 such that : @xmath1176 let @xmath30 be a countable dense subset of @xmath811 $ ] . choosing a diagonal subsequence we can conclude that there is a sequence @xmath1003 with @xmath1004 such that for all @xmath1177 ( [ equ : ln - volume - limit ] ) holds . using the continuity of the right hand side of ( [ equ :
deriv - of - volume - scaled ] ) in @xmath273 and @xmath7 we conclude that ( [ equ : ln - volume - limit ] ) holds for all @xmath835 $ ] . hence given @xmath1178 for all @xmath1179 $ ] there exists an integer @xmath10 such that for @xmath1180 we have : @xmath1181 integrating ( [ equ : ln - volume - est ] ) over @xmath811 $ ] we conclude the volume ( or mass ) estimate for @xmath1180 : @xmath1182 in light of the volume estimate ( [ equ : volume - est ] ) we can return to case ( 3 ) above and observe that it does not occur .
hence we conclude that for all @xmath1179 $ ] there is a sequence @xmath1003 with @xmath1004 such that : @xmath1183 [ cor : constructlimitflow ] for each @xmath835 $ ] there is a sequence @xmath1184 with @xmath1004 such that the lagrangian rectifiable varifolds @xmath1185 coverge weakly as radon measures to the lagrangian rectifiable varifold @xmath3 with : @xmath1186 the volume @xmath1187 is a non - increasing , lower semi - continuous function of @xmath7 such that for a.e .
@xmath7 : @xmath1188 it could be the case that @xmath1189 as @xmath303 ; in this case @xmath1190 for all @xmath1191 $ ] with @xmath1192 .
[ thm : betacontrol ] under the same hypotheses as theorem [ thm : volumedecrease ] , for each even integer @xmath921 , on a fixed interval @xmath1193 $ ] , given any @xmath1063 there is an @xmath1194 such that for @xmath1195 : @xmath1196 in particular , for any @xmath1195 and any @xmath1197 $ ] with @xmath1198 : @xmath1199 moreover @xmath1066 as @xmath1200 . for each @xmath273-flow , @xmath1201 we now use propositions [ divergenceformula ] and [ powerrule ] to trade the divergence operator for weak derivatives : @xmath1202\cdot d(\beta^{\varepsilon})_{\varepsilon}d{\lvert v^{\varepsilon}\rvert}\\ \int(\beta^{\varepsilon})^p{\operatorname{div}}jd(\beta^{\varepsilon})_{\varepsilon}d{\lvert v^{\varepsilon}\rvert}&=-\int\left[p(\beta^{\varepsilon})^{p-1}b^{\varepsilon}-(p-1)(\beta^{\varepsilon})^ph^{\varepsilon}\right]\cdot jd(\beta^{\varepsilon})_{\varepsilon}d{\lvert v^{\varepsilon}\rvert } \end{aligned}\ ] ] now we apply the harvey - lawson formula @xmath1203 in each term : @xmath1204\cdot d(\beta^{\varepsilon})_{\varepsilon}d{\lvert v^{\varepsilon}\rvert}\\ & & = -p\int \left[-(p-1)(\beta^{\varepsilon})^{p-2}jh^{\varepsilon}+(\beta^{\varepsilon})^{p-1}h^{\varepsilon}\right]\cdot d(\beta^{\varepsilon})_{\varepsilon}d{\lvert v^{\varepsilon}\rvert}\\ & & = -p(p-1)\int(\beta^{\varepsilon})^{p-2}h^{\varepsilon}\cdot jd(\beta^{\varepsilon})_{\varepsilon}d{\lvert v^{\varepsilon}\rvert}-p\int(\beta^{\varepsilon})^{p-1}h^{\varepsilon}\cdot d(\beta^{\varepsilon})_{\varepsilon}d{\lvert v^{\varepsilon}\rvert}\\ \end{aligned}\ ] ] @xmath1205\cdot jd(\beta^{\varepsilon})_{\varepsilon}d{\lvert v^{\varepsilon}\rvert}\\ & & = -\int\left[-p(\beta^{\varepsilon})^{p-1}jh^{\varepsilon}+(\beta^{\varepsilon})^ph^{\varepsilon}\right]\cdot jd(\beta^{\varepsilon})_{\varepsilon}d{\lvert v^{\varepsilon}\rvert}\\ & & = p\int ( \beta^{\varepsilon})^{p-1}h^{\varepsilon}\cdot d(\beta^{\varepsilon})_{\varepsilon}d{\lvert v^{\varepsilon}\rvert}-\int(\beta^{\varepsilon})^ph^{\varepsilon}\cdot jd(\beta^{\varepsilon})_{\varepsilon}d{\lvert v^{\varepsilon}\rvert } \end{aligned}\ ] ] adding these terms together we obtain @xmath1206 now by brakke s perpendicularity result , we can write @xmath1207 in this integrand , so that we obtain @xmath1208 both terms are of the form @xmath1209 .
@xmath1210 hence , @xmath1211 using ( [ equ : integral - bound - h ] ) if @xmath1152 then @xmath1212 using the uniform sup bound on @xmath963 and ( [ equ : integral - bound - h ] ) the argument proceeds exactly as in the proof of theorem [ thm : volumedecrease ] .
the result follows .
[ cor : beta - lipschitz ] under the same hypotheses as theorem [ thm : betacontrol ] , the conclusions hold for @xmath1213 . in particular , given any @xmath1063
there is an @xmath1194 such that for @xmath1195 and any @xmath1197 $ ] with @xmath1198 : @xmath1214 moreover @xmath1066 as @xmath1200 .
[ cor : norm - beta - decreasing ] for each @xmath835 $ ] there is a function @xmath1215 that is the scalar lift of the lagrangian angle on @xmath3 .
the norms @xmath1216 are non - increasing , lower semi - continuous functions of @xmath7 and for a.e .
@xmath7 @xmath1217 the existence of the scalar lifts of the lagrangian angle follows from theorem [ thm : convergence - beta ] . the non - increasing , lower semi - continuous statement and derivative bound follow from corollary [ cor : beta - lipschitz ]
in this section we show how to extend the lagrangian varifold @xmath1218 , constructed above on a compact time interval @xmath811 $ ] , to all @xmath4 .
we show that for a.e . @xmath4
the lagrangian varifold @xmath3 satisfies @xmath537 and has maslov index zero .
unfortunately we do not have uniform bounds on the mean curvature of @xmath3 , however , theorem [ thm : good - times - mc - bounds ] and theorem [ thm : h2-bounded - infinity ] below are partial replacements .
recall that @xmath1219 .
@xmath1220 : \limsup_{{\varepsilon}\to 0 } { \lambda}_{{\varepsilon}}(t ) > m \}\ ] ] @xmath1221 : \limsup_{{\varepsilon}\to 0 } { \lambda}_{{\varepsilon}}(t ) = \infty \}\ ] ] [ thm : first - bound - mc ] suppose we have @xmath1222 } \lvert v(t)\rvert\geq \alpha>0 $ ] .
the one dimensional lebesgue measure of @xmath1223 , denoted @xmath1224 satisfies the bound : @xmath1225 consequently , @xmath1226 first assume that for all @xmath50 and @xmath1179 $ ] that @xmath1227 . using the uniform sup bound on @xmath1228 and the behavior of mollification as @xmath55 we have for each @xmath1179 $ ] that there exists an @xmath1229 such that for @xmath1230 : @xmath1231 using the continuity in @xmath54 and @xmath7 of @xmath1232 and @xmath1233 we can find an @xmath1234 such that the inequality ( [ equ : scaled - inequality ] ) holds for any @xmath1235 and all @xmath1179 $ ] .
hence for all @xmath1179 $ ] and all @xmath1235 we have : @xmath1236 this inequality includes the cases when @xmath1237 .
hence we have for all @xmath1179 $ ] and all @xmath1235 : @xmath1238 integrate to derive for @xmath1235 : @xmath1239 it follows that for for @xmath1235 : @xmath1240 set : @xmath1241 @xmath1242 then , for a countable dense subset @xmath1243 of @xmath1244 there is a sequence @xmath1003 with @xmath1004 such that for @xmath1245 , @xmath1246 . from the continuity of @xmath1233 in @xmath7
it follows that for all @xmath1247 : @xmath1248 therefore as @xmath1004 we have @xmath1249 and @xmath1250 .
consider ( [ equ : inequ - on - small - set ] ) with @xmath1251 .
then multiply both sides by @xmath1252 to get : @xmath1253 \label{equ : inequ - on - small - set - scaled } & & < { \lambda}_{{\varepsilon}_i}({{{\mathcal}s}_m})^{-2 } m_{{\varepsilon}_i}({{{\mathcal}s}_m})^{-1 } 2 || v^{{\varepsilon}_i}(0 ) ||.\end{aligned}\ ] ] clearly for all @xmath1247 @xmath1254 therefore as @xmath1004 the left hand side of ( [ equ : inequ - on - small - set - scaled ] ) is greater than or equal to @xmath1255 the limit of the right hand side is bounded above by @xmath1256 therefore @xmath1257 the result follows .
set : @xmath1258 : \limsup_{{\varepsilon}\to 0 } { \lambda}_{{\varepsilon}}(t ) \leq m \}\ ] ] [ thm : bdd - first - variation ] for each @xmath1259 the lagrangian varifold @xmath3 : 1 .
satisfies @xmath537 .
the generalized mean curvature of @xmath3 , @xmath1260 satisfies the bound @xmath1261 + 3 .
the lagrangian angle @xmath1215 has weak derivative @xmath1262 and therefore @xmath3 has maslov index zero .
the first two items follow from allard s compactness theorem for varifolds theorem [ thm : compact - hp - varifolds ] .
the last item follows from theorem [ thm : convergence - beta ] and corollary [ cor : norm - beta - decreasing ] .
note that the first and last items in theorem [ thm : bdd - first - variation ] are independent of the value of @xmath27 and that as @xmath1263 , @xmath1264 .
therefore : [ cor : aezeromaslov ] for almost every @xmath835 $ ] the lagrangian rectifiable varifold @xmath3 satisfies the conclusions ( 1 ) and ( 3 ) of theorem [ thm : bdd - first - variation ] . the limit flow @xmath1218 has the property that for almost every @xmath1179 $ ] each lagrangian varifold satisfies @xmath537 and has maslov index zero .
however there are no uniform bounds on @xmath1265 .
this failure is partial rectified by the following result : [ thm : good - times - mc - bounds ] suppose @xmath1222 } \lvert v(t)\rvert\geq \alpha$ ] .
then : @xmath1266 from the proof of theorem [ thm : first - bound - mc ] there is an @xmath1267 such that for each @xmath1179 $ ] and @xmath1268 : @xmath1269 using the notation @xmath1233 and @xmath1270 as defined above we have : @xmath1271 now @xmath1272 .
therefore : @xmath1273 recall that as @xmath854 , @xmath1274 strongly in @xmath1275 .
therefore for any @xmath7 and @xmath273 , as @xmath854 : @xmath1276 it follows that for any @xmath835 $ ] and @xmath1277 $ ] there is an @xmath1278 such that for @xmath1279 : @xmath1280 in particular , there is an @xmath1194 such that for @xmath1195 : @xmath1281 hence for any @xmath835 $ ] and @xmath1282 : @xmath1283 thus when @xmath1284 we can integrate ( [ equ : mcintegral - ineq ] ) to derive : @xmath1285 recall that for each @xmath54 the masses @xmath1286 are bounded independent of @xmath1179 $ ] by a constant @xmath1287 and that as @xmath55 , @xmath1288 .
since @xmath1289 it follows that , @xmath1290 let @xmath1003 be a sequence such that @xmath1291 as @xmath1004 .
then : @xmath1292 given a lagrangian integer varifold @xmath1293 that satisfies @xmath537 with maslov index zero there exists a lagrangian flow @xmath1218 with @xmath1294 that exists for all time @xmath1295 . by the constructions of the previous sections given an interval @xmath811 $ ]
there is a lagrangian flow @xmath1218 with @xmath1294 on @xmath811 $ ] .
if @xmath1296 for some @xmath1297 $ ] then extend the flow by the zero varifold . otherwise there is a constant @xmath203 such that @xmath1298 for all @xmath1179 $ ] .
theorem [ thm : good - times - mc - bounds ] applies .
it follows that there is a @xmath1299 $ ] such that @xmath1300 is a lagrangian integer varifold with maslov index zero satisfying @xmath537 .
apply our construction with initial varifold @xmath1300 to construct a lagrangian flow on @xmath1301 $ ] .
iterating , the result follows .
[ thm : h2-bounded - infinity ] given any lagrangian flow @xmath1218 , if @xmath1298 for @xmath1302 $ ] then @xmath1303 in particular , if @xmath1218 is a lagrangian flow with @xmath1298 for all @xmath1058 then @xmath1304 we conclude : the flow @xmath1305 can be extended for all @xmath4 while preserving the properties : 1 .
each @xmath3 is a lagrangian integer @xmath21-varifold .
2 . @xmath6 is a nonincreasing , lower semicontinuous function of @xmath7 .
3 . for each @xmath3 the lagrangian angle , @xmath1306 , admits a scalar - valued lift so that @xmath1307 is a nonincreasing , lower semicontinuous function of @xmath7 .
4 . for almost every @xmath4
, @xmath3 satisfies @xmath537 .
5 . for almost every @xmath4
, @xmath1306 has a weak derivative in @xmath1308 , i.e. @xmath3 has maslov index zero .
now we will show how to adapt our construction of the hamiltonian flow of lagrangian varifolds to a flow of integral cycles in a calabi - yau manifold .
observe that the definitions of weak derivative and maslov index zero are sensible in this context .
first consider a lagrangian varifold @xmath1 with @xmath2 and maslov index zero in a closed calabi - yau manifold @xmath1309 .
we define @xmath1310 where @xmath1311 is the riemannian distance with respect to the calabi - yau metric on @xmath17 .
we will assume that @xmath273 is always less than the injectivity radius of @xmath17 .
we mollify according to @xmath1312 since @xmath1313 is smaller than the injectivity radius of @xmath17 , @xmath277 and hence @xmath280 is smooth .
if the ball @xmath1314 has compact closure in a convex coordinate neighborhood @xmath1315 , then alternately we could use the euclidean distance @xmath1316 between @xmath1317 . however in @xmath1318 the euclidean distance and the distance in the calabi - yau metric are equivalent .
the constructions and estimates in previous sections all carry over to this mollifier up to fixed constants .
in particular , we have the following theorem : [ thm : varifold - flow - cy ] given any lagrangian varifold @xmath1 with @xmath2 and maslov index zero in a closed calabi - yau manifold @xmath1309 , there is a one - parameter family @xmath3 of lagrangian varifolds in @xmath17 , defined for all @xmath4 , with @xmath1319 and 1 .
each @xmath3 is a lagrangian integer varifold .
2 . @xmath6 is a nonincreasing , lower semicontinuous function of @xmath7 .
3 . each @xmath3 admits a scalar lift of the lagrangian angle , @xmath1306 , so that @xmath1307 is a nonincreasing , lower semicontinuous function of @xmath7 .
4 . for almost every @xmath4
, @xmath3 satisfies @xmath537 .
5 . for almost every @xmath4
, @xmath1306 has a weak derivative @xmath1308 , i.e. @xmath3 is maslov index zero .
the basic object we will use is the integer multiplicity current as described in @xcite and originally introduced by federer and fleming @xcite . as described in @xcite
the integer multiplicity currents are obtained by assigning an orientation to the approximate tangent space @xmath305 of an integer multiplicity rectifiable varifold @xmath1 .
this allows the integration of @xmath21-forms along the varifold making the varifold into a current .
we will always require that the associated integer multiplicity rectifiable varifold @xmath1 is lagrangian and satisfies @xmath537 .
let @xmath562 be an integer multiplicity current . if the rectifiable varifold @xmath1320 associated to @xmath562 is lagrangian and satisfies @xmath537 , we will call @xmath562 a _ lagrangian integer multiplicity current _ or briefly a _ lagrangian current_. if @xmath1321 , we will call @xmath562 a _ lagrangian integer multiplicity cycle _ or briefly a _
lagrangian cycle_. the essential compactness result for lagrangian currents and cycles is a consequence of results of federer - fleming @xcite and allard @xcite .
proofs can be found in @xcite and @xcite .
we recall the definition of mass and weak convergence of currents : let @xmath562 is an @xmath21-current in the open set @xmath181
. denote the set of @xmath21-forms with support in @xmath181 by @xmath1322 .
the mass of the current @xmath562 , denoted @xmath1323 is given by : @xmath1324 more generally for any open @xmath194 we set : @xmath1325 we say a sequence @xmath1326 converges weakly to a current @xmath562 , denoted @xmath1327 , if @xmath1328 for all @xmath1329 [ thm : current - compactness ] let @xmath1315 be open .
if @xmath1330 is a sequence of lagrangian currents ( cycles ) supported in @xmath181 with : @xmath1331 then there is a lagrangian current ( cycle ) @xmath58 supported in @xmath181 and a subsequence @xmath1332 such that @xmath1333 .
the proof follows from theorems 27.3 and 42.7 in @xcite and allard s integral compactness theorem in @xcite . the lagrangian condition is easily seen to be preserved under weak convergence of currents .
[ defn : zero - maslov - current ] an lagrangian current ( cycle ) with associated lagrangian varifold that has maslov index zero will be called a _
lagrangian maslov index zero current ( cycle)_. given a lagrangian maslov index zero cycle @xmath10 whose associated varifold @xmath1 satisfies @xmath537 , the construction of the @xmath273-flows in section [ sec : epsilon - flows ] can be applied to the associated lagrangian varifold to produce , for each @xmath50 , a one - parameter family of hamiltonian diffeomorphisms @xmath815 , hence a one - parameter family of lagrangian maslov index zero cycles @xmath1334 .
we define , analogously to corollary [ cor : constructlimitflow ] , @xmath12 to be the cycle with @xmath1335 where we use the compactness theorem [ thm : current - compactness ] to ensure that there is such a limit . in the manifold case
we have : let @xmath27 be a compact calabi - yau manifold .
let @xmath1336 be a lagrangian homology class ( a class that can be represented by a lagrangian cycle ) .
suppose that @xmath1337 can be represented by a lagrangian maslov index zero cycle .
then the lagrangian cycles @xmath1338 of the @xmath273-flows and the lagrangian cycles @xmath1339 of the limit lagrangian flow all have maslov index zero and all represent @xmath1337 .
@xmath1340 is related to @xmath10 by a one - parameter family of hamiltonian diffeomorphisms , hence represents the same homology class .
moreover weak convergence of currents preserves homology class .
suppose the lagrangian homology class @xmath1336 is non - trivial .
set @xmath1341 where @xmath562 is a cycle ( not necessarily lagrangian ) that represents @xmath1337 .
then @xmath1342 is a lower bound on the mass of all the varifolds @xmath1343 of the @xmath273-flows and @xmath1344 of the limit lagrangian flow . if @xmath1345\neq 0\in h_n(m,{\mathbb r})$ ]
, then there is a sequence of times @xmath1346 so that @xmath1347 converges to a stationary current , which we may take to have maslov index zero . by theorem [ thm : h2-bounded - infinity ]
, we have : @xmath1348 so there is a sequence of times @xmath1346 with @xmath1349 . by theorem [ thm : varifold - flow - cy ] we may choose the sequence @xmath1350 so that each @xmath1347 has maslov index zero . by theorem [ thm : varifold - flow - cy ]
there are uniform bounds on @xmath1351 .
since @xmath1349 , we obtain uniform bounds on the weak derivative @xmath1352 .
the corresponding @xmath1347 have uniformly bounded mass , so by the compactness theorems [ thm : convergence - beta ] and [ thm : current - compactness ] , passing to a subsequence we obtain some limit cycle @xmath16 and associated maslov index zero varifold @xmath1353 with @xmath1354 .
denote the limit stationary cycle by @xmath16
in this section we prove : [ thm : convergence ] @xmath16 is the sum of finitely many special lagrangian integral cycles . a smooth lagrangian submanifold is special lagrangian if and only if it is stationary ( and in this case it has vanishing maslov class ) ; we have the following generalization of this fact to our setting .
[ prop : sum - sl ] suppose @xmath10 is a lagrangian current with finite mass and no boundary in the ball @xmath1355 , such that the corresponding varifold @xmath1 is stationary and has maslov index zero .
then there are special lagrangian currents , @xmath1356 , with @xmath1357 in @xmath1355 .
since @xmath1 is stationary , by allard s regularity theorem there are an open dense set of @xmath1358 such that for each @xmath344 there is some @xmath1359 such that in @xmath1360 , @xmath1361 is smooth and hence special lagrangian .
in particular we can select such a @xmath344 with @xmath1362 .
call the phase of this special lagrangian @xmath1363 .
the idea is to show that @xmath1364 , where @xmath40 is the lagrangian angle of @xmath1 , contains some definite positive amount of the mass of @xmath1 .
however to make sense of @xmath1365 we need @xmath40 to be an ambient lipschitz function .
therefore we mollify @xmath40 to construct a smooth ambient function @xmath1366 .
[ lem : coarea ] for almost every @xmath1367 , @xmath1368 for each @xmath359 , we may apply the coarea formula for rectifiable sets , namely @xmath1369 the conclusion follows from the error estimate on @xmath1370 : @xmath1371 where ( appealing to theorem [ thm : harvey - lawson - varifold ] ) @xmath1372 and @xmath1373 are both zero because @xmath1 is stationary .
the remaining term satisfies @xmath1374 however , as @xmath854 @xmath1375 by lemma [ lem : sequence - allard2 ] . the result follows . _ proof of proposition [ prop : sum - sl ] _
consider , for each @xmath1376 , @xmath1377 each @xmath1378 an integral current whose support lies in the support of @xmath1 .
we seek a limit of @xmath1378 as @xmath1379 in the sense of integral currents and in the sense of varifolds .
the masses of the @xmath1378 are dominated by the mass of @xmath1 .
note that by hypothesis , @xmath10 is without boundary .
thus the only part of @xmath1380 in @xmath1381 , has support in @xmath1382 now we can apply lemma [ lem : coarea ] to choose @xmath1383 for which @xmath1384 in particular , there is a sequence of the @xmath1378 converging to some @xmath1385 , an integral lagrangian current without boundary in @xmath1381 , with @xmath1386 . let @xmath538 be the associated lagrangian varifold to @xmath1385 then we have @xmath1387 @xmath538-a.e .. thus , @xmath538 is stationary and we can apply the monotonicity formula about @xmath344 with radii @xmath659 and @xmath1388 : @xmath1389 the last inequality follows because @xmath344 is a regular point of @xmath1 .
this inequality shows that @xmath1385 accounts for at least @xmath1390 of the mass of @xmath1 .
hence we can repeat the argument for some @xmath1391 to find @xmath1392 , and so on until the mass of @xmath1 is exhausted . the above proof is inspired by the proof of proposition 5.1 of @xcite . to establish theorem [ thm : convergence ] ,
we need only note that @xmath1353 is stationary ; then we may apply proposition [ prop : sum - sl ] to @xmath16 obtain the desired result . in the decomposition @xmath1393
, the phases of the special lagrangians may differ .
in particular , the configuration of special lagrangians may not be minimizing .
classical compact mean curvature flows in euclidean space must encounter singularities in finite time , but these singularities may not be caused by global volume collapse . here
we show that lagrangian varifold flow exhibits the same behavior , in fact collapsing to zero mass in finite time .
[ thm : collapse ] if @xmath287 lies in a ball of radius @xmath1388 , then @xmath1394 lies in a ball of radius @xmath1395 .
consider the function @xmath1396 .
compute , for each @xmath773 , @xmath1397 for almost every @xmath7 , as @xmath303 , @xmath1398 . for such @xmath7
, we have @xmath1399 now an elementary computation shows that @xmath1400 , so we have @xmath1401 hence @xmath1402 with equality holding only if @xmath3 is stationary or the zero varifold .
thus for small enough @xmath273 , @xmath1403 is nonincreasing in @xmath7 .
letting @xmath303 , we have @xmath1404 nonincreasing in @xmath7 . + similarly , for any @xmath1405 , we have @xmath1406 and @xmath1407 and @xmath1408 , so that @xmath1409 letting @xmath303
, we conclude that @xmath1410 . letting @xmath1411 ,
we conclude that @xmath1412 . +
but since @xmath279 is continuous , we have that for @xmath1413 , @xmath1414 the proof of theorem [ thm : collapse ] follows that of brakke @xcite .
it is almost but not precisely a barrier argument .
the rate of collapse @xmath1415 is slower than that ( @xmath1416 ) determined by mean curvature flow applied to the @xmath1417-sphere bounding the ball of radius @xmath1388 which contained the support of the initial varifold . for any initial varifold with compact support
, the flow @xmath3 becomes the zero varifold in finite time .
hyperplanes are barriers for the lagrangian varifold flow . by a similar argument one has the following result , which controls the rate of collapse . if the support of @xmath1 is disjoint from the ball of radius @xmath1388 , then the support of @xmath3 is disjoint from the ball of radius @xmath1395 .
[ thm : maintopological ] let @xmath27 be a closed calabi - yau manifold and @xmath1418 . if @xmath1419 can be represented by a lagrangian cycle with maslov index zero and @xmath1420 , then @xmath1421 , for some @xmath1422 , each of which can be represented by a special lagrangian cycle .
let @xmath17 be a closed calabi - yau manifold .
if an integral lagrangian homology class @xmath18 can be represented by an immersed lagrangian submanifold with vanising maslov class , then @xmath19 where each @xmath20 is a lagrangian homology class that can be represented by a special lagrangian current .
the phases of the calibrating @xmath21-forms may be different for each @xmath22 .
[ cor : thomas - yau ] if @xmath80 is an embedded lagrangian submanifold with vanishing maslov class , then there is a mass - decreasing flow of lagrangian currents starting from @xmath80 and converging ( in infinite time ) in a sum @xmath1426 of special lagrangian cycles .
let @xmath1427 be a closed calabi - yau manifold of complex dimension @xmath21 with khler form @xmath1428 .
there are two reasonable definitions of lagrangian homology class .
the one used in the introduction : a class in @xmath23 is called _ a lagrangian homology class _ if it can be represented by a simplex consisting of simplices with all @xmath21-simplices lagrangian . the topological definition : a class in @xmath1429 is called _ a lagrangian homology class _ if @xmath1430 = 0 $ ] .
see @xcite for results relating these two definitions .
the results on lagrangian homology stated in the introduction apply with either definition .
denote the subspace of lagrangian homology classes by @xmath24 .
let @xmath25 be the subspace of the lagrangian homology that is generated by the special lagrangian cycles with any phase .
[ thm : homology - image ] let @xmath17 be a closed calabi - yau manifold of complex dimension @xmath21 . if an integral lagrangian homology class @xmath18 can be represented by the image of a smooth map @xmath26 , where @xmath27 is a simply connected closed @xmath21-manifold then @xmath28 . in particular , the image of the hurewicz homomorphism @xmath29 in the lagrangian homology lies in @xmath30 .
suppose the image of a smooth map @xmath26 represents a lagrangian homology class ( for either definition ) .
then @xmath1431 ) = 0 $ ] .
using the h - principle @xcite @xcite @xmath279 is homotopic to a lagrangian immersion @xmath1432 . perturbing the lagrangian immersion we can suppose it has a at worst finite number of isolated double points .
since @xmath27 is simply connected it follows that @xmath1433 is a lagrangian cycle with maslov index zero and @xmath1434 .
the result follows .
we can assume @xmath1438 .
we will show that if the class @xmath18 can be represented by the image of a smooth map @xmath26 then @xmath11 can be represented by the image of a smooth map @xmath1439 where @xmath1440 is simply - connected .
first we observe that the map @xmath279 can be assumed to be an immersion with , at worst , double points .
if @xmath27 is not simply - connected there is a homotopically non - trivial curve @xmath1441 .
we can assume that @xmath1442 is imbedded in @xmath17 and therefore spans an imbedded disc in @xmath17 .
we write @xmath1443 for this imbedded disc and note that @xmath1444 and @xmath1445 along @xmath1446 . in particular , @xmath1447 is continuous and represents @xmath1429 .
let @xmath378 be a tubular neighborhood of @xmath1446 .
attach a @xmath1448-handle @xmath1449 to @xmath1450 by gluing @xmath1451 to @xmath181 with @xmath1452 identified with @xmath111 .
denote the resulting space by @xmath97 .
set @xmath1453 .
then @xmath1440 is an @xmath21-manifold .
there is a deformation retract @xmath659 of @xmath97 onto @xmath1450 .
therefore there is a continuous map @xmath1454 given by the composition of @xmath659 and @xmath1455 .
denote the restriction of this map to @xmath1440 by @xmath1456
. then @xmath1457 represents @xmath11 in homology .
the map @xmath1439 is continuous . by whitney approximation
we can approximate @xmath1458 by a smooth map and hence by an immersion .
iterate this process until there are no non - trivial curves @xmath1446 in @xmath27 . if @xmath1435 then by r. thom s work on the steenrod problem @xcite every class in @xmath23 can be represented by the image of a smooth map @xmath1459 .
the result now follows in the cases @xmath1460 from theorem [ thm : homology - image ] .
this conjecture should be contrasted with the related but quite different problem of characterizing the classes in @xmath24 that can be represented by a special lagrangian variety with fixed phase .
this is a version of a special lagrangian hodge conjecture . for this problem
it is known that there are lagrangian classes in a @xmath1461 surface that can not be represented a special lagrangian variety with fixed phase @xcite .
though by corollary [ cor : k3 ] all lagrangian classes in any @xmath1461 surface lie in @xmath30 and therefore can be represented by a sum of special lagrangian varieties possibly with differing phases .
if @xmath1462 is a lagrangian mean curvature flow and @xmath1463 has vanishing maslov class , so does each @xmath1464 ( see e.g. @xcite ) .
in this situation the lagrangian angle @xmath1306 of each time - slice @xmath1464 can be locally extended to an ambient smooth function ( still denoted @xmath1306 ) so that @xmath1464 is generated by the hamiltonian motion @xmath1465 . in this case , the gmt estimates we rely on all have improved versions coming from the fact that the @xmath1464 satisfy uniform @xmath1466 bounds on compact time intervals , so it can be shown that @xmath1467 uniformly as @xmath303 . if @xmath1462 is a lagrangian mean curvature flow with vanishing maslov class , the lagrangian varifold flow @xmath3 starting from the integer - rectifiable varifold corresponding to @xmath1463 is unique and @xmath3 is the integer - rectifiable varifold corresponding to @xmath1464 for each @xmath1468 .
classical mean curvature flow in general is expected to encounter singularities in finite time .
our construction , on the other hand , exists for all @xmath4 .
thus we may think of the lagrangian varifold flow as a way to extend classical lagrangian mean curvature flow past singularities , in the case of vanishing maslov class .
+ we also have a local agreement theorem : suppose @xmath1469 is an open set and @xmath3 is a lagrangian varifold flow so that for each @xmath1470 $ ] , @xmath1471 is the integer - rectifiable varifold corresponding to the intersection of some embedded lagrangian submanifold @xmath1464 with @xmath181 .
@xmath1464 satisfy mean curvature flow on @xmath1472 $ ] .
consider a lagrangian flow @xmath1473 .
suppose at time @xmath1474 the lagrangian varifold @xmath1475 has maslov index zero and satisfies @xmath537 .
then we can construct a new lagrangian flow @xmath1476 with @xmath1477 .
we do not know if the lagrangian flows @xmath1478 and @xmath1476 coincide .
suppose that at time @xmath1479 the lagrangian varifold @xmath1480 is stationary and therefore consists of components each with constant lagrangian angle .
the varifold @xmath1480 is not , in general , a minimizer of volume .
therefore there may be a time @xmath1481 such that for @xmath1482 the volume of the lagrangian varifolds @xmath1483 is less than the volume of @xmath1480 .
almgren , f. , q - valued functions minimizing dirichlet s integral and the regularity of area - minimizing rectifiable currents up to codimension 2 , almgren s big regularity paper , world scientific monograph series in mathematics , vol .
1 , world scientific publishing co. inc . ,
river edge , nj ( 2000 ) .
joyce , d. , conjectures on bridgeland stability for fukaya categories of calabi - yau manifolds , special lagrangians , and lagrangian mean curvature flow , ems surveys in mathematical sciences * 2 * ( 1 ) 1 - 62 ( 2015 ) . | we introduce a notion of vanishing maslov index for lagrangian varifolds and lagrangian integral cycles in a calabi - yau manifold .
we construct mass - decreasing flows of lagrangian varifolds and lagrangian cycles which satisfy this condition .
the flow of cycles converges , at infinite time , to a sum of special lagrangian cycles .
we use the flow of cycles to obtain the fact that special lagrangian cycles generate the part of the lagrangian homology which lies in the image of the hurewicz homomorphism .
we also establish a weak version of a conjecture of thomas - yau regarding lagrangian mean curvature flow . |
this work was supported by national science foundation grant no .
we thank pawel hawrylak for many insightful conversations .
goldberg , d. heiman , a. pinczuk , l. pfeiffer , and k. west , _ phys .
lett . _ * 65 * , 641 ( 1990 ) ; a.j .
turberfield , s.r .
haynes , p.a .
wright , r.a .
ford , r.g .
clark , j.f .
ryan , j.j .
harris , and c.t .
foxon , _ phys .
lett . _ * 65 * , 637 ( 1990 ) ; h. buhmann , w. joss , k. von klitzing , i.v .
kukushkin , g. martinez , a. plaut , k. ploog , and v.b .
timofeev , _ phys .
lett . _ * 65 * , 1056 ( 1990 ) . * figure 1 . * photoluminescence spectra in the lcp ( @xmath16 ) polarization at t=0.53k for the sqw with @xmath31 and @xmath32 .
the spectra at @xmath1 is highlighted to show the low - energy fine - structure .
the inset to fig .
1 displays the lowest - energy feature tuning to lower energy for increasing magnetic field .
* figure 2 .
* individual spectra at @xmath1 displaying low energy anomalies oa@xmath21 and oa@xmath22 .
the fundamental recombinations from the 0th and 1st landau level are labeled ll@xmath21 and ll@xmath22 respectively .
note the strong intensity of the oa@xmath21 peak relative to the fundamental ll@xmath21 .
* figure 3 . * recombination peak positions as function of magnetic field at t=0.53k .
the low energy features are labeled oa@xmath21 and oa@xmath22 .
the fundamental recombination from the 0th landau level is labeled ll@xmath21 . also shown is the line e=-1.65mev / t @xmath30b+e@xmath21 as explained in the text .
* figure 4 . * magnetic field development of optical anomalies between @xmath2 and @xmath3 .
all spectra at taken in the @xmath16 polarization at t=0.53k . for ease of comparison ,
all spectra have been normalized to have the same peak intensity in the ll@xmath21 transition . | we report an anomalous doublet structure and low - energy satellite in the magneto - photoluminescence spectra of a two - dimensional electron system .
the doublet structure moves to higher energy with increasing magnetic field and is most prominent at odd filling factors @xmath0 and @xmath1 .
the lower - energy satellite peak tunes to _ lower _ energy for _ increasing _ magnetic fields between @xmath2 and @xmath3 .
these features occur at energies below the fundamental band of recombination originating from the lowest landau level and display striking magnetic field and temperature dependence that indicates a many - body origin .
drawing on a recent theoretical description of hawrylak and potemski , we show that distinct mechanisms are responsible for each feature . over the last decade
a substantial body of work has focused on the investigation of the photoluminescence spectra originating from the recombination of a two - dimensional electron system with a valence - band hole in the quantum hall regime . in conjunction with transport measurements ,
much of this work has concentrated on examining the nature of the incompressible ground states that arise at integral and fractional filling.@xcite more recently , optical techniques have proven sensitive to excited states of the 2deg inaccessible to transport .
in particular , two groups have recently reported on new low - energy structure in the magneto - photoluminescence in gaas structures occurring at energies below the fundamental recombination from the lowest landau level ( lll).@xcite while clearly indicative of many - body effects , these new experimental results have led to differing proposals for the mechanisms responsible for the various spectral anomalies and a consensus has yet to be reached .
these features have alternatively been attributed to a perturbative many - body shake - up process@xcite and to a non - perturbative final state resonance of the 2deg.@xcite it is the purpose of this paper to present the results of a comprehensive experimental study of the polarization , magnetic field and temperature dependence of these optical anomalies and elucidate the distinct mechanisms responsible for each .
we report on the observation of unusual low - energy structure in the magneto - photoluminescence spectra for filling factors @xmath4 .
we identify and analyze two distinct features : ( i ) a doublet structure in the fundamental recombination from the lowest landau level develops at odd integral filling factors @xmath0 and @xmath1 , ( ii ) a significantly smaller satellite peak which red shifts for increasing magnetic field between @xmath2 and @xmath3 is observed at approximately 2mev below the doublet structure . while these features are separated by rather small energy differences and occur roughly in the same regime of magnetic field , we shall show that the data conclusively point to distinct mechanisms for each process .
the two - dimensional electron system under investigation is a single - side @xmath5-modulation doped al@xmath6ga@xmath7as - gaas 250@xmath8 single quantum well ( sqw ) .
samples from three different wafers have been studied with similar behavior seen in all samples .
all three samples are of extremely high quality : electronic densities n@xmath9 range from @xmath10 and mobilities @xmath11 range from @xmath12 .
the samples were excited at 740 nm with the output of a tunable ti : sapphire laser .
the laser was stabilized and the incident power density was kept below @xmath13 .
we have verified that all measurements were made in the linear regime by collecting data with incident power densities at @xmath14 with no change in the observed spectra .
excitation delivery and signal collection were accomplished via an optical fiber system that resides in he@xmath15 refrigerator mounted in a 13 t magnet .
base temperature of our configuration is 500mk .
all measurements were made in the transmission geometry and polarization analysis is done _ in situ _ with a circular polarizer placed immediately following the sample but prior to the collection fiber .
the signal is dispersed in a 0.64 m monochromator and detected with a liquid nitrogen - cooled charged - coupled - device ( ccd ) camera with 0.2mev resolution .
figure 1 displays the low temperature magneto - photoluminescence spectra in the left - circular ( @xmath16 ) polarization in the field range of 0 t to 4 t for a sample with n@xmath9=@xmath17 and @xmath18 .
the use of polarization analysis is crucial since the high quality of these samples results in narrow line - widths in which spin - splittings are resolved above 1.5 t . detecting in the @xmath16 polarization guarantees that the fundamental recombination from the lll occurs between a @xmath19 electron level and a @xmath20 hole level .
the application of a quantizing magnetic field causes the zero field spectrum to split into the landau fan structure .
the gross behavior of these spectra and its relationship to the quantum hall effect have been studied extensively.@xcite we focus our discussion on the low energy structure between @xmath0 and @xmath3 appearing below the main peak labeled ll@xmath21 in figure 2 . within the context of a non - interacting model , ll@xmath21
is associated with the ground - state recombination of an electron and hole , each sitting on the lowest landau level .
figure 2 shows an individual spectra taken at @xmath1 and t=0.53k in which two distinct low - energy features are visible below the fundamental ground - state recombination ll@xmath21 . our analysis will focus on the origin of these two optical anomalies , labeled oa@xmath21 and oa@xmath22 .
we begin our discussion with the lowest energy feature , labeled oa@xmath22 , which we attribute to the process of shake - up in a magnetic field . in a shake - up process a recombining electron - hole pair perturbs the electron gas causing another electron to be excited across the cyclotron gap from the lowest landau level to the first landau level .
energy conservation requires that the emitted photon s energy be lower than the fundamental recombination by @xmath23 @xmath24 .
shake - up in a magnetic field has been studied experimentally with similar behavior reported in ingaas - inp quantum wells@xcite , and most recently , in gaas structures.@xcite the theory of shake - up has also been extensively investigated.@xcite most theoretical approaches have assumed a significant amount of disorder in which electron - electron interactions can be treated perturbatively and only cyclotron gaps are considered .
zeeman gaps , which are typically unresolved in such systems , are ignored , and consequentially the shake - up process is independent of the spin configuration of the 2deg .
as shown in figure 3 , oa@xmath22 moves to lower energy for increasing magnetic fields with a slope approximately equal to -1.65mev / t , corresponding to an inter - landau level excitation in gaas.@xcite we note that the absolute separation between oa@xmath22 and ll@xmath21 is actually greater than @xmath24 : at @xmath1 the recombinations are separated by @xmath23 6mev while @xmath24 at this field is approximately 4mev .
this behavior has been observed previously in the ingaas - inp system@xcite and attributed to the fact that for non - zero wave - vectors the excitation energy of the n=1 magneto - plasmon mode actually exceeds @xmath24.@xcite for comparison , we have drawn in figure 3 a line through oa@xmath22 that corresponds to an excitation dispersing with a slope equal to the ideal ( non - interacting ) landau level separation in gaas of 1.65mev / t .
the correspondence is quite good .
the intensity of oa@xmath22 is also quite small ; its peak intensity is only @xmath25 of the peak intensity seen in ll@xmath21 , and between @xmath2 and @xmath3 , oa@xmath22 shows very little intensity variation .
the smallness of the observed intensity in oa@xmath22 and its complete lack of significant variation over this field range clearly identify it as a perturbative shake - up process .
we turn now to oa@xmath21 , the largest and most striking feature in figure 2 . at @xmath1 oa@xmath21
contains nearly 50 percent of the spectral weight seen in the fundamental recombination band , ll@xmath21 . the intensity of oa@xmath21 exhibits complicated magnetic field dependence which is clearly seen in figure 4 . at @xmath2
there is no indication of the oa@xmath21 but at @xmath0 it is clearly developed .
this change happens quite dramatically ; the feature appears with a change of magnetic field of only 0.1 t .
the feature does nt disappear at @xmath26 but rather its intensity is suppressed relative to @xmath0 and it _ blueshifts _ towards ll@xmath21 .
oa@xmath21 rapidly gains spectral weight at @xmath1 , again in a very narrow regime of magnetic field of approximately 0.1 t . for higher magnetic fields the feature
is greatly suppressed and largely gone at @xmath3 .
this behavior is summarized in figure 3 where the magnetic field dependence of all recombinations is displayed .
oa@xmath21 does not move to lower energy for increasing field as one would expect from a simple shake - up process : it shows very little dispersion except for blueshifts seen at @xmath26 and @xmath3 .
the temperature dependence of this novel structure is also quite telling and is displayed in figure 5 .
oa@xmath21 , which appears as broad shoulder to the ll@xmath21 recombination at t=4.2k , does nt change significantly down to temperatures of 2k .
the most dramatic changes occur between 1.5k and 0.53k .
this energy scale of 1k @xmath23 0.08mev is much smaller than any cyclotron gap and indicates the importance of many - body and/or spin effects .
the most striking aspects of oa@xmath21 are its huge intensity at odd filling factor @xmath1 and the complex dependence of its intensity and energy position on magnetic field .
the strong resonance seen at odd filling suggests that the spin of the electron system plays an important role .
our understanding of this doublet structure follows the recently developed model of hawrylak _
et al._@xcite which has been used to explain a similar behavior seen in the data of gravier and co - workers.@xcite this model has the advantage that it is applicable in the limit of low disorder in which both zeeman and cyclotron gaps are resolved and electron - electron interactions are expected to be of prime importance .
the recombination of a conduction - band electron and valence - band hole leaves a hole in the final state of the electron system . in the non - interacting limit
this hole represents a well - defined quasiparticle .
nevertheless , this hole left behind in the final state of the 2deg lies well below the fermi level and constitutes a highly excited state .
if this hole is degenerate with other elementary excitations of the electron gas and electron - electron interactions are strong , the spectral function of the fully interacting hole may not be perturbatively related to the hole in the non - interacting system .
this is the fundamental finding of hawrylak s work : the doublet structure observed in luminescence at odd filling factors is due to the non - quasiparticle behavior of the 2deg hole spectral function .
the two peaks observed in luminescence are shown to be due to a resonant many - body interaction of the hole in the electron gas with a continuum of spin - wave excitations .
thus spin plays a crucial role in the physics of this model ; the splitting of the spectral function only occurs at odd fillings where a continuum of low - lying spin waves is resonant and the spin of the hole can be compensated for by a spin - flip excitation .
hawrylak has deemed this situation an `` anderson - fano '' resonance in which the hole strongly interacting with the continuum of low - lying spin waves is mapped onto the classic solid - state physics problem of a localized state interacting with an unbound continuum .
we believe that such theoretical considerations describe qualitatively our experimental observations of the doublet ll@xmath21 and oa@xmath21 , and clearly distinguish it from the perturbative shake - up process oa@xmath22 .
the hole spectral function has been calculated numerically@xcite and the lower energy peak of the doublet is found to contain 60 percent of the spectral weight of the fundamental recombination band at @xmath1 .
this splitting into a resonant doublet structure is consistent with the experimentally observed behavior .
additionally this theory accounts well for the observed blueshifts and intensity reductions seen in oa@xmath21 at even filling @xmath26 and @xmath3 where the spectral function is expected to collapse to a single peak in the absence of low - lying intra - landau level spin wave excitations .
both experiment and theory point to the important role played by the spin magnetization of 2deg for final state interactions . in summary , we have presented a systematic experimental study of the low energy structure of the photoluminescence spectra from a low disorder 2deg at integral filling @xmath4 .
our findings identify two distinct features occurring below the fundamental band of recombination from the lowest landau level .
the lowest energy satellite is consistent with a perturbative shake - up process in a magnetic field .
the splitting of the fundamental recombination line at odd filling factors is shown to be related to a non - perturbative splitting of the final state spectral function of the 2deg . |
one of the fundamental questions of quantum chaos is why , in the semiclassical limit , almost all classically hyperbolic systems display universal spectral fluctuations , only depending on their symmetries .
this universality was initially conjectured by bohigas , giannoni and schmit @xcite and is by now supported by overwhelming experimental and numerical evidence @xcite .
examples for experimental tests range from nuclear physics and atomic and molecular spectroscopy to classical microwave billiards studied in the limit of large wavenumbers . even though random - matrix theory provides a phenomenological description of these universal features ,
a derivation from first principles is still lacking .
also in other areas of research , such as mesoscopic quantum transport , the reasons for the amazing success of random - matrix theory are only beginning to emerge @xcite . to tackle this long - lasting challenge ,
several approaches have been suggested such as parametric level dynamics @xcite or an extension of field theoretical methods used in the theory of disordered systems @xcite . in the present paper , following an ansatz pioneered in @xcite , we relate quantum spectral statistics to the properties of classical periodic orbits .
we treat the spectral form factor ( _ i.e. _ the fourier transform of the spectral two - point correlation function ) .
according to random - matrix theory , it has the following form for systems belonging to the orthogonal universality class ( _ i.e. _ hyperbolic systems with no symmetries except a time - reversal symmetry whose time - reversal operator squares to unity ) @xmath0 here , @xmath1 is the time measured in units of the heisenberg time @xmath2 ( where @xmath3 is the average level density ) , and we are only considering the range @xmath4 . using gutzwiller s trace formula for the level density @xcite , the form factor can be expressed as a double sum over periodic orbits @xmath5 , @xmath6 @xmath7 where @xmath8 is the classical stability amplitude , @xmath9 the period and @xmath10 the maslov index of the periodic orbit @xmath5 ; the brackets @xmath11 denote averaging over the energy and over a small time window .
note that as the semiclassical limit is taken for fixed @xmath12 , the period of the contributing orbits tends to infinity proportionally to @xmath13 .
the crucial point is that the summand connected to each pair of orbits has a phase given by their action difference divided by @xmath14 .
thus in the semiclassical limit , the phases will be randomly distributed and most terms will interfere destructively . a contribution to the form factor can only arise from pairs of orbits whose action difference is of the order of planck s constant .
thus , there is a deep relation between correlations among the actions of _ classical periodic orbits _ and correlations in _ quantum spectra _ @xcite .
first success in this direction was reported by berry @xcite , who derived the leading term @xmath15 in the series expansion of @xmath16 from pairs of orbits which are either identical or related by time reversal ( diagonal approximation ) . starting with results by argaman _
@xcite , the search was on to identify further families of orbit pairs with similar action , expected to give rise to higher - order contributions to the form factor ( see @xcite and references therein ) .
a breakthrough was recently achieved by sieber and richter , who proposed orbit pairs in which one of the orbits contains a self - crossing in configuration - space with a small angle @xmath17 @xcite .
its partner narrowly avoids that crossing , approximately following one loop of the first orbit , and following the other loop in the time - reversed sense ( cp .
[ fig : pair ] ) .
the action difference between the two is quadratic in @xmath17 and thus can be arbitrarily small . since it is required that the time reversal of a classical orbit loop is again a classical orbit loop , such pairs of orbits can only exist in time - reversal invariant systems . using these pairs ,
sieber and richter derived the leading off - diagonal contribution @xmath18 to the form factor for a uniformly hyperbolic billiard ( a billiard where all orbits have the same lyapunov exponent @xmath19 ) , the so - called hadamard - gutzwiller model , _
i.e. _ geodesic motion on a tesselated surface of constant negative curvature of genus 2 .
they summed over the contributions of all these pairs using two ingredients , the action difference between the two partners and the density of crossing angles .
the @xmath20-term arises due to a correction to the latter of next - to - leading order in the orbit period .
it stems from the fact that in the hadamard - gutzwiller model , a loop with a small crossing angle must have a minimal traversal time @xmath21 . in their derivation
these authors made use of several system - specific niceties of the hadamard - gutzwiller model , which the form factor , being universal , can not depend on .
first steps towards an extension to other systems have been taken for quantum graphs in @xcite , where both the @xmath20- and the @xmath22-contribution to the spectral form factor were shown to originate from similar orbit pairs . in the present paper , extending results we first presented in @xcite , we go beyond these idealized models and derive the leading off - diagonal contribution to the form factor for general two - dimensional hyperbolic systems with a hamiltonian of the form @xmath23 .
the main novel ideas needed to establish this universality are : + ( i ) the relation between the partner orbits can be formulated elegantly in terms of the invariant manifolds , which also determine the action difference within the orbit pair .
+ ( ii ) the maslov indices of the partner orbits can be shown to coincide .
+ ( iii ) in general systems , a logarithmic correction to the angle distribution arises involving the lyapunov exponent of the system ( as defined below ) .
+ ( iv ) in systems with conjugate points , the one - to - one correspondence between crossings and orbit pairs is broken . there are crossings related to almost self - retracing loops without an associated partner orbit , and `` braids '' of crossings with a common partner .
we show how to overcome these problems and reveal the distribution of crossings for which a partner orbit does exist as universal even in the presence of conjugate points .
+ ( v ) the universal contribution to the form factor follows from a relation between the invariant manifolds and the lyapunov exponent .
+ our findings are substantiated by numerical results for two billiard systems .
note that the use of crossings in configuration space constitutes no conceptual problem _
e.g. _ concerning canonical invariance , as what we are using is in fact the geometry of the invariant manifolds .
as pointed out in @xcite , it appears natural to work in configuration space since it is singled out by the conventional time - reversal operator .
interestingly , the results of @xcite imply that our considerations immediately carry over to systems with non - conventional time - reversal invariance , as long as they can be canonically transformed to a hamiltonian of the above structure .
an alternative approach avoiding a projection to configuration space will be presented in @xcite .
this paper is organized as follows .
we first determine the action difference between the two partner orbits ( sect .
2 ) and show that their maslov indices coincide ( sect .
3 ) . in section 4
we investigate the statistics of crossings angles , phase - space locations and loop times in systems without conjugate points . in section 5
we generalize our findings to systems with conjugate points and clarify the relation between crossings and orbit pairs in such systems .
finally , we will derive the @xmath20-contribution to the form factor in section 6 .
a long periodic orbit typically has a huge number of self - crossings in configuration space . for a given small - angle self - crossing
, we will show that a related orbit with similar action exists , provided the two orbit loops separated by the crossing are long .
this partner orbit is obtained from the initial orbit by time reversal of one loop and a local deformation close to the crossing .
further away , the deviation between the initial orbit ( respectively its time - reversed ) and the partner decays exponentially .
as proposed in @xcite , the relation between these two orbits can be expressed elegantly in terms of stable and unstable manifolds .
the main advantage of this method is that the action difference within the orbit pair can be shown to depend only on the crossing angle and the local behavior of the invariant manifolds at the phase - space location of the crossing .
the complicated dependence on stability matrices of the two loops derived in @xcite is thus drastically simplified in the limit of long loops .
recall that two phase - space points lie on each other s stable ( unstable ) manifold if trajectories starting from these points come infinitely close for @xmath24 ( @xmath25 ) . for two - freedom systems , the local behavior of these invariant manifolds at a phase - space point
@xmath26 can be characterized by just one number , called `` direction '' or `` curvature '' of the given manifold @xcite .
this direction is defined as the ratio between the momentum and configuration - space components @xmath27 and @xmath28 of an infinitesimal phase - space deflection along the stable ( unstable ) manifold @xmath29 here , the deflection is assumed orthogonal to the trajectory passing through @xmath26 .
note that time reversal @xmath30 exchanges stable and unstable manifolds and changes the sign of their directions such that @xmath31 .
now , our condition of a local deformation can be formulated in terms of the invariant manifolds .
let @xmath32 and @xmath33 denote the phase - space points corresponding to the two traversals of the crossing .
on the partner orbit with time - reversed right loop , we single out a phase - space point @xmath34 , located where the perpendicular from the crossing hits that orbit ( see fig . [
fig : pair ] ) .
@xmath34 must approximately lie on the stable manifold of @xmath32 ( and vice versa ) , as trajectories starting there approach for a long time as @xmath24 .
conversely , @xmath34 must lie on the unstable manifold of the time reversal of @xmath33 . to determine @xmath34
, it is convenient to work in a poincar section @xmath35 defined by the above perpendicular .
the coordinates of a phase - space point @xmath36 in that section will be denoted by @xmath37 .
since @xmath38 must be stable and @xmath39 unstable , we obtain the following system of linear equations in @xmath40 , @xmath41 involving the directions of the invariant manifolds @xmath42 furthermore , we know that @xmath32 and @xmath33 coincide in configuration space , _ i.e. _ also @xmath43 , and that the difference in orthogonal momenta of @xmath32 and @xmath44 is related to the crossing angle by @xmath45 . using this , the solution to ( [ leq ] ) can be written as @xmath46 @xmath41 follows trivially . and @xmath47 and two points half - way through the loops divide the orbit into four parts @xmath48 , @xmath49 , @xmath50 , and @xmath51 . ]
one can now expand the action of the orbit containing the crossing around that of its partner avoiding it .
as the action of a periodic orbit is stationary , the first - order term vanishes .
we restrict ourselves to the quadratic order in @xmath52 .
we first consider the `` upper '' side of the crossing , _
i.e. _ the orbit parts @xmath48 and @xmath49 separated by @xmath35 and two points half - way through the two loops as shown in figure [ fig : pair ] .
the second derivatives of their action can be related to the stable and unstable directions .
for example , @xmath53 generates the motion from the configuration - space point @xmath54 on @xmath35 to the point @xmath55 half - way through the left loop , thus @xmath56 and therefore @xmath57 .
these derivatives are taken for constant @xmath55 , and can in the limit of long loops be approximated by derivatives along the stable manifold .
we thus have @xmath58 and analogously @xmath59 the same reasoning can be repeated for the `` lower '' side of the crossing in a slightly different poincar section orthogonal to the `` lower '' part of the orbit in a phase - space point @xmath47 ( see fig .
[ fig : pair ] ) .
a taylor expansion now shows that avoiding a crossing with angle @xmath17 located at @xmath26 reduces the action by @xmath60 for small angles , we can neglect the difference of the stable resp .
unstable directions at @xmath32 , @xmath34 , @xmath44 , and @xmath61 ( which we will denote collectively by @xmath26 ) , as long as @xmath62 and @xmath63 are sufficiently smooth close to the crossing location .
this condition is usually fulfilled unless _ e.g. _ immediately after the crossing , the different branches of one loop enclose a singularity of the flow .
apart from these exceptional cases , ( [ actiondiffsing ] ) simplifies to @xmath64 this may also be negative , meaning that the action can also be increased by avoiding the crossing . using that @xmath31 one easily sees that this result is invariant under time reversal of @xmath26 , as it should be , since we must obtain the same action difference if we insert for @xmath26 the other , almost time - reversed traversal of the crossing .
assuringly , the action difference in the hadamard - gutzwiller model follows from ( [ actiondiff ] ) as a special case . in that model
we have @xmath65 ( where @xmath66 is mass ) for all phase - space points @xmath26 , and thus @xmath67 in accordance with @xcite .
the contribution of a pair of orbits to the form factor is determined not only by the difference of their actions , but also by the difference of their maslov indices .
we want to show that for sieber - richter pairs of orbits , at least in the limit of small angles the latter vanishes .
note that this was trivial in the hadamard - gutzwiller model , where all maslov indices are zero .
we work in the framework of a beautiful geometric interpretation of the maslov index of a periodic orbit due to creagh _
@xcite , again in a ( in the present context two - dimensional ) poincar section orthogonal to the orbit . here
, the invariant manifolds locally have the form of lines through the origin which rotate around the origin as we move along the orbit .
the maslov index now equals the net number of clockwise half - rotations of the stable or , equivalently , the unstable manifold ( _ i.e. _ the difference of the numbers of clockwise and counter - clockwise half - rotations ) . for our argument
, we also define the maslov index of a finite non - periodic orbit such as an orbit loop in sieber - richter theory .
it is the sum of the net rotation angles of the stable and the unstable manifold around the origin , divided by @xmath68 .
even though not canonically invariant , non - integer , and depending on the units chosen in the poincar section , this definition is very useful because it makes the maslov index of a loop invariant under time reversal .
obviously , time reversal leaves the absolute value of a rotation angle invariant .
the same is true for the sense of rotation , since time reversal inverts the motion on the poincar section in direction ( turning a clockwise rotation into a counter - clockwise one and vice versa ) , but also changes the sign of the momentum ( turning the sense of rotation back to the original one ) .
an alternative analytic proof can be found in @xcite .
in addition , time reversal exchanges the stable and unstable manifolds , which also can not affect the _ sum _ of their rotation angles . reassuringly , this newly defined maslov index coincides with the usual one in case the orbit happens to be periodic .
moreover , it is additive for subsequent orbit pieces and smooth under small deformations of the orbit as long as the invariant manifolds are smooth .
we can now show that the maslov indices of the two partner orbits coincide .
they can be expressed as sums over the maslov indices of the two loops .
due to additivity and time - reversal invariance of the loop maslov indices , formal time reversal of one loop leaves the maslov index of the orbit invariant . in the limit of small crossing angles , any subsequent local deformation as described in section 2 can at most lead to a small change of the maslov indices of the loops .
since , however , the maslov index of a periodic orbit is an integer quantity , the maslov indices of the two partners have to coincide .
we have seen that in non - uniformly hyperbolic systems the action difference within a sieber - richter pair of orbits depends both on the angle and on the phase - space location of the crossing . anticipating that like in the hadamard - gutzwiller model , also the traversal times of the loops
play a crucial role , we investigate the density @xmath69 of loops with crossing angle @xmath17 , time @xmath70 , and initial phase - space point @xmath26 ( the latter being one of the two traversal points of the crossing ) in a periodic orbit of period @xmath71 .
this density is normalized in such a way that integration over all possible values of @xmath26 , @xmath17 , and @xmath70 yields the average number of loops , and thus twice the average number of crossings , in a periodic orbit of period close to @xmath71 . to make our exposition more clear , in the present section we will still limit ourselves to systems without conjugate points ; our findings will be generalized to systems with conjugate points in section 5 . from the ergodicity of the flow we can , along the lines of @xcite ,
deduce the following approximation @xmath72 where @xmath73 is the volume of the energy shell .
the proportionality to @xmath74 reflects the fact that orthogonal parts of the orbit intersect more frequently than almost parallel or antiparallel ones . however , like in the hadamard - gutzwiller model there is a correction to that ergodic prediction because loop times below a certain angle and location dependent minimum are impossible
. we will first give a general argument why this minimal loop time is generic , and later substantiate our point by numerical evidence for billiards and discuss certain system - specific limitations .
in systems without conjugate points , two trajectories starting from the same point in configuration space with a small opening angle can not gather in another point as long as the phase - space separation between them can be approximated as a linear function of the initial separation .
for an orbit loop to close it is thus necessary that , during half the loop time the separation between one of its two branches and the time - reversed of the other branch grows far enough to make that approximation invalid . apart from exceptional cases to be discussed later
, this typically means that their separation must reach a classical phase - space scale of the order of some finite fraction of the maximal phase - space separation . if the branches enclose a small crossing angle @xmath17 , this requires the loop to be long .
let @xmath75 denote the phase - space separation between one branch of the loop and the time - reversed of the other one in an arbitrary norm @xmath76 at the location of the crossing ( thus @xmath77 ) and @xmath78 the separation after half the loop time .
as long as the linearized approximation is applicable both are for large @xmath70 related by @xmath79 where @xmath19 is the lyapunov exponent of the system .
this asymptotic behavior is due to the fact that long loops ergodically explore the whole energy shell .
the lyapunov exponent governing the asymptotic fate of a small deviation @xmath80 ( with non - vanishing unstable component ) at a phase - space point @xmath26 coincides for almost all @xmath26 , since it can be expressed as the average of the so - called local stretching rate over an infinite trajectory starting at @xmath26 @xcite .
thus , in case of an ergodic flow it coincides with a phase - space average almost everywhere ( except _ e.g. _ on periodic orbits ) .
note that due to pesin s theorem , @xmath19 also coincides with the kolmogorov - sinai entropy @xcite . in uniformly hyperbolic systems
such averaging is trivial since the local stretching rate is constant and therefore even the lyapunov exponents of periodic orbits coincide .
demanding that the deviation @xmath81 reaches the limit for a breakdown of the linear approximation just after half the loop time , we obtain a minimal loop time of @xmath82 here , @xmath83 is constant with respect to the angle but may depend on the phase - space point @xmath26 immediately preceding the loop .
however , the exact value of @xmath83 will turn out to be irrelevant for the form factor .
the second loop starts approximately at @xmath84 , thus its time must fulfill @xmath85 . incorporating these minimal loop times in a straight - forward way , we obtain the following density of loops @xmath86 integration over @xmath87 yields the density of crossing angles and phase - space locations@xmath88 we want to clearly point out the scope of our approximation .
first , the minimal loop time derived above gives the threshold after which loops have a _ chance _ to close , but does not predict the exact behavior of the shortest loops . thus in the immediate vicinity of the minimal loop time , system specific structures appear which are not described by ( [ angledist ] ) and ( [ angle ] ) .
second , we will see that singularities can let the linear approximation for the separation between the branches of a loop break down even before a typical phase - space scale is reached ( as also discussed in @xcite ) .
thus , singularities give rise to exceptional , system - specific crossings which may ignore the minimal loop time @xcite . for our numerics
, we are also interested in the statistics of self - crossings of non - periodic orbits . in full analogy to the above considerations for periodic orbits , we can show that the density of loops with crossing angle @xmath17 , time @xmath70 , and initial phase - space point @xmath26 in non - periodic orbits with traversal time @xmath71 reads @xmath89 integration yields the density of crossing angles and phase - space locations @xmath90 we will now present numerical evidence for this crossing distribution for a special billiard system , the desymmetrized diamond billiard ( see fig . [
fig : diamsing ] ) .
it can be regarded as the empty space between four overlapping disks @xcite la sinai , cut into eight equal pieces .
the distance between the disks is chosen as one , and we choose their radius @xmath91 so that its interior angles become @xmath92 , @xmath93 and @xmath94 .
the desymmetrized diamond billiard is non - uniformly hyperbolic and belongs to the class of semi - dispersing billiards ( _ i.e. _ it is surrounded by a boundary which consists of locally concave and straight segments ) .
therefore , it is free of conjugate points .
it has a circumference @xmath95 and an area @xmath96 .
santal s formula @xcite gives its mean free path as @xmath97 . by averaging over the lyapunov exponents of random non - periodic trajectories ,
we numerically obtain the lyapunov exponent of the system as @xmath98 .
a few words are in order about our numerical technique .
the crossing statistics is determined by averaging over @xmath99 non - periodic trajectories of length @xmath100 ( _ i.e. _ very long orbits compared to the typical length scales @xmath101 and @xmath102 ) with random initial conditions .
for the delicate statistics of angles @xmath103 , even @xmath104 trajectories were considered .
note that for billiards , it is useful to work in dimensionless coordinates with mass and velocity equal to one ; then the traversal time @xmath71 of each orbit equals @xmath105 , and the same holds true for the the traversal times @xmath70 and lengths @xmath106 of the loops .
we now turn to our results . for small angles ,
the basic idea of a minimal loop time depending logarithmitically on the crossing angle can be verified at a glance at figure [ fig : diama]a .
here we have plotted the density of angles and loop times ( up to @xmath107 ) of crossings occurring anywhere on the energy shell trajectories . ] , and for convenience divided out the term @xmath74 arising from the ergodic approximation . sufficiently far above that logarithmic threshold , marked by a dashed line in figure [ fig : diama]a ,
this density is almost uniform . in agreement with ( [ np1 ] )
, it decays linearly towards larger loop time .
note that according to ( [ minloop ] ) , the minimal loop time weakly depends on the location on the energy shell due to the factor @xmath83 .
thus , the minimum in figure [ fig : diama]a is slightly smeared out . below the dashed line ,
the density of crossings diminishes fast before vanishing completely inside the white region . and crossing angles @xmath17 in the desymmetrized diamond billiard :
a ) for @xmath108 , b ) for all angles .
the density is normalized such that multiplied by @xmath109 , it gives the number of crossings in the respective intervals of the crossing angle @xmath17 and the loop time @xmath70 inside one orbit of traversal time @xmath110 ; the resulting scale is shown in c ) . for small angles ,
we observe a threshold logarithmic in @xmath17 , as indicated by a dashed line.,title="fig : " ] and crossing angles @xmath17 in the desymmetrized diamond billiard : a ) for @xmath108 , b ) for all angles .
the density is normalized such that multiplied by @xmath109 , it gives the number of crossings in the respective intervals of the crossing angle @xmath17 and the loop time @xmath70 inside one orbit of traversal time @xmath110 ; the resulting scale is shown in c ) .
for small angles , we observe a threshold logarithmic in @xmath17 , as indicated by a dashed line.,title="fig : " ] and crossing angles @xmath17 in the desymmetrized diamond billiard : a ) for @xmath108 , b ) for all angles .
the density is normalized such that multiplied by @xmath109 , it gives the number of crossings in the respective intervals of the crossing angle @xmath17 and the loop time @xmath70 inside one orbit of traversal time @xmath110 ; the resulting scale is shown in c ) . for small angles
, we observe a threshold logarithmic in @xmath17 , as indicated by a dashed line.,title="fig : " ] in addition to this expected behavior , a rich variety of system - specific structures are seen .
as announced , the crossing density shows system - specific inhomogeneities in the immediate vicinity of the minimal loop time .
in addition , below the minimal loop time , we observe exceptional crossings related to singularities of the flow , most importantly the tangential singularity .
namely , the linear approximation for the separation between the two branches of a loop is already violated if one branch reflects _
e.g. _ at the circular part of the boundary , and the other branch narrowly avoids the circle . in this case , the separation between the branches of the loop need not reach a typical phase space scale for the loop to close .
an example for such exceptional loops ignoring the minimal loop time is shown in figure [ fig : diamsing ] . however , our numerical results indicate that their effect on the crossing distribution is minute . at larger angles , which in the semiclassical limit give no contribution to the form factor ,
further structures are seen ( cp .
[ fig : diama]b ) , such as + ( i ) discrete lines at @xmath111 corresponding to periodic orbits , which are nothing but loops with a `` crossing angle '' @xmath112 .
they are deformed and broadened when going to smaller angles .
thus , most loops close to the minimal time are obtained by deformation of the shortest periodic orbits .
+ ( ii ) four families of loops which start at zero loop time , and as shown in @xcite are close to corners
. one of these families of loops involves only reflections at the two straight - line segments of the boundary and has a crossing angle of @xmath92 for all its members .
note that most of these structures were absent in the idealized example of the hadamard - gutzwiller model . in that model ,
the analog of figure [ fig : diama ] just consists of dispersionless logarithmic curves corresponding each to the family of loops obtained by deformation of one periodic orbit @xcite . further evidence for the minimal loop time is shown in figure [ fig : l10 ] , this time also taking into account the phase - space dependence of the crossing statistics .
we restrict ourselves to crossings which have angles smaller than a maximal angle @xmath113 and take place inside a bin whose area is @xmath114 of the total energy shell , and consider the density of their loop times @xmath70 .
this density must have a gap for small @xmath70 whose width is given by the minimal loop time and thus increases logarithmitically when @xmath113 is reduced .
incidentally , this density of loop times can be shown to decay linearly for large @xmath70 in the case of non - periodic orbits .
our numerical results confirm these predictions and again reveal system - specific structures close to the minimal loop time .
the angle density @xmath115 ( averaged over the whole energy shell and over bins with sizes of @xmath114 and @xmath116 of the energy shell ) agrees with our predictions as well ( cp . fig .
[ fig : angle ] ) .
for sufficiently small angles , the distribution is sinusoidal . upon division by @xmath74 ,
in accordance with ( [ pnp ] ) a logarithmic correction due to the minimal loop time becomes visible .
the relative weight of that correction ( seen as the slope in the logarithmic plot in fig .
[ fig : angle]b ) coincides for all three samples .
we thus see that indeed the crossing distribution in a billiard is homogeneous on the energy shell apart from system - specific oscillations for large @xmath17 and apart from the factor @xmath83 .
based on the relative weight of the logarithm , fitting yields a minimal loop time @xmath117 where @xmath118 agrees perfectly with the lyapunov exponent @xmath19 . in this appendix , we want to show that in the cardioid billiard , the minimal loop time is ignored by a system - specific class of crossings related to the singularity at the cusp .
we will argue that for the evaluation of the form factor , these crossings have to be disregarded as well .
an example for these crossings is shown in figure ( [ fig : plot ] ) .
for one of the two loops separated by the crossing , the two branches stay very near until being reflected on opposite sides of the cusp .
thus , the linear approximation for the separation between these branches breaks down before their separation reaches a typical phase - space scale .
then , in many cases , the loop closes by going around the boundary like in a whispering gallery .
the length ( and thus time ) of the shortest such loops is approximately given by the circumference of the billiard .
in fact , they can be seen as deformations of finite orbits starting and ending at the cusp as introduced in @xcite . in spite of their short time
, the crossing angle can be arbitrarily small .
so for these special loops , there is no minimal time depending logarithmitically on the angle .
however , they do give rise to pairs of orbits with similar action , as their symbol sequences are not time - reversal invariant . nevertheless , we have to disregard these crossings for a semiclassical evaluation of the form factor , as even the validity of the gutzwiller formula for orbits coming so close to the cusp is questionable . in general , each time a periodic orbit comes closer to the boundary of a billiard than some length scale of the order of planck length , so - called penumbra corrections to the trace formula play an important role , as shown in @xcite for dispersing billiards .
the breakdown of the trace formula for families of orbits approaching the cusp of the cardioid is discussed in @xcite . due to these problems
, we have to distinguish in our numerics between generic and cusp - related crossings .
this distinction is done based on symbolic dynamics ( compare fig .
[ fig : plot ] ) .
the loop parts preceding the cusp are almost time - reversed with respect to each other and thus have symbol sequences @xmath124 and @xmath125 .
in contrast , the part following the cusp is almost symmetric with respect to the symmetry line of the cardioid , _
i.e. _ its symbol sequence @xmath126 has to be invariant under inverting the ordering of symbols @xcite .
usually , it undergoes a series of subsequent either clockwise or counter - clockwise reflections at the boundary , _
i.e. _ @xmath126 just consists of several identical symbols @xmath127 or @xmath128 .
in addition , for the symbols between @xmath124 , @xmath126 and @xmath125 one has to take into account the effect of one branch being reflected close to the cusp and the other one narrowly avoiding it .
these symbols can be read off from the example shown in figure [ fig : plot ] .
thus , we see that cusp - related crossings have symbol sequences of the form @xmath129 , @xmath121 , @xmath122 , and @xmath123 , where @xmath124 is arbitrary and @xmath126 invariant under inverting the ordering of symbols .
h. primack , h. schanz , u. smilansky , i. ussishkin , phys .
lett . * 76 * , 1615 ( 1996 ) ; same authors , j. phys .
a : math . gen . * 30 * , 6693 ( 1997 ) ; see also references therein for other semiclassical approaches to diffraction | we reason in support of the universality of quantum spectral fluctuations in chaotic systems , starting from the pioneering work of sieber and richter who expressed the spectral form factor in terms of pairs of periodic orbits with self - crossings and avoided crossings . dropping the restriction to uniformly hyperbolic dynamics , we show that for general hyperbolic two - freedom systems with time - reversal invariance the spectral form factor is faithful to random - matrix theory , up to quadratic order in time .
we relate the action difference within the contributing pairs of orbits to properties of stable and unstable manifolds . in studying the effects of conjugate points , we show that almost self - retracing orbit loops do not contribute to the form factor .
our findings are substantiated by numerical evidence for the concrete example of two billiard systems .
pacs numbers : 05.45.mt , 03.65.sq |
the mid - infrared ( ir ) data for nearby early - type galaxies ( etgs ) become increasingly important because many observational data of etgs show a clear excess over stellar photospheric emission at 8@xmath324 @xmath10 m ( e.g. , knapp et al . 1992 ; athey et al . 2002 ; bressan et al .
2006 ; ko et al .
2009 , 2012 , 2013 ; shim et al . 2011
; hwang et al . 2012 ; martini et al . 2013 ; rampazzo et al . 2013 ) . in old stellar systems ( i.e. , etgs ) ,
the most likely origin of such mid - ir excess is circumstellar dust around asymptotic giant branch ( agb ) stars ( e.g. , bressan et al .
1998 ; piovan et al . 2003 ; villaume et al .
for example , villaume et al . (
2015 ) showed that @xmath11 mid - ir photometric and spectroscopic data for etgs ( from martini et al .
2013 and rampazzo et al .
2013 ) are well explained by their agb dust model in the wavelength range where there are no emission lines .
they also show that dust models without agb dust can not fit to the mid - ir data for etgs detected at 70 and 160 @xmath10 m .
although the detailed contribution of agb dust to the mid - ir spectral energy distribution ( sed ) of a galaxy is not well understood yet ( e.g. , kelson & holden 2010 ; melbourne & boyer 2013 ) , recent models consistently suggest that the mid - ir emission from agb dust could be a proxy for the existence of intermediate - age stars in old stellar systems with little diffuse dust .
this is because the mid - ir sed is very sensitive to the ( even small ) amount of stellar mass formed within @xmath70.3@xmath33 gyr , allowing us to trace recent star formation ( e.g. , see villaume et al .
2015 and references therein )
. however , to better understand the mid - ir excess emission of etgs , it is necessary to quantify the amount of intermediate - age stars in etgs where the mid - ir emission is attributed to agb dust .
the most promising way is to investigate their stellar populations derived from age - sensitive features in optical absorption line spectroscopy ( e.g. , temi et al . 2005 ; bregman et al .
2006 ; conroy et al .
2014 ) .
ko et al . ( 2013 ; hereafter ko13 ) studied the mid - ir properties of local etgs on the optical red sequence .
we found that around half of 648 quiescent red - sequence galaxies have mid - ir excess emission over the stellar component .
if we consider only bright ( @xmath1 @xmath2 @xmath321.5 ) etgs , the fraction of mid - ir excess is still 39% .
we also found that the etgs with mid - ir excess have the strength of the 4000 break slightly smaller than those without mid - ir excess ( i.e. , existence of more young stars ) .
however , we did not find any significant difference in balmer absorption line index h@xmath6 between the two because the signal - to - noise ratio ( s / n ) of each individual sdss spectrum is not high enough to examine the difference .
here we use the etg sample of ko13 to quantify the amount of intermediate - age stars ; we perform a stellar population analysis of the stacked sdss spectra of these etgs . we focus on the mid - ir excess of etgs and quantify the amount of intermediate - age stars through a stellar population analysis of the sdss stacked spectra .
if the mid - ir emission of etgs is produced by intermediate - age stars formed within the last 1@xmath35 gyr , the absorption features in the optical spectra can provide important information on the age of their stars .
section 2 describes the observational data we use . in section 3
we examine spectral features of the etgs with mid - ir excess emission , and conclude in section 4 . throughout
, we use the ab magnitude system , and adopt flat @xmath12cdm cosmological parameters : @xmath13 km s@xmath14 mpc@xmath14 , @xmath15 and @xmath16 .
we use the observational data and the etg sample constructed in ko13 . here
we give a brief summary of the data and galaxy sample . in a spectroscopic sample of galaxies in the sdss data release 7 ( dr7 ; abazajian et al .
2009 ) , we first identified etgs using the korea institute for advanced study ( kias ) value - added galaxy catalog that provides galaxy morphological information ( choi et al .
2010 ) . to identify optically quiescent etgs , we plot these etgs in the color - magnitude diagram ( cmd ) , and select the galaxies in the red sequence
( see figure 4 in ko13 ) .
we then removed the galaxies with h@xmath0 equivalent width @xmath2 @xmath31 to select only `` quiescent '' galaxies .
we further removed the galaxies with active galactic nuclei ( agns ) because they can produce mid - ir emission .
most agns are removed from the h@xmath0 equivalent width cut .
the remaining agns are removed using the spectral types determined by the criteria of kewley et al .
we also used the _ wise _ color - color selection criteria of jarrett et al .
( 2011 ) to remove dusty agns .
we also excluded the highly inclined ( @xmath17-band isophotal axis ratio @xmath18 less than 0.6 ; see choi et al .
2007 ) galaxies to reject dust - reddened star - forming galaxies . in the result
, we obtain a sample of 397 bright ( @xmath1 @xmath2 @xmath321.5 ) , quiescent ( without h@xmath0 emission ) , red etgs without agns ( see section 2.2 in ko13 for details ) .
we compile the multi wavelength data for these galaxies including _ wise _ mid - ir and _ galex _ near - uv for the following analysis ( hwang et al .
2010 ) .
@xmath3[12 ] vs. nuv@xmath8 colors of bright ( @xmath1 @xmath2 @xmath321.5 ) , quiescent ( without h@xmath0 emission ) etgs .
circles with different colors indicate different subclasses .
asterisks represent bright e+a galaxies with early - type morphology from choi et al .
for reference , the extinction - free colors of models ( which have different prescription of agb dust ) are overplotted : the swire templates of polletta et al .
( 2007 ) including three ellipticals ( 2 , 5 , and 13 gyr ) , five spirals ( s0 , sa , sb , sc , sd ) , and two starbursts ( m82 and arp220 ) ; and single stellar population templates of piovan et al .
( 2003 ) along light - weighted mean stellar ages ( 2 , 5 , and 13 gyr ) for 40% solar metallicity ( z=0.008 ) .
the vertical and horizontal dashed lines indicate the mid - ir color threshold of [ 3.4]@xmath3[12 ] = @xmath31.3 and the near - uv color threshold of nuv@xmath8 = 5.4 , respectively .
the arrow shows the mean value of a reddening vector of e(b - v ) = 0.1 for the swire templates , and the cross in the the lower left corner indicates median color errors.,width=529 ] figure 1 shows the distribution of [ 3.4]@xmath3[12 ] and nuv@xmath8 colors for 397 bright etgs .
they show a wide range of colors in the near - uv and mid - ir .
we classify these etgs into three subclasses based on nuv@xmath8 and [ 3.4]@xmath3[12 ] colors .
we adopt the cut of [ 3.4]@xmath3[12 ] = @xmath31.3 ( ko13 ) ; galaxies redder than this cut are considered to be relatively young ( @xmath19 5 gyr ) in the mid - ir weighted mean stellar ages . because we have selected the galaxies without h@xmath0 emission , agns , and highly inclined disks , the mid - ir emission is dominated by the circumstellar dust around agb stars .
the nuv@xmath8 cut is nuv@xmath8 = 5.4 ( schawinski et al . 2007 ) ; galaxies bluer than this threshold are considered to have @xmath201% of young ( @xmath19 1 gyr ) populations ( see fig . 6 in ko13 ) .
the three subclasses can be summarised as follows .
@xmath21 etgs with near - uv excess and mid - ir excess ( ` neme ' in fig .
1 ) : nuv@xmath8 @xmath2 5.4 and [ 3.4]@xmath3[12 ] @xmath22 @xmath31.3 . these galaxies have both mid - ir and near - uv excess emissions .
@xmath21 etgs without near - uv excess but with mid - ir excess ( ` nnme ' in fig.1 ) : nuv@xmath8 @xmath22 5.4 and [ 3.4]@xmath3[12 ] @xmath22 @xmath31.3 .
these galaxies have only mid - ir excess emission over the stellar component .
there are five etgs with near - uv excess and without mid - ir excess , but we do not analyze them because of small number statistics .
three of them are close to the boundaries , suggesting that they may belong to other subclasses if we consider their errors in colors .
@xmath21 etgs without near - uv excess and mid - ir excess ( ` nnmn ' in fig .
1 ) : nuv@xmath8 @xmath22 5.4 and [ 3.4]@xmath3[12 ] @xmath2 @xmath31.3 . these galaxies are completely quiescent in the optical , near - uv and mid - ir .
note that the galaxies with [ 3.4]@xmath3[12 ] @xmath2 @xmath31.7 and nuv@xmath3r @xmath22 5.7 are undetected at 12 @xmath10 m and at nuv , respectively .
therefore , most galaxies in this subclass are indeed quiescent . for comparison , we also show e+a galaxies from choi et al .
( 2009 ) with similar absolute magnitudes ( @xmath1 @xmath2 @xmath321.5 ) and morphology ( i.e. etgs ) in figure 1 .
e+as are post - starburst ( within @xmath71 gyr ) systems characterized by strong balmer absorption lines and weak h@xmath0 emission line ( e.g. , yang et al . 2008 ) .
most e+a galaxies are located in the neme region ( except two galaxies that have only mid - ir excess emission ) .
however , e+as have much bluer nuv@xmath8 and redder [ 3.4]@xmath3[12 ] colors than neme galaxies . because e+as are not red - sequence galaxies
, this means that they are not in the same evolutionary stage as neme galaxies . in other words , e+as have much more young stars than neme galaxies , thus e+as have mid - ir weighted mean stellar ages much younger than neme galaxies .
here we use the sdss optical spectra of the etgs in the three subclasses for a stellar population analysis . to increase the s / n of the spectra , we stack the galaxy spectra in each subclass . for each subclass of etgs and e+as
, we stack the sdss spectra .
we normalize the individual spectrum at rest - frame 4150@xmath34250 and take the median for the stacking .
the wavelength coverage for the stacked spectra is 3800@xmath37650 .
the s / n of the stacked spectra is about 71 - 281 at 4092@xmath34111 ( around the h@xmath23 absorption passband ) , an order of magnitude higher than the typical s / n of individual spectrum ( @xmath711 ) .
we measure the strength of 4000 break , @xmath54000 , and lick indices of balmer absorption lines .
we adopt the bandpass of @xmath54000 defined by balogh et al .
( 1999 ) , and the lick indices defined by worthey & ottaviani ( 1997 ) .
we adopt the continuum bandpasses of moustakas & kennicutt ( 2006 , table 9 therein ) for balmer absorption lines .
figure 2 shows the stacked spectra of three subclasses and e+as in the wavelength range 3700@xmath35000 .
the stacked spectra of three subclasses show that all of them are typical quiescent systems with strong 4000 break , but the stacked spectrum of e+as shows different features including a strong h@xmath23 absorption line .
the strength of 4000 break correlates well with stellar ages , and strong h@xmath23 absorption lines are attributed to recent ( 0.1 @xmath3 1 gyr ) bursts of star formation ( bruzual & charlot 2003 ; bc03 ) .
kauffmann et al .
( 2003 ) showed that the combination of @xmath54000 and h@xmath6 is a good indicator of star formation activity of galaxies over the past 1@xmath32 gyr with little metallicity effect . in figure 3
, we show the @xmath54000 versus h@xmath6 for the bc03 simple stellar population ( ssp ) models with a solar metallicity and the star formation history of sfr(@xmath24 ) @xmath25 exp(@xmath26/@xmath27 ) .
we use a formation time from 10 gyr to 0.1 gyr , and a prompt star formation history of @xmath27 = 0.5 gyr that is generally expected for sdss etgs ( e.g. , thomas et al .
2010 ; li et al . 2015 ) .
e+as lie in a region , clearly different from the region of etgs where old populations are dominant with no significant recent bursts .
the etgs with mid - ir excess ( neme and nnme ) appear younger than the etgs without mid - ir excess ( nnmn ) .
this is consistent with our previous results that the mid - ir flux is sensitive to young and intermediate - age ( @xmath281 gyr ) populations .
figure 3 also shows a hint of slightly younger ages of neme galaxies than nnme galaxies , consistent with the difference in the mid - ir weight mean stellar age in figure 1 .
this means that galaxies with near - uv excess at a given mid - ir flux have more young ( @xmath91 gyr ) stars than those without near - uv excess , if the dust attenuation effects are negligible ( see figure 6 of ko13 ) .
this can suggest that neme galaxies evolve into nnme galaxies within @xmath71 gyr , consistent with previous studies ; villaume et al .
( 2015 ) show that the mid - ir light of etgs is enhanced by a factor of 3 because of stars formed within @xmath70.3@xmath33 gyr , while the near - uv light of etgs remains bright only for @xmath71.5 gyr
( yi et al . 2005 ) . vs. @xmath54000 for the sdss stacked spectra of three subclasses of etgs and e+as ( large symbols ) .
colors are the same as in figure 1 .
gray filled and open symbols represent individual spectra for three subclasses and e+as , respectively .
the large cross indicates median errors of individual spectrum .
the solid line represents a model of continuous star formation histories with a solar metallicity , a fixed @xmath29 gyr , and different formation time ( 10 gyr to 0.1 gyr ) .
we indicate two formation ages of 5 and 10 gyrs.,width=529 ] from a two - component ssp analysis of etgs in ko13 , we found that the small amount of young ( @xmath91 gyr ) populations can make etgs a significant change in [ 3.4]@xmath3[12 ] colors of etgs , similar to the case of nuv@xmath8 colors ( see figure 6 of ko13 ) .
the ssps with agb dust further show that only the mid - ir data can trace star formation longer than @xmath201 gyr even though both near - uv and mid - ir are sensitive to star formation over the past @xmath71 gyr . to study correlations between optical spectral features and near - uv / mid - ir data as recent star formation indicators , we plot @xmath54000 , h@xmath6 , and @xmath30h@xmath31 as a function of [ 3.4]@xmath3[12 ] and nuv@xmath8 colors of etgs in figure 4
; @xmath30h@xmath31 is the average of three lick indices of balmer absorption lines ( h@xmath32 , h@xmath33 and h@xmath34 ) .
we measure the spectral features of both stacked spectra ( large color circles ) and individual spectrum ( gray circles ) .
the spearmann rank correlation test for the data from the individual spectrum suggests that the correlation between the optical features and [ 3.4]@xmath3[12 ] color is stronger than that between the optical features and nuv@xmath8 colors .
interestingly , [ 3.4]@xmath3[12 ] colors of nnme galaxies differ from those of other subclasses ( i.e. , nnme galaxies show redder colors than nnmn galaxies , but show bluer colors than neme galaxies ) , but nuv@xmath8 colors of nnme galaxies are similar to those of nnmn galaxies .
these results support that the mid - ir data probe the star formation activity of etgs within the last 1@xmath32 gyrs better than the near - uv data .
therefore , many etgs without young ( @xmath19 1 gyr ) stars ( i.e. , nnme ) seem to have excess emission in the mid - ir because of their intermediate - age ( @xmath20 1 gyr ) stars or / and to have no near - uv excess emission because of the small fraction of young ( @xmath19 1 gyr ) stars if the dust attenuation effects are negligible . for each stacked spectrum
, we perform a decomposition of stellar populations using the spectral fitting code starlight ( cid fernandes et al .
2005 ) in the wavelength range 3800@xmath37650 .
we first use 45 templates of bc03 model covering 15 ages ( between 1 myr and 13 gyr ) and 3 metallicities ( [email protected] , 0.02 , 0.05 ) , generated from stelib library ( le borgne et al .
2003 ) with padova ( 1994 ) evolutionary tracks and chabrier ( 2003 ) initial mass function ( imf ) .
figure 5 shows the result of the starlight fit to the stacked spectra of the three subclasses and e+as .
we perform the fit with 100 different seeds for generating random numbers ( see the starlight manual for details ) , and adopt the median of the results from the 100 fits .
as expected , all etgs are dominated by a @xmath710 gyr stellar population .
the mass fraction of young ( @xmath91 gyr ) and intermediate - age ( @xmath71@xmath35 gyr ) populations for the etgs with mid - ir excess ( i.e. , nnme and neme ) is small ( @xmath74@xmath35% ) , but is not negligible compared to the etgs without mid - ir excess ( nnmn ) . e+as
have young ( @xmath71 gyr ) populations , much more than the etgs with mid - ir excess ( by a factor of seven ) .
this amount of young component makes the e+as not in the red sequence of the optical cmd .
we note that the light fraction of @xmath71 gyr population for the etgs with mid - ir / near - uv excess reaches up to @xmath720% , which explains why we could easily identify the etgs with recent star formation in the mid - ir / near - uv color space .
actually , the two - component ssp analysis of etgs ( ko et al .
2013 ) showed that only @xmath75% of 1 gyr population can make significant changes in nuv@xmath8 and [ 3.4]@xmath3[12 ] colors , thus we can easily distinguish them from the etgs without young populations .
distributions from the 100 fits with different seeds .
, width=566 ] to examine the dependence of the fit on the choice of spectral templates , we fit to the spectra with various combinations of age / metallicity distributions ( i.e. , different star formation history ) , stellar population models ( e.g. , bc03 and charlot & bruzual , in preparation , hereafter cb07 ) , and imfs ( e.g. , chabrier 2003 and salpeter 1955 ) . figure 6 shows the resulting mass fractions from the fits with various age and metallicity distributions of the templates .
the plot shows that the mass fraction of young and intermediate - age stars does not change much with different combinations of age and metallicity distributions .
only when we use the templates with two metallicities without the sub - solar one ( see 2nd column in figure 6 ) , the mass fraction of young and intermediate stars is about twice higher than the other cases .
figure 7 also shows the mass fractions of young and intermediate - age stars from the fits , but with different stellar population models ( bc03 or cb07 ) and imfs ( chabrier or salpeter ) .
we fix the age and metallicity distributions of the templates to the ones in figure 5 .
the plot again shows that the mass fractions of young and intermediate - age stars do not differ much ; they are in the range @xmath74@xmath311% .
although the optical spectra of the etgs in our sample do not show any features of star - forming galaxies , there could still be a contribution to the mid - ir excess emissions from deeply embedded sources ( e.g. , small , dense , central disks ) .
however , among the 145 etgs with mid - ir excess emission in our sample , 122 galaxies ( 84% ) are not detected at 22 @xmath10 m ( i.e. , signal - to - noise ratio at 22 @xmath10 m @xmath2 3 ) ; the remaining 23 etgs detected at 22 @xmath10 m have small 22 @xmath10 m flux densities ( see figure 7 of ko et al . 2013 ) .
this suggests that most etgs with mid - ir excess emission are not dusty , star - forming systems . instead , their mid - ir excess emission results probably from the existence of young / intermediate age stars .
interestingly , the fit to the stacked spectra shows that the stellar populations of etgs with both mid - ir and near - uv excess ( neme ) do not significantly differ from those with only mid - ir excess ( nnme ) .
the difference in intrinsic reddening between neme and nnme is also negligible ( mean a@xmath36 values derived from the 100 fits are 0.190 @xmath37 0.017 and 0.192 @xmath37 0.019 , respectively ) , suggesting that nnme is not a simply dust - reddened neme .
there could be several explanations why neme galaxies have stronger nuv emission than nnme galaxies even though the stellar populations from the spectra show no significant difference between the two .
first , if recent star formation occurs @xmath201 gyr ago in the central regions of galaxies ( within the sdss fiber 3@xmath38 , corresponding 2 kpc in radius at @xmath4 = 0.07 ) and the dust attenuation effects are negligible , the existence of extended uv structures ( e.g. , salim et al .
2009 ) in the outer regions could make the overall nuv@xmath8 color bluer ( assuming an inside - out behavior of recent star formation ) . in this case
, galaxies with strong uv emission in the outer regions could be detected as neme galaxies , but those with weak uv emission in the outer regions would be classified as nnme galaxies ( e.g. , partial recent star formation ; jeong et al . 2009 ) . however , the spectral features of neme and nnme galaxies could be similar because the optical spectra probe only the central regions where recent star formation occurs around 1 gyr ago in both populations . the reason why neme and nnme galaxies have different uv emissions in the outer regions could be because nnme galaxies are more evolved systems than neme galaxies . in other words ,
neme galaxies have bluer nuv@xmath8 colors because of recent star formation occurred within @xmath71 gyr in the outer regions ; nnme galaxies have relatively red nuv@xmath8 colors because near - uv excess fades with time ( i.e. , more than @xmath71 gyr after star formation stops ) , but they still show mid - ir excess because of intermediate - age stars .
second , if the dust attenuation effects are not negligible ( particularly for nnme ) or / and dust geometry is different between the two , the different nuv@xmath8 colors can result from localized dust absorption . in this case ,
nnme galaxies probably contain dusty regions mixed with young stars .
then their overall nuv@xmath8 colors can not be bluer because of dust attenuation even if the amount of young stars in nnme galaxies is similar to that in neme galaxies .
this interpretation is supported by the fact that 70% of 22@xmath10m - detected etgs with mid - ir excess are nnme galaxies ( i.e. , 16 out of 23 ) .
we study the optical spectral features of bright ( @xmath1 @xmath2 @xmath321.5 ) etgs in the local universe that are on the optical red sequence but have different amounts of near - uv and mid - ir excess emissions . to quantify the amount of young and intermediate - age stars in those etgs
, we perform a stellar population analysis using the sdss stacked spectra .
we find that the mid - ir colors ( [ 3.4]@xmath3[12 ] ) correlate well with the strength of @xmath54000 and with the balmer absorption line h@xmath6 that are good tracers of star formation activity over the past 12 gyr .
these suggest that the mid - ir emission can be a useful diagnostic tool for probing young and intermediate - age stars in etgs .
actually there are various mechanisms that could contribute to the mid - ir emission in galaxies ( draine & li 2007 ; da cunha et al .
this includes the sf driven dust continuum and pah emissions ( smith et al . 2007 ;
rieke et al .
2009 ) , agn heated hot dust continuum ( netzer et al .
2007 ; mullaney et al . 2011 ) , and agb dust emission ( bressan et al . 1998 ; piovan et al . 2003 ) . because we removed the galaxies with star formation and nuclear activity from our sample ( see section 2.1 ) ,
the contribution from the star formation and nuclear activity to the mid - ir emission is negligible in our sample .
this is also supported by that the majority ( i.e. , 122 out of 145 ) of the etgs with mid - ir excess emission are not detected at 22 @xmath10 m ( see section 3.3 ) .
the fit to the stacked spectra with a spectral fitting code , starlight , shows that the mass fraction of @xmath71@xmath35 gyr stars in the etgs with mid - ir excess is @xmath74@xmath311% and the light fraction is up to @xmath39 .
it should be noted that the mass fraction of @xmath71@xmath35 gyr stars could be smaller than the current estimates ( i.e. , 4@xmath311% ) if the metallicity of etgs is very low ( e.g. , [email protected] ) .
it is because the contribution of intermediate - age ( i.e. , 1@xmath35 gyr ) agb stars to the mid - ir emission becomes small in a very low metallicity tail ( e.g. , villaume et al .
in such cases , the mid - ir emission may not be a useful tool for probing recent star formation . however , this is not a general case for nearby , massive etgs including our sample that typically have metallicities close to solar value ( gallazzi et al .
these results support the idea that the etgs with mid - ir excess emission have experienced star formation within the last 1@xmath35 gyr .
we thank the anonymous referee for useful comments that greatly improved this paper .
j.k . and j.c.l .
are the members of dedicated researchers for extragalactic astronomy ( dream ) in korea astronomy and space science institute ( kasi ) .
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 .
funding for sdss - iii has been provided by the alfred p. sloan foundation , the participating institutions , the national science foundation , and the u.s .
department of energy office of science .
the sdss - iii web site is http://www.sdss3.org/. sdss - iii is managed by the astrophysical research consortium for the participating institutions of the sdss - iii collaboration including the university of arizona , the brazilian participation group , brookhaven national laboratory , carnegie mellon university , university of florida , the french participation group , the german participation group , harvard university , the instituto de astrofisica de canarias , the michigan state / notre dame / jina participation group , johns hopkins university , lawrence berkeley national laboratory , max planck institute for astrophysics , max planck institute for extraterrestrial physics , new mexico state university , new york university , ohio state university , pennsylvania state university , university of portsmouth , princeton university , the spanish participation group , university of tokyo , university of utah , vanderbilt university , university of virginia , university of washington , and yale university .
abazajian , k. n. , adelman - mccarthy , j. k. , ageros , m. a. , et al .
2009 , , 182 , 543 athey , a. , bregman , j. , bregman , j. , temi , p. , & sauvage , m. 2002 , , 571 , 272 balogh , m. l. , morris , s. l. , yee , h. k. c. , carlberg , r. g. , & ellingson , e. 1999 , , 527 , 54 bregman , j. n. , temi , p. , & bregman , j. d. 2006 , , 647 , 265 bressan , a. , granato , g. l. , & silva , l. 1998 , , 332 , 135 bressan , a. , panuzzo , p. , buson , l. , et al .
2006 , , 639 , l55 bruzual , g. , & charlot , s. 2003 , , 344 , 1000 chabrier , g. 2003 , , 115 , 763 choi , y. , goto , t. , & yoon , s .- j . 2009 , , 395 , 637 choi , y .- y . ,
han , d .- h .
, & kim , s. s. 2010 , journal of korean astronomical society , 43 , 191 choi , y .- y . ,
park , c. , & vogeley , m. s. 2007 , , 658 , 884 cid fernandes , r. , mateus , a. , sodr , l. , stasiska , g. , & gomes , j. m. 2005 , , 358 , 363 conroy , c. , graves , g. j. , & van dokkum , p. g. 2014 , , 780 , 33 da cunha , e. , charlot , s. , & elbaz , d. 2008 , , 388 , 1595 draine , b. t. , & li , a. 2007 , , 657 , 810 gallazzi , a. , charlot , s. , brinchmann , j. , & white , s. d. m. 2006 , , 370 , 1106 hwang , h. s. , elbaz , d. , lee , j. c. , et al .
2010 , , 522 , a33 hwang , h. s. , geller , m. j. , diaferio , a. , & rines , k. j. 2012 , , 752 , 64 jarrett , t. h. , cohen , m. , masci , f. , et al .
2011 , , 735 , 112 jeong , h. , yi , s. k. , bureau , m. , et al .
2009 , , 398 , 2028 kauffmann , g. , heckman , t. m. , white , s. d. m. , et al .
2003 , , 341 , 33 kelson , d. d. , & holden , b. p. 2010
, , 713 , l28 kewley , l. j. , groves , b. , kauffmann , g. , & heckman , t. 2006 , , 372 , 961 knapp , g. r. , gunn , j. e. , & wynn - williams , c. g. 1992 , , 399 , 76 ko , j. , i m , m. , lee , h. m. , et al .
2009 , , 695 , l198 ko , j. , i m , m. , lee , h. m. , et al .
2012 , , 745 , 181 ko , j. , hwang , h. s. , lee , j. c. , & sohn , y .- j .
2013 , , 767 , 90 le borgne , j .- f . ,
bruzual , g. , pell , r. , et al .
2003 , , 402 , 433 li , c. , wang , e. , lin , l. , et al . 2015 , , 804 , 125 martini , p. , dicken , d. , & storchi - bergmann , t. 2013 , , 766 , 121 marigo , p. , & girardi , l. 2007 , , 469 , 239 melbourne , j. , & boyer , m. l. 2013 , , 764 , 30 moustakas , j. , & kennicutt , r. c. , jr .
2006 , , 164 , 81 mullaney , j. r. , alexander , d. m. , goulding , a. d. , & hickox , r. c. 2011 , , 414 , 1082 netzer , h. , lutz , d. , schweitzer , m. , et al .
2007 , , 666 , 806 piovan , l. , tantalo , r. , & chiosi , c. 2003 , , 408 , 559 polletta , m. , tajer , m. , maraschi , l. , et al .
2007 , , 663 , 81 rampazzo , r. , panuzzo , p. , vega , o. , et al .
2013 , , 432 , 374 rieke , g. h. , alonso - herrero , a. , weiner , b. j. , et al .
2009 , , 692 , 556 salim , s. , & rich , r. m. 2010 , , 714 , l290 salpeter , e. e. 1955 , , 121 , 161 schawinski , k. , kaviraj , s. , khochfar , s. , et al .
2007 , , 173 , 512 shim , h. , i m , m. , lee , h. m. , et al .
2011 , , 727 , 14 smith , j. d. t. , draine , b. t. , dale , d. a. , et al .
2007 , , 656 , 770 temi , p. , mathews , w. g. , & brighenti , f. 2005 , , 622 , 235 thomas , d. , maraston , c. , schawinski , k. , sarzi , m. , & silk , j. 2010 , , 404 , 1775 villaume , a. , conroy , c. , & johnson , b. d. 2015 , , 806 , 82 worthey , g. , & ottaviani , d. l. 1997 , , 111 , 377 yang , y. , zabludoff , a. i. , zaritsky , d. , & mihos , j. c. 2008 , , 688 , 945 yi , s. k. , yoon , s .- j . , kaviraj , s. , et al .
2005 , , 619 , l111 | we present a stellar population analysis of quiescent ( without h@xmath0 emission ) and bright ( @xmath1 @xmath2 @xmath321.5 ) early - type galaxies ( etgs ) with recent star formation .
the etgs are selected from a spectroscopic sample of sdss galaxies at 0.04 @xmath2 @xmath4 @xmath2 0.11 with _
wise _ mid - infrared ( ir ) and _ galex _ near - ultraviolet ( uv ) emissions .
we stack the optical spectra of etgs with different amounts of mid - ir and near - uv excess emissions to measure the strength of 4000 break @xmath54000 and the width of balmer absorption line h@xmath6 that are indicative of recent ( @xmath71 gyr ) star formation activity .
the _ wise _ [ 3.4]@xmath3[12 ] colors show stronger correlations with the spectral features than nuv@xmath8 colors .
we fit to the stacked spectra with a spectral fitting code , starlight , and find that the mass fraction of young ( @xmath91 gyr ) and intermediate - age ( @xmath71@xmath35 gyr ) stars in the etgs with mid - ir excess emission is @xmath74@xmath311% , depending on the template spectrum used for the fit .
these results show that the etgs with mid - ir excess emission have experienced star formation within the last 1@xmath35 gyr and that the mid - ir emission is a useful diagnostic tool for probing recent star formation activity in etgs . |
sunspots are regions of concentrated strong magnetic fields comprising of dark central region , umbra surrounded by a less darker region called the penumbra .
penumbral microjets ( pmjs ) are one of the prominent fine - structure dynamical features observed in the sunspot at the chromospheric height . using the sot @xcite h filter images , @xcite first reported that these micro - jets occur ubiquitously above penumbra .
they have lengths of 1 to 4 mm and lifetimes of up to a minute .
they generally move very fast and have apparent speeds over 100 km s@xmath0 .
the magnetic field in the chromospheric penumbral region consists of a combination of spines ( more vertical field ) and interspines ( more horizontal field ) .
@xcite found that pmjs generally originate near the bright structures in between two dark penumbral filaments and propagate upward along the direction of the spine .
so , they proposed that pmjs could originate as a result of magnetic reconnection between the spine and interspine magnetic fields . using the iris @xcite observations
, @xcite found the presence of subarcsecond bright dots ( bds ) above sunspot penumbra in the tr .
they appear ubiquitously and mostly have a lifetime less than few minutes .
they sometimes appear slightly elongated along the penumbral filaments and also move along the filaments with speeds of 1040 km s@xmath0 .
they proposed that some of them could be due to impulsive reconnection in the tr and chromosphere at footpoints of coronal magnetic loops and others are probably due to falling plasma .
it is still unclear how bds are formed and if they show any signatures in the lower and upper atmosphere .
@xcite studied the multi - wavelength signatures of pmjs and found that pmjs show spatial offset from the chromosphere to tr in the direction of the pmj .
hence , they proposed that pmjs may progressively heat up to tr temperature .
@xcite could not find noticeable signature of normal pmjs in any atmospheric imaging assembly ( aia ) @xcite passbands , although a few strong and large pmjs show bd / bd - like signatures in the 1600 and high - resolution coronal imager ( hi - c ) 193 images .
@xcite found that bds observed in hi - c are on average slower , dimmer , larger in size and longer lived than iris penumbral bds .
they also found that most of the bds observed in hi - c 193 may correspond to tr .
@xcite found that the locations of most of the penumbral bds show downflow .
their statistical analysis shows that bds do not have consistent brightening response in the chromosphere .
following @xcite , they also suggested that tr penumbral bds are manifestation of falling plasma from coronal heights along more vertical and dense magnetic loops or small - scale impulsive magnetic reconnection at tr or higher heights . @xcite and observed heating events which are associated with the strong downflows in the tr and proposed a similar scenario . @xcite
have studied small transient brightening events in the penumbra and found that it show redshifts in the 1402.77 line , an inward motion towards the umbra in iris 1400 images and have a multi - thermal component .
they proposed the triggering mechanism as magnetic reconnection at low coronal heights .
a component of the plasma from the reconnection site may move downward and reach the tr , which is confirmed by the observed redshifts in the 1402.77 line and the inward motions as seen in the iris 1400 images .
finally , it reaches the chromosphere and appears as ribbon - like brightening . in this work , while combining multi - wavelength observations covering chromosphere , tr / corona we try to find the source of the bds and their relation to pmjs .
we use the data obtained from a coordinated observation taken on 19 march 2014 , using the hinode / sot ( hop-250 ) @xcite , iris ( iris-3840007453 ) @xcite and the aia @xcite on board the solar dynamics observatory .
the coordinated observation between all the instruments was performed from 08:51 ut to 09:46 ut .
we selected a fov of @xmath1x@xmath2 centered at @xmath3 and @xmath4 .
the fov is limited due to the sot observations .
iris slit - jaw images ( sji ) centered at 2796 and 1400 are dominated by the k and emission lines , respectively .
the sot filter is dominated by h line , which forms at the lower chromosphere where the temperature is below 10@xmath5 k. the iris 2796 images are dominated by the emission from a plasma at temperatures @xmath610,000 - 15,000 k and represent the middle chromosphere .
the iris 1400 passband is sensitive to tr @xmath660,000 - 80,000 k. aia filtergram images centered at 304 ( sensitive to @xmath60.05 mk ) and 193 ( @xmath61.25 mk ) are dominated by and emission lines , respectively .
sot h data was taken with 0.3 sec exposure and 1.58 sec cadence .
iris sji s were obtained with 4 sec exposure and a cadence of 10.5 sec .
aia 304 and 193 observations were taken with 2 sec exposure and 12 sec cadence .
aia data were then co - aligned and de - rotated to the start time ( 08:51 ut ) of observation to compensate for the solar rotation .
the pixel size of sot , sji s and aia are @xmath7 , @xmath8 and @xmath9 , respectively .
we have interpolated sot and aia data in space and time to match the iris sji cadence ( 10.50 sec ) and spatial resolution ( @xmath10 ) .
the iris sjis and sot images were co - aligned using iris 2796 and sot h images .
the time difference between the sot and iris image is of the order of sub - seconds .
after that iris and aia were co - aligned using iris 1400 and aia 1600 images @xcite .
figure [ sot_iris_aia ] shows the image of a sunspot , where the bottom rows ( g - k ) correspond to running difference images .
the yellow contours show the location of bds in the sunspot penumbra .
these contours are obtained from the 1400 image after subtracting the same image with a 8x8 pixel smoothing .
evolution of two bds inside the box b1 and b2 ( as marked in 1400 ) are shown in the figure [ ev_b1ab2 ] ( also in online movie ) .
the bds inside the box b1 including box b11 , b12 and b13 are related to pmjs .
the bd inside b2 is not related to any visible dynamics in the sot .
figure [ ev_b1ab2 ] depicts the temporal evolution of two bds .
shows the evolution of the bd inside the b1 box in the figure [ sot_iris_aia ] .
panel a , c and e show the evolution as seen in the iris 1400 , sot h and iris 2796 , whereas the b , d and f correspond to running difference images
. the clear presence of a bd is seen in the in the 1400 .
a jet like feature is also seen in the h and 2796 images .
similarly , show the evolution of the b2 region as marked in the figure [ sot_iris_aia ] .
here , we can clearly find the presence of a bd in the 1400 channel but no clear signature of intensity enhancement in the h. we have identified several bds and performed a statistical analysis to compile the properties of these bds and their signatures in the other channels . at first , we identify manually an isolated bd and look for the time frame where it shows maximum intensity enhancement .
we use that as our central frame and make a movie ( see animated figure of fig .
[ ev_b1ab2 ] , covering a 5x5 mm region for a duration of 210 sec ) to follow the bds and their direction of propagation .
the bds often show apparent movements along the bright filamentary structures roughly in the radial direction of the sunspot . now to find the properties of the bds , we follow a similar method as used by @xcite .
two examples of the analysis are shown in figure [ xt_b1ab2 ] & .
we have studied 180 penumbral bds and try to find if they have any signature in the sot h and iris 2796 channels .
we use the central image to compute the intensity enhancement , length and width of a bd , as shown in figure [ xt_b1ab2 ] .
we plot the intensities along the green cut and red cut ( panel a ) .
the green cut is placed along the radial direction and the red cut is perpendicular to it .
the intensity values are shown by red and green diamond symbols . a fitted gaussian is also overplotted with the same color .
the fwhm of the green profile provides an estimate of the length of the bd , whereas the red profile provides the width .
percentage intensity enhancement is calculated from the peak intensity and the linear background of the fitted gaussian .
the smallest between the two values ( green and red ) measures the intensity enhancement .
we plot space - time ( xt ) diagram ( panel c ) to study their temporal behavior and to measure their apparent speed .
a 2@xmath11 intensity contour is drawn on the xt map .
the inclined blue line is drawn inside the contour to measure the apparent velocity and lifetime of the bd .
the starting time of the blue line is considered as the initiation of the bd .
the long intensity strip at the center of the xt map is due to high intensity enhancement at that particular time frame .
we follow a similar method for the h and 2796 channels ( d and g ) .
the intensity enhancement in these channels is only a few percent and very difficult to find the location of the intensity enhancement .
so , we use the running difference images to place our green and red cut ( see e and h ) on the original images . the xt map is produced after removing a smoothed background from each images .
an elongated bright structure is seen in the xt plot .
these have been conventionally referred as penumbral micro - jets ( pmjs ) by @xcite .
the lifetimes of these jets are very small .
most of them appear as sudden brightening in the xt plot and hence it is very difficult to determine their direction of propagation and the speed . in this work
we do not focus on the speed of these jets . using earlier convention
we have assumed that these brightenings are propagating outward form the sunspot , though it is not clearly established from our analysis that the bright structures are moving outward from the sunspot .
we use a 2@xmath11 intensity contour to find the location and extent of the pmj .
the blue line is drawn along the direction of intensity enhancement .
the starting time of the blue line is considered as the initiation and the length of the line is considered as the length of the pmj .
we can clearly see that the bd appears before the pmj in this example .
similarly , in figure [ xt_b1ab2 ] , we study another bd .
following same methodology we study the evolution of the bd .
we could not find a clear signature of the bd in the h though a faint signature is seen in the 2796 channel . in the following subsection
a statistical study is performed to find out if there are two types bds present in our data .
we analyze the properties of many bds and their signature in the other channels . out of the total 180 events ,
90 are identified as having a correlated signature in the ca ii h and 2796 channels , while the other 90 do not .
we have separated these two classes so as to find out if they have any significant differences in their physical properties .
figure [ stat_bd ] ( a)-(f ) show the distributions of the observed parameters .
the red color in the histograms represent the bds which do not have a counterpart in the h images whereas the blue represent the bds which are related to the h pmjs .
the mean value of each of the parameters for both the types are also printed .
we find that the bds which are related to pmjs , generally , are longer and have higher intensity enhancements .
they are also more elongated ( the ratio between length and width is higher ) .
one of the distinguishable feature is that most of the bds which are related to pmjs show negative velocities ( drifts towards the center of the sunspot ) .
we have also calculated a few physical parameters of pmjs .
the position and length of each pmjs are measured in both the 2796 and h channels .
the starting point of the pmjs is considered as the location close to the center of the sunspot .
the locations of the pmj as seen in the h and 2796 are marked ( yellow arrows ) in the figures [ pos]a , b respectively .
the length of the arrow measures the length of the pmjs .
the bds which are related to these pmjs are also shown by the small dots .
red represents the dots which moves inward and blue represents those that move outward from the sunspot center .
the pmjs as seen in 2796 generally show spatial offset from the h pmjs along the pmjs direction ( outer penumbral side ) . in figure [ pos]c
, we show the location of the bds which are not related to any pmjs .
figure [ pos](d ) show the distribution of the length of the pmjs as seen in the 2796 and h images .
the pmjs in h have similar value as reported by @xcite though they appear little longer in the 2796 images .
figure [ pos](e ) show the distribution of the length between the bds and the footpoint of the pmjs and figure [ pos](f ) show the time delay between the appearance of the bds and pmjs .
a negative value means bds appear earlier than the pmjs .
we find that most of the bds appear before the pmjs .
we have analyzed the properties of 180 bds as seen in the 1400 images above a sunspot . using our coordinated observation ,
we have also positively identified pmjs in the sot h and iris 2796 images .
we find that 90 bds are related to pmjs whereas others are not .
a detail analysis show that the bds which are related are generally longer and more elongated and have higher intensity enhancements .
most of these bds show negative velocities ( moves inward direction ) and appear at top of the pmjs .
these bds are found to be appearing before the generation of the pmjs .
these results may indicate that bds could originate from a magnetic reconnection occurring at low coronal heights and/or due to falling plasma .
a component of the plasma from the reconnection site may move downwards and reach the tr and show inward motion as seen in the iris 1400 images . finally , it reaches the chromosphere and appears as ribbon - like brightening ( pmj ) which could explain the time delay in the appearance of the pmjs and their short lifetime .
@xcite , @xcite and @xcite proposed that this kind of small magnetic reconnection occurs in the low corona .
this is in contrast to the reconnection model as proposed by @xcite .
the progressive heating mechanism of the pmjs as proposed by @xcite also may not explain the inward motions of the bds , appearance of bds before pmjs and the longer lifetime of the bds .
it is still unclear how the non - related bds originate .
we thank yukio katsukawa for useful discussions and suggestions .
h. tian is supported by the recruitment program of global experts of china , nsfc under grant 41574166 , and the max planck partner group program .
h.tian thanks issi bern for the support to the team `` solar uv bursts : a new insight to magnetic reconnection '' .
we thank the hinode , iris and sdo team for proving the data in the public domain .
hinode is a japanese mission developed and launched by isas / jaxa , with naoj as domestic partner and nasa and stfc ( uk ) as international partners .
it is operated by these agencies in co - operation with esa and nsc ( norway ) .
iris is a nasa small explorer mission developed and operated by lmsal with mission operations executed at nasa ames research center and major contributions to downlink communications funded by the norwegian space center ( nsc , norway ) through an esa prodex contract . | recent high - resolution observations reveal that subarcsecond bright dots ( bds ) with sub - minute lifetimes appears ubiquitously in the transition region ( tr ) above sunspot penumbra .
the presence of penumbral micro - jets ( pmjs ) in the chromosphere have also been reported earlier .
it was proposed that both the pmjs and bds are formed due to magnetic reconnection process and may play an important role in heating of the penumbra .
using simultaneous observation of the chromosphere from the solar optical telescope ( sot ) aboard hinode and the tr from the interface region imaging spectrograph ( iris ) , we study the dynamics of bds and their relation with pmjs .
we find two types of bds , one which is related to pmjs and the others which do not show any visible dynamics in the sot h images . from a statistical analysis
we show that these two types have different properties .
the bds which are related to pmjs always appear at the top of the pmjs , the vast majority of which show inward motion and originate before the generation of the pmjs .
these results may indicate that the reconnection occurs at the lower coronal / tr height and initiates pmjs at the chromosphere .
this formation mechanism is in contrast with the currently believed formation of pmjs by reconnection in the ( upper ) photosphere between differently inclined fields . |
quasar pairs are known to represent three classes of objects : chance alignments , gravitational lenses , and binary quasars .
binary quasars are rare ; in the lbqs sample of @xmath610@xmath7 quasars , hewett et al .
( 1998 ) find two binaries . a similar rate of occurrence is found for all quasars in the literature : 11 qso pairs in @xmath6 10@xmath8 quasars ( kochanek , falco , and muoz 1999 ) .
this rate is high compared to a naive extrapolation of the quasar quasar two - point correlation function at megaparsec separations to separations less than 100 kpc ( djorgovski 1991 ) .
the existence of binary quasars , as distinct from the gravitationally lensed pairs , has been established by the occurrence of quasar pairs with only one radio loud member ( e.g. djorgovski et al .
1987 ) and by the statistics of qso pairs with large ( @xmath9 30 ) separations ( kochanek 1999 ) .
kochanek et al .
find that quasar pairs with @xmath10@xmath11 @xmath12 30 are gravitational lenses and that most quasar pairs with @xmath10@xmath11 @xmath13 30 are binary quasars .
the leading explanation for binary quasars is the fueling of quasar activity when two galaxies with massive black holes in their nuclei collide ( kochanek 1999 ) .
the observation that two nearby quasar images have similar emission - line spectra is insufficient to allow classification as a lens ( e.g. mortlock , webster , and francis 1999 ) ; and gravitational lens pairs can have different emission - line spectra due to reddening or time delays .
however , quasar pairs with almost identical redshifts but very different emission - line spectra or bals are probably binary quasars .
the cloverleaf qso , h1413 + 114 , with @xmath14 , is a four image gravitational lens ; all four images show bals ( e.g. monier , turnshek , and lupie 1998 ) .
hs 1216@xmath155032 with @xmath16 is a binary qso with one balqso and one non
balqso ( hagen et al . 1996 ) . here
we report the serendipitous discovery that lbqs 0103@xmath02753 is a binary quasar with a projected separation of 030 on the sky .
the two components show large differences in their emission - line spectra , and only one has bals .
lbqs 0103@xmath02753 is one of eight balqsos observed in a program of hst / stis and ground based spectrophotometry of low and medium redshift balqsos for studies of chemical abundances .
the selection criteria were the presence of bals ( both high and low ionization balqsos were candidates ) , redshift in the range 0.4 @xmath17 @xmath1 @xmath17 1.8 , and high apparent brightness .
, @xmath18 = 0.848 , was chosen based on its apparent magnitude , @xmath19 = 18.1 , and its lbqs survey spectrum showing broad , deep mg ii @xmath202798 absorption around @xmath21 = 0.769 ( morris et al . 1991 ) .
observations of lbqs 0103@xmath02753 were obtained on 1999 july 1718 ( ut ) .
figure 1 shows the stis ccd acquisition image , which involved a 10s integration time in the f28x50lp mode ( transmission from about 5000 to 10,000 , @xmath22 = 7228.5 ) with a standard 100 by 100 subimage readout .
the image has total counts of 1113 and 356 dn for a and b ( formal errors 1.8% , 4.1% ) , respectively , using a 3 pixel radius photometry aperture and a much larger annulus for sky .
the a / b flux ratio is 3.1 , probably a slight underestimate due to the wings of image a leaking into image b. the centroids are separated by @xmath23=0295 @xmath24 0.011 .
the fwhms of the images are 2.1 and 2.3 pixels for a and b , respectively , consistent with point sources .
the position angle ( pa ) of the image @xmath25 axis is pa = @xmath0149.7 , and the pa connecting a and b is @xmath0149.9@xmath26 or + 30.1@xmath26 @xmath24 2.2@xmath26 , with a north and east of b. the qso pair is almost perfectly aligned with the stis slit , itself aligned along the image + y axis .
a three step peak up into the 52x0.2 ( 02 wide ) aperture was used for the spectroscopic observations with the stis nuv mama using the g230l grating .
four orbits gave a total integration time of 10,907 s. the spectra of the two qsos are shown in figure 2 .
the spectra were extracted from the nuv
mama image using the standard stsdas / calstis routine `` x1d '' with an 11 pixel wide extraction box for each qso .
the measured separation of the two spectra near the center of the nuv
mama image is 11.9 pixels or 0295 , in agreement with the acquisition image .
inspection of plots perpendicular to the dispersion shows that the two spectra have very little ( @xmath17 5% ) contamination from each other .
the nominal resolution at 2400 is 2.1 pixels fwhm or 3.3 for the 1.55 per pixel sampling .
the useful wavelength range of the two spectra is from 1650 ( where the s / n @xmath6 1 ) to 3145 .
the s / n in the continuum around 2600 is about 20 per pixel for component a and 11 per pixel for b. an optical spectrum of lbqs 0103@xmath02753 ( sum of both components ) was obtained obtained on 22 august 1998 ut using the ctio blanco 4 m telescope .
the r c spectrograph with the blue air schmidt and loral 3k ccd with a 632 line mm@xmath27 grating was used to obtain a 3.0 fwhm resolution spectrum covering 3050 to 5975 .
the spectrum is shown in figure 3 , along with the lbqs spectrum obtained on 26 august 1987 ut by morris et al .
at @xmath1 = 0.848 , the projected separation of 030 amounts to 2.3 kpc (= 70 mpc@xmath27 , @xmath28 , @xmath29 ) . for comparison ,
the next smallest reliable binary quasar known , j1643@xmath03156 , has an apparent separation of 15 kpc ( brotherton et al .
table 1 compares lbqs 0103@xmath02753 to the next smallest binary quasars and other bal / non bal quasar pairs ( also see lists by mortlock et al . 1999 and kochanek et al .
1999 ) .
the stis spectra of lbqs 0103@xmath02753 a and b show that the two components of this qso pair have very different emission , continuum , and intrinsic absorption characteristics .
component a shows strong bals of c iv @xmath301548,1551 ( @xmath31 to 27,000 ) , si iv @xmath201400 , n v @xmath201240 , and o vi @xmath201034 .
the emission line shapes , equivalent widths , and intensity ratios are different between components a and b. the two components of other binary qsos agree in emission - line redshift within about 600 ( kochanek et al .
1999 ) . for lbqs
0103@xmath02753 the emission redshifts , @xmath2 = 0.834 and @xmath3 = 0.858 , differ by 3900 .
the centroids of the c iv @xmath201549 emission line profiles above 50% of peak flux level were used to measure the redshifts .
the formal one @xmath32 errors in redshift for components a and b , in velocity units , are 370 and 210 respectively ; and the systematic errors , estimated from different choices for the continuum , are @xmath33 and @xmath34 . for component
b , the ly @xmath35 line gives the same redshift as c iv . for component a , the ly @xmath35 and n v @xmath201240 emission lines are blended with an unknown ratio of line strengths and an uncertain profile because the n v bal truncates the feature on the blue side .
the si iv and o iv ] @xmath201400 and o vi @xmath201034 emission lines in component a , with positions marked in figure 2 based on @xmath2 = 0.834 , are probably present but weak and the wavelengths are uncertain because the shapes of the overall features are dominated by the adjacent bals .
the redshift of component a depends entirely on the measurement of the c iv line . however , c iv emission lines are often blueshifted from the systemic redshift ( e.g. espey et al .
1989 ) ; and weak low contrast lines , like the c iv emission in a , tend to be shifted the most ( brotherton et al . 1994 ) .
if there was a large blueshift from the systemic value for a and little blueshift from the systemic value for b , the systemic redshifts could be close .
further observations are needed to determine the systemic redshift difference between a and b. the optical spectrum of lbqs 0103@xmath02753 ( the summed components shown in fig .
3 ) has undergone a large change in 11 years in the observed frame or 6 years in the qso frame .
the mg ii @xmath202798 bal , which is the strongest of the low ionization bals ( voit et al . 1993 ) , has changed from about 50% deep in 1987 to less than 10% deep in 1998 .
the spectrum of lbqs 0103@xmath02753 obtained in august , 1998 , matches the average non balqso spectrum ( weymann et al . 1991 ) to within 3% between @xmath36 and the base of mg ii .
also the continuum has apparently become bluer from 1987 to 1998 .
low ionization balqsos tend to have redder spectra than balqsos without low ionization lines ( sprayberry and foltz 1992 ) .
the stis spectrum of lbqs 0103@xmath02753 , obtained about 11 months after the optical spectrum of august , 1998 , shows very strong high ionization bals , but no apparent low ionization bals like c ii @xmath201335 . in 6 years in the qso frame , lbqs 0103@xmath02753 a has changed from a low ionization balqso to a high ionization balqso .
this is a large change compared with typical balqso variability ( barlow 1993 ) .
the unusual bal variability was noted ( junkkarinen , cohen , and hamann 1999 ) before the stis observations were obtained that show lbqs 0103@xmath02753 is a binary qso .
if we assume power - law continua @xmath37 for each component , with @xmath38 , the flux ratios of a and b at @xmath39 ( stis ) and @xmath40 ( image ) give @xmath41 , so that b is slightly bluer than a. using the ctio fluxes at @xmath204150 and @xmath42 , we find @xmath43 = -0.66 , -0.86 and @xmath44 in units of @xmath45 . extrapolating the observed spectra to 5500(1+z ) , we find that components a , b have absolute magnitudes @xmath46 , again for @xmath47 , @xmath48 , and @xmath49 .
the odds of finding two unrelated qsos so close in direction and redshift are negligible .
the very different ultraviolet spectra of lbqs 0103@xmath02753a , b show that this is a true binary , not a lensed qso .
variability , microlensing , and differential reddening might give different spectra or colors for components of a lensed qso .
however , for our pair the expected time delay for a lens is @xmath610 days , and luminous qsos rarely show significant variability over such times .
unequal reddening can not explain the different absorption and emission lines of lbqs 0103@xmath02753a , b ( see lopez 2000 for a related discussion ) . what is the rate of occurrence of such close binary qsos ?
the snapshot survey for gravitationally lensed qsos observed 498 qsos with @xmath50 and @xmath51 ( = 100 , = 0.5 ; maoz 1993 , and references therein ) .
no binary qsos with separations less than 10 were found .
our pair slightly misses the redshift cutoff for the snapshot survey , but the angular diameter distance is not a strong function of redshift above z = 0.86 ( see hogg 1999 , and references therein ) .
lbqs 0103@xmath02753 also misses the absolute magnitude cutoff by @xmath52 magn .
however , with regard to angular separation and magnitude difference , the snapshot survey provides some guidance . at a separation of 03 ,
the snapshot survey could detect pairs with a magnitude difference @xmath53 at a pa of 0 relative to the image trail , the least favorable pa for detection ( bahcall 1992 ) .
a pair with the spacing and magnitude difference ( @xmath54 ) of lbqs 0103@xmath02753 should have been detected at any pa .
if the snapshot survey accounts for most of the opportunities to detect such a pair to date , the example of lbqs 0103@xmath02753 is consistent with a rate of roughly 1/500 for separations
@xmath603 . among the @xmath55 known qsos ,
there likely are of order 10 unrecognized binaries in the 03 range .
is lbqs 0103@xmath02753 likely to be a chance projection of a wider binary ?
the probability of a qso having a binary companion at 3 to 10 arcseconds is @xmath56 ( djorgovski 1991 ; kochanek 1999 , and references therein ) .
the cases listed by kochanek have a median separation of @xmath57 ( 40 kpc ) .
the probability that a pair with this physical spacing will have a projected spacing of 03 or less , because it is oriented nearly along the line of sight , is ( coincidentally ) @xmath56 .
the probability that a randomly selected qso will be such a pair , and that it will be so oriented , is then @xmath58 .
then , in a sample of 500 qsos , the probability of finding even one pair at 03 is only @xmath56 .
this suggests that lbqs 0103@xmath02753 is not a chance projection of a @xmath59 kpc binary , but rather has a true spacing not much greater than 2 kpc .
close binary qsos presumably represent mergers of galaxies in which the nuclei in both galaxies are currently active .
observations of nearby galaxies show that the nuclei of large galaxies typically contain supermassive black holes ( , kormendy & richstone 1995 ) .
the merger will likely produce a supermassive binary black hole ( begelman , blandford , & rees 1980 ) .
simulations of colliding disk galaxies by barnes and hernquist ( 1996 ) show little perturbation of the incoming galaxies until the first pericenter passage . however , gas accumulates in the nucleus of each galaxy as the two orbit away from each other . the time to orbit to an apocenter @xmath60 kpc and
fall back is @xmath61 for galactic masses relevant here ( see below ) .
subsequent orbital loops are much smaller , and the central regions of the galaxies merge quickly , entailing a massive concentration of gas in the nucleus .
we speculate that the observed 3 to 10 binary qsos mostly represent galaxy pairs undergoing the loop following the first close encounter , and that lbqs 0103@xmath02753 is in a later stage when the nuclei of the two galaxies are coalescing .
is the incidence of @xmath62 binaries consistent with this scenario ?
the time for the orbit to decay from radius @xmath63 to the center by dynamical friction is @xmath64 ( binney & tremaine 1987 ; mortlock 1999 , and references therein ) . here
@xmath65 is the mass of the orbiting black hole ( units @xmath66 ) , assumed to be the smaller hole on a circular orbit through the stellar background of the larger galaxy ( velocity dispersion @xmath67 ) .
let us assume that each qso is shining at one - third of the eddington limit ( see kaspi 2000 ) and take @xmath68 following laor ( 1998 ) .
then @xmath69 for a , b ; and the black hole masses are @xmath70
. the recently discovered correlation of black hole mass with bulge velocity dispersion , @xmath71 ( gebhardt 2000 , ferrarese and merritt 2000 ) gives @xmath72 .
the total bulge masses should be @xmath73 ( kormendy & richstone 1995 ) .
tidal radius considerations suggest that the dense core of the stellar nucleus of galaxy b remains intact around black hole b , with a total mass @xmath74 .
this shortens the dynamical friction time , and the above formula then suggests that the orbit decays on the dynamical timescale of @xmath75 years .
the discovery of lbqs 0103@xmath02753 then indicates that the probability of both nuclei being active as qsos during the 2 kpc stage of a merger is @xmath76 times higher than during the 40 kpc phase .
simulations of galactic mergers including supermassive black holes could help to clarify these timescales .
ultraluminous infrared galaxies ( uligs , @xmath77 ) typically involve mergers , the mean nuclear spacing being @xmath4 kpc ( sanders and mirabel 1996 ; genzel and cesarsky 2000 ) . for high luminosities , the predominant power source appears to be agn activity that is heavily obscured by a large mass of nuclear gas and dust . from this perspective , an important aspect of lbqs 0103@xmath02753 is the lack of heavy extinction , perhaps because the high qso luminosity has dispersed the nuclear gas .
infrared observations of lbqs 0103@xmath02753 would be most interesting , as would observations that could detect a large mass of atomic or molecular gas expanding away from the nucleus .
the discovery of lbqs 0103@xmath02753 suggests that binaries with separations of several kpc occur with a frequency of order one or two per thousand qsos .
the detection of just a few more cases would provide a better estimate of this rate , providing an important constraint on galactic mergers and the fueling of active galactic nuclei .
we thank r. blandford , k. gebhardt , l. hernquist , j. kormendy , c. kochanek , c. mihos , s. phinney , d. richstone , and r. weymann for helpful discussions .
we also thank r. weymann for providing in digital form the lbqs 0103@xmath02753 spectrum from the lbqs survey , and k. gebhardt for the use of a computer routine .
g.a.s . gratefully acknowledges the hospitality of lick observatory and stanford university during summer 2000 .
support for this work was provided by nasa through grants go-07359 , go-07359.02 from the space telescope science institute , which is operated by the association of universities for research in astronomy under nasa contract nas5 - 26555 .
llrrcrll + lbqs 0103@xmath02753 & 0.848 & 0.3 & 2.3 & 18.2 & 1.2 : & bal / non bal pair & 1 + j1643@xmath153156 & 0.586 & 2.3 & 15 & 18.4 & 0.8 & radio loud / radio quiet pair & 2 + ctq 839 & 2.24 & 2.1 & 17 & 18.2 & 1.9 & radio quiet / radio quiet pair & 3 + + + lbqs 2153@xmath02056 & 1.85 & 7.8 & 66 & 17.9 & 3.4 & bal / non bal pair & 4 + hs 1216@xmath155032 & 1.45 & 9.1 & 77 & 17.2 & 1.8 & bal / non bal pair & 5,6 + q1343@xmath152640 & 2.03 & 9.5 & 79 & 20.2 & 0.1 & weak bal / non bal pair & 7 + | imaging and spectroscopy with show that lbqs 0103@xmath02753 ( v = 17.8 , @xmath1 = 0.848 ) is a binary quasar with a separation of 03 or 2.3 kpc .
this is by far the smallest separation binary quasar reported to date .
the two components have very different spectra , including the presence of strong broad absorption lines ( bals ) in component a only .
the emission - line redshifts , based on the broad high ionization c iv lines , are @xmath2 = 0.834 and @xmath3 = 0.858 ; their difference is 3900 in velocity units . the broad c iv lines , however , are probably not a good indicator of systemic redshift ; and lbqs 0103@xmath02753 a and b could have a much smaller systemic redshift difference , like the other known binary quasars .
if the systemic redshift difference is small , then lbqs 0103@xmath02753 would most likely be a galaxy merger that has led to a binary supermassive black hole .
there is now one known 03 binary among roughly 500 qsos that have been observed in a way that would reveal such a close binary .
this suggests that qso activity is substantially more likely for black hole binaries at spacings @xmath4 kpc than at @xmath5 to 60 kpc . between 1987 and 1998 ,
the observed mg ii bal disappeared . |
in the past we have calculated inclusive cross section for heavy quarks production at hadron colliders .
these calculations were performed using an approach based on the unintegrated parton distributions functions @xcite .
it is known that gluon - gluon fusion is the dominant mechanism at high energy .
however , other mechansims were not carefully studied in the literature .
it is the aim of this work to present contributions of several subleading terms usually neglected in the analysis of @xmath0 production .
we wish to include contributions of photon - gluon ( gluon - photon ) as well as purely electromagnetic contributions of photon - photon fusion .
we wish to discuss also diffractive processes ( single and central ) in the framework of ingelman - schlein model corrected for absorption .
such a model was used in estimation of several diffractive processes @xcite .
the absorption corrections turned out to be necessary to understand a huge regge - factorization breaking observed in single and central production at tevatron . recently a surprisingly large cross section for exclusive @xmath0 production has been reported @xcite .
her we will show results for rhic and lhc energies .
in the leading - order ( lo ) approximation within the collinear approach the quadruply differential cross section in the rapidity of @xmath8 ( @xmath9 ) , in the rapidity of @xmath10 ( @xmath11 ) and the transverse momentum of one of them ( @xmath12 ) can be written as @xmath13 above , @xmath14 and @xmath15 are the familiar ( integrated ) parton distributions in hadron @xmath16 and @xmath17 , respectively .
there are two types of the lo @xmath18 subprocesses which enter eq.([lo_collinear ] ) : @xmath2 and @xmath19 .
the first mechanism dominates at large energies and the second one near the threshold .
in particular for the gluon - gluon fusion the cross section formula takes a simple form : @xmath20 there are three ( @xmath21 , @xmath22 and @xmath23 ) diagrams in the leading order @xcite .
the parton distributions are evaluated at : @xmath24 , @xmath25 , where @xmath26 .
the formulae for matrix element squared averaged over the initial and summed over the final spin polarizations can be found e.g. in ref.@xcite . the inclusive heavy quark / antiquark production can be also calculated in the framework of the @xmath27-factorization . in this approach transverse momenta of initial partons are included and emission of gluons is encoded in so - called unintegrated gluon distributions ( ugdfs ) @xcite . in the leading - order ( lo )
approximation within the @xmath27-factorization approach the quadruply differential cross section in the rapidity of @xmath8 ( @xmath9 ) , in the rapidity of @xmath10 ( @xmath11 ) and the transverse momentum of @xmath8 ( @xmath28 ) and @xmath8 ( @xmath29 ) can be written as @xmath30 where @xmath31 and @xmath32 are so - called unintegrated gluon ( parton ) distributions .
now the unintegrated parton distributions must be evaluated at : @xmath33 , @xmath34 , where @xmath35 .
as discussed above the dominant contributions are initiated by gluons or quarks and antiquarks . in general even photon
can be a constituent of the proton .
this was considered only in one work by martin , roberts , stirling and thorne @xcite .
below we repeat the main aspects of their formalism .
the factorization of the qed - induced collinear divergences leads to qed - corrected evolution equations for the parton distributions of the proton @xcite : @xmath36 where @xmath37 and the parton distributions fullfil momentum conservation : @xmath38 if the photon is a constituent of the nucleon then other mechanisms presented in fig.[fig : new_diagrams ] are possible . + + here
the cross section can be calculated similarly as for the gluon - gluon fusion . a corresponding triple differential
cross section can be written as : @xmath39 for gluon - photon , photon - gluon and photon - photon contributions , respectively . compared to gluon - gluon case here
only @xmath22 and @xmath23 diagrams occur .
the above contributions include only cases when nucleons do not survive a collision and nucleon debris is produced instead .
the case when nucleon survives a collision has to be considered separately . in this case
one can include corresponding photon distributions where extra `` el '' index will be added to denote that situation .
corresponding contributions can be then written as : @xmath40 the elastic contributions are calculated using drees - zepenfeld ( elastic ) parametrizations of photon fluxes @xcite which include nucleon electromagnetic form factors .
in fig.[fig : xg ] we show three different leading - order gluon distributions from the literature @xcite ( left panel ) and photon distributions @xcite ( right panel ) as a function of longitudinal momentum fraction @xmath41 for a fixed scale @xmath42 10 gev relevant for @xmath0 production . above @xmath43 10@xmath44
all the distributions coincide . for smaller values of @xmath41
they diverge and can be different by almost an order of magnitude .
what are consequences of this divergence for @xmath0 pair production ?
this will be discussed below . [ ! thb ] before we go to cross sections in transverse momentum and rapidities , in fig.[fig : dsig_dxi1dxi2 ] we present distribution of the cross section in @xmath45 = log@xmath46 and @xmath47 = log@xmath48 for two different energies @xmath49 = 500 gev ( updated rhic ) and @xmath49 = 14 tev ( nominal lhc energy ) .
one can clearly see that the @xmath50 and @xmath51 values are strongly correlated .
typical values at @xmath49 = 500 gev are @xmath52 0.5 @xmath53 10@xmath44 and at @xmath49 = 14 tev are @xmath52 10@xmath54 . in the latter case
@xmath41 s as small as 10@xmath55 may appear in the forward @xmath6 or @xmath7 region .
this is clearly a region of @xmath41 which was never studied so far .
[ ! thb ] now let us present distributions in transverse momentum of @xmath6 ( or @xmath7 ) for gluon - gluon fusion mechanism for different gluon distributions and different popular choices of scales ( @xmath56 , invariant mass of the @xmath0 system @xmath57 ) .
we show our results for @xmath49 = 500 gev ( fig.[fig : dsig_dpt_500 ] ) and @xmath49 = 14 tev ( fig.[fig : dsig_dpt_14000 ] ) .
one can clearly see that for some choices of gluon distribution function and scales the results for @xmath49 = 14 tev are not physical .
this shows how badly known are gluon distributions at the low @xmath41 .
[ ! thb ] [ ! thb ] in fig.[fig : dsig_dpt_gamma ] and in fig.[fig : dsig_dy_gamma ] , we show results for different gluon distribution functions for the rhic energy @xmath49 = 500 gev and nominal lhc energy @xmath49 = 14 tev , respectively . at the lhc energy
the results for different gdfs differ considerably which is a consequence of the small-@xmath41 region as discussed in the previous section .
the differences at the nominal lhc energy @xmath49 = 14 tev are particularly large which can be explained by the fact that a product of gluon distributions ( both at small @xmath41 ) enters the cross section formula
. a new measurement of @xmath0 at the nominal lhc energy will be therefore a severe test of gluon distributions at small @xmath41 and not too high factorization scales not tested so far .
similar uncertainties for the @xmath58 and @xmath59 are smaller as here only one gluon distribution appears in the corresponding cross section formula .
thb ] [ !
thb ] it is very difficult to quantify uncertainties related to the photon pdfs as only one set of pdfs includes photon as a parton of the proton . here the isospin symmetry violation ( not well known at present ) would be an useful limitation .
our collection of the results for the photon induced mechanisms show that they are rather small and their identification would be rather difficult as the different distributions are very similar to those for the gluon - gluon fusion .
our intension here is to document all the subleading terms in one publication .
our etimation shows that the sum of all the photon induced terms is less than 0.5 % and is by almost 2 orders of magnitude smaller than the uncertainties of the dominant leading - order gluon - gluon term .
the mechanisms of the ordinary as well as diffractive production of heavy quarks ( @xmath0 ) are shown in figs.[fig : mechanism_sd],[fig : mechanism_dd ] . the mechanism of single - diffractive production of @xmath0 .
, title="fig:",scaledwidth=35.0% ] the mechanism of single - diffractive production of @xmath0 .
, title="fig:",scaledwidth=35.0% ] the mechanism of central - diffractive production of dileptons .
, scaledwidth=35.0% ] in the following we apply the ingelman and schlein approach production @xcite . ] .
in this approach one assumes that the pomeron has a well defined partonic structure , and that the hard process takes place in a pomeron proton or proton pomeron ( single diffraction ) or pomeron pomeron ( central diffraction ) processes .
we calculate triple differential distributions as @xmath60 , \nonumber \\
\label{00}\end{aligned}\ ] ] @xmath61 , \nonumber \\ \label{sd}\end{aligned}\ ] ] @xmath62 \nonumber \\ \label{dd}\end{aligned}\ ] ] for ordinary , single - diffractive and central - diffractive production , respectively .
we do not calculate the higher - order contributions and include them effectively with the help of a so - called @xmath63-factor .
we have checked that this procedure is precise enough in the case of ordinary drell - yan process .
the @xmath63-factor is calculated as for the drell - yan process @xmath64 here the running coupling constant @xmath65 is evaluated at @xmath66
. the diffractive quark distribution of flavour @xmath67 can be obtained by a convolution of the flux of pomerons @xmath68 and the parton distribution in the pomeron @xmath69 : @xmath70 the flux of pomerons @xmath68 enters in the form integrated over four momentum transfer @xmath71 with @xmath72 being kinematic boundaries . both pomeron flux factors
@xmath73 as well as quark / antiquark distributions in the pomeron were taken from the h1 collaboration analysis of diffractive structure function and diffractive dijets at hera @xcite . the factorization scale for diffractive parton distributions
is taken as @xmath74 .
let us start presentation of our results for diffractive mechanisms . in fig.[fig
: diff_dsig_dpt_500 ] we show transverse momentum distributions of charm quarks ( or antiquarks ) .
the distribution for single diffractive component is smaller than that for the inclusive gluon - gluon fusion by almost two orders of magnitude .
our results include gap survival factor .
corresponding values are taken the same as in ref .
the cross section for inclusive central diffractive component is smaller by additional order of magnitude .
in addition we show the cross section for fully state hadronization leads to more complicated states @xcite . ]
exclusive mechanism discussed in section [ sec : exclusive_production ] .
below we shall use the following notation : @xmath75 for standard nondiffractive component , @xmath76 or @xmath77 for single diffractive and @xmath78 for central diffractive components . [ !
thb ] [ !
thb ] in fig.[fig : diff_dsig_dpt_14000 ] we show similar results for nominal lhc energy @xmath49 = 14 tev .
the situation and the interrelations between different components is qualitatively the same . here
somewhat smaller gap survival factors were used @xcite .
the distributions for all components are somewhat broader than those for the rhic energy shown above . [ !
thb ] in fig.[fig : diff_dsig_dy1_14000 ] we show distributions in quark ( antiquark ) rapidity .
we show separately contributions of two different single - diffractive components , which give the same distributions in transverse momentum in fig.[fig : diff_dsig_dpt_14000 ] .
when added together they produce a distribution similar in shape to the standard inclusive case . here
different parton distributions functions give similar result .
the distributions for different proton gluon distributions are quite different .
this was already observed when discussing photon induced components in section [ sec : exclusive_production ] .
[ ! thb ] also two dimensional distributions can be interesting as here different mechanisms may occupy different parts of the phase space . in fig.[fig :
ysummll ] we show distributions in the rapidity of the pair and quark - antiquark invariant mass .
although the distributions are somewhat different the differences occur in the regions which may be difficult to measure .
the spread in the pair rapidity for the central diffractive component is much smaller then that for the inclusive case .
[ ! thb ] finally we show distributions in quark and antiquark rapidities . the distribution for the inclusive central diffractive mechanism
are concentrated at midrapidities .
this is a rather universal feature of diffractive processes .
[ ! thb ] the cross section for single and central diffraction is rather small .
however , a very specific final state should allow for its identification by imposing special conditions on the one - side ( single - diffractive process ) or on both - side ( central diffractive process ) rapidity gaps .
we hope that such an analysis is possible at lhc .
special care should be devoted to the observation of the exclusive @xmath0 production where the observation of @xmath79 mesons associated by a few pions would be a proper signal @xcite . without a special analysis of the final state multiplicity
the exclusive @xmath0 production may look like an inclusive central diffraction . at present
there is no analysis of the final state production for the exclusive @xmath0 .
a comparison of cross sections for both components will be done in the next section .
there is recently a growing theoretical interest in studying central exclusive mechanisms of different particles production at high energies , which constitute a special category of double - diffractive processes . to date , only a few exclusive processes have been measured so far at the tevatron collider ( see @xcite and references therein ) . in particular , central exclusive production ( cep ) of the higgs boson is a flag process of special interest and importance in the upcoming higgs searches at the lhc ( see e.g. ref .
@xcite ) .
generally , in the case of the central exclusive production @xmath80 with the leading protons , the central system @xmath81 should necessarily be produced in the color singlet state , such that the proton remnants and the @xmath81 system are disconnected in the color space and their hadronisation occurs independently giving rise to rapidity gaps @xcite . from the experimental point of view , cep procesess are very attractive , because of the rare clean experimental environment , related to @xmath82 selection rule , and great mass resolution of the centraly produced object .
such unique features give a new possibility to exploit @xmath83 high branching ratio decay channel of the higgs boson , which is rather impossible in standard inclusive measurements , due to very large qcd background .
therefore , the qcd mechanism of central exclusive heavy quark dijets is a source of the irreducible background to the exclusive higgs boson production .
central exclusive production of @xmath0 and @xmath83 pairs was studied in detail in our previous papers @xcite . in these calculations the @xmath84 reaction , illustrated in fig .
[ fig : edd_mechanism ] , was considered as a genuine 4-body process with exact kinematics .
the applied perturbative model of theoretical predictions is based on the khoze - martin - ryskin ( kmr ) approach used previously for the exclusive higgs boson production @xcite .
total cross sections and differential distributions for heavy quarks are calculated by using @xmath85-factorization approach with help of the kmr unintegrated gluon distribution functions . ,
title="fig:",width=226 ] , title="fig:",width=226 ] this qcd model works very good in the case of exclusive dijets and charmonia production , what was confirmed by cdf data @xcite . however , estimated uncertainties related to gluon densities , factorization and renormalization scales , as well as due to absorption corrections are quite large .
it makes the situation somewhat clouded and prevents definite conlcusions , especially in the case of the exclusive production of heavy quark pairs . in this context
, the promising idea , how to clarify and calibrate purely - known parameters of the theoretical model , is to study @xmath0 cross section by exclusive measurements of @xmath86 meson pairs .
such experimental studies are being performed now at tevatron and could be also available in run ii experiments at rhic and at lhc . therefore , it is also very interesting , from both , theoretical and experimental side , to compare mechanism of central exclusive production of charm quarks with standard single and double diffractive processes .
such an analysis of differential cross sections has never been done before but could bring important informations about differences in kinematics and in production rates between them , what is crucial for future measurements . according to the kmr approach
@xcite we write the amplitude of the exclusive diffractive @xmath87 pair production @xmath88 as @xmath89 where @xmath90 are helicities of heavy @xmath91 and @xmath92 , respectively , @xmath93 are the momentum transfers along each proton line , @xmath94 and @xmath95 are the transverse momenta and the longitudinal momentum fractions for active and screening gluons , respectively . above @xmath96
are the off - diagonal ugdfs related to both nucleons .
the vertex factor @xmath97 is the production amplitude of a pair of massive quark @xmath91 and antiquark @xmath98 with helicities @xmath99 , @xmath100 and momenta @xmath101 , @xmath102 , respectively .
the longitudinal momentum fractions of active gluons are calculated based on kinematical variables of outgoing quark and antiquark : @xmath103 and @xmath104 , where @xmath105 and @xmath106 are transverse masses of the quark and antiquark , respectively , and @xmath107 and @xmath108 are corresponding rapidities . the off - diagonal ugdfs are written as @xcite @xmath109 where @xmath110 accounts for the single @xmath111 skewed effect @xcite .
the factor @xmath112 here can not be calculated from first principles in the most general case of off - diagonal ugdfs .
it can be estimated only in the case of off - diagonal collinear pdfs when @xmath113 and @xmath114 and then @xmath115 . in the considered kinematics
the diagonal unintegrated densities can be written in terms of the conventional ( integrated ) densities @xmath116 as @xcite @xmath117 \ ; , \ ] ] where @xmath118 is the conventional sudakov survival factor which suppresses real emissions from the active gluon during the evolution , so the rapidity gaps survive . in the framework of the @xmath27-factorization approach @xcite the hard subprocess
@xmath119 gauge invariant amplitude @xmath120 reads @xmath121 where @xmath122 is the c.m.s .
proton energy , @xmath123 are the color group generators in the fundamental representation , @xmath124 and @xmath125 are on - shell quark and antiquark spinors , respectively , @xmath126 and @xmath127 are the effective vertices arising from the feynman rules in quasi - multi - regge kinematics ( qmrk ) approach @xcite : @xmath128 where @xmath129 is the effective three - gluon vertex .
the effective @xmath130-vertices are canceled out when projecting the @xmath87 production amplitude eq .
( [ qqamp ] ) onto the color singlet state .
since we will adopt the definition of gluon polarization vectors proportional to transverse momenta @xmath131 , i.e. @xmath132 ( see below ) , then we must take into account the longitudinal momenta in the numerators of effective vertices ( see eq .
( [ bb ] ) ) .
the su(3 ) clebsch - gordan coefficient @xmath133 in eq .
( [ qqamp ] ) projects out the color quantum numbers of the @xmath87 pair onto the color singlet state .
factor @xmath134 provides the averaging of the matrix element squared over intermediate color states of quarks .
therefore , we have the following amplitude @xmath135 in the present calculations we use the gjr08 set of collinear gluon distributions @xcite . in the analogy to the cep of higgs boson , where renormalization and factorization scales are advocated to be @xmath136 @xcite
, we apply the following prescription @xmath137 .
absorption corrections to the bare @xmath138 amplitude , which are necessary to be taken into account ( to ensure exclusivity of the process ) , are included approximately by multiplying the cross section by the gap survival factors @xmath139 for rhic and @xmath140 for the lhc energy .
more details about exclusive production of heavy quarks can be found in our original paper @xcite .
let us come now to presentation of our results . in fig.[fig :
dsig_dy_exc ] we show rapidity distribution of @xmath6 quarks from the exclusive mechanism shown in fig .
[ fig : edd_mechanism ] ( solid line ) .
we show the results for leading order ( upper curves ) and next - to - leading order collinear gluon distributions @xcite .
we observe large difference of results for lo and nlo gluon distribution especially at lhc . for comparison
we show the contribution of central diffractive component discussed in section [ sec : diffractive_production ] . in this calculation
we have included gap survival factors @xmath141 = 0.1 for @xmath49 = 500 gev and @xmath141 = 0.03 for @xmath49 = 14 tev .
the cross section for the exclusive mechanism is similar to that for the inclusive central diffractive mechanism .
the exclusive production starts to dominate only at large @xmath6 quark rapidities .
therefore a measurement of the cross section with double ( both side ) rapidity gaps may be not sufficient to single out the exclusive mechanism .
clearly other cuts would be necessary . [ !
thb ] quark / antiquark for the exclusive component at @xmath49 = 500 gev ( left panel ) and @xmath49 = 14 tev ( right panel ) .
for comparison we show the central diffractive contribution ( dashed line ) .
different collinear gluon distributions were used to obtain the unintegrated gluon distribution according to the kmr prescription . [
fig : dsig_dy_exc ] , title="fig:",width=226 ] quark / antiquark for the exclusive component at @xmath49 = 500 gev ( left panel ) and @xmath49 = 14 tev ( right panel ) . for comparison
we show the central diffractive contribution ( dashed line ) .
different collinear gluon distributions were used to obtain the unintegrated gluon distribution according to the kmr prescription .
[ fig : dsig_dy_exc ] , title="fig:",width=226 ] corresponding distributions in the @xmath6 quark ( @xmath142 antiquark ) transverse momentum are shown in fig.[fig : dsig_dpt_exc ] .
the distribution for exclusive component extends to higher transverse momentum than that for the central inclusive diffractive one .
a lower cut on @xmath6 quark ( d meson ) transverse momentum may therefore help to identify the exclusive component but will exclude a measurment of the integrated cross section for this component .
a detailed monte carlo studies of final states of both components may help to find a better criterion to separate experimentally the two components . [ !
thb ] quark / antiquark for the exclusive component at @xmath49 = 500 gev ( left panel ) and @xmath49 = 14 tev ( right panel ) tev .
different collinear gluon distributions were used to obtain the unintegrated gluon distribution according to the kmr prescription . for comparison
we show the inclusive central diffractive contribution ( dashed line ) .
[ fig : dsig_dpt_exc ] , title="fig:",width=226 ] quark / antiquark for the exclusive component at @xmath49 = 500 gev ( left panel ) and @xmath49 = 14 tev ( right panel ) tev .
different collinear gluon distributions were used to obtain the unintegrated gluon distribution according to the kmr prescription . for comparison
we show the inclusive central diffractive contribution ( dashed line ) .
[ fig : dsig_dpt_exc ] , title="fig:",width=226 ]
in the present paper we have calculated differential distributions for different subdominant contributions usually neglected in the literature when calculating production of @xmath0 pairs .
single and double photon induced processes are first class of mechanisms considered here . in calculating single particle distributions
we have used a special set of parton distributions which includes photon as a parton of the proton .
the calculation of the cross section is therefore very similar to that for the gluon - gluon fusion .
the difference is only in color factors and a lack of the s - channel diagrams for photon induced processes .
we have also included contributions when emitted photon which enters a hard process leaves a proton in a ground state .
those `` elastic '' mechanisms give similar contribution as the `` deeply inelastic '' mechanisms considered in the qcd - improved parton model .
we have found that although individual contributions are very small , when added together they can give cross section of about 1 % of the inclusive one dominated by gluon - gluon fusion . in our analysis
we have neglected resonance contributions when the photon leaves the remaining object in a proton excited state , e.g. in @xmath143(1220 ) resonance or other nucleon resonances .
we have also discussed single and central diffractive production of @xmath0 pairs in the ingelman - schlein model . in these calculations
we have included diffractive parton distributions obtained by the h1 collaboration at hera and absorption effects neglected in some early calculations in the literature .
the absorption effects which are responsible for the naive regge factorization breaking cause that the cross section for diffractive processes is much smaller than that for the fully inclusive case , but could be measured at rhic and lhc by imposing special condition on rapidity gaps .
finally we have discussed a fully exclusive diffractive production of @xmath0 .
it was advocated recently that the cross section for this mechanism may be substantial .
we have found here that both at rhic and lhc its contribution is smaller than that for single diffractive one . in our opinion
it is very timely to analyze if this contribution could be measured .
this equires an analysis of the final state .
we expect that the final state in single and exclusive production are different enough to set criteria to pin down the fully exclusive component .
it is , however , not obvious if the central diffractive and purely exclusive mechanisms could be differentiated experimentally .
they may look similar as far as rapidity gap structure is considered .
we predict that the total contribution of central diffractive mechanism is similar to that for the exclusive one .
in contrast the final state multiplicity can be expected to be different .
a better analysis requires a monte carlo studies .
we have not discussed an impact of diffractive mechanisms considered in the present paper on the fully inclusive cross section for @xmath0 pair production .
this is a rather difficult task and goes beyond the scope of the present paper .
p. bruni , g. ingelman , phys.lett.*b311 * 317 ( 1993 ) ; + l. alvero , j.c .
collins , j. terron and j.j .
whitmore , phys .
* d59 * ( 1999 ) 074022 ; + r.j.m .
covolan and m.s .
soares , phys .
* d60 * ( 1999 ) 054005 , phys . rev .
* d67 * ( 2003 ) 017503 ; + m.b .
gay ducati , m.m . machado and m.v.t machado , phys .
rev . * d75 * ( 2007 ) 114013 .
r. enberg , g. ingelman , n. timneanu , phys.rev.*d67 * ( 2003 ) 011301 ; + s. erhan , v.t .
kim and p.e .
schlein,[arxiv : hep - ph/0312342 ] ; + m.b .
gay ducati , m.m . machado and g.g .
silveira,[arxiv : hep - ph/1101.5602 ] .
r. maciua , r. pasechnik and a. szczurek , phys .
rev . * d83 * , 054014 ( 2011 ) r. maciua , r. pasechnik and a. szczurek , phys
. rev . * d83 * , 114034 ( 2011 ) r. enberg , g. ingelman , a. kissavos _ et al .
_ , phys .
lett . * 89 * , 081801 ( 2002 ) .
v. a. khoze , a. d. martin and m. g. ryskin , phys .
* b401 * , 330 ( 1997 ) ; + a. b. kaidalov , v. a. khoze , a. d. martin and m. g. ryskin , eur .
j. * c33 * , 261 ( 2004 ) .
a. dechambre , o. kepka , c. royon and r. staszewski , phys .
rev . * d83 * , 054013 ( 2011 ) r. maciua , r. pasechnik and a. szczurek , arxiv:1109.5517 .
r. s. pasechnik , a. szczurek and o. v. teryaev , phys .
* b680 * , 62 ( 2009 ) ; + r. s. pasechnik , a. szczurek and o. v. teryaev , phys .
d * 81 * , 034024 ( 2010 ) .
a. de .
roeck et al . , eur .
j. * c25 * ( 2002 ) 391 ; + s. heinemeyer et al .
j. * c53 * ( 2008 ) 231 .
a. g. shuvaev , v. a. khoze , a. d. martin and m. g. ryskin , eur .
j. * c56 * , 467 ( 2008 ) [ arxiv:0806.1447 [ hep - ph ] ] .
a. d. martin and m. g. ryskin , phys . rev . * d64 * , 094017 ( 2001 ) .
a. g. shuvaev , k. j. golec - biernat , a. d. martin and m. g. ryskin , phys .
* d60 * , 014015 ( 1999 ) .
p. hagler , r. kirschner , a. schafer , l. szymanowski and o. teryaev , phys .
* d62 * , 071502 ( 2000 ) ; + p. hagler , r. kirschner , a. schafer , l. szymanowski and o. v. teryaev , phys .
lett . * 86 * , 1446 ( 2001 ) . | at high - energies the gluon - gluon fusion is the dominant mechanism of @xmath0 production .
this process was calculated in the nlo collinear as well as in the k@xmath1-factorization approaches in the past .
we show that the present knowledge of gluon distributions does not allow to make a precise predictions for @xmath0 production at lhc , in particular at forward rapidities . in this paper
we study production of @xmath0 pairs including several subleading mechanisms .
this includes : @xmath2 , @xmath3 , @xmath4 , @xmath5 . in this context
we use mrst - qed parton distributions which include photon as a parton in the proton as well as elastic photon distributions calculated in the equivalent photon approximation .
we present distributions in the @xmath6 quark ( @xmath7 antiquark ) rapidity and transverse momenta and compare them to the dominant gluon - gluon fusion contribution .
we discuss also inclusive single and central diffractive processes using diffractive parton distribution found from the analysis of hera diffractive data . as in the previous case we present distribution in @xmath6 ( @xmath7 ) rapidity and transverse momentum .
finally we present results for exclusive central diffractive mechanism discussed recently in the literature .
we show corresponding differential distributions and compare them with corresponding distributions for single and central diffractive components . |
there is an extensive literature on the development and the use of high - energy electron - induced knockout of both one or two nucleons from the ground states of stable nuclei as probes of nucleonic correlations and of pair correlations .
relevant and recent reviews include references @xcite , which cite the essential experimental developments and data sets : see also the citations in @xcite .
quite recently , new conclusions have been drawn regarding the observation of enhanced short - ranged correlations ( src ) between nucleon - nucleon pairs from @xmath3 experiments on a carbon target @xcite .
specifically , measurements that selected high momentum transfer and large missing momentum events from the few - body final - state phase - space found evidence that @xmath2 pairs were more than an order of magnitude more prevalent than like - nucleon pairs in this selection ; deduced from the corresponding @xmath4 and @xmath5 yields .
the probabilities of these high relative momentum two - nucleon components have been computed , using the microscopic ground - state wave functions from _ ab - initio _ variational monte carlo calculations , for systems with masses @xmath6 @xcite and , using a linked cluster expansion , for @xmath7 @xcite .
these computations involve the wave functions of the nucleons over the entire volume of the nuclei concerned .
these theoretical studies are strongly suggestive that the tensor force , in the spin @xmath8 , isospin @xmath9 , @xmath2 channel , plays the major role in generating these enhanced @xmath2-spatial correlations and their associated high - relative - momenta signatures in the two - nucleon density distribution . in such @xmath3 measurements , the specific final - state phase space selection ( of high relative momentum back - to - back two - nucleon events )
provides the leverage and the amplification of the short - ranged pair - correlation sensitivity . the behavior of the wave functions of single nucleons and of pairs of nucleons ( in a mass @xmath10 projectile ) can also be probed if the nucleons are removed ( suddenly ) in fast collisions with a light target nucleus . within this strong - interaction probe case ,
the sensitivity is now to the nuclear wave functions at and near the nuclear surface .
such processes are also referred to as one- and two - nucleon knockout ( or removal ) reactions .
these direct reaction mechanisms , combined with @xmath11-decay spectroscopy , to determine the final state of the reaction residues , and their momentum distributions , are being actively exploited as a spectroscopic tool to study the evolution of nucleonic single - particle structure near the two ( often very displaced ) fermi surfaces of exotic nuclei .
they are proving to be robust techniques , see e.g. @xcite . unlike the final - state - exclusive measurements in electron induced reactions , observables in such nuclear - induced fast nucleon removal reactions are inclusive with respect to the final states of the removed nucleon(s ) and of the fate of the struck light target nucleus .
the cross sections are thus relatively large . measurements usually consist of the total removal reaction yield , often the momenta of the fast , forward travelling projectile - like residues , and sometimes the differential yields to the ground and bound excited states of the mass @xmath12 or mass @xmath13 residual nuclei ; the latter obtained by @xmath11-ray spectroscopy .
however , in such collisions between the projectile and a light composite target nucleus , such as beryllium or carbon , the reaction is _ geometrically _ very selective @xcite , and the removal of two nucleons will be enhanced if nucleon pairs have a strong spatial correlation ( and localization ) in the projectile ground state .
an interesting question therefore is whether like and unlike two - nucleon removal under such conditions also exhibits any evidence of enhanced @xmath2 over @xmath14 and @xmath15 spatial correlations on a longer length scale than implied by the ( src ) observations of the electron knockout data ? here , the spatial ( geometrical ) selectivity of the reaction mechanism would provide the leverage and probe of the presence of spatially localized pairs and evidence of an enhanced @xmath2 correlation .
experimental data are available in the form of high - energy , primary beam measurements of the inclusive cross sections to the bound states of the residues after @xmath2 , @xmath14 and @xmath15 removal @xcite .
we will show that these data reveal a significant enhancement of unlike - pair yields , @xmath16 , relative to those , @xmath17 and @xmath18 , for like - nucleon pairs , the enhancement being significantly greater than would be expected ( trivially ) from the numbers of such pair combinations available .
the removal of two ( well - bound ) nucleons of the deficient species from asymmetric nuclei has been shown to proceed as a direct reaction @xcite . a description of the reaction making use of configuration - mixed shell - model wave functions and elastic and inelastic breakup contributions to the removal cross section has been shown to reproduce experiment @xcite .
such calculations have also recently been extended to describe the reaction - residue momentum distributions @xcite , where it was shown that the shapes and widths of the residue momentum distributions are indicative of both the total angular momentum ( @xmath19 ) and the total orbital angular momentum components @xmath20 ) of the removed nucleon pair @xcite . here
we will exploit the eikonal reaction model in the isospin formalism @xcite for the removal of like ( @xmath21 ) and unlike ( @xmath22 ) nucleon pairs .
we also discuss the possible role of indirect population of the residue final states of interest , i.e. by single - nucleon knockout paths that populate particle - unbound states of the intermediate mass @xmath13 + 1 system .
such indirect paths severely limit the number of systems that , in the absence of an empirical means to distinguish between direct and indirect knockout events , may be studied quantitatively with unlike pair knockout .
these indirect contributions to final state yields are expected , in general , to be large in @xmath2 removal from asymmetric systems @xcite where one or other of the nucleon thresholds will be at relatively low excitation energy .
attractive @xmath2-removal test cases thus suggest symmetric light nuclei , e.g. or . for such less - massive cases one must however consider core - recoil effects in the reaction and the analogous center - of - mass corrections to the ( fixed center ) shell - model two - nucleon amplitudes .
both are expected to affect , in detail , the absolute magnitudes of the cross sections .
these are discussed here .
regardless of these two ( relatively small ) effects , comparisons of the relative like- and unlike - pair knockout yields may be made . in section
[ sec : carbon ] we outline the specific features of the reactions using @xmath0c projectiles .
the necessary formalism has been presented elsewhere and will only be outlined in section [ sec : formalism ] , exploiting the notation used in previous work .
the calculated results for are discussed in section [ sec : results ] and a summary made in section [ sec : summary ] .
our current expectation is that a quantitative discussion of pair - correlation effects can only be made when the residue final states are populated predominantly via a single - step direct reaction .
it is thus vital to select examples that minimize the indirect ( evaporative ) contributions to the reaction yield : and that we assume can not ( at present ) be distinguished from the direct reactions of interest .
we must therefore consider systems with large ( and symmetric ) nucleon separation thresholds , such that the single - nucleon removal strength to particle - unbound states will be weak .
for the same reason , we require the projectile nucleus to be relatively light to minimize population of the ( bound ) final residue states via the removal of deeply bound ( non - valence ) nucleons .
ideally , any experiment would also measure single - nucleon and like - pair removal in addition to the unlike pair removal , in order to verify shell- and reaction - model predictions for the distribution of single - nucleon spectroscopic strength and for the direct @xmath21 pair knockout .
the above considerations severely constrain the potential candidates , the best examples being and .
the former of these will be considered in detail here .
the choice of an @xmath23 projectile poses the additional ( experimental ) complication of distinguishing the mass @xmath13 residue from the incident beam , the residue and projectile having identical mass - to - charge ratios .
consideration of two - nucleon knockout from @xmath0c is valuable for two reasons .
first , its shell model description and that of the residual nuclei @xmath1c , @xmath1be and @xmath1b are extensively studied and so establish a valuable point of reference .
second , the existing experimental cross sections for two - nucleon knockout from @xmath0c @xcite are accurate to @xmath24 10% and were taken at high energies , where the eikonal model used here is at its most reliable .
the experimental data were obtained using reactions of a carbon beam on a carbon target at 250 , 1050 and 2100 mev per nucleon incident energies .
these show an expected enhancement of the @xmath1b production cross sections ( @xmath2 removal ) over those for the @xmath14 ( to @xmath1c ) and @xmath15 ( to @xmath1be ) removal reactions ( see table [ 12c ] ) .
the significance of this observed enhancement is quantified here .
in addition to two - nucleon removal , single - nucleon removal cross sections were also measured at high energy ( and previously studied theoretically , see ref .
these verify the expectation , from the shell - model , that the majority of single - particle removal strength is exhausted in transitions to final states lying below the nucleon separation thresholds of @xmath25c and @xmath25b .
this point will be developed further in later sections .
the relevant one- and two - nucleon ( and @xmath26-particle ) separation thresholds are illustrated in fig .
[ fig : c12_seps ] .
-particle separation thresholds relevant to nucleon - removal reactions from .
the two - nucleon separation energies are @xmath27 , @xmath28 and @xmath29 mev .
proton evaporation from the single - neutron removal residue @xmath25c would be expected to be the largest indirect pathway , but the shell - model calculations suggest very little spectroscopic strength to states above 8.689 mev in . [
fig : c12_seps ] ] the simplest of the three two - nucleon removal cases is two - neutron ( @xmath14 ) removal , leading to the @xmath1c residual nucleus .
this has only two bound final states below the first proton threshold , the 0@xmath30 ground state and a 2@xmath30 excited state at 3.354 mev @xcite .
the level scheme of @xmath1be , the residual nucleus in the @xmath15 removal case , is only slightly more complex .
we must consider population of the 0@xmath30 ground state , the 2@xmath30 states at 3.368 and 5.958 mev , and a second 0@xmath30 state at 6.179 mev , all below the neutron threshold of 6.812 mev .
the most complicated final state is that for @xmath1b , the @xmath2 knockout residue . the @xmath31-shell
shell - model calculations used include even - parity states up to a maximum of spin 3 .
for we have shown states up to the proton separation threshold , however the low @xmath26-particle separation threshold means that the @xmath9 states at 4.774 ( 3@xmath30 ) , 5.180 ( 1@xmath30 ) and 5.920 mev ( 2@xmath30 ) are reported to decay ( with branching ratios of near 100% ) by @xmath26-emission .
we will show the cross sections for population of these states but will assume that they do not contribute to the calculated yield .
the @xmath21 , 2@xmath30(5.184 mev ) state is reported to have a 16% @xmath26 emission branch .
these states are illustrated in fig .
[ fig : levels ] .
we note that the spectrum also contains several negative parity states which are not expected to be populated by the nucleon removal reaction mechanism .
residues populated in the two - nucleon knockout .
the spin and isospin labels ( @xmath32,@xmath33 ) are indicated .
all states included are of positive parity .
levels assumed to be part of the isospin multiplet are connected by dashed lines .
the lowest particle thresholds are also indicated .
states above the @xmath26-particle threshold in are expected to decay via @xmath26 emission , with the exception of the 5.164 mev , @xmath21 , @xmath34 state which has an 84% @xmath11-decay branch .
[ fig : levels ] ] direct , unlike - pair - removal cross sections are expected to be larger than those for like - pairs . assuming the simplest @xmath35shell @xmath36 ^ 4 \nu[0p_{3/2}]^4 $ ] structure for , the pn - format tnas for pair removal to a residue final state with spin @xmath32 are given by the appropriate coefficients of fractional parentage , with value @xmath37 .
the cross sections to a given final state , being proportional to the square of these tna , suggest unlike : like two - nucleon inclusive cross section ratios @xmath38 if all strength leads to bound final states . the same result is obtained from the counting of available valence - nucleon pairs .
the primary motivation for the present study is that this simple combinatorics expectation is not corroborated by the available data ; the ratio @xmath39 ( @xmath40 ) from the experiments at 250 , 1050 and 2100 mev / nucleon are 8.1 ( 8.9 ) , 5.3 ( 6.3 ) and 6.0 ( 8.5 ) , respectively @xcite , the observed enhancement of the @xmath2 removal cross section being significantly larger than expected .
clearly our simple estimate , from the number of available pairs of each type , assumes that all of the removal strength leads to bound configurations in the residue of interest , which is not the case , and part of the enhancement noted above may be attributable to this ( channel - dependent ) fraction of events leading to particle - unbound states .
here we seek to address this question quantitatively , by use of a direct reaction model for two - nucleon removal .
specifically , our aims are to quantify : ( i ) whether this enhancement is accounted for by those two - nucleon correlations that are included in the truncated , @xmath31-space shell model calculation , and ( ii ) to make a simple estimate of the contributions one might reasonably expect from indirect reaction pathways in this model picture .
the formalism used is based on that developed in refs .
isospin - format two - nucleon amplitudes ( tnas ) will be used and are expected to offer a good description of the light symmetric systems considered here .
thus , in the unlike pair removal case , we assume a common set of nucleon orbital wave functions , to be discussed below .
we evaluate the cross sections for transitions from the projectile initial ( ground ) state @xmath41 , with spin @xmath42 , to particular residue final states @xmath43 .
the residue is assumed to be a spectator in the sudden reaction description and its state is not coupled to the reaction dynamics .
the direct reaction will then probe the two - nucleon overlap .
we denote the @xmath13-body final states by @xmath44 , where the label @xmath45 includes the angular momentum projection @xmath46 .
the two - nucleon wave function ( two - nucleon overlap ) of the removed nucleons 1 and 2 is written as @xmath47 , where @xmath48_{i \mu}^{t\tau}\ .
\label{eqn : overlap}\end{aligned}\ ] ] here the set of available active two - nucleon configurations , with counter @xmath26 , consists of particular pairs of single - particle orbitals [ @xmath49 , where the index @xmath50 denotes the state s spherical quantum numbers .
the @xmath51 are the two - nucleon amplitudes ( tnas ) that express the parentage of the residue final state , when coupled to a particular two - nucleon configuration @xmath26 , in the projectile ground state . here , as elsewhere , these are taken from a truncated - basis shell model calculation
. expressed in @xmath52-coupling , the antisymmetrized two - nucleon wave function of eq .
( [ eqn : overlap ] ) is @xmath53_{i\mu}^{t \tau } = & d_{\alpha}\hat{j}_1\hat{j}_2 \sum_{\substack{l\lambda{s\sigma}\\ \lambda_1\lambda_2}}{(\ell_1\lambda_1\ell_2\lambda_2|l\lambda ) } { ( l\lambda{s}\sigma|i\mu ) } \ , \hat{l}\hat{s } \ , \chi_{s\sigma } ( 1,2 ) \chi_{t\tau } ( 1,2)\nonumber\\ \times & [ { \phi_{\ell_{1}j_{1}}^{\lambda_{1}}({1 } ) } { \phi_{\ell_{2}j_{2}}^{\lambda_{2}}({2})}-(-)^{s+t } { \phi_{\ell_{1}j_{1}}^{\lambda_{1}}({2})}{\phi_{\ell_{2}j_{2}}^{\lambda_{2}}({1})}]\left\ { \begin{array}{ccc } \ell_1 & s & j_1 \\
\ell_2 & s & j_2 \\ l & s & i \\ \end{array } \right\}\ , \label{eqn : anti_wf_ls}\end{aligned}\ ] ] with @xmath54 . the angular momentum and isospin couplings used
are summarized in fig .
[ fig : coupling ] .
the nucleon - wave functions @xmath55 are @xmath56 -couplings used in this paper are presented .
[ fig : coupling ] ] we note that no explicit account is taken of additional ( short , medium , long range ) correlations other than those contained within the shell - model description , and that enter eq .
( [ eqn : overlap ] ) .
these are seen to arise from : ( a ) antisymmetry and angular momentum coupling of the nucleon pair , and ( b ) the ( shell - model ) two - nucleon overlap , via the weights and phases of the contributing tnas .
if there are significant additional strong - interaction generated pair correlations , missing from our description , then we might expect empirical cross sections to deviate significantly ( and be enhanced ) relative to the shell - model - correlated model used here .
spectroscopic factors calculated within the fixed - center shell - model basis require a center - of - mass correction factor to be applied @xcite . essentially , the shell - model two - nucleon overlaps are calculated relative to the centre of mass of all @xmath10 nucleons and not relative to the remaining @xmath13 nucleons .
such corrections were discussed by pinkston @xcite ( see also ref .
@xcite ) . in previous , predominantly @xmath57-shell applications
such corrections were expected to be small , but for the examples discussed here they may be more significant , particularly when one has an ambition to compare absolute cross sections with experiment . in the simplest ( harmonic oscillator ) limit ,
the requirement is to multiply the shell - model tnas by a factor @xmath58^{n/2}$ ] , with @xmath59 the number of oscillator quanta of the orbitals of the active nucleons . in the present case , of ( @xmath602n )
, this would enhance all cross sections by @xmath61 , independent of the reaction channel and the residue .
thus , conclusions regarding the _ relative strengths _ of the different two - nucleon removal channels will be unaffected and we do not apply these corrections . clearly , a more complete understanding will be necessary for a precise consideration of the absolute cross sections .
we exploit eikonal reaction dynamics with the required nucleon- and residue - target @xmath62-matrices calculated in the optical limit of glauber theory @xcite .
these elastic @xmath62-matrices are thus calculated assuming that the residue and nucleons travel on straight line paths through the interaction field of the target .
the residue - target @xmath62-matrices are calculated by double - folding their densities with an effective nucleon - nucleon ( nn ) interaction .
the absorptive nature of these interactions naturally localizes the reaction to the projectile surface @xcite .
a detailed derivation of the two - nucleon removal cross section is presented elsewhere @xcite , as is the formalism for their momentum distributions @xcite .
these are not reproduced here .
the key approximations are as follows .
the removal of nucleons is sudden , i.e. their co - ordinates are assumed to vary slowly compared to the timescale of the reaction , and can be assumed frozen during the interaction . in previous work
we have also ( reasonably ) made the no - recoil ( heavy residue ) approximation in the diffraction - stripping terms of the removal cross section @xcite ; i.e. we have assumed coincidence of the residue and projectile centers of mass and impact parameters , @xmath63
. these core recoil effects will certainly introduce corrections for the lighter residues in this work , however , as will be quantified later , at the high projectile energies considered here these diffractive - stripping terms contribute a significantly smaller fraction of the cross section than the two - nucleon stripping terms . thus , recoil will have a relatively minor effect on the calculated removal cross sections and the discussion of this paper .
this effect will be fully quantified in future work .
the experimental information available consists of final state inclusive cross sections to bound states of the mass @xmath64 residues and some associated residue momentum distribution information .
we are able to identify each known positive parity final state ( @xmath65 ) and negative parity final state ( @xmath66 ) with a well - defined shell model state .
the shell - model calculations were performed with oxbash @xcite .
we will consider the results when using the wbp interaction @xcite , and also compare these with the pjt interaction @xcite outcomes , in the @xmath35shell model space .
the former interaction was used in previous work on nucleon removal from @xcite .
the shell - model spectroscopic factors @xmath67 and the two - nucleon spectroscopic amplitudes , the @xmath68 , as enter eq .
( [ eqn : overlap ] ) , were calculated for each residue final state . to discuss possible indirect processes
we must consider the population of highly - excited , unbound states in the single - nucleon removal cases .
the @xmath35shell model space allows removal of nucleons from the active @xmath69 and @xmath70 single - particle orbitals .
the radial wave functions used to describe these nucleonic states are calculated in a woods - saxon potential well .
the radius and diffuseness parameters were fixed to be @xmath71 fm and @xmath72 fm , as were used in the single - nucleon knockout analysis of ref .
in addition , we include a spin - orbit potential of depth @xmath73 mev . the ground - state to ground - state @xmath14 , @xmath15 and @xmath2 separation energies , @xmath74 , from are 31.841 mev , 27.184 mev and 27.41 mev , respectively .
the nucleon bound state wave functions for each reaction and transition , with final - state excitation energy @xmath75 , were calculated using effective nucleon separation energies of @xmath76 for single nucleon removal and @xmath77 for each nucleon in two - nucleon removal . in the @xmath2-removal case we assumed a common set of radial wave functions for the neutrons and protons .
these were calculated as above but assuming each nucleon carried a charge of 0.5@xmath78 .
the required nucleon- and residue - target @xmath62-matrices were calculated using the optical limit of glauber theory @xcite - i.e. by double folding each constituent particle and target density with an effective nucleon - nucleon ( nn ) interaction .
the essential input parameters , given the dominance of the stripping mechanism , are the free @xmath14 ( @xmath79 ) and @xmath2 cross sections , taken from the parametrization of ref .
@xcite , and the residue and target densities .
the former are given in table [ param ] .
the effective nn interaction was assumed to be zero - range , represented by a @xmath80-function , and the real : imaginary forward scattering amplitude ratios @xmath81 and @xmath82 were obtained from a polynomial fit to the values tabulated in ref .
these are listed in table [ param ] for the three energies of interest .
the point - nucleon density distributions of the target and the residues were assumed to have gaussian shapes , with root mean squared ( rms ) radii consistent with glauber - model analyses of the measured interaction cross sections @xcite . for simplicity , we assume a single matter rms radius for each mass nucleus ; for @xmath0c target and @xmath66 and @xmath65 residues , the mass radii were 2.32 , 2.11 and 2.30 fm respectively .
the latter value is that quoted for in ref .
@xcite .
.nucleon - nucleon effective interaction parameters used for the calculation of the nucleon- and residue - target @xmath62-matrices .
[ param ] [ cols="^,^,^,^,^",options="header " , ] ( dashed line ) , @xmath83 ( dot - dashed line ) and @xmath84 ( dotted line ) .
the inclusive cross section , their sum , is shown by the black solid line .
the data is for a target whereas the calculations assume a target .
the calculations have been offset by @xmath6030 mev / c and the have been scaled to match the experimental two - proton removal cross section on the target , of 5.97 mb .
the tnas used were calculated using the wbp interaction.[fig : greiner ] ] the calculated residue momentum distributions for two - proton removal are compared to the data in fig .
[ fig : greiner ] , the data having been scanned from figure 1 of ref .
the data provided are for a beryllium target while the calculations assume a carbon target .
any differences in the momentum distributions are expected to be very small for these two light targets .
both the exclusive contributions from the four final states involved and their sum are shown .
the figure shows clearly that the inclusive gaussian - like distribution is formed of exclusive cross sections of very different widths for the different final states .
thus , residue momentum distributions for these individual final states would be of considerable value in validating , or revealing deficiencies in the details of the shell - model wave functions and tnas used here , including our description of the reaction as a direct single - step process . details of these calculated full width at half maximum ( fwhm ) widths in the case of @xmath2-removal at 2100 mev per nucleon , for the six @xmath11-decaying final states of ,
are shown in fig .
[ fig:10c ] .
results from both the wbp ( solid lines and filled points ) and pjt ( dashed lines and open points ) shell - model interaction tna are shown .
the widths from the two interactions are remarkably similar for all transitions , indicative that the @xmath52-contributions to each final state are very similar @xcite .
the momentum distributions for the first and second @xmath9 , 1@xmath30 states of were also presented in figure 3 of ref .
@xcite , calculated from the wbp interaction .
the corresponding exclusive cross sections ( upper panel of the figure ) also show very similar trends for the two shell - model interactions .
they differ in detail , however , particularly with regard the ratio of the cross sections to the ground @xmath85 and first excited @xmath86 states .
we note also that the two interactions predict very similar cross sections in the two @xmath21 states , the differences being more evident in the @xmath9 states yields .
once again , this suggests interesting information would be gained from exclusive final states yields .
-decaying final states of the residues , for reactions at 2100 mev per nucleon .
the tnas used were obtained using the wbp interaction ( solid lines and solid points ) and the pjt interaction ( dashed lines and open points ) .
the significant variations predicted : ( i ) for the widths of the momentum distributions to the different final states , and ( ii ) in the sensitivity of the , @xmath9 final state yields to the effective interaction used , are evident . ]
the predicted final state - dependent variations of the fwhm widths from the model used also offer considerable scope for further testing of the reaction inputs , the shell - model wave functions used , and of the nucleon - nucleon correlations content of these wave functions .
we have considered in detail theoretical expectations for the cross sections of two - nucleon removal reactions , from incident on a carbon target , at beam energies of 250 , 1050 and 2100 mev per nucleon .
the calculated inclusive cross sections for two like - nucleon ( @xmath21 ) removal are broadly consistent with available experimental data . for @xmath2-pair
removal the analogous @xmath21 , final states are also shown to be insensitive to the shell - model interactions used .
the calculated inclusive @xmath2-pair removal cross sections at all energies _ underestimate _ the data by approximately a factor of two .
theoretical calculations of the widths of the final - state - inclusive residue momentum distributions on the other hand are consistent with the available experimental data , including for the @xmath2-removal channel .
further measurements , of final - state - exclusive cross sections and residue momentum distributions , would allow a much more detailed scrutiny and validation of the direct nature of the reaction , including the identification of any indirect reaction components .
the calculated @xmath2-removal cross sections to the @xmath9 , final states were also shown to have sensitivity to the shell - model effective interaction used ; for example , the ratio of the calculated cross sections to the ground @xmath85 and first @xmath86 excited states .
the overall conclusion from the present analysis is that the existing data suggest that the @xmath9 , @xmath2-spatial correlations present in the wave functions used are insufficient .
we have shown that exclusive measurements would offer a means to interrogate these shell - model inputs , in particular the @xmath2-channel , @xmath9 wave functions , and the direct reaction mechanism predictions in considerable detail .
a similar two - nucleon knockout study can also be performed using an projectile , for which np - pair knockout using electromagnetic probes has been measured @xcite and that supports the src observations for . in this case
the residue is also of interest , given that only the @xmath87 ground state is bound , and where the results of refs .
@xcite indicate a correspondingly small cross section and a narrow momentum distribution , consistent with theoretical expectations . additionally , possible @xmath88 components in the ground state wave function add interest to the magnitudes of the absolute cross sections in this case . the assistance of b.a .
brown and p. fallon is acknowledged .
this work was supported by the united kingdom science and technology facilities council ( stfc ) under grant no .
st / f012012 .
e.c.s is grateful for support from the united kingdom engineering and physical sciences research council ( epsrc ) under grant number ep / p503892/1 .
d. bazin , b.a .
brown , c.m .
campbell , j.a .
church , d.c .
dinca , j. enders , a. gade , t. glasmacher , p.g .
hansen , w.f .
mueller , h. olliver , b.c .
perry , b.m .
sherrill , j.r .
terry , j.a .
tostevin , phys .
91 * , 012501 ( 2003 ) . | reactions that involve the direct and sudden removal of a pair of like or unlike nucleons from a fast projectile beam by a light target nucleus are considered .
specifically , we study the three two - nucleon removal channels from @xmath0c that populate final states in the @xmath1be , @xmath1b and @xmath1c reaction residues .
the calculated two - nucleon removal cross sections and the residue momentum distributions are compared with available high energy data at 250 , 1050 and 2010 mev per nucleon , data that are inclusive with respect to the bound final - states of the residues .
the measured @xmath2-removal cross sections only are significantly greater than the values calculated , suggesting that the reaction mechanism observes enhanced @xmath2 spatial correlations compared to those present in the shell - model wave functions . |
the stochastic model considered in this paper is a generalization of the stacked contact process introduced in @xcite and studied analytically in @xcite .
the stacked contact process is a spatial stochastic process based on the framework of interacting particle systems that describes the co - evolution of a population of hosts together with their symbionts .
individuals are located on the @xmath1-dimensional integer lattice and interact with their nearest neighbors .
the model assumes that the symbionts can only live in association with their host ( obligate relationship ) and are transmitted both vertically from associated hosts to their offspring and horizontally from associated hosts to nearby unassociated hosts .
the stacked contact process @xcite also assumes that all the hosts give birth and die at the same rate regardless of whether they are associated with a symbiont or not , meaning that the symbionts have no effect on the fitness of their host .
in contrast , this paper considers the natural generalization of the stacked contact process in which associated and unassociated hosts have different birth rates : symbionts that increase the birth rate of their host , and therefore have a beneficial effect , are referred to as a mutualists , whereas symbionts that decrease the birth rate of their host , and therefore have a detrimental effect , are referred to as pathogens .
formally , the state of the system at time @xmath2 is a spatial configuration @xmath3 where state 0 means empty , state 1 means occupied by an unassociated host , and state 2 means occupied by a host associated with a symbiont .
letting @xmath4 be the fraction of nearest neighbors of vertex @xmath5 which are in state @xmath6 , hosts and symbionts co - evolve according to the spin system whose transition rates at vertex @xmath5 are given by @xmath7 the first four transition rates indicate that unassociated hosts give birth at rate @xmath8 , hosts associated with a symbiont give birth at rate @xmath9 and , regardless of whether they are associated or not , all the hosts die at the normalized rate one .
an offspring produced at @xmath5 is sent to a vertex chosen uniformly at random among the nearest neighbors but the birth is suppressed when the target site is already occupied , which models competition for space .
the offspring is always of the same type as its parent , indicating that the symbiont is always transmitted vertically .
the process described by these four transitions is the multitype contact process @xcite .
the effect of the symbiont on the host is modeled by the choice of the two birth rates : the symbiont is @xmath10 the last two transition describe the symbiont dynamics within the host population . the symbiont spreads to adjacent unassociated hosts at rate @xmath11 , which corresponds to a horizontal transmission of the symbiont .
finally , hosts associated with a symbiont become unassociated at rate @xmath12 , which we simply call the recovery rate even when the symbiont is a mutualist .
the stacked contact process @xcite is obtained by setting @xmath13 .
this corresponds to the neutral case in which the symbionts have no effect on the fertility of their hosts , i.e. , all the hosts have the same birth rate .
the analysis of this special case in @xcite is somewhat facilitated by the fact that the process is attractive and monotone with respect to its parameters .
in contrast , in the general case when @xmath14 , standard couplings to compare processes starting from different configurations or with different parameters fail to show attractiveness and monotonicity . + * mean - field approximation . * before studying the spatial stochastic process , we first look at its non - spatial deterministic counterpart called mean - field approximation .
this new model is obtained by assuming that the population is well - mixing , which results in a system of ordinary differential equations for the density @xmath15 of vertices in state @xmath16 . since @xmath17 , in order to study the mean - field model , it suffices to focus on two densities , say @xmath18 and @xmath19 .
for the transition rates given in , this system consists of the following coupled equations : @xmath20 clearly the feasible region @xmath21 is forward invariant . in addition
, the line segment @xmath22 is forward invariant , as well as the line segment @xmath23 when @xmath24 .
three boundary equilibria are possible : 1 .
@xmath25 , 2 .
@xmath26 and 3 .
@xmath27 . to account for the additional invariant line segments ,
define @xmath28 then we have the following ( fairly complete ) description of the dynamics .
[ th : mf ] under the following conditions , the stated equilibrium @xmath29 is globally stable on @xmath30 . 1 .
[ mf : ext ] ( extinction ) @xmath31 , if @xmath32 ( here @xmath29 is stable on all of @xmath33 ) .
[ mf : coex ] ( coexistence ) some unique @xmath29 in the interior of @xmath33 , if either 1 .
[ mf : coex1 ] @xmath34 , @xmath35 and @xmath36 , 2 .
[ mf : coex2 ] @xmath34 , @xmath37 and @xmath38 3 .
[ mf : coex3 ] @xmath24 , @xmath37 , inequality holds and either @xmath39 or @xmath40 3 .
[ mf:1s ] ( 1s win ) @xmath41 , if @xmath24 , @xmath37 , does not hold and holds .
[ mf:2s ] ( 2s win ) @xmath42 , if @xmath24 , @xmath43 , does not hold and holds .
some of these conditions can be intuitively understood as follows : * if @xmath37 , then inequality says that the 2s can invade the 1s in equilibrium . *
if @xmath24 and @xmath43 , then inequality says that the 1s can invade the 2s in equilibrium .
theorem [ th : mf ] is proved in section [ sec : mf ] . +
* spatial stochastic process .
* as in the mean - field model , the spatial stochastic process can exhibit four regimes : either the host population survives or goes extinct and , in case of survival , the symbiont population either goes extinct , occupies some but not all hosts , or occupies all hosts .
in addition , when the recovery rate @xmath34 , survival of associated hosts implies coexistence of associated and unassociated hosts .
as pointed out previously , the stacked contact process obtained by setting @xmath44 has several nice properties such as attractiveness and monotonicity .
in contrast , these properties do not seem to hold for our process , i.e. , standard couplings fail to show these properties .
coupling can however be used to compare the process with other popular interacting particle systems and collect interesting results .
we start by comparing the process with the basic and the multitype contact processes using simple coupling techniques .
we also show that the process inherits some of the properties of the forest fire model though , because of the lack of monotonicity , this does not simply follow from a standard coupling argument . in the limiting case
when the recovery rate @xmath45 , all the symbionts die instantaneously so the host dynamics reduces to the basic contact process @xmath46 there exists a critical value @xmath47 such that above this critical value the host population survives whereas at and below this critical value the population goes extinct @xcite .
assume from now on that the recovery rate is finite and let @xmath48 be the process that keeps track of the hosts and the process that keeps track of the hosts associated to a symbionts , respectively .
the transitions for the first process satisfy @xmath49 while @xmath50 one . in particular , this process can be coupled with the basic contact process described above to deduce that , for all @xmath51 , @xmath52 this follows from theorem iii.1.5 in @xcite , which applies to general two - state spin systems , together with obvious inequalities relating the transition rates of our process and their counterpart for the basic contact process .
similarly , the transitions for the second process satisfy @xmath53 while @xmath50 at rate @xmath54 , from which it follows that , for all @xmath51 , @xmath55 the four parameter regions in are illustrated in the four diagrams of figure [ fig:3d - diag ] .
so far , the behavior of the stochastic process agrees with the behavior of the mean - field model described in theorem [ th : mf ] , if we think of the mean - field model as having @xmath56 note that @xmath57 implies holds . setting @xmath58 , the process reduces to the multitype contact process completely analyzed when the death rates are equal in @xcite .
the transition rates become @xmath59 in this case , the type with the larger birth rate outcompetes the other type provided its birth rate is also strictly larger than the critical value of the single - type contact process .
this result has first been proved in @xcite using duality techniques and again in @xcite using also a block construction in order to prove that the long - term behavior of the process is not altered by small perturbations of the parameters .
using in addition a coupling argument , it can be easily deduced from ( * ? ? ? * propositions 3.13.2 ) that , the recovery rate @xmath12 being fixed , for all @xmath51 , @xmath60 the parameter region in is shown in the bottom left diagram of figure [ fig:3d - diag ] .
although it is not recorded in theorem [ th : mf ] , this behavior also agrees with that of the mean - field model , which can be verified by an examination of the nullclines described in section [ sec : mf ] here the assumption that @xmath11 is small is important .
the forest fire model , also referred to as epidemics with recovery , is the three - state spin system with a cyclic dynamics described by the following three transitions : @xmath61 the three states are interpreted as 0 = alive , 2 = on fire and 1 = burnt , but can also be thought respectively as healthy , infected and immune in the context of epidemics .
this process has been studied in @xcite which proves the existence of a critical value @xmath62 such that , in @xmath63 , @xmath64 because the dynamics is cyclic , basic couplings between the forest fire model and our process do not lead to any useful stochastic ordering between the two systems .
however , the proof in @xcite easily extends to our process in a certain parameter region . indeed ,
in addition to general geometrical properties and percolation results which are not related to the specific dynamics of the forest fire model , the key estimates in @xcite rely on the following two ingredients : * the set of burning trees dominates its counterpart in the process with no regrowth ( @xmath65 ) provided both processes start from the same configuration . * in regions that have not been on fire for at least @xmath66 units of time , the set of trees which are alive dominates a product measure with density @xmath67 .
now , fix @xmath68 , let @xmath69 and consider the spin system on the two - dimensional integer lattice whose dynamics is described by the five transitions @xmath70 in this process , trees burn for an exponential amount of time with rate @xmath71 as in the original forest fire model .
it follows that the domination property ( a ) remains true : the set of burning trees in this new process dominates its counterpart in the forest fire model with no regrowth and in which the fire spreads by contact at rate @xmath72 . since the transition @xmath50 again occurs spontaneously at rate @xmath73 , the domination of the product measure ( b ) remains true as well .
in particular , @xmath74 this holds for all @xmath75 . since the proof in @xcite
is based on a block construction , which supports small perturbations of the system , we also obtain coexistence in the process under the same assumptions and provided @xmath8 is sufficiently small . in conclusion , in @xmath63 , @xmath76 the parameter region in is shown in the two diagrams on the right of figure [ fig:3d - diag ] .
we now focus on the parameter region where @xmath77 which is not covered by our previous results . in this case
, the host population survives without the symbiont , which is a pathogen , and the main question is whether associated and unassociated hosts coexist .
theorem [ th : mf ] says that in the mean - field model they do coexist , provided @xmath11 is large enough .
interestingly , our results show that for the stochastic process , the answer depends on the spatial dimension . to begin with , we show that , even when infected hosts are sterile , coexistence is possible in spatial dimensions @xmath0 but we also point out that our proof relies on geometric arguments that do not hold in one dimension . more precisely , we have the following theorem .
[ th:2d ] assume that @xmath34 and @xmath0 .
then , there is @xmath78 large such that @xmath79 note that , when @xmath80 , the result directly follows from so the theorem is only interesting when infected hosts have a low birth rate .
the proof relies on a block construction , a technique introduced in @xcite .
having a tiling of the lattice into large squares and letting @xmath81 and @xmath82 be two adjacent squares , the main ingredient is to prove that , whenever @xmath81 contains a large number of pathogens while @xmath82 does not have too many pathogens , * there is a large connected component void of pathogens included in @xmath82 and * the number of pathogens on the boundary of this connected component is large .
this is used together with the fact that @xmath34 to show that , in a short time , one of the infected hosts along the boundary recovers .
this recovery event is followed with probability close to one by a wave of births and then a wave of infections which , together , ensure that a positive fraction of the square @xmath82 becomes infected .
large deviations estimates are used to make this precise .
in contrast with theorem [ th:2d ] , when @xmath24 and @xmath83 , and even when the infection rate and the birth rate of healthy hosts are very large , the pathogen is unable to survive .
[ th:1d ] assume that @xmath77 and @xmath24 and @xmath83 .
then , starting from any configuration with infinitely many vertices in state 1 , @xmath84 the two parameter regions covered by theorems [ th:2d ] and [ th:1d ] are illustrated in the last three phase diagrams of figure [ fig:3d - diag ] .
we point out that theorem [ th:1d ] holds even when @xmath85 , i.e. , when the pathogen invades instantaneously nearby healthy hosts . also , in addition to the statement of the theorem , our proof gives specific estimates on the rate of extinction of the pathogens and the rate of expansion of the healthy hosts .
the first part of the proof shows that there exists a constant @xmath86 such that , uniformly in all configurations @xmath87 with infinitely many 1s and for any site @xmath5 , @xmath88 where @xmath89 refers to the law of the process starting from configuration @xmath87 . in other words , there exists a uniform ( over all sites ) exponentially growing ( in time ) neighborhood of any site which is eventually void of pathogens . to describe the long - term behavior of the healthy hosts ,
let @xmath90 denote the one - dimensional nearest - neighbor ( supercritical ) contact process with parameter @xmath8 starting from the all - one configuration .
also , let @xmath91 denote the edge speed in this contact process as defined in @xcite .
then , under the assumptions of the theorem , there exist * a random site @xmath92 and an almost surely finite time @xmath93 that depend on @xmath87 , * site - valued processes @xmath94 defined for @xmath95 and satisfying @xmath96 , * a coupling of the processes @xmath97 and @xmath90 such that , @xmath89-almost surely , @xmath98 \times [ t , \infty ) . \end{array}\ ] ] in other words , as long as @xmath87 has an infinite number of 1s then , @xmath89-almost surely , eventually there arises a stable population of 1s that behaves like the basic contact process on an interval that grows linearly in time . from , we also obtain a complete convergence theorem . indeed ,
letting @xmath99 denote the upper invariant measure of the contact process @xmath90 and @xmath100 denote the measure that concentrates on the all - zero configuration , since the distribution of the contact process converges weakly to @xmath99 ( * ? ? ?
vi ) , we deduce the following for the distribution @xmath101 of the process @xmath97 .
let @xmath77 and @xmath24 and @xmath83 . then , as @xmath102 , @xmath103 in particular , all invariant measures are convex combinations of @xmath100 and @xmath99 .
in this section we prove theorem [ th : mf ] . part [ mf : ext ] is easily proved by examining @xmath104 . adding the equations in and recalling @xmath105
, we find @xmath106 therefore , if @xmath32 , then @xmath107 which implies that @xmath108 as @xmath102 . since @xmath109
, this means that @xmath110 and part [ mf : ext ] is proved . to prove the other parts we need a few more ingredients .
[ lem : dulac ] suppose we have the planar system @xmath111 let @xmath112 be a simply connected region in the plane and @xmath113 be a @xmath114 function ( called a dulac function ) such that the divergence of @xmath115 is not zero and has constant sign on @xmath116 .
then , the system @xmath117 has no closed orbits on @xmath116 .
this is a standard result .
the proof uses green s theorem and can be found in @xcite , theorem 2 in section 3.9 .
[ th : pb ] let @xmath33 be a compact invariant set of the planar system @xmath117 with @xmath118 , such that there is at most a finite number of equilibria in @xmath33 . for @xmath119 , let @xmath120 denote the @xmath121-limit set of the positive semi - orbit starting from @xmath5 .
then the following trichotomy holds : 1 .
@xmath122 is an equilibrium , 2 .
@xmath122 is a periodic orbit , or 3 .
@xmath122 is the union of at least one equilibrium and at least one homoclinic or heteroclinic orbit , and any heteroclinic orbits form a single heteroclinic cycle connecting the equilibria in @xmath120 .
this is theorem 1.5 in @xcite , and is proved therein .
[ lem : omega ] if @xmath33 is a compact and simply connected invariant set of the planar system @xmath117 such that dulac s criterion holds on @xmath33 , and there are at most a finite number of equilibria in @xmath33 , then for any @xmath123 , @xmath120 is an equilibrium . under dulac
s criterion , in the trichotomy of theorem [ th : pb ] , option 2 is ruled out , and homoclinic orbits are ruled out .
suppose @xmath122 contains a heteroclinic orbit , then @xmath122 contains at least two equilibria .
this implies existence of a heteroclinic cycle connecting the equilibria in @xmath122 .
but this is also ruled out by dulac s criterion .
+ + we now return to our model .
letting @xmath117 denote and @xmath124 , @xmath125 is strictly negative on the interior of @xmath33 , therefore @xmath126 is a dulac function for .
thus in order to prove the rest of theorem [ th : mf ] , it is enough to identify all equilibria in @xmath33 , then to show that all of them except @xmath29 attract no points in @xmath30 .
if this can be shown , then by lemma [ lem : omega ] , trajectories have no choice but to converge to @xmath29 . +
as noted in the introduction , we have the three boundary equilibria , that we now label 1 .
@xmath127 , 2 .
@xmath128 and 3 .
@xmath129 . and these are the only equilibria on the boundary on @xmath33 (
note that @xmath130 if @xmath131 ) .
simple conditions ensure instability of these equilibria .
[ lem : unst ] in each case , the stated equilibrium attracts no points in @xmath30 .
1 . @xmath132 , if @xmath37 or @xmath36 .
2 . @xmath133 , if holds .
@xmath134 , if holds .
it is easy to check that 1 . if @xmath37 then @xmath135 , and if @xmath36 then @xmath136 .
2 . if holds then @xmath137 .
3 . if holds then @xmath138 . letting @xmath29 denote the relevant equilibrium , in each of the above cases , the result follows from the fact that * @xmath139 and * there exists a @xmath30-neighborhood @xmath140 of @xmath29 such that @xmath141 for all @xmath142 .
this completes the proof .
+ + next , we examine the nullclines . write as @xmath143 and @xmath144
. + + * @xmath145-nullcline .
* solving @xmath146 gives the pair of conditions @xmath147 we distinguish three cases : 1 . if @xmath148 then ( ii ) is void .
if @xmath149 and @xmath150 then ( ii ) reads @xmath151 .
3 . if @xmath152 then ( ii ) gives @xmath153 in conclusion , the @xmath19-nullcline is a pair of straight lines @xmath154 where @xmath155 and @xmath156
. + + * @xmath157-nullcline . * solving @xmath158 gives the quadratic @xmath159 we distinguish two cases depending on the value of @xmath12 .
+ * case 1 : @xmath160 .
* in this case , factors to give the pair of conditions @xmath161 we again distinguish three sub - cases : 1 .
if @xmath162 then ( ii ) is void .
if @xmath163 and @xmath150 then ( ii ) gives @xmath164 that does not intersect @xmath33 .
if @xmath165 then ( ii ) gives @xmath166 in conclusion , the intersection of the @xmath18-nullcline with @xmath33 is a pair of straight lines @xmath167 where @xmath168 and @xmath169 .
+ * case 2 : @xmath170 .
* let @xmath171 denote the intersection of the @xmath18-nullcline with @xmath172 , excluding the origin .
there are a couple of possibilities . 1 .
if @xmath162 then degenerates to @xmath173 , so @xmath174 .
2 . if @xmath175 then is hyperbolic , which can be checked from the hyperbolicity condition @xmath176 for a quadratic of the form @xmath177 . ignoring lower order terms gives @xmath178 which means there is a vertical asymptote and a slant asymptote of the form @xmath179 for the same @xmath180 as before , and some unknown @xmath181 .
since @xmath182 and thus @xmath183 , as a function of @xmath18 , the slant asymptote has a negative slope . ignoring the terms without @xmath19 , we find that the vertical asymptote is given by @xmath184 therefore , the vertical asymptote is @xmath185 and in particular , @xmath186 lies to the right of the line @xmath187 . since @xmath188
, the @xmath18-nullcline can not cross @xmath187 except possibly at @xmath132 .
in addition , @xmath189 from which it follows that the @xmath18-nullcline must intersect both @xmath132 and @xmath190 , and is otherwise disjoint from the straight line @xmath191 .
combining these observations , the only option is that @xmath171 is a single convex curve such that 1 . if @xmath192 then @xmath171 emerges from @xmath132 and asymptotes to @xmath186 , and 2 . if @xmath193 then @xmath171 emerges from @xmath133 and asymptotes to @xmath186 .
+ note that if @xmath186 intersects @xmath133 then @xmath194 .
we are now ready to prove the theorem .
+ + theorem [ th : mf ] + * part [ mf : ext]*. this was proved at the beginning of this section . + * part [ mf : coex]*. under any of the conditions in part [ mf : coex ] , from lemma [ lem : unst ] it follows that no boundary equilibria attract any points in @xmath30 . using dulac s criterion and lemma [ lem : omega ] , any trajectory in @xmath30 converges to an equilibrium in @xmath33 .
therefore , there must be at least one interior equilibrium . to obtain uniqueness
, it suffices to show in each case that @xmath195 is at most a single point . + * part [ mf : coex1 ] .
* suppose @xmath34 , @xmath35 and @xmath36 .
then @xmath196 is not vertical and intersects the line @xmath197 at the point @xmath134 that lies above @xmath132 .
if @xmath162 then @xmath171 is a straight line that intersects @xmath132 , so @xmath195 is at most a single point .
otherwise , @xmath171 emerges from @xmath25 into @xmath172 and asymptotes to @xmath186 .
since @xmath196 is a straight line and @xmath171 is convex , a quick sketch confirms that the intersection @xmath195 reduces a single point . + * part [ mf : coex2 ] .
* suppose @xmath34 , @xmath37 and holds .
regardless of the position of @xmath186 relative to @xmath133 , @xmath171 is a convex ( possibly straight ) curve emerging from @xmath133 into @xmath172 . if @xmath196 is vertical then , since @xmath198 the asymptote @xmath186 lies to the left of @xmath133 ,
so @xmath171 is not vertical and @xmath195 is at most a single point .
if @xmath196 is not vertical then , since it lies above @xmath133 , the intersection @xmath195 reduces to a single point . +
* part [ mf : coex3 ] .
* suppose @xmath24 , @xmath37 and holds .
in addition , since @xmath182 and @xmath193 , @xmath171 intersects the ray @xmath199 , and since condition holds , either @xmath196 lies to the left of @xmath133 if it is vertical or above @xmath133 if it is not vertical .
* suppose @xmath39 .
if @xmath196 is vertical then clearly @xmath195 is a single point .
if @xmath196 is not vertical then , since it lies above @xmath133 and intersects @xmath197 below @xmath132 , we must have @xmath200 , and again the intersection @xmath195 reduces to a single point . *
if @xmath43 and holds , then @xmath171 intersects @xmath197 above @xmath134 . since @xmath196 intersects @xmath191 above @xmath133 , @xmath171 and @xmath196 have different slopes , so @xmath195 is again a single point .
* part [ mf:1s ] .
* suppose @xmath24 , @xmath37 , does not hold and holds .
since @xmath37 , @xmath132 attracts no points in @xmath30 , and if @xmath134 is an equilibrium then , since holds , it attracts no points in @xmath30 . to conclude that all trajectories in @xmath30 converge to @xmath133
, it is enough to show there is no interior equilibrium .
since @xmath24 and @xmath165 , @xmath171 is a straight line with finite slope as a function of @xmath18 . if @xmath196 is vertical then , since does not hold , it either intersects or lies to the right of @xmath133 , in which case there is no interior equilibrium .
if @xmath196 is not vertical and @xmath39 then , since @xmath201 and does not hold , on @xmath202 , @xmath196 lies either on or below the line @xmath191 and there is no interior equilibrium .
if @xmath43 then , since holds , @xmath171 intersects @xmath197 above @xmath134 , so on @xmath30 , @xmath196 lies below @xmath171 and there is no interior equilibrium .
+ * part [ mf:2s]*. suppose @xmath24 , @xmath43 , does not hold and holds .
essentially the same argument as in part 3 , with the roles of type 1 and 2 exchanged , shows that @xmath134 attracts all points in @xmath30 .
+ the proof is complete .
throughout the paper , we think of the process as being generated from a percolation sub - structure , also called harris graphical representation @xcite .
that is , we attach independent poisson processes with appropriate rate to each vertex and oriented edge of the @xmath1-dimensional integer lattice .
the process is then constructed by assuming that , at the times of these poisson processes , either a birth or an infection or a death or a recovery occurs whenever the configuration of the system at that time is compatible with the event .
table [ tab : harris ] shows how to construct the percolation substructure when @xmath203 , in which case the symbiont is a pathogen .
ccp280pt rate & symbol & effect on the process + @xmath204 & @xmath205 & birth at @xmath206 when @xmath5 is occupied by a healthy host and @xmath206 is empty + @xmath207 & @xmath208 & birth at @xmath206 when @xmath5 is occupied and @xmath206 is empty + @xmath209 & @xmath210 & infection at @xmath206 when @xmath5 is infected and @xmath206 is occupied by a healthy host + 1 & @xmath211 at @xmath5 & death at @xmath5 when @xmath5 is occupied + @xmath12 & @xmath212 at @xmath5 & recovery at @xmath5 when @xmath5 is infected + note that the results in can be proved by coupling different processes using this graphical representation rather than theorem iii.1.5 in @xcite . for instance , the contact process @xmath213 with parameter @xmath9 can be constructed from the graphical representation in the table by assuming that births can only occur through type 2 arrows , while the contact process @xmath214 with parameter @xmath8 can be constructed by assuming that births occur through both type 1 and type 2 arrows . constructing our modified stacked contact process and these two contact processes from this common graphical representation results in a coupling such that @xmath215 at all times @xmath2 provided this holds at time zero .
this shows that when @xmath203 .
this property when the inequality is reversed as well as and are proved similarly by using other graphical representations which are designed based on the ordering of the parameters .
the background idea to prove theorem [ th:2d ] is a block construction , which consists in coupling the process suitably rescaled in space and time with oriented site percolation .
this technique has been introduced in @xcite and is reviewed in @xcite . even though it has been widely used in the field of interacting particle systems , finding
an appropriate couping is rarely straightforward because it strongly depends on the process under consideration . to prove the theorem
, we need a series of estimates , starting with some geometrical properties of the configurations . + for simplicity , we focus on the two - dimensional case but our proof easily extends to higher dimensions .
note however that the proof relies on geometrical properties which are not true for the one - dimensional system . to begin with ,
we let @xmath216 be a large integer to be fixed later and partition the two - dimensional lattice using the collection of boxes @xmath217 letting @xmath218 be the first unit vector , the first step of the proof is to show that if the number of symbionts in the spatial region @xmath219 is not too small while the number of symbionts in @xmath220 is not too large , then there exists a connected component void of symbionts that contains many vertices in the region @xmath220 and whose boundary contains many symbionts . to make this statement precise , for every configuration @xmath87 of the system , we introduce @xmath221 then , we have the following lemma .
[ lem : geometry ] assume that @xmath222 and @xmath223 .
then , @xmath224 to prove the first inequality , we first observe that at most @xmath225 rows in @xmath220 can be occupied by a symbiont so at least @xmath226 rows have no symbiont .
reasoning similarly with the columns , we deduce that there exist coordinates @xmath227 such that @xmath228 for all vertices @xmath229 belonging to @xmath230 since each of the columns @xmath231 intersects each of the rows @xmath232 , the union of all these sets forms a connected set .
in addition , the number of vertices in this connected set exceeds half of the number of vertices in the box @xmath220 .
more precisely , @xmath233 the right - hand side of figures [ fig : geometry-1][fig : geometry-2 ] , in which the grey squares represent the symbionts , shows a picture of @xmath234 in white dots . in particular , any connected component of the set of vertices not occupied by a symbiont that does not contain @xmath234 has less vertices than @xmath234 .
it directly follows that @xmath235 to prove the second inequality , we let @xmath236 be the coordinates of the rows , respectively , the coordinates of the columns , in @xmath219 that have at least one symbiont . when there are @xmath237 symbionts in @xmath219 , we must have @xmath238 where @xmath239 and @xmath240 .
assume ( a ) and let @xmath241 the horizontal coordinates @xmath242 for all @xmath243 are shown in the rightmost grey squares on the left - hand side of figure [ fig : geometry-1 ] . by construction ,
we have @xmath244 since in addition @xmath245 , we have @xmath246 therefore @xmath247 now , assume ( b ) . in this case ,
* since @xmath248 , there exists @xmath249 such that @xmath250 * for each @xmath251 , there exists @xmath252 such that @xmath253 is occupied by a symbiont and has a neighbor connected to @xmath254 by a vertical path of vertices not occupied by a symbiont . a particular choice of the vertical coordinates @xmath252 for @xmath251 and the vertical paths to @xmath255 are shown on the left - hand side of figure [ fig : geometry-2 ] . since in addition each vertex in @xmath255 is connected by a path of vertices not occupied by a symbiont to each vertex in @xmath256 , we deduce that @xmath257 in particular , we conclude as previously that @xmath258 this completes the proof .
+ + using lemma [ lem : geometry ] , we now prove that if the number of symbionts in @xmath219 is not too small and the number of symbionts in @xmath220 not too large then , with probability close to one , one of the infected hosts along the boundary of the largest connected component void of symbionts will recover after a short time .
this is done in lemma [ lem : switch ] .
after this recovery event , with probability close to one , healthy hosts quickly invade the connected component void of symbionts when their natural birth rate is sufficiently large .
this provides a habitat for the symbionts which , again with probability close to one , quickly invade the connected component in a short time when the infection rate is sufficiently large .
this creates in particular a large cluster of symbionts in the target region @xmath220 , as shown in lemma [ lem : invade ] . to quantify these statements
, we set @xmath259 also , we introduce the stopping time @xmath260 in the next lemma , we give an upper bound for @xmath261 we point out that our bound holds regardless of the configuration in @xmath220 and therefore without conditioning on the event @xmath262 . the reason for the extra conditioning
is that the conditional probability in is the one that appears naturally in the proof of lemma [ lem : block ] below .
the two events in the conditioning are convenient because they give information about the number of symbionts at _ all _ times until time 1 rather than at a fixed time but we will prove that , with probability close to one when the parameter @xmath216 is large , * if there are at least @xmath216 symbionts at some time before time 1 in the box @xmath220 , which is the event @xmath263 , then there are at least @xmath264 symbionts in this box at time 1 , * if there are at least @xmath264 symbionts at time 0 in box @xmath219 then there are at least @xmath265 symbionts in this box at all times until time 1 , which is the event @xmath266 .
this will be used to couple the process with oriented site percolation .
[ lem : switch ] for all @xmath267 , we have @xmath268 according to the second part of lemma [ lem : geometry ] , @xmath269 since in addition infected hosts recover at rate @xmath12 , @xmath270 which completes the proof .
[ lem : invade ] assume that @xmath271 . then , @xmath272 for all @xmath216 sufficiently large . first , we define the stopping times @xmath273 also , for @xmath274 , we let @xmath275 be independent . since both the connected set @xmath276 and its boundary have at most @xmath277 vertices , the conditional probability given @xmath278 that this connected set becomes fully occupied by healthy hosts before time 3/4 is bounded from below by the probability that @xmath277 consecutive births of healthy hosts occur in less than 1/4 unit of time and before any birth of infected hosts , death , infection or recovery occurs .
this gives the lower bound : @xmath279 for all @xmath216 large .
note that , in the last inequality , we have used that @xmath280 using the same argument but for infections rather than births of healthy hosts , we obtain that the conditional probability that @xmath276 becomes fully infected before time 1 is @xmath281 for all @xmath216 large . combining
, we deduce that @xmath282 in other respects , according to the first part of lemma [ lem : geometry ] , @xmath283 since in addition @xmath284 , it follows that the two events @xmath285 are mutually exclusive .
this , together with , implies that @xmath286 for all @xmath216 sufficiently large .
+ + in order to ultimately compare the process with oriented site percolation , the next step is to attach a collection of events to the collection of boxes defined in . declare site @xmath287 to be good when the event @xmath288 that the region @xmath289 contains at least @xmath264 symbionts at time @xmath290 occurs . to prove that the set of good sites dominates the set of wet sites in an oriented site percolation process on @xmath291 , we need to relate the events to the good event .
more precisely , we need to prove that both @xmath292 are close to one , which follows from the next lemma .
[ lem : keep ] for all @xmath293 and @xmath294 , @xmath295 uniformly in all possible configurations outside @xmath289 .
since the total rate at which a host associated with a symbiont either dies with its symbiont or recovers is equal to @xmath54 , we have @xmath296 by independence , we deduce that , conditional on @xmath297 , @xmath298 in distribution . using the large deviation estimate @xmath299 with @xmath300 and @xmath301 , together with
, we conclude that @xmath302 this completes the proof .
+ + with lemmas [ lem : switch][lem : keep ] in hands , we are now ready to prove that if a site is good then its neighbors at the next level are good with probability arbitrarily close to one .
[ lem : block ] for all @xmath303 , there exists @xmath216 large such that @xmath304 the conditional probability to be estimated is bounded by @xmath305 the first and last terms on the right - hand side of can be bounded using lemma [ lem : keep ] .
indeed , according to this lemma , we have @xmath306 for all @xmath216 large .
lemma [ lem : keep ] also implies that @xmath307 for all @xmath216 large .
in addition , by lemmas [ lem : switch][lem : invade ] , for all @xmath216 large , @xmath308 when @xmath271 . combining
, we conclude that @xmath309 for all @xmath216 sufficiently large .
+ + lemma [ lem : block ] , together with ( * ? ? ?
* theorem 4.3 ) and obvious symmetry , implies that , at least for @xmath216 large , the set of good sites dominates stochastically the set of wet sites in a one - dependent oriented site percolation process on @xmath291 in which sites are closed with probability at most @xmath310 .
since such a process is supercritical , i.e. , percolation occurs with probability one when starting with a positive density of open sites at level zero , and since @xmath303 in the statement of the lemma can be chosen arbitrarily small , we deduce that the density of boxes with at least @xmath264 symbionts converges to a positive limit .
this shows survival of the symbionts , and therefore coexistence of the hosts with their symbionts , under the assumptions of theorem [ th:2d ] .
this section is devoted to the proof of which , together , imply theorem [ th:2d ] . as pointed out in the introduction ,
the theorem is true even in the limiting case @xmath85 , i.e. , when the pathogen invades instantaneously the nearby healthy hosts .
note that , in this case , if @xmath311 and a @xmath312 birth event occurs along the edge @xmath313 at time @xmath2 then @xmath314 the same applies in the reflected scenario ; we refer to this as _
invasion_. for convenience s sake , in the limiting case @xmath85 , we ignore the set of configurations @xmath315 since the complement of this set is invariant under the dynamics .
theorem [ th:1d ] is proved in several steps , but the basic recipe is to focus on some collection of 2s until they collide with the 1s ( or on some collection of 1s , until they collide with the 2s ) , then to reset and try again . in the absence of collision , things are nice , so if we can show that the time to collision is short and the probability of collision is less than one , then we need only reset a finite number of times before we can say something about the asymptotics . for the sake of exposition , we first study the `` segregated '' case where the 1s and the 2s lie on separate sides .
we then treat the general case in the same way as the segregated case , using a simple comparison property . +
* segregated case . * given a configuration @xmath97 , we let @xmath316 be the rightmost type 2 and the leftmost type 1 , respectively .
say that configuration @xmath97 is _ segregated _ if either @xmath317 or this is true after reflecting across the origin .
since @xmath24 , if interactions are nearest - neighbor then segregation is preserved , that is , @xmath318 letting @xmath319 , another way to express this is : @xmath320 implies that @xmath321 for all @xmath322 . for a segregated initial configuration , and noting that @xmath323 is left skip - free , let @xmath324 be the first time a 1 and a 2 are adjacent after separation .
when the infection rate is infinite , @xmath323 is never equal to one , so in this case we instead define @xmath325 since this implies that either a 1 showed up , and was instantaneously invaded by a 2 , or else a 2 advanced and then invaded some 1s at time @xmath2 . in lemmas [ lemma1][lemma2 ] below ,
we show that there are positive constants @xmath326 and @xmath327 so that uniformly over all segregated configuration @xmath87 , @xmath328 thinking of the event that @xmath329 as a success , the first estimate says that there is a success with positive probability and the last estimate that , if they happen , failures happen quickly , which will be used later to prove that the first success occurs quickly as well .
the second estimate indicates that , in case of a success , the pathogens die exponentially fast . using these estimates
, we can then prove the upcoming propositions [ prop1 ] and [ prop3 ] , that are the main results in the segregated case . in the following proofs , @xmath330 and @xmath327
are strictly positive constants such that @xmath331 decrease from step to step . to check
that decreasing @xmath331 will be justified , note that the probability estimates used are upper bounds , and that except at a few specific points when @xmath330 is fixed while @xmath327 is varied , the events being measured do not increase when @xmath331 are decreased . for an illustration of some of the random variables and processes introduced in the proof , we refer the reader to figure [
fig : edges ] . using a pair of independent substructures ,
define a pair of contact processes @xmath214 and @xmath213 with respective birth rates @xmath8 and @xmath9 and initial configurations @xmath336 there is a natural coupling of our process @xmath97 with the two contact processes @xmath214 and @xmath213 obtained by defining @xmath97 using both substructures up to time @xmath337 , then using , say , only the first substructure for all times @xmath338 .
this coupling has the property that @xmath339 for all @xmath340 .
now , define @xmath341 so that @xmath342 and @xmath343 for all @xmath340 . by the coupling , @xmath344
and we point out that the right - hand side of can be estimated in terms of the two independent contact processes @xmath214 and @xmath213 , which is the key to proving the first estimate .
+ in the estimates that follow , @xmath345 and @xmath346 represent universal , positive constants and @xmath347 may get smaller , respectively , larger from step to step . by monotonicity of the contact process , for @xmath348 , the set of occupied sites @xmath349 is dominated by the pure birth process in which particles do not die and give birth onto neighboring sites at rate @xmath350 , so the advance of type @xmath6 into uncharted territory grows like at most @xmath351 . in particular , @xmath352 and applying a standard large deviations estimate , we get @xmath353 also , for each @xmath322 , @xmath354 to control @xmath355 , we use a known estimate at integer times , then a poisson estimate at in - between times . from @xcite and the assumption
@xmath356 , for any integer @xmath290 , @xmath357 in addition , since the displacement in one unit of time is dominated by a poisson random variable with parameter @xmath9 , for any integer @xmath358 , we have @xmath359 ) \leq c \,e^{-ck}.\ ] ] combining and , we deduce that @xmath360 ) \leq c \,e^{-cn}\ ] ] then , combining with evaluated at @xmath361 , for each integer @xmath362 , @xmath363 ) \leq c \,e^{-cn}.\ ] ] to deduce the first estimate , we distinguish two cases : + * case 1 . * in case @xmath364 , the above bound on @xmath365 is at least two for all @xmath290 . combining the observation with and and summing over @xmath366 , @xmath367 * case 2 . *
assume now that @xmath368 .
in this case , @xmath369 by monotonicity and translation invariance , the same holds in general : @xmath370 using the markov property to combine with gives the first part of the lemma .
+ now , in above , for @xmath371 for some @xmath372 , absorb @xmath290 into @xmath373 by decreasing @xmath330 , then increase @xmath327 to account for @xmath374 .
then , for any @xmath322 , summing over @xmath375 , @xmath376 on the event @xmath377 , we have @xmath342 for all @xmath378 . using this and , @xmath379 uniformly in all segregated @xmath87 .
define recursively @xmath382 and @xmath383 we will use the obvious inclusion @xmath384 because the probability of the event on the right - hand side is easier to estimate . using the two bounds and above , uniformly in all configurations with @xmath385 , @xmath386 and for @xmath290
large enough the above bound on @xmath387 is at least @xmath290 ( it is exponential in @xmath290 but this is good enough ) . for @xmath388 with @xmath2 large enough and using ,
@xmath389 this completes the proof .
+ + in the next proposition , we use the estimates of the previous two lemmas to prove that , when starting from a segregated configuration , the position of the rightmost pathogen goes to @xmath390 exponentially fast .
this is the analog of the second estimate in lemma [ lemma1 ] but without the conditioning .
this result is then used in the subsequent lemma to show that the probability that the rightmost pathogen moves @xmath290 steps to the right of its initial position decays exponentially with @xmath290 .
define recursively @xmath393 , @xmath394 and @xmath395 then , by lemma [ lemma1 ] , @xmath396 where @xmath397 means stochastically smaller .
moreover , by lemma [ lemma2 ] , @xmath398 where @xmath399 is a sequence of independent , identically distributed random variables independent of the random variable @xmath216 and such that @xmath400 therefore , letting @xmath401 and applying a large deviations bound , @xmath402 we distinguish two cases depending on the value of the infection rate . + * case 1 .
* assume first that @xmath403 .
recall that @xmath404 by assumption .
comparing the set of sites in state 2 to a pure birth process with no deaths and with birth to adjacent sites at rate @xmath11 , a large deviations estimate gives @xmath334 so that @xmath405 for any @xmath86 , there is @xmath406 so that @xmath407 for all @xmath408 . therefore , @xmath409 for all @xmath408 . in other respects ,
@xmath410 combining equations gives the desired estimate when @xmath408 . if @xmath411 then , after increasing @xmath327 if necessary , the estimate holds for all values of @xmath2 .
+ * case 2 .
* now , let @xmath85 .
by assumption , @xmath412 .
we have @xmath413 for time @xmath414 , the process @xmath415 grows at rate at most @xmath9 , so repeating the reasoning in case 1 with the rate @xmath9 replacing the rate @xmath11 , we get @xmath416 it remains to control the jumps @xmath417 .
note that , for all @xmath418 , @xmath419 where the random variables @xmath420 are also independent .
a straightforward coupling argument shows that , for any @xmath6 , the set @xmath421 is dominated by product measure with density @xmath422 satisfying the initial - value problem @xmath423 letting @xmath424 , the unique solution is @xmath425 conditioned on @xmath426 , @xmath427 bounding @xmath422 by @xmath428 on @xmath429 $ ] , using the fact that @xmath430 is stochastically decreasing in @xmath2 to integrate against @xmath431 in place of @xmath337 , then using the bound @xmath432 , @xmath433 letting @xmath434 , a large deviations estimate gives @xmath435 for the same @xmath290 , and using with @xmath436 , @xmath437 combining with , then noticing that @xmath438 when @xmath408 for some @xmath406 that depends on @xmath330 , and increasing @xmath327 if necessary to account for @xmath439 , we have the desired result : @xmath440 this completes the proof . we again distinguish two cases .
+ * case 1 .
* assume that @xmath403 .
comparing to a pure birth process as above , @xmath442 for all @xmath290 large enough , @xmath443 for all @xmath444 , therefore using proposition [ prop1 ] and large deviations for the poisson distribution with parameter @xmath11 , @xmath445 ) \vspace*{4pt } \\
\hspace*{20pt } \leq \ p \,(r_m > r_0 - e^{-cm } ) + p \,(r_s > r_m + m \ \hbox{for some } \ s \in [ m , m + 1 ] ) \leq c \,e^{-cm}. \end{array}\ ] ] summing over @xmath446 gives the desired estimate . +
* now , assume that @xmath85 .
the idea is the same as above , and the complications are the same as in the proof of proposition [ prop1 ] , namely , that we need to account for invasion the only difference is that on a fixed interval of time @xmath447 $ ] we can have as many as a @xmath448 number of invasions .
we omit the details .
+ * general case .
* we now use the result proved in the segregated case to study the process starting from a general initial configuration . to deal with this general context
, it is useful to have a notion of _ descendant _ and _ ancestor_. suppose @xmath449 for some @xmath450 and @xmath451 .
then @xmath452 is a descendant of @xmath453 , and @xmath453 is an ancestor of @xmath454 if there are times and sites @xmath455 such that the following two conditions hold : in case @xmath85 , there may be a chain of multiple sites at a fixed time , but the reader may verify that the ancestry is still well - defined , in the obvious way .
+ for a site @xmath5 , let @xmath464 denote the descendants of @xmath465 , which all have type @xmath466 . by default ,
the set of descendants is empty if site @xmath5 is initially in state 0 . since interactions are nearest - neighbor , it is easy to check that @xmath467 \ \hbox{and } \ t \geq 0,\ ] ] and similarly with the roles of states 1 and 2 reversed
. let @xmath470 and @xmath471 be two copies of our modified stacked contact process constructed from the same graphical representation and define @xmath472 in case @xmath473 , since @xmath474 , by comparing to a ( single - type ) contact process with parameter @xmath8 , it is clear that @xmath475 with this comparison property in hand , the proof follows that of lemmas [ lemma1 ] and [ lemma2 ] and propositions [ prop1 ] and [ prop3 ] , so we only point out the adjustments
. + let @xmath476 and @xmath477 as well as @xmath478 also , let @xmath479 and @xmath480 .
figure [ fig : descendant ] gives an illustration of these quantities .
then , define recursively the sequence of stopping times @xmath393 and @xmath481 for all @xmath482 . as in lemma
[ lemma1 ] , we define a pair of independent processes @xmath214 and @xmath213 , this time copies of the modified stacked contact process , with initial configurations @xmath483 and values @xmath484 defined in the obvious way , so that as before , the process @xmath97 can be coupled to @xmath214 and @xmath213 in such a way that @xmath485 can be computed in terms of @xmath214 and @xmath213 . using the comparison property on @xmath214 , we proceed as before , establishing estimates on both sides ( i.e. , for @xmath486 as well as @xmath487 ) to obtain . then , in , we take @xmath346 large enough that , when @xmath364 , the probability is less than one half . using the comparison property ,
we show that , when @xmath488 and with probability at least @xmath489 , @xmath490 .
then , using the markov property , we get @xmath491 the adaptation of the rest of lemma [ lemma1 ] , and of lemma [ lemma2 ] , is straightforward and relies on establishing estimates on both sides and taking a union bound .
+ the adaptation of propositions [ prop1 ] and [ prop3 ] is also straightforward .
note that the above comparison property also supplies an upper bound on the size of the jumps @xmath492 and @xmath493 , in the limiting case @xmath85 . from the analogue of proposition [ prop1 ]
, we obtain the first statement of the proposition , since @xmath317 implies @xmath494 . finally , the second statement is just the analogue of proposition [ prop3 ] , and the proof is complete . + + next , we use proposition [ prop5 ] to prove , which says that in an exponentially growing neighborhood of any site , eventually there are no sites in state 2 .
we also record an exponential estimate .
[ prop7 ] for a site @xmath495 , let @xmath496 then , there exist @xmath334 such that , for any @xmath87 and @xmath406 , @xmath497 also , there exists @xmath86 so that , for any configuration @xmath87 and any site @xmath5 , @xmath498 throughout this proof , @xmath206 refers to a site which is initially in state 2 .
let @xmath347 be two constants as in proposition [ prop5 ] , so that @xmath499 using a union bound over @xmath500 $ ] and that @xmath501 is an absorbing property , @xmath502 let @xmath372 be such that @xmath371 implies @xmath503 . if @xmath504 for integer @xmath505 and all @xmath371 , then we have @xmath506 taking @xmath327 larger if necessary makes the previous inequality true also for all @xmath374 .
then , taking a union bound and increasing @xmath327 at the last step gives @xmath507 combining the estimates gives @xmath508 \ \hbox{and } \
t \in [ n , n + 1 ] ) \leq c \,e^{-cn/2}.\ ] ] given @xmath406 , summing over @xmath509 gives the first statement .
summing over all @xmath290 and using the borel - cantelli lemma finishes the proof .
+ + from now on , we call an _ event _ a poisson event in the graphical representation .
events are referred to by their effect on the target site , so for example , a @xmath312 event is an edge event at time @xmath2 along a directed edge @xmath229 such that , if @xmath510 and @xmath511 then @xmath512 .
also , a death event refers to a @xmath513 event , while a birth event is a @xmath514 event , where @xmath515 . note that since the graphical representation consists of at most a countably infinite number of poisson point processes , with probability one , no two events occur at the same time .
let @xmath520 with @xmath521 ; a similar argument applies to @xmath522 .
using proposition [ prop5 ] gives the existence of constants @xmath334 so that , for all @xmath303 , @xmath523 on the complement of the above event , the descendants of @xmath524 are contained in @xmath525 a quick sketch ( see figure [ fig : rectangle ] ) shows that the rectangle @xmath33 is disjoint from the set @xmath526 provided the top left corner of @xmath33 lies to the right of the line @xmath527 , which is the condition @xmath528 since @xmath529 , this condition is satisfied if @xmath530 . summing over @xmath521 for both @xmath520 and @xmath531 then gives the desired result .
+ + a related notion to the descendants is the _ cluster _
, that we need only define for type 1 , as follows .
suppose @xmath532 for some @xmath450 and @xmath451 .
then , we say that @xmath452 belongs to the cluster of @xmath453 if there are times and sites @xmath455 such that the following two conditions hold : in contrast to the definition of descendants , it is permitted to have @xmath534 .
+ if @xmath535 then , for @xmath536 , let @xmath537 denote the cluster of @xmath453 at time @xmath2 , that is , @xmath538 and denote it @xmath539 for @xmath540 . again , since interactions are nearest - neighbor , @xmath541 \quad \hbox{and } \quad t \geq s.\ ] ] as a warm - up to , we prove the following .
[ lemma8 ] let @xmath542 and @xmath543 . for @xmath403 , let @xmath544 and for @xmath85 define @xmath337 in a similar way .
then , there are @xmath545 such that @xmath546 uniformly in all @xmath87 such that @xmath547 .
define a pair of independent copies @xmath470 and @xmath471 of the modified stacked contact process with initial configurations @xmath548 define also @xmath549 so that @xmath337 can also be expressed as @xmath550 to show that @xmath329 with positive probability , first we fix @xmath290 and consider the case @xmath551 . using large deviations estimates for the poisson distribution , we can show that @xmath552 to do so , it suffices to first make an estimate for @xmath553 , then for @xmath554 $ ] for each @xmath446 , then to take a union bound .
then , taking @xmath555 and the same @xmath290 as in the statement of lemma [ lemma7 ] , we find that @xmath556 since in addition @xmath557 taking @xmath290 large enough , we find that if @xmath518 for @xmath519 then @xmath558 for @xmath87 such that @xmath547 , the probability @xmath559 is at least the probability that , on the time interval @xmath429 $ ] , there are no birth events along edges touching @xmath560 $ ] , there is no death event at 0 , and there is a death event at every @xmath5 with @xmath561 , and this probability is at least @xmath562 for some @xmath332 . using the markov property and the estimate on @xmath329 in the previous case then gives @xmath563 as desired . + to deduce the estimate on @xmath380
, we note that @xmath564 which can be proved by applying an estimate at each integer time @xmath565 and summing over @xmath290 .
then , combining with the first statement in proposition [ prop7 ] and noting that @xmath566 we deduce the estimate on @xmath380 . + + we are now ready to establish which states the existence of a linearly growing region starting from a random space - time point in which the process agrees with the contact process with parameter @xmath8 .
this will also complete the proof of theorem [ th:1d ] .
+ + given @xmath453 , for @xmath536 recall that @xmath537 denotes the cluster of @xmath453 at time @xmath2 .
let @xmath393 and @xmath567 be any site with @xmath568 .
without loss of generality , suppose that the set @xmath569 is infinite , and define @xmath570 and @xmath571 recursively for @xmath403 by letting @xmath572 with the value of @xmath570 being unimportant if @xmath573 .
note that if time @xmath574 then site @xmath575 is well - defined due to the fact that @xmath576 is an invariant set for the dynamics .
let @xmath577 . applying the strong markov property and using lemma [ lemma8 ]
, we obtain that @xmath216 is at most geometric with parameter @xmath29 .
in addition , by the second part of lemma [ lemma8 ] , for @xmath578 , @xmath579 and the random variables @xmath420 are independent . in particular , @xmath580 is almost surely finite .
let @xmath581 and let @xmath582 , and let @xmath583 and @xmath584 .
+ recall that @xmath90 denotes the process with initial configuration @xmath585 for all @xmath5 . since @xmath203 by assumption
, a straightforward coupling argument shows that , for any configuration @xmath586 , @xmath587 therefore , @xmath588 whenever @xmath574 . by definition of time
@xmath589 , the set @xmath590 is the set of infected sites in a ( single - type ) contact process started from the single infected site @xmath570 at time @xmath571 , so a coupling of @xcite shows that if @xmath574 then @xmath591 \ \hbox{and all } \
\tau_i < t < \tau_{i + 1}.\ ] ] in @xcite , it is shown that , for the contact process ( which in this context means in the absence of any interaction with 2s ) , @xmath592 and @xmath593 so the same is true here provided @xmath594 , which is the case for @xmath595 .
the proof is now complete . | the stacked contact process is a three - state spin system that describes the co - evolution of a population of hosts together with their symbionts . in a nutshell ,
the hosts evolve according to a contact process while the symbionts evolve according to a contact process on the dynamic subset of the lattice occupied by the host population , indicating that the symbiont can only live within a host .
this paper is concerned with a generalization of this system in which the symbionts may affect the fitness of the hosts by either decreasing ( pathogen ) or increasing ( mutualist ) their birth rate .
our main results show the existence of a parameter region in which hosts and symbionts , in this case pathogens , coexist in spatial dimensions @xmath0 whereas only the host survives on the one - dimensional integer lattice .
the mean - field approximation of the process is also studied in detail and compared to the spatial model . |
many - body effects in the condensed matter systems usually generate a plethora of quasiparticles of either bosonic ( like phonons , magnons ) or fermionic ( for instance polarons ) nature .
recently there has been enormous interest in the exotic fermionic quasiparticles , that resemble the majorana - type fermions @xcite .
these emergent quasiparticles , identical with their own antiparticles , can appear under specific conditions in the symmetry broken states @xcite . from practical point of view
the considerable attention has been motivated by their non - abelian character , because the nonlocal nature makes them immune to decoherence that is desirable for quantum computing and brand new spintronic devices @xcite .
possible realizations of the majorana quasiparticles have been predicted in variety of systems , e.g. vortices of superfluids @xcite , three - dimensional @xcite or two - dimensional @xcite topological insulators coupled to superconductors , noncentrosymmetric superconductors
@xcite , electrostatic defects in topological superconductors @xcite , @xmath2-wave superconductors @xcite , the semiconducting @xcite or ferromagnetic @xcite nanowires with strong spin - orbit interaction coupled to @xmath3-wave superconductors , josephson junctions @xcite , ultracold atom systems @xcite , and other .
so far the most reliable experimental evidence for the majorana quasiparticles has been reported by the tunneling measurements with use of the atomic nanowires coupled to bulk @xmath3-wave superconductors @xcite .
subgap tunneling conductance has revealed the zero - bias enhancement at the edges of such wires @xcite which has been interpreted as signature of the zero - energy majorana mode , however , in alternative scenarios one can eventually assign it to disorder @xcite , kondo effect in a crossover from the doublet to singlet configurations @xcite or other effects @xcite . since unambiguous identification of the majorana quasiparticles
is currently needed there have been suggested proposals , based on optomechanical detection @xcite , shot noise measurements @xcite , josephson phase - tunable spectroscopy @xcite and other ideas relevant to tunneling heterostructures comprising the quantum dot ( qd ) with the side - attached nanowire hosting the majorana quasiparticles , as shown in fig .
[ schematics ] . in the case
when both of external leads are metallic there has been predicted reduction ( by half ) of the quantum dot conductance @xcite , the vanishing seebeck coefficient ( due to a perfect particle - hole symmetry ) @xcite and unique interferometric features @xcite .
another proposal , applicable to vortices in topological superconductors , indicated the spin - selective andreev scattering @xcite as a convenient tool . in such method the incoming electron with a given spin is converted to a pair ( of equal spin electrons ) in topological superconductor , simultaneously reflecting a hole ( of the same spin ) back to the conducting electrode .
such andreev scattering , has recently indeed provided evidence for the majorana modes in bi@xmath4te@xmath5/nbse@xmath4 heterostructure @xcite . for the t - shape configurations ( displayed in fig .
[ schematics ] ) it has been recently suggested @xcite that fingerprints of the majorana - type fermions would be detectable in spin - dependent spectrum , when qd is coupled between superconducting and metallic leads . such setup has been already considered by several authors @xcite , however , the spin - dependent effects have not been addressed . on the other hand the ordinary double quantum dot arranged in the same setup between the
superconducting and conducting electrodes has been studied by us @xcite and followed by other groups @xcite .
these studies have revealed nontrivial interplay of the quantum interference ( fano - lineshapes ) , the on - dot electron paring ( andreev / shiba states ) and the strong correlations ( kondo effect ) . in this work
we extend our previous analysis @xcite , considering the side - attached majorana quasiparticle and address its influence on the spin - resolved qd spectrum and the measurable subgap ( andreev ) spectroscopy .
in particular , we show that subgap transport is sensitive to the spin - polarization ( spin - resolved ) and can unambiguously detect the interferometric features driven by the majorana quasiparticle .
we also study whether such interferometric majorana structures are robust against the kondo effect appearing at the same zero - energy .
the paper is organized as follows . in section
ii we formulate the microscopic model and briefly study the interferometric effects for the uncorrelated qd case .
next we discuss the spin - dependent electronic spectrum and , in section iii , we analyze the majorana signatures in the spin - resolved andreev tunneling .
section iv discussed the correlated qd in the kondo regime . in section
v we point out the conclusions and in appendices we present some useful technical details .
due to chiral properties the ends of topological superconducting wire , that host a pair of majorana fermions , are spin polarized @xcite .
for this reason we assume that in the t - shape configuration ( fig .
[ schematics ] ) only one of the electron spin components is directly coupled to the majorana quasiparticle @xcite . when the central quantum dot ( qd ) is coupled to both normal conducting electrodes
the spin @xmath0 and @xmath1 transport channels can be regarded as independent , at least in absence of the correlations @xcite .
this is no longer true , if one ( or both ) of external reservoirs is ( are ) superconducting because of the proximity effect that mixes the particle with hole degrees of freedom @xcite and affects both of the spin components . in consequence ,
any physical process that engages electron of a given spin simultaneously affects its opposite spin partner .
such mechanism will prove to be extremely important when considering the quantum interference driven by side - coupled majorana quasiparticle , which for brevity shall be referred as the majorana quantum dot ( mqd ) .
we have previously shown @xcite that local pairing induced in the t - shape double quantum dot ( dqd ) on interface between the metallic and superconducting electrodes is manifested by two lineshapes , detectable in the subgap andreev transport .
spectrum of the central qd reveals : the fano - type resonance formed near the energy @xmath6 of the side - coupled qd @xcite and another anti - fano structure appearing at @xmath7 on opposite side of the fermi level @xcite .
motivation of the present study is to investigate whether similar features can show up when the central quantum dot is coupled to the majorana quasiparticle , whose generic nature is related to only one spin ( say @xmath0 ) electrons .
we extend here the previous study of n - dqd - s system @xcite , considering the quantum interference induced by side - attached mqd . in order to clarify the asymmetric features appearing in the qd spectrum ( section ii ) and the spin - selective andreev transport ( section iii ) , we shall revisit the usual n - dqd - s setup imposing the spin polarized inter - dot hopping .
such toy model provides the clear physical interpretation of the fano and anti - fano resonances . in section
iv we shall also address the correlation effects in the kondo regime .
two - terminal system with the central qd embedded between the metallic and superconducting electrodes and side - coupled to the edge of superconducting topological nanowire hosting the majorana quasiparticle ( fig .
[ schematics ] ) can be described by the anderson impurity hamiltonian @xmath8 the external bath @xmath9 consists of the metallic @xmath10 and superconducting @xmath11 reservoirs , where electron energies @xmath12 are measured from the fermi level @xmath13 .
we describe the central qd by @xmath14 , where @xmath15 is the energy level and @xmath16 stands for the repulsive interaction between opposite spin electrons .
its hybridization with the external reservoirs is given by via @xmath17 , where @xmath18 denote the matrix elements .
focusing on the deep subgap regime ( i.e. energies @xmath19 ) it has be shown @xcite that the superconducting electrode is responsible for inducing the static pairing , therefore we can replace @xmath20 by the effective , , proximized quantum dot @xmath21 .
this simplification is fairly acceptable for our considerations of the spin - dependent subgap spectrum and the andreev spectroscopy . on the other hand ,
the low - energy theory for the side - attached majorana quasiparticle can be formulated as @xcite @xmath22 where the operators @xmath23 describe the edge states of the topological superconducting wire and @xmath24 accounts for their overlap .
it is convenient to represent the exotic majorana operators @xmath25 , @xmath26 by the standard fermionic ones @xcite @xmath27 , @xmath28 . in this representation the term ( [ majorana_part ] )
takes the following form @xmath29 where @xmath30 . to distinguish the consequences caused by the fact that tunneling to the mqd involves only the spin @xmath0 electrons from other effects due to their specific majorana - type nature we first examine the setup in which ts is replaced by the usual quantum dot ( qd@xmath4 )
@xmath31 imposing the spin - dependent interdot hoping @xmath32 . under such circumstances
the fourier transform of the retarded green s function for the uncorrelated central quantum dot can be expressed as @xcite @xmath33 \label{ord_g } \\ & = & \left ( \begin{array}{cc } \omega-\epsilon + \frac{i\gamma_n}{2 } - \frac{t^{2}_{\sigma}}{\omega-\epsilon_2 } & \frac{\gamma_s}{2 } \\
\frac{\gamma_s}{2 } & \omega+\epsilon + \frac{i\gamma_n}{2 } -\frac{t^{2}_{\bar{\sigma}}}{\omega+\epsilon_2 } \end{array}\right)^{-1 } \nonumber\end{aligned}\ ] ] where @xmath34 is inverse spin to @xmath35 . for the weak identical couplings @xmath36 the spectral function of central quantum dot @xmath37
is characterized by two interferometric structures at the qd@xmath4 level and on the opposite side of a fermi level @xcite .
the feature at @xmath38 has the usual fano - type resonant lineshape @xcite , whereas its companion at @xmath7 has anti - resonant ( antifano ) shape . obviously for @xmath36 the spectral functions @xmath39 of both spins are identical .
when we forbid the tunneling of @xmath1 electrons @xmath40 , the toy model ( [ toy_model ] ) would be analogous to the original setup ( fig .
[ schematics ] ) with spin @xmath0 of the central quantum dot coupled to ts wire . in such case
there survives only the single interferometric structure in each of the spectral functions @xmath39 . for @xmath0 electrons ( directly coupled to the side - attached dot )
we observe the fano - type interference pattern at @xmath41 and for the opposite spin @xmath1 electrons there appears the anti - fano structure at @xmath42 .
this result can be understood if we anticipate that the anti - fano feature is an indirect response of the spin @xmath0 electrons . in other words , even though the spin @xmath1 electrons are not directly coupled to the side - attached quantum dot , due to the induced local pairing they feel a feedback from the opposite ( @xmath0 ) spin electrons .
figure [ ord_zero ] illustrates the spin - resolved spectral functions of the uncorrelated central quantum dot asymmetrically coupled to the ordinary qd@xmath4 whose energy level is @xmath43 . in this case
the fano and anti - fano lineshapes appear at the same position .
the next figure [ ordsep ] corresponds to the situation @xmath44 , when these features appear either in the particle or hole regions .
as we shall see in the next subsection this behaviour is partly similar to the case with majorana quasiparticle . to simplify the notation
let us denote the particle @xmath45 and hole @xmath46 green s functions of the central qd coupled to the metallic and superconducting leads in absence of the majorana quasiparticle . for the uncorrelated case ( [ ord_g ] )
these functions read @xmath47 similarly we also denote in the inverse particle and hole propagators of the isolated majorana quasiparticle by @xmath48 and @xmath49 , respectively . using the equation of motion ( eom )
approach we calculated the matrix green s function @xmath50 defined in the following matrix notation @xmath51 . for spin @xmath0 electrons we find ( see appendix a for details ) @xmath52 where @xmath53 $ ] and @xmath54 .
we also determined the green s function @xmath55 .
below we present explictly only the green s function @xmath56 that is the most relevant for this work .
it takes the following form @xmath57 ^ 2\frac{amn-2t_{m}^2\omega}{abmn-2t_{m}^2\omega(a+b ) } , \label{gdown}\end{aligned}\ ] ] where @xmath58 is the normal green s function for the case when qd is coupled only to the metallic lead ( for @xmath59 ) .
let us remark that differences between @xmath60 and @xmath55 originate from the fact that only spin @xmath0 electrons are directly coupled to the majorana mode .
this difference vanishes for @xmath61 when @xmath62 reproduce the result @xcite obtained for dot coupled only to n and s electrodes @xmath63^{-1}$ ] . on the other hand , for @xmath64 in absence of a superconducting electrode ( @xmath65 ) the solution for
spin @xmath1 electrons is identical with solution for qd coupled only to normal metal , regardless of the coupling strength to ts wire .
this is because spin @xmath1 electrons are not affected by the side - coupled majorana quasiparticle ( so without electron pairing they do not , , feel any interference ) .
this result clearly indicates that the interference patterns appearing in the spectrum of @xmath1 electrons originate solely from the pairing with electrons of the opposite spin .
[ mdqd ] illustrates the spectral function @xmath39 of the central qd weakly coupled to the majorana quasiparticle , in the case @xmath66 .
interference pattern appearing at @xmath67 in the spectral function @xmath68 resembles a resonant lineshape .
however , from a careful examination we clearly notice that it is not really the true fano resonance ( like the one for @xmath0 electrons shown in fig .
[ ord_zero ] ) .
such line - shape indicates , that electron waves resonantly scattered by the majorana quasiparticle ( to be regarded as half of the physical electron ) change their phase only by the fraction of @xmath69 value ( which is typical for scattering caused by the side - coupled ordinary quantum dots @xcite ) .
underlying mechanism responsible for this fractional interferometric feature has thus the same origin as @xmath70-periodicity of the josephson junctions made of two majoranized superconducting wires @xcite . in rashba nanowires of a finite length the majorana quasiparticles partly overlap with one another , inducing some energy splitting between the edge modes ( @xmath71 ) . in fig .
[ overlap ] we show the interference patterns obtained for @xmath72 .
spectrum of the spin @xmath0 electrons reveals two fractional fano - like resonances ( solid line in fig .
[ overlap ] ) , whereas the spin @xmath1 electrons are characterized by two anti - fano features ( dashed lines in fig .
[ overlap ] ) at the same energies .
let us emphasize , that such behavior is qualitatively different from the results for t - shape heterojunction with the ordinary quantum dot ( displayed in fig .
[ ordsep ] ) .
interferometric effects could thus be useful for unambiguous detection of the majorana quasiparticles .
the interferometric fano - like structures displayed in figs [ mdqd ] and [ overlap ] occur when the central qd is very weakly coupled to the side - attached majorana quasiparticle . upon increasing the interdot coupling @xmath73 the nanoscopic ( qd and mqd )
objects are expected to develop new spectroscopic signatures , characteristic for the entire ` molecular ' complex .
evolution of the spectral function @xmath39 vs @xmath73 is presented in fig .
[ molecular ] . for increasing @xmath73
we observe that the initial two andreev peaks ( originating from the mixed particle and hole degrees of freedom @xcite ) and the fano / antifano lineshapes ( caused by the majorana qd ) gradually change into the three - peak structure . in both spin components
we clearly see emergence of the zero - energy peak at expense of reducing the spectral weight of the initial andreev states .
formation of the zero - energy peak signifies a ` leakage ' of the majorana quasiparticle into the central qd , in some analogy to what has been discussed in ref .
it is amazing that in the present case such proximity induced zero - energy state shows up in both spin sectors , despite that mqd is directly coupled only to the spin @xmath0 electrons .
interference effects caused by the majorana quasiparticle can be practically observed in the setup ( fig .
[ schematics ] ) by measuring the tunneling current under nonequilibrium conditions @xmath74 .
when applied bias @xmath75 is smaller in magnitude than @xmath76 the charge current @xmath77 is contributed by spin @xmath0 ( @xmath78 ) and spin @xmath1 ( @xmath79 ) electrons transmitted between the external electrodes via the andreev scattering @xmath80 , \label{i_a}\end{aligned}\ ] ] where @xmath81^{-1}$ ] is the fermi distribution .
the transmittance of each spin sector @xmath82 describes a probablity of converting the electron with spin @xmath35 into the hole with spin @xmath34 in the metallic lead .
the differential conductance @xmath83 is usually enhanced near the subgap ( andreev / shiba ) states @xcite .
andreev spectroscopy can also reveal other subgap features , in particular due to the quantum interference @xcite .
[ tra1 ] shows the spin - resolved andreev transmittance @xmath84 obtained for @xmath85 ( upper panel ) and @xmath86 ( bottom panel ) . in the first case we observe the fractional resonance appearing for each spin sector at zero - bias ( although of opposite shape ) . in the case
@xmath87 we notice two interferometric structures at @xmath88 . for spin @xmath0 sector
there appears the well pronounced resonance at @xmath89 and another shallow structure at @xmath90 .
the andreev transmittance of @xmath1 sector has an opposite shape , i.e. @xmath91 .
the spin - resolved andreev transport could thus estimate the overlap @xmath92 between the edge modes of ts nanowire . for
@xmath85 such spectroscopy can clearly distinguish the fractional interferometric lineshapes caused by the majorana quasiparticle from the typical fano / antifano lineshapes due to the ordinary quantum dots @xcite .
since the interferometric feature caused by the side - attached majorana quasiparticle shows up at the fermi level ( for the case @xmath85 ) it is natural to inspect its relationship with the kondo effect , whose signature ( narrow peak ) appears at the same energy .
the many - body kondo effect occurs at low temperatures due to the effective exchange interaction induced between the qd and normal lead ( n ) electrons .
its subtle nature in a subgap regime has been addressed by variety of methods ( see the recent discussion @xcite and other references cited therein ) . in the present context
we shall treat the correlations using the decoupling scheme for the green s functions @xcite that proved to be satisfactory , at least qualitatively @xcite . to account for the kondo effect we start from the results obtained for the uncorrelated problem ( section iii ) and proceed with approximations for the electron - electron interactions . in absence of the side - attached mqd
, we again introduce the abbreviations for particle @xmath93 and hole @xmath94 propagators .
following our previous study @xcite of the single quantum dot ( n - qd - s ) setup we approximate these propagators by @xmath95^ { * } } , \label{new_a } \\
\tilde{b } & = & \omega + \epsilon + [ \sigma_n(-\omega)]^ { * } + \frac{(\gamma_s/2)^2}{\omega-\epsilon - \sigma_n(\omega ) } , \label{new_b}\end{aligned}\ ] ] where the selfenergy @xmath96 accounts for the coupling of qd to the normal lead taking into account the interactions between electrons @xmath97 .
one possible approximation for @xmath96 can be based on the popular decoupling scheme for the set of green s functions ( see appendix b in ref .
@xcite ) @xmath98 } { [ \omega-\epsilon][\omega-\epsilon - u-\sigma_3(\omega ) ] + i\frac{\gamma_{n}}{2}u } , \nonumber\end{aligned}\ ] ] where @xmath99 .
\label{sigmas } \end{aligned}\ ] ] instead of ( [ sigman ] ) one can obviously describe the correlation effects for the quantum dot coupled to the free fermion gas using some different techniques @xcite . substituting the inverse green s functions ( [ new_a ] , [ new_b ] ) with specific @xmath96 to the matrix greens function ( [ gr44 ] ) , whose derivation is outlined in appendix
a , we obtain @xmath100 ^ 2\frac{\tilde{a}mn-2t^2\omega}{\tilde{w } } , \label{gdown_bis}\end{aligned}\ ] ] where @xmath101 . in our setup
the correlated quantum dot is connected to the superconducting reservoir , which ( by proximity effect ) induces the on - dot electron pairing . on the other hand
the repulsive coulomb interactions disfavor any double occupancy , suppressing the local pairs . even though the pairing and correlations are strongly antagonised one can find some regime of the model parameters , for which the kondo physics coexists with the on - dot pairing @xcite ( the latter is necessary for activating the andreev tunnelling that could probe the subgap states ) .
this regime is particularly important if we want to confront the kondo state with the interferometric structures due to side - attached majorana quasiparticle .
it has been earlier indicated @xcite that the optimal conditions where the kondo effect coexists with the on - dot pairing can be tuned by : @xmath102 ( that controls qd occupancy ) and the ratio between couplings to external the electrodes @xmath103 ( that is crucial for the effective exchange potential @xcite ) . for specific calculations we focus here on the strong coulomb potential @xmath104 and
choose @xmath105 .
we have checked that in such situation the kondo effect coexists with the on - dot pairing for slightly asymmetric couplings @xmath106 . with this in mind , we thus fixed the ratio @xmath107 . in absence of mqd ( i.e. for n - qd - s configuration ) the narrow kondo peak at @xmath67
coexists then with the subgap andreev quasiparticle peaks at @xmath108 whose broadening ( inverse life - time ) is proportional to @xmath109 @xcite .
influence of the side - coupled majorana quasiparticle on the spin - resolved spectral functions @xmath39 of the correlated qd is illustrated in figures [ kondo_up ] and [ kondo_down ] .
the upper panels correspond to the case of overlapping majorana modes @xmath110 . in analogy to the noninteracting situation ( fig .
[ overlap ] ) we observe the fractional fano and anti - fano lineshapes appearing at @xmath111 in the spectrum of spin @xmath0 and @xmath1 electrons , respectively . for @xmath1 electrons
we also notice that both anti - fano resonances are much less pronounced as compared to @xmath112 case .
this is a consequence of the strong coulomb interactions suppressing the on - dot pairing , that is indirectly responsible for the interferometric structures in the spectrum of @xmath1 electrons .
the most intriguing case occurs for @xmath85 , when the kondo and interferometric structures coincide at exactly the same energy .
spectrum of @xmath0 electrons , that are directly coupled to the majorana quasiparticle clearly show the dominant and destructive influence of the quantum interference on the kondo state ( see the bottom panel in fig .
[ kondo_up ] , where the kondo peak is completely washed out ) . as regards the spectrum of @xmath1 electrons , the anti - fano interferometric structure constructively combines with the kondo peak , enhancing the zero - energy feature .
scattering mechanism driven by the zero - energy majorana quasiparticle side - attached to the correlated quantum dot has thus very interesting effect on the kondo state .
for the spin @xmath0 sector ( directly coupled to the majorana quasiparticle ) the ongoing quantum interference has destructive character . in other words , the fractional fano - type resonance induced by
the side - coupled majorana quasiparticle is robust against the kondo peak .
on contrary , in the spin @xmath1 sector ( where electrons are not directly coupled to the majorana quasiparticle ) the kondo state is promoted by the quantum interference .
such exotic spin - resolved quantum interference effects might be useful for experimental detection of the majorana quasiparticle . from a physical point of view , the spin - resolved screening of the correlated quantum dot is due to the following process : the spin @xmath0 electrons , , flee to the side - couple majorana structure ( so there is nothing left to be screened ) , whereas the local pairing compensates this loss in density of @xmath1 electrons ( whose screening becomes more efficient ) .
we have studied the interferometric structures induced by the majorana quasiparticle side - coupled to the quantum dot on interface between the superconducting and normal electrodes . due to the superconducting proximity effect
such interferometric lineshapes appear simultaneously in both spin sectors , although in each of them they have completely different character . in the case of uncorrelated quantum dot
the spin component directly coupled to the majorana quasiparticle reveals the fractional fano - type resonances .
fractionality comes here from the fact that majorana quasiparticle represents a half of the true electronic state . on the other hand ,
the opposite spin sector shows up the antifano lineshapes at the same energies .
we have also investigated an interplay of the spin - resolved interferometric features with the kondo effect , originating from the strong correlations .
we have found that scattering caused by the majorana quasiparticle either suppresses or enhances the kondo effect , depending on orientation of the electron spins .
this mechanism might enable a spin - selective detection of the majorana quasiparticles .
we hope that our results can stimulate experimental efforts to use the quantum interference as a tool for probing the majorana quasiparticles .
further theoretical studies of this setup are needed , to treat the correlation effects in more sophisticated manner .
we acknowledge p. stefaski for valuable discussions and thank k. kapcia for technical assistance . this work is supported by the national science centre in poland through the project dec-2014/13/b / st3/04451 ( td ) .
in this appendix we outline procedure for determination of the matrix green s function ( [ gr44 ] ) . starting from the model hamiltonian ( [ hand ] ) we consider the uncorrelated quantum dot @xmath112 coupled to between the metallic ( n ) and superconducting ( s ) electrodes and
additionally side - coupled to the edge of topological wire . in the deep subgap regime
the superconducting electrode induces the static on - dot pairing and ( [ hand ] ) simplifies to @xmath113 fourier transform of the retarded green s function can be computed from the equation of motion @xmath114_{+}\rangle + \langle\langle [ a , h]_{- } ; b\rangle\rangle$ ] where @xmath115 denote anticommutator / commutator , respectively .
the particle propagator @xmath116 for spin @xmath0 electrons of the central qd is mixed with the operators of ts wire and with the anomalous green s function @xmath117 as @xmath1 electrons are not directly coupled to ts wire the anomalous green s function @xmath118 does not generate any propagator containing @xmath119 operators . using eom , it can be expressed via the hole propagator @xmath120 using ( [ a3 ] ) we can rewrite equation ( [ g1up ] ) as @xmath121 we can notice that expression in the bracket on left hand side is the inverse particle propagator for n - qd - s system in absence of the ts wire . for brevity
we denote it by symbol @xmath122 , that is presented in eqn ( [ a_symbol ] ) .
the other green s functions , where @xmath123 is mixed with @xmath119 and @xmath124 operators can be found from the eom as @xmath125 equations ( [ m],[n ] ) generate the new anomalous function @xmath126 .
we write down the equation of motion for this function @xmath127 and determine the new function @xmath128 as @xmath129 now the propagator @xmath130 can be represented as @xmath131 expression appearing in a bracket on the left hand side is the inverse hole propagator of n - qd - s system without ts wire .
we have denoted it by symbol @xmath132 in the main text , that is explicitly given by eqn ( [ b_symbol ] ) .
finally , we obtain the following set of equations @xmath133 using the abbreviations @xmath134 , @xmath135 and denoting @xmath136 these green s functions can be recast in the following matrix form @xmath137 the entire matrix presented in eqn ( [ gr44 ] ) can be obtained by solving 4 similar sets of the equations .
we have previously investigated @xcite the normal double quantum dot ( dqd ) in the same t - shape configuration ( fig [ schematics ] ) assuming the spin independent couplings @xmath36 .
our study has indicated a destructive influence of the fano interferometric structure on the kondo resonance whenever the qd@xmath138 energy @xmath139 approached the fermi level . in this appendix
b we consider the spin - polarized coupling @xmath32 , focusing on the limit of vanishing @xmath140 .
figures [ kondo_ord1 ] and [ kondo_ord2 ] show the spectral function @xmath39 obtained in the kondo regime ( at temperature @xmath141 ) for spin @xmath0 ( red line ) and @xmath1 ( blue line ) electrons . in the case when @xmath6 is far from the kondo peak ( fig .
[ kondo_ord1 ] ) we observe that the fano - type resonance ( seen in @xmath68 at @xmath6 ) and its anti - fano companion ( present in @xmath142 at @xmath7 ) practically coexist with the zero - energy kondo peak .
the situation changes dramatically , when energy @xmath6 qd@xmath4 coincides with the fermi level ( fig .
[ kondo_ord2 ] ) . in both spin sectors
the kondo peak is then completely destroyed by the interferometric lineshape .
this effect proves that the quantum interference is dominant , whenever it coincides with the zero - energy kondo peak .
let us notice that such tendency is distinct from the interferometric features induced by the majorana quasiparticle ( figs [ kondo_up],[kondo_down ] ) .
s. tewari , s. das sarma , c. nayak , c.w .
zhang , and p. zoller , _ quantum computation using vortices and majorana zero modes of a @xmath143 superfluid of fermionic cold atoms _ , phys .
lett . * 98 * , 010506 ( 2007 ) .
m. wimmer , a.r .
akhmerov , m.v .
medvedyeva , j. tworzydo , c.w.j .
beenakker , _ majorana bound states without vortices in topological superconductors with electrostatic defects _ , phys . rev .
lett . * 105 * , 046803 ( 2010 ) .
l. jiang , t. kitagawa , j. alicea , a.r .
akhmerov , d. pekker , g. refael , j.i .
cirac , e. demler , m.d .
lukin , and p. zoller , _ majorana fermions in equilibrium and in driven cold - atom quantum wires _ , phys .
lett . * 106 * , 220402 ( 2011 ) .
v. mourik , k. zuo , s.m .
frolov , s.r .
plissard , e.p.a.m .
bakkers , & l.p .
kouwenhoven , _ signatures of majorana fermions in hybrid superconductor - semiconductor nanowire devices _ , science * 336 * , 1003 ( 2012 ) . s. nadj - perge , i.k .
drozdov , j. li , h. chen , s. jeon , j. seo , a.h .
macdonald , b. andrei bernevig , & a. yazdani , _ observation of majorana fermions in ferromagnetic atomic chains on a superconductor _ , science * 346 * , 602 ( 2014 ) .
r. pawlak , m. kisiel , j , klinovaja , t. maier , s. kawai , t. glatzel , d. loss , and e. meyer , _ probing atomic structure and majorana wave - functions in mono - atomic fe - chains on superconducting pb - surface _ , arxiv:1505.06078 ( 2015 ) .
j. liu , a.c .
potter , k.t .
law , and p.a .
lee , _ zero - bias in the tunneling conductance of spin - orbit - coupling superconducting wires with and without majorana end - states _ , phys .
* 109 * , 267002 ( 2012 ) .
gong , s .- f .
zhang , z .- c .
li , g. yi , and y .- s .
zheng , _ andreev reflection in a t - shaped double - quantum - dot structure induced by majorana bound states _ , j. phys .
jpn . * 83 * , 034706 ( 2014 ) ; w.j .
gong , s .- f .
zhang , z .- c .
li , g. yi , and y .- s .
zheng , _ detection of a majorana fermion zero mode by a t - shaped quantum - dot structure _ , phys .
b * 89 * , 245413 ( 2014 ) .
wang , y .- x . li , &
liu , _ resonant andreev reflection in a normal - metal / quantum - dot / superconductor system with coupled majorana bound states _ , chin .
b * 25 * , 037304 ( 2016 ) ; s .- x .
wang , y .- x .
li , n. wang , & j .- j .
liu , _ andreev reflection in a t - shaped double quantum - dot with coupled majorana bound states _ , acta phys
. sin .
* 65 * , 137302 ( 2016 ) . | we investigate the subgap spectrum and transport properties of the quantum dot placed on interface between metallic and superconducting leads and additionally side - coupled to the edge of topological superconducting ( ts ) chain , hosting the majorana quasiparticle . due to chiral nature of the majorana
states only one spin component , say @xmath0 , of the quantum dot electrons is directly hybridized with the ts wire .
we study the spin - selective interferometric patterns driven by the majorana quasiparticle .
the proximity induced on - dot pairing effectively transmits the interference onto both spin components , but each of them is characterized by completely different ( even opposite ) signatures .
to explain their origin we confront our results with the toy model , where ts chain is replaced by the usual quantum and the tunneling of @xmath1 electrons is prohibited .
we also address unique interplay of the majorana and kondo features , both appearing at zero energy . |
relativistic heavy ion collisions , involving two nuclei moving close to the speed of light , produce thousands of particles .
the collective motion of the outgoing particles is of special interest because it carries important information about the bulk properties of the sytem .
a handle into the study of such collective motion makes use of the azimuthal distributions of detected particles in non - central collisions @xcite . for a quantitative description
, such distribution is conventionally decomposed @xcite by a fourier series @xmath2 where @xmath3 denotes the angle between the particle s azimuthal angle in momentum space and the reaction plane angle .
the reaction plane is defined by the collision axis and the line connecting the centers of two nuclei .
the various coefficients in this expansion can be calculated as @xcite : @xmath4 the first and the second coefficients are refered to as directed @xmath5 and elliptic flow @xmath6 , respectively , and they both play important roles in describing the collective expansion in azimuthal space .
elliptic flow is produced by the conversion of the initial coordinate - space anisotropy into the momentum - space anisotropy , due to the developed large in - plane pressure gradient @xcite .
directed flow , which is the focus of this study , describes the side splash " of particles measured off mid - rapidity@xcite .
it probes the dynamics of the system in the longitudinal direction .
the shape and magnitude of directed flow in the vicinity of midrapidity are of special interest , specially for protons , because they are sensitive to the equation of state ( eos ) and may carry phase transition signal@xcite .
directed flow has been measured at the ags @xcite , at the cern sps @xcite , and at rhic @xcite .
in particular , the directed flow of protons at rhic has been studied in detail by the star collaboration , for au+au collisions at @xmath7 gev @xcite and at lower energies @xcite .
it is worthy of notice , that at energies studied by star , protons and antiprotons exhibit different @xmath8 slope in close - to - central collisions ( 5 - 30% at 200 gev and 10 - 40% at low energies ) , and at low energies the proton @xmath8 slope changes its sign from negative as in peripheral collisions to positive as in close - to - central collisions .
the sign - change of @xmath8 slope has been proposed as a possible signal of phase transition , it is thus necessary to identify other causes that may give rise to the similar sign - change of directed flow in the vicinity of y=0 .
indeed it is mentioned in @xcite that the observed @xmath1 for protons is a convolution of the directed flow of produced protons with that of protons transported from beam rapidity .
since transported protons come from the original projectile and target nuclei , they preserve the directed flow of spectators . produced protons , on the other hand ,
tend to flow with the bulk .
the flow direction of spectators and bulk are not necessarily equal ; the final @xmath1 observed would then be the result of a competition between the two different proton sources . in this work
we study the interplay between the directed flow of transported protons and that of non - transported protons , within the framework of the ultra - relativistic quantum molecular dynamics model ( urqmd ) @xcite .
note that although @xmath8 calculated with urqmd does not reproduce the measurement from au+au collisions at @xmath7 gev , as already pointed out in @xcite , the study of the contributions from transported and non - transported protons to the overall proton @xmath1 can offer insights into the competing dynamics which lead to the final observation of the sign of @xmath8 and its slope in the vicinity of midrapidity .
urqmd is a microscopic transport model based on the covariant propagation of all hadrons along classical trajectories in combination with stochastic binary scatterings , color string formation and resonance decay .
this microscopic transport model describes the phenomenology of hadronic interactions at low and intermediate energies ( @xmath9 gev ) in terms of interactions between known hadrons and their resonances . at higher energies(@xmath10 gev ) , the excitation of color strings and their subsequent fragmentation into hadrons dominates the multiple production of particles in the urqmd model @xcite . fig .
[ cartoon ] schematically sketches the exitation and subsequent fragmentation ( @xmath10 gev ) of a baryon - string @xcite : in the first case , shown on top of the figure , the exitation energy is too low to produce new quark pair(s ) , and there is only one baryon a in the final state . in the second case ,
one diquark - antidiquark pair and one quark - antiquark pair are created .
one of the initial quarks ( white circle ) combines with a newly produced diquark ( black circles ) to form the baryon b , the newly produced anti - diquark ( brown circles ) combines with a newly produced anti - quark to form the antibaryon c , and the newly produced quark together with the initial diquark form another baryon d. the third case is similar to the second one , and it shows the produced baryons and a meson . at the end baryon a has three initial quarks ( 3 iq , where iq stands for initial quark " ) , baryon d and h each have two initial quarks ( 2 iq ) , baryon b has one initial quark ( 1 iq ) , and baryon g has zero initial quark ( 0 iq ) .
the two anti - baryons ( c and g ) each have zero initial quark ( 0 , iq ) .
the more initial quarks can be found in a particle , the more initial information that particle carries . by tracing the number of initial quarks in a particle ,
we classify particles as produced or transported particles . in this article , baryons with three initial quarks
are regarded as transported baryons , and all other baryons that have at least one produced quark are regarded as non - transported .
a particular subgroup of the latter are the produced baryons with zero initial quarks .
produced baryons should be similar to anti - baryons in many respects as both groups are made of produced quarks .
after the process of fragmentation excitation of baryon - strings , baryons produced ( mostly unstable baryons ) are subject to multiple scatterings .
when the energy of a binary collision is lower than 5 gev ( @xmath9 gev ) , there will be no strings involved and there are no new quark pairs produced . during this process
the unstable baryons do decay but the number of constituent quarks does not change .
for example , in the case of a @xmath11 decay into a proton and a pion , one can still identify how many quarks in the daughter proton are originated from initial quarks by tracing the quark constituent of its parent , @xmath11 .
gev ( left ) and 200 ( right ) gev . ] fig .
[ yield ] shows the rapidity distribution of protons and anti - protons for au+au collisions at @xmath12 gev and 200 gev , calculated with the urqmd model .
the selection on centrality ( 10%-70% ) and transverse momentum @xmath13 ( @xmath14 gev/@xmath15 ) are chosen to match those used in @xcite .
the distribution of transported , non - transported , and produced protons are drawn separately .
as expected , the transported protons peak at beam rapidity and the produced ones , at midrapidity . not surprisingly ,
the distribution of produced protons is similar to that of anti - protons ( to first order ) , and the remaining difference comes from their separate production mechanisms .
for example , for each produced baryon there is an accompanying anti - baryon produced but that statement is not necessarily true for anti - baryons .
in addition , antiprotons are subject to annihilation while protons are not .
the transported ( produced ) protons accounts for 12% ( 2.1% ) of inclusive protons at 11.5 gev , and 0.91% ( 50% ) at 200 gev .
the baryon stopping effect is seen at both energies . at midrapidity , transported protons dominate over produced protons at 11.5 gev , while at 200 gev , the opposite is true .
note that the sum of the yield of transported and produced protons does not give the inclusive proton yield because some of the protons can be regarded neither as transported nor produced .
those are protons that have 2 or 1 initial quarks , and they account for a large fraction of the total number of protons . by definition
the inclusive proton yield is a sum of transported and non - transported proton yields . for protons and anti - protons in au+au collisions at @xmath12 gev for centrality 10%-70% , from urqmd calculation .
see text for explanation on individual protons groups . ] to illustrate the effect of the mixture of initial and produced quarks on the shape of @xmath8 figure [ v1_y ] shows @xmath8 plotted for all proton ( antiproton ) groups for au+au collisions collisions at @xmath12 gev .
it is seen that for transported protons ( 3 iq ) , the @xmath8 slope is positive at midrapidity . with the replacement of even only one initial quark by a produced quark ,
as shown by the case ( 2 iq ) , the @xmath8 shape becomes extremely flat at midrapidity . as more initial quarks are replaced , the @xmath8 slope changes sign to negative and continues to decrease monotonically with descreasing number of initial quarks . for
produced protons ( 0 iq ) , the sign of the @xmath8 slope agrees with that of antiprotons as expected because both are produced particles .
the @xmath8 of inclusive protons is a convolution of @xmath8 and the relative yields of each individual proton group . for clarity
, the figures shown below do not display the two intermediate proton groups ( 1 iq and 2 iq groups ) .
slope as a function of centrality for protons and anti - protons for au + au collisions at @xmath16 11.5 ( left ) and 200 ( right ) gev , calculated with the urqmd model . ] in fig .
[ v1_cent ] we present the centrality dependence of @xmath17 slope at midrapidity in au+au collisions at @xmath12 gev and 200 gev , for transported , non - transported , produced , and inclusive protons .
the slope for antiprotons is also shown for comparison . here
the slope is extracted from the normalized ( @xmath18 ) rapidity distribution , where @xmath19 is the beam rapidity .
our results are obtained by integrating over transverse momentum ( @xmath20 ) from 0.4 to 1.0 gev as done in @xcite . at 11.5 gev ,
as the overlap between colliding ions range from central to peripheral , the @xmath17 slope of transported protons increases rapidly , reaches a maximum at @xmath2120% centrality and then decreases .
the opposite trend is seen for produced protons .
the slope of @xmath17 extracted from non - transported protons is negative in most centralities , ranges between positive and negative values , as non - transported protons also include intermediate proton groups other than produced protons .
the @xmath17 slope of inclusive protons is the result of competing effects from directed flow between transported and non - transported protons . because the yield of non - transported protons dominate over that of transported protons , in peripheral collisions the sign of the @xmath17 slope agree with that of non - transported protons . as the centrality decreases , transported proton @xmath17 slope increases .
this effect gradually overcomes the effect of directed flow from non - transported protons and change the sign of the @xmath17 slope to positive . in most central collisions , @xmath1 from all groups
vanishes due to the symmetrical shape of the collision .
this explains the sign change of @xmath17 seen in data @xcite .
similarly , the trend of the @xmath17 slope at 200 gev can be explained with similar arguments .
note that at 200 gev , the @xmath17 slopes are negative at all centralities .
this can be due to the fact that at 200 gev the transported proton yield is too small and the overall effect is driven by non - transported protons . for the same reason , a difference between proton and antiproton @xmath1 is observed as seen in the inset , similar to what is shown in @xcite .
slope ( d@xmath22 ) around mid - rapidity ( @xmath23 ) as a function of incident energy . ] the energy dependence of directed flow for protons and anti - protons is presented in fig .
[ energydependence ] . based on urqmd calculation ,
the slopes of @xmath17 extracted from produced protons are negative and they follow more or less a similar behavior as the one extracted from antiprotons , while the @xmath17 slopes for transported protons are all positive .
the @xmath17 slope for inclusive protons is a convoluted effect produced by transported and non - transported protons .
it is positive at 7.7 , 11.5,19.6 , 27 and 39 gev , decreases monotonically with increasing energy for @xmath011.5 gev and beyond , and becomes negative at 62.4 gev .
the sign change of @xmath17 slope is also seen in experimental data @xcite , although the energy at which the @xmath17 slope changes sign is lower ( 11.5 gev ) .
these findings suggest that the apparent collapse " of @xmath1 , defined by a flat shape of the proton @xmath17 , or the sign - change of the proton @xmath17 slope , can be explained by the interplay between transported protons and non - transported protons .
directed flow of inclusive , transported protons and non - transported protons ( including produced protons ) , as well as antiprotons , have been studied with the urqmd model at rhic energies .
it is found that the integrated directed flow decreases monotonically as a function of collision energy for @xmath011.5 gev and beyond .
however , the sign - change of directed flow of inclusive protons , as a function of centrality and collision energy , can be explained by the competing effect of directed flow between transported and non - transported protons .
similarly , the difference in directed flow between protons and antiprotons can be explained .
our study identified alternative cause of the @xmath1 sign - change other than the anti - flow component of protons alone which is argued to be linked to a phase transition .
this work is supported in part by national natural science foundation of china under grants 11075060 and 11135011 .
a. tang is supported by u.s .
department of energy under grants de- ac02 - 98ch10886 and de - fg02- 89er40531 .
we wish to thank g. wang , d. keane and j. y. chen for their valuable comments and suggestions .
we thank r. debbe for english corrections .
99 for a recent review , see : s. voloshin , a. poskanzer and r. snellings , volume 23 , in _ relativistic heavy ion physics _ , published by springer - verlag .
edited by r. stock .
doi : 10.1007/978 - 3 - 642 - 01539 - 7 . arxiv:0809.2949 .
s. a. bass , m. belkacem , m. bleicher , m. brandstetter , l. bravina , c. ernst , l. gerland , m. hofmann , s. hofmann , j. konopka , g. mao , l. neise , s. soff , c. spieles , h. weber , l. a. winckelmann , h. stocker , w. greiner , ch .
hartnack , j. aichelin and n. amelin , prog .
nucl . phys . *
41 * , 255 ( 1998 ) .
+ m. bleicher , e. zabrodin , c. spieles , s.a .
bass , c. ernst , s. soff , l. bravina , m. belkacem , h. weber , h. stocker , w. greiner , j. phys . g. nucl .
* 25 * , 1859 ( 1999 ) . | the directed flow of inclusive , transported and non - transported ( including produced ) protons , as well as antiprotons , has been studied in the framework of ultra - relativistic quantum molecular dynamics approach ( urqmd model ) for au+au collisions at @xmath07.7 , 11.5 , 19.6 , 27 , 39 , 62.4 and 200 gev . the rapidity , centrality and energy dependence of directed flow for various proton groups are presented .
it is found that the integrated directed flow decreases monotonically as a function of collision energy for @xmath011.5 gev and beyond .
however , the sign - change of directed flow of inclusive protons , seen in experimental data as a function of centrality and collision energy , can be explained by the competing effect of directed flow between transported and non - transported protons .
similarly the difference in directed flow between protons and antiprotons can be explained .
our study offers a conventional explanation on the cause of the @xmath1 sign - change other than the antiflow component of protons alone which is argued to be linked to a phase transition . |
single dish millimeter observations of hcn and cs emission from nearby galaxies show that most spiral galaxies , not just starburst galaxies , have an appreciable amount of dense ( @xmath1 10@xmath10 @xmath9 ) gas in their bulges ( helfer & blitz 1993 and references therein ) .
these observations are in good agreement with what is seen in the inner @xmath11 parsecs of the milky way , where strong , diffuse emission from cs ( bally et al .
1987 ) and hcn ( jackson et al .
1996 ) is observed despite the moderate star formation rates there .
( hcn and cs trace gas densities of @xmath12 10@xmath10 @xmath9 in galaxies ; this is two orders of magnitude higher than the density required to excite co. ) the ubiquity of large - scale emission from hcn and cs over hundreds of parsecs in the centers of galaxies is surprising when compared with the known properties of giant molecular clouds ( gmcs ) in the solar neighborhood , which contain few and relatively small ( @xmath13 pc - scale ) clumps of dense gas and these only where stars are actively forming or where stars have formed very recently ( i.e. where there is a local source of pressure , not common to the gmc as a whole ) . in galaxies ,
single - dish millimeter wavelength beams typically cover tens to hundreds of gmc - sized diameters . as part of our program to measure the degree to which different galactic environments affect the properties of molecular clouds , we began a program with the berkeley - illinois - maryland association ( bima ) interferometer to survey eight nearby galaxies in hcn emission in order to measure the distribution and amount of dense gas on size scales of individual gmcs or small associations of gmcs .
we consider these galaxies to be `` normal '' since they do not have prolific circumnuclear starbursts or active galactic nuclei ; however , like most spiral galaxies , these galaxies have low - level nuclear line emission ( ho , filippenko & sargent 1993 ) . in this paper
, we present the results of the bima survey .
we also present a comparison of the hcn emission from one of the sources , ngc 6946 , with its co j = 1 - 0 emission ( regan & vogel 1995 ) . a comparison with the co emission from the other sources will be the topic of a future paper .
the sources and their coordinates are listed in table 1 .
the sources were selected from the single - dish survey of helfer & blitz ( 1993 ) as galaxies with strong co emission that were also detected in hcn with a single pointing using the nrao 12 m telescope .
the data were collected with the bima interferometer ( welch et al .
1996 ) , which then comprised 6 antennas , between 1994 february 15 and 1995 may 02 ; the observations included data from up to three array configurations . for each source , the receivers were tuned to the redshifted frequency of the hcn j = 1 - 0 transition ( @xmath14@xmath15 = 88.61 ghz ) .
the data were processed using the miriad package ( sault , teuben , & wright 1995 ) .
the time variations of the amplitude and phase gains were calibrated using short observations of nearby quasars every @xmath1 30 minutes ; a planet or strong quasar was observed to set the absolute flux scale as well as to calibrate the spectral dependence of the gains across the if passband .
the digital correlator was configured to achieve a maximum spectral resolution of 1.56 mhz ( 5.3 km s@xmath16 ) .
for some of the observations , there was an intermittent phase lock on one of the antenna receivers that was not discovered until after the observations ; in these cases all baselines involving that antenna were eliminated from further data reduction .
for each source , the calibrated data were smoothed to 21.1 km s@xmath16 resolution , then gridded and fourier transformed using natural weighting .
the visibilities were also weighted by the inverse of the noise variance , which was determined from the system temperatures and from the gains of the individual antennas .
we cleaned the maps using the standard hgbom algorithm .
maps of integrated intensity were made by summing those channels with emission after clipping the channel maps at a 1 @xmath17 level .
the beam sizes and noise levels of the final maps are listed in table 2 .
the noise levels are somewhat underestimated relative to the formal uncertainties , since they were determined from the clipped moment maps .
the absolute flux calibration in all maps is probably accurate to @xmath3 30% .
we have also used bima to image co in the centers of three of the galaxies ( ngc 3628 , ngc 4826 , and m83 ) with detected hcn emission .
a fourth galaxy , ngc 6946 , has been imaged in co using bima by regan & vogel ( 1995 ) .
the co emission , unlike the hcn emission ( see below ) , is significantly `` resolved out '' in the interferometric images of these galaxies ; it is therefore necessary to fill in the zero - spacing flux with a single - dish telescope .
we are currently still making these measurements for the first three sources ; we therefore defer further discussion of the co results for these galaxies to a future paper . in the case of ngc 6946 , we measured the co short - spacing flux with the nrao 12 m telescope and combined these data with regan & vogel s bima map .
we observed on 95 june 2021 in the newly implemented `` on - the - fly '' ( otf ) scheme at the nrao 12 m ( emerson et al .
, in preparation ) and covered a region roughly 5 on a side .
we observed orthogonal polarizations using two 256 channel filterbanks , each with a spectral resolution of 2 mhz per channel .
the data were gridded and a linear baseline removed from the resulting data cubes in aips ; the data were then transferred to the miriad package for further processing .
details of the single dish and interferometric data combination were very similar to those described in helfer & blitz ( 1995 ) for the case of ngc 1068 .
we detected hcn in five of the eight galaxies : ngc 3628 , ngc 4826 , ngc 5194 ( m51 ) , ngc 5236 ( m83 ) , and ngc 6946 ; their images are presented in figure [ hcnmaps ] along with optical images from the digitized sky survey .
the spectra from the positions of peak hcn emission in each of the detected sources are shown in figure [ spectra ] . in each case , the hcn images show compact structure , with the detected emission confined to the central @xmath1 500 pc diameter ; the 510 fwhm synthesized beam sizes correspond to linear resolutions of 125200 pc at the distances of the galaxies . while the hcn emission is compact , it appears resolved by the interferometer in each map . in order to investigate the effect of the lack of @xmath18 sampling at small spatial visibilities , we compared the integrated intensities measured with the interferometer to the single - dish fluxes of helfer & blitz ( 1993 ) .
table 3 lists the integrated intensities measured at bima along with those measured at the nrao 12 m and the fraction of the single - dish flux recovered by the interferometer . with the possible exception of m51
( see @xmath19 3.1 ) , the interferometer appears to recover all the single - dish flux measured by helfer & blitz .
this means that the maps shown in figure [ hcnmaps ] are a reliable representation of the distribution of the hcn emission , constrained of course by the usual limitation of the signal to noise ratios . allowing for low - level , more diffuse emission detected at @xmath20
, we conclude that the hcn emission is confined to the central kiloparsec of the galaxies ( it is perhaps slightly more extended in m51 ) .
the spatial extent of the hcn emission is in good agreement with what is seen in the milky way ( jackson et al . 1996 ) and in ngc 1068 ( helfer & blitz 1995 ) . the galaxies in this survey were originally selected from a list of the brightest extragalactic co emitters ( see helfer & blitz 1993 ) .
the following are brief descriptions of the individual sources .
ngc 3628 this galaxy is a member of the leo triplet ( along with ngc 3623 and ngc 3627 ) .
it is a nearly edge - on sbc galaxy with a prominent and irregular dust lane .
its nuclear region contains a modest starburst ( e.g. condon et al .
1982 , braine & combes 1992 ) .
the co from ngc 3628 ( young , tacconi , & scoville 1983 ; boiss , casoli , & combes 1987 ; and israel , baas , & maloney 1990 ) is strongly peaked in the inner kiloparsec and has a similar extent and mass as that in the center of the milky way ( boiss et al .
the hcn in ngc 3628 is resolved and appears elongated in the east - west direction .
there is emission at the 2 @xmath17 level to the northwest of the central source that may be associated with gas further along the major axis .
the hcn , co and radio continuum centers of ngc 3628 all peak some 21 to the southeast of the optical nucleus ; however , the optical nucleus is heavily obscured and its position is highly uncertain ( boiss et al .
1987 ) .
ngc 4826 this sab galaxy has been called variously the `` black eye , '' the `` evil eye , '' or somewhat more optimistically ( rubin 1994 ) the `` sleeping beauty '' galaxy for its conspicuous dust lane ( see sandage 1961 ; see also the cover of _ the astronomical journal _ , 1994 , 107 , 1 ) .
ngc 4826 has gained notoriety recently for the discovery that the inner kiloparsec - scale disk is counterrotating with respect to the larger scale rotation of the galaxy ( braun , walterbos , & kennicutt 1992 ; rubin 1994 ; braun et al . 1994 ) .
the co emission ( casoli & gerin 1993 ) is confined to the inner @xmath21 .
the hcn emission appears symmetric and centrally peaked in ngc 4826 .
ngc 5194 ( m51 )
m51 is the prototypical grand design spiral and is one of the best studied galaxies in co emission .
interferometric co images of the nuclear region of m51 ( e.g. lo et al .
1987 ; rand & kulkarni 1990 ; and adler et al . 1992
; the last includes zero - spacing flux ) show a notable lack of a single central concentration of co , despite the strong molecular emission associated with the spiral arms in the nuclear region .
m51 is the only galaxy of the five detected in hcn at bima that does not have a strong central concentration of co. in contrast to the co emission , the hcn emission does appear to be centrally concentrated , though there is a significant contribution to the total flux from low - level ( @xmath20 ) emission .
we note that kohno et al .
( 1996 ) mapped the central concentration as well as more extended structure in their hcn map of m51 observed with the nobeyama millimeter array .
the linear extent of the low - level ( @xmath20 ) hcn emission in the bima map is somewhat larger than those of the other four galaxies studied here , and the interferometer may have `` resolved out '' a nonnegligible contribution to the total flux in this source ( table 3 ) . for m51 ,
the flux of structures larger than @xmath1 18 is attenuated by @xmath12 50% .
ngc 5236 ( m83 ) this well - studied source is a grand design , sc / sbb spiral galaxy with strong circumnuclear star formation within the central few hundred pc ( e.g. gallais et al . 1991 ) .
the hcn emission is strongest at the position of the peak of the radio continuum emission ( condon et al . 1982 ; turner & ho 1994 ) ; the condensations to the east and to the south of the strongest hcn emission are also apparent in the radio continuum .
the synthesized beam of the interferometer is quite elongated because of the foreshortening of the north - south baselines toward this low declination source .
the north - south elongation of the hcn emission is therefore almost certainly an artifact of the observations .
ngc 6946 this late - type , grand design spiral galaxy contains a moderate starburst ( turner & ho 1983 ) . within the inner 1.5 kpc diameter
, the co in this galaxy has a non - axisymmetric , north - south distribution that has been interpreted as a bar ( regan & vogel 1995 and references therein ) .
however , regan & vogel ( 1995 ) combined new co and k - band observations and showed that the co traces gas on the trailing side of spiral arms ; their observations are consistent with what is expected for the gas and stellar response to a spiral density wave rather than a bar
. the co peaks up strongly in the inner 300 pc diameter of ngc 6946 .
the hcn emission is detected in this region and is resolved and slightly extended in an east - west direction , with an additional elongation towards the northwest ( there is a similar northwest extension in the co map of regan & vogel 1995 ) .
we discuss this source more fully in the following section .
the ratio of the 3 mm integrated intensities , @xmath23 , may be used as a qualitative measure of the molecular gas density ( e.g. helfer & blitz 1996 ) . in order to determine any line ratio from interferometric measurements
, one must first take into account the possibility that the flux measured with the interferometer is missing a significant contribution from large - scale structures in the maps . in ngc 6946
, the interferometer recovers all the single - dish hcn flux to within the errors of the measurement ( table 3 ) . for the co , the regan & vogel ( 1995 ) bima map recovered about half the single - dish flux ; we therefore modeled the short spatial frequency visibilities from the nrao 12 m data ( @xmath19 2.2 ) and combined these with the bima co map .
the co distribution and flux in the resulting map did not change appreciably interior to r = 15 ; at larger radii , the most dramatic flux increases were distributed over radii from 2050 , though the shape of the structures remained about the same . with the fully - sampled co map
, we can now make a legitimate comparison of the hcn and co intensities for this source .
to determine the ratio @xmath23 in ngc 6946 , we convolved the co map to match the resolution of the hcn image , and we converted both intensities to a main beam brightness scale ( @xmath24 @xmath25 ) ; the ratio was computed only for regions with detected hcn emission ( i@xmath26 , or r @xmath13 15 ) .
the resulting ratio map of i@xmath6/i@xmath7 is presented in figure [ n6946rat ] . although the emissions from co and hcn both peak at the center of ngc 6946 , the distribution of the hcn / co ratio is saddle - shaped , with the highest values to the east and west of the nucleus by about 7 ( 175 pc ) and the lowest values to the northwest and southeast of the nucleus by about 5 ( 125 pc ) . at first
, it seems surprising that the ratio does not rise monotonically to the central position .
however , at the small size scales resolved by the interferometer ( @xmath27 pc ) , the characteristics of individual molecular clouds start to dominate the distribution , rather than the integrated effects of dozens of gmcs
. it may be that the small - scale ratio is dominated by local effects from individual clouds ; this effect is seen within the central few hundred pc of the milky way , where the i@xmath28/i@xmath7 ratio ( helfer & blitz 1993 ) and the i@xmath6/i@xmath7 ratio ( jackson et al . 1996 ) look very clumpy and irregularly distributed . in the milky way , features like the sgr a and b gmcs are characterized by relatively high values in the ratio maps . the average ratio over the extent of the hcn emission ( @xmath29 , or 300 pc diameter ) in ngc 6946 is @xmath30 , with a peak value of 0.19 and a minimum of 0.049 .
the range of i@xmath6/i@xmath7 in ngc 6946 is very similar to what is seen in the milky way by jackson et al .
( 1996 ) ; on small scales within the inner few degrees of the milky way , i@xmath6/i@xmath7 ranges from 0.04 to 0.12 , and the average over the extent of the hcn emission , or r @xmath31 300 pc , is 0.08 ( see below ) .
we can compare the ratio we measure in the central 300 pc of ngc 6946 with those measured with larger apertures : the single dish ratios of i@xmath6/i@xmath7 in ngc 6946 are 0.063 @xmath3 0.007 at a resolution of 24 ( 600 pc ) ( nguyen - q - rieu et al .
1989 ; weliachew , casoli , & combes 1988 ) and 0.025 @xmath3 0.003 at a resolution of @xmath32 ( 1500 pc ) ( helfer & blitz 1993 ) .
figure [ n6946.ratio_vs_r]@xmath33 shows this radial distribution of the average integrated i@xmath6/i@xmath7 ratio .
it is important to note that the points shown in figure [ n6946.ratio_vs_r]@xmath33 represent the _ average _ ratios over the area enclosed at the radius @xmath34 , i.e. that the plot represents the _ integrated _ ratios i@xmath6/i@xmath7 as a function of @xmath34 .
the monotonic falloff in the ratio with radius is simply a result of the confinement of the hcn emission to the inner @xmath34 = 150 pc , while the co is distributed over a much larger radius ( there is detectable emission at least to @xmath34 = 3.5 kpc , tacconi & young 1989 ) .
what is perhaps a more interesting quantity physically is the _ annular _ ratio measured as a function of radius ; that is , if i@xmath6/i@xmath7 is @xmath35 measured as an average from @xmath36 pc , what is the value of i@xmath6/i@xmath7 from @xmath5 pc ( where 800 pc is the radius of the nrao beam in helfer & blitz 1993 ) ?
we can set an upper limit to i@xmath6/i@xmath7 in this annulus by comparing the bima data with the nrao 12 m hcn flux . since the bima measurement recovered @xmath37 of the single - dish hcn flux measured at the nrao 12 m ( table 3 ) , let us assume that 20% of the single - dish hcn flux is distributed at radii larger than the interferometer was able to measure , yet within the half power beam area of the the nrao 12 m
that is , radii within the annulus @xmath5 pc .
( this is a conservative estimate , since any `` missing '' large - scale flux could also contribute to the flux at the central position . ) if we then measure the flux in the regan & vogel co map from @xmath5 pc , we find that the ratio i@xmath6/i@xmath7 in this annulus is at most 0.01 .
this ratio is an order of magnitude lower than that measured over the central @xmath2 pc , as shown in figure [ n6946.ratio_vs_r]@xmath38 .
we did not detect hcn emission from ngc 4321 ( m100 ) , ngc 4527 , or ngc 4569 .
while it is possible that the dense structure in these galaxies is so extended that the interferometer resolves out the single - dish hcn emission ( for these sources , the flux of structures larger than @xmath1 20 is attenuated by @xmath12 50% ) , it is also likely that the observations simply were not sensitive enough to detect the hcn from these sources .
for those observations which suffered from an intermittent phase lock ( see @xmath19 2.1 ) , the antenna that was flagged was one of the two that made up the shortest baseline pair ; thus the calibration solution may not have been reliable and also the zero spacing problem may have been exacerbated for these observations .
spectroscopic studies of cs and hcn emission in normal external galaxies suggest that most spiral galaxies contain an appreciable amount of gas at densities of @xmath39 in their centers ( mauersberger et al .
1989 ; sage , solomon , & shore 1990 ; nguyen - q - rieu et al .
1992 ; israel 1992 ; helfer & blitz 1993 ) .
the maps in figure 1 show the distribution of that dense gas , namely , that it is confined to the central kiloparsec of the sources imaged .
this situation appears to be very similar to that seen in the milky way , where widespread emission from the dense gas tracers cs ( bally et al .
1987 ) and hcn ( jackson et al .
1996 ) is found only within the central @xmath1 500 pc diameter .
what does the distribution of dense gas tell us about the physical conditions in the molecular gas as a function of its location in a galaxy ? to investigate the physical conditions rather than the total gas content , we normalize the hcn to that of the co and consider the ratio @xmath23 . if the kinetic temperature of the gas responsible for the cospatial hcn and co emissions is about the same , then @xmath23 may be considered as a qualitative indicator of the density or thermal pressure in the gas .
( indeed , since the j = 1 state lies only 5.5 k above the ground state for co and 4.3 k above ground for hcn , even rather cold gas has the energy to populate the j = 1 state for both molecules .
it is the molecular density that is more important in determining the excitation .
see helfer & blitz 1996 . ) in ngc 6946 ( @xmath19 3.2 ) , @xmath23 is 0.11 @xmath3 0.01 averaged over the central @xmath40 pc , whereas we deduce an upper limit of @xmath41 averaged in the annulus @xmath5 pc , a region that includes the inner part of the disk .
a similar radial dependence of @xmath23 is seen in the unusual seyfert / starburst hybrid galaxy ngc 1068 , where the ratio approaches 0.6 in the central @xmath34 = 175 pc ( helfer & blitz 1995 ) , and the ratio falls off monotonically to about 0.1 at the large reservoir of molecular gas at about 1 kiloparsec from the nucleus . in the milky way ,
the ratio @xmath23 is about 0.081 to table 4 . ]
@xmath3 0.004 averaged over the central @xmath42 pc ( jackson et al . 1996 ) ; between @xmath43 kpc in the plane of the milky way , the average ratio is @xmath1 0.026 @xmath3 0.008 , and in solar neighborhood gmcs , we measure @xmath23 ratios of 0.014 @xmath3 0.020 when averaged over @xmath1 50 pc gmcs ( helfer 1995 and these are upper limits to the ratio averaged over several hundred parsecs ) .
these numbers are summarized in table 4 .
what seems apparent from these comparisons is that @xmath23 is a strong function of galactocentric radius or more precisely , that there is at least a bimodal distribution in the ratio in normal galaxies : the ratio at the center is substantially higher than elsewhere in the galaxy . even though local effects can dominate on scales of individual gmcs ( @xmath19 3.2 ) , the average ratio of dense gas emission is highest at a galaxy s center and drops at larger distances from the center . furthermore , the general agreement between the ratios in the central @xmath1 500 pc of ngc 6946 and the milky way suggest that the physical conditions in the centers of the two galaxies are similar .
the measurement that the ratio @xmath23 is 5 to 10 times higher in the bulge regions of the milky way and ngc 6946 than in their disks suggests that the physical conditions of the molecular gas in the bulge and disk regions are very different .
furthermore , the presence of dense gas on size scales of @xmath1 500 pc suggests that the internal pressure is very high in the molecular gas .
let us assume that the intrinsic line widths of the clouds are dominated by nonthermal , bulk motions as in local clouds , and that their intrinsic linewidths are @xmath44 1 km s@xmath16 .
although we can not model the density accurately with the observation of a single transition of hcn or co , a simple lvg analysis of the line ratios suggests that a line ratio of @xmath23 = 0.11 implies densities of 10@xmath45 @xmath9 for gas at a kinetic temperature in the range t@xmath46 = 15 70 k. these densities are consistent with what is measured in the milky way bulge molecular clouds , where n(h@xmath47 ) @xmath48 ( gsten 1989 ) .
if we take the gas densities in ngc 6946 to be @xmath49 , then the typical kinetic pressure throughout the molecular gas is @xmath50
@xmath9 k. spergel & blitz ( 1992 ) considered the effects of the extended , hot coronal gas in the bulge of the milky way on the thin molecular layer that is embedded within it , and they argued that the pressure in the center of the galaxy is two to three orders of magnitude greater than that in the solar neighborhood .
how does such an extraordinary difference in the environmental pressure affect the properties of molecular clouds ?
in disk gmcs , the external pressure of the ambient ism ( @xmath51 k ) is small compared with the pressure from the self - gravity of a cloud ( @xmath52 k ) , and a source of `` local '' pressure ( i.e. , ongoing or recent star formation ) is required to support any localized high - density clumps within the cloud . in the high - pressure ( @xmath53 k , estimated for the milky way from the x - ray measurements of yamauchi et al .
1990 ) environments of bulge molecular gas , on the other hand , the gmcs need not be self - gravitating in fact , only the most massive clouds _ could _ have the jeans masses required to be self - gravitating . in bulge clouds , then , it is the external pressure of the environment that dominates , and these high pressures can support the dense gas throughout the molecular component ( regardless of whether there is any active star formation in the gmcs ) .
these arguments are easily extended to observations of external galaxies , which typically also have x - ray emission associated with their bulges ( fabbiano , kim , & trinchieri 1992 ) . from our observations of hcn in the bulges of external galaxies
, it appears that the physical conditions in molecular gas in the centers of galaxies are much more similar to each other than they are to local gmcs in the milky way . in the milky way ,
there is a pronounced offset of @xmath1 80 pc in the position of the peak radio continuum emission ( sgr a * , @xmath54 ) and the centroid of the dense gas ( traced by cs and hcn ) distribution ( @xmath55 ) .
there is also evidence for an offset of @xmath1 100 pc in ngc 1068 between the peaks of the hcn emission and the radio continuum emission ( helfer & blitz 1995 ) , and a kinematic analysis of the molecular gas suggests the existence of an @xmath56 mode ( i.e. a dipole asymmetry in the mass distribution ) in ngc 1068 . whether it is the radio continuum emission or the molecular gas that traces the center of the mass distribution ,
the dynamical timescale of the offsets is brief ( @xmath57 10@xmath58 years ) . if the offsets are indeed ephemeral , and not a steady - state condition of the galaxies , then it is appropriate to look for a source of the instability that causes them .
how common are such offsets between the peaks of the dense molecular gas distribution and the radio continuum emission in galaxies ?
we compared the positions of peak hcn emission with those of the peak radio continuum in our sample ( figure 1 , table 5 ) .
while three of the galaxies show a reasonable coincidence between the two positions , two of the five sources , ngc 4826 and m51 , show significant offsets : in ngc 4826 , the offset is @xmath59 or 74 pc ; in m51 , it is @xmath60 or 130 pc .
( in m83 , there is a @xmath61 or 40 pc offset , but because of the elongated beams and extended distribution in both hcn and the radio continuum , the peak positions are less certain . )
it appears that these offsets are a common feature in spiral galaxies . in this sample ,
two out of five detected galaxies have offsets ; we have already mentioned the offsets in the milky way and in ngc 1068 . in a recent study of the k - band morphology in 18 face - on spiral galaxies ,
rix & zaritsky ( 1995 ) found that about one third of the disks have significant @xmath56 modes at 2.5 disk exponential scale lengths .
it may be significant that the two galaxies in our sample that show offsets , ngc 4826 and m51 , also share the characteristic that they have low - level nonstellar nuclear activity ( both are liners ; these are low - level active galactic nuclei , or agn ) ; the other three galaxies detected here ( ngc 3628 , m83 , and ngc 6946 ) have low - level starburst activity instead ( l. ho , private communication ) .
a recent study by ho et al .
( 1993 ) suggests that up to 80% of the 500 brightest galaxies in the northern sky harbor some kind of activity in their nuclei ; of these , about half are classified as liners , and half are starbursts .
it is surprising both that _ most _ spiral galaxies appear to show some kind of nuclear activity and also that the activity seems to be roughly evenly divided between stellar and nonstellar mechanisms .
the offsets of the dense gas from the peak of the radio continuum in ngc 4826 and m51 , as well as that in ngc 1068 ( a galaxy with a more energetic agn , but similar conceptually to the liners ) may help to distinguish empirically the kind of activity that dominates in a given galaxy s nucleus .
we have presented the results of a survey of hcn emission from eight nearby spiral galaxies made with the bima interferometer .
these observations comprise the first high - resolution survey of dense ( @xmath62 @xmath9 ) gas from a sample of relatively normal galaxies , rather than galaxies with prolific starburst or nuclear activities .
we imaged five of the eight galaxies in hcn : ngc 3628 , ngc 4826 , ngc 5194 ( m51 ) , ngc 5236 ( m83 ) , and ngc 6946 . to within the uncertainties ,
the interferometer recovers all of the single - dish flux measured for each source in a single pointing at the nrao 12 m telescope ( helfer & blitz 1993 ) ; this implies that there is not a significant contribution to the hcn fluxes from extended emission in the disks of the galaxies . in
all the galaxies observed , the hcn emission is confined to the central kiloparsec of the sources .
we added zero - spacing data from the nrao 12 m telescope to the bima co map of ngc 6946 by regan & vogel ( 1995 ) in order to compare the ratio of hcn to co intensities in this galaxy .
the ratio @xmath23 ranges from 0.050.2 within the central @xmath2 pc ; the average ratio over this region is @xmath63 .
a comparison with single - dish observations allows us to place an upper limit of @xmath64 in the annulus @xmath5 pc in ngc 6946 .
the extent of hcn emission in ngc 6946 , @xmath65 pc , and the ratio @xmath66 in this region are similar to what is observed in the milky way ( @xmath11 pc , @xmath67 , jackson et al .
1996 ) ; this suggests that the physical conditions in the centers of these two galaxies are similar .
furthermore , in ngc 6946 , ngc 1068 , and the milky way , the ratios at the centers are 5 to 10 times higher than those in the disks of these galaxies .
this result is consistent with an enhancement of two to three orders of magnitude in the pressure of the bulges compared with the disks ( spergel & blitz 1992 ) .
it appears that the physical regions in the centers of galaxies are much more like each other than the conditions in the center of a galaxy relative to its disk . in ngc 4826 and m51 , as in the milky way and in ngc 1068
, there is a linear offset of @xmath1 100 pc between the dense gas distribution and the peak of the radio continuum emission .
these offsets appear to be a common feature in galaxies and may indicate that their disks are non - axisymmetric
. it may be significant that of the five galaxies imaged in hcn , ngc 4826 and m51 are liners , whereas the other three are starbursts .
we did not detect hcn in three galaxies with positive single - dish hcn emission : ngc 4321 ( m100 ) , ngc 4527 , and ngc 4569 .
it could be that the emission is extended enough in these sources that the interferometer resolved out any detectable emission ; however , we can not rule out the possibility that there was some intrinsic problem with the observations of these sources .
we thank the referee , paul ho , for his careful reading and suggestions ; these helped us to improve the manuscript .
we thank mike regan for providing us with the bima ngc 6946 co data , and we thank darrel emerson , phil jewell , tom folkers and the staff of the nrao 12 m telescope for assistance with the otf observations and data reduction .
luis ho helped with the early stages of the bima observations .
we thank kotaro kohno for kindly providing us with the nro map of hcn in m51 prior to publication .
tth thanks jack welch for hospitality while visiting uc - berkeley .
this research was partially supported by a grant from the national science foundation , with additional support from the state of maryland .
adler , d.s .
, lo , k.y . ,
wright , m.c.h .
, rydbeck , g. , plante , r.l .
, & allen , r.j .
1992 , , 392 , 497 bally , j. , stark , a.a . ,
wilson , r.w . , &
henkel , c. 1987 , , 65 , 13 . 1988 , , 324 , 223 boiss , p. , casoli , f. , & combes , f. 1987 , , 173 , 229 braine , j. & combes , f. 1992 , , 264 , 433 braun , r. , walterbos , r.a.m .
, & kennicutt , r.c .
1992 , nature , 360 , 442 braun , r. , walterbos , r.a.m . ,
kennicutt , r.c . , & tacconi , l.j .
1994 , , 420 , 558 casoli , f. & gerin , m. 1993 , , 279 , l41 condon , j.j . ,
condon , m.a .
, gisler , g. , & puschell , j.j .
1982 , , 252 , 102 fabbiano , g. , kim , d .- w . , &
trinchieri , g. 1992 , , 80 , 531 gallais , p. , rouan , d. , lacombe , f. , tiphene , d. , & vauglin , i. 1991 , , 243 , 309 helfer , t.t .
1995 , ph.d .
thesis , u. maryland helfer , t.t . & blitz , l. 1993 , , 419 , 86 . 1995 , , 450 , 90 ho , l.c . ,
filippenko , a.v . , & sargent , w.l.w .
1993 , in iau symp .
159 , multi - wavelength continuum emission of agn , eds .
blecha & t.j .- l .
courvoisier ( dordrecht : reidel ) , 275 israel , f. p. 1992
, , 265 , 487 israel , f.p . ,
baas , f. , maloney , p.r .
1990 , , 237 , 17 jackson , j.m . ,
heyer , m.h . ,
paglione , t.a.d . , & bolatto , a.d .
1996 , , 456 , 91 kohno , k. , kawabe , r. , tosaki , t. , & okumura , s.k .
1996 , , in press lo , k.y . , ball , r. , masson , c.r . ,
phillips , t.g . , scott , s. , and woody , d.p .
1987 , , 317 , l63 mauersberger , r. , henkel , c. , wilson , t. l. , & harju , j. 1989 , , 226 , l5 nguyen - q - rieu , jackson , j. m. , henkel , c. , truong - bach , & mauersberger , r. 1992 , , 399 , 521 rand , r.j . &
kulkarni , s.r .
1990 , , 349 , l43 regan , m. , & vogel , s.n .
1995 , , 452 , 21 rix , h .- w . &
zaritsky , d. 1995 , , 447 , 82 rubin , v.c .
1994 , , 107 , 173 sage , l. j. , shore , s. n. , & solomon , p. m. 1990 , , 351 , 422 sandage , a. 1961 , the hubble atlas of galaxies ( washington : carnegie institute of washington ) sault , r.j . ,
teuben , p.j . , & wright , m.c.h .
1995 , in astronomical data analysis software and systems iv , eds .
shaw , h.e .
payne , & j.j.e .
hayes , a.s.p .
conference series 77 , 433 spergel , d.n . & blitz , l. 1992 , nature , 357 , 665 tacconi , l.j .
& young , j.s . , 1989 , , 71 , 455 turner , j.l .
& ho , p.t.p . 1983 , , 268 , l79 turner , j.l .
& ho , p.t.p .
1994 , , 421 , 122 welch , w.j . ,
1996 , , 108 , 93 weliachew , l. , casoli , f. , & combes , f. 1988 , , 199 , 29 yamauchi , s. , kawada , m. , koyama , k. , kunieda , h. , & corbet , r.h.d . 1990 , , 365 , 532 young , j.s . ,
tacconi , l.j . , & scoville , n.z .
1983 , , 269 , 136 ngc 3628 & 11@xmath6820@xmath6916.@xmath7027 & 13@xmath7135@xmath7239.@xmath730 & 9 & 847 ngc 4321 ( m100 ) & 12 22 54.80 & 15 49 23.0 & 17 & 1550 ngc 4527 & 12 34 08.80 & 02 39 10.3 & 20 & 1734 ngc 4569 & 12 36 50.02 & 13 09 53.1 & 17 & -235 ngc 4826 & 12 56 44.25 & 21 40 52.3 & 5 & 408 ngc 5194 ( m51 ) & 13 29 53.30 & 47 11 50.0 & 10 & 463 ngc 5236 ( m83 ) & 13 37 00.23 & -29 52 04.5 & 5 & 516 ngc 6946 & 20 34 51.91 & 60 09 11.9 & 5 & 52 ngc 3628 & 4.7 @xmath74 4.1 & 8.13 & 2.0 ngc 4321 ( m100 ) & 8.2 @xmath74 6.5 & 2.91 & 2.1 ngc 4527 & 13.5 @xmath74 10.7 & 1.07 & 3.6 ngc 4569 & 5.5 @xmath74 4.2 & 6.85 & 1.0 ngc 4826 & 10.5 @xmath74 8.0 & 1.87 & 3.2 ngc 5194 ( m51 ) & 7.9 @xmath74 6.5 & 3.03 & 1.5 ngc 5236 ( m83 ) & 12.5 @xmath74 4.1 & 3.03 & 3.6 ngc 6946 & 5.9 @xmath74 5.0 & 5.31 & 1.7 ngc 3628 & 49.0 @xmath3 17.5 & 1.5 @xmath3 0.2 & 1.01 @xmath3 0.36 ngc 4826 & 39.0 @xmath3 16.7 & 1.3 @xmath3 0.2 & 0.93 @xmath3 0.41 ngc 5194 ( m51 ) & 26.4 @xmath3 10.5 & 1.2 @xmath3 0.2 & 0.68 @xmath3 0.27 ngc 5236 ( m83 ) & 103 @xmath3 23 & 2.7 @xmath3 0.2 & 1.18 @xmath3 0.27 ngc 6946 & 39.2 @xmath3 5.5 & 1.5 @xmath3 0.2 & 0.81 @xmath3 0.17 | we present images of the hcn j = 1 - 0 emission from five nearby spiral galaxies made with the berkeley - illinois - maryland association interferometer .
the hcn observations comprise the first high - resolution ( @xmath0 @xmath1 5 10 ) survey of dense molecular gas from a sample of normal galaxies , rather than galaxies with prolific starburst or nuclear activities .
the images show compact structure , demonstrating that the dense gas emission is largely confined to the central kiloparsec of the sources . to within the uncertainties ,
70 - 100% of the single - dish flux is recovered for each source ; this implies that there is not a significant contribution to the hcn flux from low - level emission in the disks of the galaxies . in one of the galaxies , ngc 6946 , the ratio of hcn to co integrated intensities ranges from 0.050.2 within the extent of the hcn emission ( @xmath2 pc ) , with an average value of 0.11 @xmath3 0.01 over the whole region
; the range and average values of the ratios in ngc 6946 are very similar to what is observed in the central @xmath4 pc of the milky way .
a comparison with single - dish observations allows us to place an upper limit of 0.01 on the ratio of integrated intensities in the region @xmath5 pc in ngc 6946 . in ngc 6946 , ngc 1068 and the milky way , the ratio i@xmath6/i@xmath7 is 5 to 10 times higher in the bulge regions than in their disks ; this suggests that the physical conditions in their bulges and disks are very different .
furthermore , the presence of dense gas on size scales of @xmath1 500 pc in the centers of these nearby galaxies and the milky way suggests that the internal pressure is at least 10@xmath8 @xmath9 k in their centers ; this is some three orders of magnitude greater than the pressure in the local interstellar medium in the milky way , and it is two orders of magnitude greater than the pressure from the self - gravity of a solar neighborhood giant molecular cloud . in ngc 4826 and m51 , as in the milky way and ngc 1068 , there is a linear offset of @xmath1 100 pc between the dense gas distribution and the peak of the radio continuum emission .
we did not detect hcn towards three additional spiral galaxies . |
together with quintessence ( a scalar field with some potential minimally coupled to gravity ) @xcite and other physical and geometrical ( modified gravity ) models of dark energy ( see , e.g. , the reviews @xcite for exact definitions ) , the generalized chaplygin gas @xcite ( hereafter gcg ) is one widely studied model among those proposed to describe the observed accelerated expansion of the universe @xcite . in its generalized form , the chaplygin gas is a barotropic fluid with the following equation of state : @xmath6 where @xmath7 and @xmath3 are positive constants ( for the original chaplygin gas @xmath8 ) .
the generalized equation of state ( [ gcgeos ] ) has been introduced in @xcite and analysed in @xcite and in many other subsequent papers .
in contrast to many models describing dark energy alone , the gcg gives a unified description of dark matter and dark energy , enrolling itself in the class of so - called quartessence or unified dark matter ( udm ) cosmological models .
it allows to interpolate between a dust dominated phase of the evolution of the universe in the past and an accelerated one at recent time , see eq .
( [ gcgrhoevo ] ) below .
that is why the gcg has attracted much attention in cosmology .
the evolution of the energy density as a function of the cosmic scale factor @xmath9 of the friedmann lematre robertson walker ( flrw ) cosmological model is easily obtained from ( [ gcgeos ] ) and from the energy conservation equation : @xmath10 where @xmath11 is an integration constant usually chosen to be positive ( if @xmath12 , the weak energy condition is violated and phantom cosmology takes place , see for example @xcite ) . from ( [ gcgeos ] ) and ( [ gcgrhoevo ] ) ,
the parameter @xmath13 and the square sound velocity @xmath14 have the following expressions : @xmath15^{-1 } , \label{gcgw}\\ c_{\rm sch}^{2 } & \equiv & { \frac { { \rm d } \pp}{{\rm d } \rr } } = -\alpha w_{\rm ch}. \label{gcgcs}\end{aligned}\ ] ] note that for @xmath0 , @xmath16 may exceed the velocity of light ( set to unity in our notations ) in the course of the recent or future evolution of the universe , while it was non - relativistic at large redshifts , during the matter dominated stage , for all values of @xmath3 .
cosmological models based either on one or on more than one fluid are essentially different .
the reason for this is that the various fluid components interact indirectly via geometry and may also interact by direct energy exchange .
direct non - gravitational interactions are typically neglected but the indirect ones are always present and can give rise to totally different evolutions , especially at the perturbative level .
we will see how this works for the gcg together with baryons .
the pure gcg , namely the cosmological model based on the gcg alone , has passed many tests of standard cosmology .
quite recently , updated constraints for the gcg parameters have been published @xcite . however , the behaviour of the gcg under perturbations is still problematic @xcite . in @xcite
the study of the power spectrum of large scale structures seems to indicate that the best fit value of @xmath3 is very close to zero , rendering the gcg indistinguishable from @xmath2cdm [ the latter is indeed the limit of gcg for @xmath17 , as one can infer from ( [ gcgh ] ) below ] .
this feature has to be attributed to the gcg sound velocity which , during the cosmic evolution , grows from 0 to @xmath18 driving inhomogeneities to oscillate ( if @xmath19 ) or to blow up ( if @xmath20 ) .
this characteristic seems to be common to all udm models @xcite which , as a consequence , might appear to be ruled out .
however , as shown in @xcite , it appears that adding a baryon component to the gcg the problem disappears and the model is in agreement with observations .
but the debate still goes on and the authors of @xcite have corroborated their claim about ruling out gcg and , in general , all quartessence models through gravitational lensing measurements on the basis of the current value of the cosmological parameter @xmath21 ( the rms mass dispersion on a sphere of radius @xmath22 @xmath23 ) . still ,
the parameter @xmath21 need not necessarily be a confident discriminant among different cosmological models since it is clearly a quantity which is strongly affected by non linear effects , while all the calculations performed in @xcite are carried out in the linear regime .
this point was also raised much earlier in @xcite in a more general context . in the present paper
we investigate the issue of cosmological perturbations in gcg models using the gauge invariant formalism @xcite .
we consider both the gcg alone and in the presence of baryons , confirming the results of @xcite and @xcite .
in addition , the gcg seems to favour structure formation when its sound velocity is superluminal , in particular when @xmath4 .
the structure of the paper is the following : 1 .
in section [ giform ] we briefly outline bardeen gauge invariant formalism and derive the equations that we will numerically solve in the rest of the paper .
2 . in section [ gcgruled ]
we investigate structure formation in a gcg dominated universe and display the role of the sound velocity .
3 . in section [ gcgpowers ]
we calculate the theoretical power spectrum in a gcg dominated universe and compare it with the observed one .
4 . in section [ gcgandbar ]
we consider a two fluid model based on gcg in presence of baryons .
for this variant of the model we perform the same analysis of the previous two sections and compare the results .
in section [ concl ] we present our conclusions and discuss in more detail why a possible superluminal sound velocity of the gcg is not prohibited by causality arguments . 6 . for completeness ,
we add an appendix in which , using the technique developed in @xcite we write down the exact solution for perturbations of dust - like matter and generalized chaplygin gas in the newtonian approximation at the matter dominated stage .
the cosmological perturbations issue was first tackled and studied by lifshitz in 1946 in the synchronous gauge @xcite ( see also @xcite ) . in 1980 bardeen developed gauge invariant perturbation theory @xcite by constructing suitable combinations of metric and stress energy tensor perturbations which are invariant under a generic gauge transformation . the gauge invariant formalism has been used and reviewed by many authors . here
we follow the notation of @xcite and study only scalar perturbations .
the first order einstein equations have the following form : @xmath24\delta^{i}_{j } - \\ - \frac{1}{2}\gamma^{ik}\left(\phi - \psi\right)_{|kj } = a^{2}\delta p\delta^{i}_{j } - \sigma^{|i}_{|j } , \end{array } \right.\ ] ] where @xmath25 is the scale factor as a function of the conformal time @xmath26 .
its present value is normalized to unity ; @xmath27 , where the prime denotes derivation with respect to the conformal time ; @xmath28 and @xmath29 are the bardeen gauge
invariant potentials ; @xmath30 and @xmath31 are the gauge invariant expressions of the perturbations of the energy density and pressure ; @xmath32 is the gauge invariant expression of the scalar potential of the velocity field ; @xmath33 represents the shear . finally , @xmath34 ( @xmath35 ) is the spatial part of the flrw metric and @xmath36 is its curvature parameter which corresponds respectively to flat , close and open geometry .
the vertical bar denotes covariant derivation with respect to @xmath34 .
units are chosen such that @xmath37 .
see @xcite for a detailed derivation of system ( [ muksys ] ) .
we introduce two assumptions which simplify ( [ muksys ] ) : * we assume @xmath38 ; * we neglect shear perturbations , namely we assume @xmath39 .
this implies @xmath40 . for a cosmological model based on @xmath41 non interacting fluids we write for the background quantities : @xmath42
while for the perturbations in the linear regime : @xmath43 the mixed time
space component of the perturbed energy
momentum tensor will look like @xmath44 , since each fluid will contribute with its own velocity .
adiabatic perturbations of a barotropic fluid are characterized by the following equation of state : @xmath45 where @xmath46 is the sound velocity at constant entropy for the generic component @xmath47 . in the cosmological perturbation theory it is customary to deal with the density contrast : @xmath48 where @xmath49 is the background energy density of the component @xmath47 .
since we work in the linear regime it is useful to take the spatial fourier transform of ( [ muksys ] ) which allows us to treat each mode independently . with the above assumptions and trading the conformal time for the scale factor @xmath50 , system ( [ muksys ] ) for a generic multi
fluid model is rewritten as follows : @xmath51 where the dot denotes derivation with respect to the scale factor @xmath50 ( for the sake of simplicity we adopt the same notation for the fourier transformed quantities as for the original ones ) .
if @xmath52 , the system is determined and can be solved by eliminating @xmath53 and by extracting the following second order equation for @xmath54 : @xmath55 once solved , we can use the solution of ( [ eqphi ] ) in the first equation of ( [ muksys2 ] ) to find the solution for @xmath53 . instead , if @xmath56 we need @xmath57 additional equations to make ( [ muksys2 ] ) determined .
these equations should take into account the possible interactions and the energy exchanges among the different fluids .
since we have assumed that they are not mutually interacting directly , the required additional equations to be added can be simply chosen as the energy conservation equations , @xmath58 , given separately for each further component .
namely the continuity equation @xmath59 : @xmath60 and the euler equation @xmath61 : @xmath62\dot { } + \frac{3}{a}\left(\rho_{i } + p_{i}\right)v_{i } - \frac{c_{{\rm s}i}^{2}\rho_{i}}{{\mathcal{h}}}\delta_{i } - \frac{\left(\rho_{i } + p_{i}\right)}{{\mathcal{h}}}\phi = 0,\ ] ] where @xmath63 .
both ( [ conteq ] ) and ( [ euleq ] ) are fourier transformed , that is why there is no @xmath64 subscript in ( [ euleq ] ) .
it is possible to write the combination of ( [ muksys2 ] ) , ( [ conteq ] ) and ( [ euleq ] ) as a system of @xmath41 second order equations , one for each @xmath65 .
one needs to derive ( [ conteq ] ) , then to express @xmath66 , @xmath67 , @xmath68 and @xmath69 in terms of the other @xmath65 s through ( [ muksys2 ] ) , ( [ euleq ] ) and again ( [ conteq ] ) . in the case
@xmath70 the resulting system has the following form : @xmath71 where the coefficients have the following expressions : @xmath72}{k^{4 } + 3\left(3{\mathcal{h}}^{2 } + k^{2}\right)\left({\mathcal{h}}\dot{{\mathcal{h}}}a - { \mathcal{h}}^{2}\right)},\\ \fl d_{\rm 1 } = -3\left(1 + w_{\rm 1}\right)c_{\rm s2}^{2}\rho_{\rm 2}+ \nonumber \\
\fl 3{\mathcal{h}}^{2}\left(2 + 3c_{\rm s1}^{2}\right)\rho_{\rm 2}\left(1 + w_{\rm 1}\right)\frac{k^{2 } - 3\left(c_{\rm s2}^{2 } - w_{\rm 2}\right)\left(3{\mathcal{h}}^{2 } + k^{2}\right)}{k^{4 } + 3\left(3{\mathcal{h}}^{2 } + k^{2}\right)\left({\mathcal{h}}\dot{{\mathcal{h}}}a - { \mathcal{h}}^{2}\right ) } \nonumber\\ \fl -\left(k^{2 } + 3{\mathcal{h}}^{2 } - 6{\mathcal{h}}\dot{{\mathcal{h}}}a\right)\rho_{\rm 2}\left(1 + w_{\rm 1}\right)\frac{k^{2 } + 3\left[{\mathcal{h}}\dot{{\mathcal{h}}}a - { \mathcal{h}}^{2 } - 3{\mathcal{h}}^{2}\left(c_{\rm s2}^{2 } - w_{\rm 2}\right)\right]}{k^{4 } + 3\left(3{\mathcal{h}}^{2 } + k^{2}\right)\left({\mathcal{h}}\dot{{\mathcal{h}}}a - { \mathcal{h}}^{2}\right)}.\end{aligned}\ ] ] the coefficients @xmath73 , @xmath74 , @xmath75 and @xmath76 have the same form , but with the interchange @xmath77 in the subscripts .
as shown in @xcite , at the matter dominated regime , i.e. when @xmath78 , @xmath79 and @xmath80 , the above coefficients are far simpler and system ( [ gensys ] ) can be rewritten in the following quasi
newtonian form : @xmath81 this approximation is sufficient to study the formation of all gravitationally bound objects ( galaxies , in particular ) which become nonlinear for @xmath82 . as pointed out in @xcite , it is possible to recover the system ( [ gensyssimplified ] ) by considering the limit @xmath83 , too , but it is not a necessary constraint in the matter dominated regime when @xmath84 and @xmath85 are small .
it is interesting that the system ( [ gensyssimplified ] ) can be exactly solved for the two fluid model considered in sec .
[ gcgandbar ] .
the solution is exhibited in the appendix .
in the gcg dominated universe the hubble parameter scales as follows : @xmath86 where @xmath87 and @xmath88 is the hubble constant . in a typical analysis of the gcg model
it is customary to take @xmath89 since the square sound velocity tends to @xmath3 for @xmath90 , as can be seen from ( [ gcgcs ] ) .
for @xmath8 the sound velocity will tend to unity in the far future , at @xmath91 .
the new regime that we investigate in the present paper is @xmath0 , thereby making gcg superluminal after a sufficiently long but still finite time . from ( [ gcgcs ] )
it is straightforward to calculate at which redshift the transition to the superluminal gcg would occur . for generic values of @xmath92 and @xmath3 : @xmath93^{\frac{1}{3\left(\alpha + 1\right ) } } - 1.\ ] ] given @xmath94 and @xmath92 , it is plausible to expect that some , hopefully observable , cosmological effect would take place at that redshift .
it is crucial to point out that @xmath3 and @xmath92 are not independent , but linked by ( [ gcgh ] ) , once we have some constraints about the background expansion . in the present paper we consider @xmath95 , i.e. the redshift at which the transition to the accelerated phase of the expansion takes place . we derive its expression from ( [ gcgh ] ) : @xmath96^{\frac{1}{3\left(\alpha + 1\right ) } } - 1.\ ] ]
it follows from the most recent sn ia observations combined with cmb and baryon acoustic oscillations ( bao ) data that @xmath97 with a rather large uncertainty ( at least @xmath98 or more at the @xmath99 confidence level ) , see e.g. fig . 7 in @xcite and
also @xcite . for comparison ,
if only sn ia are used , then the gold+hst dataset @xcite leads to @xmath100 , see fig . 2 in @xcite and also @xcite ) while other datasets ( snls and essence ) alone produce values of @xmath101 , too ( see @xcite ) .
however , these numbers strongly depend on the present value of the non - relativistic matter density @xmath102 which , in turn , has to be expressed through @xmath92 in the gcg case ( the values given above correspond to @xmath103 )
. direct calculation of @xmath104 for udm models @xcite shifts this quantity to even larger redshifts : @xmath105 . actually ,
in the latter case , the authors consider only silence quartessence ( namely the effective sound velocity is assumed to be zero ) and the parameter @xmath3 is constrained to be @xmath106 .
then from ( [ acctransz ] ) we can write down an explicit relation between @xmath92 and @xmath3 : @xmath107 note that the uncertainty @xmath108 of the transition redshift propagates on @xmath92 through the following formula : @xmath109^{2}(1 + z_{\rm tr})^{3\alpha+4}}\sigma_{z_{tr}},\ ] ] therefore the larger is @xmath3 , the smaller is the bias on @xmath92 .
a widely used cosmological parameter stemming from cmb observation is @xmath110 , the comoving distance to the last scattering surface scaled to @xmath102 , namely : @xmath111 where @xmath112 is the last scattering surface redshift . in udm models
, the quantity @xmath102 should be taken from the asymptotic behaviour of @xmath113 for @xmath114 at the matter dominated stage , so @xmath115 ( or @xmath116 if baryons are taken into account , too ) . at the @xmath117 confidence level , @xmath118 ( see for example @xcite ) .
an interesting limiting case is the super
duperluminal one : @xmath119 .
then @xmath120 for @xmath121 and @xmath122 for larger @xmath123 . in this case
, one can obtain an analytic expression for @xmath110 : @xmath124 then the above mentioned observational window for @xmath110 is reached for @xmath125 . combining ( [ acctransz ] ) with ( [ transz ] )
we find that @xmath126 if @xmath127 , then @xmath128 .
for @xmath129 , @xmath130 but @xmath131 approaches @xmath95 once more for @xmath119 . for a fixed @xmath95
, the maximal value of @xmath131 is reached for @xmath132 when @xmath133 .
thus , for @xmath134 , the transition from subluminal to superluminal gcg occurs approximately at the same time as the transition to the accelerated phase of expansion of the universe .
we now compare the evolution of perturbations in the gcg dominated universe for different values of @xmath3 .
we choose an integration range which starts at the decoupling era , namely @xmath135 ( @xmath136 ) , and ends up when the first structures , namely protogalaxies , were formed ( @xmath137 , or @xmath138 ) , so that the linear approximation holds true .
we numerically solve ( [ eqphi ] ) choosing @xmath139 $ ] as normalized initial conditions for @xmath140 $ ] . in our calculations we use ( [ acctransza ] ) to properly choose @xmath92 as a function of @xmath3 .
we show the results using @xmath141 and @xmath142 . for different values of @xmath3 and for @xmath143 h mpc@xmath144 .
the transition redshift is @xmath145,title="fig:",width=491 ] + , with @xmath146,title="fig:",width=491 ] + the plots in figures [ fig1 ] and [ fig2 ] display the evolution of the density contrast in the gcg , which we call @xmath147 , for different choices of the parameter @xmath3 and for @xmath143 h mpc@xmath144 , which corresponds to a scale of order @xmath148 @xmath149 @xmath150 , which is typical of a protogalaxy .
the case @xmath151 corresponds to the @xmath2cdm model .
indeed , here we are first investigating if formation of gravitationally bound objects including galaxies takes place at all .
that is why we have chosen so small a scale and we have stopped at @xmath152 .
the gcg udm model , like all cdm like ones , belongs to the class of bottom up , or hierarchical clustering , scenarios of structure formation where small scale structures are formed first .
the question of large scale structure in this model at scales which remain linear for @xmath153 , of course , requires integration up to @xmath154 .
it is considered in the next section .
part of the results found in @xcite are already confirmed by this analysis , without a deeper study involving the power spectrum .
the development of oscillations takes place too early for @xmath155 , thus preventing structure formation .
however , two further interesting features can be extracted from figures [ fig1 ] and [ fig2 ] .
first , it seems that there exists a critical value , namely @xmath156 , for which the deviation of the gcg from the @xmath151 linear growth is maximal .
second , the larger the value of @xmath3 ( in the range @xmath157 ) the smaller the deviation from the @xmath151 behaviour . indeed ,
when @xmath0 , i.e. in the superluminal regime , the oscillations are absent and the growth seems even to be enhanced . both the above mentioned features can be explained by the behaviour of the gcg sound velocity as a function of @xmath3 , which is displayed in figure [ fig3 ] for @xmath141 ( the plots for @xmath142 are very similar ) . ) plotted as a function of @xmath50 for different values of @xmath3 .
@xmath92 is given by ( [ acctransza ] ) with @xmath141.,title="fig:",width=491 ] + for @xmath158 the sound velocity becomes non negligible much earlier than at other values of @xmath3 .
the range @xmath159 appears thus to be ruled out since structure formation is prevented .
however , to understand what happens out of this range it is necessary to study the clustering properties of matter at the present time , namely the power spectrum .
this will be done in the following section .
in the previous section we have shown that in the gcg model galaxy formation is possible only when @xmath3 is very small or @xmath134 . in this section
we try and constrain these bounds further by studying the clustering properties of gcg perturbations , namely the power spectrum @xmath160 , and comparing it with observation .
we exploit the sdss ( sloan digital sky survey ) data as analysed in @xcite .
the sample considered by the authors consists of 205,443 galaxies observed before july 2002 .
the data consist of ra ( right ascension ) , dec ( declination ) and redshift @xmath123 for each galaxy .
the redshift is converted to comoving distance by using a flat cosmological model with a cosmological constant @xmath161 ( it can be shown that our results are robust to this assumption ) . the power spectrum is computed at present time , namely at @xmath162 .
the data processed in @xcite are displayed in .
@lllllll * k * & * low k * & * high k * & * @xmath160 * & * bias * & * @xmath163 * + 0.01819 & 0.01502 & 0.02440 & 33254.63896 & 1.16745 & 24572.79357 + 0.02405 & 0.01984 & 0.03136 & 38360.58264 & 1.16674 & 13320.48227 + 0.02776 & 0.02300 & 0.03571 & 24143.07699 & 1.16613 & 10047.21709 + 0.03201 & 0.02663 & 0.04048 & 19709.29306 & 1.16532 & 7413.82995 + 0.03691 & 0.03094 & 0.04601 & 12595.77528 & 1.16428 & 5485.84481 + 0.04257 & 0.03591 & 0.05215 & 13558.60084 & 1.16294 & 4077.68223 + 0.04912 & 0.04162 & 0.05983 & 18311.08054 & 1.16124 & 2974.15048 + 0.05670 & 0.04824 & 0.06873 & 12080.73574 & 1.15910 & 2140.39514 + 0.06527 & 0.05561 & 0.07867 & 9217.46084 & 1.15647 & 1580.01728 + 0.07529 & 0.06455 & 0.08999 & 9750.49986 & 1.15317 & 1127.95803 + 0.08698 & 0.07507 & 0.10351 & 9529.63935 & 1.14906 & 818.49019 + 0.10037 & 0.08670 & 0.11898 & 6384.82545 & 1.14409 & 601.70752 + 0.11581 & 0.09985 & 0.13679 & 5294.92081 & 1.13813 & 446.65015 + 0.13360 & 0.11478 & 0.15748 & 4629.83669 & 1.13109 & 335.07950 + 0.15412 & 0.13181 & 0.18139 & 3573.92724 & 1.12290 & 254.13547 + 0.17774 & 0.15114 & 0.20891 & 3393.82814 & 1.11358 & 195.23431 + 0.20489 & 0.17288 & 0.24062 & 2298.10000 & 1.10320 & 152.73819 + 0.23592 & 0.19690 & 0.27653 & 1597.14976 & 1.09193 & 123.50981 + 0.27062 & 0.22219 & 0.31395 & 1105.42903 & 1.08019 &
107.20069 + 0.30618 & 0.23123 & 0.34782 & 1012.72489 & 1.06912 & 110.33632 + we compute the gcg power spectrum for different values of @xmath3 , by first solving ( [ eqphi ] ) , thus finding the transfer function and then applying it to the following prior : @xmath164^{2}\left[1 + 3.89q + \left(16.1q\right)^{2 } + \left(5.46q\right)^{3 } + \left(6.71q\right)^{4}\right]^{-1/2},\ ] ] where @xmath41 is a normalization constant and @xmath165 where @xmath166 is the cold dark matter ( cdm ) energy density fraction evaluated today , and @xmath167 is the hubble constant in units of @xmath168 @xmath169 @xmath170 @xmath171 . in our calculations
we shall set @xmath172 .
since the gcg has the same asymptotic behaviour as the @xmath2cdm model in the matter dominated stage , then the cold dark matter fraction is given by @xmath173 the prior ( [ bbks ] ) is obtained by applying the bbks transfer function @xcite to the scale invariant harrison
zeldovich spectrum .
the normalization constant @xmath41 is computed through a best fit of the data for each @xmath3 . in our calculation , @xmath92 is constrained by ( [ acctransza ] ) using @xmath141 and @xmath142 .
compared with the measured one .
@xmath92 is given by ( [ acctransza ] ) with @xmath141 .
note that the plots for @xmath151 and @xmath174 are superposed .
we have extracted the envelopes of the oscillations of the density contrasts for @xmath127 and @xmath175 and have used them for the calculations . for the other cases this procedure was unnecessary since
no oscillations were present.,title="fig:",width=491 ] + , with @xmath142.,title="fig:",width=491 ] + in figures [ fig4 ] and [ fig5 ] the power spectra have been computed by considering the envelope of the oscillations of the density contrast @xmath176 and by using it in the subsequent calculations .
this procedure serves to clearly compare the different slopes of the calculated power spectra with respect to the observed one .
moreover , it has been carried out only for @xmath127 and @xmath175 .
for the other ones it was unnecessary as no oscillations were present
. an interesting feature that can be drawn from figures [ fig4 ] and [ fig5 ] is that the larger is @xmath3 the more the power spectrum tends to a limiting behaviour which is systematically below that of the @xmath2cdm one .
this happens because the suppression of the gcg transfer function relative to that of the @xmath2cdm model behaves approximately as @xmath177 for large values of @xmath3 after averaging over oscillations ( actually , the exponent is slightly larger than @xmath178 because of the accelerated expansion at recent times ) . compared with the measured one .
@xmath92 is given by ( [ acctransza ] ) with @xmath141 .
note that the plots for @xmath151 and @xmath174 are superposed.,title="fig:",width=491 ] + , with @xmath142.,title="fig:",width=491 ] + figures [ fig4bis ] and [ fig5bis ] show what happens if the oscillations in the density contrasts are taken into account .
the power spectra for @xmath127 and @xmath175 are highly irregular and their slopes are hardly recognizable . in conclusion , the pure gcg does not seem to work from a perturbative viewpoint , even when it is superluminal .
however , a reasonable possibility is to allow the gcg to have non
adiabatic perturbations , which is a natural assumption since it is not a pressureless fluid .
an attempt in this direction has already been performed in @xcite and @xcite ( here also a baryon component is considered ) .
however , the results are based on the _ ad hoc _ assumption of silent perturbations ( namely @xmath179 ) , which seems questionable .
in this section we consider a two fluid model including baryons together with gcg . the friedmann equation governing this model
has the following form : @xmath180^{1/1 + \alpha}\right\}a^{2},\ ] ] where subscript b refers to the baryonic component and subscript @xmath181 indicates that the corresponding quantity is evaluated at the present epoch ( @xmath182 ) . system ( [ gensys ] ) is solved numerically in the same integration range of the previous sections and setting @xmath183 . before proceeding
it is necessary to find a relation between @xmath92 and @xmath3 which ensures the preservation of the correct background expansion , e.g. the acceleration redshift @xmath184 . from ( [ chbhomega ] )
it is possible to work out the following expression : @xmath185 - 1/2}{\left\{\bar{a}\left[(1+z_{\rm tr})^{-3(\alpha+1)}-1\right ] + 1\right\}^{\alpha/\alpha + 1 } } = \frac{\omega_{\rm b0}}{2\left(1-\omega_{\rm b0}\right)},\ ] ] which can not be inverted to give @xmath92 as a function of @xmath3 .
however , since @xmath186 , is small we can use again ( [ acctransza ] ) to that purpose . as in the previous section ,
we choose ( [ bbks ] ) as a prior for the power spectrum , but since this time we have the presence of baryons , @xmath166 has to be chosen differently .
we exploit then sugiyama s shape correction @xcite : @xmath187\times \nonumber\\ \exp\left(-\omega_{b0 } - \frac{\sqrt{2h}\omega_{b0}}{\omega_{b0 } + \left(1 - \omega_{b0}\right)\left(1 - \bar{a}\right)^{\frac{1}{1 + \alpha}}}\right),\end{aligned}\ ] ] which properly takes into account the presence of baryons . in the appendix
we show an exact solution , in the matter dominated regime , for the evolution of perturbations in the model of the present section . with @xmath145.,title="fig:",width=491 ]
+ with @xmath188.,title="fig:",width=491 ] + in figures [ fig6 ] and [ fig7 ] we show the baryonic power spectra computed for different values of @xmath3 compared to the observed one .
we have chosen as normalized initial conditions the following values : @xmath189 = \left[1 , 1 , 1 , 1\right]$ ] .
we have chosen the same initial conditions for all the quantities essentially because after the equivalence era the gcg behaved as dust and therefore its density contrast grew linearly ( as a function of the scale factor ) . unlike what happened in the previous section , this time there were no oscillations in @xmath190 , since the baryonic density contrast is unaffected by the instabilities of the gcg one .
in fact , the results are different from those inferred from figures [ fig4 ] and [ fig5 ] .
the agreement with the observed power spectrum is better .
it is particularly good again for very small @xmath3 but also in the range @xmath0
. therefore the gcg seems to work better either when it is practically indistinguishable from @xmath2cdm or when its sound velocity is superluminal .
the good agreement of the theoretical power spectrum with the observed one can be explained realizing that gcg perturbations are indistinguishable from those of baryons at the beginning of the matter dominated stage .
therefore they really act as dark matter and compel baryons perturbations to grow and form structures . on the other hand , when the gcg sound velocity increases ,
the growth of the baryons inhomogeneities is damped , but does not oscillate .
this fact can be observed in figures [ fig8 ] and [ fig9 ] .
moreover , structures which have already formed are not influenced by the gcg sound velocity , so neither is the large scale power spectrum .
this was already pointed out in @xcite . for different values of @xmath3 and for @xmath143 @xmath167 @xmath171 .
note that the plot for @xmath151 and @xmath127 are superposed .
@xmath145.,title="fig:",width=491 ] + , with @xmath188.,title="fig:",width=491 ] + in figures [ fig8 ] and [ fig9 ] we display the evolution of @xmath191 for different values of @xmath3 choosing again @xmath192 = \left[1 , 1 , 1 , 1\right]$ ] as normalized initial conditions . as in section [ gcgruled ]
we have chosen @xmath143 @xmath167 @xmath171 .
note that the plots corresponding to @xmath151 and @xmath127 are superposed .
we infer from figures [ fig8 ] and [ fig9 ] that the gcg sound velocity has a damping effect on the growth of baryon inhomogeneities .
these are not compelled to oscillate , but structure formation is delayed , if not prevented . in order to render structure formation as similar as possible to that of @xmath2cdm , we need either very small values of @xmath3 or @xmath129 . the latter range for @xmath3 just leads to @xmath130 as was shown in section [ gcgruled ] .
from our analysis it emerges that : * the generalized chaplygin gas cosmological model , with no additional fluid components , is compatible with structure formation and large scale structure clustering properties ( encoded in the power spectrum ) only for @xmath3 sufficiently small ( @xmath1 ) , in which case it is practically indistinguishable from the @xmath2cdm model . *
adding to the generalized chaplygin gas model a baryon component we find that the growth of the density contrast and the large scale structure clustering properties are compatible with observations for all values of @xmath3 .
however very small values of @xmath3 and the case @xmath4 are favoured .
note , however , that the integrated sachs - wolfe ( isw ) effect ( not considered in this paper ) leads to rather large enhancement of low cmb multipoles @xmath193 in udm models , in contrast to a modest increase of @xmath193 in the @xmath2cdm model @xcite , that results in the exclusion of the gcg with @xmath194 @xcite ( see also the recent paper @xcite for an analytical treatment of the isw effect for udm models ) .
it may be interesting to investigate if it remains so large for all @xmath195 .
now , since most of the new results of this paper refer to the superluminal case @xmath0 , it is natural to consider the question if this range is physically admissible in more detail .
the fact that the group velocity @xmath196 in media may exceed the light velocity in vacuum ( unity in our notations ) is not , in itself , unphysical .
not only it is not prohibited from the theoretical point of view , but this phenomenon has been actually observed in laboratories in case of light propagation through dispersive media . as was proved long ago
( see e.g. @xcite ) , what is really necessary in order not to violate causality is that the signal ( or , the wavefront ) velocity should not exceed @xmath197 .
then , what is the signal velocity for the gcg ? it does not appear possible to answer this question at the hydrodynamic level since the gcg phenomenological equation of state in the normal ( non - phantom ) case requires @xmath198 , so that the limit @xmath199 corresponding to a wavefront expanding into vacuum can not be studied .
therefore , to determine the value of the signal velocity for the gcg , one has to use some underlying microscopic field theoretical model from which the equation of state ( [ gcgeos ] ) arises in macroscopic hydrodynamics .
now we are able to consider the limit @xmath203 corresponding to a wavefront expanding into vacuum with no @xmath5-essence in front of it . to prove that the signal velocity is equal to unity in this case , it is sufficient to show that the wave equation corresponding to ( [ tach ] ) , namely @xmath204 has plane - wave solutions of the type @xmath205 in flat space - time ( since the signal velocity is also the infinite momentum limit of the phase velocity of a wave )
. it is straightforward to check that the sufficient condition for this to hold is that both @xmath206 and @xmath207 are finite at @xmath208 ( a similar remark is made in the recent paper @xcite , see also @xcite ) .
in addition , @xmath209 should be non - negative in order for ghosts not to be present . in the superluminal case @xmath0 , the lagrangian density ( [ tach ] )
does not satisfy this condition since @xmath206 and @xmath207 diverge at @xmath203 .
however , without going into a more detailed analysis of the singularity structure at a null hyper
surface in this case , we propose a simple _ sufficient _ way to cure the causality problem for the gcg .
namely , let us assume that the expression ( [ tach ] ) changes to @xmath210 for @xmath211 where @xmath212 and that it smoothly matches ( [ tach ] ) at @xmath213 .
then it is clear that the signal velocity is exactly unity .
the lagrangian density ( [ change ] ) describes a massless scalar field minimally coupled to gravity and a cosmological constant .
the correspondingly corrected macroscopic gcg remains the same in the observable region @xmath214 , or @xmath215 ( note that @xmath216 is of the order of the present day critical energy density in the universe ) but will become a mixture of a cosmological constant and the extremely stiff ideal fluid ( @xmath217 ) in the far future when @xmath218 will drop to @xmath219 .
with the same accuracy , one can say that the corrected gcg becomes the usual chaplygin gas when its density is very close to @xmath216 , much closer than the one which occurs presently .
thus , by slightly changing the gcg equation of state in the unobservable region , the causality problem is cured . as a result ,
the present gcg superluminal sound velocity is a _ transient _ phenomenon which disappears at @xmath220 .
finally , note that in the limit @xmath119 , the ggg realizes the parker - raval scenario @xcite of a sudden transition from the matter dominated stage with @xmath221 to the de sitter expansion with a constant curvature ( modulo a small term @xmath222 ) .
the original idea of parker and raval to produce such a scenario using non - perturbative vacuum polarization of a light minimally coupled scalar field is questionable , both regarding derivation of the effective action and stability of the resulting specific kind of the @xmath223 theory of gravity ( here @xmath110 is the ricci scalar ) .
but the same scenario based on the limiting case of the gcg does not have these problems .
this shows once more that the superluminal gcg deserves further , more detailed study .
the work of ayk and aas was partially supported by the research programme `` astronomy '' of the russian academy of sciences , by the rfbr grant 05 - 02 - 17450 and by the grant lss-1157.2006.2 .
aas also thanks prof .
misao sasaki and the yukawa institute for theoretical physics , kyoto university , for hospitality during the middle part of this project .
at the matter dominated regime , the system ( [ gensyssimplified ] ) describing gcg and baryons takes the following form : @xmath224 where @xmath225 is the gcg square sound velocity given by ( [ gcgw ] ) and ( [ gcgcs ] ) .
this system can be exactly solved using the technique developed in @xcite for cosmological perturbations of two fluid models in the newtonian approximation .
let us introduce the following variables : @xmath226 where @xmath3 is the gcg parameter .
it is possible to extract from system ( [ solstarsys ] ) a fourth order equation for @xmath227 : @xmath228\delta_{\rm ch } = 0,\end{aligned}\ ] ] where @xmath229 . as found in @xcite , the general solution of ( [ solstareq ] ) can be represented in terms of the meijer g functions ( the generalized hypergeometric functions ) : @xmath230 where : @xmath231 the @xmath232 ( @xmath233 ) are integration constants and the meijer g functions are defined as follows @xcite : @xmath234 where @xmath235 is the gamma function and the integration contour @xmath236 circuits all the poles of @xmath237 anti clockwise and all the poles of @xmath238 clockwise ( all these poles lie along the real axis ) .
the solution for @xmath191 has the same functional form as ( [ deltachsolovstaro ] ) , with : @xmath239 where @xmath240 since we assume a spatially flat universe .
36 peebles p j e and ratra b , _ cosmology with a time variable cosmological constant _ , 1988 _ astroph .
j. ( lett . ) _ * 325 * l17 wetterich c , _ cosmology and the fate of dilatation symmetry _
, 1988 _ nucl .
b _ * 302 * 668 caldwell r r , dave r and steinhardt p j , _ cosmological imprint of an energy component with general equation of state _ , 1998 _ phys .
lett . _ * 80 * 1582 starobinsky a a , _ how to determine an effective potential for a variable cosmological term _ , 1998 _ jetp lett _ * 68 * 757 sahni v and starobinsky a a , _ the case for a positive cosmological lambda term _ , 2000
d _ * 9 * 373 sahni v and starobinsky a a , _ reconstructing dark energy _ , 2006 _ int
d _ * 15 * 2105 kamenshchik a y , moschella u and pasquier v , _ an alternative to quintessence _
, 2001 _ phys . lett .
b _ * 511 * 265 riess a g _ et al . _ , _ cosmological results from high - z supernovae _ ,
2003 _ apj _ * 594 * 1 perlmutter s
_ , _ measurements of omega and lambda from 42 high - redshift supernovae _ , 1999 _ apj _ * 517 * 565 bilic n , tupper g b and viollier r d , _ unification of dark matter and dark energy : the inhomogeneous chaplygin gas _ , 2002
b _ * 535 * 17 bento m c , bertolami o and sen a a , _ generalized chaplygin gas , accelerated expansion , and dark - energy - matter unification _ , 2002 _ phys .
d _ * 66 * 043507 gorini v , kamenshchik a y and moschella u , _ can the chaplygin gas be a plausible model for dark energy ? _ , 2003 _ phys .
d _ * 67 * 063509 zhang h and zhu z , _ interacting chaplygin gas _
, 2006 _ phys .
d _ * 73 * 043518 bouhmadi
lopez m and jimenez madrid j a , _ escaping the big rip ? _ , 2005 _ jcap _ * 0505 * 005 wu p and yu h , _ generalized chaplygin gas model : constraints from hubble parameter versus redshift data _ , 2007 _ phys . lett .
b _ * 644 * 16 wu p and yu h , _ constraints on the unified dark energy dark matter model from latest observational data _ , 2007 _ jcap _ * 0703 * 015 davis t _ et al . _ , _ scrutinizing exotic cosmological models using essence supernova data combined with other cosmological probes _ , 2007 _ astroph
. j. _ * 666 * 716 bean r and dor o , _ are chaplygin gas serious contenders for the dark energy ? _ , 2003 _ phys .
d _ * 68 * 023515 sandvik h b , tegmark m , zaldarriaga m and waga i , _ the end of unified dark matter ? _ , 2004 _ phys .
d _ * 69 * 123524 beca l
m g , avelino p p , de carvalho j p m and martins c j a p , _ role of baryons in unified dark matter models _ , 2003 _ phys .
d _ * 67 * 101301 peacock j a 1999 _ cosmological physics _ ( cambridge university press ) bardeen j m , _ gauge invariant cosmological perturbation _ , 1980 _ phys .
d _ * 22 * 1882 soloveva l v and starobinsky a a , _ gravitational instability of a two component medium in an expanding universe _ , 1985 _ sov .
* 29 * 367 lifshitz e m , _ on the gravitational stability of the expanding universe _ , 1946 _ j. phys .
( ussr ) _ * 10 * 116 lifshitz e m and khalatnikov i m , _ investigations in relativistic cosmology _ , 1963 _ adv . phys .
_ * 12 * 185 mukhanov v f , feldman h a and brandenberger r h , _ theory of cosmological perturbation _ , 1992 _ phys .
rep . _ * 215 * 205 alam u , sahni v and starobinsky a a , _ exploring the properties of dark energy using type ia supernova and other datasets _
, 2007 _ jcap _ * 0702 * 011 melchiorri a , pagano l and randolfi s , _ when did cosmic acceleration start ? _ , 2007 _ phys .
d _ * 76 * 041301(r ) daly r a _ et al .
_ , _ improved constraints on the acceleration history of the universe and the properties of dark energy _ , 2007 _ preprint _ arxiv:0710.5345 [ astro - ph ] riess a g _ et al .
_ , _ new hubble space telescope discoveries of type ia supernovae at @xmath241 : narrowing constraints on the early behaviour of dark energy _ , 2007 _ astroph .
j _ * 659 * 98 daly r a and djorgovsky s g , _ the acceleration history of the universe and properties of dark energy _ , 2007 _ preprint _ arxiv:0710.5690 [ astro - ph ] wang y and mukherjee p , _ observational constraints on dark energy and cosmic curvature _ , 2007 _ phys
d _ * 76 * 103533 tegmark m _
_ , _ the 3d power spectrum of galaxies from the sdss _ , 2002 _ astrophys
. j. _ * 606 * 702 bardeen j m , bond j r , kaiser n and szalay a s _ the statistics of peaks of gaussian random fields _ , 1986 _ astrophys
. j. _ * 304 * 15 reis r r r , waga i , calvao m o and joras s e , _ entropy perturbations in quartessence chaplygin models _ , 2003 _ phys .
061302 amendola l , waga i and finelli f , _ observational constraint on silent quartessence _ , 2005 _ jcap _ * 11 * 009 sugiyama n , _ cosmic background anisotropies in cdm cosmology _ , 1995 _ astrophys . j. suppl . _ * 100 * 281 brillouin l 1960 _ wave propagation and group velocity _
( academic press , new york ) gorini v , kamenshchik a , moschella u , pasquier v and starobinsky a , _ stability properties of some perfect fluid cosmological models _ , 2005 _ phys . rev .
* 72 * 103518 babichev e , mukhanov v and vikman a , _ k - essence , superluminal propagation , causality and emergent geometry _ , 2007 _ preprint _ arxiv:0708.0561 [ hep - th ] kang j u , vanchurin v and winitzki s , _ attractor scenarios and superluminal signals in k - essence cosmology _ , 2007 _ phys .
d _ * 76 * 083511 parker l and raval a , _ non - perturbative effects of vacuum energy on the recent expansion of the universe _
, 1999 _ phys .
d _ * 60 * 063512 erdelyi a , magnus w , oberhettinger f and tricomi f g 1981 _ higher transcendental functions _ vol 1 ( new york : krieger ) kofman l a and starobinsky a a , _ cosmological constant and large - scale anisotropy of the relic background radiation _ , 1985 _ sov
lett . _ * 11 * , 271 carturan d and finelli f , _ cosmological effects of a class of fluid dark energy models _ , 2003 _ phys .
d _ * 68 * 103501 amendola l , finelli f , burigana c and carturan d , _ wmap and the generalized chaplygin gas _ , 2003 _ jcap _ * 68 * 005 bertacca d and bartolo n , _ isw effect in the unified dark matter scalar field cosmologies : an analytical approach _ , 2007 _ jcap _ * 0711 * , 026 | we exploit the gauge invariant formalism to analyse the perturbative behaviour of two cosmological models based on the generalized chaplygin gas describing both dark matter and dark energy in the present universe . in the first model we consider the generalized chaplygin gas alone , while in the second one we add a baryon component to it .
we extend our analysis also into the parameter range @xmath0 , where the generalized chaplygin gas sound velocity can be larger than that of light . in the first model we find that the matter power spectrum is compatible with the observed one only for @xmath1 , which makes the generalized chaplygin gas practically indistinguishable from @xmath2cdm . in the second model
we study the evolution of inhomogeneities of the baryon component .
the theoretical power spectrum is in good agreement with the observed one for almost all values of @xmath3 .
however , the growth of inhomogeneities seems to be particularly favoured either for sufficiently small values of @xmath3 or for @xmath4 .
thus , it appears that the viability of the generalized chaplygin gas as a cosmological model is stronger when its sound velocity is superluminal .
we show that in this case the generalized chaplygin gas equation of state can be changed in an unobservable region in such a way that its equivalent @xmath5-essence microscopical model has no problems with causality .
_ keywords _ : chaplygin gas , cosmological perturbation theory , gauge invariant formalism , large scale structure power spectrum |
the model computations are based on the 2 dimensional , general relativistic mhd code harm2d , described in ref . .
the nuclear equation of state is discussed in detail in ref . .
the goal of our preliminary test calculations is to investigate the overall structure of an accretion disk in which nuclear reactions take place and the gas looses energy via neutrino cooling .
we assume the gas to be in beta equilibrium , so that the ratio of proton to neutron satisfies the balance between forward and backward nuclear reactions .
we assume neutrino cooling via electron , muon and tau neutrinos in the plasma opaque to their absorption and scattering .
neutrinos are formed in the following processes : urca process , electron positron pair annihilation , nucleon - nucleon bremsstrahlung , plasmon decay .
we solve grmhd equations using numerical code harm2d . in our model
, the gas is composed of mixture of @xmath0 and @xmath1 , so the total fluid energy is @xmath2 , where @xmath3 is the total rest mass energy density and @xmath4 is the total internal energy density and where @xmath5 .
the total gas pressure is @xmath6 .
all particles have a common temperature : @xmath7 ( @xmath8 ) and we assume polytropic equation of state @xmath9 , with adiabatic index @xmath10 .
we do not include the radiation and neutrino pressure in the dynamical calculation , but we do account for a neutrino cooling .
after each timestep , we compute the new gas composition and neutrino cooling rates @xmath11 as described in ref . .
the total internal energy is reduced using explicit method with @xmath12-sub - cycles in each time step .
the cooling function @xmath11 also accounts for neutrino absorption and scattering on free neutrons and protons .
the optical thicknesses for absorptions and scatterings are calculated assuming approximate disk vertical thickness @xmath13 and local proton and neutron number densities .
therefore in the dense disk regions the cooling is effectively zero .
we start the numerical calculation from the equilibrium model based on ref . around a spinning black hole .
the central black hole mass is @xmath14 and the total mass of the surrounding gas is about 0.1 @xmath15 .
harm2d code is designed to solve the magnetohydrodynamic equations in the stationary metric .
we assume dimensionless spin of a black hole to be @xmath16 . for higher disk masses , a fully general relativistic model including the evolution of the space - time and spin of the black hole should be considered .
the initial magnetic field is assumed with toroidal component of vector potential @xmath17 ( magnetic filed lines follow the constant density surfaces i.e. they form loops concentrated around the pressure maximum ) and normalized with @xmath18 .
the torus around the spinning black hole at hyper - eddington rates is cooled by neutrinos ( e.g. , as shown in 1-d model in ref . ) .
here we also study such tori , however to quantify the effect of neutrino cooling in mhd simulations , we run a starting test model with no cooling assumed . in figure [ fig1 ] , we show the structure of such a torus ( model 1 ) .
the snapshots present the baryon density ( @xmath19 ) , gas temperature t and magnetic @xmath20 parameter .
we also show the topology of magnetic field lines . a turbulent and dense rotating torus forms in the equatorial plane and extends towards the marginally stable orbit of the black hole .
the polar regions are quickly evacuated and an electromagnetically dominated jet is formed above the poles of the black hole , as shown before in refs . - .
the mass accretion rate is @xmath21 .
in figure [ fig2 ] we show model 2 , in which the neutrino cooling is accounted for in the dynamical evolution .
the initial conditions in model 2 are the same as in model 1 .
the mass accretion rate remains the same as in model 1 , but the structure of the disk changes .
the disk is geometrically thinner and strongly magnetized in comparison to model 1 .
we compute the neutrino emissivity at t=0.02 s. the total neutrino luminosity corrected for absorption and scattering processes , is @xmath22 erg s@xmath23 ( figure [ fig2 ] , left panel ) .
this number is comparable to numbers obtained from relativistic hydrodynamical simulations , e.g. in refs . - .
also , the relativistic mhd simulations in ref . reported the neutrino luminosity on the order of @xmath24 erg s@xmath23 , depending on black hole spin ( @xmath25 ) and torus mass . to compute the electromagnetic luminosity of the observed grbs
, one needs to consider the efficiency of neutrino - antineutrino annihilation process , as well as swallowing of some fraction of neutrinos by the black hole due to the curvature effects . in figure [ fig2 ] ( most right panel ) , we show the effective neutrino cooling .
the most neutrinos are formed within @xmath26 .
we do not see a clear neutrinosphere ( a surface at which the optical thickness for neutrino absorption and scattering is 1.0 ) .
its presence as well as the effect of neutrino trapping will be accounted when the neutrino pressure is incorporated in the dynamical calculations .
this research was supported in part by grant nn 203 512638 from the polish ministry of science and higher education . | gamma ray bursts ( grb ) are the extremely energetic transient events , visible from the most distant parts of the universe
. they are most likely powered by accretion on the hyper - eddington rates that proceeds onto a newly born stellar mass black hole .
this central engine gives rise to the most powerful , high lorentz factor jets that are responsible for energetic gamma ray emission .
we investigate the accretion flow evolution in grb central engine , using the 2d mhd simulations in general relativity .
we compute the structure and evolution of the extremely hot and dense torus accreting onto the fast spinning black hole , which launches the magnetized jets .
we calculate the chemical structure of the disk and account for neutrino cooling .
our preliminary runs apply to the short grb case ( remnant torus accreted after ns - ns or ns - bh merger ) .
we estimate the neutrino luminosity of such an event for chosen disk and central bh mass . |
skyrmions were predicted to occur in certain magnetic systems @xcite and were subsequently experimentally identified in the chiral magnet mnsi @xcite . since this initial discovery
there has been tremendous growth in the field as an increasing number of materials have been found that can support a skyrmion phase @xcite .
there are also numerous proposals on how to stabilize skyrmion states by utilizing different materials properties or bilayers @xcite .
direct imaging of skyrmions with lorentz microscopy @xcite and other techniques @xcite show that the skyrmions form a triangular lattice and have particle - like properties similar to vortices in type - ii superconductors @xcite . as an external magnetic field
is increased , skyrmions emerge from a spiral state and their density initially increases and then decreases with field until the sample enters a uniform ferromagnetic state @xcite . in bulk samples ,
skyrmions form three - dimensional ( 3d ) line objects and occur in a limited range of fields and temperatures @xcite , while for thin samples the skyrmions exhibit two - dimensional ( 2d ) properties and are stable over a much larger range of fields and temperatures extending close to room temperature @xcite .
skyrmions can be set into motion through the application of an external current @xcite , and it has been shown that there is a critical current above which the skyrmions depin into a sliding state @xcite .
skyrmion motion can be directly observed with lorentz microscopy @xcite or deduced from changes in the transport properties , permitting the construction of effective skyrmion velocity versus applied force curves that show that the skyrmion velocity increases with increasing current @xcite .
other methods to move skyrmions include the use of temperature gradients @xcite , electric fields @xcite , and coupling to a magnetic tip @xcite . from an applications standpoint
, skyrmions are attracting attention due to their potential use in racetrack memory devices where they would play a role similar to that of magnetic domain walls @xcite .
skyrmions have several advantages over domain walls due to their size and the fact that the current needed to depin a skyrmion can be orders of magnitude smaller than that needed to move domain walls @xcite .
it has been experimentally demonstrated that individual skyrmions can be created or annihilated with a magnetic tip , indicating that it is feasible to read and write skyrmions @xcite .
developing applications of skyrmions will require an understanding of how skyrmions interact with and move along tailored landscapes , so examining skyrmion dynamics on periodic substrates is an important step in this direction .
skyrmions have many similarities to vortices in type - ii superconductors , such as the effective skyrmion - skyrmion interaction , which is repulsive and favors triangular ordering , and the fact that both skyrmions and vortices can be driven by an external current . in the presence of quenched disorder ,
skyrmions exhibit pinning - depinning transitions @xcite , as observed in experiments @xcite and simulations @xcite , and these are similar to the depinning transitions observed for vortices in type - ii superconductors @xcite .
there are important differences between the two systems , particularly the dominant role that non - dissipative effects can take in skyrmion motion . for superconducting vortices , non - dissipative effects such as a magnus force
are typically small , permitting the system to be effectively described as obeying overdamped dynamics @xcite . in a skyrmion system , by contrast , the magnus term can strongly affect how the skyrmions interact with pinning sites and how they move under an external drive @xcite .
numerical simulations using continuum and particle - based models for skyrmions interacting with pinning have shown that the magnus term reduces the effective pinning in the system by creating a velocity component that is perpendicular to the force induced by a pinning site @xcite . as a result
, skyrmions have a tendency to swing around the edge of a pinning site and escape , while for overdamped systems a particle moves toward the center of a pinning site and is much more likely to be pinned . relatively little
is known about how particles with a strong magnus term move over a periodic substrate , and skyrmions are an ideal system to study such effects . in certain limits ,
a skyrmion can be effectively modeled as a point - like particle utilizing an equation of motion @xcite derived from thiele s equation @xcite .
comparison between continuum - based models and particle - based models of skyrmions moving in the presence of pinning have shown good agreement @xcite . in this work
we examine the dynamics of a skyrmion moving over a square periodic substrate using a particle - based description given in section ii .
we specifically examine the effect of changing the importance of the magnus term relative to that of the damping term . despite the apparent simplicity of this system ,
we show in section iii that the magnus term can induce a remarkably rich variety of dynamical behaviors that are absent in the overdamped limit .
we find that the hall angle for the skyrmion motion is dependent on the external drive amplitude and approaches the substrate - free limit only at higher drives .
since the skyrmion is moving over a periodic substrate , as the hall angle changes the motion becomes locked to specific symmetry directions of the substrate , producing a series of steps in the transport curves corresponding to integer and rational fractional ratios of the skyrmion velocity in the directions parallel and perpendicular to the drive direction . at the transitions into the different directional locking phases
, the skyrmion velocity exhibits a pronounced cusp accompanied by negative differential mobility in which the skyrmion velocity _ decreases _ with increasing external driving force .
we map the extent of the locking phase as a function of the external drive and the ratio of the magnus and dissipative terms , and find a rich structure of integer and fractional locking effects .
we also describe a speedup effect for skyrmions interacting with a substrate where the skyrmion velocity can be higher than the external drive .
this effect is most prominent just above depinning and is caused by the magnus term ; it is absent in the overdamped case . in section
iv we consider a skyrmion scattering from a single pinning site for varied impact parameters to show how the hall angle is reduced by the pinning due to a side - step phenomenon where the skyrmion trajectory is shifted by the magnus term as the skyrmion moves through the pinning site . for increasing drive amplitude ,
the skyrmion trajectories become less curved and the size of the side step is reduced , so that the hall angle approaches the clean limit for higher drives .
our results for the speedup effect and side steps are in agreement with recent theoretical and computational studies by m " uller and rosch , who considered a single skyrmion interacting with a defect site @xcite . in that work ,
the pinning potential is of a different form than the pinning sites we consider ; however , the consistency of the two studies indicates that the hall angle dependence on external drive and the speedup phenomenon are generic features of skyrmions interacting with pinning .
there are other examples of particles moving over ordered substrates , such as vortices in type - ii superconductors with periodic pinning arrays @xcite or colloids placed on optically created periodic substrates @xcite . in these systems
the dynamics is overdamped ; however , there can be directional locking effects in which the particles preferentially move along symmetry directions of the underlying substrate as the direction of drive is rotated with respect to the substrate lattice @xcite .
such directional locking effects can be exploited to perform particle separation in colloidal systems by setting up a system in which one particle species locks to the substrate while the second species does not , causing the two species to move at an angle with each other @xcite .
the direction of motion of the locked particles undergoes a series of steps as a function of the effective angle of drive with respect to the substrate .
the steps are centered at integer and rational ratios of the angle of drive and form a devil s staircase structure @xcite . in the skyrmion system , we observe directional locking effects when the direction of external drive is _ fixed _ with respect to the substrate and the drive amplitude is varied . the directional locking and transitions into the locking states in the skyrmion system exhibit a number of features that have not been observed in overdamped systems , such as an overshoot effect in the locking direction and negative differential mobility at the locking transition .
the steps in the transport response are also distinct from the shapiro steps found in systems that can be effectively described as a particle moving over a periodic substrate under superimposed ac and dc drives @xcite . in the system we consider in this work ,
there is no imposed ac drive .
we consider a skyrmion moving over a 2d square periodic substrate and utilize a particle - based description of the skyrmion from a recently developed equation of motion for skyrmions @xcite .
the dynamics of the skyrmion is obtained by integrating the equation of motion : @xmath5 here @xmath6 is the skyrmion velocity , and @xmath7 is a damping term representing spin precession and dissipation of electrons localized in the skyrmion . the substrate force @xmath8 is applied by placing @xmath9 skyrmions in a square lattice with lattice constant @xmath10 and fixing them in place .
these skyrmions interact repulsively with the mobile skyrmion .
such a substrate could be created experimentally by placing an array of magnetic dots on the sample to create a background of effectively immobile skyrmions .
alternatively , in recent continuum - based simulations it was shown that a defect site formed by removing a single spin creates a potential with a long range repulsion and a short range attraction @xcite .
our system can then be viewed as containing a periodic array of such pinning sites in the limit where the moving skyrmion does not experience the shorter range attraction of the pins .
we also consider a model in which the substrate potential is represented by a 2d sinusoidal array and find behavior very similar to that of a skyrmion moving through a pinned skyrmion lattice , indicating that our results are robust and capture the general features of skyrmions moving over periodic substrates .
the force from the pinned skyrmions has the form @xmath11 , where @xmath12 is the distance between the driven skyrmion @xmath13 and an immobile skyrmion @xmath14 . here
@xmath15 is the modified bessel function , which falls off exponentially for large @xmath12 , and @xmath16 is a screening length which we take to be @xmath17 in dimensionless units . the sample size is @xmath18 with @xmath19 , and the lattice constant of the substrate is @xmath20 . in the second system we model the substrate with the 2d periodic form @xmath21 $ ] , where @xmath20 is the substrate lattice constant and @xmath22 is the amplitude of the substrate force . in both systems we employ periodic boundary conditions .
the magnus term @xmath23 produces a force that is perpendicular to the skyrmion velocity , where @xmath24 is the magnitude of the magnus term .
the driving force is @xmath25 where @xmath26 is the direction of the applied drive .
such a force could arise due to the application of an external current to the skyrmion @xcite . in most of this work , we take @xmath27 .
we measure the skyrmion velocity parallel , @xmath1 , and perpendicular , @xmath0 , to the drive . in the absence of a substrate or in the overdamped limit @xmath28 , the skyrmion moves in the direction of the drive , while for a finite @xmath29 the skyrmion moves at an angle @xmath30 with respect to the drive , where @xmath31 . increasing @xmath29 produces a larger angle @xmath30 for the skyrmion motion with respect to the external drive . to quantify the direction of motion we measure @xmath32 so that the hall angle @xmath30 is given by @xmath33 .
we increase the external drive in small increments of @xmath34 to @xmath35 and wait several thousand simulation time steps before measuring the average velocity in order to ensure that the skyrmion velocity is in a steady state .
we find that for smaller increments @xmath36 our results do not change . throughout this work
we impose the constraint @xmath37 in order to maintain a constant magnitude of the skyrmion velocity for varied ratios of @xmath29 . in the overdamped limit with @xmath38 moving over a periodic substrate .
the drive is applied in the @xmath39-direction ; the system geometry is illustrated in fig .
[ fig:2 ] .
@xmath1 is the velocity component parallel to the applied drive and @xmath0 is the velocity component perpendicular to the drive .
here there is a depinning transition into a sliding state where the skyrmion moves strictly in the direction of the applied drive .
the hall term @xmath40 in this case .
dashed line : @xmath1 response in the absence of a substrate .
, width=336 ] with @xmath41 where the skyrmion ( large red dot ) is driven in the @xmath39-direction under an applied drive @xmath42 .
the black line is the skyrmion trajectory and the smaller blue dots indicate the locations of the potential maxima in the periodic square substrate lattice . ,
in fig . [ fig:1 ] we plot @xmath1 and @xmath0 versus @xmath42 for a skyrmion moving over a periodic substrate in the overdamped case where @xmath43 and @xmath44 .
figure [ fig:2 ] shows the system geometry , highlighting the motion of the mobile skyrmion through the background of potential maxima . in fig .
[ fig:1 ] there is a depinning transition to a sliding state for @xmath45 as indicated by @xmath46 . here
the skyrmion moves strictly in the direction of the applied drive so that @xmath47 for all @xmath48 and @xmath49 .
the dashed line in fig .
[ fig:1 ] is the expected value of @xmath1 in the clean limit , and we find that as @xmath48 increases , @xmath1 gradually approaches the clean limit value .
figure [ fig:2 ] shows the skyrmion trajectory for fixed @xmath50 , where motion occurs in a straight line along the driving direction . and @xmath1 vs @xmath42 for the same system as in fig .
[ fig:1 ] but with @xmath51 .
( b ) @xmath52 vs @xmath42 in the same sample .
the dashed line indicates that in the clean limit @xmath53 .
there are a series of dips in the velocity - force curves correlated with jumps in @xmath52 , indicating that the skyrmion undergoes a series of transitions between locking to different symmetry directions of the substrate . the largest steps in @xmath52 ,
corresponding to @xmath54 ratios of @xmath55 , @xmath56 , @xmath57 , @xmath58 , @xmath59 , and @xmath60 , are marked ; there are also additional smaller higher order steps at other integer values of @xmath3 and @xmath4 .
, width=336 ] with @xmath51 .
( a ) @xmath61 state at @xmath62 .
( b ) @xmath63 state at @xmath64 .
( c ) @xmath65 state at @xmath66 .
( d ) @xmath67 state at @xmath68 .
, width=336 ] in fig .
[ fig:3 ] we illustrate the transport behavior in a sample with @xmath51 , @xmath69 , and @xmath70 . in the absence of a substrate the skyrmion would move at an angle of @xmath71 with respect to the drive direction and would have @xmath72 .
figure [ fig:3](a ) shows @xmath1 and @xmath0 vs @xmath42 , and in fig .
[ fig:3](b ) we plot @xmath52 vs @xmath42 .
here there is a transition from a pinned to sliding state just above @xmath73 .
the skyrmion moves strictly in the direction of the drive for @xmath74 , which corresponds to the @xmath75 regime shown in fig .
[ fig:3](b ) for the same interval of @xmath48 .
figure [ fig:4](a ) shows the skyrmion trajectory in the @xmath75 region for @xmath62 for the system in fig .
[ fig:3](a ) .
the skyrmion no longer passes through the centers of the potential minima but its trajectory shifts closer to the bottom row of potential maxima and develops an oscillation in the @xmath76 direction that was absent in the @xmath38 case shown in fig .
[ fig:2 ] .
as @xmath42 is further increased the skyrmion shifts even closer to the row of potential maxima and its velocity parallel to the drive decreases , as indicated by the dip in @xmath1 near @xmath77 .
as @xmath48 increases further , both @xmath78 and @xmath1 increase , indicating that the skyrmion is now translating in the @xmath76 direction as well as in the @xmath39 direction .
additional dips in both @xmath1 and @xmath0 occur at higher @xmath48 , with occasional regions containing multiple closely spaced smaller dips .
figure [ fig:3 ] indicates that these dips correlate with the jumps in @xmath52 , and that between the dips , @xmath52 remains constant .
this means that the direction of motion or hall angle of the skyrmion changes in a discrete fashion with increasing @xmath42 .
the steps in @xmath52 appear at values that are rational ratios of @xmath79 of the form @xmath54 , where @xmath3 and @xmath4 are integers . in fig .
[ fig:3](b ) we highlight the steps at @xmath80 , 1/3 , 2/5 , 3/7 , and @xmath60 .
there are also numerous additional steps in @xmath52 for smaller intervals of @xmath42 for higher order rational ratios of @xmath81 . in general , the larger the values of @xmath3 and @xmath4 , the smaller the interval in @xmath42 over which the step appears . during each step interval in @xmath52
, the skyrmion follows an ordered periodic orbit , translating @xmath3 substrate plaquettes in the direction perpendicular to the drive for every @xmath4 plaquettes it translates in the direction parallel to the drive .
figure [ fig:4](a ) shows the orbit of a skyrmion in the @xmath61 state , while fig .
[ fig:4](b ) illustrates the orbit in the @xmath80 state at @xmath64 . here
the skyrmion moves periodically through the system at an angle @xmath82 with respect to the drive direction . in each period of the motion ,
the skyrmion translates to the right by four plaquettes in the drive or @xmath39 direction and down by one plaquette in the perpendicular or @xmath76 direction , giving @xmath83 . at the transition out of the @xmath80 state ,
the skyrmion slows down , producing a cusp in the velocity curves in fig .
[ fig:3](b ) and a dip in the net velocity near @xmath84 .
a similar ordered orbit occurs in the @xmath85 state , where the skyrmion moves three plaquettes in the @xmath39-direction and one plaquette in the @xmath76 direction during every period .
there is another dip in the net skyrmion velocity near @xmath86 that occurs when the system transitions to the @xmath87 state .
figure [ fig:4](c ) illustrates the orbit at this value of @xmath52 for @xmath66 . here
the periodic orbit is more extended and the skyrmion moves two plaquettes in the negative @xmath76-direction for every five plaquettes in the @xmath39-direction . above the @xmath87 state ,
the system enters the @xmath88 state , followed by some higher order steps which occur over small intervals of @xmath42 as shown in fig .
[ fig:3](a ) near @xmath89 .
the system then reaches the @xmath90 state illustrated in fig .
[ fig:4](d ) at @xmath68 , where the skyrmion moves two plaquettes in the @xmath39-direction for every one plaquette in the @xmath76-direction . at the other higher - order locking steps ,
similar ordered orbits appear . the dips in @xmath1 and @xmath0 in fig .
[ fig:3 ] are associated with transitions in the skyrmion orbit from one directionally locked state to another , with a corresponding change in the hall angle .
the symmetry of the square lattice determines the specific directions along which the skyrmion motion locks , so that for other geometries such as a triangular substrate , the locking directions will be different . at the transitions between locking steps , the net skyrmion velocity @xmath91 decreases with increasing external drive @xmath42 .
this phenomenon is known as negative differential mobility , and it has been observed in other systems where particles are driven over a periodic substrate , such as superconducting vortices moving over periodic pinning arrays where there are transitions between different dynamical phases @xcite .
negative differential mobility is also a common feature in semiconductor devices @xcite and can be useful for creating logic devices .
the ability to control differential mobility in skyrmion systems could open new approaches for applications . vs @xmath42 curve for the same system in fig .
[ fig:3 ] at @xmath51 , where we examine the locking for drives up to @xmath92 .
the system is locked to the @xmath93 value but then jumps to @xmath94 and @xmath95 before reaching @xmath96 .
as @xmath42 is further increased , @xmath52 gradually approaches the clean limit value , indicated by the dashed line .
, width=336 ] at @xmath98 in fig .
[ fig:3](b ) , @xmath99 , indicating the skyrmion is moving at a hall angle of @xmath100 , which is _ higher _ than the clean limit value of @xmath101 or @xmath102 .
this phenomenon occurs when the clean value of @xmath30 is oriented close to but slightly below a strong symmetry locking direction of the substrate .
the skyrmion locks to the substrate and its hall angle slightly exceeds that expected for the clean limit . as @xmath42 is increased , the effectiveness of the substrate is reduced and the direction of motion of the skyrmion gradually approaches the clean limit value , as shown in fig .
[ fig:5 ] where we plot a portion of the @xmath52 vs @xmath42 curve for the same system from fig .
[ fig:3](b ) but for drives up to a higher value of @xmath92 .
the dashed line in fig .
[ fig:5 ] indicates the clean value limit of @xmath101 . for @xmath103
, there is a jump from the @xmath104 state to a lower value of @xmath105 , followed by another jump to the @xmath106 state .
there is then a small region where the system locks to the @xmath107 state before jumping to @xmath108 .
as @xmath48 increases further , @xmath52 gradually approaches the clean limit value , as indicated by the dashed line in fig .
[ fig:5](b ) .
this shows that there can be an overshoot effect in the locking behavior for a certain range of drives , giving rise to non - monotonic behavior of @xmath52 as a function of @xmath42 .
we have observed similar overshoot effects for other values of @xmath29 . and @xmath1 vs @xmath42 for ( a ) @xmath109 and ( b ) @xmath110 .
( c , d ) the corresponding @xmath52 vs @xmath42 curves .
dashed line in ( d ) : the clean limit value for @xmath52 . in ( c )
we highlight the @xmath61 and @xmath111 steps , and in ( d ) we highlight the @xmath112 , 4/3 , 3/2 , 5/3 , 7/4 , @xmath113 and @xmath114 steps . ,
width=336 ] step for @xmath109 for the system in fig .
[ fig:6](a , c ) .
( b ) the @xmath115 step for @xmath116 for the system in fig .
[ fig:6](b , d ) .
( c ) the @xmath117 state for @xmath118 for the system in fig .
[ fig:8](a , c ) .
( d ) the @xmath119 state for @xmath118 in fig .
[ fig:8](a , c ) .
, width=336 ] and @xmath0 vs @xmath42 at @xmath118 .
( b ) @xmath1 vs @xmath48 for @xmath120 .
( c ) @xmath52 vs @xmath48 for the sample in panel ( a ) with @xmath121 .
the steps at @xmath122 , 3/1 , 7/2 , 4/1 , 17/4 , and 9/2 are highlighted .
( d ) @xmath52 vs @xmath48 for the sample in panel ( b ) with @xmath120 , where the steps at @xmath123 , 6/1 , 7/1 , 8/1 , 17/2 , and 9/1 are highlighted . ,
width=336 ] the specific features in the velocity force curves are strongly dependent on the value of @xmath29 . in fig .
[ fig:6](a , b ) we show @xmath0 and @xmath1 for @xmath109 and @xmath110 , respectively , while in fig .
[ fig:6](c , d ) we plot the corresponding @xmath52 vs @xmath42 curves . for @xmath109 ,
the system is predominantly locked to the @xmath112 step as shown in fig .
[ fig:6](b ) . on this step ,
the skyrmion moves along the @xmath124 direction as illustrated in fig .
[ fig:7](a ) for @xmath125 . here
the skyrmion follows a sinusoidal trajectory , moving by one plaquette in the @xmath39-direction and one plaquette in the @xmath76-direction in a single period . at values of @xmath42
higher than those plotted in fig .
[ fig:6](a ) , further locking steps occur as @xmath126 approaches @xmath127 . at @xmath116 , a larger number of steps
occur , as indicated in fig .
[ fig:6](d ) where we highlight the @xmath112 , 4/3 , 3/2 , 5/3 , 7/4 , 5/9 , and @xmath114 steps . for @xmath48 values
higher than those shown in fig .
[ fig:6](d ) , additional steps appear as @xmath126 approaches the clean limit value .
figure [ fig:7](b ) illustrates the skyrmion trajectory on the @xmath128 step for the system in fig .
[ fig:6](b ) .
the skyrmion moves five plaquettes in the direction perpendicular to the drive for every three plaquettes in the direction parallel to the drive during a single period .
[ fig:8](a ) shows @xmath1 and @xmath0 versus @xmath42 for a sample with @xmath118 , and the corresponding @xmath52 vs @xmath42 curve appears in fig .
[ fig:8](c ) . in this case , the clean limit hall angle is large , and there is no longer a phase where the skyrmions move only in the direction of the drive , so the @xmath55 step is lost . instead , above depinning the motion jumps straight into the @xmath129 state .
this is followed by steps at @xmath117 , 3/1 , 7/2 , 4/1 , 17/4 , and @xmath130 .
there are also numerous smaller intermediate steps corresponding to higher order fractions . in fig .
[ fig:7](c ) we plot the skyrmion orbit on the @xmath122 step for the system in fig .
[ fig:8](a , c ) , while fig . [ fig:7](d ) shows the skyrmion orbit at @xmath131 for the same system . in general
, as @xmath29 increases , more steps become visible . in fig . [
fig:8](b ) we plot only @xmath1 versus @xmath42 for a sample with @xmath132 to show more clearly the increased number of features in the velocity - force curve .
figure [ fig:8](d ) shows the corresponding @xmath52 vs @xmath42 curve , where we highlight the steps at @xmath133 , 6/1 , 7/1 , 8/1 , 17/2 , and @xmath134 .
the largest observable integer step has a value of @xmath3 that is the largest integer which is smaller or equal to the value of @xmath29 .
state for the system in fig .
[ fig:8](b , d ) at @xmath132 , and ( b ) the @xmath135 state for the same system . ,
width=336 ] in fig .
[ fig:9](a ) we plot the skyrmion trajectories at @xmath136 for the system in fig .
[ fig:8](b , d ) with @xmath120 , showing that the skyrmions circle around every fifth substrate maximum . at @xmath137 , shown in fig .
[ fig:9](b ) , the trajectories are much straighter and the skyrmion moves at an angle that causes it to translate by eight plaquettes in the negative @xmath76 direction for every one plaquette in the positive @xmath39-direction . vs @xmath29 where the pinned phase and the @xmath138 , 1/1 , 2/1 , and @xmath139 locking states occur .
, width=336 ] vs @xmath29 for the @xmath140 side of the @xmath111 locking step from fig .
[ fig:10 ] .
( b ) @xmath42 vs @xmath29 showing only the @xmath111 locking phase from fig . [ fig:10 ] .
the width of the step decreases for large @xmath42 .
, width=336 ] by conducting a series of simulations for varied @xmath29 we can examine the evolution of the different locking phases . in fig .
[ fig:10 ] we show the evolution of the pinned phase and the @xmath61 , 1/1 , 2/1 , and @xmath139 locking phases for @xmath141 and @xmath142 .
the depinning threshold for the pinned phase remains roughly constant as a function of @xmath29 . for small @xmath29 ,
the skyrmion motion is strictly in the @xmath39 direction as indicated by the presence of the @xmath61 phase . as @xmath42 increases , the range and the width of the lower order locking phases decreases while new locking phases appear .
the regions in which the locking occurs have the characteristic features of arnold tongues , which occur in dynamical systems when there are two competing frequencies @xcite . in our system , the two frequencies are given by the inverse rate of translation along the x direction and the inverse rate of translation along the y direction when the system is on a locking step .
these frequencies are quantized on the step due to the periodicity of the substrate ; unlike the case of shapiro steps , we do not apply any ac drive and the frequencies arise from the combination of dc motion and the substrate .
if the substrate were rectangular instead of square , a different set of arnold tongues would arise . at higher values of @xmath29 , a larger number of locking steps
appear in the velocity - force curves .
for example , at @xmath143 , as @xmath48 increases the system passes through a small region of @xmath61 followed by the @xmath111 locking phase and then by the @xmath117 phase .
another feature of the phase diagram is that at higher drives the width of each locking phase decreases . to illustrate this more clearly , we focus on the width of the @xmath144 step as a function of @xmath42 , as highlighted in fig .
[ fig:11](a ) where we plot the location of the @xmath111 step over the narrow range @xmath145 out to @xmath146 , much higher than the maximum value of @xmath48 shown in fig .
[ fig:10 ] . the lower edge of the locking regime , which was dropping to lower values of @xmath29 with increasing @xmath48 in fig .
[ fig:10 ] , bends back around for higher @xmath48 as shown in fig .
[ fig:11 ] , and shifts to higher @xmath29 with increasing @xmath48 .
the resulting nose structure is the origin of the overshoot phenomenon , where over a certain range of @xmath29 , the system locks to the @xmath111 step for lower @xmath48 only to drop out of that step as @xmath48 increases . here
the skyrmions can lock to the @xmath111 direction even though their resulting motion follows a higher angle @xmath30 than would occur in a clean system . as @xmath42 increases above the edge of the @xmath111 step , the skyrmion jumps to smaller values of @xmath126 in a series of steps . on each step
the skyrmion locks to different symmetry directions which are closer to the clean system value of @xmath147 .
the overall shape of the @xmath111 step is shown in fig .
[ fig:11](b ) , where we plot the range @xmath148 and @xmath149 .
there is a decrease in the extent of the locking phase at higher drives and at higher @xmath29 values .
the width of the @xmath111 step gradually decreases and approaches a point centered at the @xmath150 value for the highest drives .
we find that the regions over which the other integer locking phases occur have similar shapes to that shown in fig .
[ fig:11](b ) .
vs @xmath42 for the @xmath140 regime of the @xmath111 locking phase from fig .
[ fig:11](b ) . from top right to bottom right , @xmath151 , 0.8418 , 0.8668 , 0.98041 , 0.81475 , and @xmath152 .
the @xmath111 step is marked , and the dashed line indicates the value of @xmath52 for @xmath151 in the clean limit .
( b ) @xmath52 vs @xmath42 for the @xmath153 regime of the @xmath111 locking phase from fig .
[ fig:11](b ) . from top right to bottom right , @xmath154 , 1.1 , 1.134 , 1.17 , 1.207 , 1.246 , and @xmath155 .
the @xmath111 step is marked and the dashed line indicates the value of @xmath52 for @xmath156 in the clean limit . , width=336 ] a new set of locking steps arises for values of @xmath42 greater than that where the @xmath111 locking occurs in fig .
[ fig:11](a ) . in fig .
[ fig:12](a ) we plot @xmath52 versus @xmath42 for different values of @xmath29 on the @xmath140 side of the @xmath111 locking phase shown in fig .
11(b ) , where an overshoot effect occurs .
the dashed line is the value of @xmath52 at @xmath157 in the clean limit ; similar overshoots occur for the other values of @xmath29 .
the black curve shows that for @xmath157 in the presence of a substrate , the skyrmion locks to the @xmath111 direction for @xmath158 and then jumps to the @xmath159 state . for @xmath160 ,
@xmath52 jumps off of the @xmath161 step and , for higher values of @xmath42 , further small jumps in @xmath52 occur as @xmath52 approaches the clean value limit . as @xmath29 is increased toward @xmath150 , the width of the interval of @xmath42 over which the system is locked to the @xmath111 step increases since the clean limit hall angle is closer to the @xmath111 angle of @xmath124 . in each case , when the system jumps out of the @xmath111 step for increasing @xmath48 , it can jump into a series of other locking steps as @xmath52 gradually approaches the clean value . in fig .
[ fig:12](b ) we plot @xmath52 versus @xmath42 for the @xmath153 side of the @xmath111 locking phase from fig . [ fig:11](b ) .
the dashed line indicates the clean limit value of @xmath52 for @xmath156 .
as @xmath29 approaches @xmath17 from above , the extent of the region over which the system is locked to the @xmath111 state again increases . after jumping out of the @xmath111 state ,
the system jumps to larger values of @xmath126 in a series of smaller locking steps as it approaches the clean limit value of @xmath52 .
we observe similar changes in @xmath52 for the other integer locking phases such as @xmath117 and @xmath119 . vs @xmath29 highlighting the fractional locking steps with @xmath80 , 1/3 , 1/2 , 2/3 , 3/4 , 5/4 , 4/3 , 3/2 , 5/3 , and @xmath162 steps along with the integer matching steps with @xmath163 , 1/1 , and @xmath164 .
( b ) integer locking phases only for @xmath48 vs @xmath29 , for @xmath61 through @xmath165 from left to right .
, width=336 ] in fig .
13(a ) we plot the locking phases as a function of @xmath48 and @xmath29 over the range @xmath166 , highlighting the fractional locking steps at the @xmath80 , 1/3 , 1/2 , 2/3 , 3/4 , 5/4 , 4/3 , 3/2 , 5/3 , and @xmath162 states falling between the integer steps at @xmath61 , 1/1 , and 2/1 . here
the widths of the fractional locking states behave similarly to those of the integer locking states .
figure [ fig:13](a ) shows that for certain values of @xmath29 , the fractional steps will be the dominant feature observed in transport .
the higher order fractional steps , not shown in the figure , exhibit similar features .
the overshoot effect described for the @xmath111 locking step in fig . [ fig:11 ] occurs for all of the integer and fractional locking steps as well , with the overall width of each of the steps decreasing with increasing @xmath42 . in fig .
[ fig:13](b ) we plot only the integer locking regions from @xmath61 to @xmath165 as a function of @xmath48 and @xmath29 for the range @xmath167 . between each of the integer steps
, there is a series of fractional steps ( not shown ) similar to that illustrated in fig .
[ fig:13](a ) .
the fractional steps have the form @xmath168 , where @xmath9 are the integer steps .
vs @xmath42 for @xmath132 ( upper curve ) , @xmath169 ( lower curve ) , and the clean limit value ( dashed line ) .
here there are regions of the @xmath120 curve where @xmath170 exceeds the clean limit value , indicating a speedup or acceleration effect .
( b ) @xmath171 for @xmath172 ( upper line ) , @xmath169 ( lower line ) , and the clean limit value ( dashed line ) . the speedup effect is indicated by regions in which @xmath173 , width=336 ] the magnus term can produce an acceleration or speedup effect of the skyrmion , in which the speed of the skyrmion is higher in the presence of a substrate than it would be in the absence of a substrate or in the overdamped limit . in fig .
[ fig:14](a ) we plot the net skyrmion velocity @xmath174 versus @xmath42 for @xmath132 , the overdamped case of @xmath41 , and the clean limit . in the overdamped case
, @xmath170 is always smaller than the clean limit value , indicating that the substrate does not accelerate the skyrmion .
in contrast , for @xmath132 there are clear regions where @xmath170 is _ higher _ than the clean value .
this effect is most prominent for values of @xmath42 just above the depinning threshold . to show the speedup more clearly , in fig .
[ fig:14](b ) we plot @xmath171 for @xmath132 and 0.0 . for the overdamped case , @xmath175 over the entire range of @xmath48 .
the lowest values of @xmath176 occur just at depinning , and then @xmath176 gradually approaches zero as @xmath48 increases .
the behavior of the overdamped system is similar to the velocity - force curves observed in other overdamped systems such as colloids driven over periodic substrate arrays@xcite . for @xmath132
, @xmath176 shows a series of peaks which are correlated with transitions between the different locking regimes . for @xmath177 , @xmath178 due to the speedup effect , where there is an enhancement in the net velocity of up to twice the velocity in the clean limit . as @xmath42
is further increased the magnitude of the speedup effect decreases and there are several intervals of @xmath42 where the skyrmion is moving significantly slower than it would in the clean limit . for further increases in @xmath42 , @xmath176 approaches zero .
the speedup effect is generated by the non - dissipative terms in the equation of motion which cause the skyrmions to be accelerated through certain portions of the substrate potential .
vs @xmath29 for fixed @xmath179 .
the dashed line indicates the clean limit value of @xmath180 .
the speedup effect occurs when @xmath181 and increases in magnitude with increasing @xmath29 .
, width=336 ] in fig .
[ fig:15 ] we plot @xmath170 versus @xmath29 for a fixed @xmath179 , where the dashed line indicates the clean limit value of @xmath182 .
values of @xmath181 indicate a speedup effect . for @xmath38 , in the overdamped limit , @xmath183 .
we find a series of oscillations in @xmath170 corresponding to the different locking phases through which the system passes as a function of @xmath29 . here
, the magnitude of the speedup effect increases on average with increasing @xmath29 .
a speedup effect for a driven skyrmion interacting with a defect has also been observed in simulations by m " uller and rosch , who find that for some cases the defect causes a net increase in the skyrmion velocity @xcite .
they also find that the magnitude of the speedup effect decreases when an externally imposed skyrmion drift velocity is increased .
this is similar to what we observe for the periodic substrate case where for increasing external driving force the speedup effect is reduced as shown in fig .
[ fig:14](b ) . in ref .
@xcite the speedup was up to an order of magnitude higher than the velocity in the clean limit , while we observe a velocity enhancement of only up to a factor of 2 .
this is because in our system there is a minimum critical force required to depin the skyrmion , whereas in ref .
@xcite there was no lower bound on the imposed drift velocity .
since the speedup is increased for lower drives , m " uller and rosch could access lower drives and obtain larger velocity enhancements from the speedup effect . -direction .
, width=336 ] for @xmath184 and @xmath118 .
( a ) the net skyrmion velocity @xmath170 vs @xmath42 shows a series of cusps . dashed line : the clean value limit .
( b ) @xmath0 and @xmath1 vs @xmath42 for the same system .
( c ) @xmath52 vs @xmath42 for the same system with the @xmath112 , 2/1 , 3/1 , and @xmath185 steps highlighted .
there are also numerous smaller scale fractional locking steps .
, width=336 ] in order to understand how general our results are for different details of the periodic substrate , we next consider a 2d analytic substrate with the same lattice constant @xmath10 used in obtaining the previous results .
the force from the substrate is given by @xmath186 $ ] , where @xmath22 is the maximum force exerted by the substrate and @xmath20 is the lattice constant as in the system shown in fig .
[ fig:2 ] .
figure [ fig:16 ] shows a gray scale of a portion of the substrate , where the potential maxima are highlighted . in fig .
[ fig:17](a ) we plot @xmath170 versus @xmath42 for a skyrmion moving over a substrate with @xmath184 for @xmath118 .
the dashed line indicates the clean limit value of @xmath170 .
figure [ fig:17](b ) shows @xmath1 and @xmath0 vs @xmath42 for the same system , while in fig .
[ fig:17](c ) we plot the corresponding @xmath52 vs @xmath42 curve with a dashed line indicating the value of @xmath52 in a clean system and with the locking steps at @xmath112 , 2/1 , 3/1 , and @xmath185 highlighted .
although there are some differences in the details , we find the same general transport features for the analytic substrate as we observed for the pinned skyrmion substrate at @xmath118 , including the locking of the skyrmion motion to different symmetry directions and the steps in @xmath52 at integer and fractional ratios of @xmath79 .
figure [ fig:17](a ) shows that there are a similar series of dips in @xmath170 corresponding to transitions between the different locking phases .
we find that with an analytic potential , a somewhat larger number of small looking steps can be resolved , as shown for @xmath187 in fig .
[ fig:17](a ) , and the steps in @xmath52 in fig .
[ fig:17](c ) for this region have a devil s staircase structure .
figure [ fig:17](a ) also shows that @xmath170 exhibits regions where it is higher than the clean limit ( dashed line ) , indicating that the same type of speedup effect occurs for the analytic potential .
we have also examined the locking effects on the analytic potential for other values of @xmath29 and @xmath22 , and find that all the features highlighted in fig . [ fig:17 ] are robust .
this indicates that the directional locking effect is a generic feature of skyrmions moving over periodic substrates .
interacting with a single pinning site of radius @xmath188 and maximum pinning force of @xmath189 for an external drive @xmath48 applied in the positive @xmath76-direction .
( a ) @xmath28 , @xmath190 , and @xmath191 .
the skyrmions that interact with the pinning site are captured .
( b ) @xmath41 , @xmath192 , and @xmath193 .
the skyrmions can escape from the pinning site .
( c ) @xmath194 , @xmath190 , and @xmath191 .
the dashed line is a trajectory that a skyrmion would follow in the absence of the pinning site . for a certain range of impact parameters ,
the skyrmion is captured . for other impact parameters , skyrmions that escape from the pinning site
have their trajectories shifted as highlighted by the thick line .
( d ) @xmath195 , @xmath192 , and @xmath196 . here
the shift in the trajectories of the skyrmions that pass through the pinning site is reduced and there is less curvature in the trajectories . , width=336 ] in order to better understand the dependence of the hall angle on the external drive , we consider the case of a skyrmion scattering from a single pinning site .
we drive the skyrmion toward the pinning site for varied impact parameters @xmath197 , and measure the resulting shift in the skyrmion position perpendicular to the driving force for the outgoing state .
we model the pinning site as a parabolic trap with radius @xmath198 and a maximum force of @xmath189 .
the skyrmion is driven from a point outside the trap towards the trap with an external drive @xmath42 applied in the positive @xmath76-direction . in fig .
[ fig:18](a ) we show the overdamped case with @xmath28 , @xmath190 , @xmath188 , and @xmath191 , we highlight the trajectory of a skyrmion with an impact parameter of @xmath199 . for this value of @xmath42 ,
all skyrmion trajectories that contact the pinning site form straight lines directed toward the equilibrium position of the pinned state along the @xmath200 line .
the pinned equilibrium position is shifted in the positive @xmath76-direction from the center of the pinning site due to the applied external force .
figure [ fig:18](b ) shows the same system with higher @xmath193 .
the trajectories are shifted toward the center of the pinning site , but since @xmath201 the skyrmions can escape from the pinning site .
the trajectory for an impact parameter @xmath202 comes in and out of the pinning site at @xmath200 , so there is no shift , while the shift of the trajectories for non - zero impact parameters are symmetric across @xmath203 .
18(c ) shows the trajectories for the same system at @xmath194 for @xmath191 , where the external drive is still applied in the positive @xmath76-direction . in this case , the trajectories that do not intersect with the pinning site move at an angle of @xmath204 with respect to the @xmath76-axis .
the dashed line represents the trajectory a skyrmion would follow in the absence of the pinning site .
when the skyrmion moves through the pinning site , the trajectory is no longer straight as in the overdamped case but is now strongly curved due to the magnus force .
the skyrmions that encounter the pinning site on the upper left side become trapped by entering a spiral trajectory that carries them to the equilibrium pinned position in the upper portion of the pinning site .
skyrmions that enter the pinning site on the lower left side can escape from the pin ; however , the position of the outgoing skyrmions are shifted in the positive @xmath76 direction as indicated by the highlighted trajectory in fig . [ fig:18](c ) .
this shift is similar to the side jump effect that occurs for electron scattering @xcite , where the interaction with pinning or disorder shifts the outgoing trajectory of the particle relative to its incoming trajectory . in the case of a periodic substrate
, a skyrmion would be repeatedly shifted as it moves through the system , so that for @xmath205 the average direction of skyrmion motion follows a hall angle that is less than the clean value limit of @xmath206 .
figure [ fig:18](c ) also shows that when the skyrmion is captured it becomes pinned along the @xmath200 line at a position above the center of the pin .
the shift of the pinned equilibrium position to this location is caused by the bias from the external drive , and the location of the equilibrium position is not changed by the inclusion of a finite magnus term .
figure [ fig:18](d ) shows the trajectory for @xmath205 with a higher drive of @xmath193 . here
, the size of the side jump is reduced , so that the hall angle that would be observed for motion through a periodic substrate is closer to the pin - free value of @xmath206 .
this indicates that the hall angle increases with increasing driving force , as also observed in the simulations with a periodic substrate .
the shift is reduced at the higher drives in fig .
[ fig:18](d ) , since the skyrmion is moving more rapidly through the pinning site and the trajectory spends less time being bent . , the magnitude of the trajectory shift , vs @xmath42 for a skyrmion entering a single pinning site with an impact factor of zero .
( a ) @xmath190 , @xmath188 , and @xmath194 .
@xmath207 approaches zero at higher drives .
( b ) the same for @xmath150 .
, width=336 ] to quantify the the dependence of the magnitude @xmath207 of the shift or side jump of the trajectory as a function of the external drive , in fig . [ fig:19 ]
we plot @xmath207 for a skyrmion approaching the pinning site with @xmath202 .
figure [ fig:19](a ) shows the shift for @xmath190 , @xmath188 , and @xmath205 .
the shift is highest at low @xmath48 , starting near @xmath208 in fig .
[ fig:19](a ) and gradually approaching zero as @xmath48 increases . in fig .
[ fig:19](b ) we plot @xmath207 versus @xmath48 for a sample with @xmath190 , @xmath188 , and @xmath150 . here
the initial value of @xmath207 is smaller due to the smaller value of @xmath29 . from an initial value of @xmath209 ,
the magnitude of the shift decreases with increasing drive . for @xmath28 (
not shown ) , @xmath210 for all values of @xmath48 .
if we apply a fit to the decrease of the shift as a function of drive , we find @xmath211 , with @xmath212 depending on the choice of the low drive cutoff . in the work of m " uller and rosch for a skyrmion scattering off a single defect @xcite , they found through both simulations and perturbation that the shift as a function of drive goes as @xmath213 . even though the pinning potential we use has a different form than that used in ref .
@xcite , our results are consistent with the pinning site inducing a trajectory shift . vs impact parameter @xmath197 for a skyrmion moving through a pinning site with @xmath190 and @xmath214 .
( a ) @xmath38 for @xmath196 , 0.16 , and 0.20 , from upper left to lower left .
@xmath207 is symmetric across @xmath202 .
( b ) @xmath215 for @xmath216 , 0.09 , 0.10 , 0.11 , 0.12 , 0.14 , 0.16 , 0.18 , and 0.20 , from upper left to lower left . here
the shift is asymmetric across @xmath202 . for values of @xmath197 where there are no points ,
the skyrmion is captured by the pinning site .
( b ) @xmath205 for @xmath217 , 0.05 , 0.08 , 0.12 , and 0.20 , from upper left to lower left .
the shifts become larger with increasing @xmath29 .
, width=336 ] in order to understand the shift of the skyrmion and when the skyrmion will be captured by a pinning site , in fig .
[ fig:20 ] we plot @xmath207 as a function of the impact parameter @xmath197 .
figure [ fig:20](a ) shows the overdamped case @xmath28 for @xmath190 and @xmath218 at @xmath193 , 0.16 , and @xmath219 . at @xmath202 , @xmath210 and
there is no shift in the trajectory , while shifts for positive and negative values of @xmath197 are symmetric across @xmath202 so that the integrated shift over all impact parameters is zero .
for @xmath220 the skyrmion is always captured by the pin .
figure [ fig:20](b ) shows @xmath207 versus @xmath197 for a sample with @xmath150 at drives ranging from @xmath221 to @xmath222 . here
, @xmath207 decreases with increasing @xmath42 , and the shifts are asymmetric for positive and negative values of @xmath197 , so that the integrated shift over all impact parameters is positive . at @xmath223 and @xmath224 , values of @xmath197
at which there are no points on the curve indicate that the skyrmion was captured by the pinning site . for higher values of @xmath42 , @xmath225 is larger for @xmath226 than for @xmath227 . the results in fig .
[ fig:20](b ) show that increasing @xmath228 reduces the range of impact parameters where an incoming skyrmion is captured by the pin , and that for some ranges of @xmath197 a skyrmion can escape the pin even when @xmath229 , in contrast to the overdamped case where the skyrmion is always trapped whenever @xmath230 .
figure [ fig:20](c ) shows @xmath207 versus @xmath197 for @xmath194 over the range @xmath231 to @xmath232 .
here , the ability of the pinning site to capture the skyrmion is further reduced , and the skyrmion does not become pinned for more than half the impact parameters at @xmath231 even though this drive is substantially smaller than the maximum pinning force @xmath192 . at @xmath191
a skyrmion can only become captured if @xmath233 .
the overall shifts are positive over a much wider range of @xmath42 , and the magnitude of @xmath207 decreases with increasing @xmath42 .
this result shows that inclusion of a magnus term in the dynamics reduces the ability of pinning sites to capture particles since the magnus term induces a side jump or shift that permits the particles to escape from the pin .
we have numerically examined a skyrmion under a dc drive moving over a two - dimensional square pinning substrate for varied ratios of the magnus force term @xmath24 to the damping @xmath7 . in the overdamped limit @xmath38
, there is a single depinning transition into a sliding state where the skyrmion moves in the direction of the applied drive . for a finite @xmath29
we find that the skyrmion direction of motion or hall angle depends on the magnitude of the external drive and gradually approaches the substrate - free limit at high drives . due to the symmetry of the underlying substrate ,
the hall angle does not change continuously but passes through a series of steps as the skyrmion motion becomes locked to certain symmetry directions of the substrate .
these steps occur at integer and rational fractional ratios @xmath54 of the perpendicular to parallel velocity components of the skyrmion motion with respect to the direction of the applied drive , where @xmath3 and @xmath4 are the number of plaquettes the skyrmion moves in the perpendicular and parallel directions , respectively .
we find that the hall angle generally increases with increasing external drive , but that there can be an overshoot effect in which the hall angle is larger than expected for the substrate - free or clean limit . in this case , as the drive increases the hall angle jumps back to a lower value closer to the clean limit value . at the transitions between different directional looking steps , the skyrmion velocity shows
a striking series of cusps or dips where the skyrmion slows down for increasing @xmath42 , producing a negative differential mobility at the transitions . as @xmath29 increases , the number of transitions between different looking steps increases , as we map out in a series of phase diagrams .
the directional locking effects exhibited by the skyrmions are very distinct from the directional locking effects previously observed for overdamped particles such as vortices and colloids interacting with periodic substrates . in the overdamped system , the angle of the external drive
must be changed with respect to the symmetry direction of the substrate in order to induce locking steps , whereas for the skyrmion system the driving direction remains fixed ; only the magnitude of the driving force is changed . additionally , the overdamped systems exhibit neither the negative mobility phenomenon at the transitions between steps nor the overshoot effect .
we find that the skyrmion motion can exhibit a speedup or acceleration effect where the interaction of the skyrmion with a pinning site can accelerate the skyrmion such that the skyrmion velocity is higher than the value that would be induced by the external drive alone .
this effect is generally enhanced at the lower drives and suppressed near the directional locking transitions .
it contrasts strongly with the behavior of overdamped systems , where interactions with pinning or a substrate always decrease the velocity of an overdamped particle .
we find that all of these features are robust for different forms of the periodic substrate .
to better understand how the pinning can induce a hall angle dependence on the magnitude of the drive , we consider a skyrmion scattering from a single pinning site . in the overdamped limit with @xmath41 ,
the skyrmion becomes trapped by the pin whenever the external drive is less than the magnitude of the maximum pinning force that can be exerted by the pinning site . for external drives larger than this value
, the skyrmion escapes from the pin and its trajectory is shifted perpendicular to the direction of skyrmion motion .
the shift is symmetric for positive and negative impact parameters , so there is no net shift of the trajectory when all impact parameters are integrated .
in contrast , at a finite @xmath29 , the magnitude of the trajectory shift depends asymmetrically on the impact parameter , producing a nonzero net shift of the skyrmion as it moves through the pinning site when all impact parameters are integrated .
this shift is similar to the side jump phenomenon found in electron scattering . as the external drive
is increased , the net shift decreases and the skyrmion motion approaches the pin - free trajectory track .
the trajectory shift is responsible for the modified hall angle experienced by the skyrmion at low drives , and the gradual decrease in shift with increasing drive causes the hall angle to gradually approach the clean limit value . for the case of a periodic substrate ,
the additional symmetry of the substrate prevents the angle of the skyrmion motion from changing continuously ; instead , the motion changes in a series of steps with increasing drive as it approaches the clean limit value .
we also find that for finite @xmath29 , depending on the impact parameter the skyrmion may or may not become pinned regardless of whether the maximum pinning force is smaller than the driving force , and that for increasing @xmath29 , the range of impact parameters over which the particle is not pinned even when the driving force is smaller than the maximum pinning force increases .
this is one of the reasons that pinning of skyrmions is much weaker than pinning in overdamped systems such as superconducting vortices .
j. gutierrez , a.v .
silhanek , j. van de vondel , w. gillijns , and v.v .
moshchalkov , phys .
b * 80 * , 140514 ( 2009 ) ; s. avci , z.l .
xiao , j. hua , a. imre , r. divan , j. pearson , u. welp , w.k .
kwok , and g.w .
crabtree , appl .
. lett . * 97 * , 042511 ( 2010 ) . | we examine the dynamics of a skyrmion moving over a two - dimensional periodic substrate utilizing simulations of a particle - based skyrmion model .
we specifically examine the role of the non - dissipative magnus term on the driven motion and the resulting skyrmion velocity - force curves . in the overdamped limit
, there is a depinning transition into a sliding state in which the skyrmion moves in the same direction as the external drive . when there is a finite magnus component in the equation of motion , a skyrmion in the absence of a substrate moves at an angle with respect to the direction of the external driving force .
when a periodic substrate is added , the direction of motion or hall angle of the skyrmion is dependent on the amplitude of the external drive , only approaching the substrate - free limit for higher drives . due to the underlying symmetry of the substrate
the direction of skyrmion motion does not change continuously as a function of drive , but rather forms a series of discrete steps corresponding to integer or rational ratios of the velocity components perpendicular ( @xmath0 ) and parallel ( @xmath1 ) to the external drive direction : @xmath2 , where @xmath3 and @xmath4 are integers .
the skyrmion passes through a series of directional locking phases in which the motion is locked to certain symmetry directions of the substrate for fixed intervals of the drive amplitude . within a given directionally locked phase ,
the hall angle remains constant and the skyrmion moves in an orderly fashion through the sample .
signatures of the transitions into and out of these locked phases take the form of pronounced cusps in the skyrmion velocity versus force curves , as well as regions of negative differential mobility in which the net skyrmion velocity decreases with increasing external driving force .
the number of steps in the transport curve increases when the relative strength of the magnus term is increased .
we also observe an overshoot phenomena in the directional locking , where the skyrmion motion can lock to a hall angle greater than the clean limit value and then jump back to the lower value at higher drives .
the skyrmion - substrate interactions can also produce a skyrmion acceleration effect in which , due to the non - dissipative dynamics , the skyrmion velocity exceeds the value expected to be produced by the external drive .
we find that these effects are robust for different types of periodic substrates .
using a simple model for a skyrmion interacting with a single pinning site , we can capture the behavior of the change in the hall angle with increasing external drive .
when the skyrmion moves through the pinning site , its trajectory exhibits a side step phenomenon since the magnus term induces a curvature in the skyrmion orbit . as the drive increases
, this curvature is reduced and the side step effect is also reduced . increasing the strength of the magnus term reduces the range of impact parameters over which the skyrmion can be captured by a pinning site , which is one of the reasons that strong magnus force effects reduce the pinning in skyrmion systems . |
the _ herschel space observatory _ has revealed a wide - ranging network of elongated ( filamentary ) structures in molecular clouds ( e.g. , * ? ? ? * ; * ? ? ? * ) . even though filamentary structures in molecular clouds were already well established ( e.g. , * ? ? ?
* ) , the herschel continuum maps of dust emission at 70 - 500 @xmath0 m have achieved unprecedented sensitivity and revealed a deeper network of filaments , in both star - forming and non - star - forming molecular clouds .
this implies that the filamentary network is an imprint of initial conditions , likely turbulence , rather than the result of pure gravitational instability .
furthermore , the prestellar cores and protostars , when present , are preferentially found along massive filaments . much interpretation of the filaments has been based on the assumption that they are isothermal cylinders . this simplifies their analysis as their observed shape is then independent of most viewing angles and one can rely on established theoretical results about the equilibrium or collapse of infinite cylinders . @xcite interpreted the observations in terms of the critical line mass of an isothermal cylinder @xmath1 , where @xmath2 is the isothermal sound speed . for a mass per unit length @xmath3 ,
a cylinder undergoes indefinite collapse as long as the gas is isothermal , and for @xmath4 it can settle into an equilibrium structure , although still unstable to clumping along its length into jeans length sized fragments @xcite .
@xcite argue that star formation is initiated when @xmath3 .
a challenge to the view of filaments as cylinders is the magnetic field alignment inferred from polarized emission .
@xcite find that large scale magnetic fields are aligned perpendicular to the long axis of the massive star - forming filaments ( see also * ? ? ? * ) .
this makes a circular symmetry of a cylinder about the long axis unlikely unless the magnetic field strength is dynamically insignificant .
a more natural configuration is a magnetic ribbon , a triaxial object that is flattened along the direction of the large - scale magnetic field with its shortest dimension in that direction . in the lateral direction to the magnetic field
, elongated structures can form due to turbulence and gravity .
indeed , simulations of turbulence accelerated star formation in a strongly magnetic medium @xcite show the formation of ribbon - like structure in a layer that is flattened along the magnetic field direction .
magnetic ribbons have recently been investigated theoretically by @xcite and @xcite .
they study magnetohydrostatic equilibria of ribbons that arise from a parent filament of radius @xmath5 , which is a free parameter in the problem .
they find that a critical line - mass - to - flux ratio exists for collapse , in analogy to the critical mass - to - flux ratio for axisymmetric three - dimensional objects @xcite .
a further challenge to filaments modeled as isothermal cylinders comes from the dust emission measurement of the fwhm of the mean column density profile relative to the axis of a filament @xcite .
for example , figure 7 of @xcite shows that the fwhm values for 90 filamentary structures in low mass star forming regions cluster around a mean of @xmath6 pc with some scatter over two orders of magnitude range of mean column density pc for filaments in velocity - integrated emission and @xmath7 pc for filaments in individual velocity channels @xcite ] .
however , @xcite showed that the central half - mass radius of an equilibrium isothermal cylinder is @xmath8 , essentially the jeans length , where @xmath9 is the central density .
the projected column density of such a circularly symmetric configuration has a central flat region of size @xmath10 and column density @xmath11 ( see * ? ? ?
* ) , so that we can also write @xmath12 . therefore , the approximate observed relation @xmath13 is unlike the expected @xmath14 .
however , the observed set of values of the fwhm radii also intersect the line of jeans length at the median log column density , which implies that the jeans length may not be wholly unrelated to them . in this paper
, we explore the consequences of a magnetic ribbon model for molecular cloud filaments for the measured relation between apparent width and the observed column density .
we argue that this is a more natural model for filaments and is based on the interplay between turbulence , strong magnetic fields , and gravitationally - driven ambipolar diffusion , rather than pure gravity and thermal pressure .
we extend the analytic model of @xcite for the formation of magnetic ribbons that is based on numerical simulations .
we derive a lateral width of a magnetic ribbon and use it to calculate a synthetic observed relation between apparent width in projection versus observed column density .
dynamically important magnetic fields , corresponding to mass - to - flux ratios that range from subcritical to mildly supercritical , will lead to flattening along the magnetic field direction , and subsequent evolution will be primarily perpendicular to the magnetic field @xcite .
even highly turbulent three - dimensional simulations @xcite show that the turbulence is eventually dominated by motions perpendicular to the ambient magnetic field .
observations of some filaments ( e.g. , * ? ? ?
* ) that show a large - scale magnetic field along the short dimension of the filament also support the idea of flattening along the field . in this paper
, we adopt the scenario of turbulent compression acting primarily perpendicular to the magnetic field direction in an initially subcritical molecular cloud .
this leads to the paradigm of turbulence accelerated star formation , in which star formation occurs with globally low efficiency and in turbulent compressed regions .
these regions oscillate about an approximate force - balanced state until ambipolar diffusion creates supercritical pockets that collapse to form stars .
we explore the consequences of this scenario by extending a semi - analytic model of @xcite that is based on numerical simulations .
we consider local pressure balance of a compressed region in a subcritical cloud and neglect thermal pressure in comparison to magnetic pressure and the ram pressure of the flow .
we assume that the cloud is stratified in the _ z_-direction , with compression happening primarily in the @xmath15 plane . here
we simplify the analysis of the compression by limiting it to one direction , the _
x_-axis ( fig .
[ ribbon ] ) , as done by @xcite .
the initial magnetic field strength is @xmath16 and the field strength increases upon compression until the magnetic pressure within the compressed ribbon balances the external ram pressure and magnetic pressure .
hence the compression ends ( and oscillations may ensue ) when @xmath17 where @xmath18 is the nonlinear flow speed . assuming that the gas has adequate time to settle into hydrostatic equilibrium along the _ z_-direction , the half thickness of the cloud is @xmath19 @xcite .
now if the ambipolar diffusion time is longer than the compression time @xcite , flux freezing is valid during compression , i.e. , @xmath20 for the surface density @xmath21 , equation ( [ fluxfreezing ] ) can be rewritten as @xmath22 using equation ( [ scaleheight ] ) and equation ( [ fluxfreezing1 ] ) in equation ( [ balance eq ] ) and with some simplifications we get @xmath23 where @xmath24 is the square of the initial alfvn speed of the cloud . the consequence of such compression results in the formation of magnetic ribbons of width @xmath25 and thickness @xmath26 , as they are flattened along the direction of magnetic field ( see fig . [ ribbon ] ) . for conservation of mass per unit length in the ribbon during the compression of the cloud @xmath27 where @xmath28 is the initial width ( along the _
x_-axis ) and @xmath29 is the initial thickness of the cloud in the vertical direction ( _ z_-axis ) . using equation ( [ scaleheight ] )
, we can simplify the above equation to @xmath30 by using equation ( [ densitylength ] ) in equation ( [ densityvelocity ] ) , we can express the final width of the filament as @xmath31^{-1}.\ ] ] analysis of zeeman measurements of the magnetic field in molecular clouds presented by @xcite shows that the turbulent line width is comparable to the alfvn speed @xcite .
if we make the plausible estimate that the flow speed is comparable to the alfvn speed , i.e. , @xmath32 , the filament width becomes @xmath33 the result illustrates the fact that the final width of a filament is _ independent _ of the density of the medium .
instead it is a fraction of the initial length scale @xmath28 of the compressed region .
[ ribbon ] in the above theory , the final ribbon width @xmath25 is independent of its density , but does depend on the initial compression scale @xmath28 associated with turbulence .
the origin and physics of @xmath25 is then quite different than that of the jeans length .
what is @xmath28 then ? in the turbulent scenario that we adopt in this study , it would be associated with the dominant mode in the turbulent flow field in a molecular cloud . at this point , no first principles theory exists to calculate @xmath28 as the preferred mode of an instability that leads to molecular cloud turbulence .
hence , we take guidance from observations to make an empirical estimate for @xmath28 .
the column density maps used by @xcite to determine the mean filament width can also be used to estimate a mean spacing between filaments , which we identify with @xmath28 in our model .
figure 3b of @xcite identifies 27 filaments in a dust emission map of the cloud ic 5146 .
we use this same map to make an approximate measurement of @xmath28 . for each of the 27 filaments
, we identify a center along the length ( spine ) of the filament .
then for each filament we measure the distance to the nearest center point of another filament .
we obtain a set of 20 unique distance measurements ( eliminating double counting in cases where two filaments are mutually each other s nearest neighbor ) .
we ignore effects of an inclination angle @xmath34 in this analysis , which could mean that the measured distances are less than the actual distances by a factor @xmath35 .
our measured filament spacings ( which we equate with @xmath28 ) have a minimum value 0.5 pc , a maximum value 2.2 pc , a median value 0.9 pc , and a mean value 1.0 pc .
another way to constrain @xmath28 is through the star formation timescale in molecular clouds .
since this number is widely accepted to be in the range 1 - 3 myr @xcite , and star formation is often coordinated along a filament over this timescale , we can place an upper limit ( again empirically ) on the compression timescale @xmath36 of a few myr .
therefore the initial length scale that can trigger a compression can be written as @xmath37 where @xmath36 is as above and again using the alfvnic nature of turbulence .
this further simplifies to @xmath38 where we have used pressure balance along the magnetic field , @xmath39 and the normalized mass - to - flux ratio @xmath40 .
if we consider the initial cloud to be mildly subcritical , i.e. , @xmath41 , and a sound speed @xmath42 , then @xmath43 myr leads to @xmath44 pc .
the simple arguments of the previous section show that the length scale at which the ribbon formation is initiated is of the order of a parsec .
our semi - analytic model then implies that the final width of the ribbon given by equation ( [ intrinsic width ] ) is @xmath45 pc .
however , the shortest dimension is flattened along the direction of the magnetic field and has a thickness @xmath26 that _ does _ depend on the column density .
therefore , the observed shape will depend on the viewing angle .
below we calculate the observed width for a particular viewing angle and then calculate a synthetic plot of observed ribbon width versus observed column density for a collection of random viewing angles .
our objective is to gain insight into the form of the observed correlation , and how it compares with the standard jeans length scaling and with the observational results presented by @xcite .
the value of @xmath28 can be considered a free parameter and physically may vary from one cloud to another and have a distribution of values within a single cloud .
while we do not advocate a specific individual value for @xmath28 , we use the empirical estimate that it should be @xmath46 pc to determine the shape and approximate quantitative values of an observed correlation .
let the normal to the filament , along the
_ z_-axis , be inclined at an angle @xmath47 to the observer as shown in fig .
[ orientation ] . for a ribbon - like filament of intrinsic width @xmath25 and half thickness @xmath48 the projected width @xmath49 as seen by the observer is @xmath50 if the ribbon is viewed face on , i.e. , @xmath51 , the observed width is just the intrinsic width @xmath25 . when viewed side on i.e. , @xmath52 the observed width is the thickness @xmath26 of the ribbon along the _ z _ axis . for any other intermediate angles one sees the projection in the _ y - z _ plane i.e. , equation ( [ l_observed ] ) , as shown in fig .
[ orientation ] .
+ from our analysis we have already shown in equation ( [ intrinsic width ] ) that the intrinsic width @xmath25 is a fraction of the initial compression length scale @xmath28 .
the thickness @xmath26 of the magnetic ribbon is evaluated using the hydrostatic equilibrium , equation ( [ scaleheight ] ) , along the direction of the magnetic field ( i.e. , perpendicular to the filament width ) . for column density @xmath21 , the half - thickness of the clouds
is estimated to be @xmath53 for a ribbon of any particular column density @xmath54 we can estimate the corresponding half - thickness @xmath48 , which is essentially the jeans scale , using equation ( [ hsigma ] ) . for example , @xmath55 pc for @xmath56 and @xmath57 in which @xmath58 .
the observed column density @xmath59 will be different from the intrinsic column density @xmath54 depending on the angle at which the ribbon is being viewed . if the observer is situated at angle other than @xmath60 ( as shown in fig .
[ ribbon ] ) , the length along the line of sight changes thus affecting the observed column density . in the following section we will analyze the variation of @xmath59 with the viewing angles @xmath47 and @xmath61 .
we neglect the variation of @xmath62 within the ribbon . for the beam incident on the face of the ribbon at an angle @xmath64 and @xmath65 , ( refer to fig .
[ orientation]a , ) the observed column density is @xmath66 since the intrinsic column density @xmath21 , we get @xmath67 thus only for @xmath68 , i.e. , when the ribbon is viewed face on , @xmath69 . for @xmath70 , @xmath71 . for the beam entering at a critical angle @xmath73 and @xmath65 ( fig .
[ orientation]b ) , the observed column density is @xmath74 rearranging the above equation , we find that @xmath75 ) .
middle panel : case 2 ( @xmath72 ) when the ribbon is observed at a critical angle as shown .
bottom panel : case 3 ( @xmath76 ) .
, height=491 ] the critical angle @xmath73 separates the two sets of angles that have separate expressions for @xmath77 . for the beam entering through the shorter dimension of the ribbon ( see fig . [
orientation]c ) at an angle @xmath78 and @xmath65 , the observed column density is @xmath79 using equation ( [ scaleheight ] ) , and @xmath21 , we get @xmath80 for @xmath81 , i.e. , when the ribbon is viewed side on , @xmath82 . for @xmath83 , @xmath71 if @xmath84 , i.e. , when the ribbon width is greater than the thickness of the ribbon , the observed column density is greater than the actual column density .
furthermore , different angular orientation of the magnetic ribbons in the _ x - y _ plane will also alter the observed column density . if the long axis ( _ y _ ) of the ribbon is not perpendicular the line of sight ( see fig . [ ribbon ] ) , i.e. , @xmath85 the observed column density will further increase . for any random orientation in the _ x - y _ plane ,
the modified column density is @xmath86 @xmath87 however , different orientations in the _ x - y _ plane do not affect the observed ribbon width .
the resultant projection of the ribbon width on the _ y - z _ plane is independent of the azimuthal angle @xmath61 . since the observations of @xcite
reveal a relatively flat relation between observed width and column density , it is instructive to use our model to make a synthetic map of these quantities .
for simplicity we consider @xmath88 in this analysis .
we take a sample of 100 ribbons with number column density distributed uniformly in the range @xmath89 .
furthermore we take viewing angles randomly chosen in the range @xmath90 .
each pair of values @xmath91 yields a pair of values @xmath92 represented as blue dots in fig .
[ scatter ] .
we obtain @xmath49 using equation ( [ l_observed ] ) and @xmath59 using equation ( [ sigma1 ] ) or equation ( [ sigma2 ] ) for @xmath93 or @xmath94 , respectively . the black dashed line in fig .
[ scatter ] is the locus of points obtained by taking 100 randomly chosen values of @xmath47 for each value of @xmath95 and calculating the mean values of @xmath49 and @xmath96 .
this line is similar to the result of taking equation ( [ l_observed ] ) and inserting the mean values of @xmath97 and @xmath98 , which are both equal to @xmath99 , and replacing @xmath54 with the mean value of @xmath59 .
however , the mean value of @xmath59 across all angles is not exactly equal to @xmath54 .
both the set of individual synthetic data points shown in blue dots as well as the average relation in the black dashed line show a relatively flat relation over two orders of magnitude variation in @xmath96 . fig .
[ scatter ] also shows the analytic relation for two limiting cases .
the black dotted line corresponds to the face on view ( @xmath68 ) where @xmath100 pc and is independent of the column density .
the blue dot - dashed line corresponds to @xmath101 where @xmath102 , essentially the jeans length . versus observed column density @xmath96 .
each blue dot corresponds to a magnetic ribbon with intrinsic column density @xmath95 and observing angle @xmath47 , chosen as described in sec .
[ obscorr ] .
the black dashed line is the mean ribbon width for the entire range of values of @xmath96 .
the black dotted line is the width when the ribbon is viewed at @xmath68 .
the blue dot - dashed line is the width for the side on view i.e. , @xmath101 .
, height=207 ]
we have presented a minimum hypothesis model for the width of a filament in a molecular cloud in which magnetic fields and magnetohydrodynamic turbulence are initially dominant .
a turbulent compression leads to a magnetic ribbon whose thickness is set by the standoff between ram pressure and magnetic pressure region .
gravitationally - driven ambipolar diffusion then leads to runaway collapse of the densest regions in the ribbon , where the mass - to - flux ratio has become supercritical .
this process has been demonstrated in published simulations of trans - alfvnic turbulence in a cloud with an initial subcritical mass - to - flux ratio ( e.g. , * ? ? ?
* ; * ? ? ?
we have extended the semi - analytic model of @xcite to estimate their lateral ( perpendicular to magnetic field and ribbon long axis ) width .
this quantity is independent of the density of the ribbon .
this lateral width can also be used to estimate the parent filament radius @xmath5 in the theoretical magnetic ribbon model of @xcite . in our model ,
the thickness parallel to the magnetic field is essentially the jeans scale and does depend on density .
hence , we calculate a distribution of apparent widths seen in projection assuming a random set of viewing angles .
the resulting distribution of apparent widths versus apparent column density is relatively flat ( unlike expectations based on the jeans length ) over the range @xmath103 @xmath104
@xmath105 @xmath104 , in rough agreement with the observations of @xcite .
other models have been introduced to explain the apparent near - uniform width of observed filaments .
@xcite introduce an external pressure to an isothermal cylinder and find that the fwhm versus column density is a peaked function and approximately flat in the regime where @xmath106 is well below @xmath107 and the external pressure is comparable to the central pressure .
however , filaments with @xmath108 would be in a time - dependent state of dynamical collapse .
@xcite develop a model of a cylindrical self - gravitating filament that is accreting at a prescribed rate .
a near - uniform radius is derived based on assumption of a steady - state balance between energy input from accretion and dissipation of energy by ion - neutral friction at the filament radius scale .
@xcite also develops a model of accretion at the free - fall rate onto a filament with @xmath109 and uses various prescribed forms of internal structure to find that the fwhm has a peaked dependence on column density . a series of simulation papers @xcite use hydrodynamic or mhd simulations ( with supercritical mass - to - flux ratio ) and analyze filament widths at particular snapshots in time .
although their filament widths cluster at @xmath6 pc with some scatter , there is a mild to strong density dependence of the widths , and the filaments are single time snapshots in a situation of continuing collapse .
@xcite suggests that @xmath6 pc is special since the linewidth - size relation of @xcite would lead to subsonic turbulence below that scale , but it is not clear if his simulations satisfy this scaling internally .
we believe that the magnetic ribbon model provides an alternative simplified interpretation that accounts for turbulence and strong magnetic fields .
we have developed a method to estimate the width of a magnetic ribbon based on the characteristic scale and amplitude of mhd turbulence .
such ribbons can have a line mass that exceeds the hydrodynamic limit @xmath110 and still be in a dynamically oscillating quasi - equilibrium state . however , gravity still leads to star formation in the dense interior through rapid ambipolar diffusion .
we thank the anonymous referee for comments that improved the discussion in this paper .
this work was supported by a discovery grant from nserc . | we develop a magnetic ribbon model for molecular cloud filaments . these result from turbulent compression in a molecular cloud
in which the background magnetic field sets a preferred direction .
we argue that this is a natural model for filaments and is based on the interplay between turbulence , strong magnetic fields , and gravitationally - driven ambipolar diffusion , rather than pure gravity and thermal pressure .
an analytic model for the formation of magnetic ribbons that is based on numerical simulations is used to derive a lateral width of a magnetic ribbon .
this differs from the thickness along the magnetic field direction , which is essentially the jeans scale .
we use our model to calculate a synthetic observed relation between apparent width in projection versus observed column density .
the relationship is relatively flat , similar to observations , and unlike the simple expectation based on a jeans length argument . |
the diamond channel introduced by schein @xcite consists of a broadcast channel ( bc ) from a source to two relays and a multiple access channel ( mac ) from the two relays to a destination .
the capacity of the diamond channel is not known in general . to simplify the problem ,
let us consider a diamond channel having bc with two orthogonal links and gaussian mac . in this setup , there is a tension between increasing the amount of information sent over the bc and increasing the coherent combining gain for the mac .
two coding schemes corresponding to the extremes would be partial decode - and - forward , where independent partial messages are sent to the relays , and decode - and - forward ( df ) , where the whole message is sent to each of the relays . by incorporating multicoding at the source , @xcite , @xcite proposed a coding scheme in which the relays send independent partial messages using dependent codewords and showed that this coding scheme strictly outperforms the df and partial df in some regime .
furthermore , @xcite showed an upper bound by taking into account the correlation between the two relay signals , which is strictly tighter than the cutset bound .
this upper bound was shown to coincide with the lower bound of @xcite , @xcite for some channel parameters . in this paper , we consider the degraded gaussian diamond - wiretap channel presented in fig .
[ fig : physical ] and present lower and upper bounds on the secrecy capacity by exploiting the correlation between the two relay signals .
we identify several ranges of channel parameters where these bounds coincide with useful intuitions and investigate the effect of the presence of an eavesdropper on the capacity .
we note that this model is a natural first step to studying diamond - wiretap channel because the sum secrecy capacity of the multiple access - wiretap channel has been characterized only for the degraded gaussian case @xcite .
a practical situation corresponding to this model is the side channel attack @xcite where the eavesdropper attacks by probing the physical signals such as timing information and power consumption leaked from the legitimate destination . in the presence of an eavesdropper , the technique of utilizing randomness is widely used to confuse the eavesdropper .
we consider the following two scenarios regarding the availability of randomness : 1 ) a common randomness of rate @xmath0 is available at the source and the two relays and 2 ) a randomness of rate @xmath0 is available only at the source and there is no available randomness at the relays .
see @xcite , @xcite for the related works assuming restricted randomness at encoders . for the upper bound
, we generalize the upper bound on the capacity of the diamond channel @xcite and the upper bound on the sum secrecy capacity of the multiple access - wiretap channel @xcite . for the lower bound
, we propose two types of coding schemes : 1 ) a decode - and - forward ( df ) scheme where the relays cooperatively transmit the message and the fictitious message and 2 ) a partial df scheme incorporated with multicoding in which each relay sends an independent partial message and the whole or partial fictitious message using dependent codewords .
if there is no secrecy constraint , our partial df scheme incorporated with multicoding falls back to that in @xcite , @xcite .
interestingly , in the presence of the eavesdropper , the availability of randomness at the encoders is shown to affect the optimal selection of correlation coefficient between the two relay signals in our proposed schemes .
the remaining this paper is organized as follows . in section [ sec :
model ] , we formally present the model of the degraded gaussian diamond - wiretap channel . our main results on the secrecy capacity are given in section [ sec : main ] . in section [ sec :
proof ] , we derive our upper and lower bounds on the secrecy capacity .
we conclude this paper in section [ sec : conclusion ] .
consider the degraded gaussian diamond - wiretap channel in fig . [ fig : physical ] that consists of a source , two relays , a legitimate destination , and an eavesdropper .
the source is connected to two relays through orthogonal links of capacities @xmath1 and @xmath2 and there is no direct link from the source to the legitimate destination or eavesdropper .
the channel outputs @xmath3 and @xmath4 at the legitimate destination and the eavesdropper , respectively , are given as @xmath5 and @xmath6 , where @xmath7 , @xmath8 and @xmath9 are the channel inputs from relay 1 and relay 2 , respectively , @xmath10 is the gaussian noise with zero mean and unit variance at the legitimate destination , and @xmath11 is the gaussian noise with zero mean and variance of @xmath12 at the eavesdropper .
@xmath10 and @xmath11 are assumed to be independent .
the transmit power constraint at relay @xmath13 is given as @xmath14 , where @xmath15 denotes the number of channel uses . note that the channel output at the eavesdropper is a physically degraded version of the channel output at the legitimate destination .
we consider the following two scenarios regarding the availability of randomness . in the first scenario , a common fictitious message @xmath16 of rate @xmath0 ,
i.e. , @xmath17$]$ ] for two integers @xmath18 and @xmath19 denotes the set @xmath20 . ]
is available at the source and the two relays . in this case ,
a @xmath21 secrecy code consists of a message @xmath22 $ ] , an encoding function at the source that maps @xmath23\times [ 1:2^{nr'}]$ ] to @xmath24\times [ 1:2^{nc_2}]$ ] , an encoding function at relay @xmath13 that maps @xmath25\times [ 1:2^{nr'}]$ ] to @xmath26 , and a decoding function at the legitimate destination that maps @xmath27 to @xmath28 $ ] . in the second scenario , a fictitious message @xmath16 of rate
@xmath0 is available only at the source and the encoding at the two relays is restricted to be deterministic . in this case , the encoding function at relay @xmath13 maps @xmath29 $ ] to @xmath26 . for both scenarios , the probability of error
is given as @xmath30 .
a secrecy rate of @xmath31 is said to be achievable if there exists a sequence of @xmath32 codes such that @xmath33 and @xmath34 .
the secrecy capacity is the supremum of all achievable secrecy rates .
let @xmath35 and @xmath36 denote the secrecy capacity for the first scenario and for the second scenario , respectively . because the legitimate destination and the eavesdropper do not cooperate , the secrecy capacity in fig .
[ fig : physical ] is the same as that of stochastically degraded case in fig .
[ fig : stochastic ] , in which @xmath4 is given as @xmath37 , where @xmath38 has zero mean and unit variance and is independent of @xmath10 .
in this section , we present main results of this paper on the secrecy capacity of the degraded gaussian diamond - wiretap channel described in section [ sec : model ] . for the brevity of presentation ,
let us define the following functions : @xmath39 [ eqn : f_def ] where the domain of @xmath40 , @xmath41 , and @xmath42 is @xmath43 $ ] and that of @xmath44 and @xmath45 is @xmath46 $ ] for @xmath47
. becomes negative infinity when @xmath48 . ]
the following two theorems give upper and lower bounds on @xmath35 , respectively , whose proofs are in section [ sec : proof ] .
[ thm : common_ub ] for @xmath49 , @xmath35 is upper - bounded by @xmath50 where @xmath51 for @xmath52 .
we note that the functions @xmath53 s for @xmath54 $ ] are defined in ( [ eqn : f_def ] ) .
[ thm : common_lb ] for @xmath55 $ ] and @xmath56 , @xmath35 is lower - bounded by @xmath57 where @xmath58 we note that the functions @xmath53 s for @xmath54 $ ] are defined in ( [ eqn : f_def ] ) .
in theorem [ thm : common_ub ] , we note that the upper bound @xmath59 is the same as that in @xcite that assumes no secrecy constraint .
this is natural because the secrecy capacity is upper - bounded by the capacity without secrecy constraint , which is not affected by the common randomness at the encoders . to derive the upper bound @xmath60
, we generalize the bounding techniques @xcite and @xcite taking into account the secrecy constraint and the available randomness at the encoders .
in theorem [ thm : common_lb ] , @xmath61 is achieved by using a df scheme where the source sends the message to both relays and the relays cooperatively transmit the message and the common fictitious message over the wiretap channel . on the other hand ,
@xmath62 is achieved by a partial df incorporated with multicoding ( pdf - m ) where each relay sends an independent partial message and the common fictitious message using dependent codewords .
the source performs multicoding as follows : the message @xmath63 is represented as two partial messages @xmath64 , a codebook for relay @xmath13 consisting of independently generated @xmath65 sequences is constructed for each @xmath66 and @xmath67 , and the source finds a jointly typical sequence pair @xmath68 and sends @xmath69 to relay @xmath70 for @xmath13 . a more detailed explanation for the pdf - m scheme is given in section [ sec : proof ] .
let @xmath71 denote the partial df ( pdf ) rate without multicoding at the source . to compare our lower and upper bounds ,
let us consider sufficiently large @xmath0 and symmetric channel parameters , i.e. , @xmath72 and @xmath73 for some nonnegative @xmath74 and @xmath75 .
it can be easily proved that 1 ) the pdf scheme , which achieves , @xmath73 , and @xmath76 , @xmath77 and @xmath78 become redundant . ]
@xmath79 , is optimal for @xmath80 , i.e. , the bc cut is the bottleneck , and 2 ) the df scheme , which achieves @xmath81 , is optimal for @xmath82 , i.e. , the mac cut is the bottleneck .
when neither the bc cut nor the mac cut is the bottleneck , the pdf - m scheme strictly outperforms the pdf and df schemes for some range of @xmath75 as shown in fig .
[ fig : common ] . for example , when @xmath83 and @xmath84 , the pdf - m scheme strictly outperforms the pdf and df schemes for @xmath85 .
furthermore , fig .
[ fig : common ] shows that the pdf bound gets close to the upper bound in theorem [ thm : common_ub ] as @xmath74 increases . the following theorem states that the pdf scheme is indeed asymtotically optimal as @xmath86 or @xmath87 tends to infinity
, whose proof is relegated to the end of this section .
+ [ thm : asymptotic ] for the first scenario with @xmath88 and @xmath89 or @xmath90 , and @xmath2 are not necessarily fixed and can be arbitrary functions of @xmath86 and @xmath87 . ]
the pdf scheme is asymptotically optimal .
next , the following two theorems give upper and lower bounds on @xmath36 , respectively , whose proofs are in section [ sec : proof ] .
[ thm : no_ub ] for @xmath49 , @xmath36 is upper - bounded by @xmath91 where @xmath92 we note that the functions @xmath53 s for @xmath54 $ ] are defined in ( [ eqn : f_def ] ) , @xmath47 , and @xmath52 .
[ thm : no_lb ] for @xmath55 $ ] such that @xmath56 , @xmath36 is lower - bounded by @xmath93 where @xmath94 we note that the functions @xmath53 s for @xmath95 $ ] are defined in ( [ eqn : f_def ] )
. note that in both the upper and lower bounds for the first scenario , the term @xmath96 , which corresponds to the required rate of randomness to confuse the eavesdropper , appears only with @xmath97 , which signifies the amount of information sent through the mac .
in contrast , in both the upper and lower bounds for the second scenario , because the fictitious message has to be sent through the bc , @xmath96 appears in common for all terms .
this affects sufficient ranges of correlation coefficient for the lower bounds for large enough @xmath0 as remarked in the following .
[ remark : negative ] for large enough @xmath0 , sufficient ranges of correlation coefficient @xmath98 for the lower bounds in theorem [ thm : common_lb ] and theorem [ thm : no_lb ] are different . for the first scenario , note that the df rate is maximized at @xmath99 and that it is enough to consider nonnegative @xmath98 for the pdf - m scheme . on the other hand , for the second scenario , because the minus term @xmath100 is common for all terms , considering smaller @xmath98 can be beneficial by decreasing @xmath96 and we need consider all @xmath101 . in the df scheme for the second scenario , the source sends to both relays the fictitious message as well as the message .
hence , @xmath102 is obtained from @xmath103 by replacing @xmath1 and @xmath2 by @xmath104 and @xmath105 , respectively . for a partial df scheme incorporated with multicoding for the second scenario , a straightforward extension from that for the first scenario
is to let the source send the fictitious message @xmath67 as well as the partial message @xmath66 and the relay codeword index @xmath106 to relay @xmath70 for @xmath13 .
since each relay decodes a partial genuine message and a whole fictitious message , we call this scheme as pdf - df - m scheme . note that @xmath107 is obtained by replacing @xmath1 and @xmath2 by @xmath104 and @xmath105 , respectively , in @xmath108
. however , since the same fictitious message is sent to both relays , there exists inefficiency in the use of the bc . to resolve this inefficiency ,
we let each of relay codebooks be indexed by independent partial fictitious message , i.e. , codebook for relay @xmath13 is constructed for each @xmath109 by representing @xmath67 as two partial fictitious messages @xmath110 . by using this pdf - pdf - m scheme where each relay decodes a partial genuine message and a partial fictitious message ,
we show that @xmath111 is achievable , which has @xmath112 intead of @xmath113 in @xmath107 .
we note that having independent fictitious message at each relay reduces the achievable rate region over the mac , which results in additional contraints @xmath114 and @xmath115 in @xmath111 . nevertheless , as long as @xmath116
, @xmath111 is always higher than or equal to @xmath107 because @xmath117 , which should be satisfied if @xmath118 , implies @xmath114 and @xmath115 . if @xmath119 , @xmath107 can be strictly higher than @xmath111 as illustrated in fig .
[ fig : pdfdf ] .
let @xmath120 and @xmath121 denote the rates of pdf - df and pdf - pdf schemes ( without multicoding ) .
similarly as for the first scenario , let us consider sufficiently large @xmath0 and symmetric channel parameters .
since @xmath116 , we only consider the df , pdf - pdf - m , and pdf - pdf schemes for the lower bounds .
it can be easily proved that the df scheme , which achieves @xmath122}\min ( c , f_4(\rho))-f_5(\rho)$ ] , is optimal for @xmath123 , i.e. , the mac cut is the bottleneck .
we can see in fig .
[ fig : no ] that the pdf - pdf rate coincides with the pdf - pdf - m rate at one point .
this is because a negative correlation between the two relay signals is helpful for small @xmath75 due to the reason in remark [ remark : negative ] , i.e. , the bc cut is the bottleneck , and positive correlation becomes beneficial as @xmath75 increases , i.e. , the mac cut becomes bottleneck .
[ fig : no ] also shows that the pdf - pdf - m rate is zero up to some threshold value of @xmath75 due to the constraint @xmath124 in @xmath111 and the threshold value decreases as @xmath74 decreases .
indeed , we can prove that the threshold value tends to zero as @xmath74 tends to zero .
furthermore , fig .
[ fig : no ] shows that the pdf - pdf - m rate coincides with the upper bound in theorem [ thm : no_ub ] for some range of @xmath75 , e.g. , @xmath125 when @xmath126 and @xmath84 .
the following theorem gives a condition where the pdf - pdf - m rate coincides with the upper bound in theorems [ thm : no_ub ] , whose proof is relegated to the end of this section .
+ [ thm : cap_condition ] for the second scenario with sufficiently large @xmath0 and symmetric channel parameters , the pdf - pdf - m rate in theorem [ thm : no_lb ] coincide with the upper bound in theorem [ thm : no_ub ] , and the secrecy capacity is given as @xmath127 for @xmath128 such that that at least one of @xmath129 and @xmath130 is satisfied , where @xmath52 and @xmath131 $ ] is such that @xmath132 . ) , @xmath131 $ ] such that @xmath132 exists . ]
theorem [ thm : cap_condition ] indicates that the upper and lower bounds in theorems [ thm : no_ub ] and [ thm : no_lb ] coincide for @xmath125 when @xmath126 and @xmath84 and for @xmath133 when @xmath134 and @xmath84 .
for @xmath135 , the bounds in theorems [ thm : common_ub]-[thm : no_lb ] fall back to those in @xcite .
now , a natural question is how the presence of an eavesdropper affects the capacity .
we partially answer this question by comparing our results with the lower and upper bounds in @xcite that are derived without secrecy constraint .
note that when there is no secrecy constraint , the availability of randomness at the encoders does not affect the capacity .
hence , the capacity without secrecy constraint is higher than or equal to the secrecy capacity with secrecy constraint both for the first and the second scenarios .
we compare the bounds in fig .
[ fig : compare ] for sufficiently large @xmath0 and symmetric channel parameters . first , as illustrated in fig . [ fig : compare]-(a ) , the upper bound without secrecy constraint and the lower bound for the first scenario coincide up to @xmath136 .
this indicates that , when there is a sufficient amount of common randomness between the source and the relays , there is no decrease in capacity due to an eavesdropper for some range of @xmath75 . on the other hand , for the same channel parameters , fig .
[ fig : compare]-(b ) shows that the lower bound without secrecy constraint is strictly higher than the upper bound for the second scenario for all range of @xmath137 .
this indicates that , when there is no randomness at the relays , the secrecy capacity for the second scenario can be strictly smaller than the capacity without secrecy constraint for all range of @xmath75 .
+ the bound in theorem [ thm : common_ub ] is further upper - bounded as follows : @xmath138 where @xmath139 is because @xmath140 and @xmath141 are decreasing functions of @xmath142 $ ] .
furthermore , for any @xmath143 $ ] , we have @xmath144 hence , the secrecy capacity for the first scenario when @xmath89 or @xmath90 is asymtotically upper - bounded by @xmath145 which is clearly achievable by the pdf scheme .
let us first show that @xmath146 . for symmetric channel parameters , @xmath147 and @xmath148 can be rewritten as follows : @xmath149 let us define functions @xmath150 and @xmath151 of @xmath152 $ ] as follows : @xmath153 note that we can rewrite the condition in ( [ eqn : cap_condition ] ) as @xmath154 and @xmath155 .
since @xmath150 and @xmath151 are monotonically decreasing function and monotonically increasing function of @xmath152 $ ] , respectively , the condition in ( [ eqn : cap_condition ] ) implies that there exists @xmath131 $ ] such that @xmath156 .
hence , we have @xmath157 .
now , let us show @xmath158 .
since both @xmath159 and @xmath160 for @xmath161 $ ] have the maximum at @xmath162 , we have @xmath163 if @xmath129 or @xmath130 . hence , under the conditions in theorem
[ thm : cap_condition ] , the upper bound in theorem [ thm : no_ub ] becomes @xmath127 .
now , it remains to show @xmath127 is achievable .
we have @xmath164 where @xmath139 is because @xmath132 and @xmath165 imply @xmath166 .
this completes the proof .
in this section , we prove the upper and lower bounds on the secrecy capacity presented in section [ sec : main ] . we note that the upper bound @xmath59 , which the same as the upper bound in @xcite on the capacity without secrecy constraint , is easily obtained by noting that the secrecy capacity is upper - bounded by the capacity without secrecy constraint and that common randomness at the encoders does not affect the capacity when there is no secrecy constraint . nevertheless , we provide a direct proof for the upper bound @xmath167 as well as the upper bound @xmath168 since it can be useful for bounding in other related problems .
the proof generalizes those in @xcite and @xcite taking into account the secrecy constraint and the available randomness at the encoders . for @xmath169
$ ] and @xmath170 $ ] , let @xmath171 and let @xmath172 .
let @xmath173 $ ] and @xmath174 $ ] be such that @xmath175 and @xmath176 .
we use @xmath177 to denote a function of @xmath15 such that @xmath177 tends to zero as @xmath15 tends to infinity . by applying similar bounding techniques as in @xcite
, we have @xmath178 for sufficiently large @xmath15 , where @xmath139 is from the fano s inequality , @xmath179 is because @xmath180 and @xmath16 are independent , @xmath181 is because @xmath182 is a function of @xmath183 and @xmath16 , @xmath184 is because the gaussian distribution maximizes the differential entropy given the power constaint , @xmath185 is due to the concavity of the logarithm function , and @xmath186 is from the definition of @xmath187 .
similarly , we can obtain @xmath188 for sufficiently large @xmath15 .
we also have for sufficiently large @xmath15 , @xmath189 where @xmath139 is from the fano s inequality , @xmath179 is because @xmath180 and @xmath16 are independent , and @xmath181 is because @xmath182 and @xmath190 are functions of @xmath16 and @xmath180 and the markov relationship @xmath191 holds .
furthermore , for any random variable @xmath192 generated through a conditional pmf @xmath193 , we have @xmath194 by applying the above lower bound to ( [ eqn : fifth ] ) , we obtain @xmath195 for sufficiently large @xmath15 , we have @xmath196 where @xmath139 is from the secrecy constraint , @xmath179 is due to the fano s inequality , @xmath181 is due to the degradedness of the channel , and @xmath184 is from the entropy power inequality .
we note that ( [ eqn : f4 ] ) is a nondecreasing function of @xmath197 .
@xmath197 is further upper - bounded as follows : @xmath198 from the cauchy - schwarz inequality , we have @xmath199 similarly , we have @xmath200 .
hence , we obtain @xmath201 now we are ready to prove theorem [ thm : common_ub ] .
define @xmath202 $ ] and @xmath203 $ ] as follows .
first , @xmath204 is determined from @xmath205 .
@xmath206 if @xmath207 otherwise , @xmath204 is such that @xmath208 next , @xmath209 is determined from @xmath197 .
@xmath210 if @xmath211 otherwise , @xmath209 is such that @xmath212 let us first show that @xmath213 if @xmath206 , from ( [ eqn : f1 ] ) , ( [ eqn : f2 ] ) , ( [ eqn : fifth ] ) , ( [ eqn : fourth ] ) , and ( [ eqn : mu_zero ] ) , we have @xmath214 .
consider @xmath215 . from @xmath216 , ( [ eqn : mu_nonzero ] ) , and ( [ eqn : upper_min ] ) , we have @xmath217 then , from ( [ eqn : f1 ] ) , ( [ eqn : f2 ] ) , ( [ eqn : fifth ] ) , ( [ eqn : fourth ] ) , and ( [ eqn : mu_nonzero ] ) , we obtain @xmath218 . if @xmath204 further satisfies @xmath219 , we let @xmath220 , where @xmath221 is an i.i.d .
gaussian random variable with zero mean and variance of @xmath222 .
, @xmath223 is nonnegative . ] then , the mutual information terms in ( [ eqn : f3 ] ) are bounded as follows : @xmath224 where @xmath139 is from the conditional entropy power inequality . substituting the above bounds to ( [ eqn : f3 ] ) , we obtain @xmath225 .
hence , we have @xmath226 for @xmath219 .
this concludes the proof of ( [ eqn : thm1_first ] ) .
now , let us show @xmath227 if @xmath210 , from ( [ eqn : f1 ] ) , ( [ eqn : f2 ] ) , ( [ eqn : fifth ] ) , ( [ eqn : f4 ] ) , and ( [ eqn : nu_zero ] ) , we have @xmath228 . consider @xmath229 . from
( [ eqn : nu_nonzero ] ) and ( [ eqn : upper_min ] ) , we have @xmath230 then , from ( [ eqn : f1 ] ) , ( [ eqn : f2 ] ) , ( [ eqn : fifth ] ) , ( [ eqn : f4 ] ) , and ( [ eqn : nu_nonzero ] ) , we obtain @xmath231 . if @xmath209 further satisfies @xmath232 , we consider the following bound by adding the inequalities ( [ eqn : fourth ] ) and ( [ eqn : f3 ] ) : @xmath233 where @xmath139 holds when @xmath234 .
we let @xmath235 , where @xmath236 is an i.i.d .
gaussian random variable with zero mean and variance of @xmath237 .
then , by substituting ( [ eqn : nu_nonzero ] ) and similar bounds as in ( [ eqn : gamma1])-([eqn : gamma3 ] ) to ( [ eqn : f3f4 ] ) , we obtain @xmath238 . hence , we have @xmath239 , @xmath240 for @xmath232 .
this concludes the proof of ( [ eqn : thm1_second ] ) .
let us first assume that the channel from the relays to the legitimate destination and the eavesdropper is a discrete memoryless channel with a conditional pmf @xmath241 .
fix @xmath242 and let @xmath243 fix @xmath244 .
we use @xmath245 to denote a function of @xmath246 such that @xmath245 tends to zero as @xmath246 tends to zero . in the df scheme
, the source sends the message to both relays , which requires @xmath247 .
once the relays share both the message and the fictitious message , we can treat the channel from the relays to the legitimate destination and the eavesdropper as a classical wiretap channel @xcite , @xcite with randomness of rate @xmath0 in ( [ eqn : ach_fictitious ] ) and hence the secrecy rate of @xmath248 is achievable . by combining two inequalities for @xmath31 , we conclude the following secrecy rate is achievable : @xmath249 the pdf - m scheme is described in the following .
* codebook generation : we represent the message @xmath250 $ ] as the partial message pair @xmath251\times [ 1:2^{nr_2}]$ ] for some @xmath252 and @xmath253 such that @xmath254 i.e. , @xmath255 for @xmath256 $ ] is uniformly distributed over @xmath257 $ ] and @xmath258 and @xmath259 are independent .
consider @xmath260 for @xmath256 $ ] .
for each @xmath169 $ ] and @xmath261\times [ 1:2^{nr'}]\times [ 1:2^{n\tilde{r}_k}]$ ] , generate @xmath262 independently according to @xmath263 . *
encoding at the source : for message @xmath264 and fictitious message @xmath67 , the source finds an @xmath265 such that @xmath266 for @xmath256 $ ] , the source sends @xmath69 to relay @xmath70 . * encoding at relay @xmath256 $ ] : note that fictitious message @xmath67 is given at relay @xmath70 . after receiving @xmath69 from the source
, relay @xmath70 sends @xmath262 . * decoding at the legitimate destination : the legitimate destination finds @xmath267 such that @xmath268 the legitimate destination declares that @xmath269 is the message .
* error analysis : from the mutual covering lemma @xcite , the encoding error at the source averaged over the codebooks tends to zero as @xmath15 tends to infinity if @xmath270 for @xmath256 $ ] , the transmission of @xmath69 from the source to relay @xmath70 requires @xmath271 from the standard error analysis , the decoding error at the legitimate destination averaged over the codebooks tends to zero as @xmath15 tends to infinity if @xmath272 * secrecy analysis : we can show @xmath273 if ( [ eqn : ach_fictitious ] ) and the following inequalities are satisfied .
@xmath274 see section [ sub : secrecy_analysis ] for the detail .
therefore , there exists a sequence of codes such that @xmath275 tends to zero and @xmath276 as @xmath15 tends to infinity if ( [ eqn : ach_fictitious ] ) , ( [ eqn : ach_eq1])-([eqn : ach_sec3 ] ) are satisfied . by performing fourier - mozkin elimination to ( [ eqn : ach_fictitious ] ) , ( [ eqn : ach_eq1])-([eqn : ach_sec3 ] ) and by taking @xmath277 ,
the pdf - m rate of @xmath278 is obtained . from the standard discretization procedure @xcite , @xmath61 and @xmath62
are obtained by evaluating ( [ eqn : dfd_1 ] ) and ( [ eqn : pdfmd_1 ] ) for the degraded gaussian diamond - wiretap channel discussed in section [ sec : model ] and a jointly gaussian distribution @xmath242 such that @xmath279 for @xmath256 $ ] has zero mean and variance of @xmath280 and the correlation coefficient between @xmath8 and @xmath9 is @xmath55 $ ] .
we note that the upper bound ( [ eqn : f4_pre ] ) continues to hold when the fictitious message is given only at the source .
then , we have @xmath281 where @xmath187 is defined in the proof of theorem [ thm : common_ub ] , @xmath139 is because @xmath182 is a function of @xmath183 , and @xmath179 is from some similar steps as in the derivation of ( [ eqn : f1 ] ) .
similarly , we can obtain @xmath282 where @xmath283 is defined in the proof of theorem [ thm : common_ub ] . for any random variable @xmath192 generated through a conditional pmf @xmath193
, we have @xmath284 where @xmath139 is because @xmath285 is a function of @xmath286 for @xmath256 $ ] and @xmath179 is from some similar steps as in the derivation of ( [ eqn : intro_u ] ) .
note that we have the following lower and upper bounds on @xmath287 : @xmath288 hence , there exists @xmath289 $ ] such that @xmath290 then , we have the following lower bound on @xmath291 : @xmath292 from the entropy power inequality . now , we are ready to prove theorem [ thm : no_ub ] .
first consider @xmath293 .
then , from ( [ eqn : f4_sec])-([eqn : four ] ) , ( [ eqn : rho_def ] ) , and ( [ eqn : secrety_minus ] ) , we have @xmath294 .
next , consider @xmath143 $ ] .
then , due to similar reasons as in the proof of theorem [ thm : common_ub ] , we have @xmath295 . then , from ( [ eqn : f4_sec])-([eqn : four ] ) , ( [ eqn : rho_def ] ) , and ( [ eqn : secrety_minus ] ) , we have @xmath296 . now , assume that @xmath98 further satisfies @xmath297.$ ] we choose @xmath298 , where @xmath299 is an i.i.d .
guassian random variable with zero mean and variance of @xmath300 .
then , by substituting ( [ eqn : secrety_minus ] ) and similar bounds as ( [ eqn : gamma1])-([eqn : gamma3 ] ) to ( [ eqn : f4f5 ] ) , we obtain @xmath301 . hence , we have @xmath302 for @xmath297 $ ] .
this concludes the proof of theorem [ thm : no_ub ] . as in the proof of theorem [ thm : common_lb ] , we first assume that the channel from the relays to the legitimate destination and the eavesdropper is a discrete memoryless channel with a conditional pmf @xmath241 .
fix @xmath242 and @xmath244 .
let @xmath303 for the df scheme , by letting the source send both the message and the fictitious message to the relays , an achievable secrecy rate of @xmath304 is obtained from ( [ eqn : dfd_1 ] ) by replacing @xmath1 and @xmath2 by @xmath305 and @xmath306 , respectively .
similarly , for the pdf - df - m scheme , by letting the source send the fictitious message as well as the partial message and the relay codeword index to relay @xmath70 for @xmath13 in the pdf - m scheme for the first scenario , an achievable secrecy rate of @xmath307 is obtained from ( [ eqn : pdfmd_1 ] ) by replacing @xmath1 and @xmath2 by @xmath305 and @xmath306 , respectively .
the pdf - pdf - m scheme is described in the following . * codebook generation : we represent the message @xmath250 $ ] and the fictitious message @xmath308 $ ] as a partial message pair @xmath251\times [ 1:2^{nr_2}]$ ] and a partial fictitious message pair @xmath309\times [ 1:2^{nr_2'}]$ ] , respectively , for some nonnegative rates @xmath310 , and @xmath311 such that @xmath312 consider @xmath260 for @xmath256 $ ] . for each
@xmath169 $ ] and @xmath313\times [ 1:2^{nr'_k } ] \times [ 1:2^{n\tilde{r}_k}]$ ] , generate @xmath314 independently according to @xmath263 . *
encoding at the source : for message @xmath264 and fictitious message @xmath110 , the source finds an @xmath265 such that @xmath315 for @xmath256 $ ] , the source sends @xmath316 to relay @xmath70 . * encoding at relay @xmath256 $ ] : after receiving @xmath316 from the source , relay @xmath70 sends @xmath314 . * decoding at the legitimate destination : the legitimate destination finds @xmath317 such that @xmath318 the legitimate destination declares @xmath269 is the message . *
error analysis : from the mutual covering lemma , the encoding error at the source averaged over the codebooks tends to zero as @xmath15 tends to infinity if @xmath270 for @xmath256 $ ] , the transmission of @xmath316 from the source to relay @xmath70 requires @xmath319 from the standard error analysis , the decoding error at the legitimate destination averaged over the codebooks tends to zero as @xmath15 tends to infinity if @xmath320 * secrecy analysis : we can show @xmath273 if ( [ eqn : ach_fictitious2 ] ) and the following inequalities are satisfied .
@xmath321 see section [ sub : secrecy_analysis ] for the detail . therefore , there exists a sequence of codes such that @xmath275 tends to zero and @xmath276 as @xmath15 tends to infinity if ( [ eqn : ach_fictitious2 ] ) , ( [ eqn : ach_eq2])-([eqn : ach_sec3_no ] ) are satisfied . by performing fourier - mozkin elimination to ( [ eqn : ach_fictitious2 ] ) , ( [ eqn : ach_eq2])-([eqn : ach_sec3_no ] ) and by taking @xmath277 , a secrecy rate of @xmath322 subject to the constraints @xmath323 is obtained . from the standard discretization procedure , @xmath324 , @xmath107 , and @xmath111
are obtained by evaluating ( [ eqn : dfd_2 ] ) , ( [ eqn : pdfmd_2 ] ) , ( [ eqn : pdfmd2 ] ) , and ( [ eqn : pdfmd2_constr ] ) for the degraded gaussian diamond - wiretap channel discussed in section [ sec : model ] and a jointly gaussian distribution @xmath242 such that @xmath279 for @xmath256 $ ] has zero mean and variance of @xmath280 and the correlation coefficient between @xmath8 and @xmath9 is @xmath55 $ ] .
let @xmath325 denote the random codebook . for message @xmath180 , fictitious message @xmath16 , and chosen relay codeword indices @xmath326 , we have @xmath327 for sufficiently lage @xmath15 , where @xmath139 is because the eavesdropper who already knows @xmath180 and @xmath328 can decode @xmath16 and @xmath329 with high probability when ( [ eqn : ach_sec1])-([eqn : ach_sec3 ] ) are satisfied for the first scenario and when ( [ eqn : ach_sec1_no])-([eqn : ach_sec3_no ] ) are satisfied for the second scenario .
hence , we have @xmath273 .
in this paper , we derived nontrivial upper and lower bounds on the secrecy capacity of the degraded gaussian diamond - wiretap channel under two scenarios regarding the availability of randomness .
our upper bound was obtained by taking into account the correlation between the two relay signals and the availability of randomness at each encoder , which generalizes both the upper bound on the capacity of the diamond channel without secrecy constraint @xcite and the upper bound on the sum secrecy capacity of the mac wiretap channel @xcite . for the lower bound , we proposed df scheme and partial df scheme incorporated with multicoding that is called pdf - m scheme for the first scenario and pdf - df - m and pdf - pdf - m schemes for the second scenario depending on whether the relay decodes the whole or partial fictitious message . in the first scenario , pdf - m scheme with strictly positive correlation coefficient
was shown to outperform df and pdf ( without multicoding ) schemes for some channel parameters .
we also showed that the pdf scheme is asymptotically optimal for the first scenario when at least one of relay power constraint tends to infinity .
for the second scenario , we presented a condition for channel parameters where the pdf - pdf - m scheme is optimal . furthermore , because the fictitious message has to be sent through the bc for the second scenario , it was shown to be befinicial to consider negative correlation in all df , pdf - df - m , pdf - pdf - m schemes when the bc cut becomes the bottleneck .
furthermore , we investigated the effect of the presence of an eavesdropper on the capacity . if there is a sufficient amount of common randomness between the source and the relays , it was shown that there is no decrease in capacity due to an eavesdropper for some range of @xmath75 . as a final remark
, it seems to be straightforward to combine our df scheme and partial df scheme incorporated with multicoding by using superposition coding , but the resultant rate expression would be rather complicated with less useful insights .
d. kobayashi , h. yamamoto , and t. ogawa , `` how to attain the ordinary channel capacity securely in wiretap channels , '' in _ proc .
ieee information theory workshop on theory and practice in information - theoretic security _
2005 , pp . 1318 .
r. a. chou and m. r. bloch , `` uniform distributed source coding for the multiple access wiretap channel , '' in _ proc .
ieee conference on communications and network security ( cns ) _ , oct .
2014 , pp . 127132 . | in this paper , we present nontrivial upper and lower bounds on the secrecy capacity of the degraded gaussian diamond - wiretap channel and identify several ranges of channel parameters where these bounds coincide with useful intuitions .
furthermore , we investigate the effect of the presence of an eavesdropper on the capacity .
we consider the following two scenarios regarding the availability of randomness : 1 ) a common randomness is available at the source and the two relays and 2 ) a randomness is available only at the source and there is no available randomness at the relays .
we obtain the upper bound by taking into account the correlation between the two relay signals and the availability of randomness at each encoder . for the lower bound
, we propose two types of coding schemes : 1 ) a decode - and - forward scheme where the relays cooperatively transmit the message and the fictitious message and 2 ) a partial df scheme incorporated with multicoding in which each relay sends an independent partial message and the whole or partial fictitious message using dependent codewords . wiretap channel , diamond channel , diamond - wiretap channel , multicoding |
it is well known that the lyman alpha forest shows very little gunn - peterson absorption along lines of sight to distant quasars .
the standard interpretation of these observations is that the intergalactic medium ( igm ) becomes highly ionized before redshift @xmath1 .
the approximate reionization redshift is among the most important unknown quantities relevant to large scale structure and cosmology .
there have been many attempts to simulate the reionization process , but they are strongly dependent on the input physics and are hence unreliable .
in fact , it is still quite plausible to have reionization occur anywhere between redshifts @xmath1 and @xmath2 in most reasonable cosmological models .
it is observations that constrain the reionization epoch to this range : the lack of the compton @xmath3-distortion of the cmb excludes @xmath4 and the lack of gunn - peterson absorption excludes @xmath5 . because of the uncertainty in the reionization epoch
, it is important to be able to observe the reionization event in some way .
we propose a new signal in the cosmic infrared background ( cib ) that is potentially capable of giving the redshift of reionization .
while the signal is small relative to the current limits at 1 - 10 @xmath6 m , the predicted frequency dependence makes detection possible .
the rate of hydrogen recombinations is peaked at the reionization epoch , and this is the source of the background we seek .
the fraction of neutral hydrogen undergoes a drop of a factor of at least 10@xmath7 in about a tenth of a hubble time , thus the gunn - peterson optical depth undergoes a drastic change in a very small redshift interval .
this abrupt change in absorption can give an accurate value for the redshift of reionization . because we observe an all - sky background , we are only concerned with the effects absorption has on the spectrum .
it serves to brighten some lines while dimming others .
because the effect turns off very rapidly , sharp features are produced in the spectrum .
in order to be specific , we adopt a cdm+@xmath0 cosmological model as a framework for our investigations .
we fix the cosmological parameters as follows : @xmath8 , @xmath9 , and @xmath10 .
we study two values of the baryon density parameter , @xmath11 and 0.055 . we allow for a small tilt in the primordial spectrum of @xmath12 , and we normalize the model to cobe according to bunn & white ( 1997 ) . in order to accurately determine the initial conditions ,
we compute the linear transfer functions for the model using the linear boltzmann code linger from the cosmics package ( bertschinger 1995 ) .
the simulations were performed with the slh - p@xmath7 m cosmological hydrodynamic code ( gnedin 1995 ; gnedin & bertschinger 1996 ) .
the physical modeling incorporated in the code is described in gnedin & ostriker ( 1997 ) , and includes dynamics of the dark matter and cosmic gas , evolution of the spatially averaged uv- and x - ray background radiation , star formation , stellar feedback , non - equilibrium ionization and thermal evolution of primeval plasma , molecular hydrogen chemistry , equilibrium metal cooling , and self - shielding of the gas .
we have performed six runs with different box sizes and numerical resolutions to assess the importance of different scales and estimate the uncertainty due to the finite resolution of our simulations as well as to test different physical assumptions included in the simulations .
the adopted parameters of these runs are compiled in table [ tabnumpar ] .
the first four runs were stopped at @xmath13 , since at that moment the rms density fluctuation at the box scale reaches 0.3 - 0.4 , and the absence of fluctuations with wavelengths larger than the box size renders the simulations unreliable .
the two runs e and f were continued until @xmath14 , and are used to assess the evolution of the recombination rate at low redshifts . because of their low resolution , they can not be used for calculating the reionization history of the universe .
the largest of our simulations , run a , has the highest resolution , and is our fiducial run .
its spatial and mass resolutions are sufficient to accurately predict the epoch of reionization and the evolution of the recombination rate throughout the reionization period
. two smaller runs , b and c , differ by the shape of the spectrum emitted by the sources of ionization .
specifically , run c includes only young stars as sources of ionization , whereas run b includes quasars in the ionizing radiation .
finally , run d has a lower value of the baryon density .
this run is discussed in detail in gnedin & ostriker ( 1997 ) .
the space averaged equation of radiative transfer in an expanding universe is @xmath15 where @xmath16 .
we need the intensity of a sharp line at frequency @xmath17 as a function of its redshift .
the source for an infinitely sharp line is given by @xmath18 where @xmath19 is the volume rate of photon emission and @xmath20 is the dirac delta function .
neglecting absorption , the intensity today in a flat @xmath21 universe is given by @xmath22 ^{-1/2}r(t_i ) .
\label{jnu}\ ] ] where @xmath23 is defined by @xmath24 and @xmath25 is the scale factor normalized to unity today .
the case a recombination rate is computed from the simulation data by obtaining the average @xmath26 over the simulation box .
the fit to @xmath27 from hui and gnedin ( 1997 ) was used .
this quantity drops steadily before reionization as the universe expands .
there is a rise at the reionization epoch as the hydrogen becomes nearly completely ionized .
after reionization , the rate declines steadily until the end of the simulation . from the case a recombination rate we need to obtain the direct recombination rates to the individual levels .
we use the rates in ferland et al .
( 1992 ) for the mean temperature of the gas over the simulation box , which is about 10@xmath28 k. for levels @xmath29 we have used the following fit which reproduces the summed values well at least to @xmath30 .
@xmath31 these three values correspond to the three summed recombination rates at 10@xmath28 k in ferland et al .
( 1992 ) for @xmath32 , @xmath33 , and @xmath34 .
these values reproduce the @xmath35 recombination coefficient well , but the @xmath35 rate was not used in making the fit . at large @xmath36 this fit becomes negative , but this does not happen until @xmath37 or so .
we use the approximation given by johnson ( 1971 ) for the radiative transition rates in hydrogen . knowing the rates of direct recombinations to various levels
, we can then compute the strengths of lines .
we find that the emission is dominated by the direct recombination photons .
we have used a simple approximation to account for the absorption . because we are investigating a diffuse signal , we are only concerned with how energy is redistributed among different spectral lines .
we assume that the optical depth is very large before reionization and very small afterwards .
this is a reasonable approximation because the simulations indicate that the neutral hydrogen fraction drops by three orders of magnitude in about a tenth of a hubble time .
we refer to this as the reionization event .
the justification of our approximation is as follows .
the gunn - peterson optical depth is given by @xmath38 .
the current observational limit indicates that @xmath39 , giving @xmath40 .
we neglect this relatively small optical depth . before reionization however , the neutral fraction is a factor of at least 10@xmath41 higher , possibly as much as a factor of @xmath42 higher , giving a gunn - peterson optical depth of at least @xmath43 , and probably 10@xmath28 .
the other lines in the lyman series will have smaller optical depths due to their smaller cross sections , but this effect does not refute this argument .
for example , the optical depth of ly-7 is only decreased by a factor of 100 .
this justifies our approximation .
we assume that any transition that is not in the lyman series is optically thin .
there are a large number of neutral hydrogen atoms for each photon , so the hydrogens will be virtually all in the ground state . each time
a lyman series transition is excited , there is some chance that it will decay through other states
. these photons can escape unless they are also in the lyman series .
the net effect of this is that photons are removed from the lyman series and broken up into lower energy photons and ly@xmath44 .
we now estimate how many times this process occurs .
the probability to remain in the lyman series is simply @xmath45 , and @xmath46 , where @xmath47 is the probability of remaining after one interaction .
the branching ratios for the @xmath48 transitions are all roughly 95% , with the @xmath49 transitions making up most of the remaining 5% , thus @xmath50 , giving @xmath51 . through the reionization event
, @xmath52 drops from about @xmath53 to about 1 , so previous to reionization , lyman series photons suffer as many as @xmath42 absorptions .
we approximate this process by recomputing the branching ratios , taking into account the fact that radiative transitions in the lyman series are forbidden , except for ly@xmath44 , where there is no other possibility .
these modified branching ratios are used until the reionization event .
after even 100 such decays , the probability that every single one was a lyman series decay is about @xmath54 .
our approximation for the branching ratios is thus justified .
we find that ly@xmath44 becomes about three times brighter and that h@xmath44 becomes about thirty times brighter .
we neglect opacity due to dust .
dust is not associated with the highly ionized gas we are studying , and in any event it is unlikely that there is any significant dust formation at the high redshifts we are concerned with .
for example , the average metallicities in the lyman - alpha forest at redshift @xmath55 are not much in excess of 0.01 solar ( songaila & cowie 1996 ) .
more recent dust formation may obscure a cib , but we expect that the dust covers only a small fraction of the sky , so we neglect dust absorption in our calculation . in any event ,
dust absorption can not erase a spectral feature , though it may decrease the total amplitude .
we have attempted to extrapolate the emission rate to @xmath14 in several cases . as a simple exercise , we first neglect temperature evolution , thus fixing @xmath44 . if the bulk of the emission comes from bound objects , then @xmath56 . if the emission is coming from growing mode perturbations in the linear regime , then @xmath57 . finally , if the emission is coming from a uniform background , then @xmath58 .
we use runs e and f to investigate the low redshift evolution . at late times , after reionization , the case a recombination rate varied as roughly @xmath59 in these simulations .
the temperature in the predominant emission regions is thus rising .
we use this result in extrapolating the higher resolution simulations to @xmath14 .
the emission will be suppressed in the highest density regions due to the greatly increased temperature and thus greatly decreased recombination rate .
we can make a simple argument to motivate this scale factor dependence .
the overall temperature of the igm is around 10@xmath28 k. the recombination coefficient varies roughly as @xmath60 at temperatures of 10@xmath61 k. the temperature in a cluster can be 10@xmath62 or 10@xmath63 k. if a cluster is overdense by a factor of 10@xmath7 , the square of the density is up by 10@xmath61 .
the temperature is up by a factor of roughly 10@xmath7 , and the total volume of gas is down by a factor of @xmath64 .
crudely , the cluster produces as much emission as the volume of igm from which it formed .
if we assume unresolved clusters dominate the emission , we can then assign a power law evolution to the mean temperature of the igm today , which is as large as 10@xmath63 k. at redshift @xmath65 in the simulations , the temperature is 10@xmath28 k and today it is 10@xmath63 k , this means that the temperature rises like @xmath66 , so @xmath67 , and for clusters , which are bound objects , the total rate varies as @xmath68 .
there are two quantities that we track at the reionization epoch , the space averaged recombination rate and the neutral hydrogen fraction . the hydrogen becomes ionized in roughly a tenth of a hubble time at a redshift of about @xmath69 , depending on the model .
the recombination rate is at a maximum at this time , but the peak is quite broad . to illustrate this point ,
we have plotted the recombination rate and the neutral hydrogen fraction together in figure 1 for simulation a. the other simulations have similar behavior . in figure 2 we plot @xmath70 from recombinant emission from the first 20 levels of hydrogen for simulations a - d .
the recombination rate has been extrapolated in the high resolution simulations from @xmath71 to @xmath14 according to a power law , @xmath72 where we have used @xmath73 , the fit obtained from simulations e and f. inspecting equation ( [ jnu ] ) we see that @xmath19 must fall more quickly than about @xmath74 at late times in the flat model being considered .
if @xmath19 does not fall quickly enough , the spectrum is dominated by the most recent emission , and any feature from the reionization epoch is obscured .
it is the absorption that gives the best indication of the redshift of the reionization event . before reionization
, the optical depth in the lyman series is very high .
all lyman series lines except ly@xmath44 are absorbed immediately and redistributed .
this makes ly@xmath44 and the balmer series significantly brighter because they receive the energy of the lyman series .
once reionization occurs , the lyman series becomes optically thin and the brightness of ly@xmath44 and the balmer series drops considerably .
this is clearly seen in the spectrum .
as we look redward , crossing ly@xmath44 at about one micron or h@xmath44 at about six microns from the reionization event , the brightness suddenly increases due to the effect of the absorption and redistribution of the higher lyman series lines .
these features give a measurement of the redshift of the reionization event .
the features are reasonably sharp , with @xmath75 .
haiman , rees & loeb ( 1997 ) estimate a similar effect whereby radiative excitations serve to brighten ly@xmath44 .
we find that recombinations give a significantly brighter signal .
the recombination rate is proportional to the square of the gas density , and thus is sensitive to the clumpiness of the gas , while the excitation rate scales linearly with the gas density . in a hierarchically clustered universe
we can expect that the recombination rate should be larger than the excitation rate .
the positions of the broad maxima from the lyman and balmer series provide a check on the redshift of the sharp absorption features .
these can also be used to relate the redshift of the reionization event to the evolution of the recombination rate .
these can provide valuable information about the evolution of the ionizing background and the temperature evolution at the reionization epoch .
these effects will be investigated in a future paper ( baltz , gnedin , & silk 1997 ) .
this signature is quite weak , though it may be feasible to see some of these features in the near infrared , where dust emission is lowest . estimating the broadband intensity @xmath76 for spectra a - d at the position of the ly@xmath44 feature , we get results of @xmath77 nw sr@xmath78 m@xmath79 . the signal is currently below observational upper limits by a factor of @xmath80 ( kashlinsky et al .
1996b ) , but it may not be completely obscured according to theoretical models of the correlated component of the diffuse light from young galaxies ( kashlinsky et al .
1996b , veeraraghavan 1996 ) .
all of these predict that the broadband diffuse infrared light should be of the order of @xmath81 nw sr@xmath78 m@xmath79 .
the present model also predicts correlations in the diffuse light . in principle
measuring these correlations can distinguish between reionization by stars and reionization by quasars , as the fluctuation length in the quasar scenario is much longer .
there are two avenues which seem to greatly increase our chances of detecting these spectral features .
the first is a spaceborne mission such as the egbirt or desire proposals ( mather & beichmann 1996 ) , which plan to put an infrared telescope at 3 au where the interplanetary dust scattering and emission is reduced by a factor of at least 100 .
since such dust is the primary foreground , such a mission would be quite fruitful .
the second involves measuring the anisotropy of the cib .
it has been shown that while foregrounds are far brighter than any expected extragalactic infrared background , the extragalactic sources contribute most of the fluctuations at the degree scale down to the arcminute scale ( kashlinsky , mather & odenwald 1996a , kashlinsky et al .
if it were possible to measure fluctuations on scales smaller than this , the recombinant emission should dominate due to its high redshift .
in fact , the angular size of the simulation boxes at the reionization events are all on the order of a few seconds of arc .
for example , in runs b and c the reionization event occurs at redshift @xmath82 .
the comoving nonlinear scale at this time is 100 @xmath83 kpc .
this region subtends @xmath84 today .
any fluctuations at this scale should be produced at the reionization epoch , long before the arcminute scale fluctuations characteristic of galaxy formation .
the spectra we have presented are only meant to be representative of what might be observed .
there are large theoretical uncertainties concerning the reionization epoch , and we hope a measurement like the one described in this work can help to sort out the details of this important event in the history of structure formation .
e. b. was supported by nasa grant 1 - 443839 - 23254 - 2 .
n. g. was supported by the uc berkeley grant 1 - 443839 - 07427 .
simulations were performed on the ncsa power challenge array under the grant ast-960015n and on the ncsa origin2000 mini super - computer under the grant ast-970006n . | emission lines in hydrogen can be used to measure the approximate redshift of the reionization of the universe .
this is an important measurement given the lack of a convincing theoretical prediction of this epoch .
there is a rapid change in absorption at the reionization event , which can leave sharp features in the spectrum of recombining hydrogen .
these features in the near infrared background permit us to measure the redshift at which reionization occurs .
we have calculated these signals based on numerical simulations of the @xmath0cdm cosmogony .
# 1 |
shell - type supernova remnants ( snr ) have long been considered as prime candidates for accelerating cosmic rays up to at least 100 tev , but until recently , this statement was only supported by indirect evidence , namely non - thermal x - ray emission interpreted as synchrotron radiation from very - high - energy ( vhe ) electrons in a few objects @xcite . a more direct proof is provided by the emission of high - energy gamma - rays produced either by inverse compton ( ic ) scattering of high - energy electrons on ambient photons or by nuclear interactions of high - energy protons or ions in the interstellar medium and subsequent @xmath2 meson decays . however , in the @xmath3 energy range , the energetic gamma - ray experiment telescope ( egret ) onboard the compton gamma - ray observatory could not provide an unambiguous detection of a shell - type snr , due to its poor angular resolution and to the difficulty of separating signals of extended objects from the diffuse galactic gamma - ray background . in the very - high - energy range ( @xmath4 200 gev ) on the other hand , the situation is more favorable @xcite : recent imaging atmospheric cherenkov telescopes have achieved angular resolutions of the order of a few arc minutes and the diffuse background is expected to decrease more rapidly with energy than the gamma - ray signal .
the first confirmed gamma - ray signal from a shell - type snr was that of rx j1713.7 - 3946 detected by the cangaroo - i and cangaroo - ii experiments @xcite as well as by the h.e.s.s .
collaboration @xcite .
the latter experiment provided the first detailed morphological and spectral study of this source @xcite .
a second shell - type supernova remnant , rx j0852.0 - 4622 ( also named g266.2 - 1.2 ) , was recently detected by cherenkov telescopes : the announcement of a signal from the north - western part by the cangaroo collaboration @xcite was shortly followed by the publication of a complete gamma - ray map of this object by the h.e.s.s .
collaboration @xcite , making it the largest extended source ( 2@xmath5 angular diameter ) ever resolved by a cherenkov telescope .
this previous h.e.s.s .
result was obtained from a short observation campaign ( 3.2 hours ) in 2004 . in this article , we present the results of much longer observations of this source in 2005 ( @xmath6 20 hours ) with the full h.e.s.s . array .
+ rx j0852.0 - 4622 is located in the south - eastern corner of the vela snr and its study in x - rays as well as in radio is complicated by the superposition of the highly structured emission of the vela remnant .
its discovery in the rosat all - sky survey @xcite relied on the restriction to energies greater than 1.3 kev where the signal stands out above the soft thermal emission from the vela snr . in x - rays , rx j0852.0 - 4622 appears as a roughly circular disk with a diameter of 2@xmath5 with a brightening towards the north - western , western and south - eastern rims . since its discovery
, its distance and age have been a matter of controversy .
low values of these quantities have been inferred from the detection by comptel @xcite of the 1.157 mev gamma - ray line of @xmath7ca due to the decay chain @xmath7ti@xmath8sc@xmath8ca characterized by the @xmath7ti lifetime , whose average value , based on measurements by independent groups , is @xmath9 years @xcite .
on the basis of the @xmath7ti yield and of the x - ray diameter , an age of @xmath6 680 yr and a distance of @xmath6 200 pc , thus close to that of the vela remnant , was proposed @xcite .
it should be noted that this result was obtained by adopting a mean expansion velocity of 5000 km s@xmath10 based on the assumption of a purely thermal x - ray spectrum .
however , further observations of the brightest parts of the shell by asca @xcite , xmm - newton @xcite and chandra @xcite demonstrated the non - thermal nature of the x - ray emission . in this framework , different models interpreting the x - ray spectrum yield absorbing column densities for rx j0852.0 - 4622 significantly larger than that of the vela snr .
moreover , the significance level of the @xmath7ti yield was later questioned in a reanalysis of comptel data @xcite .
the sc - k emission at about 4 kev expected from the @xmath7ti yield is also controversial : evidence for this line was first claimed from asca sis0 data @xcite , but not confirmed by sis1 data from which only an upper limit could be deduced @xcite , whereas a detection at the 4 @xmath11 level was obtained from xmm - newton data @xcite .
the doubt on the detection of the @xmath7ca and @xmath7sc lines thus affects the interpretation of rx j0852.0 - 4622 as a young and close - by supernova remnant .
+ in contrast , on the basis of the absorbing column density deduced from the x - ray spectrum , @xcite argue in favor of a distance much larger than 200 pc , with the restriction that the remnant be in front of the vela molecular ridge .
otherwise , this concentration of giant molecular clouds , revealed by co data and located at a distance of @xmath6 12 kpc , should produce significant absorption in x - rays in the eastern rim of rx j0852.0 - 4622 at a level which is not observed .
new estimates of the age and distance of this source were recently proposed by @xcite , on the basis of the observation of very thin hard x - ray filaments in the north - western edge with the high angular resolution of the chandra satellite . using an empirical relation @xcite between the filament width on the downstream side of the shock @xmath12 , the roll - off frequency @xmath13 of the synchrotron spectrum and the snr age
, the authors derive an age in the range of @xmath14 yr and a distance of @xmath15 kpc .
+ the possible presence of a compact remnant of the supernova explosion at the center of rx j0852.0 - 4622 , first suggested from rosat observations @xcite , was confirmed by beppo - sax @xcite and chandra @xcite ; if this object is interpreted as a neutron star , as proposed by @xcite , rx j0852.0 - 4622 would be the remnant of a core - collapse supernova .
the absorbing column density obtained from the spectrum of this central object is also significantly higher @xcite than those measured for the vela remnant , supporting larger distances as suggested by @xcite .
however , @xcite recently interpreted the radio counterpart of the central object as due to a planetary nebula ; therefore , rx j0852.0 - 4622 may also be the result of a thermonuclear explosion .
+ to summarize , there remains a large uncertainty on the distance of rx j0852.0 - 4622 ; this object could be as close as the vela snr ( @xmath6 290 pc ) and possibly in interaction with vela , or as far as the vela molecular ridge ( @xmath6 1 kpc ) .
even the nature of the explosion of the progenitor remains unclear .
in addition , the superposition of the vela snr and rx j0852.0 - 4622 makes the interpretation of radio and x - ray data difficult . on the other hand , due to its very weak radio - emission @xcite and to the non - thermal nature of its x - ray spectrum , rx
j0852.0 - 4622 shows remarkable similarities with rx j1713 - 3946 , also detected in the very - high - energy range .
+ this article is organized as follows . in section 2 , the main characteristics of the h.e.s.s .
telescope array are reviewed and the rx j0852.0 - 4622 data set is described .
section 3 is devoted to the analysis method ( gamma - ray selection , angular resolution and spectral resolution ) .
results on the gamma - ray morphology of the source are given in section 4 , whereas section 5 is concerned with the spectral analysis .
section 6 reviews and summarises observations relevant to the multiwavelength study of this object , in particular from the x - ray and radio bands .
section 7 attempts to derive some general constraints on the energetics and emission mechanisms in this source , sections 8 and 9 then discuss the electronic and hadronic scenarios respectively and finally in section 10 we draw some general conclusions .
h.e.s.s . is an array of four 13 m diameter imaging cherenkov telescopes located in the khomas highlands in namibia , 1800 m above sea level @xcite .
each telescope has a tesselated mirror with an area of 107 m@xmath16 @xcite and is equipped with a camera comprising 960 photomultipliers @xcite covering a field of view of 5@xmath5 diameter . during the observations , an array level hardware trigger requires each shower to be observed by at least two telescopes within a coincidence window of 60 ns @xcite . due to the efficient rejection of hadronic showers provided by stereoscopy
, the complete system ( operational since december 2003 ) can detect point sources at flux levels of about 1% of the crab nebula flux near zenith with a significance of 5 @xmath11 in 25 hours of observation .
this high sensitivity , the angular resolution of a few arc minutes and the large field of view make h.e.s.s . ideally suited for the study of the gamma - ray morphology of extended sources .
the region of the supernova remnant rx j0852.0 - 4622 was observed with the complete h.e.s.s .
array between december 2004 and may 2005 for a total observation time of 33 hours of on - source runs .
the data were recorded in runs of typical 28 minute duration in the so - called `` wobble mode '' , where the source is offset from the center of the field of view .
the offset angles both in right ascension and declination ( @xmath170.7@xmath5 , @xmath171.1@xmath5 and @xmath171.56@xmath5 ) were chosen in order to provide a full coverage of this widely extended supernova remnant . in order to reduce systematic effects due to varying observational conditions ,
quality selection criteria were applied on a run - by - run basis resulting in a total of 20 hours of high - quality data at zenith angles between 20@xmath5 and 50@xmath5 ( with an average of 30@xmath5 ) .
the energy threshold of the system increases with the zenith angle : for the observations presented here , the average threshold was around 250 gev .
the data were calibrated as described in detail in @xcite . in a first stage ,
a standard image cleaning was applied to shower images to remove the contamination due to the night sky background .
several independent analysis methods are used within the h.e.s.s .
collaboration @xcite to cross - check all results .
the results presented in this paper were obtained using a 3d - modeling of the light - emitting region of an electromagnetic air shower , a method referred to as `` the 3d - model analysis '' @xcite .
all results were verified and confirmed using the standard h.e.s.s .
analysis described in detail in @xcite .
we briefly recall the main characteristics of these methods : * the standard stereoscopic analysis is based on the hillas parameters of shower images @xcite .
the incident direction as well as the shower impact on the ground are reconstructed from the image axes , whereas parameters directly related to the widths and lengths of the images ( mean reduced scaled width and mean reduced scaled length ) are used for gamma - hadron separation .
the gamma - ray energy is estimated from the image intensity taking into account the reconstructed shower geometry , in particular the impact distance .
the performance of this analysis method as applied to extended sources can be found in @xcite . * in the 3d - model analysis , the cherenkov light emitted by a shower is modeled in the following way : the photon origins ( photosphere ) are distributed according to a 3d - gaussian law and their directions are assumed to follow an anisotropic angular distribution , with the overall requirement of rotational symmetry characteristic of an electromagnetic shower .
the expected number of cherenkov photons collected by each phototube of a given telescope is then calculated by integrating the light yield over the corresponding line of sight .
a comparison of the observed images to the expected ones by means of a maximum likelihood method allows to reject a large fraction of hadronic showers , namely those which are not compatible with rotational symmetry . an additional discrimination between gamma - rays and hadrons is provided by the lateral spread of the photosphere ( or 3d - width ) obtained from the likelihood fit . in practice
, we use a dimensionless parameter directly related to this quantity , the `` reduced 3d - width '' , whose distribution is almost zenith angle independent .
the energy @xmath18 of the primary gamma - ray is then reconstructed calorimetrically from the number of cherenkov photons obtained from the fit .
a complete review of the performance of this analysis method is given in @xcite .
for the generation of the excess skymaps for rx j0852.0 - 4622 , two different methods of background subtraction have been applied .
the first one is classic : the background level is estimated from off - source runs , observing sky regions without any gamma - ray sources in the field of view .
for this purpose , 20 hours of off runs taken at similar zenith angles are used .
all events passing the gamma - ray cuts of the different analysis methods , i.e. gamma - ray like background events , are used to estimate the background .
the second method of background subtraction is more recent and is called the `` weighting method '' @xcite . in this method , the signal and the background are estimated simultaneously in the same portion of the sky . in each sky bin
( treated independently ) , the signal and the background are estimated from those events originating from this bin exclusively .
each event is characterized by a discriminating parameter , the reduced 3d - width defined above .
since its distribution is fairly different for gamma - rays and hadrons , the respective numbers in each population are derived by a likelihood fit .
therefore , no cut on the reduced 3d - width is necessary .
the advantage of these background subtraction methods is that no assumption on the gamma - ray content in the field of view is necessary .
the bin size used in this analysis is 0.05@xmath19
@xmath20 0.05@xmath19 .
the images were further smoothed by a gaussian distribution with a standard deviation of 0.06@xmath19 to reduce the statistical fluctuations .
this procedure was chosen in order to match the h.e.s.s .
angular resolution and the binning of the images .
the resulting excess maps are in units of integrated excess counts per gaussian sigma of the smoothing function . in the study of the morphology of an extended source , one of the major objectives
is to have the best possible angular resolution . in an array such as h.e.s.s . including 4 telescopes placed in a square formation
, events triggering 4 telescopes are concentrated in the central region of the array , whereas events triggering 2 telescopes , being peripheral , are not so accurately reconstructed as the central ones .
therefore , to obtain a high angular resolution ( of the order of 0.06@xmath19 ) , one can restrict the analysis to events triggering at least 3 telescopes , which also further reduces the hadronic background .
the excess skymap in figure [ fig::velajr_34tels ] shows the gamma - ray image of rx j0852.0 - 4622 in a @xmath21 field of view , obtained with the 3d - model and the weighting method by keeping only 3 and 4 telescope events .
the gamma - ray content in this skymap presents a much higher statistics than the one obtained with the h.e.s.s .
dataset from february 2004 @xcite .
the significance is about @xmath22 with an excess of @xmath23 events , keeping all events inside a radius of @xmath24 around the center of the remnant . in order to explore the robustness of the result
, the data set was analyzed using the same calibration and analysis method but applying different sets of cuts , which resulted in different resolutions and statistics , all results being compatible with each other . additionally , the morphology was cross - checked using the standard analysis method for the reconstruction and the on - off method for the background subtraction . the comparison of the results obtained by the 2 methods in a region of @xmath25 radius around the center of the supernova remnant yields a correlation coefficient of @xmath26 .
these tests show that the gamma - ray morphology of the remnant remains consistent when analyzed with different sets of cuts or with different background subtraction methods . + the morphology appearing from the excess skymap in figure [ fig::velajr_34tels ] reveals a very thin shell of 1@xmath19 radius and @xmath60.2@xmath19 thickness .
it shows several bright regions in the north , north - western and south - eastern parts of the supernova remnant .
another feature is the remarkably circular general shape of this shell , even if the southern part shows a more broken ( non - uniform ) structure than the northern one .
this regular morphology resembles very much the image that one would expect from a homogeneously emitting shell . in order to investigate the projection effect of the 3d - source into a 2d - skymap ,
a simple geometrical model ( `` toy model '' ) of a thin and homogeneous shell has been used and adapted to the data . after calculating the projection ,
the obtained skymap is smoothed according to the average point spread function in this dataset .
the radial profiles ( i.e the number of events per unit solid angle as a function of the distance to the center of the remnant ) obtained with the `` toy - model '' for different values of the shell thickness are then fitted to those obtained from h.e.s.s .
data ( restricting to 3 and 4 telescopes events ) in the northern part of the remnant ( declination higher than @xmath27 ) which exhibits a clear shell as seen on the gamma - ray excess map . for each value of the shell thickness ,
the only parameter of the fit is the outer radius of the shell .
figure [ fig::radprof_toymodel ] shows that the bright shell observed by h.e.s.s .
is well reproduced by the simple geometrical model . in figure
[ fig::radprof_toymodel ] , the remarkable point to note is the small value of the shell thickness giving the best fit : it is equal to @xmath28 of the radius of the remnant and between 12.5@xmath29 and 22.5@xmath29 at 95% confidence level .
this contrasts with rx j1713.7 - 3946 , another shell - type supernova remnant resolved by h.e.s.s . , in which the shell thickness that best suited the data was about
45@xmath29 of the radius of the remnant .
this good match of the toy model and the h.e.s.s .
data clearly shows that the gamma - ray emission detected comes from a thin shell .
figures [ fig::velajr_inf500gev ] and [ fig::velajr_sup500gev ] show the morphology of rx j0852.0 - 4622 in two distinct energy bands , [email protected] tev and [email protected] tev , keeping only events triggering at least 3 telescopes .
the two energy bands were chosen in order to have approximately the same statistics in both datasets . clearly , the morphology of the remnant is the same in the two energy bands .
the overall radial profile in the two energy bands shown in figure [ fig::energyband_radprof ] confirms that the morphology does not vary significantly with energy .
for the spectral analysis , the source region ( on region ) is defined by a circle of 1@xmath19 radius centered on the position ( @xmath31 = 8@xmath3252@xmath33 , @xmath34 = @xmath3522 ) . in the study of a point - like source , the restricted selection of events due to the knowledge of the exact gamma - ray origin and the reduced offsets of the source from the center of the camera
improve the energy resolution .
however , in the present case , these two characteristics are lost , which results in an average energy resolution @xmath36e / e @xmath6 25@xmath29 for the 3d - model , slightly larger than for a point - like source . the energy resolution can be improved by restricting to 3 and 4 telescope events at the expense of a smaller statistics ( but with the same statistical significance ) . in this study ,
the two possibilities ( restricting or not to 3 and 4 telescope events ) were used in order to verify the stability of our results .
+ the spectral analysis requires selection criteria slightly different from those of the morphological study .
two cuts were applied independently of the telescope multiplicity : a cut on the reduced 3d - width and a cut on the image size at 80 photoelectrons .
all events passing the cuts and with reconstructed direction within a region of 1@xmath19 radius from the center of the remnant are considered as on events .
+ for the background estimation , off events were selected from the same field of view and in the same runs as the on events by selecting the region symmetric to the on region with respect to the camera center ( `` mirrored background '' ) .
a minimum distance of 0.1@xmath19 between the two regions is required in order to avoid any gamma - ray contamination in the off data .
this approach ensures that background events are taken at the same zenith angle and offset angles as the on events , which is crucial because of the strong dependence of the effective area upon these two quantities .
+ the energy spectrum of the gamma - ray excess is then obtained by the method of @xcite . in this method ,
an a priori spectral shape is assumed , whose parameters are obtained by fitting the expected distribution to the reconstructed energy distribution . in this procedure ,
gamma - ray acceptances and resolutions calculated from simulations are taken into account .
+ the differential energy spectrum thus obtained is shown in figure [ fig::velajr_spec_pl ] .
it extends from 300 gev up to 20 tev .
the spectral parameters were obtained from a maximum likelihood fit of a power law hypothesis dn / de = @xmath37 @xmath38 to the data , resulting in an integral flux above 1 tev of ( @xmath39 ) @xmath20 @xmath40 and a spectral index of 2.24 @xmath41 .
the present data include much more statistics especially at high energy than the previous h.e.s.s .
measurement @xcite and a slight deviation from a pure power law can be seen in figure [ fig::velajr_spec_pl ] .
this can explain that the average spectral index is slightly higher than the one from the previous measurement : @xmath42 . to confirm this spectrum , two other estimates of the spectrum of the whole remnant
were obtained by using other techniques . these checks
ensure that the systematics introduced by the spectral analysis technique or by the background estimation are small .
for these tests , the standard analysis method was applied by using a cut on the `` scaled parameters '' and on the image size at 80 photoelectrons . in figure
[ fig::velajr_allana ] , two different estimates of the spectrum are superimposed to the one obtained with the 3d - model .
one was computed by using the same background estimation with a mirror region and the other by using a background estimation based on off runs taken at similar zenith angles .
all spectra are compatible with each other and all show an indication of deviation from a power - law at high energy .
asca archival data of rx j0852.0 - 4622 were used to study the non - thermal emission of the remnant in the x - ray band .
figure [ fig : asca_fov ] shows the asca gis ( gis2 and 3 ) mosaic image of rx j0852.0 - 4622 in the @xmath43@xmath44 kev energy band @xcite obtained by using seven distinct pointings ( n1n7 ) which almost cover the entire remnant .
standard quality criteria ( screening procedures ) were applied to asca gis data , and spectra were extracted from the seven non - overlapping regions shown in figure [ fig : asca_fov ] . since the soft thermal emission from the vela snr foreground is spatially variable , background spectra were produced by using blank sky event files , which are considered to contain both non x - ray background and cosmic x - ray background , except for the central pointing ( n2 ) . for this region , due to different observation conditions , the background spectrum was extracted from large sky survey ( lss ) data observed during the asca pv phase @xcite . to derive the spectral parameters ,
the emission was modeled with two components .
a thermal model was used to account for the soft thermal emission from the vela snr , with a column density fixed at @xmath45 @xcite and a temperature of @xmath46(low ) @xmath47 kev , typical values for vela(low ) and @xmath48(vela ) . ] .
an absorbed power law was used for the hard emission .
a simultaneous spectral fitting was performed with both gis instruments from @xmath43 up to @xmath49kev .
figure [ fig : asca_specn1 ] shows the spectrum extracted from region n1 with the best - fit models ; it exhibits some line - like feature below @xmath50 kev , which originates from thermal emission from the vela snr and/or rx j0852.0 - 4622 .
the resulting spectral parameters of the fit are the photon index , @xmath51 , the column density for the non - thermal component , @xmath52 and the high temperature of the vela snr thermal component , @xmath46(high)=0.56@xmath53kev .
these values are well consistent with previous results ( slane et al .
2001 ; iyudin et al . 2005 ; bamba et al .
the derived non - thermal flux from region n1 in the 210kev energy band is 1.87@xmath54 while the thermal flux is @xmath630 times smaller than the non - thermal flux in 210kev band .
some residuals are still visible around 1 kev and suggest that another thermal component , which might originate from rx j0852.0 - 4622 , is needed . for the other six regions ,
spectra have been fitted with the same procedure as described above .
the resultant photon indices are between 2.5 and 2.8 and the total flux for the seven regions is 8.3@xmath55 ( @xmath50@xmath56 kev ) .
all errors described in this section are statistical and given at 90% confidence level .
the systematic error on the flux implied by the procedure has been estimated as follows : the error due to the uncertainty in the instrumental response is @xmath1710% and the one due to the uncertainty on the surface brightness variation in rx j0852.0 - 4622 is @xmath1720% .
figure [ fig::velajr_rosat ] presents the gamma - ray excess map obtained by h.e.s.s . with the superimposed contours of the x - ray data from the rosat
all sky survey .
the overall gamma - ray morphology seems to be similar to the one seen in the x - ray band , especially in the northern part of the remnant where a brightening is seen in both wavebands .
the correlation coefficient between the gamma - ray and the x - ray counts in bins of 0.2@xmath19 @xmath20 0.2@xmath19 is found to be equal to 0.60 and between 0.54 and 0.67 at 95@xmath29 confidence level .
the use of asca data gives a similar result .
the data of the various instruments were then compared to each other in six different sectors defined in figure [ fig::asca_camembert ] . in each region , the radial profiles were determined .
a binning of 0.1@xmath19 for the radial profiles was used ; this value , larger than the point spread function of each instrument , allows to safely compare their data .
all the radial profiles were normalized to unity .
the results of this study are shown in figure [ fig::radprof_xray ] .
one should note that , due to an incomplete coverage of the snr in regions 4 , 5 and 6 , the asca data are not reliable at distances larger than @xmath57 degree in these regions and were thus not included ; this incomplete coverage is visible in figure [ fig::asca_camembert ] .
the different radial profiles are in good agreement with each other in all regions
. differences between x - rays and tev gamma - rays seem to appear mainly in region 3 where a peak is seen only in the tev regime .
unfortunately , as the asca data are incomplete and the rosat data are contaminated by the vela snr , a more quantitative conclusion ( for example on the question of the boundaries of the snr ) can not be drawn .
mosaic observations of rx j0852.0 - 4622 with the australian telescope compact array ( atca ) were undertaken in november 1999 at frequencies of 1384 mhz and 2496 mhz @xcite .
figure [ fig::atca ] is a mosaic image of rx j0852.0 - 4622 obtained from 110 pointings at 1384 mhz .
the image suffers from sidelobes originating from the strong radio source ctb 31 ( rcw 38 ) . a certain similarity between the h.e.s.s .
image ( figure [ fig::velajr_34tels ] ) and the radio image of rx j0852.0 - 4622 ( figure [ fig::atca ] ) can be noticed at first glance .
the overall morphology appears to be similar : many features seen in the tev skymap coincide well with the radio image , such as the bright region in the northern part of the shell and the central filamentary structures inside the snr .
indeed , a high correlation can be seen when superimposing the radio contours to the h.e.s.s .
images , as shown in figure [ fig::hess_atca ] .
the atca contours directly match the structures from the h.e.s.s .
skymap and the features obtained inside the remnant are also well reproduced .
co data at 2.6 mm wavelength of the region of the vela molecular ridge and its surroundings were taken with the 4-m , mm and sub - mm telescope nanten in 1999 @xcite .
figure [ fig::fikui ] shows the integrated molecular column density derived from co data , in the rx j0852.0 - 4622 region .
a high density is clearly visible in the eastern part of the remnant ( which corresponds to regions 3 and 4 defined in figure [ fig::asca_camembert ] ) due to the presence of the vela molecular ridge ( vmr ) .
unfortunately , in region 4 , uncertainties concerning the x - ray data ( incomplete coverage and contamination by the vela snr foreground ) preclude any firm conclusion concerning a possible anti - correlation with the co data . on the other hand , no clear evidence of interaction between very high energy particles and the vmr
is seen in the h.e.s.s .
data since the gamma - ray flux does not vary by more than a factor of 2 from the eastern to the western sides , whereas the molecular column density drops by a factor of @xmath620 .
figure [ fig::correl_co_hess ] shows the correlation coefficient between the co and the h.e.s.s .
data calculated in different intervals of distance varying from @xmath58 to 3.2 kpc , in the 6 regions defined previously .
distances were estimated by using a galactic rotation model @xcite .
no clear correlation can be seen in this figure [ fig::correl_co_hess ] .
as stated above , the parameters of the supernova remnant rx j0852.0 - 4622 are not well known , in particular its age and its distance .
indeed , the remnant could be as close as the vela snr ( @xmath59 pc ) and very young , or as far as the vela molecular ridge ( @xmath60 kpc ) .
this leads to a change in the prediction of the x - ray or gamma - ray luminosity by a factor 25 according to the assumed distance .
however , beside the distance , other observational characteristics of the remnant provide some helpful constraints .
the analysis of asca data enabled @xcite to derive a limit on the thermal content of the remnant emission and therefore a limit on the density @xmath61 of the thermally emitting material in the remnant . using the assumption of thermal equilibrium ,
the limit obtained is @xmath62 , where @xmath63 is the filling factor of a sphere taken as the emitting volume in the region chosen .
it should be noted that this limit is restricted to a gas with temperature above 1 kev because of the contribution of the thermal emission from the vela snr at lower energies ; higher densities of cooler material are thus not excluded .
furthermore , if the shocks are strongly modified by the accelerated particles , the shock heating is substantially reduced and the x - ray data could be consistent with higher densities
. another important piece of information comes from the measured width of the filaments observed by the chandra satellite .
if we assume that these filaments are due to the outer shock , their width determines the downstream magnetic field @xmath64 .
the thickness of the filament observed by chandra in the 2 - 10 kev energy band is : @xmath65 .
this led @xcite to suggest a value of @xmath66 for a distance of 300 pc . the field evaluation based on the work of @xcite
gives values @xmath67 .
such values imply that the magnetic field is highly amplified . finally , the morphological analysis of the h.e.s.s .
data sets an upper limit on the thickness of the shell @xmath68% of the radius of the supernova remnant . for electrons , which rapidly lose their energy by synchrotron and inverse compton radiation , the
escape time out of the shell into the remnant interior ( by diffusion and convection ) should be larger than the characteristic time of energy losses .
otherwise , their interactions with photons from the cosmic microwave background would lead to a gamma - ray emission region more extended than that observed by h.e.s.s .. for protons , for which the time scale for energy losses is longer , the escape time must be larger than the age of the supernova remnant .
the characteristic escape time of particles is calculated with the formula @xmath69 , @xmath70 being the diffusive escape time in the bohm diffusion regime and @xmath71 the convective escape time .
the diffusive escape time is given by @xmath72 with @xmath73 the diffusion coefficient for @xmath74 tev particles and @xmath75 the upper limit on the width of the shell as resolved by h.e.s.s . ; the convective escape time is given by @xmath76 with @xmath77 the flow velocity into the interior in the shock frame , as derived from @xcite and reported in table [ tab : evol3 ] .
this allows us to calculate a lower limit on the magnetic field @xmath78 , reported in table [ tab : evol3 ] , in order to confine particles of @xmath79 within the thin shell resolved by h.e.s.s .. however , it should be noted that the shell thickness in the case of a proton model also reflects the thickness of the gas target and is not necessarily a good indicator of thickness of the zone filled by accelerated particles ( unlike the electron case where the cmb radiation provides a uniform target ) .
nevertheless the condition that the acceleration time scale associated with 100 tev particles be less than the age of the remnant , or the equivalent condition that the associated diffusion length scales be small compared to the shock radius , lead , within factors of order unity , to identical lower limits on the magnetic field strength .
.magnetic field @xmath78 , total energy of accelerated protons @xmath80 , efficiency and age of the supernova remnant ( assuming an energy explosion of @xmath81 ) for different values of distance @xmath82 , density of the medium @xmath83 and ejected mass @xmath84 . for the latter ,
we have chosen two possible values for rx j0852.0 - 4622 : @xmath85 which is typical for snia and @xmath86 as an average value for snii .
@xmath78 is the lower limit on the magnetic field allowing to confine particles of @xmath87 in the thin shell resolved by h.e.s.s .. it has been calculated from the value of the shock velocity @xmath88 indicated in this table as derived from @xcite .
[ cols="^,^,^,^,^,^,^,^ " , ] another constraint comes from the broadband spectral energy distribution ( from radio to gamma - rays ) as interpreted by a model of emission processes taking place in the supernova remnant .
the objective is to constrain parameters like the magnetic field , the density of the medium and the injection spectrum of the primary particles , thanks to a multi - wavelength study . the gamma - ray spectrum obtained by analyzing the h.e.s.s .
data is well described by a power - law with a photon index of @xmath89 : @xmath90 this flux can be translated into a global energy flux @xmath91 between 1 and 10 tev by using the formula : @xmath92 which corresponds to a @xmath93-ray luminosity @xmath94 .
if one assumes that the gamma - ray flux is entirely due to proton - proton interactions , we can estimate the total energy @xmath95 of accelerated protons in the range @xmath96 tev required to produce this gamma - ray luminosity . in this energy range , the characteristic cooling time of protons through the @xmath2 production channel is approximately independent of the energy and can be estimated by @xcite : @xmath97 .
thus:@xmath98 assuming that the proton spectrum continues down to @xmath99 gev with the same spectral slope as that of the photon spectrum , the total energy injected into protons is estimated to be : @xmath100 values of @xmath80 are reported in table [ tab : evol3 ] for different distances and densities of the ambient medium .
+ the x - ray spectral analysis of the whole remnant in the 2 - 10 kev energy band was presented in section [ lab : asca ] : the non - thermal spectrum is well described by a power - law with a spectral index of @xmath101 and a flux @xmath102 .
the fluxes at 1.40 ghz and 2.42 ghz were taken from the analysis of the parkes data by @xcite .
later on , we shall use these multi - wavelength measurements when comparing the h.e.s.s .
spectral data to broadband models . in the simple model used here
, we assume that primary particles ( protons and electrons ) are injected at a constant rate with the same spectral shape , namely a power - law with an exponential cut - off at the energy @xmath18 , into a spherical shell of fixed thickness @xmath75 . as seen previously , the h.e.s.s .
data require @xmath68% of the remnant radius .
the injection is supposed to last a time t ( the age of the supernova remnant ) in a region of magnetic field @xmath103 and ambient density @xmath83 .
the electron to proton ratio @xmath104 is a free parameter .
the energy distribution of the electrons is calculated at a fixed time @xmath105 by taking into account energy losses due to synchrotron radiation , inverse compton scattering and bremsstrahlung , while the proton spectrum is calculated by taking into account the escape out of the shell by diffusion ( in the bohm diffusion regime ) and convection , as described in section [ lab : cons ] .
adiabatic losses are neglected .
the broadband spectrum of the source is then derived by taking into account p - p interactions , synchrotron radiation ( of primary and of secondary electrons produced via p - p interactions ) , inverse compton scattering and bremsstrahlung . concerning the energy density of the target photons in the inverse compton process , we added the contribution of the cosmic microwave background , @xmath106 , and that of the galactic seed photons , namely on average : @xmath107 for the optical star light and @xmath108 for the infra - red background @xcite .
+ it is clear that such a model oversimplifies the acceleration process in an expanding remnant , as discussed by e.g. @xcite and @xcite . to this
must be added the uncertainties introduced by the dynamics of the ejecta , the nonuniform structure of the ambient medium , and the complexities of the reaction of the accelerated particles on both the magnetic field and the remnant dynamics . however , as a starting point for estimates such a simple model is still , we feel , useful , at least for those cases where the remnant evolution is relatively smooth and the emission is not dominated by relic particles injected and accelerated at earlier times @xcite .
+ in this study , we explored two different cases of distance ( 200 pc and 1 kpc ) , both for the electronic process ( gamma - rays mainly produced by inverse compton scattering ) and for the hadronic process ( gamma - rays mainly produced by p - p interactions ) . using the free expansion and sedov - taylor phase equations @xcite , one can easily find that , in the nearby case , the supernova remnant should be very young ( @xmath109 years ) . on the contrary , in the distant case
, the supernova remnant would be rather old ( @xmath110 years ) . in the following scenarios , the total energy injected into the protons
is fixed to @xmath111 erg ( i.e 10% of the energy of explosion of an average supernova ) , and the width of the shell must be smaller than 22.5% of the radius of the remnant . the age of the remnant is assumed to be 500 years at 200 pc and 5000 years at 1 kpc .
the other parameters , namely the characteristics of the injection spectrum ( spectral index and cut - off energy ) , the electron / proton ratio @xmath104 at the injection level , the density of the medium and the magnetic field are free parameters in the fit .
if the tev emission is mainly due to inverse compton scattering , one should note that , independently of the distance assumed , the ratio of the x - ray flux and the gamma - ray flux determines the value of the magnetic field @xmath103 . in the case of rx
j0852.0 - 4622 , assuming that inverse compton emitting electrons are contained in a volume equal to the one responsible for synchrotron emission , one can easily deduce that the magnetic field has to be close to @xmath112 .
figure [ fig : fitlepton ] a ) shows the best fit obtained , together with the measurements at different wavelengths : the injection spectrum follows a power - law of index 2.4 and an exponential cut - off at 40 tev ; the value of the magnetic field is @xmath113 and the density of the medium should be lower than @xmath114 so that the gamma - ray flux produced by p - p interactions should not be significant .
one can clearly notice that the radio flux predicted by our model is about 3 times larger than the one observed in the radio range by parkes .
even more constraining is the thickness of the shell observed in gamma - rays , which is inconsistent with the observations .
indeed , at 200 pc , our limit on the width of the shell implies @xmath115 pc , which leads to an escape time by diffusion and convection of about 300 years for an ambient density of @xmath114 and an energy of 40 tev . this value is lower than the age of the remnant ( @xmath109 years ) but also lower than the synchrotron loss time of 8700 years .
electrons above @xmath116 tev will escape the shell and thus automatically produce gamma - rays by inverse compton scattering on the cosmic microwave background .
these gamma - rays have on average an energy @xmath117 greater than @xmath118 gev .
therefore , we expect to observe a much thicker shell for @xmath119 gev . the present analysis of h.e.s.s .
data does not show any variation of the morphology of the remnant with the energy , which highly disfavours this scenario . in the case of a distant object , the magnetic field must also be close to @xmath113 ; the only difference comes from the fact that energy losses are no more negligible since the supernova remnant is older ( @xmath110 years ) .
these energy losses tend to steepen the electron spectrum and , in order to compensate this effect , the cut - off energy has to be increased in comparison to the preceding case .
the parameters of our best fit is an injection spectrum following a power - law with an index of 2.4 , a cut - off energy of 80 tev and an electron / proton ratio @xmath120 ( figure [ fig : fitlepton ] b ) ) .
the different multi - wavelength data are reasonably reproduced despite a radio flux three times larger than the observational data from parkes .
this last point is not critical since the fit could be improved by including non - linear acceleration effects , which are expected to lead to a steeper rise of the synchrotron sed with frequency above the radio range .
+ in this case , the characteristic time of synchrotron losses is 3700 years for the maximal energy 80 tev , while the age of the snr varies from 4000 years to 9000 years depending on the density of the medium in which it evolves .
one can easily find that the escape time is either larger than the synchrotron loss time or larger than the age of the remnant , and is thus irrelevant in this case .
first , one can see from table [ tab : evol3 ] that the only way to explain the entire gamma - ray flux by proton - proton interactions in a homogeneous medium is to assume that rx j0852.0 - 4622 is a nearby supernova remnant ( d @xmath121 pc ) .
indeed , for larger distances and a typical energy of the supernova explosion , the acceleration efficiency would be excessive ( assuming a uniform ambient density compatible with the limit implied by the non - detection of thermal x - rays ) .
nevertheless , a distance of 1 kpc should also be considered if rx j0852.0 - 4622 is assumed to be the result of a core collapse supernova which exploded inside a bubble created by the wind of a massive progenitor star , as proposed by @xcite for the snr rx j1713.7 - 3946 . according to stellar wind theory @xcite
, the size of the bubble evolves according to the formula : @xmath122 pc . for a density of 1 @xmath123
, the radius of this bubble would be equal to 45 pc . in the case of a close by supernova
remnant , its size would be significantly lower than the size of the bubble and the hypothesis of a homogeneous medium would be satisfactory . on the opposite , for larger distances ( @xmath124 kpc ) , the presence of the vela molecular ridge can produce a sudden increase of the density leading to a smaller bubble ( @xmath125 pc for a density of @xmath126 ) , which would make the proton - proton interactions efficient at the outer shock .
+ in any case , independently of the distance of the remnant , the extension of the h.e.s.s .
spectrum up to 20 tev implies an energy cut - off @xmath18 higher than 100 tev . if we assume that energy losses are negligible over the lifetime of the remnant , the synchrotron spectrum can then be approximated by the formula @xcite : @xmath127\ ] ] with @xmath128 this relation implies that the magnetic field should be lower than @xmath129 to obtain a synchrotron peak centered at an energy lower than 2 kev . the gamma - ray flux would then be entirely produced by inverse compton scattering ( as seen previously in the electronic process ) which enables to exclude this situation .
therefore , the magnetic field should be high enough to produce significant energy losses during the lifetime of the remnant @xmath130 ( @xmath109 years ) and create a break in the synchrotron spectrum . refering to @xcite , to obtain a break at an energy @xmath131 close to @xmath132 kev
, one would need a magnetic field higher than 40 @xmath133 : @xmath134 in this case , our best fit is obtained for an injection spectrum in the form of a power - law with an index of 2.1 , a cut - off at 110 tev and a very low electron / proton ratio of @xmath135 . the density of the medium is 0.2 @xmath136 and the magnetic field amounts to 120 @xmath137 : these two values are compatible with both the limit implied by the absence of thermal x - rays and with the thin filaments resolved by chandra . on the other hand
, one can note in figure [ fig : fithadron ] a ) that such a model with the above parameters does not provide a good description of the asca data , since the energy losses tend to steepen the electron spectrum .
however , knowing the difficulty of the x - ray spectral analysis , this point can not be used to exclude this scenario .
furthermore , a better agreement could be obtained by simply relaxing the assumption used in the model that electrons and protons have similar injection spectra . in this last case ,
if the density were low enough as to be compatible with the absence of thermal x - rays , table [ tab : evol3 ] shows that the values of @xmath80 required to account for the total observed gamma - ray flux exceed the total energy of the supernova explosion assumed in the present calculation or would require an anomalously energetic explosion .
as stated previously , a way out of this difficulty would be to consider the case of a bubble created by the wind of the massive progenitor .
actually , our best fit is obtained for an injection spectrum in the form of a power - law with an index @xmath138 , a cut - off energy at @xmath139 tev and an electron / proton ratio of @xmath140 ( figure [ fig : fithadron ] b ) ) .
the density of the medium is found to be @xmath141 , which is acceptable in the framework of the bubble scenario .
the magnetic field , @xmath142 , is compatible with the very thin shell resolved by h.e.s.s .. finally , one should note that our model perfectly reproduces the radio and the h.e.s.s .
data , but only approximately the slope coming from the asca spectral analysis .
we have firmly established that the shell - type supernova remnant rx j0852.0 - 4622 is a tev emitter and for the first time we have resolved its morphology in the gamma - ray range . the thin shell observed by h.e.s.s .
is highly correlated with the emission observed in x - rays with the rosat all - sky survey and asca but is also very similar to the morphology resolved in radio by atca .
the overall gamma - ray energy spectrum extends over two orders of magnitude , providing the direct proof that particles of @xmath143 tev are accelerated at the shock .
there is an indication of deviation from a pure power law at high energy , but the lack of statistics does not enable us to draw any firm conclusions on this point .
this spectrum is very similar to that of the other shell - type supernova remnant resolved by h.e.s.s .
, rx j1713.7 - 3946 , although the morphology of the latter was very different with a much thicker shell .
+ the question of the nature of the particles producing the gamma - ray signal observed by h.e.s.s . was also addressed . despite the large uncertainty concerning the parameters of rx j0852.0 - 4622 , the h.e.s.s .
data already give some strong constraints . in the case of a close by remnant ,
the results of the morphological study combined with our spectral modeling highly disfavour the electronic scenario which is unable to reproduce the thin shell observed by h.e.s.s . and the thin filaments resolved by chandra .
the hadronic scenario can approximately reproduce the data at the expense of a very low electron / proton ratio . in the case of a medium distance , the explosion energy needed to explain the gamma - ray flux observed by h.e.s.s .
, taking into account the limit on the density implied by the absence of thermal x - rays , would disfavour the hadronic process . at larger distances , both the electronic and the hadronic scenario are possible , at the expense , for the electronic process , of a low magnetic field of @xmath144 .
such a small magnetic field exceeds typical interstellar values only slightly and is difficult to reconcile with the theory of magnetic field amplification at the region of the shock @xcite .
+ finally , it appears clearly from figures [ fig : fitlepton ] and [ fig : fithadron ] , that the flux expected for lower energy gamma - rays ( @xmath145 gev ) for the electronic process ( synchrotron + ic scattering ) or for the hadronic process ( proton - proton interactions ) are significantly different .
the results which should hopefully be obtained by integral , glast or h.e.s.s .
ii will therefore have a great interest for the domain .
the support of the namibian authorities and of the university of namibia in facilitating the construction and operation of h.e.s.s .
is gratefully acknowledged , as is the support by the german ministry for education and research ( bmbf ) , the max planck society , the french ministry for research , the cnrs - in2p3 and the astroparticle interdisciplinary programme of the cnrs , the u.k . particle physics and astronomy research council ( pparc ) , the ipnp of the charles university , the south african department of science and technology and national research foundation , and by the university of namibia .
we appreciate the excellent work of the technical support staff in berlin , durham , hamburg , heidelberg , palaiseau , paris , saclay , and in namibia in the construction and operation of the equipment .
m. filipovic would like to thank milorad stupar for his work on the atca data .
aharonian f. a. , atoyan a.m. , kifune t. , 1997 , mnras , 291 , 162 aharonian f. a. , in very high energy cosmic gamma - ray radiation , world scientific ed . , 2004 ,
aharonian f. a. et al .
( _ h.e.s.s .
collaboration _ ) 2004 , nature , 432 , 75 aharonian , f. , et al .
( _ h.e.s.s .
collaboration _ ) 2004 , aph , 22 , 109 aharonian , f. , et al .
( _ h.e.s.s .
collaboration _ ) 2005b , a&a , 430 , 865 aharonian , f. , et al .
( _ h.e.s.s .
collaboration _ ) 2005c , a&a , 437 , l7 aharonian , f. , et al .
( _ h.e.s.s .
collaboration _ ) 2006 , a&a , 449 , 223 aschenbach b. , 1998 , nature , 396 , 141 aschenbach b. , iyudin a. f. & schnfelder v. , 1999 , a&a , 350,997 bamba a. et al .
, 2005a , apj , 621 , 793 bamba a. , yamazaki r. & hiraga j. s. , 2005b , apj , 632 , 294 bell a. r. , lucek s. g. , 2001 , mnras , 321 , 433 berezhko e. g. , elshin , v. k. & ksenofontov , l. tl , 1996 , jetph , 82 , 1 berezhko e. g. , vlk h. j. , 2004 , a&a , 419 , l27 berezhko e. g. , vlk h. j. , 2006 , a&a , 451 , 981 bernlhr , k. , et al .
2003 , aph , 20 , 111 brand j. , blitz l. , 2003 , a&a , 275 , 67 chen , w. and gehrels , n. , 1999 , apj , 514 , l103 - 106 chevalier r.a . , liang e.p . , 1989 , apj , 344 , 332 combi j. a. , romero j. e. , benaglia p. , 1999 ,
apj , 519 , l177 duncan a. r. & green d. a. , 2000 , a&a , 364 , 732 drury l. oc . ,
markiewicz , w. j. & vlk , h. j. , 1989 , a&a , 225 , 179 drury l. oc .
, aharonian f. a. & vlk h. , 1994 , a&a , 287 , 959 enomoto r. et al . 2002 , nature 416 , 823 funk , s. , et al .
2004 , aph , 22 , 285 hashimoto , t. , et al .
2001 , nucl
a , 686 , 591 hinton , j. a. 2004 , newar , 48 , 331 iyudin a. f. et al . , 1998 , nature , 326,142 iyudin a. f. et al . , 2005 , a&a , 429,225 kargaltsev o. et al .
2002 , apj 580 , 1060 katagiri h. et al .
, 2005 , apj , 619 , l163 koyama k. et al . , 1995 ,
nature , 378 , 255 koyama k. et al .
, 1997 , pasj , 49 , l7 lemoine - goumard , m. et al .
2005 , in proceedings of the conference `` towards a network of atmospheric cherenkov detectors vii '' , ed .
b. degrange & g. fontaine ( palaiseau : ecole polytechnique ) , 173 lemoine - goumard , m. , & degrange , b. 2005 , in proceedings of the conference `` towards a network of atmospheric cherenkov detectors vii '' , ed .
b. degrange & g. fontaine ( palaiseau : ecole polytechnique ) , 209 lemoine - goumard , m. , degrange , b. , tluczykont , m. , 2006 , astropart .
phys . , 25 , 195 lu f.j . & aschenbach b. , 2000 .
, 362 , 1083 mathis j.s . ,
mezger p.g .
, panagia n. , 1983 , a&a , 128 , 212 mereghetti s. , 2001 , apj , 548 , l213 moriguchi , y. et al .
2001 , pasj , 53 , 1025 muraishi h. et al . , 2000 ,
a&a , 354 , l57 de naurois , m. , et al .
2005 , in proceedings of the conference `` towards a network of atmospheric cherenkov detectors vii '' , ed .
b. degrange & g. fontaine ( palaiseau : ecole polytechnique ) , 149 pavlov , g. g. , et al . , 2001 ,
apj , 559 , l131 piron , f. , 2000 , phd thesis , universit paris xi reynolds , s.p . , 1998 ,
apj , 493 , 375 reynoso , e. et al .
2006 , a&a , 449 , 243 schnfelder v. et al .
2000 , in aip conf .
510 , fifth compton symp . ,
m. l. mcconnell & j. m. ryan ( new - york : aip ) , 54 slane p. et al . , 2001 , apj , 548 , 614 stupar , m. et al .
2005 , advances in space research 35 , 1047 truelove j. k. , mckee c. f. , 1999 , apjs , 120 , 299 tsunemi h. et al . , 2000 , pasj , 52 , 887 ueda , y. , takahashi , t. , inoeu , h. , et al . , 1999 ,
apj , 518 , 656 vincent , p. , et al .
2003 , in proceedings of the 28th international cosmic ray conference , t. kajita et al .
( universal academy press , tokyo , 2003 ) , 2887 yamazaki , r. , kohri , k. , bamba , a. , et al . , 2006 ,
mnras in press , astro - ph/0601704 .
the magnetic field amounts to @xmath146 and the density of the ambient medium is @xmath147 .
b ) on the right , the modeling was done by using an injection spectrum in the form of a power - law of index 2.4 , an exponential cut - off at 80 tev and an electron / proton ratio @xmath148 .
the magnetic field amounts to @xmath149 , and the density of the ambient medium is @xmath150 .
the parkes data @xcite in the radio range , the asca data and the h.e.s.s .
data are indicated .
red lines correspond to electrons , and blue lines to protons .
the following processes have been taken into account : synchrotron radiation of primary ( solid red line ) and secondary ( dotted blue line ) electrons , ic scattering ( dotted red line ) , bremsstrahlung ( dotted - dashed red line ) and proton - proton interaction ( solid blue line ) . ] .
the magnetic field amounts to @xmath146 and the density of the ambient medium is @xmath147 .
b ) on the right , the modeling was done by using an injection spectrum in the form of a power - law of index 2.4 , an exponential cut - off at 80 tev and an electron / proton ratio @xmath148 .
the magnetic field amounts to @xmath149 , and the density of the ambient medium is @xmath150 .
the parkes data @xcite in the radio range , the asca data and the h.e.s.s .
data are indicated .
red lines correspond to electrons , and blue lines to protons .
the following processes have been taken into account : synchrotron radiation of primary ( solid red line ) and secondary ( dotted blue line ) electrons , ic scattering ( dotted red line ) , bremsstrahlung ( dotted - dashed red line ) and proton - proton interaction ( solid blue line ) . ] .
the magnetic field amounts to @xmath151 , and the density of the ambient medium is @xmath152 .
b ) on the right , the modeling was done by using an injection spectrum in the form of a power - law of index 2.0 , an exponential cut - off at 100 tev and an electron / proton ratio @xmath140 .
the magnetic field amounts to @xmath142 , and the density of the ambient medium is @xmath141 .
the parkes data @xcite in the radio range , the asca data and the h.e.s.s .
data are indicated .
red lines correspond to the electrons , and blue lines to protons .
the following processes have been taken into account : synchrotron radiation of primary ( solid red line ) and secondary ( dotted blue line ) electrons , ic scattering ( dotted red line ) , bremsstrahlung ( dotted - dashed red line ) and proton - proton interaction ( solid blue line ) . ] .
the magnetic field amounts to @xmath151 , and the density of the ambient medium is @xmath152 .
b ) on the right , the modeling was done by using an injection spectrum in the form of a power - law of index 2.0 , an exponential cut - off at 100 tev and an electron / proton ratio @xmath140 .
the magnetic field amounts to @xmath142 , and the density of the ambient medium is @xmath141 .
the parkes data @xcite in the radio range , the asca data and the h.e.s.s .
data are indicated .
red lines correspond to the electrons , and blue lines to protons .
the following processes have been taken into account : synchrotron radiation of primary ( solid red line ) and secondary ( dotted blue line ) electrons , ic scattering ( dotted red line ) , bremsstrahlung ( dotted - dashed red line ) and proton - proton interaction ( solid blue line ) . ] | the shell - type supernova remnant rx j0852.0 - 4622 was observed with the high energy stereoscopic system ( h.e.s.s . ) of atmospheric cherenkov telescopes between december 2004 and may 2005 for a total observation time of 33 hours , above an average gamma - ray energy threshold of 250 gev .
the angular resolution of @xmath0 ( for events triggering 3 or 4 telescopes ) and the large field of view of h.e.s.s .
( @xmath1 diameter ) are well adapted to studying the morphology of the object in very high energy gamma - rays , which exhibits a remarkably thin shell very similar to the features observed in the radio range and in x - rays .
the spectral analysis of the source from 300 gev to 20 tev is also presented .
finally , the possible origins of the very high energy gamma - ray emission ( inverse compton scattering by electrons or the decay of neutral pions produced by proton interactions ) are discussed , on the basis of morphological and spectral features obtained at different wavelengths . |
malicious cryptography and malicious mathematics malicious are an emerging field [ 2 , 3,4 ] that finds its origin in the initial work of young and yung on the use of asymmetric cryptography in the design of dedicated offensive functions for money extorsion ( cryptovirology ) [ 5 ] . but this initial approach is very limited and gives only a very small insight of the way mathematics and cryptography can be actually perverted by malware .
malicious cryptology and malicious mathematics make in fact explode young and yung s narrow vision .
this results in an unlimited , fascinating yet disturbing field of research and experimentation .
this new domain covers several fields and topics ( non - exhaustive list ) : * use of cryptography and mathematics to develop `` _ super malware _ ''
( _ ber - malware _ ) which evade any kind of detection by implementing : * * optimized propagation and attack techniques ( e.g. by using biased or specific random number generator ) @xcite . * * sophisticated self - protection techniques .
the malware code protect itself and its own functional activity by using strong cryptography - based tools @xcite . * * code invisibility techniques using testing simulability @xcite . * * code mutation techniques ( by combining the different methods of the previous three items ) . * use of complexity theory or computability theory to design undetectable malware @xcite . * use of malware to perform applied cryptanalysis operations ( theft of secrets elements such as passwords or secret keys , static or on - the - fly modification of systems to reduce their cryptographic strength @xcite ... ) by a direct action on the cryptographic algorithm or its environment . * design and implementation of encryption systems with hidden mathematical trapdoors .
the knowledge of the trap ( by the system designer only ) enables to break the system very efficiently . despite the fact that system is open and public
, the trapdoor must remain undetectable .
this can also apply to the keys themselves in the case of asymmetric cryptography @xcite . to summarize
, we can define malicious cryptography and malicious mathematics as the interconnection and interdependence of computer virology with cryptology and mathematics for their mutual benefit .
the possibilities are virtually infinite .
it is worth mentioning that this vision and the underlying techniques can be translated for `` non - malicious '' use as the protection of legitimate codes ( e.g. for copyright protection purposes ) against static ( reverse engineering ) and dynamic ( sandboxing , virtualization ) analyses . if simple and classical obfuscation techniques are bound to fail as proved by barak and al .
@xcite theoretically ( for a rather restricted model of computation ) , new models of computation @xcite and new malware techniques @xcite have proved that efficient code protection can be achieved practically .
if the techniques arising from this new domain are conceptually attractive and potentially powerful , their operational implementation is much more difficult and tricky .
this requires a very good mastery of algorithms and low - level programming . above all the prior operational thinking of the attack
is unavoidable .
the same code implementing the same techniques in two different contexts will have quite different outcomes .
in particular , we must never forget that the code analysis is static ( disassembly / decompilation ) but also dynamic ( debugging , functional analysis ... ) .
code encryption usually protects possibly against static code analysis only . in this paper
we will show how the techniques of malicious cryptography enable to implement total amoring of programs , thus prohibiting any reverse engineering operation .
the main interest of that approach lies in the fact that transec properties are achieved at the same time . in other words ,
the protected binary has the same entropy as any legitimate , unprotected code .
this same technique can also achieve a certain level of polymorphism / metamorphism at the same time : a suitable 59-bit key stream cipher is sufficient to generate up to @xmath0 variants very simply .
more interestingly , the old fashioned concept of decryptor which usually constitutes a potential signature and hence a weakness , is totally revisited .
these techniques have been implemented by the author in the libthor which has been written and directed by anthony desnos @xcite .
the paper is organized as follows .
section [ cacm ] recalls some important aspect about code protection and discusses key points about code armouring and code mutation .
then section [ cs ] presents the context we consider to apply our technique : we aim at protecting critical portions of code that are first transformed into an intermediate representation ( ir ) , then into the bytecode .
this code itself is that of virtual machine used to provide efficent protection against dynamic analysis .
section [ mprng ] presents the different models and techniques of malicious pseudo - random number generator ( prng ) .
finally section [ imp_k ] presents implementation issue which must be considered to achieve protection against both static and dynamic analyses .
we will not recall in details what these two techniques are . the reader may consult @xcite for a complete presentation with respect to them . as far as malicious cryptography
is concerned , we are just going to discuss some critical points which are bottlenecks in their effective implementation .
most of the times the attempts to play with these techniques fail due to a clumsy use of cryptography .
code armouring @xcite consists in writing a code so as to delay , complicate or even prevent its analysis .
as for code mutation techniques they strive to limit ( polymorphism ) or to remove ( metamorphism ) any fixed component ( invariant ) from one mutated version to another . the purpose is to circumvent the notion of antiviral signature , may it be a a simple sequence of contiguous or nor contiguous bytes or meta - structures such as control flow graphs ( cfg ) and other traces describing the code functional ( behavioral ) structure . in all these cases ,
the general approach is to encrypt the code to protect or to use special techniques for generating random data .
but encryption and generation of randomness relates to two major practical problems : their characteristic entropy profile and the secret elements ( keys ) management .
the whole problem lies in the fact that code armouring and code mutation involve random data .
these must be generated on - the - fly and in the context of metamorphism , the generator itself must be too .
for sake of simplicity , we shall speak of _ pseudo - random number generator _ ( prng ) to describe both a random number generator and an encryption system .
the difference lies in the fact that in the latter case either random data produced from the expansion of the key are combined with the plaintext ( stream ciphers ) or they are the result of the combination of the key with the plaintext ( block ciphers ) .
the whole issue lies in the generation of a so - called `` good '' randomness . except that in the context of malicious cryptography @xcite , the term `` good '' does not necessarily correspond to what cryptographers
usually mean .
in fact , it is better yet a simplified but sufficient reduction as a first approximation to use the concept of entropy @xcite . in the same way
, the term of random data will indifferently describe the random data themselves or the result of encryption . consider a ( malicious ) code as an information source @xmath1 . when parsed , the source outputs characters taking the possible values @xmath2 , each with a probability @xmath3 $ ] .
then the entropy @xmath4 of the source is the following sum - grams and would compute entropy when @xmath5 . ] : @xmath6 random data , by nature will exhibit a high entropy value . ] . on the contrary ,
non random data exhibit a low entropy profile ( they are easier or less difficult to predict ) . from the attacker
s point of view the presence of random data means that something is hidden but he has to make the difference between legitimate data ( e.g. use of packers to protect code against piracy ) and illegitimate data ( e.g. malware code ) . in the nato terminology at the present time it is the most precise and accurate one as far as infosec is concerned random data relate to a comsec ( _ communication security _ ) aspect only . for the attacker ( automated software or human expert ) ,
the problem is twofold : first detect random data parts inside a code and then decrypt them . in this respect ,
any code area exhibiting a high entropy profile must be considered as suspicious . to prevent attention to be focused on those random parts ,
is it possible to add some transec ( _ transmission security _ ) aspect .
the most famous one is steganography but for malware or program protection purposes it is not directly usable ( data can not be directly executed ) and we have to find different ways .
the other solution is to use malicious statistics as defined and exposed in @xcite .
it is also possible to break randomness by using noisy encoding techniques like perseus @xcite . in this case
the code remains executable while being protected and exhibiting low entropy profile at the same time ( or the entropy profile of any arbitrary kind of data ) . as for an exemple , on a 175-bytes script , the unprotected code has an entropy of 3.90 .
an encrypted version ( by combining simple transpositions with simple substitutions ) of that code has an entropy equal to 5.5 .
when protected by perseus the entropy is about 2.57 .
so it is very close to a normal program and thus evade entropy - based detection .
this applies well on any data used for code mutation ( e.g. junk code insertion ) , including specific subsets of code as cfgs : randomly mutated cfg must exhibit the same profile as any normal cfg would .
otherwise , considering the comsec aspect only is bound to make the code detection very easy .
encrypting a code or a piece of code implies its preliminary deciphering whenever it is executed .
but in all of the cases except those involving money extortion introduced young and yung @xcite the key must be accessible to the code itself to decipher .
consequently in a way or another it is contained in a more or less obfuscated form inside the code .
therefore is it accessible to the analyst who will always succeed in finding and accessing it . instead of performing cryptanalysis , a simple decoding /
deciphering operation is sufficient .
it is therefore necessary to consider keys that are external to the encrypted code .
two cases are possible @xcite : * environmental key management introduced in @xcite and developped in @xcite .
the code gathers information in its execution environment and calculates the key repeatedly .
the correct key will be computed when and only when the suitable conditions will be realized in the code environment which is usually under the control of the code designer .
the security model should prohibit dictionary attacks or environment reduction attacks ( enabling reduced exhaustive search ) by the code analyst .
consequently the analyst must examine the code in an controlled dynamic area ( sandbox or virtual machine ) and wait until suitable conditions are met without knowing when they will be .
however it is possible to build more operational scenarii for this case and to detect that the code is being analyzed and controlled @xcite . *
use of @xmath7-ary codes @xcite in which a program is no longer a single monolithic binary entity but a set of binaries and non executable files ( working in a serial mode or in a parallel mode ) to produce a desired final ( malicious or not ) action .
then the analyst has a reduced view on the whole code only since generally he can access a limited subset of this @xmath7-set . in the context of ( legitimate ) code protection , one of the files will be a kernel - land module communicating with a userland code to protect .
the code without the appropriate operating environment with a user - specific configuration by the administrator will never work .
this solution has the great advantage of hiding ( by outsourcing it ) , the encryption system itself .
it is one particular instance with respect to this last solution that we present in this paper .
in this section we are going to illustrate our approach with simple but powerful cases . without loss of generality and for sake of clarity , we consider a reduced case in which only a very few instructions are protected against any disassembly attempt . of course any real case as we did in libthor will consider far more instructions , especially all critical ones ( those defining the cfg for instance ) .
the libthor library initiated and mainly developped by anthony desnos @xcite uses virtual machines ( vm ) as the core tool .
virtual machines offer a powerful protection for an algorithm or anything else that we would like to protect against reverse engineering .
we can have a lot of different vms piled up ( like russian dolls ) into a software .
this technique must be combined with classical techniques since it is just a one more step towards code protection . in libthor ,
the main goal of vms is to build a simple code which interprets another one .
the idea is to take a piece of x86 assembly instructions and to have a simple , dynamic , metamorphic vms very quickly to interpret it .
it is worth stressing on the fact that you can embed different vm into the same program to protect differents parts . in desnos libthor vms
are codes which interpret an intermediate representation ( ir ) derived from the reil language @xcite .
so a translation between x86 assembly instructions and the chosen ir is performed .
the vm is totaly independent from the libc or anything like that .
the main goal of the vm is to run an algorithm into an encapsulated object ( which can be loaded anywhere ; in fact it is a simple shellcode ) , but the main feature of the vm is to have a different version of it at each generation for a same code so if we want to protect a simple instruction @xmath1 , we can build a new vm every time to protect the same @xmath1 .
this new vm will be different from the previous one .
moreover everything must be dynamic .
therefore we must have simple working rules : * the bytecode must be dynamic and hidden by using the malicious cryptography tools ( malicious pnrg ) we are exposing hereafter ; * information must be dynamic ( area for each context , the offset for each register in the context ... ) ; * the code must be transformed with the internal libthor metamorphic package ( in which malicious pnrg can used used too ) ; * integer constants can be hidden with libthor internal junk package .
this is very useful to hide opcode values , register offsets or anything which represent any invariant in a program . here
again malicious prngs play a critical role . during the code generation ( mutation ) , we use the libthor metamorphic package a lot of times on a function .
we have several steps during a generation : 1 .
obfuscation of c source code .
2 . compiling the c source code of the vm into a library .
3 . extraction of interesting functions from the library .
4 . transformation of x86 assembly instructions into ir .
obfuscation of ir by using metamorphic package
transformation of ir into bytecode
creation of dynamic functions to handle the bytecode
obfuscation of functions by using the metamorphic package .
the mutated code is produced here and the malicious prng is mainly involved here .
assembly of all parts to have a simple shellcode .
our prng is essentially dedicated to the protection of the bytecode which is the final result of the transformation : x86 asm @xmath8 reil ir @xmath8 bytecode .
here is an illustrating example : .... [ x86 asm ] mov eax , 0x3 [ b803000000 ] [ reil ir ] str ( 0x3 , b4 , 1 , 0 ) , ( eax , b4 , 0 , 0 ) [ bytecodes ] 0xf1010000 0x40004 0x3 0x0 0x6a .... we have five fields in the bytecode corresponding respectively to : * 0xf1010000 * * 0xf1 : the opcode of the instruction ( str ) , * * 0x01 : specifies that it is an integer value , * * 0x00 : useless with respect to this instruction , * * 0x00 : specifies that it is a register . * 0x40004 * * 0x04 : the size of the first operand , * * 0x00 : useless with respect this instruction , * * 0x04 : the size of the third operand , * 0x3 : direct value of the integer , * 0x0 : useless with respect to this instruction , * 0x6a : value of the register . in this setting the 0x00 values ( useless with respect this instruction ) contribute directly to the transec aspect .
now that the working environment is defined , let us explain how a malicious prng can efficiently protect the opcode values while generating them dynamically .
sophisticated polymorphic / metamorphic or obfuscation techniques must rely on prng ( pseudo - random number generator ) . in the context of this paper ,
the aim is to generate sequences of random numbers ( here bytecode values ) on - the - fly while hiding the code behavior .
sequences are precomputed and we have to design a generator which will afterwards output those data .
the idea is that any data produced by the resulting generator will be first used by the code as a valid address , and then will itself seed the pnrg to produce the next random data .
three cases are to be considered : 1 .
the code is built from any arbitrary random sequence ; 2 .
the sequence is given by a ( non yet protected ) instance of bytecode and we have to design an instance of pnrg accordingly ; 3 .
a more interesting problem lies in producing random data that can be somehow interpreted by a prng as meaningful instructions like jump 0x89 directly .
this relates to interesting problems of prng cryptanalysis .
we are going to address these three cases . from a general point of view
it is necessary to recall that for both three cases the malware author needs reproducible random sequences . by reproducible ( hence the term of pseudo - random ) , we mean that the malware will replay this sequence to operate its course of actions .
the reproducibility condition implies to consider a _ deterministic finite - state machine _ ( dfsm ) .
the general scheme of how this dfsm is working is illustrated as follows . without the dfsm , any instruction data whenever executed produced a data used the next instruction ( e.g. an address , an operand ... ) .
@xmath9 any analysis of the code will easily reveal to the malware analyst all the malware internals since all instructions are hardcoded .
but if a few data / instructions are kept under an encrypted form , and deciphered at execution only , the analysis is likely to be far more difficult ( up to decryptor and the secret key protection issue ) .
it is denied of a priori analysis capabilities .
so we intend to have @xmath10 where @xmath11 for all i. upon execution , we just have to input data @xmath12 into the dfsm which will then output the data @xmath13 .
a few critical points are worth stressing on 1 .
no key is neither required nor used ; 2 .
instructions can similarly protected as well .
of course to be useful as a prevention tool against analysis , the dfsm must itself be obfuscated and protected against analysis . but this last point is supposed to be fulfilled .
we will show a more powerful implementation by using @xmath14-ary malware in section [ imp_k ] . in this case
, the sequence is arbitrary chosen before the design of the code and hence the code is written directly from this arbitrary sequence .
this case is the most simple to manage .
we just have to choose carefully the dfsm we need .
one of the best choice is to take a congruential generator since it implies a very reduced algorithm with simple instructions .
let us consider @xmath15 an initial value and the corresponding equation @xmath16 where @xmath17 is the multiplier , @xmath18 is the increment and @xmath19 is the modulus . since the length of the sequence involved in the malware design is rather very short ( up to a few tens of bytes ) , the choice of those parameters is not as critical as it would be for practical cryptographic applications . in this respect
, one can refer to knuth s seminal book to get the best sets of parameters @xcite .
here are a few such examples among many others : standard minimal generator : : @xmath20 .
vax - marsaglia generator : : @xmath21 .
lavaux & jenssens generator : : @xmath22 .
haynes generator : : @xmath23 .
kuth s generator : : @xmath24 and @xmath25 .
of course the choice of the modulus is directly depending on the data type used in the malware .
another interesting approach is to consider hash functions and s / key .
the principle is almost the same .
we take a @xmath26 hash function @xmath27 which produces a @xmath28-bit output from a @xmath29-bit input with @xmath30 . in our case
we can build @xmath29 in the following way ....
m = < data to protect><padding of random data><size of data > .... or equivalently ....
m = d_i < random data > |d_i| .... then we choose a @xmath29-bit initialization vector ( iv ) and we compute the random sequence as follows @xmath31 the iteration value @xmath32 can be used to get one or more required arbitrary value thus anticipating the next case .
of course the nature of the hash function is also a key parameter : you can either use existing hash function ( e.g md5 , sha-1 , ripemd 160 , sha-2 ... ) and keep only a subset of the output bit ; or you can design your own hash function as explained in @xcite . in this slightly different case ,
the sequence is determined by a ( non yet protected ) instance of a code .
this issue is then to design or use an instance of prng accordingly .
this is of course a far more difficult issue which implies cryptanalytic techniques . to formalize the problem we have a sequence @xmath33 which represents critical data ( addresses , asm instructions , operands ... ) of a particular instance of a ( malware ) code . as for example let us consider three series of 32-bit integers describing bytecode values as defined in section [ cs ] : .... 0x2f010000
0x040004 0x3 0x0 0x89 ( 1 ) 0x3d010000 0x040004 0x3 0x0 0x50 ( 2 ) 0x5010000 0x040004 0x3 0x0 0x8d ( 3 ) .... they are just different instances of the same instruction @xcite .
the aim is to have these data in the code under a non hardcoded form .
then we intend to code them under an obfuscated form , e.g. @xmath34 we then have to find a dfsm such that @xmath35 the notation @xmath36 directly suggests that the quantity input to the dfsm is a key in a cryptographic context but these keys have to exhibit local low entropy profile at the same time .
so the malicious prng must take this into account as well . in this case
, we have to face a two - fold cryptanalytic issue : * either fix the output value @xmath37 and find out the key @xmath36 which outputs @xmath37 for an arbitrary dfsm , * or for an arbitrary set of pairs @xmath38 design a unique suitable dfsm for those pairs .
the first case directly relates to a cryptanalytic problem while the second refers more to the problem of designing cryptographic dfsms with trapdoors . in our context of malicious cryptography ,
the trapdoors here are precisely the arbitrary pairs of values @xmath38 while the dfsm behaves for any other pair as a strong cryptosystem @xcite .
this second issue is far more complex to address and will not be exposed in this paper ( research under way ; to be published later ) .
we will focus on the first case which has been partially addressed for real - life cryptosystem like _ bluetooth _
e0 @xcite in the context of zero knowledge - like proof of cryptanalysis .
but in our case we do not need to consider such systems and much simpler dfsm can be built conveniently for our purposes : sequences of data we use are rather short .
we have designed several such dfsms and we have proceeded to their cryptanalysis to find out the keys @xmath36 which output the values @xmath37
. as we will see , those dfsm have to exhibit additional features in order to * be used for code mutation purposes , * exhibit transec properties . in other words , if we have @xmath39 , then @xmath1 and @xmath40 must have the same entropy profile .
replacing @xmath1 with a @xmath40 having a higher entropy profile would focus the analyst s attention . without loss of generality , let us consider the mathematical description of a 59-key bit stream cipher ( we currently work on block cipher which should offer more interesting features ; to be continued ) we have designed for that purpose .
other dfsms with larger key size ( up to 256 ) have been also designed ( available upon request ) .
the principle is very simple yet powerful : three registers @xmath41 and @xmath42 are filtered by a boolean function @xmath43 , thus outputing bits @xmath44 ( or bytes in a vectorized version ) that are combined to the plaintext ( figure [ combsys ] ) . the value @xmath36 initializes the content of registers @xmath41 and @xmath42 at time instant @xmath45 , and outputs bits @xmath46 which represent the binary version of values @xmath37 .
the linear feedback polynomials driving the registers are the following : @xmath47 where @xmath48 denotes the exclusive or .
the combination function is the _ majority function
_ applied on three input bits and is given by @xmath49 an efficient implementation in c language is freely available by contacting the author . for our purposes , we will use it in a procedure whose prototype is given by ....
void sco(unsigned long long int * x , unsigned long long int k ) { / * k obfuscated value ( input ) , x unobfuscated value ( output ) * / / * ( array of 8 unsigned char ) by sco * / ... } .... now according to the level of obfuscation you need , different ways exist to protect your critical data inside the code ( series of integers ( 1 ) , ( 2 ) and ( 3 ) above ) .
we are going to detail two of them .
the dfsm outputs critical data under a concatenated form to produce chunks of code corresponding to the exact entropy of the input value @xmath50 , thus preventing any local increase of the code entropy . for the dfsm
considered , it means that we output series ( 1 ) , ( 2 ) and ( 3 ) under the following form ....
1)-- > 0x2f01000000040004000000030000000000000089
2)-- > 0x3d01000000040004000000030000000000000050 3)-- > 0x050100000004000400000003000000000000008d .... let us detail the first output sequence ( 1 ) .
it will be encoded as three 59-bit outputs @xmath51 and @xmath52 ....
m_1 = 0x0bc04000000ll ; m_2 = 0x080008000000060ll ; m_3 = 0x000000000000089ll ; .... to transform @xmath51 and @xmath52 back into five 32-bit values @xmath53 and @xmath54 , we use the following piece of code : .... / * generate the m_i values * / sco(&m_1 , k_1 ) ; sco(&m_2 , k_2 ) ; sco(&m_3 , k_3 ) ; x_1 = m_1 > > 10 ; / * x_1 = 0x2f010000l * / x_2 = ( ( m_2 > > 37 ) | ( m_1 < < 22 ) ) & 0xffffffffl / * x_2 = 0x00040004l * / x_3 = ( m_2 > > 5 ) & 0xffffffffl ; / * x_3 = 0x3 * / x_4 = ( ( m_3 > > 32 ) | ( m_2 < < 27 ) ) & 0xffffffffl ; / * x_4 = 0x0 * / x_5 = m_3 & 0xffffffffl ; / * x_5 = 0x89 * / .... values @xmath51 and @xmath52 will be stored in the code as the values @xmath55 and @xmath56 with @xmath57 : .... k_1 = 0x6aa006000000099ll ; k_2 = 0x500403000015dc8ll ; k_3 = 0x0e045100001eb8all ; .... similarly we have for sequence ( 2 ) .... m_1 = 0x0f404000000ll ; k_1 = 0x7514360000053c0ll ; m_2 = 0x080008000000060ll ; k_2 = 0x4c07a200000a414ll ; m_3 = 0x000000000000050ll ; k_3 = 0x60409500001884all ; .... and for sequence ( 3 ) .... m_1 = 0x01404000000ll ; k_1 = 0x76050e00001f0b1ll ; m_2 = 0x080008000000060ll ; k_2 = 0x00000010c80c460ll ; m_3 = 0x00000000000008dll ; k_3 = 0x000000075098031ll ; .... the main interest of that method is that the interpretation of code is not straightforward .
code / data alignment does not follow any logic ( that is precisely why we have chosen a 59-bit fsm which is far better that a seemingly more obvious 64-bit fsm ; any prime value is optimal ) .
moreover , as we can notice , the @xmath36 values are themselves sparse as unobfuscated opcodes are ( structural aspect ) .
additionally , their entropy profile ( quantitative aspects ) is very similar to the @xmath58 values ( and hence the @xmath37 ones ) .
this implies that any detection techniques based on local entropy picks is bound to fail . due to the careful design of our 59-bit dfsm
, we managed to obtain a unicity distance which is greater than 59 bits ( the unicity distance is the minimal size for a dfsm output to be produced by a single secret key ) . in our case a large number of different 59-bit keys can output an arbitrary output sequence .
here are the results for the three series ( 1 ) , ( 2 ) and ( 3 ) ( table [ res ] ) : .number of possible keys for a given output value [ cols="^,^,^",options="header " , ] this implies that we can randomly select our 9 @xmath58 values and thus we have @xmath59 different possible code variants .
it is approximatively equal to @xmath60 variants .
the different files ( @xmath61 with @xmath62 ) whose name is given in right column of table [ res ] are freely available upon request .
to build a variant , you just have to select data in the nine files randomly according to the following piece of code ( code extract to generate the @xmath63 values for the first serie ( 1 ) only ; refer to section [ imp_k ] for a secure implementation ) : .... f1 = fopen("res11","r " ) ; f2 = fopen("res12","r " ) ; f3 = fopen("res13","r " ) ; randval = ( 314.0*(rand()/(1 + rand_max ) ) ; for(i = 0 ; i < randval ; i++ ) fscanf(f1 , % lx % lx % lx\n , & y1,&y2 , & y3 ) ; k_1 = y1 | ( y2 < < 17 ) | ( y3 < < 36 ) ; / * do the same for values m_2 and m_3 of serie ( 1 ) * / .... / * repeat the same for series ( 2 ) and ( 3 ) * / .... / * generate m_1 value for series(1 ) * / sco(&m_1 , k_1 ) ; .... the different files ( @xmath61 with @xmath62 ) can be stored in a program whose size is less than 400 kb . in this second case ,
the dfsm outputs 59-bit chunks of data whose only the 32 least significant bits are useful .
in this case we output five 59-bit chunks of data @xmath64 and @xmath65 .
for sequence ( 1 ) we have .... m_1 = 0x???????2f010000ll ; m_2 = 0x???????00040004ll ; m_3 = 0x???????00000003ll ; m_4 = 0x???????00000000ll ; m_5 = 0x???????00000089ll ; .... where ` ? ` describes any random nibble .
we get the @xmath37 values from the @xmath58 values with @xmath66 the main interest of that method is that it naturally and very simply provides increased polymorphism compared to the previous approach . indeed about @xmath0 5-tuples @xmath67 whenever input in our dfsm produces 5-tuples @xmath68
. then we can produce a huge number of different instances of the same code by randomly choosing any possible 5-tuples . by increasing size of the memory of the fsm
we even can arbitrarily increase the number of possible polymorphic instances .
the obfuscation techniques we have presented before , which are based on malicous cryptography ( malicious , biased prng ) and cryptanalysis techniques ( to precompute inputs to the dfsm ) require obviously to protect the dfsm itself very strongly . indeed obtaining the @xmath37 values from the @xmath36
is straightforward whenever we have the dfsm .
we would then come back to the weak existing implementations and reduce the problem to a simple decoding scheme .
but this obfuscation is impossible to reverse in the context of @xmath7-ary malware @xcite . in this case , the viral information is not contained in a single code as usual malware do , but it is split into @xmath7 different innocent - looking files whose combined action -
serially or in parallel - results in the actual malware behavior and in the metamorphic code instance generation .
just imagine a 2-ary code ( we can take of course @xmath69 ) made of parts @xmath70 and @xmath71 .
part @xmath70 just embeds the dfsm and wait in memory for values @xmath36 coming from part @xmath71 .
its role is to compute @xmath72 values and send them back to part @xmath71 according to the protocol summarized by figure [ fig2 ] .
the interpretation of data by the reverser ( and hence the reversing of the code ) is no longer possible since the dfsm is deported in the part @xmath70 and the analysis of part @xmath71 would require to reconstruct the dfsm and to break it .
this is of course impossible since the code does not contain a sufficient quantity of encrypted information . in the case of metamorphism
, the part @xmath70 will output a random @xmath36 value instead during the code instance generation . from a practical point of view
, we have considered different implementations .
communication pipes : : due to the relationship between father and children processes , only parallel class a or c @xcite k - ary codes can be implemented .
it is not the most optimal solution .
named communication pipes : : in this case , k - ary parallel class b codes can be efficiently implemented ( the most powerful class since there is no reference in any part [ hence information ] to other any part ) .
system v ipc : : this is the most powerful method since everything is located into shared memory
. the source codes of our implementation will be soon publicly available . in the context of code protection for legitimate purpose , the part @xmath71 will be a kernel - land module can be user / host specific .
it can also be a program located on a server outside the code analyst s scope .
we have shown in this paper that it is possible to prevent code analysis from reversing while ensuring a high level of metamorphism . of course , we have to apply those techniques to all critical parts of the code ( especially those which determine the execution flow ) . in this case
it is obvious that static analysis is no longer possible . as far as dynamic analysis is concerned , the analyst has to perform debugging and sandboxing . but using delay detection technique @xcite can trigger a different code behaviour and code mutation .
our current research and experimentation work is related to non deterministic fsms . in this case
any internal state of the fsm results in many possible outcome ( next state at time instant @xmath73 ) .
even if is likely to be far beyond any av software detection capability , human analysis becomes impossible .
non deterministic contexts result in untractable complexity .
thanks to anthony desnos for his support with respect to virtual machines and ir and for many other reasons : geoffroy gueguen ( and his awful vanwinjgaarden formal grammars ) , eloi vanderbken ( aka _ baboon _ ) for his funny way of thinking security .
50 b. barak , o. goldreich , r. impagliazzo , s. rudich , a. sahai , s. vadhan and k. yang ( 2001 ) . on the ( im)possibility of obfuscation programs . in : _ advances in cryptology - crypto
2001 _ , lncs 2139 , pp . 1 - 8 , springer verlag .
philippe beaucamps and eric filiol ( 2006 ) . on the possibility of practically obfuscating programs - towards a unified perspective of code protection .
_ journal in computer virology _
, 2(4 ) , wtcv06 special issue , g. bonfante editor , springer verlag .
anthony desnos ( 2009 ) .
implementation of @xmath14-ary viruses in python .
hack.lu 2009 conference .
anthony desnos ( 2010 ) .
dynamic , metamorphic ( and open source ) virtual machines . to appear in the _ proceedings of the hack.lu 2010 conference _ , luxembourg .
thomas dullien and sebastian porst ( 2009 ) .
reil : a platform independent intermediate representation of disassembled code for static code analysis .
cansecwest 2009 , vancouver , canada .
available on http://www.zynamics.com/downloads/csw09.pdf robert erra and christophe grenier ( 2009 ) . how to choose rsa keys ( the art of rsa : past , present and future ) ?
iawacs 2009 conference .
available on http://www.esiea-recherche.eu/iawacs_2009_papers.html eric filiol .
strong cryptography armoured computer viruses forbidding code analysis : the bradley virus . in : _
proceedings of the 14th eicar conference _ , pp.- 201217 , may 2005 , malta .
eric filiol ( 2005 ) .
_ computer viruses : from theory to applications _ , iris international series , springer verlag france , isbn 2 - 287 - 23939 - 1
eric filiol ( 2007 ) . _
techniques virales avances _ , collection iris , springer verlag france , isbn 978 - 2 - 287 - 33887 - 8 .
eric filiol and sbastien josse . a statistical model for viral detection undecidability .
eicar 2007 special issue , v. broucek ed .
, _ journal in computer virology _ , 3 ( 2 ) .
springer verlag .
eric filiol , edouard franc , alessandro gubbioli , benoit moquet and guillaume roblot ( 2007 ) .
combinatorial optimisation of worm propagation on an unknown network .
_ international journal in computer science _ , 2 ( 2 ) , pp .
124130 , 2007 .
eric filiol ( 2007 ) .
formalisation and implementation aspects of @xmath14-ary ( malicious ) codes .
_ journal in computer virology _ ,
3 , issue 3 .
springer verlag .
eric filiol ( 2007 ) .
zero knowledge - like proof of cryptanalysis of bluetooth encryption . _
international journal in information theory _ , 3 ( 4 ) , pp .
40ff , http://www.waset.org/journals/ijit/v3/v3-4-40.pdf eric filiol and fred raynal ( 2008)._malicious cryptography ... reloaded - and also malicious statistics_. cansecwest conference , vancouver , canada , march 2008 .
available on http://cansecwest.com/csw08/csw08-raynal.pdf eric filiol .
the malware of the future : when mathematics are on the bad side .
_ opening keynote talk _ , hack.lu 2008 , luxembourg , october 2008 .
available on the hack.lu website .
eric filiol ( 2010 ) anti - forensics techniques based on malicious cryptography . in : _ proceedings of the 9th european conference in information warfare eciw 2010 _ , thessaloniki , greece .
eric filiol ( 2010 ) dynamic cryptographic trapdoors . submitted .
shafi goldwasser and guy n. rothblum ( 2007 ) . on best - possible obfuscation . in : _ proceedings of the 4th theory of cryptography conference - tcc07_. lncs 4392 ,
194213 , springer verlag .
donald e. knuth ( 1998 ) .
_ the art of computer programming : seminumerical algorithms _ , volume 2 , addison - wesley .
libthor website ( 2010 ) _ playing with virology for fun and profit _ http://libthor.avcave.org .
available on october 2010 .
perseus homepage .
james riordan and bruce schneier ( 1998 ) . environmental key generation towards clueless agents . in : _ mobile agents and security _ , g. vigna ed . ,
lecture notes in computer science , pp . 1524 , springer verlag .
adam young and moti yung ( 1999 ) . _ malicious cryptography - exposing cryptovirology_. wiley .
isbn 0 - 7645 - 4975 - 8 . | fighting against computer malware require a mandatory step of reverse engineering .
as soon as the code has been disassemblied / decompiled ( including a dynamic analysis step ) , there is a hope to understand what the malware actually does and to implement a detection mean .
this also applies to protection of software whenever one wishes to analyze them . in this paper , we show how to amour code in such a way that reserse engineering techniques ( static and dymanic ) are absolutely impossible by combining malicious cryptography techniques developped in our laboratory and new types of programming ( k - ary codes ) .
suitable encryption algorithms combined with new cryptanalytic approaches to ease the protection of ( malicious or not ) binaries , enable to provide both total code armouring and large scale polymorphic features at the same time .
a simple 400 kb of executable code enables to produce a binary code and around @xmath0 mutated forms natively while going far beyond the old concept of decryptor . |
the recent developments in the cooling and trapping techniques of neutral atoms have opened the way to the realization of fermionic atomic samples at temperatures well below the degeneracy temperature @xcite .
this suggests that atomic gases are ideal candidates for the study of the physics of degenerate many - fermion systems . with respect to solid state ones ,
atomic systems offer in fact a better isolation from external disturbances such as material defects , a better knowledge of the microscopic details of the systems , as well as a wider range of tunability of the parameters , in particular the interparticle interactions . by tuning the external magnetic field around a feshbach resonance ,
the atom - atom scattering length @xmath0 can be varied from @xmath1 to @xmath2 ( @xmath3 being the fermi momentum ) opening the way towards a comprehensive study of the pairing transition both in the regime @xmath4 in which a bose - condensate ( bec ) of tightly - bound molecules is present , and in the regime @xmath5 ( bcs ) in which a condensate of cooper pairs is formed .
diatomic molecules have been created and observed by several experimental groups @xcite .
bose - einstein condensation of tightly bound diatomic molecules has been recently reported @xcite .
the crossover region between bec and bcs is currently under experimental investigation @xcite and first evidences of pairing in the crossover region have been reported in @xcite .
from the theoretical point of view , a large effort is currently made to establish the main features of the pairing for high values of the scattering length @xmath6 , regime in which the atomic gas shows strong correlations @xcite .
in particular , the dependance of the transition temperature on the interaction strength in this crossover region is still an open problem .
the present paper reports a numerical study of the condensation of pairs in a regime of relatively strong interactions , so to characterize the consequences of the transition on the different observables of the system and identify specific features which may represent unambiguous signatures of the onset of condensation of pairs .
the calculations have been performed by applying the quantum monte carlo ( qmc ) method developed in @xcite to a one - dimensional lattice model of fermions with attractive on - site interactions .
a short description of the model under examination is given in sec .
[ sec : model ] , while the numerical algorithm used for the calculations is presented in sec .
[ sec : qmc ] .
numerical results are presented in sec .
[ sec : results ] and then extensively compared to the predictions of a perturbative expansion in the interaction coupling constant ( sec .
[ sec : pert ] ) , and of existing approximate approaches ( sec.[sec : exist ] ) , such as the bcs theory @xcite , two versions of the random phase approximation ( rpa ) @xcite as well as the nozires schmitt - rink theory @xcite .
several among the most relevant correlation functions of the fermi gas have been considered here , in particular the opposite - spin density - density correlation function @xmath7 , the first - order pair coherence function @xmath8 and the second - order momentum space correlation function @xmath9 .
the density - density correlation function has been already the object of several papers studying the experimental signatures of the bcs transition in atomic fermi systems , e.g. @xcite , while the first - order pair coherence function is the counterpart , in a non - symmetry - breaking approach , of the order parameter of the phase transition in a landau - ginzburg theory @xcite .
a one - dimensional low energy two - component fermi gas can be modeled by the hamiltonian : @xmath10 the spatial coordinate @xmath11 runs on a discrete lattice of @xmath12 points with periodic boundary conditions ; @xmath13 is the total length of the quantization box and @xmath14 is the length of the unit cell of the lattice .
the spin index runs over the two @xmath15 spin states .
the system is taken as spatially homogeneous , @xmath16 is the atomic mass , and interactions are modeled by a two - body discrete delta potential with a coupling constant @xmath17 .
the field operators @xmath18 satisfy the usual fermionic anticommutation relations @xmath19 and can be expanded on plane waves according to @xmath20 with @xmath21 restricted to the first brillouin zone of the reciprocal lattice . in order for the discrete model to correctly reproduce the underlying continuous field theory , the grid spacing @xmath22 must be smaller than all the relevant length scales of the system , e.g. the thermal wavelength and the mean interparticle spacing . in the present one - dimensional case ,
the relation between the coupling constant on the lattice and the physical 1d coupling constant @xmath23 is : @xmath24 which , in the limit @xmath25 reduces to the expected one @xmath26 @xcite .
this condition is satisfied in the monte carlo simulations presented in this paper .
we also note that two particles interacting in free space with a attractive delta potential in 1d have a bound state of energy @xmath27 . in the numerical examples of this paper ,
the fermi energy is much larger than this binding energy so that we are not investigating the condensation of preformed pairs but rather a bcs regime .
we assume the gas to be at thermal equilibrium at a temperature @xmath28 in the canonical ensemble , so that the unnormalized density operator @xmath29 with @xmath30 . from textbook statistical physics , we know that such a density operator can be obtained by means of an imaginary - time evolution : @xmath31 { \label{eq : imagtev}}\ ] ] during a `` time '' interval @xmath32 starting from the initial state corresponding to the infinite temperature case where @xmath33 , @xmath34 being the identity matrix in the @xmath35-body hilbert space .
as it has been recently shown in @xcite , the exact solution of the imaginary - time evolution ( [ eq : imagtev ] ) can be written as a statistical average of hartree - fock dyadics of the form : @xmath36 for @xmath37 , @xmath38 ( @xmath39 ) are hartree - fock orbitals for the @xmath35 fermions , in the sense that : @xmath40 the creation operator corresponding to the wavefunction @xmath41 being defined as : @xmath42 for the model hamiltonian ( [ eq : hamilt ] ) , the imaginary - time evolution of each of the orbitals @xmath38 can be reformulated in terms of ito stochastic differential equations of the form : @xmath43 + \\ -\frac{g_0}{2}\sum_j\sum_{x',\sigma'}dx'\ , \frac{1}{\|\phi_j{^{(\alpha)}}\|^2\,\|\phi_i{^{(\alpha)}}\|^2}\ , \big [ \phi_i^{(\alpha)*}(x',\sigma')\,\phi_j^{(\alpha)*}(x',-\sigma')\ , \phi_j{^{(\alpha)}}(x',-\sigma')\,\phi_i{^{(\alpha)}}(x',\sigma')+ \\ \left .
-\phi_i^{(\alpha)*}(x',\sigma')\,\phi_j^{(\alpha)*}(x',-\sigma')\ , \phi_j{^{(\alpha)}}(x',\sigma')\,\phi_i{^{(\alpha)}}(x',-\sigma ' ) \big]\,\phi_i{^{(\alpha)}}(x,\sigma)\right\}+db_i{^{(\alpha)}}(x,\sigma),\end{gathered}\ ] ] where @xmath44 represents the momentum operator on the grid and the norm @xmath45 is defined as @xmath46 .
the deterministic part is simply the mean - field hartree - fock equation in imaginary time , while the correlation functions of the zero - mean noise @xmath47 are given by : @xmath48 .
{ \label{eq : noise}}\ ] ] the projector @xmath49 projects orthogonally to the subspace spanned by the wavefunctions @xmath50 .
a possible noise with the required correlation function ( [ eq : noise ] ) is : @xmath51 with @xmath52 independent zero - mean gaussian noises with @xmath53 , @xmath54 .
it can be proven @xcite that this set of stochastic differential equations reproduces , in the average over the noise , the exact evolution of the hartree - fock dyadic @xmath55 during @xmath56 : @xmath57.\ ] ] the initial state @xmath58 can be written as a functional integral over all possible sets of orthonormal wavefunctions @xmath59 ( @xmath39 ) : @xmath60 this writing of the identity operator can be used as a starting point for an exact simulation of the fermionic many - body problem . to this purpose
, we have to numerically solve the stochastic differential equations ( [ eq : ito_phi ] ) for imaginary times going from @xmath61 to @xmath62 .
this is done by splitting the imaginary - time interval into a large enough number @xmath63 of time steps ; @xmath64 is the noise terms at the time - step @xmath65 ( @xmath66 ) on the site @xmath11 .
the expectation values of any observable at temperature @xmath28 is then obtained as an average over all the possible values of the initial wavefunctions @xmath67 and the elementary noise terms @xmath64 .
for example , the partition function @xmath68 $ ] is obtained as : @xmath69=\overline { { \langle \phi{^{(2)}}_1\ldots\phi{^{(2)}}_n | \phi{^{(1)}}_1\ldots\phi{^{(1)}}_n \rangle } } \ ] ] or , equivalently , as the determinant @xmath70 $ ] of the matrix @xmath71 whose entries are @xmath72 . in the following , we shall be mainly interested in the one- and two - body correlation functions of the gas . by making use of the jacobi theorem @xcite ,
these can be usefully rewritten in the following compact forms : @xmath73\,\sum_{ij}\,\big(m^{-1}\big)_{ij}\,\phi^{(2)*}_i(x,\sigma)\ , \phi^{(1)}_j(x',\sigma')\ ] ] and @xmath74\,\sum_{ijkl } \,\textrm{det}\left [ \begin{array}{cc } \big(m^{-1}\big)_{il } & \big(m^{-1}\big)_{ik } \\ \big(m^{-1}\big)_{jl } & \big(m^{-1}\big)_{jk } \\ \end{array } \right ] \,\phi^{(2)*}_i(x,\sigma)\,\phi^{(2)*}_j(x',\sigma')\ , \phi^{(1)}_k(x'',\sigma'')\ , \phi^{(1)}_l(x''',\sigma''')\end{gathered}\ ] ] in a practical simulation , the averages are performed by means of monte carlo techniques .
a description of the details of the numerical algorithm used is given in appendix [ app : mc ] .
a monte carlo code based on the stochastic approach described in the previous section has been used to numerically compute the expectation values of some one- and two - body correlation functions for a one - dimensional fermi gas with attractive binary interactions as described by the hamiltonian ( [ eq : hamilt ] ) with @xmath75 .
the results of analogous calculations performed with a very similar monte carlo algorithm have been reported recently in @xcite .
other quantum monte carlo schemes have also been applied to the numerical study of the fermionic hubbard model with attractive interactions at finite temperature . in particular , the determinantal qmc algorithm @xcite has been used to study the correlation functions in 2d @xcite and the transition temperature to a pair condensate state in 2d @xcite and in 3d @xcite . for our simulations ,
a lattice of @xmath76 points was taken , with a total number of @xmath77 atoms .
a number @xmath63 of imaginary - time steps comprised between @xmath78 and @xmath79 has been used .
as already mentioned , the ensemble in which observables are calculated is the canonical one ; note that the number of particles in each of the spin state can fluctuate , only the total number of particles is fixed .
as the two spin components are equivalent , the mean densities in each of the spin components are equal : @xmath80 the state of the gas in the absence of interactions and at @xmath81 is depicted in fig .
[ fig : schema ] : in a given spin component , the 5 lowest - lying single particle energy levels are totally filled , whereas the two degenerate states of wavevectors @xmath82 and @xmath83 are half - filled .
more precisely , 10 atoms are frozen in the states of @xmath84 , and the two remaining atoms are distributed among the 4 degenerate states , @xmath85 , which can be done in @xmath86 different ways . in presence of attractive interactions ,
this degeneracy will obviously be lifted and the configurations with one atom @xmath87 and one atom @xmath88 with opposite momenta in the degenerate multiplicity are favorable to the formation of a cooper pair . atoms and @xmath76 grid points .
each mode is a plane wave with a wavevector @xmath89 and an energy @xmath90 , where the integer @xmath91 ranges from @xmath92 to @xmath93 .
the modes with @xmath94 are totally filled , whereas the modes with @xmath95 are half - filled , the higher energy modes being empty .
@xmath96 is the fermi energy .
the energies are here in units of @xmath97 .
[ fig : schema],width=377 ] for different values of the temperature @xmath98 ( circles , squares , diamonds ) .
@xmath77 atoms on a @xmath76 points lattice .
coupling constant @xmath99 .
[ fig : g1uumc],width=377 ] the simplest observable to compute is the one - body correlation function in a single spin state @xmath55 ( normalized to the density @xmath100 ) : @xmath101 the monte carlo prediction is plotted in fig.[fig : g1uumc ] for different values of the temperature : as expected , this correlation function is short - ranged , coherence extending only on a length of the order of the fermi length @xmath102 for @xmath103 .
this correlation function is indeed the fourier transform of the momentum distribution of the gas .
as the interactions affect the momentum distribution only in a thin region around the fermi surface ( the fermi points in our one - dimensional geometry ) , they do not significantly modify its shape as compared to the ideal fermi distribution .
because of the rotational symmetry of the density operator in the spin space , the one - body correlation function in different spin states : @xmath104 is instead always identically vanishing .
density - density correlation functions are another observable of interest . both the single - spin density - density correlation function : @xmath105 and the opposite - spin one : @xmath106 with @xmath107 , have been calculated by monte carlo and plotted as a function of @xmath11 respectively in fig.[fig : g2mc]a and in fig.[fig : g2mc]b . in fig.[fig : g2udc_t ] , we have plotted @xmath108 as a function of temperature .
the magnitude of actual density correlations is quantified by the difference @xmath109 . and @xmath110 for different values of the temperature @xmath98 ( circles , squares , diamonds ) .
same system parameters as in fig.[fig : g1uumc ] .
[ fig : g2mc],title="fig:",width=302 ] and @xmath110 for different values of the temperature @xmath98 ( circles , squares , diamonds ) .
same system parameters as in fig.[fig : g1uumc ] .
[ fig : g2mc],title="fig:",width=302 ] : @xmath111 as a function of the temperature @xmath28 in the canonical ensemble .
circles : monte carlo results . dotted ,
short dashed , long dashed , solid lines : perturbative results upto order respectively 0 , 1 , 2 , and 3 .
same system parameters as in fig.[fig : g1uumc ] .
the vertical line is the somewhat arbitrary lower bound on the temperature range where perturbation theory converges rapidly .
[ fig : g2udc_t],width=377 ] on one hand , the density correlations in a single spin state described by @xmath112 show a short - range hole ( pauli hole ) of width similar to the bump of the one - body correlation function @xmath113 and are weakly affected by the interactions and by the temperature variations ( fig.[fig : g2mc]a ) .
on the other hand , the density correlations between opposite spins described by @xmath114 show an interesting temperature dependence in the presence of attractive interactions .
the lower is the temperature , the most effective are in fact the interactions and therefore the stronger the bunching of opposite spin particles on a given lattice site . in fig.[fig : g2mc]b we have plotted the spatial profile of @xmath115 for different values of the temperature : for the lowest value of @xmath116 , notice not only the increase of @xmath117 , but also the appearance of oscillations as a function of @xmath11 . as we shall see in the next subsection , at this temperature a condensate of pairs is present .
the oscillations then result from the contribution of two distinct effects : the friedel oscillations in the correlation functions of the normal phase which follow from the sharpness of the fermi surface @xcite , and the oscillations shown by the cooper pair wavefunction described within the bcs theory by the pairing function @xmath118 . in fig.[fig :
g2udc_t ] we have summarized the values of @xmath108 as a function of the temperature . notice that @xmath119 is appreciable already at the highest temperature considered in fig.[fig : g2udc_t ] , which , as we shall see in the next subsection , is much higher than the critical temperature @xmath120 for the appearance of long - range order .
it is believed in statistical physics that the superfluid transition in two - component fermi systems with attractive binary interactions is related to the appearance of long - range order in the so - called _ anomalous _ averages @xcite . in symmetry breaking theories such as the bcs one , this feature corresponds to a non - vanishing value for the gap function defined as : @xmath121 which plays the role of the order parameter of the phase transition in a ginzburg - landau approach . in number conserving approaches , quantities like ( [ eq : anomalous ] ) are zero .
the phase transition however still appears in the long - range behaviour of correlation functions of the form : @xmath122 a similar criterion was used in @xcite to determine the transition temperature .
a simple physical interpretation of @xmath123 can be provided as the first order correlation function of pairs : the operator @xmath124 annihilates in fact a pair of particles in opposite spin states at the spatial position @xmath125 and the operator @xmath126 creates them back at @xmath11 .
this correlation function is therefore formally equivalent to the first order coherence function of a composite boson formed by a pair of fermions with opposite spins . from this point of view
, the non - vanishing long - range limit of @xmath127 is a signature of a quantum condensation of pairs . for two different temperatures @xmath128 ( left panel ) and @xmath129 ( right panel ) in the canonical ensemble .
circles : monte carlo results . dotted , dashed , solid lines in left panel : perturbative results upto order respectively 0 , 1 and 2 ( orders 1 and 2 are undistinguishable ) .
dotted , short dashed , long dashed , solid lines in right panel : perturbative results upto order respectively 0 , 1 , 2 , and 3 .
same system parameters as in fig.[fig : g1uumc ] .
[ fig : g1pairmc],title="fig:",width=302 ] for two different temperatures @xmath128 ( left panel ) and @xmath129 ( right panel ) in the canonical ensemble .
circles : monte carlo results . dotted , dashed , solid lines in left panel : perturbative results upto order respectively 0 , 1 and 2 ( orders 1 and 2 are undistinguishable ) .
dotted , short dashed , long dashed , solid lines in right panel : perturbative results upto order respectively 0 , 1 , 2 , and 3 .
same system parameters as in fig.[fig : g1uumc ] .
[ fig : g1pairmc],title="fig:",width=302 ] pair coherence function @xmath130 as a function of the temperature @xmath28 .
circles : monte carlo results in the canonical ensemble .
solid line : bcs theory .
same system parameters as in fig.[fig : g1uumc].,width=377 ] monte carlo simulations for this quantity are shown in fig.[fig : g1pairmc ] . at low temperatures , @xmath127 has a finite value also for @xmath131 , i.e. at the largest distance from @xmath125 allowed by the finite size of the box . on the other hand , at higher temperatures , but still much lower than the fermi temperature , @xmath130 becomes very small and the long - range order is destroyed . to make this cross - over more apparent , we have plotted in fig .
[ fig : g1pairc_t ] the value of @xmath130 as a function of the temperature : a sudden rise of this quantity appears at low temperatures .
this behavior qualitatively corresponds to the one expected for a bcs transition : although a bcs transition can not occur in one dimension in the thermodynamical limit because of long wavelength fluctuations destroying the long range order @xcite , it can however be observed in our simulations because of the finite size of the system . as the system is finite ,
the transition temperature @xmath120 is not precisely defined and the long - range order has an analytic dependance on temperature .
notice that the opposite spin density - density correlation described by @xmath108 are already important at @xmath132 and for @xmath133 they only get slightly reinforced .
another observable that has been recently proposed as a possible way of detecting the transition to a pair condensate state is the second - order momentum space correlation function @xcite : @xmath134 where the operator @xmath135 gives the occupation of the plane wave @xmath21 with spin component @xmath55 . as discussed in @xcite
, bcs theory predicts that correlations should be absent above @xmath136 , that is @xmath137 , while the transition to a condensate state should be observable as the appearance of a non - vanishing value of @xmath138 , sharply peaked around @xmath139 .
second - order momentum space correlation function @xmath140 for different values of the temperature .
empty circles , squares , diamonds : monte carlo results for @xmath141 .
filled circles and squares at @xmath142 : perturbative expansion up to order 3 for @xmath143 respectively . for @xmath144 a perturbative expansion to an order higher than 3
would be required to observe convergence .
, width=377 ] second - order momentum space correlation function at @xmath139 : @xmath145 as a function of the temperature @xmath28 .
circles : monte carlo results in the canonical ensemble .
solid line : bcs theory in the grand canonical ensemble . dashed line : ideal gas in the canonical ensemble .
same system parameters as in fig.[fig : g1uumc].,width=377 ] in fig.[fig : g2kkmc ] , we have plotted monte carlo results for @xmath138 as a function of @xmath21 for different values of the temperature . at all temperatures ,
the quantity is indeed strongly peaked at @xmath139 , and nearly vanishes at the other values .
a summary of the temperature - dependence of @xmath145 is plotted in fig.[fig : g2kkc_t ] . at temperatures above the transition temperature @xmath120 ,
correlations are negative and increase as the temperature is lowered .
the negative correlation simply follows from the fact that we are working in the canonical ensemble , that is at a fixed total number of particles ( see the ideal fermi gas result in fig.[fig : g2kkc_t ] ) .
as the temperature drops below @xmath120 , the correlations change sign becoming large and positive .
this is a signature of pairing : because of the attractive interactions , the states with a filled fermi sphere plus two particles in states of opposite momenta and spins are in fact energetically favoured . in this state ,
the fluctuations of the occupation numbers of the @xmath146 and @xmath147 states are positively correlated .
in this section , we explain how to calculate the pair coherence function @xmath148 and the density correlation function @xmath149 by means of a series expansion in powers of the coupling constant @xmath17 .
the same procedure was used in fig.[fig : g2kkmc ] to obtain a series expansion for the momentum - space second order correlation function @xmath138 although we do not give here the details of the calculation .
we expect this perturbative approach performed around the ideal fermi gas to be efficient mainly at @xmath132 that is in absence of a condensate of pairs . for @xmath133
we indeed found numerically that the series ( up to order 3 ) is slowly convergent .
note that such a series expansion can however be shown to be convergent at non - zero temperature for our model system with a finite number of modes , see below . as we wish to compare to the quantum monte carlo results
, we have in principle to perform the perturbative treatment directly in the canonical ensemble with @xmath35 particles .
the resulting averages in the ideal fermi gas thermal state are however difficult to evaluate analytically .
we therefore apply the following trick to ` canonize ' the grand canonical ensemble .
we introduce an unnormalized grand canonical thermal density operator defined as @xmath150 where @xmath151 is the total number operator , @xmath152 is an angle and @xmath153 is the chemical potential of the ideal fermi gas with an average number of @xmath35 particles . taking the fourier component of @xmath154 over the harmonic @xmath155 amounts to projecting @xmath154 over the subspace with exactly @xmath35 particles .
the canonical expectation value of an operator @xmath156 is therefore exactly given by : points has an upper limit of @xmath157 .
if one excludes the cases @xmath158 and @xmath159 , one can replace the integrals over @xmath152 by discrete sums over the values @xmath160 where the integer @xmath161 ranges from @xmath125 to @xmath162 . furthermore , the symmetry @xmath163 may be used to reduce the range of @xmath161 . ]
@xmath164 } { \int_0^{2\pi } d\theta\ , e^{-i\theta n } \mbox{tr}[\sigma_{gc}(\theta)]}. \label{eq : can_afo_gc}\ ] ] we then expand @xmath165 in powers of the interaction potential @xmath166 , here up to third order : @xmath167 \label{eq : expansion}\end{aligned}\ ] ] where @xmath168 is the kinetic energy operator of the gas and the imaginary time interaction picture for an operator @xmath169 is defined as @xmath170 for a non - zero temperature and a finite number of grid points , the norm of the operator @xmath171 is finite as both @xmath166 and @xmath172 have a finite norm . as a consequence , the norm of the @xmath173-order contribution to the series expansion eq .
( [ eq : expansion ] ) can be bounded from above by @xmath174 where @xmath175 is some number , and the series eq .
( [ eq : expansion ] ) is absolutely convergent . where @xmath176 is the difference between the largest and the smallest eigenvalues of @xmath177 and where the norm of @xmath166 is @xmath178 . ]
the calculation of the numerator of eq .
( [ eq : can_afo_gc ] ) then involves the @xmath152-dependent grand canonical partition function of the ideal fermi gas and @xmath152-dependent expectation values in the grand canonical ideal fermi gas : @xmath179 = \prod_{k } \left(1+e^{-\beta[\hbar^2k^2/(2m)-\mu_0]}e^{i\theta}\right)^2 \\
\langle x\rangle_0(\theta ) & \equiv & \frac{1}{\xi_0(\theta)}\mbox{tr}[\sigma_{\rm gc}^0(\theta)x ] \end{aligned}\ ] ] where the square originates from the presence of two spin components .
the operator @xmath169 is one of the terms inside the square brackets of eq.([eq : expansion ] ) .
the expectation values can be evaluated using wick s theorem and involve the following particle and hole correlation functions : @xmath180 the explicit expressions of the relevant expectation values in terms of @xmath181 and @xmath182 are given in the appendix [ appen : expect ] .
the integrals over the ` times ' @xmath183 and the sums over the grid points associated to each factor @xmath184 are performed numerically . as each integral
is discretized in 256 steps and there are 16 grid points in the lattice , the calculation of the third order correction involves the summation of about @xmath185 terms for a given value of @xmath152 .
the perturbative results for the @xmath186 pair distribution function @xmath187 are plotted against the monte carlo results as a function of temperature in fig .
[ fig : g2udc_t ] for various orders of the perturbative expansion .
the agreement with the second order expansion is perfect at high temperature , whereas the third order contribution is required to have agreement at lower temperatures .
limit ; this can be checked to be the case for the first order correction to @xmath187 in the grand canonical ensemble , this correction diverging as @xmath188 .
we therefore restrict the perturbative expansion to temperatures larger than @xmath189 in dimensionless units . ] for a given temperature , the @xmath11 dependence of @xmath190 predicted by the perturbative expansion is also in good agreement with the exact monte carlo results . for @xmath191 ,
the agreement is also good , at high temperature in fig .
[ fig : g1pairmc]a , as well as at a temperature @xmath133 in fig .
[ fig : g1pairmc ] .
the fact that the third order prediction is very close to the quantum monte carlo results even when long range order is present may be fortuitous : it significantly differs from the second order prediction so that a calculation of the fourth order correction is required to justify the truncation of the series at this order .
it is actually interesting to perform also the perturbative expansion in the grand canonical ensemble : simpler analytical formulas can be obtained , which can be used to test existing approximate theories applicable to the grand canonical ensemble .
the unnormalized density operator of the gas is now @xmath192 the perturbative expansion has to be performed for a fixed value of the mean total number of particles equal to @xmath35 . as a consequence
the value of the chemical potential @xmath193 is not known in advance and has to be adjusted order by order in the perturbative expansion . to this end
, we write @xmath194 where @xmath153 is the chemical potential of the ideal fermi gas having on the mean a number @xmath35 of particles
. this amounts to performing the following splitting : @xmath195 where the perturbation is now @xmath196 both terms in @xmath197 being of order @xmath17 .
we shall restrict here for simplicity to a second order expansion . from eq.([eq :
expansion ] ) we get @xmath198 where @xmath199/\textrm{tr}[\sigma_{\rm gc}]$ ] stands for the expectation value in the grand canonical density operator of the interacting gas eq .
( [ eq : sigma_gc ] ) and @xmath200 stands for the expectation value in the grand canonical density operator @xmath201 $ ] of the ideal fermi gas . expanding the inverse of the denominator in eq.([eq : exp_frac ] ) and keeping terms up to second order , one obtains @xmath202 where we have introduced the irreducible averages of products of operators @xmath175 , @xmath203 , @xmath204 : @xmath205 and where we used the identity @xmath206 opposite spin density - density correlation function at @xmath186 : @xmath111 as a function of the temperature @xmath28 in the grand - canonical ensemble .
solid line : perturbative result at order 2 . dashed line : density - density rpa .
dotted line : nozires schmitt - rink theory .
dot - dashed line : @xmath207 rpa .
thin solid line : bcs theory .
the vertical line is an approximate lower bound on @xmath28 where the second order perturbative theory is accurate ; its position was determined by a comparison to the quantum monte carlo results of fig .
[ fig : g2udc_t ] . same system parameters as in fig.[fig : g1uumc ] .
the mean number of particles is fixed to @xmath208.,width=377 ] to calculate @xmath209 up to second order , we express the fact that the mean density of spin up particles is fixed in @xmath186 . as the system has translational and spin symmetry ,
this is equivalent to fixing the mean total number of particles .
we therefore specialize eq.([eq : exp_joli ] ) to the case @xmath210 and obtain where @xmath211 is the density in one spin component of the ideal fermi gas with a chemical potential @xmath153 . ]
@xmath212 where @xmath22 is the spatial step of the grid , @xmath213 is the total density and the following notations were introduced : @xmath214 for integers @xmath215 equal to @xmath216 or @xmath217 and with the convention @xmath218 .
note that the term of order @xmath17 in @xmath209 coincides with the hartree - fock mean field prediction . in a second step ,
we calculate @xmath187 by taking @xmath219 .
eliminating @xmath209 from the resulting expression gives : @xmath220 in the case of a negative coupling constant @xmath17 , both the first and second order terms are positive , leading to a spatial bunching of opposite spin particles , as expected for attractive interaction . from the comparison with the quantum monte carlo calculations in the canonical ensemble , we know the temperature range over which the second order perturbative expansion gives accurate predictions for @xmath187 . for the grand canonical ensemble with the same mean number of particles , we expect the same conclusion to apply .
we therefore use a numerical integration of eq.([eq : g2gc ] ) as a test of existing approximate theories that will be reviewed in sec.[sec : exist ] . as is apparent in fig .
[ fig : g2udgc_t ] , the density - density rpa is in very good agreement with the perturbative result , whereas the @xmath221 rpa overestimates @xmath187 and the nozires - schmitt - rink prediction clearly underestimates it . at temperatures above the bcs critical temperature , the bcs theory reduces to the mean - field hartree - fock theory which gives for @xmath187 simply the ideal fermi gas result , @xmath222 .
for the grand canonical ensemble , several approximate many - body theories exist which can be used to obtain predictions for the correlation functions of the interacting fermi gas . in the next subsections ,
some among the most famous ones are discussed , namely the mean - field bcs theory @xcite , two versions of the random - phase approximation ( rpa ) @xcite , and the nozires - schmitt rink ( nsr ) theory @xcite developed to study the bcs - bec crossover in strongly interacting gases .
a quantitative comparison with the prediction of a grand canonical version of the perturbative expansion of section [ sec : pert ] will be performed . in order for the comparison to be meaningful ,
the many - body theories under investigation have been specialized to the specific case of the discrete lattice hamiltonian ( [ eq : hamilt ] ) with exacly the same discretization parameters as used in the previous sections . in the bcs theory ,
the equilibrium density matrix is determined in a self - consistent way from the mean - field quadratic hamiltonian of the grand canonical ensemble at a chemical potential @xmath193 : @xmath223 where the mean density @xmath100 in a given spin component and the gap function @xmath224 are defined as usual as : @xmath225 the quadratic hamiltonian ( [ eq : hamiltq ] ) is easily diagonalized by a bogoliubov transformation in the plane wave basis : @xmath226 where the @xmath227 , @xmath228 operators satisfy fermi anticommutation rules and are related to the fermi field operators by : @xmath229 where the positive coefficients @xmath230 , @xmath231 of the bogoliubov transformation are defined by : @xmath232 the quasi - particle energies @xmath233 are given by : @xmath234 and the chemical potential is shifted as @xmath235 so as to take into account the mean - field energy .
the self - consistency equation for the gap @xmath224 is : @xmath236 where the @xmath237 are the quasi - particle occupation numbers @xmath238 . for high temperature @xmath239 , eq.([eq : gap ] )
has no solution , so that the system is in the _ normal _ phase @xmath240 and the bcs theory reduces to a hartree - fock theory . at low temperature @xmath241 ,
the gap equation is solved for a non - vanishing value of @xmath224 .
this value grows as the temperature decreases .
the expansion of the field operator ( [ eq : bcs_field_exp ] ) in terms of quasi - particle creation and destruction operators can be used to obtain a prediction for the correlation functions .
for instance , the bcs prediction for the one - body correlation function @xmath242 is given by : @xmath243 .
{ \label{eq : g1uubcs}}\ ] ] from wick s theorem , the single - spin density - density correlation function @xmath244 is @xmath245 both quantities are affected in a weak way by the attractive interactions and eventually by the appearance of a non - vanishing gap @xmath224 .
a richer physics can be found in the opposite spin density - density correlation function @xmath110 .
for this quantity , the bcs theory predicts : @xmath246 where the anomalous correlation function @xmath247 is defined as : @xmath248 as @xmath249 , @xmath117 has the simple expression : @xmath250 for @xmath239 , this quantity is identically @xmath251 , which means that the bcs theory does not predict any correlation between the densities in opposite spin states .
these appear only for @xmath252 as a consequence of the non - vanishing bcs gap .
as one can see in fig.[fig : g2udgc_t ] , this result is in qualitative disagreement with the perturbative expansion which gives a significant degree of correlation also for @xmath253 . the bcs prediction for the first - order pair coherence function @xmath127 is : @xmath254 and is characterized by a short - ranged bump of spatial size of the order of @xmath102 , and a non - vanishing long - range limit .
as one can see in fig.[fig : g1pairc_t ] , the long distance behaviour of @xmath123 predicted by the bcs theory is in qualitative agreement with the monte carlo predictions . a simple way of including the fluctuations around the mean - field is to compute a response function within the mean - field theory and then invoke the fluctuation - dissipation theory to obtain the corresponding correlation function . in this subsection
, we shall give a short review of the main results of linear response theory that are required to obtain the correlation functions of our interacting fermi gas .
a complete discussion of linear response theory and fluctuation - dissipation theorem can be found in @xcite .
let @xmath175 and @xmath203 be two operators of a system characterized by a time - independent hamiltonian @xmath255 . for notational simplicity
, we assume that at equilibrium @xmath256 .
a weak perturbation of the form : @xmath257 is applied to the system and its effect on the observable @xmath203 recorded . at linear regime , this is summarized by the linear response functions : @xmath258.\ ] ] the linear susceptibilities @xmath259 have the simple expression in terms of commutators : @xmath260\rho_{\rm eq } \big\}\theta(t),\ ] ] where @xmath261 is the thermal equilibrium density matrix at @xmath30 , @xmath262 $ ] is the partition function and @xmath263 .
as the hamiltonian of the unperturbed system does not depend on time , the fourier transform of @xmath264 is the frequency - dependent response function to a harmonic perturbation of frequency @xmath265 : @xmath266 where @xmath267 has no delta singularity in @xmath268 . ]
the correlation function @xmath269 is defined as : @xmath270 if the condition : @xmath271\in { \mathbb r } , { \label{eq : reality}}\ ] ] holds for all the eigenenergies @xmath272 and @xmath273 of the hamiltonian @xmath274 where the projector @xmath275 projects onto the eigenspace of energy @xmath276 , then the fluctuation - dissipation theorem holds in its most common form ( callen - welton theorem ) relating the imaginary part of the response function @xmath277 $ ] to the correlation function @xmath269 : @xmath278=-\frac{1}{2\hbar}{\tilde s}_{ba}(\omega)\big ( 1-e^{-\beta \hbar \omega } \big ) .
{ \label{eq : fd}}\ ] ] it is easy to verify that the condition ( [ eq : reality ] ) is verified if @xmath279 or , more generally , if @xmath280 , @xmath281 being an arbitrary unitary operator such that @xmath282 . the fluctuation - dissipation theorem ( [ eq : fd ] ) implies that the correlation function @xmath283 is fixed by the knowledge of @xmath277 $ ] modulo a delta distribution in @xmath268 : @xmath284}{1-e^{-\beta \hbar \omega}}+2\pi\,c_{ba}\,\delta(\omega ) .
{ \label{eq : fluctdiss}}\ ] ] the constant @xmath285 can be written as follows : @xmath286\ ] ] in terms of the thermodynamic ( isothermal ) susceptibility @xmath287 .
as usual in thermodynamics , this is defined as the response on @xmath203 when the system is at thermal equilibrium in the presence of a weak and time - independent perturbation @xmath288 : @xmath289 resulting from the expansion to first order in @xmath290 of : @xmath291 } { \textrm{tr}[e^{-\beta ( { \mathcal h}+\epsilon^ * a+\epsilon a^\dagger ) } ] } .\ ] ] notice that while the definition of @xmath292 involves some implicit coupling to a thermal reservoir at temperature @xmath28 , @xmath293 is defined for an isolated system evolving under the hamiltonian @xmath255 .
for this reason , the thermodynamical susceptibility @xmath292 and the static limit @xmath294 in general are not equal @xcite . from the microscopic expression of @xmath285 in terms of the eigenstates of @xmath255 of energy @xmath272 :
@xmath295 , \ ] ] one concludes that @xmath285 is not vanishing in the presence of degeneracies between the eigenenergies of @xmath255 or when the diagonal matrix elements @xmath296 and @xmath297 are not vanishing .
the possibility of having a term in @xmath298 in the correlation function @xmath299 is often neglected in statistical mechanics textbooks , e.g. @xcite . although this is generally correct in the thermodynamical limit , it may lead to incorrect results for the correlation functions of finite systems .
examples of this issue are discussed in the next section and in the appendix [ app : delta(omega ) ] .
a prediction for the density - density correlation function of the fermi gas in the grand canonical ensemble can be obtained by applying the general results of the previous subsection to the operator @xmath300 giving the particle density in the spin state @xmath55 at position @xmath11 .
an approximate prediction for the density - density response functions can be obtained by linearizing the equations of the mean - field theory discussed around the thermal equilibrium state . for historical reasons ,
this approximation scheme is usually called _ random phase approximation _ ( rpa ) @xcite . for the sake of simplicity ,
we shall limit ourselves to the case @xmath239 , regime in which the vanishing of the anomalous averages considerably improves the physical transparency of the formulas .
as the system is spatially homogeneous with the same density @xmath100 in each spin - component , the different fourier components of the spatial density @xmath301 are decoupled .
taking @xmath302 and @xmath303 , the frequency - dependent susceptibility matrix has the form : @xmath304 because of the symmetry in the spin space , @xmath305 and @xmath306 , so that the eigenvectors of the susceptibility matrix are the symmetric and antisymmetric linear combination of the two spin states .
the corresponding eigenvalues are : @xmath307 conversely , the susceptibility matrix @xmath308 in the @xmath309 basis is written as a function of the @xmath310 as : @xmath311 the rpa susceptibility of an interacting gas can be calculated from the hartree - fock equation of motion @xcite , and has the simple expression : @xmath312 in terms of the susceptibility @xmath313 of a non - interacting , one component fermi gas at the same temperature and chemical potential : @xmath314 @xmath315 are the energies of the single - particle states and @xmath316)^{-1}$ ] the corresponding fermi occupation factors .
notice that the quantity inside the sum vanishes for the states @xmath161 such that @xmath317 .
an expression analogous to ( [ eq : rpa_dens - dens ] ) : @xmath318 relates the thermodynamic susceptibilities @xmath319 of the interacting gas to the thermodynamic susceptibility of the non - interacting one - component gas : @xmath320 by applying the fluctuation - dissipation theorem in its form ( [ eq : fluctdiss ] ) to the operators @xmath321 and @xmath322 , one can write the correlation function @xmath323 in terms of the imaginary part of @xmath324 and the thermodynamic susceptibility @xmath325 : @xmath326}{1-e^{-\beta\hbar\omega } } -2\pi\,k_b t\big [ \chi_{\sigma\sigma'}^{\rm therm}(k)- \lim_{\omega\rightarrow 0}{\tilde \chi}_{\sigma\sigma'}(k,\omega)\big]\,\delta(\omega ) . { \label{eq : s(k , omega)}}\ ] ] the condition ( [ eq : reality ] ) is here satisfied since @xmath175 and @xmath203 are connected by @xmath327 with @xmath281 respectively equal to the identity , if @xmath328 , or the spin - inversion operator @xmath281 exchanging the spin components @xmath329 of all the particles , if @xmath330 . finally , the rpa prediction for the desired real - space , one - time density - density correlation function can be found by inverse fourier transform of @xmath331 : @xmath332 corresponding predictions for the opposite spin density - density correlation function at @xmath186 as a function of the temperature are plotted in fig.[fig : g2udgc_t ] .
notice the excellent agreement of the rpa prediction with the one of the perturbative expansion in @xmath17 discussed in sec.[sec : pert ] .
in our finite system , the agreement strongly relies on the correct inclusion of the @xmath298 term in ( [ eq : s(k , omega ) ] ) .
an explicit calculation of this issue for the non - interacting case is presented in the appendix [ app : delta(omega ) ] . in the previous subsection
, we have obtained a prediction for the density - density correlation function @xmath335 of an interacting fermi gas by using the rpa density - density susceptibility and then invoking the fluctuation - dissipation theorem . in the present subsection
, a similar approach is used to obtain the pair coherence function @xmath127 at temperatures higher than the bcs critical temperature @xmath336 in the grand canonical ensemble .
consider the pair of operators : @xmath337 at thermal equilibrium both of them have a vanishing expectation value .
the correlation function : @xmath338 can be evaluated from the susceptibility @xmath339 .
as in the previous subsection , we introduce the spatial fourier components as : @xmath340 note the sign difference in the phase factors of ( [ eq : a_k ] ) and ( [ eq : b_k ] ) .
thanks to the spatial homogeneity of the system , the susceptibility is diagonal in @xmath21-space . from the hartree - fock - bogoliubov equation of motion for the anomalous averages
@xcite , one can obtain the following simple expression for the rpa susceptibility : @xmath341 where @xmath313 is defined as : @xmath342 and describes the ideal gas response . as previously , @xmath343 are the energies of the single - particle states , @xmath344 the fermi occupation factors , and @xmath345 the total particle density summed over both spin states .
notice how @xmath346 diverges when @xmath347 tends to @xmath348 .
this is the signature of the approaching of the bcs transition : the standard equation @xcite for the bcs critical temperature is in fact recovered if one imposes : @xmath349 as the chemical potential is a variable that can be continuously varied , all degeneracies between states with different particle number are accidental and occur only for discrete values of @xmath193 . as @xmath175 and @xmath203 have vanishing diagonal elements , the correction term in @xmath298 vanishes for all other values of @xmath193 .
as @xmath123 has a continuous dependance on @xmath193 , there is no need for calculating @xmath350 .
we therefore have : @xmath351}{1-e^{-\beta\hbar\omega}}\ ] ] and @xmath352 the condition ( [ eq : reality ] ) is here satisfied as @xmath353 . as
@xmath354 coincides with @xmath355 , we have included in fig.[fig : g2udgc_t ] also the prediction of the present @xmath221 rpa approach .
the agreement with the perturbative expansion is less good than in the case of the density - density rpa approach . in @xcite
, a non - perturbative calculation is performed for the grand potential of a two component fermi gas with attractive interactions by a resummation of a certain class of diagrams .
starting from eq .
( 20 ) of @xcite which gives the grand potential @xmath356 in terms of an integral in the complex plane , we can deform the integration contour and apply the residues formula to obtain the following expression in terms of a sum for the case of a contact interaction potential : in terms of a matsubara sum differs from the unnumbered equation between eqs .
( 19 ) and ( 20 ) of @xcite .
this is due to the omission by the authors of @xcite of the contribution to the contour integral of the half - circle of infinite radius in the @xmath357 half plane . ]
@xmath358-\frac{g_0}{2l } \sum_{k_1,k_2 } ( 1-f_{k_1}-f_{k_2 } ) , { \label{eq : nsromega}}\ ] ] where @xmath359 with @xmath360 integer ranging from @xmath361 to @xmath2 .
the function @xmath259 is defined as : @xmath362 @xmath344 being the occupation number @xmath363\}^{-1}$ ] and @xmath364 being the grand potential for the ideal two - component fermi gas .
this theory requires a self - consistent determination of @xmath193 from @xmath365 , which we perform numerically .
it gives access to @xmath117 thanks to the hellmann - feynman theorem @xcite : @xmath366 the results are plotted in fig.[fig : g2udgc_t ] . the poor agreement with the perturbative expansion
can be explained as follows .
let us expand eq.([eq : nsromega ] ) upto second order in @xmath17 at a given @xmath193 .
if one then replaces @xmath193 by its value in the nsr theory for the density under consideration , one gets from eq.([eq : hellfey ] ) a prediction for @xmath117 upto first order that the can be compared to the exact expansion eq.([eq : g2gc ] ) : @xmath367 here @xmath368 is the total density of the ideal two - component fermi gas for the chemical potential @xmath193 .
a first order expansion of @xmath356 is enough to obtain the chemical potential @xmath369 in the nsr theory and to conclude that @xmath370 differs from the exact value by a term of the order of @xmath17 which has the same sign as @xmath17 .
this just because not all the second order diagrams for @xmath356 have been included in the resummation procedure .
tends to zero when the cut - off energy tends to infinity , in which case the hartree - fock mean field term in @xmath371 tends to zero . in this regime ,
a non - perturbative resummation procedure then seems unavoidable . ]
in the present paper , we have presented the result of extensive quantum monte carlo simulations for the static correlation functions of a one - dimensional lattice model of attractively interacting two component fermions .
the numerical results obtained by qmc have been compared to existing approximate theories .
excellent agreement with the predictions of a perturbative expansion in the interaction constant has been found , as well as with the ones of the random phase approximation .
although long - range order is destroyed by phase fluctuations in one - dimensional systems in the thermodynamical limit , the finite size of the system under consideration still allows for the identification of a crossover to a condensed state at the temperature @xmath120 at which the first order coherence length of the pairs becomes larger than the system size .
we have found that a significant degree of opposite spin density - density correlations already exists at temperatures well above @xmath120 and is only slightly enhanced as the temperature goes below @xmath120 .
this means that a measurement of the density - density correlation function @xmath114 can not provide an unambiguous signature of the onset of a condensed state in the gas . on the other hand , this could be provided by a measurement of the second - order momentum space correlation function as suggested in @xcite , or , even more directly , of the long - range behaviour of the first - order pair coherence function @xmath372 . a non - vanishing limit of @xmath127 for large @xmath11 corresponds in fact to the presence of a finite condensate fraction in both weak- ( bcs ) and strong- ( bec ) interaction regimes .
a possible experimental scheme to measure @xmath123 in atomic fermi systems by means of matter - wave interferometric techniques will be the subject of a forthcoming publication .
laboratoire kastler brossel is a unit de recherche de lcole normale suprieure et de luniversit paris 6 , associe au cnrs .
we acknowledge discussions with ph .
chomaz , j. dalibard , b. derrida , t. jolicur , o. juillet , a. montina , c. mora , a. recati .
in the present appendix , we describe the numerical algorithm used for the numerical simulations .
a monte carlo technique has been used to sample the probability distribution of the initial wavefunctions @xmath373 and the elementary noise terms @xmath64 .
as the effective contributions of the different realizations to the observables involve the trace of the hartree - fock ansatz @xmath55 , i.e. the scalar product between the two @xmath35-body nartree - fock states , they can have enormously different values , so a direct draw of the random variables @xmath373 and @xmath64 would be poorly efficient . an _ importance sampling _
scheme @xcite has therefore been implemented , using the value of the modulus of the trace at the end @xmath62 of the imaginary - time evolution as the _ a priori _ probability distribution function @xmath374 : @xmath375=\big|\textrm{tr}[\sigma]\big|= \big| { \langle \phi{^{(2)}}_1(\beta)\ldots\phi{^{(2)}}_n(\beta ) | \phi{^{(1)}}_1(\beta)\ldots\phi{^{(1)}}_n(\beta ) \rangle } \big|.\ ] ] in this way , the contributions of the different realizations to the trace have the same absolute value , although their phases are still random . in order to sample @xmath374 , a _ metropolis _ scheme @xcite
has been implemented : at each step a random move is proposed for both the wavefunctions @xmath373 and the elementary noises @xmath64 . for the first one , a rotation @xmath376 in the one - body hilbert space is chosen with random rotation axis @xmath377 and angle @xmath152 , and then is applied to all the orbitals @xmath373 : @xmath378 this operation rotates the hyperplane spanned by the set of orthonormal orbitals @xmath379 describing the initial hartree - fock state . for what concerns the elementary noises @xmath64 ,
each of then is displaced to a new position @xmath380 as follows : @xmath381 @xmath382 being independent , zero - mean , complex gaussian variables such that @xmath383 , @xmath384 . this kind of random process is such that the resulting distribution of the @xmath385 s is indeed a gaussian with the required width .
the parameter @xmath386 as well as the probability distribution for the random rotation angle @xmath152 are free parameters which can be tuned to optimise the efficiency of the simulation . denoting with @xmath387 and @xmath388 the value of the a priori probability for the configurations respectively before and after the proposed move ,
this is accepted with a probability @xmath389 $ ] . as all configurations can be attained by the random motion and detailed balance
is verified , the stationary probability distribution of the stochastic process is indeed the desired one @xmath374 .
if a large enough number of moves is performed between successive realizations , these can be considered to be statistically independent .
the calculation of @xmath390 involves the following expectation values : @xmath391 where @xmath392 and @xmath393 , @xmath394 . introducing the notations for the particle and hole correlation functions @xmath395 where @xmath215 are integers from @xmath125 to @xmath396 or are equal to @xmath0
, we find @xmath397 the property that each of these expressions is a product of two similar factors , the second one deduced from the first one by the replacement @xmath398 , originates from the fact that the ideal fermi gas in the grand canonical ensemble consists of two equivalent and independent spin components .
similarly the calculation of @xmath399 involves the expectation values @xmath400 using the previous notations we then obtain @xmath401 the fact that these expressions are squares is again due to the existence of two independent and equivalent spin components .
the density - density correlation function for a one - component non - interacting fermi gas in the grand canonical ensemble at inverse temperature @xmath30 and chemical potential @xmath193 can be calculated from wick s theorem : @xmath402 where @xmath403 is the operator giving the density at point @xmath11 . a calculation based on the fluctuation - dissipation theorem neglecting the term @xmath404 term in ( [ eq : fluctdiss ] )
would give : @xmath405 where the sum @xmath406 has to be performed over the pair of states such that @xmath407 .
using the relation : @xmath408 one can see that expression ( [ eq : fdapp ] ) does not coincide with ( [ eq : wickapp ] ) because of the missing contribution of the pairs such that @xmath317 .
inclusion of the term proportional to @xmath298 in ( [ eq : fluctdiss ] ) fixes the problem , since it exactly provides the missing contribution : @xmath409 where the sum @xmath410 has to be performed over the pairs of states such that @xmath317 .
the physical meaning of the @xmath411 term which contains the contribution of the diagonal matrix elements of the perturbation is transparent : it keeps track of the total particle number fluctuations of the grand canonical ensemble . in our spatially homogeneous system , degenerate pairs of states for @xmath412 are also present .
b. de marco and d. s. jin , science * 285 * , 1703 ( 1999 ) ; a. c. truscott , k. e. strecker , w. i. mcalexander , g. b. partridge , r. g. hulet , science * 291 * , 2570 ( 2001 ) ; f. schreck , l. khaykovich , k. l. corwin , g. ferrari , t. bourdel , j. cubizolles , and c. salomon , phys .
lett . * 87 * , 080403 ( 2001 ) ; s. r. granade , m. e. gehm , k. m. ohara , and j. e. thomas , phys . rev
. lett . * 88 * 120405 ( 2002 ) ; z. hadzibabic , c. a. stan , k. dieckmann , s. gupta , m. w. zwierlein , a. grlitz , and w. ketterle , phys .
lett . * 88 * , 160401 ( 2002 ) .
j. cubizolles , t. bourdel , s. j. j. m. f. kokkelmans , g. v. shlyapnikov , and c. salomon , phys .
lett . * 91 * , 240401 ( 2003 ) ; c. a. regal , c. ticknor , j. l. bohn , d. s. jin , nature * 424 * , 47 ( 2003 ) ; s. jochim , m. bartenstein , a. altmeyer , g. hendl , c. chin , j. hecker denschlag , and r. grimm , phys .
lett . * 91 * , 240402 ( 2003 ) .
m. bartenstein , a. altmeyer , s. riedl , s. jochim , c. chin , j. hecker denschlag , r. grimm , preprint cond - mat/0401109 ; t. bourdel , l. khaykovich , j. cubizolles , j. zhang , f. chevy , m. teichmann , l. tarruell , s.j.j.m.f .
kokkelmans , c. salomon , preprint cond - mat/0403091 . c. a. regal , m. greiner , and d. s. jin , phys .
lett . * 92 * , 040403 ( 2004 ) ; m. w. zwierlein , c. a. stan , c. h. schunck , s. m. f. raupach , a. j. kerman , and w. ketterle , phys .
lett . * 92 * , 120403 ( 2004 ) .
m. holland , s. j. j. m. f. kokkelmans , m. l. chiofalo , and r. walser , phys .
lett . * 87 * , 120406 ( 2001 ) ; y. ohashi and a. griffin , phys .
lett . * 89 * , 130402 ( 2002 ) ; j. carlson , s .- y .
chang , v. r. pandharipande , and k. e. schmidt , phys .
* 91 * , 050401 ( 2003 ) . | a recently developed quantum monte carlo algorithm based on the stochastic evolution of hartree - fock states has been applied to compute the static correlation functions of a one - dimensional model of attractively interacting two component fermions .
the numerical results have been extensively compared to existing approximate approaches .
the crossover to a condensate of pairs can be identified as the first - order pair coherence extending throughout the whole size of the system .
the possibility of revealing the onset of the transition with other observables such as the density - density correlations or the second - order momentum space correlations is discussed . |
the existence and importance of bars in spiral galaxies has been recognized for the entire history of extragalactic astronomy ( curtis 1912 ) .
hubble ( 1936 ) included the barred ( sb ) and unbarred ( s ) classes as the tines of the tuning fork .
de vaucouleurs ( 1959 ) first noted that `` barredness '' is not a binary state , and invented the class sab for galaxies that are detectably , but not strongly barred .
the relevance of observed wavelength to the detection of bars was first pointed out by hackwell & schweizer ( 1983 ) who discovered a strong bar in their @xmath0-band image of the optical sab ngc 1566 .
the most extensive demonstration of this issue to date is that of eskridge et al .
one of the shortcomings of the above - cited work is that it uses subjective bar classification .
subjective classification is a useful starting point , but to move toward a physical understanding of morphology requires the general adoption of an appropriate , quantifiable and objective metric .
the problem is that bars do not lend themselves naturally to scalar quantification .
there have been a large number of `` bar - strength quantifiers '' proposed and implemented by different groups , and all of them measure somewhat different things ( e.g. , elmegreen & elmegreen 1985 ; ohta et al . 1990 ; martin 1995 ; wozniak et al .
1995 ; abraham & merrifield 2000 ; buta & block 2001 ) .
this presentation reports on the progress of a study to intercompare a set of different bar - strength quantifiers using a single large , and statistically complete sample of nearby spiral galaxies ( the osu sample see eskridge et al .
the main purposes of the study are two - fold : \1 ) to determine how well different bar - strength classifiers compare to one another and study the sorts of situations in which they fail to agree .
\2 ) to address how reliably we can study the bar fraction of high-@xmath1 galaxies .
this follows from a comparison of the results from techniques that are most demanding of the observational data with those that are the least demanding .
section 2 presents the methods included in this study .
section 3 presents some of the bivariate relations found .
section 4 includes a discussion of the findings to date and a summary of issues yet to be fully investigated .
the most data intensive method included in this study is the @xmath2 `` disk - torque '' measure of buta et al . (
it is a refinement of the @xmath3 `` bar - torque '' measure of by buta & block ( 2001 )
. both of these methods are pure nir methods by definition .
the goal of these methods is to express the strength of the bar in terms of its gravitational effect on the surrounding disk . in principle , this is an excellent physical means of measuring bar strength . in practice ,
the methods requires assumptions about disk structure that are difficult to verify .
they also require high spatial resolution , high s / n images , and are very laborious to implement .
both of these methods are based on fourier - mode decomposition . in its simplest guise
, the fourier - mode method amounts to measuring the strength of the constant - phase @xmath4 mode in an image .
this study uses analysis by d. elmegreen and her students , following elmegreen & elmegreen ( 1985 ) . a method proposed by abraham & merrifield ( 2000 ) measures bar strength by the maximum ellipticity of the inclination - corrected disk .
it is thus most sensitive to long , thin bars .
the method was refined by whyte et al .
( 2002 ) , who also applied it to the osu survey sample .
this study uses the results of whyte et al .
( 2002 ) .
the complete study will consider the neural - net method of odewahn et al .
( 2002 ) , developed for automated classification of high - redshift galaxies .
it is the most forgiving of data quality .
these results are not yet available .
figures 13 show a selection of bivariate comparisons amongst @xmath2 ( buta et al .
2004 ) , the @xmath0-band m@xmath52 fourier mode amplitudes ( from d. elmegreen ) , and the @xmath6- and @xmath0-band abraham bar parameter @xmath7 ( whyte et al .
2002 ) . in all cases ,
statistically significant correlations exists ( at better than the @xmath8 level from a variety of tests ) , but with substantial scatter and large outliers .
the dashed lines are the ordinary least - squares bisectors .
-band @xmath4 fourier mode against @xmath2.,width=384 ] vs a ) @xmath0-band @xmath7 and b ) @xmath6-band @xmath7.,width=384 ] -band fourier @xmath4 mode vs a ) @xmath0-band @xmath7 and b ) @xmath6-band @xmath7.,width=384 ]
it is a great reassurance that all the methods examined here provide statistically self - consistent results , even when comparing different methods in different wave - bands ( nir torque methods compared with @xmath6-band bar ellipticity , as an example ) .
thus any of the methods discussed here can provide a good statistical means of measuring bar strength .
this argues that bar - strength analysis of high - redshift galaxy samples can , at least in principle , be compared with analysis of nearby systems even when the observations are different rest - frame wavelengths , and at very different spatial sampling .
the inclusion of neural - net techniques to the current study promises to be a very interesting advance .
a more careful examination of the bivariate results reveals that even very similar methods tend to have significant scatter between them , and the occasional very substantial outlier .
the existence of these outliers actually gives us the opportunity to study the failure modes of the various techniques under consideration .
as an example , the two systems with the largest @xmath0-band @xmath7 for their @xmath2 values ( the points on the bottom right of fig .
2b ) are ngc 3166 and ngc 4772 .
both these galaxies have @xmath0-band morphologies of sb0 ( eskridge et al .
2002 ) , with relatively large bulges and weak disks .
the @xmath7 statistic is very good at describing such bars , but @xmath2 is not .
this is due to the relatively strong bulges causing the effective torque from the bars to be small . in the same plot , the two galaxies with the largest @xmath2 for their @xmath0-band @xmath7 values are ngc 1042 and ngc 3319 .
both these systems have grand design spiral patterns and short bars .
the shortness of the bars appears to drive the low @xmath7 values .
but the combination of the bars and the strong spiral patterns results in a large disk torque . a true @xmath3 analysis , as discussed in buta et al . ( 2003 ) , will improve the sensitivity of torque - based analysis to pure bar components .
the scatter in the bivariate plots is generally correlated with morphological patterns in the sample .
the use of multiple measures of bar strength thus offers us the opportunity to study full range of the physical properties of bars in an objective , statistical sense .
this study is made possible by the work of the authors of block et al .
( 2002 ) , buta et al .
( 2004 ) and whyte et al .
( 2002 ) , and d. elmegreen and her students .
i thank them for their efforts .
thanks are also due to david block and ken freeman for inviting me , and for organizing the conference . the osu survey was made possible by grants ast-9217716 and ast-9617006 from the us national science foundation .
this work is supported by minnesota state university and by nasa grant hst - go-09892.02 .
eskridge , p.b .
, frogel , j.a . , pogge , r.w . ,
quillen , a.c . ,
berlind , a.a . ,
davies , r.l . ,
depoy , d.l . ,
gilbert , k.m . ,
houdashelt , m.l . ,
kuchinski , l.e . ,
ramirez , s.v . ,
sellgren , k. , stutz , a. , terndrup , d.m . , & tiede , g.p .
( 2002 ) apjs , 143 , 73 | a set of objective bar - classification methods have been applied to the ohio state bright spiral galaxy survey ( eskridge et al .
2002 ) .
bivariate comparisons between methods show that all methods agree in a statistical sense .
thus the distribution of bar strengths in a sample of galaxies can be robustly determined .
there are very substantial outliers in all bivariate comparisons .
examination of the outliers reveals that the scatter in the bivariate comparisons correlates with galaxy morphology .
thus multiple measures of bar strength provide a means of studying the range of physical properties of galaxy bars in an objective statistical sense .
galaxies : fundamental parameters galaxies : spiral galaxies : statistics galaxies : structure infrared : galaxies |
aussel et al . ( 1999 ) demonstrated that the large density of mid - ir(mir ) sources detected in deep isocam ( cesarsky et al .
1996 ) data are dominated by galaxies at intermediate redshifts .
luminous infrared galaxies ( lirgs ) , though rare locally , are common in the past and suggest a previously overlooked site of significant star formation , dust enshrouded starbursts .
zheng et al .
( 2004 ) , analyzed hst / wfpc2 images of 36 distant ( @xmath11 ) isocam detected lirgs .
they classified 36% of their objects as disk galaxies , 22% as irregulars , 17% as obvious mergers and 25% compact sources .
they found a size - color relationship for the cores of their galaxies .
compact sources tend towards blue cores while the larger disk galaxies tend to have redder cores .
this suggested to the authors an evolutionary sequence of disk galaxy formation , with compact blue bulges forming first , followed by inside - out disk growth while the core reddens .
this theme was elaborated by hammer et al .
( 2005 ; hereafter h05 ) , who suggested that disk mergers are driving the evolution of both lirgs and star formation over the last 8 gyr . in their picture ,
disk merger and rebuilding account for the various morphological stages seen in lirgs at intermediate redshifts . for instance , a brief peculiar phase of heightened star formation is followed by a compact phase with dust and gas funneled into the center , growing the bulge .
subsequent settling of gas into the potential well re - grow the disk from the inside out .
the authors suggest that this scenario may have occurred in 75% of intermediate mass spirals since @xmath12 .
isocam data are only sensitive to the brightest mir sources at intermediate redshifts . with the launch of the _ spitzer _ space telescope ( werner et al .
2004 ) , much deeper ir data have become available . using mips ( rieke et al .
2004 ) 24@xmath13 m observations from _ spitzer _
, bell et al .
( 2005 ; hereafter bell05 ) study 1500 @xmath14 galaxies with photometric redshifts in the chandra deep field south . they found a somewhat different morphological make - up of lirgs than the zheng et al .
and h05 studies and drew different conclusions .
bell05 found more than half of lirgs are undisturbed , massive ( @xmath15 ) spirals and less than a third strongly interacting .
they conclude that major mergers are not the primary factor in lirg production or of the rapid decline of star formation since @xmath14 .
they suggest instead a combination of gas depletion and minor mergers . in this letter , we focus on one of the deepest mir survey fields yet taken , the mips 24@xmath13 m image ( dickinson in preparation ) of the great observatories origins deep survey ( goods , giavalisco et al .
2004 ) north field .
very deep optical _ hubble space telescope _
( _ hst _ ) imaging ( @xmath16 mag deeper than bell05 ) are used as well as spectroscopic redshifts from the team keck redshift survey ( tkrs , wirth et al .
we compare optical morphologies and luminosities with ir luminosities of @xmath0 optically selected galaxies to trace the morphological evolution of star forming galaxies since @xmath1 .
the optically selected sample is less biased against low ir luminosity objects than the previous studies .
in addition , the deeper _ hst _ data allow for more accurate morphological typing , especially of fainter tidal features from mergers .
our sample contains 3 times the number of lirgs as the zheng et al ( 2004 ) and h05 sample , and spans a wide range in redshift encompassing the bell05 narrow redshift slice .
section 3 explores the relationship between optical and ir luminosity , and its dependence on optical color , morphology and redshift .
specifically we 1 ) look for evolution in the number density and morphological make - up of lirgs since @xmath1 ; and 2 ) investigate the utility of optical color and luminosity as a predictor of ir luminosity .
section 4 synthesizes the results into a picture of lirg evolution over the last 8 gyr , similar to that of bell05 .
we adopt vega magnitudes and a flat , @xmath17 , @xmath18 cosmology throughout .
galaxies are identified in the goods - n field ( giavalisco et al .
2004 ) using the @xmath19 , and @xmath7 _ hst _ images .
spectroscopic redshifts are from tkrs ( wirth et al 2004 ) , a magnitude limited ( @xmath20 ) survey of 2018 objects in goods - n .
tkrs is 53% complete at its limiting magnitude ( wirth et al . 2004 ) .
elliptical aperture photometry from the program was generated from the 4 _ hst _ bands for all objects in tkrs ( j. melbourne , in preparation ) .
total magnitudes were calculated from the measurements using a curve of growth technique .
following the prescription in wilmer et al .
( 2005 ) , k - corrections convert apparent magnitudes to absolute @xmath6 and rest - frame color @xmath21 .
our sample contains the 789 tkrs galaxies brighter than @xmath22 grouped into three equal volume bins between @xmath23 .
this introduces a slight bias for redshifts @xmath24 , where the tkrs limiting rest - frame magnitude is greater than @xmath25 .
this should not have a major impact on the lirg detection as the majority of lirgs are brighter than this limit ( see for instance figure 10 in le floch et al . 2005 ) .
galaxies are typed into one of 4 morphological categories , spiral , peculiar / irregular , compact and early - types ( e / s0 ) .
spirals , peculiars and early - types are classified by eye .
spirals exhibit symmetry , signs of bulge and disk components and spiral arms .
peculiars are characterized by asymmetry , tidal features or obvious mergers .
ellipticals are picked out as smooth , symmetric and red and lacking an obvious disk component .
compacts are selected by having half - light radii smaller than 3 kpc . at low and intermediate redshifts ( @xmath26 )
the typing is done with the very deep @xmath27- and @xmath28-band images . at higher redshift
, the @xmath7-band image ( rest - frame @xmath6 ) is used to account for the morphological k - correction .
the typing was done by three independent researchers with 70% full agreement , and 100% agreement between at least two of three classifications .
a final classification was assigned by majority . as a check of our typing ,
we classify the 27 non - compact objects in zeng et al .
( 2004 ) and agree with all but one of the classifications .
furthermore we clone our low redshift lirgs to the pixel scale and signal to noise of the z=0.95 lirgs and find that the classifications are not affected .
[ h ] 24@xmath13 m fluxes are measured from the publicly released mips mosaic image of the goods - n field ( goods dr1 + data release , m. dickinson , in preparation ) .
the image is flux calibrated and aligned to the astrometry of the irac images of the same region ( chary et al .
2004 ) . the mips
24@xmath13 m point - spread - function ( psf ) has a full - width - half - max ( fwhm ) of @xmath29 , meaning virtually every source on the image is point - like . as a result , our photometry techniques are borrowed from those used in crowded star fields .
first we generate a psf for the image by median combining 10 flux scaled isolated sources . then for each tkrs galaxy position , we 1 ) subtract all neighbors with centroids separated by more than @xmath30 and less than @xmath31 from the location of interest .
2 ) measure the flux in a @xmath32 aperture at the location of interest .
3 ) apply an aperture correction of 1.34 to the flux measurement ( taken from the mips data handbook v2.1 ) , to account for the large amount of flux in the wings of mips psf .
4 ) apply an additional correction to the flux such that @xmath33 is within 0.5 % of the sky value . for bright objects ( @xmath34 @xmath13jy )
, this technique reproduces the results in the dr1 + bright source catalogue ( r. chary , in preparation ) to within 5% . for fainter sources ( @xmath35 @xmath13jy ) ,
the results vary with an rms of 15% . while the dr1 + catalogue supplies measurements of objects brighter than 100 @xmath13jy , the measurement strategy outlined above provides flux estimates of faint sources to 25 @xmath13jy .
in addition , because we select galaxies in the optical , we measure 24@xmath13 m fluxes for objects that may not have been identified by the standard daophot source finding routines .
elbaz et al .
( 2002 ) demonstrate that the mir is well correlated with total ir luminosity , even at cosmological distances .
we estimate the total ir luminosity , @xmath36 from the mips 24@xmath13 m flux following the prescription in le floch et al .
( 2005 ) . for each redshift and 24@xmath13 m flux ,
this method derives the monochromatic luminosity at the corresponding rest - frame wavelength , which is then translated into a total ir luminosity ( @xmath37 m ) according to libraries of ir spectral energy distributions ( seds ) .
the seds considered in this work were taken from chary & elbaz ( 2001 ) .
other templates give consistent results within 0.3 dex , which is also considered to be the accuracy of our estimates .
following the convention established in sanders & mirabel ( 1996 ) , lirgs are defined as log(@xmath38 .
using the conversion to star formation rates ( sfr ) given in kennicutt ( 1998 ) , lirgs have sfr @xmath39 .
starbursts are defined as @xmath40 and have sfr of 2 - 17 @xmath41 , while normal blue galaxies , log(@xmath42 , have sfr less than 2 solar masses per year .
figure [ fig : images ] shows acs 3-color images of sample galaxies from the intermediate redshift bin ( @xmath14 ) .
galaxies are placed into 4 morphological categories , spiral , peculiar / irregular , compact and early - types ( e / s0 not shown on image ) . for each morphological category
, the figure shows a lirg , a starburst galaxy and a normal galaxy . aside from the dearth of red early - type lirgs ( only 5 of the 119 lirgs
are classified as ellipticals ) , morphology is not a good predictor of ir luminosity .
table [ table : lirgmorph ] reports the morphological typing of the 119 lirgs binned into 3 equal volume bins ( low , intermediate and high redshift ) and 1 additional low redshift bin to highlight morphology evolution of the sample .
two major conclusions can be drawn from these numbers .
first , the number density of lirgs has dropped by 50% from the high redshift bin to the intermediate redshift bin , and continues to drop in the low redshift bins .
second the morphological fraction of the lirgs has changed . above @xmath3
the peculiar to spiral ratio for lirgs is @xmath43 , whereas below @xmath3 the peculiar to spiral ratio is @xmath44 .
a quick monte carlo simulation indicates that drawing such a skewed sample from a distribution with @xmath45 is unlikely at the 84% level , for this small sample size .
we caution that the reclassification of 4 low z objects would be enough to change these fractions to match the higher @xmath7 results
. therefore , larger samples of low-@xmath7 lirgs will be helpful in verifying if this trend is real .
while the result is marginal , figure [ fig : mbir ] , a plot of ir luminosity vs. @xmath6 for the entire sample of galaxies , provides additional evidence for a change in lirg population with redshift .
rcc|cc|cc|cc & # & % & # & % & # & % & # & % + total & 63 & 100 & 30 & 100 & 25 & 100 & 18 & 100 + spiral & 29 & 46 & 14 & 47 & 12 & 48 & 7 & 38 + peculiar & 20 & 32 & 10 & 33 & 10 & 40 & 9 & 50 + compact & 10 & 16 & 6 & 20 & 2 & 8 & 2 & 11 + elliptical & 4 & 6 & 0 & 0 & 1 & 4 & 0 & 0 + figure [ fig : mbir ] plots ir luminosity vs. @xmath6 , in low , intermediate and high redshift equal volume bins .
the red horizontal line divides starburst galaxies ( below the line ) and lirg luminosity ( above the line ) . objects with mips 24@xmath13 m measurements below @xmath46j are shown as black circles and are designated normal galaxies the morphologies of the starburst and lirg samples are indicated : spiral galaxies are blue circles , peculiar / irregular are green squares , compact galaxies are yellow stars and ellipticals are red triangles .
filled symbols are blue galaxies @xmath47 and open symbols are galaxies redder than this limit .
this plot illustrates the morphological evolution hinted at in the previous analysis . in the high redshift bin , spirals and peculiars
are well mixed , spanning similar @xmath6 and @xmath5 ranges . in the low redshift bin ,
the two populations segregate with the peculiars systematically at higher ir luminosity than the spirals .
this results in a change of the lirg peculiar to spiral ratio .
the morphological segregation of low-@xmath7 ir luminous galaxies has been observed by ishida ( 2004 ) in a statistically complete sample of 56 local lirgs drawn from iras bright galaxy sample ( soifer et al .
the ishida study finds that 100% of galaxies with log(@xmath48 show at least some evidence for tidal features .
the majority of galaxies with @xmath49 log(@xmath50 , also tend to be disturbed but @xmath51 are bright isolated spirals .
the ishida study indicates that at lower ir luminosities , spirals comprise a larger fraction of the total .
the implications of these morphological changes with redshift will be explored in the discussion section . while morphology does not appear to be a good indicator of ir luminosity for blue galaxies , @xmath47 , absolute @xmath6 magnitude is .
blue galaxies ( filled symbols in figure [ fig : mbir ] ) show a clear trend of increasing ir luminosity with @xmath6 , regardless of morphology .
this trend , shown previously without a color - cut ( chary & elbaz 2001 ; le floch et al .
2005 ) , was found to have a large scatter . with a color - cut , this scatter is significantly reduced . for our sample , the rms scatter in ir luminosity about the best - fit line ( black line , a bivariate least - squares fit to the full sample of blue galaxies , used for illustrative purposes ) drops from @xmath52 dex without the color - cut to @xmath9 dex with the cut .
this scatter is actually somewhat smaller than our roughly estimated ir luminosity error ( @xmath53 dex ) .
figure [ fig : mbir ] shows that both the optical and ir luminosities of the blue galaxies are increasing with redshift .
roughly 75% of blue galaxies brighter than the characteristic luminosity of @xmath54 ( red vertical line ) are lirgs , regardless of redshift .
only 3 of the 93 blue galaxies brighter than this limit are undetected in the mips image .
studies of galaxies in the uv and optical have shown that the average comoving star formation rate ( sfr ) density of the universe has declined by a factor of @xmath55 since @xmath1 ( lilly et al .
1996 ; madau et al . 1998 ; hogg et al . 1998
; among others ) , and observations in the sub mm wavelengths indicate that it may have peaked around @xmath56 ( chapman et al .
because ir luminosity strongly correlates with sfr ( kennicutt 1998 ) , we see a decline , by @xmath57 in the numbers of lirgs from the high redshift bin ( @xmath58 ) to the intermediate ( @xmath14 ) and low ( @xmath59 ) redshift bins .
this drop in lirg numbers matches well the results from le floch et al .
( 2005 ) , who found that the comoving ir energy density evolves as @xmath60 . assuming the drop in ir energy density were dominated by a reduction in the number of lirgs
, we expect a 50% drop in the number of lirgs of from @xmath1 to @xmath61 , which is what we find . despite this decline , @xmath62 of the brightest ( @xmath54 ) ,
blue ( @xmath63 ) galaxies are lirgs , regardless of redshift .
therefore it appears that the @xmath64-band optical and ir luminosities of galaxies are linked , with sfr as an obvious candidate for the connection .
a large gas supply is required to sustain elevated sfr at the lirg level .
a lirg event in the milky way lasting only 100 myr would use up the entire galactic hi - disk gas supply ( assuming @xmath65 of hi gas , nakanishi & sofue 2003 ) . however , roughly half of the intermediate and high redshift lirgs in our sample are comprised of large normal looking spirals like the milky way , the difference being that high-@xmath7 lirg spirals can be as luminous in the ir as peculiar galaxies .
bell05 finds that lirg spirals tend to be massive with stellar masses @xmath66 .
figure 4 of bell05 shows that the specific sfr of a handful of the most massive lirgs are at or below their past average , indicating that either these galaxies experienced multiple lirg events , or have had enough gas to continuously form stars at a lirg level .
eventually this gas , if not resupplied , will be used up . at low redshift , the make - up of the lirg population changes .
lirg spirals exhibit lower ir luminosities than their peculiar counterparts , and also become more rare .
possible explanations for this include 1 ) the gas reservoirs in spirals are depleted with time , or 2 ) gas accretion is reduced with time . with a depleted gas supply , and/or reduced gas accretion ,
large local spirals can still form stars but generally at rates and below the lirg level .
one way to build up a large , high - density gas supply at low-@xmath7 is to merge spirals together , an event revealed by peculiar morphology .
this picture of gas depletion differs from that of h05 , who suggest that major mergers of spirals are driving the star formation and creation of lirgs since @xmath2 . in the h05 picture ,
most intermediate - mass spirals undergo at least one major merger since @xmath2 , during which the morphology changes from spiral to peculiar ( merger ) to compact ( bulge formation ) to spiral again ( gas infall reforms the disk ) .
if this scenario is responsible for most lirgs , the process should maintain a roughly constant peculiar to spiral ratio , as each morphological phase leads , proportionally in time , into the next .
such a prediction is hard to reconcile with both the large fraction of undisturbed spirals at high redshift and the evolutionary changes seen in the morphological fractions of lirgs . while the changing peculiar to spiral ratio may appear at face value to be incompatible with the h05 picture
, we would like to caution the reader on several points .
first the evidence against the h05 picture is based on visual morphological classifications and thus somewhat subjective .
while identification of major mergers is relatively straight forward at low redshift , tidal features and other indications of peculiar morphology become harder to identify at higher redshift , as signal - to - noise drops . by using very deep optical imaging and cloning the low-@xmath7 objects to higher redshift , we have tried to minimize these concerns . in the future ,
adoption of quantitative morphological indicators such as concentration , asymmetry , clumpiness ( conselice 2003 ) and gini / m@xmath67 ( lotz , primack & madau 2004 ) may help alleviate these concerns .
secondly , larger samples preferably in other parts of the sky will help to account for cosmic variance and increase our relatively modest sample of low redshift lirgs .
thirdly , gas depletion at low-@xmath7 may itself change the disk rebuilding efficiency in the h05 scenario and result in altered lirg morphological fractions .
we would like to thank leonidas moustakas , mark dickinson , and ranga - ram chary , catherine ishida and sandra faber for helpful comments on the project .
this work has been supported in part by the nsf science and technology center for adaptive optics managed by uc santa cruz under the cooperative agreement no . | we combine optical morphologies and photometry from _ hst _ , redshifts from keck , and mid - infrared luminosities from _ spitzer _ for an optically selected sample of @xmath0 galaxies in goods -
n to track morphology evolution of infrared luminous galaxies ( lirgs ) since redshift @xmath1 .
we find a 50% decline in the number of lirgs from @xmath2 to lower redshift , in agreement with previous studies .
in addition , there is evidence for a morphological evolution of the populations of lirgs . above @xmath3 , roughly half of all lirgs are spiral , the peculiar / irregular to spiral ratio is @xmath4 , and both classes span a similar range of @xmath5 and @xmath6 . at low-@xmath7 , spirals account for one - third of lirgs , the peculiar to spiral fraction rises to 1.3 , and for a given @xmath6 spirals tend to have lower ir luminosity than peculiars .
only a few percent of lirgs at any redshift are red early - type galaxies . for blue galaxies ( @xmath8 ) ,
@xmath6 is well correlated with log(@xmath5 ) with an rms scatter ( about a bivariate linear fit ) of @xmath9 dex in ir luminosity . among blue galaxies that are brighter than @xmath10 , 75% are lirgs , regardless of redshift .
these results can be explained by a scenario in which at high-@xmath7 , most large spirals experience an elevated star formation rate as lirgs .
gas consumption results in a decline of lirgs , especially in spirals , to lower redshifts . |
euler s equations for irrotational incompressible ideal fluid motion under gravity with a free surface have an asymptotic expansion for shallow water waves that contains two small parameters , @xmath3 and @xmath4 , with ordering @xmath5 .
these small parameters are @xmath6 ( the ratio of wave amplitude to mean depth ) and @xmath7 ( the squared ratio of mean depth to horizontal length , or wavelength ) .
euler s equations are made non - dimensional by introducing @xmath8 for horizontal position , @xmath9 for vertical position , @xmath10 for time , @xmath11 for surface elevation and @xmath12 for velocity potential , where @xmath13 is the mean wave speed and @xmath14 is the constant gravity .
the quantity @xmath15 is the dimensionless bond number , in which @xmath16 is the mass density of the fluid and @xmath17 is its surface tension , both of which are taken to be constants . after dropping primes ,
this asymptotic expansion yields the nondimensional korteweg - de vries ( kdv ) equation for the horizontal velocity variable @xmath18 at _ linear _ order in the small dimensionless ratios @xmath3 and @xmath4 , as the left hand side of @xmath19 here , partial derivatives are denoted using subscripts , and boundary conditions are @xmath20 and @xmath21 at spatial infinity on the real line .
the famous @xmath22 traveling wave solutions ( the solitons ) for kdv ( [ kdv - eqn ] ) arise in a balance between its ( weakly ) nonlinear steepening and its third - order linear dispersion , when the quadratic terms in @xmath3 and @xmath4 on its right hand side are neglected . on the right hand side of equation ( [ kdv - eqn ] ) ,
a normal form transformation due to kodama @xcite has been used to remove the other possible quadratic terms of order @xmath23 and @xmath24 .
the remaining quadratic correction terms in the kdv equation ( [ kdv - eqn ] ) may be collected at order @xmath25 .
these terms may be expressed , after introducing a `` momentum variable , '' @xmath26 and neglecting terms of _ cubic _ order in @xmath3 and @xmath4 , as @xmath27 in the momentum variable @xmath28 , the parameter @xmath29 is given by @xcite @xmath30 thus , the effects of @xmath31dispersion also enter the nonlinear terms . after restoring dimensions in equation ( [ b - eqn ] ) and rescaling velocity @xmath32 by @xmath33 , the following @xmath34-equation " emerges , @xmath35 where @xmath36 is the dimensional momentum variable , and the constants @xmath37 and @xmath38 are squares of length scales .
when @xmath39 , one recovers kdv from the @xmath34-equation ( [ dim - b - eqn ] ) , up to a rescaling of velocity .
any value of the parameter @xmath40 may be achieved in equation ( [ dim - b - eqn ] ) by an appropriate kodama transformation @xcite .
as we have emphasized , the values of the coefficients in the asymptotic analysis of shallow water waves at quadratic order in their two small parameters only hold , modulo the kodama normal - form transformations .
hence , these transformations may be used to advance the analysis and thereby gain insight , by optimizing the choices of these coefficients .
the freedom introduced by the kodama transformations among asymptotically equivalent equations at quadratic order in @xmath3 and @xmath4 also helps to answer the perennial question , `` why are integrable equations so ubiquitous when one uses asymptotics in modeling ? ''
namely , there may be special values of the free parameters in the normal - form transformations of the asymptotics that allow integrability .
[ [ integrable - cases - of - the - b - equation - dim - b - eqn . ] ] integrable cases of the @xmath34-equation ( [ dim - b - eqn ] ) . + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + the cases @xmath41 and @xmath42 are special values . for these values ,
the @xmath34-equation becomes completely integrable as a hamiltonian system . for @xmath41 , equation ( [ dim - b - eqn ] )
specializes to the integrable ch equation of camassa and holm @xcite .
the case @xmath42 in ( [ dim - b - eqn ] ) recovers the integrable dp equation of degasperis and procesi @xcite .
these two cases exhaust the integrable candidates for ( [ dim - b - eqn ] ) , as was shown using painlev analysis .
the @xmath43family of equations ( [ dim - b - eqn ] ) was also shown in @xcite to admit the symmetry conditions necessary for integrability , only in the cases @xmath44 for ch and @xmath45 for dp .
the @xmath34-equation ( [ dim - b - eqn ] ) with @xmath41 was first derived in camassa and holm @xcite by using asymptotic expansions directly in the hamiltonian for euler s equations governing inviscid incompressible flow in the shallow water regime . in this analysis ,
the ch equation was shown to be bi - hamiltonian and thereby was found to be completely integrable by the inverse scattering transform ( ist ) on the real line .
this development of ist for ch equation ( [ dim - b - eqn ] ) with @xmath41 is discussed in section [ ist - sec ] .
camassa and holm @xcite also discovered the remarkable peaked soliton ( peakon ) solutions of ( [ singlepeakon - soln],[peakontrain - soln ] ) for the ch equation on the real line , given by ( [ dim - b - eqn ] ) in the case @xmath41 .
the peakons arise as solutions of ( [ dim - b - eqn ] ) , when @xmath46 and @xmath47 in the absence of linear dispersion .
peakons move at a speed equal to their maximum height , at which they have a sharp peak ( jump in derivative ) .
the single peakon solution is @xmath48 unlike the kdv soliton , the peakon speed is independent of its width ( @xmath49 ) .
periodic peakon solutions of ch were treated in alber _
there , the sharp peaks of periodic peakons were associated with billiards reflecting at the boundary of an elliptical domain .
these billiard solutions for the periodic peakons arise from geodesic motion on a tri - axial ellipsoid , in the limit that one of its axes shrinks to zero length .
the ch equation was found after its derivation as a shallow water equation in @xcite to fit into a class of integrable equations derived previously by using hereditary symmetries in fokas and fuchssteiner @xcite .
see fuchssteiner @xcite for an insightful history of how the shallow water equation ( [ dim - b - eqn ] ) in the integrable case with @xmath41 relates to the mathematical theory of hereditary symmetries .
equation ( [ dim - b - eqn ] ) with @xmath41 was recently re - derived as a shallow water equation by using asymptotic methods in three different approaches in dullin _
@xcite , in fokas and liu @xcite and also in johnson @xcite .
these three derivations all used different variants of the method of asymptotic expansions for shallow water waves in the absence of surface tension . only the derivation in dullin _
@xcite took advantage of the parametric freedom in the kodama normal - form transformations of the asymptotic expansion results at quadratic order .
the effects of the parameter @xmath34 on the solutions of equation ( [ dim - b - eqn ] ) were investigated in holm and staley @xcite , where @xmath34 was treated as a bifurcation parameter , in the limiting case when the linear dispersion coefficients are set to @xmath46 and @xmath47 .
this limiting case allows a variety of special solutions for different ranges of the values of @xmath34 , in which the two nonlinear terms in equation ( [ dim - b - eqn ] ) balance each other in the _ absence _ of linear dispersion .
since its first appearance in @xcite , the ch equation has been the centre of a confluence of scientific endeavors that includes water waves , integrable systems , pde analysis , asymptotics , geometry and lie groups . this confluence has led to a continued interest and opportunities for contributions from many different fields in mathematics .
the interest in ch solutions may be measured by noticing that the entry ` peakon ' has acquired many thousands of _ google hits_. this paper aims to review some of the geometric highlights of recent work on the ch equation .
it is certainly not exhaustive .
it mainly focuses on comparing the soliton theory for smooth ch solutions with the peakon theory for its singular solutions in the dispersionless limit .
section [ ist - sec ] briefly explains the application of the method of the inverse scattering transform ( ist ) for obtaining the soliton solutions of ch equation .
the set of scattering data is introduced and the formulation of the inverse scattering as a riemann - hilbert problem is outlined . the solution is expressed via the scattering data in a form that admits the peakon limit in the sections that follow . in section [ mommap - sec ]
the map between the action - angle variables ( expressed via the scattering data ) and the momentum of the ch solution is formulated as a momentum map from the symplectic action - angle variables to the dual of the lie algebra of smooth vector fields on the real line .
this is a poisson map , but the noncanonical bi - hamiltonian structure of the ch equation which led to the discovery of its isospectral problem in @xcite and its geometrical significance as geodesic flow on the diffeomorphism group are not discussed here . in section [ peakon - sec ]
we introduce the peakons as singular solutions that appear in dispersionless limit .
the @xmath2-peakon solution is governed by a finite - dimensional integrable dynamical system .
section [ peakon - limit - sec ] presents the multi - peakon solution as a limiting case of the ch multi - soliton solution and points out the similarity between the dynamics of the peakon system and the well known toda lattice . in section [ superint - sec ]
we comment briefly on the existence of additional integrals of motion of the peakon system , a property known as _ superintegrability_.
the two - peakon system is analysed explicitly .
section [ other - sing - sol - sec ] presents the compacton and pulson solutions , which are singular solutions , similar to peakons that can also be represented by the singular momentum map for peakons discussed in @xcite . in section [ ep - higher - sec ] we deal with generalizations of the ch equation in higher dimensions .
these also allow for singular solutions .
although such multidimensional generalizations are unlikely to be integrable , numerical studies show that their solutions are stable and interact elastically in collisions . in section [ conclus - sec ]
we comment briefly on some limitations of our present discussion and point out three open problems for further research .
one of the most significant results in the theory of nonlinear partial differential equations was the development of gardner , greene , kruskal and miura @xcite of a method for the exact solution of the initial - value problem for the kdv equation .
prior to their work the only known exact solutions of kdv were the travelling wave solutions .
the idea is based on a representation of the equation ( @xmath50 ) as a compatibility condition of two eigenvalue problems with a time - independent spectral parameter @xmath51 : [ kdv - isospec ] _
xx&+ & u= + _ t&=&(u_x+)-(4 + 2u)_x where @xmath52 is an arbitrary constant .
the method is conceptually analogous in many ways to the fourier transform method for solving linear equations .
this method is now known as inverse scattering transform ( ist ) .
it was recast by lax @xcite in a general framework that allows the ist to be used for solving other nonlinear pdes .
an important consequence is the relationship between the discrete eigenvalues and the characteristics of the solitons that emerge asymptotically .
another feature of soliton collisions is the preservation of soliton identities after the interaction ( asymptotically in time ) .
for example , collisions of kdv solitons only result in a phase shift from the positions where they would have been without the interaction .
the kdv was formulated as a completely integrable hamiltonian system in a work by faddeev and zakharov @xcite , the hamiltonian form was also noted by gardner @xcite . reviews of ist may be found , for example , in ablowitz et al .
@xcite , dubrovin _ et al . _
@xcite , novikov _ et al .
_ @xcite . for discussions of other related bi - hamiltonian equations ,
see @xcite . in this section
we outline the application of the ist for the ch equation .
we use the form with a linear dispersion term @xmath53 where @xmath54 is a real constant .
the equation in this form also appears as a model of unidirectional propagation of shallow water waves over a flat bottom @xcite as well as that of axially symmetric waves in a hyperelastic rod @xcite .
it can be obtained from the @xmath34-equation ( [ dim - b - eqn ] ) with @xmath41 via galilean transformation that removes the @xmath55 term .
equation ( [ eq1 ] ) admits a lax pair formulation @xcite [ eq3 ] _
xx&=&(1-(m+ ) ) + [ eq4 ] _ t&=&-(+u)_x++ where @xmath52 is an arbitrary constant .
we will use this freedom for a proper normalization of the eigenfunctions . in our further considerations
@xmath56 will be a schwartz class function , @xmath57 and @xmath58 .
then @xmath59 for all @xmath60 @xcite .
let us introduce a new spectral parameter @xmath61 , such that [ lambda ] ( k)= ( 1 + 4k^2 ) .
the spectrum of the problem ( [ eq3 ] ) under these conditions is described in @xcite .
the continuous spectrum in terms of @xmath61 corresponds to @xmath61 real . the discrete spectrum ( in the upper half plane )
consists of finitely many points @xmath62 , @xmath63 where @xmath64 is real and @xmath65 . for
all real @xmath66 a basis in the space of solutions of ( [ eq3 ] ) can be introduced , @xmath67 and @xmath68 , fixed by its asymptotic when @xmath69 @xcite : [ eq5 ] ( x , k)&=&e^-ikx+o(1 ) , x another basis can be introduced , @xmath70 and @xmath71 fixed by its asymptotic when @xmath72 : [ eq5a ] ( x , k)&=&e^-ikx+o(1 ) , x - and the relation between the two bases is @xcite [ eq8 ] ( x , k)=a(k)(x , k)+b(k)|(x , k ) .
where [ eq10 ] |a(k)|^2-|b(k)|^2=1 .
the quantity @xmath73 is called the reflection coefficient , without any danger of confusing @xmath74 with the bifurcation parameter @xmath34 introduced in ( [ dim - b - eqn ] ) .
all of the required information about the scattering , i.e. @xmath75 and @xmath74 , is provided by @xmath76 for @xmath77 only @xcite .
it is sufficient to know @xmath76 only on the half line @xmath77 , since @xmath78 , @xmath79 and thus @xmath80 . the constant @xmath52 in ( [ eq4 ] )
can be chosen for each eigenfunction in such a way that @xmath75 does not depend on @xmath60 and is a generating function of the integrals of motion @xcite . at the points of the discrete spectrum
, @xmath75 has simple zeroes @xcite , therefore @xmath81 and @xmath82 are linearly dependent ( [ eq8 ] ) : @xmath83 in other words , the discrete spectrum is simple , there is only one ( real ) eigenfunction @xmath84 , corresponding to each eigenvalue @xmath85 , and we can take this eigenfunction to be [ eq201]^(n)(x)(x , i_n ) the asymptotic behaviour of @xmath86 , according to ( [ eq5a ] ) , ( [ eq200 ] ) is [ eq203 ] ^(n)(x)&=&e^_n x+o(e^_n x ) , x- ; + [ eq204 ] ^(n)(x)&= & b_n e^-_n x+o(e^-_n x ) , x. the sign of @xmath87 obviously depends on the number of the zeroes of @xmath86 .
suppose that [ eqn ] 0<_1<_2< <_n<1/2.then from the oscillation theorem for the sturm - liouville problem @xcite , @xmath86 has exactly @xmath88 zeroes .
therefore [ eq205 ] b_n= ( -1)^n-1|b_n| .
the set [ eq206 ] \ { ( k)(k>0),_n , r_n = b_n / ia(i_n),n=1, n } is called the _
scattering data_. the hamiltonians for the ch equation in terms of the scattering data are presented in @xcite .
the time evolution of the scattering data is @xcite : [ eq19 ] ( k , t)&=&(k,0)e^-t . + [ eq207 ]
r_n(t)&=&r_n(0)(t ) , where @xmath89 is always a positive quantity @xcite and @xmath90 .
the scattering coefficient @xmath75 is analytic for @xmath91 with asymptotic @xcite @xmath92 where @xmath93 is a constant ( integral of motion ) : @xmath94 the asymptotics of the jost solutions ( or , rather the following quantities , depending on the jost solutions ) for @xmath95 have the form @xcite [ eq20 ] ( x , k ) & & ( x , k)e^=x_0(x)+++ , + ( x , k ) & & ( x , k)e^ik ( + ) = x_0(x)+++ [eq21 ] where @xmath96^{1/4}>0 $ ] , [ eq22 ] y(x)=. moreover , ( [ eq20 ] ) is analytic for @xmath97 , ( [ eq21 ] ) is analytic for @xmath91 @xcite . from ( [ eq8 ] ) with ( [ eq20 ] ) and ( [ eq21 ] )
we obtain [ eq27 ] = ( x , k)+ ( k)|(x , k)e^2iky(x)/. the function @xmath98 is analytic for @xmath91 , @xmath99 is analytic for @xmath100 .
thus , ( [ eq27 ] ) represents an additive * _ _ * riemann - hilbert problem ( rhp ) with a jump on the real line , given by @xmath101 .
the solution of this rhp follows a standard pattern , see @xcite for more details .
the solution simplifies considerably in the case @xmath102 .
this is the @xmath2-soliton solution : [ solk- ] ( x , k)&=&x_0(x)-i_n=1 ^n , k<0 , + & = & x_0(x)-i_n=1 ^n , k>0 .
[ solk+ ] from ( [ solk- ] ) one has a linear system for the quantities @xmath103 with a solution [ eq43a ] ( x ,- i_n , t)=x_0(x)_p=1^n a^-1_np[y , t ] , n=1, ,n . where [ eq43 ] a_pn[y , t]_pn+ .
taking @xmath104 in ( [ solk- ] ) produces e^-(x- ) & & ( x ,- i/2)=x_0(x)(1-_n , p=1 ^n a^-1_np[y , t ] ) . + [ -i/2 ] the substitution @xmath105 in ( [ solk+ ] ) with @xmath106 gives e^(x- ) & & = x_0(x)(1-_n , p=1 ^n a^-1_np[y , t ] ) .
+ [ i/2 ] from ( [ -i/2 ] ) and ( [ i/2 ] ) there follows a parametric representation x&=&x(y , t)+ , [ x = f / f ] + f _ & & 1-_n , p=1 ^n a^-1_np[y , t ] . [ f pm ] from
( [ eq22 ] ) and ( [ x = f / f ] ) one can compute the solution in parametric form [ eq u ] u(x(y , t),t)=x_t(y , t),x = x(y , t ) , [ sol ] where @xmath107 is given in terms of the scattering data in ( [ x = f / f ] ) , ( [ f pm ] ) . upon introducing new notations
j&=&2_j ( -++x_j0 ) + x_j0&= & [ x_0j ] + _ j & = & [ phi_j ] + _ ij&= & ( ) ^2 [ gamma ij ] one can rewrite the expression for @xmath108 ( [ f pm ] ) in the form @xcite f _ & & _ = 0,1 . the solution for @xmath56 can be obtained from ( [ sol ] ) .
first we notice that @xmath109 and also @xmath110 thus @xmath111 .
@xmath112 and @xmath113 m(x , t)&=&_-^p(y , t)(x - x(y , t ) ) y [ m(x , t ) ] + & & p(y , t)=x_t x_y- ( ) _ y [ p(y , t ) ] , + u(x , t)&=&_-^p(y , t)(-|x - x(y , t)| ) y.[u(x , t ) ] as we shall see in section [ ep - higher - sec ] , this representation is also useful in the study of multidimensional solutions .
the spectral problem ( [ eq3 ] ) is gauge equivalent to a standard sturm - liouville problem , well known from the kdv hierarchy , cf .
( [ kdv - isospec ] ) , with short notation @xmath114 : -_yy+u(y)&= & , = - , + ( y)&=&q^1/4 , = , [ chvar ] + u(y)&=&+-- .
[ ep ] note that ( [ chvar ] ) leads to two possible expressions for the change of the variables in the liouville transformation : y&= & x+_-^x ( -)dx+ , [ y1 ] + y&= & x+_^x ( -)dx+ . [ y2 ]
these two possibilities , ( [ y1 ] ) , ( [ y2 ] ) are only consistent iff _ -^ ( -)dx= , which is always the case , since the integral under question is ( up to a multiplier ) the casimir function @xmath93 ( [ eqi8 ] ) .
the matching of the ch hierarchy to kdv hierarchy requires solving the ermakov - pinney equation ( [ ep ] ) @xcite , which is not straightforward and leads to the same solution ( [ sol ] ) in parametric form .
the canonical poisson brackets for the scattering data of the ch equation are computed in @xcite where also the action - angle variables are expressed in terms of the scattering data .
let us consider the action variable for the n - soliton solution .
( these considerations can be extended easily to the variables of the continuous spectrum . )
the angle variable is @xmath115 ; it is linear in time @xmath60 and @xmath116 let us introduce @xmath117 into hamilton s principle @xmath118 , with @xmath119 = \int\bigg ( \ell[u]+\sum_{n=1}^{n}\pi_n\big(\dot{\phi}_n -\lambda_n [ u ] \big ) \bigg ) dt.\ ] ] here the lagrange multiplier @xmath120 enforces the action - angle relation for the ch scattering data as a @xmath121 shift of the angles @xmath122 at constant angular frequencies @xmath123 , with @xmath124 .
the stationary variation is
@xmath125 \big)\delta \pi_n - \sum_{n=1}^{n}\dot{\pi}_n \delta \phi_n \right ) dt.\ ] ] since by definition @xmath126 is the momentum , @xmath118 implies the eulerian representation [ mommap1 ] m==_n=1^n_n , + _
n = _ n [ u ] _ n=0 .
[ toralnaction ] relation ( [ mommap1 ] ) is the _ momentum map _
@xmath127 for the toral @xmath121 action ( [ toralnaction ] ) on the angles @xmath122 at constant angular frequencies @xmath123 .
this momentum map from the action - angle scattering variables @xmath0 to the flow momentum @xmath128 ( dual to the smooth vector fields @xmath129 on the real line ) provides the eulerian representation of the @xmath2-soliton solution of ch in terms of the scattering data and squared eigenfunctions of its isospectral eigenvalue problem .
momentum maps for hamiltonian dynamics are reviewed in @xcite , for example . by using the spectral quantities of the @xmath2-soliton solution ( recall : @xmath130 )
one may express the variation of the spectrum with respect to the ch solution in terms of the squared - eigenfunctions of the isospectral problem as @xcite @xmath131 ^ 2,\ ] ] in which @xmath132 is the eigenfunction that belongs to eigenvalue @xmath133 , see ( [ eq200 ] ) .
on the other hand , the expansion of @xmath134 over squares of eigenfunctions is given by @xcite @xmath135 ^ 2 .\ ] ] consequently , @xmath136 ^ 2,\ ] ] or @xmath137 where @xmath120 and @xmath138 denote explicitly _ n&= & = , + j_n ( x , t ) & & + & = & r_n(t)(1-^2)[|(x ,- i_n , t)]^2 .
thus , the momentum map ( [ mommap1 ] ) from the action - angle variables under going dynamics ( [ toralnaction ] ) to the eulerian representation of the momentum for the ch solution is expressed in terms of the scattering data and squared eigenfunctions of its @xmath2-soliton isospectral eigenvalue problem .
perhaps not unexpectedly , this momentum map may be applied to the action - angle representation of the solution of any integrable hamiltonian pde .
camassa and holm @xcite discovered the `` peakon '' solitary traveling wave solution for a shallow water wave , @xmath139 whose fluid velocity @xmath32 is a function of position @xmath140 on the real line and time @xmath60 .
the peakon traveling wave moves at a speed equal to its maximum height , at which it has a sharp peak ( jump in derivative ) .
peakons are an emergent phenomenon , solving the initial value problem for a partial differential equation derived by an asymptotic expansion of euler s equations using the small parameters of shallow water dynamics .
peakons are _ nonanalytic _ solitons , which superpose as @xmath141 for sets @xmath142 and @xmath143 satisfying canonical hamiltonian dynamics .
peakons arise for shallow water waves in the limit of zero linear dispersion in one dimension .
peakons satisfy a partial differential equation ( pde ) arising from hamilton s principle for geodesic motion on the smooth invertible maps ( diffeomorphisms ) with respect to the @xmath144 sobolev norm of the fluid velocity .
peakons generalize to higher dimensions , as well .
we explain how peakons were derived in the context of shallow water asymptotics and describe some of their remarkable mathematical properties .
peakons were first found as singular soliton solutions of the completely integrable ch equation .
this is equation ( [ dim - b - eqn ] ) with @xmath41 , now rewritten in terms of the velocity , as @xmath145 peakons were found in @xcite to arise in the absence of linear dispersion .
that is , they arise when @xmath146 and @xmath147 in ch ( [ ch - u - eqn ] ) .
specifically , peakons are the individual terms in the peaked @xmath148soliton solution of ch ( [ ch - u - eqn ] ) for its velocity , in the absence of linear dispersion .
each term in the sum ( [ peakontrain - soln ] ) is a solition with a sharp peak at its maximum .
hence , the name `` peakon . '' expressed using its momentum , @xmath149 , the peakon velocity solution ( [ peakontrain - soln ] ) of dispersionless ch becomes a sum over a delta functions , supported on a set of points moving on the real line .
namely , the peakon velocity solution ( [ peakontrain - soln ] ) implies @xmath150 because of the relation @xmath151 .
these solutions satisfy the @xmath34-equation ( [ dim - b - eqn ] ) for any value of @xmath34 , provided @xmath46 and @xmath47 .
as we shall discuss later , the peakon momentum relation ( [ chpeakon - m - soln ] ) is again a momentum map .
thus , peakons are _ singular momentum solutions _ of the dispersionless @xmath34-equation , although they are not stable for every value of @xmath34 . from numerical simulations @xcite ,
peakons are conjectured to be stable for @xmath152 . in the integrable cases @xmath44 for ch and @xmath45 for dp , peakons are stable singular _
soliton _ solutions .
the spatial velocity profile @xmath153 of each separate peakon in ( [ peakontrain - soln ] ) is the green s function for the helmholtz operator on the real line , with vanishing boundary conditions at spatial infinity . unlike the kdv soliton ,
whose speed and width are related , the width of the peakon profile is set by its green s function , independently of its speed . substituting the peakon solution ansatz ( [ peakontrain - soln ] ) and ( [ chpeakon - m - soln ] ) into the dispersionless ch equation , @xmath154 yields _ hamilton s canonical equations _ for the dynamics of the discrete set of peakon parameters @xmath155 and @xmath156 , @xmath157 for @xmath158 , with hamiltonian given by @xcite , @xmath159 or explicitly , _
a&=&_b=1^np_b e^-|q_a - q_b|/ [ q_a ] + _
a&=&_b=1^n p_b e^-|q_a - q_b|/(q_a - q_b ) [ p_a ] thus , one finds that the points @xmath160 in the peakon solution ( [ peakontrain - soln ] ) move with the flow of the fluid velocity @xmath32 at those points , since @xmath161 .
this means the @xmath156 are _ lagrangian _ coordinates .
moreover , the singular momentum solution ( [ chpeakon - m - soln ] ) is the lagrange - to - euler map for an invariant manifold of the dispersionless ch equation ( [ dchmom - eqn ] ) . on this finite - dimensional invariant manifold for the partial differential equation ( [ dchmom - eqn ] ) ,
the dynamics is canonically hamiltonian .
with hamiltonian ( [ h - peakon - ansatz ] ) , the canonical equations ( [ ham - peakon - eqn ] ) for the @xmath162 canonically conjugate peakon parameters @xmath155 and @xmath156 were interpreted in @xcite as describing _ geodesic motion _ on the @xmath148dimensional riemannian manifold whose co - metric is @xmath163 .
moreover , the canonical geodesic equations arising from hamiltonian ( [ h - peakon - ansatz ] ) comprise an integrable system for any number of peakons @xmath2 .
this integrable system was studied in @xcite for solutions on the real line , and in @xcite and references therein , for spatially periodic solutions .
the integrals generated by the action variables in terms of the coordinates can be recovered as tr@xmath164 , where @xmath165 is the lax operator for the peakon system @xcite .
being a completely integrable hamiltonian soliton equation , the continuum ch equation ( [ ch - u - eqn ] ) has an associated isospectral eigenvalue problem , discovered in @xcite for any values of its dispersion parameters @xmath166 and @xmath167 .
remarkably , when @xmath46 and @xmath47 , this isospectral eigenvalue problem has a purely _ discrete _ spectrum .
moreover , in this case , each discrete eigenvalue corresponds precisely to the time - asymptotic velocity of a peakon .
this discreteness of the ch isospectrum in the absence of linear dispersion implies that _ only _ the singular peakon solutions ( [ chpeakon - m - soln ] ) emerge asymptotically in time , in the solution of the initial value problem for the dispersionless ch equation ( [ dchmom - eqn ] ) .
this is borne out in numerical simulations of the dispersionless ch equation ( [ dchmom - eqn ] ) , starting from a smooth initial distribution of velocity @xcite .
figure [ peakon_figure ] shows the emergence of peakons from an initially gaussian velocity distribution and their subsequent elastic collisions in a periodic one - dimensional domain .
this figure demonstrates that singular solutions dominate the initial value problem and , thus , that it is imperative to go beyond smooth solutions for the ch equation ; the situation is similar for the epdiff equation .
[ [ peakons - as - mechanical - systems . ] ] peakons as mechanical systems .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + governed by canonical hamiltonian equations , each @xmath148peakon solution can be associated with a mechanical system of moving particles .
calogero et al .
@xcite further extended the class of mechanical systems of this type .
the r - matrix approach was applied to the lax pair formulation of the @xmath148peakon system for ch by ragnisco and bruschi @xcite , who also pointed out the connection of this system with the classical toda lattice .
a discrete version of the adler - kostant - symes factorization method was used by suris @xcite to study a discretization of the peakon lattice , realized as a discrete integrable system on a certain poisson submanifold of @xmath168 equipped with an r - matrix poisson bracket .
@xcite used the stieltjes theorem on continued fractions and the classical moment problem for studying multi - peakon solutions of the ch equation .
generalized peakon systems are described for any simple lie algebra by alber _
the limit @xmath169 in the @xmath2-soliton solution @xmath134 produces the @xmath2-peakon solution ( [ peakontrain - soln ] ) .
the limiting procedure is described in detail in @xcite .
due to ( [ lambda ] ) one can write for the discrete eigenvalues @xmath170 2_j=(1-_j)^1/2= 1-_j+ .
[ kappa - limit ] the solution ( [ x = f / f ] ) depends explicitly on @xmath171 ( [ kappa - limit ] ) and the limit can be computed with ( [ kappa - limit ] ) by taking @xmath169 and keeping the eigenvalue @xmath172 constant .
the result is the expression ( [ peakontrain - soln ] ) with p_i&=&4d_n - i+1^(0)d_n - i^(2)d_n - i+1^(1)d_n - i^(1),(i=1 , 2 , ... , n),[m_i ] + q_i&=&,(i=1 ,
2 , ... , n ) , [ x_i ] where d_n^(m)=_1i_1<i_2< ... <i_nn_n(i_1,i_2 , ... , i_n)(_i_1_i_2 ... _i_n)^m r_i_1r_i_2 ... r_i_n , + n=1 , 2 , ... , n , _
n(i_1,i_2 , ... , i_n)&=&_1l <
mn(_i_l-_i_m)^2,(n2 ) , + r_i(t)&=&r_i(0)e^t,(x_i0=r_i(0 ) ) .
the quantities @xmath173 are called * _ _ * hankel determinants . by definition @xmath174 . in general ,
hankel determinant is a determinant of @xmath175 matrix of the form @xmath176 where @xmath177 are elements of a sequence , i.e. @xmath178 in this particular case a_l=_i=1^n_i^l r_i(t ) .
[ [ similarity - of - peakon - lattice - and - toda - lattice ] ] similarity of peakon lattice and toda lattice + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + hankel determinants appear in the solutions of other integrable systems , e.g. toda lattice , e.g. see @xcite .
the toda equation @xcite & = & e^q_n+1-q_n - e^q_n - q_n-1 , + & = & p_n , n is one of the most important integrable systems .
we have a finite chain with @xmath2 nodes under the fixed ends boundary conditions @xmath179 .
the solution is in the form q_n(t)&=&q_1(0)+ , + q_1(0)&=&-d^(0)_n(0 ) .the hankel determinants are obtained by a similar sequence , a_l&=&_i=1^n_i^l r_i(t ) , + r_i(t)&=&r_i(0)e^-_it , where @xmath180 are @xmath2 different constants ( eigenvalues of the lax matrix ) and the quantities @xmath181 , @xmath182 represent another set @xmath2 of constants . due to its simple form
, the @xmath2-peakon solution can be used as an approximation of the @xmath2-soliton ch solution when the dispersion term is small ( and the term @xmath183 can be neglected ) .
similarly , the toda chain with complex dynamical variables ( the so - called complex toda chain ctc ) provides an approximation for the n - soliton solution of the nonlinear schredinger equation ( nls ) @xmath184 see @xcite for more details .
such an approximation is called adiabatic approximation and means that the @xmath2-soliton solution consists of @xmath185 well separated solitons @xmath186}}{\cosh ( 2\nu_k(x-\xi_k(t))},\ ] ] i.e. the overlap between the solitons is small . the variables @xmath187 of the ctc are related to the nls solitons parameters by @xmath188 where @xmath189 , @xmath190 , @xmath191 and @xmath192 characterize the center - of - mass position , the phase , velocity and amplitude respectively of the @xmath193-th soliton in the chain ; @xmath194 and @xmath195 are the average amplitude and velocity of the soliton train . @xmath189 and @xmath190 can be obtained as the real and imaginary parts of @xmath196 . such soliton trains and their asymptotic behavior appear to be important for the needs of soliton based fiber optics communications .
suppose we have an integrable system with @xmath162- dimensional phase space which in terms of the action - angle ( canonical ) variables can be represented as _
n=0,_n=_n , n=1,2 , , n , or , if there exists a bracket such that \{_n , _ l } = _ nl , \{_n , _ l}=\{_n , _ l}=0,[brackets]the system is hamiltonian with a hamiltonian h_n=_1 ^ 2+ _n^2 .
[ phamilt]the integrals @xmath123 , @xmath197 are clearly in involution , which guarantees the integrability of the system .
there is however another set of integrals : i_j=(_j-_j+1)(_1+
+_n)-(_j-_j+1 ) ( _ 1+ _n ) + j=1,2, ,n-1.[i_j ] if the action variables @xmath123 are all different , the set ( [ i_j ] ) is functionally independent from the set @xmath123 , @xmath197 .
in addition , the integrals ( [ i_j ] ) form another set of @xmath2-integrals in involution together with @xmath198 . due to the existence of two sets of functionally independent integrals in involution
such systems are termed superintegrable .
an example of such system is toda lattice , see the discussion in @xcite .
the peakon system is also superintegrable .
the canonical variables in terms of the scattering data for ch equation can be used in the peakon limit : @xmath199 , @xmath200 , the hamiltonian ( [ phamilt ] ) is also a peakon limit ( @xmath169 ) of the @xmath2-soliton hamiltonian @xcite h_n ( ) = ^2_n=1^n(+ ) .the poisson bracket for the ch peakon solution is @xmath201 and the scattering data satisfy ( [ brackets ] ) with respect to ( [ pb ] ) , see @xcite for the details .
more interesting are the integrals ( [ i_j ] ) . from ( [ x_i ] ) and ( [ m_i ] )
one can recover these integrals in coordinate form .
for example , when @xmath202 we have i_1 & = & + ( + ) , + & & j=(p_1-p_2)^2 + 4p_1p_2e^-|q_1-q_2|/ .
note that @xmath203 depends on both combinations @xmath204 and @xmath205 as well as the momentum variables .
the hamiltonian @xmath206 ( which depends only on @xmath205 and the momentum variables ) and @xmath203 form a complete system of integrals in involution .
the integration of the 2-peakon system with these integrals can be performed as follows .
first , one can express @xmath207 and @xmath208 in terms of @xmath203 , @xmath206 and the momentum variables : @xmath209 .
next , _ i=+.[alg]the substitution of @xmath210 , @xmath211 from ( [ q_a ] ) ( [ p_a ] ) to ( [ alg ] ) produces an algebraic equation that gives , say @xmath212 as a function of @xmath213 .
then ( [ p_a ] ) is an ode for @xmath213 of the form @xmath214 .
clearly , from practical viewpoint it is much more convenient to work with the other system of integrals in involution : @xmath206 and the conserved momentum @xmath215 .
note that @xmath216 is itself an integral .
[ [ pulsons - generalizing - the - peakon - solutions - of - the - dispersionless - b - equation - for - other - greens - functions . ] ] pulsons : generalizing the peakon solutions of the dispersionless @xmath34-equation for other green s functions .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + the hamiltonian @xmath217 in equation ( [ h - peakon - ansatz ] ) depends on the green s function for the relation between velocity @xmath32 and momentum @xmath56 . however , the singular momentum solution ansatz ( [ chpeakon - m - soln ] ) is _ independent _ of this green s function .
thus , as discovered in fringer and holm @xcite , _ the singular momentum solution ansatz ( [ chpeakon - m - soln ] ) for the dispersionless equation , @xmath218 provides an invariant manifold on which canonical hamiltonian dynamics occurs , for _ any choice _ of the green s function @xmath14 relating velocity @xmath32 and momentum @xmath56 by the convolution @xmath219 .
_ the fluid velocity solutions corresponding to the singular momentum ansatz ( [ chpeakon - m - soln ] ) for equation ( [ fhmom - eqn ] ) are the _
pulsons_. pulsons are given by the sum over @xmath2 velocity profiles determined by the green s function @xmath14 , as @xmath220 again for ( [ fhmom - eqn ] ) , the singular momentum ansatz ( [ chpeakon - m - soln ] ) results in a finite - dimensional invariant manifold of solutions , whose dynamics is canonically hamiltonian .
the hamiltonian for the canonical dynamics of the @xmath162 parameters @xmath155 and @xmath156 in the `` pulson '' solutions ( [ pulson - u - soln ] ) of equation ( [ fhmom - eqn ] ) is @xmath221 again for the pulsons , the canonical equations for the invariant manifold of singular momentum solutions provide a phase - space description of geodesic motion , this time with respect to the co - metric given by the green s function @xmath14 .
mathematical analysis and numerical results for the dynamics of these pulson solutions are given in @xcite .
these results describe how the collisions of pulsons ( [ pulson - u - soln ] ) depend upon their shape .
[ [ compactons - in - the-1alpha2to0-limit - of - ch . ] ] compactons in the @xmath222 limit of ch .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + as mentioned earlier , in the limit that @xmath39 , the ch equation ( [ ch - u - eqn ] ) becomes the kdv equation . in the opposite limit
that @xmath222 ch becomes the hunter - zheng equation @xcite @xmath223 this equation has `` compacton '' solutions , whose collision dynamics was studied numerically and put into the present context in @xcite .
the corresponding green s function satisfies @xmath224 , so it has the triangular shape , @xmath225 for @xmath226 , and vanishes otherwise , for @xmath227 .
that is , the green s function in this case has compact support ; hence , the name `` compactons '' for these pulson solutions , which as a limit of the integrable ch equations are true solitons , solvable by ist .
[ [ pulson - solutions - of - the - dispersionless - b - equation . ] ] pulson solutions of the dispersionless @xmath34-equation .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + holm and staley @xcite give the pulson solutions of the traveling wave problem and their elastic collision properties for the dispersionless @xmath34-equation , @xmath228 with any ( symmetric ) green s function @xmath14 and for any value of the parameter @xmath34 .
numerically , pulsons and peakons are both found to be stable for @xmath152 , @xcite .
the reduction to _ noncanonical _ hamiltonian dynamics for the invariant manifold of singular momentum solutions ( [ chpeakon - m - soln ] ) of the other integrable case @xmath42 with peakon green s function @xmath229 is found in @xcite .
[ [ generalizing - the - peakon - solutions - of - the - ch - equation - to - higher - dimensions . ] ] generalizing the peakon solutions of the ch equation to higher dimensions .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + in @xcite , weakly nonlinear analysis and the assumption of columnar motion in the variational principle for euler s equations were found to produce the two - dimensional generalization of the dispersionless ch equation ( [ dchmom - eqn ] ) .
this generalization is the euler - poincar ( ep ) equation @xcite for the lagrangian consisting of the kinetic energy , @xmath230dxdy \ , , \label{ke - def}\ ] ] in which the fluid velocity @xmath231 is a two - dimensional vector .
evolution generated by kinetic energy in hamilton s principle results in geodesic motion , with respect to the velocity norm @xmath232 , which is provided by the kinetic energy lagrangian . for ideal incompressible fluids governed by euler s equations , the importance of geodesic flow was recognized by arnold @xcite for the @xmath233 norm of the fluid velocity . the ep equation generated by any choice of kinetic energy norm without imposing incompressibility
is called `` epdiff , '' for `` euler - poincar equation for geodesic motion on the diffeomorphisms . ''
epdiff is given by @xcite @xmath234 with momentum density @xmath235 where @xmath236 is given by the kinetic energy , which defines a norm in the fluid velocity @xmath232 , yet to be determined . by design
, this equation has no contribution from either potential energy , or pressure .
it conserves the velocity norm @xmath232 given by the kinetic energy .
its evolution describes geodesic motion on the diffeomorphisms with respect to this norm @xcite .
an alternative way of writing the epdiff equation ( [ epdiff - eqn ] ) in either two , or three dimensions is , @xmath237 this form of epdiff involves all three differential operators , curl , gradient and divergence . for the kinetic energy lagrangian @xmath238
given in ( [ ke - def ] ) , which is a norm for _ irrotational _ flow ( with @xmath239 ) , we have the epdiff equation ( [ epdiff - eqn ] ) with momentum @xmath240 .
epdiff ( [ epdiff - eqn ] ) may also be written intrinsically as @xmath241 where ad@xmath242 is the @xmath233 dual of the ad - operation ( commutator ) for vector fields .
see @xcite for additional discussions of the beautiful geometry underlying this equation .
[ [ reduction - to - the - dispersionless - ch - equation - in-1d . ] ] reduction to the dispersionless ch equation in 1d .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + in one dimension , the epdiff equation ( [ epdiff - eqn]-[epdiffeo - eqn ] ) with lagrangian @xmath238 given in ( [ ke - def ] ) simplifies to the dispersionless ch equation ( [ dchmom - eqn ] ) .
the dispersionless limit of the ch equation appears , because we have ignored potential energy and pressure .
[ [ strengthening - the - kinetic - energy - norm - to - allow - for - circulation . ] ] strengthening the kinetic energy norm to allow for circulation .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + the kinetic energy lagrangian ( [ ke - def ] ) is a norm for irrotational flow , with @xmath239 .
however , inclusion of rotational flow requires the kinetic energy norm to be strengthened to the @xmath243 norm of the velocity , defined as @xmath244dxdy \nonumber\\ & = & \frac{1}{2 } \int \bigg[|{\mathbf{u}}|^2 + \alpha^2|\nabla{\mathbf{u}}|^2 \bigg]dxdy = \frac{1}{2 } \|{\mathbf{u}}\|^2_{h_\alpha^1 } \ , .
\label{h1-eqn}\end{aligned}\ ] ] here we assume boundary conditions that give no contributions upon integrating by parts .
the corresponding epdiff equation is ( [ epdiff - eqn ] ) with @xmath245 this expression involves inversion of the familiar _ helmholtz _ operator in the ( nonlocal ) relation between fluid velocity and momentum density .
the @xmath243 norm @xmath246 for the kinetic energy ( [ h1-eqn ] ) also arises in three dimensions for turbulence modeling based on lagrangian averaging and using taylor s hypothesis that the turbulent fluctuations are `` frozen '' into the lagrangian mean flow @xcite .
[ [ generalizing - the - ch - peakon - solutions - to - n - dimensions . ] ] generalizing the ch peakon solutions to @xmath247 dimensions .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + building on the peakon solutions ( [ peakontrain - soln ] ) for the ch equation and the pulsons ( [ pulson - u - soln ] ) for its generalization to other traveling - wave shapes in @xcite , holm and staley @xcite introduced the following measure - valued singular momentum solution ansatz for the @xmath248dimensional solutions of the epdiff equation ( [ epdiff - eqn ] ) : with @xmath249 .
in contrast , the similar expression ( [ m(x , t ) ] ) for the soliton solutions represent smooth functions @xmath250 . ]
@xmath251 these singular momentum solutions , called `` diffeons , '' are vector density functions supported in @xmath252 on a set of @xmath2 surfaces ( or curves ) of codimension @xmath253 for @xmath254 with @xmath249 .
they may , for example , be supported on sets of points ( vector peakons , @xmath255 ) , one - dimensional filaments ( strings , @xmath256 ) , or two - dimensional surfaces ( sheets , @xmath257 ) in three dimensions .
figure [ 2dstrip_plate ] shows the results for the epdiff equation when a straight peakon segment of finite length is created initially moving rightward ( east ) . because of propagation along the segment in adjusting to the condition of zero speed at its ends and finite speed in its interior , the initially straight segment expands outward as it propagates and curves into a peakon `` bubble . ''
figure [ 2dstrip_plate - multi ] shows an initially straight segment whose velocity distribution is exponential in the transverse direction , but is wider than @xmath49 for the peakon solution .
this initial velocity distribution evolves under epdiff to separate into a train of curved peakon `` bubbles , '' each of width @xmath49 .
this example illustrates the emergent property of the peakon solutions in two dimensions .
this phenomenon is observed in nature , for example , as trains of internal wave fronts in the south china sea @xcite .
for the peakon solution will break up into a train of curved peakon `` bubbles , '' each of width @xmath49 .
this example illustrates the emergent property of the peakon solutions in two dimensions . ]
substitution of the singular momentum solution ansatz ( [ m - ansatz ] ) into the epdiff equation ( [ epdiff - eqn ] ) implies the following integro - partial - differential equations ( ipdes ) for the evolution of the parameters @xmath258 and @xmath259 , @xmath260 importantly for the interpretation of these solutions , the coordinates @xmath254 turn out to be _ lagrangian _ coordinates .
the velocity field corresponding to the momentum solution ansatz ( [ m - ansatz ] ) is given by @xmath261 for @xmath262 .
when evaluated along the curve @xmath263 , this velocity satisfies , @xmath264 consequently , the lower - dimensional support sets defined on @xmath263 and parameterized by coordinates @xmath265 move with the fluid velocity .
this means the @xmath265 are lagrangian coordinates .
moreover , equations ( [ intdiffeqn - q ] ) for the evolution of these support sets are canonical hamiltonian equations , @xmath266 the corresponding hamiltonian function @xmath267 is , @xmath268 this is the hamiltonian for geodesic motion on the cotangent bundle of a set of curves @xmath269 with respect to the metric given by @xmath270 .
this dynamics was investigated numerically in @xcite to which we refer for more details of the solution properties .
one important result found _ numerically _ in @xcite is that only codimension - one singular momentum solutions appear to be stable under the evolution of the epdiff equation .
thus , we have _ stability for codimension - one : the singular momentum solutions of epdiff are stable , as points on the line ( peakons ) , as curves in the plane ( filaments , or wave fronts ) , or as surfaces in space ( sheets ) .
_ proving this stability result analytically remains an outstanding problem .
the stability of peakons on the real line is proven in @xcite .
[ [ reconnections - in - oblique - overtaking - collisions - of - peakon - wave - fronts . ] ] reconnections in oblique overtaking collisions of peakon wave fronts .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + figures [ 2dstrip_skew ] shows results of oblique wave front collisions producing reconnections for the epdiff equation in two dimensions .
figure [ 2dstrip_skew ] shows a single oblique overtaking collision , as a faster expanding peakon wave front overtakes a slower one and reconnects with it at the collision point .
[ [ the - peakon - solution - ansatz - is - a - momentum - map . ] ] the peakon solution ansatz is a momentum map .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + as shown in @xcite , the solution expressions ( [ chpeakon - m - soln ] ) in one dimension and ( [ m - ansatz ] ) in higher dimensions may be interpreted as equivariant momentum maps , from the cotangent bundle of the smooth embeddings of lower dimensional sets @xmath271 to the dual of the lie algebra of vector fields defined on these sets .
( momentum maps for hamiltonian dynamics are reviewed in @xcite , for example . )
the result that the singular solution ansatz is a momentum map helps to organize the theory , to explain previous results and to suggest new avenues of exploration .
this geometric feature underlies the remarkable reduction properties of the ch and epdiff equations , and _ explains _ why they must be lie - poisson hamiltonian equations .
this is because of the general fact that equivariant momentum maps are poisson maps .
this geometric feature also underlies the singular momentum solution ( [ m - ansatz ] ) and its associated velocity ( [ u - ansatz ] ) which generalize the peakon solutions , both to higher dimensions and to arbitrary kinetic energy metrics . as we saw in section [ mommap - sec ] ,
the soliton solution ( [ mommap1 ] ) is also a momentum map .
this soliton momentum map may be expected to apply in the action - angle representation of the solution of any integrable hamiltonian pde .
its further properties will be studied in detail elsewhere .
( 1 . ) throughout this discussion the solutions @xmath134 were confined to be functions in the schwartz class , @xmath272 .
the situation when the condition @xmath273 on the initial data does not hold is more complicated and requires separate analysis @xcite . in general
, it leads to wave - breaking @xcite .
an attempt at developing the inverse scattering theory for this case has been made by kaup @xcite , who suggested applying the inverse scattering approach separately in each interval where @xmath274 is of the same sign .
the problem however is how to join solutions that are valid in different intervals .
the peakon solution ( [ m_i ] ) was obtained from the soliton solution under the assumption @xmath273 .
thus , all @xmath275 are of the same sign , since all the eigenvalues @xmath133 in this case are positive .
however , one can formally use the same solution with eigenvalues of various signs to model @xmath275 of various signs ( mixture of peakons and ` anti - peakons ' ) and thus to study peakon
anti - peakon interactions , see e.g. @xcite .
the result is that the multi - peakon interaction ( including anti - peakons ) in general decomposes into a sequence of pairwise collisions @xcite .
the collision of a single peakon - anti - peakon pair was studied already in @xcite . when the eigenvalues are of mixed signs , the hankel determinants in the denominator of ( [ m_i ] )
may develop singularities for finite values of @xmath60 .
this ` peakon - breaking ' phenomenon is apparently the analog of the wave - breaking mentioned earlier when @xmath276 .
this needs to be investigated further .
ddh is grateful to r. camassa , j. e. marsden , t. s. ratiu and a. weinstein for their collaboration , help and inspiring discussions over the years .
ddh gratefully acknowledges partial support by the royal society of london s wolfson award scheme .
rii acknowledges funding from a marie curie intra - european fellowship .
both authors thank j. percival and m. f. staley for providing figures .
alber m s , camassa r , fedorov y , holm d d and marsden j e 1999 _ phys .
a _ * 264 * 171 ; + ibid 2001 _ comm .
phys . _ * 221 * 197 ; + alber m s , camassa r , holm d d and marsden j e 1994 _ lett . math . phys . _
* 32 * 137 + alber m s , camassa r and gekhtman m 2000 _ crm proc . and
lecture notes _ * 25 * 1 degasperis a and procesi m 1999 asymptotic integrability , in:_symmetry and perturbation theory _
, ed a degasperis and g gaeta ( singapore : world scientific ) p 23 ; + degasperis a , holm d d and hone a n w 2002 _ theoret . and math .
phys . _ * 133 * 1463 dubrovin b a 1981 _ russ .
surv . _ * 36 * 11 ; + dubrovin b a , novikov s p and krichever i m 1985 integrable systems .
i in : _ itogi nauki i tekhniki .
probl . mat .
naprav . _ * 4 * ( moscow : viniti ) ( russian ) ; 1989 _ encyclopaedia of mathematical sciences _ * 4 * ( berlin : springer ) ( engl . transl . )
holm d d and marsden j e 2004 momentum maps and measure - valued solutions ( peakons , filaments and sheets ) for the epdiff equation .
_ in : the breadth of symplectic and poisson geometry _ , _ progr
_ * 232 * ed j e marsden and t s ratiu ( boston : birkhuser ) p 203 | the relations between smooth and peaked soliton solutions are reviewed for the camassa - holm ( ch ) shallow water wave equation in one spatial dimension .
the canonical hamiltonian formulation of the ch equation in action - angle variables is expressed for solitons by using the scattering data for its associated isospectral eigenvalue problem , rephrased as a riemann - hilbert problem .
the momentum map from the action - angle scattering variables @xmath0 to the flow momentum ( @xmath1 ) provides the eulerian representation of the @xmath2-soliton solution of ch in terms of the scattering data and squared eigenfunctions of its isospectral eigenvalue problem .
the dispersionless limit of the ch equation and its resulting peakon solutions are examined by using an asymptotic expansion in the dispersion parameter .
the peakon solutions of the dispersionless ch equation in one dimension are shown to generalize in higher dimensions to peakon wave - front solutions of the epdiff equation whose associated momentum is supported on smoothly embedded subspaces .
the eulerian representations of the singular solutions of both ch and epdiff are given by the ( cotangent - lift ) momentum maps arising from the left action of the diffeomorphisms on smoothly embedded subspaces .
pacs numbers : 47.10.+g , 52.65.kj , 45.20.jj , 47.65.+a , 02.40.yy , 11.10.ef , 05.45.-a [ [ keywords ] ] keywords : + + + + + + + + + solitons , peakons , hamilton s principle , integrable hamiltonian systems , inverse spectral transform ( ist ) , singular solutions , shallow water waves , euler s equations , geodesic motion , diffeomorphisms , sobolev norms , momentum maps , compactons |
in this paper we assess primordial nucleosynthesis in light of the new lyman limit system - derived deuterium abundance data @xcite . based on our analysis
we adopt the view that currently we do not have a reliable handle on the primordial abundances of @xmath1he , @xmath2he , and @xmath3li ; whereas , the keck telescope may be providing us with a direct measurement of the abundance of the very fragile deuteron in relatively chemically - unevolved primordial material .
this is a somewhat radical view , given that much of the past discussion on big bang nucleosynthesis ( bbn ) has been predicated on the idea of concordance where a single global specification of the parameters characterizing nuclear statistical equilibrium ( nse ) freeze - out nucleosynthesis can lead the calculated abundances to agree with their observationally - inferred primordial values within errors . in a computation of nse freeze - out nucleosynthesis @xcite ,
it is necessary to specify : ( 1 ) either the entropy - per - baryon @xmath11 ( cosmic average in units of boltzmann s constant @xmath12 ; where @xmath13 is the baryon closure fraction and @xmath14 is the hubble parameter in units of @xmath15 ) , or the baryon - to - photon ratio @xmath4 ( cosmic average @xmath16 ) ; ( 2 ) the spatial distribution of either of these quantities , both on scales smaller than the particle horizon at the epoch of nse freeze - out and on super - horizon scales ; ( 3 ) the three net lepton - to - photon numbers ; ( 4 ) the ratio of the axial vector to vector weak interaction coupling constants ( derived from the neutron lifetime ) ; and ( 5 ) the expansion rate through the epoch of freeze - out .
the expansion rate is determined by the energy density in the horizon . in turn , in the expected radiation - dominated conditions of bbn , the energy density is usually parametrized as an effective number of light neutrino species @xmath9 , representing all relativistic particle degrees of freedom beyond those contributed by photons , electrons and positrons .
the usual procedure has been to compute nucleosynthesis yields as a function of @xmath4 , for specifications of @xmath9 and the neutron lifetime , _ assuming _ that the entropy is homogeneously distributed on all scales and _ assuming _ that all net lepton numbers are small ( @xmath17 ) and neutrino masses are small and there are no other relativistic degrees of freedom .
these simple occam s razor assumptions are then justified by finding a concordant @xmath4 where all the independently - determined primordial abundances line up with their values predicted in the calculations ( _ cf .
such concordance - based justification has been touted as being all the more impressive and secure , given that the predicted and observationally - determined abundances range over some ten orders of magnitude .
the claimed precisions in the determinations of the primordial abundances of @xmath0h , @xmath1he , @xmath2he , and @xmath3li have increased to the point where , if taken at face value , they invalidate the simple picture of concordance outlined above @xcite .
this is the recent so - called crisis in bbn .
in fact , there were even earlier hints at a potential problem with the standard picture of concordance @xcite . with their adopted abundances of the @xmath0h and @xmath2he , the authors of ref .
@xcite find no concordance for @xmath8 , and derive a best fit to concordance for @xmath18 with @xmath8 ruled out at the @xmath19 c.l .. these authors have suggested that perhaps this discrepancy could be eliminated with the introduction of new neutrino physics , essentially relaxing the usual assumptions regarding the above - discussed parameters ( 3 ) and ( 5 ) in bbn .
this conclusion and interpretation has been disputed by copi _ et .
@xcite , who take as a prior assumption that @xmath8 and then argue that @xmath2he has been underestimated .
cardall and fuller ( @xcite , see also @xcite ) have done a re - analysis of this problem in light of the discordant determinations of @xmath0h in ly-@xmath20 clouds and with special attention to the dependence of the bbn yields on @xmath9 and to the @xmath3li non - concordance problem ( which they conclude can not be rectified with new neutrino physics ) .
this work tends to support the copi _
as we shall see , new work on the observationally - inferred abundances of @xmath0h and @xmath2he provides further support for this view and offers a hint of a new concordance .
the deuteron is the most fragile of all nuclei , with a binding energy of only @xmath21 .
as a result , the @xmath0h yield in bbn is exponentially sensitive to @xmath4 , though only mildly sensitive to @xmath9 @xcite .
however , the fragile nature of the deuteron makes it extremely vulnerable to even small amounts of stellar processing and this , in turn , calls into question claims that we have a reliable handle on the primordial deuterium abundance .
direct measurements using the keck telescope of isotope - shifted hydrogen lines in high redshift ly - limit systems along lines of sight to distant qso s may completely circumvent these stellar processing issues @xcite .
this is because these systems have manifestly low metallicity and this argues against significant stellar processing - induced destruction of @xmath0h @xcite . at present
, however , there is no consensus on the primordial value of @xmath0h / h from this technique : the seattle - hawaii group @xcite obtains a very high range for this quantity ( @xmath22 from ref .
@xcite ) ; whereas , the san diego group @xcite examining different clouds obtains a consistently much lower range ( @xmath23 , combined results from refs .
@xcite and @xcite with @xmath24 statistical error and @xmath25 systematic error ) . here
@xmath26h / h @xmath27 .
an analysis of these discordant ranges is performed in ref .
@xcite . in this analysis ,
we have adapted the kawano bbn code ( with neutron lifetime from ref .
@xcite , reaction rates from ref .
@xcite , and a small , nearly @xmath4-independent correction to the @xmath2he yield , @xmath28 , from the time step and weak rate corrections of ref .
the bbn yields for @xmath0h , @xmath2he , and @xmath3li are plotted as functions of @xmath9 and @xmath4 .
such plots give insight into the leverage which each light element species has on concordance . since , in contrary fashion to @xmath0h , the bbn @xmath2he yield is relatively sensitive to @xmath9 and much less sensitive to @xmath4 , the tension between the overlap of these two species on our plots provides the most strigent criterion for concordance .
this analysis shows that for the typically adopted range of the @xmath2he abundance @xcite and the spite plateau @xmath3li abundance measured in old , hot pop ii halo stars @xcite , concordance is readily attained for the seattle - hawaii @xmath0h for @xmath8 . on the other hand ,
no concordance with these values of @xmath2he and @xmath3li is possible for _ any _ @xmath9 with the san diego deuterium determination .
lack of concordance with the spite plateau @xmath3li abundance may not be a serious problem : this determination of primordial @xmath3li is fraught with potential sytematic uncertainty , as @xmath3li is destroyed readily by @xmath29 at temperatures as low as @xmath30 .
therefore , rotation - induced mixing and turbulent diffusion could have destroyed most of the original @xmath3li on the surfaces of the old halo stars @xcite . however , the issue is complicated by the claimed observation of the even more fragile species @xmath31li in some of these objects@xcite .
if mixing - induced destruction ( _ depletion _ ) of @xmath3li has indeed occurred , then the @xmath31li could have been produced _ in situ _
@xcite ; alternatively , stellar wind - driven mass loss could deplete @xmath3li while leaving some @xmath31li present @xcite .
we here present plots showing the concordance situation when allowance is made for some @xmath3li depletion . in figure 1
we show the @xmath4-@xmath9 parameter space corresponding to the seattle - hawaii deuterium range ( dotted lines ) , along with the ranges for @xmath2he ( @xmath32 , solid lines ) and @xmath3li ( @xmath33 , dashed lines ) taken from ref .
@xcite . here
@xmath34 is the primordial mass fraction of @xmath2he , and @xmath35 .
this range of @xmath36 reflects @xmath24 statistical errors and @xmath37 sytematic errors , and allows for a factor of @xmath38 depletion ( contained in the sytematic error ) on the spite plateau .
concordance of all of these species is apparent for @xmath8 in figure 1 .
however , if we adopt the san diego deuterium range as primordial , with the same ranges adopted for the other species ( figure 2 ) , then the crisis is evident , as there is no statistically significant overlap for @xmath8 .
the quality of the san diego group s data and the sophistication of their self - consistent analysis can not be dismissed .
furthermore , there are a number of astrophysical problems which are partially or completely ameliorated if the san diego group s deuterium range is adopted as primordial .
this range for @xmath0h / h would correspond to a range in baryonic closure fraction @xmath39 .
in contrast , adoption of the seattle - hawaii @xmath0h / h as primordial would imply the much smaller closure fraction @xmath40 .
there are hints from the x - ray galaxy cluster problem @xcite that the higher range for @xmath41 is to be preferred_e.g . _
the coma cluster apparently has a fractional baryonic mass @xmath42 or @xmath43 for @xmath44 .
likewise , the macho project gravitational microlensing results @xcite suggest that most or all of the galactic dark halo mass ( corresponding to @xmath45 to @xmath46 ) is composed of objects with masses in the range @xmath47 to @xmath48 .
it is clear that the seattle - hawaii deuterium range implies a baryonic closure fraction which is difficult to reconcile with these macho results if the lensing objects had baryonic progenitors .
finally , observation of the cosmic background radiation doppler peaks may require a high range of @xmath49 @xcite .
an alternative to the adoption of the san diego deuterium as primordial would be to invoke super - horizon scale entropy fluctuations at the epoch of bbn ( _ i.e. _ , relax the homogeneity assumption of bbn parameter 2 ) @xcite .
though such a scheme could give a comfortably high @xmath50 @xcite , it would require a fair degree of fine tuning of the fluctuation spectrum and would _ demand _ that the cosmic average primordial deuterium abundance be high @xmath51 @xcite .
present statistics of lyman limit sytems and other uncertainties would have to improve significantly to establish such intrinsic inhomogeneity @xcite , so it seems reasonable to discount this scheme at present and see if a concordance can be found with adoption of the san diego deuterium as a homogeneous primordial value .
a recent reinvestigation ( with new data ) of the linear regression method for estimating the primordial @xmath2he abundance has called into question the systematic uncertainties assigned to @xmath52 @xcite .
in fact , this study derives @xmath53 , where the @xmath54 error is statistical .
it is clear that the central value of this result is well above twice the systematic error assigned by refs .
@xcite , confirming the suspicion @xcite that @xmath52 is not known well enough to draw sweeping conclusions regarding a lack of concordance . in figure 3
we show both the san diego and seattle - hawaii @xmath0h ranges ( dotted lines ) , along with the same @xmath3li range range ( dashed lines ) employed in figure 1 , but now with a band ( solid lines ) for @xmath2he ( @xmath55 , reflecting @xmath24 statistical errors and no systematic uncertainty ) meant to be representative of the ref .
@xcite results .
it is evident from this figure that there is now no statistically significant concordance between @xmath2he and the seattle - hawaii @xmath0h for @xmath8 , while there is now a hint of concordance for the san diego @xmath0h range for @xmath8 .
the new concordance engendered by the san diego deuterium would be even better if allowance for systematic error were to be made in the @xmath2he range .
such a new concordance would probably still require significant depletion of @xmath3li in old , hot pop ii halo stars @xcite , though the classic constraints on neutrino physics from bbn @xcite would survive intact and could be strengthened @xcite . | recent determinations of the deuterium abundance , @xmath0h / h , in high redshift lyman limit hydrogen clouds challenge the usual picture of primordial nucleosynthesis based on concordance of the calculated light element ( @xmath0h , @xmath1he , @xmath2he , @xmath3li ) nucleosynthesis yields with the observationally - inferred abundances of these species .
concordance implies that all light element yields can be made to agree with the observationally - inferred abundances ( within errors ) for single global specifications of the baryon - to - photon ratio , @xmath4 ; lepton number ; neutron lifetime ; and expansion rate ( or equivalently , effective number of light neutrino degrees of freedom @xmath5 ) .
though one group studying lyman limit systems obtains a high value of @xmath0h / h ( @xmath6 ) , another group finds consistently low values ( @xmath7 ) . in the former case , concordance for @xmath8
is readily attained for the current observationally - inferred abundances of @xmath2he and @xmath3li .
but if the latter case represents the primordial deuterium abundance , then concordance for _ any _ @xmath9 is impossible unless the primordial value of @xmath3li / h is considerably larger than the abundance of lithium as measured in old , hot pop ii halo stars .
furthermore , concordance with @xmath10 is possible for low @xmath0h / h only if either ( 1 ) the primordial @xmath2he abundance has been significantly underestimated , or ( 2 ) new neutrino sector physics is invoked .
we argue that systematic underestimation of both the @xmath3li and @xmath2he primordial abundances is the likely resolution of this problem , a conclusion which is strengthened by new results on @xmath2he . |
an assumption that the pairing state in the high - temperature superconductors is the @xmath1-wave - like one is widely discussed nowadays .
@xcite although up to now it apparently can not be treated as the uniquely established fact there exist a large body of experimental data which supports it .
@xcite such a pairing would result in presence of `` nodal '' points in the superconducting excitations spectrum in whose vicinities it has gapless phonon - like character .
it should essentially modify the damping behavior comparing to the @xmath2-wave pairing state .
the quasiparticle damping owing to impurities has been extensively studied in literature .
@xcite however , this is not the only channel , though , maybe , the more interesting one .
the excitations scattering by phonons and also by each others always contribute to the damping , as well . in the present paper
we evaluate the quasiparticle damping in two - dimensional ( 2d ) superconductors with @xmath1-wave - like pairing appeared due to their four - fermionic interaction
. it may be the on - site coulomb repulsion in the small-@xmath3 hubbard model , or the residual part of the interaction which provides the pairing in the strong correlations scheme .
@xcite in the bcs case , the damping at zero temperature has the threshold at @xmath4 .
we show that for the gap function with nodes it demonstrates gapless @xmath0 low - frequency behavior at the nodal points .
the results obtained for high frequencies depend on details of fermi surface ( fs ) parametrization .
the damping for the circular fs will increase quadratically while for the spectrum with van - hove singularities the additional linear regime appears .
being generalized onto finite temperatures , this results allow us to conclude about existence of crossovers in temperature dependence of damping in the superconducting state .
the outline of this paper is as follows . in sect .
ii we obtain general expression for damping of superconducting excitations due to their interaction . in sect .
iii we evaluate it for the case of 2d superconductor with nodes in the spectrum .
since the high - frequency behavior of damping depends strongly on fs parametrization , in the consequent subsections we analyze three principal cases : ( i ) the case of circular fs and also the case of fs of varying curvature when the gap node intersects it ( ii ) at the almost flat piece or ( iii ) in the vicinity of saddle - point of the spectrum . at last , in sect .
iv we generalize and discuss the results obtained .
it is convenient to represent the hamiltonian @xmath5 in the nambu notation @xmath6 here @xmath7 is the interaction , @xmath8 stands for the spinor with @xmath9 being the fermionic creation operator , @xmath10 and @xmath11 are the normal - state conduction band and the gap function ( gf ) , respectively , and @xmath12 , where @xmath13 denote the pauli matrices . let us define the bare matrix green function as follows @xmath14 where normal @xmath15 and anomalous @xmath16 green functions are written as @xmath17 here @xmath18 is the imaginary matsubara frequency and the excitation spectrum is given by @xmath19^{1/2}$ ]
. within this notation the dyson equation obtains the form @xmath20 , with the diagonal self - energy operator @xmath21 . the first nonvanishing contribution to @xmath22
is given by @xmath23,\end{aligned}\ ] ] where the indices @xmath24 label the intermediate frequencies and momenta and @xmath25 corresponds to the external one .
after the analytic continuation @xmath26 for the imaginary part of the self - energy at @xmath27 and @xmath28 we get @xmath29 , \end{aligned}\ ] ] where the coefficients of bogolyubov transform are defined by @xmath30
in this section we will evaluate eq .
( 5 ) in the case of 2d superconductor with unconventional pairing assuming different model parametrizations for the fermi surface .
we shall suppose that the gap function has nodal lines @xmath31 in the 2d brillouin zone .
let these lines intersect fermi surface at a set of the points @xmath32 .
then in the vicinities of such points one can write @xmath33 where @xmath34 and the index @xmath35 labels the nodal point .
we see that the excitation spectrum near these points has gapless linear character .
in general , the quasiparticle damping due to their interaction has to reveal threshold feature in frequency in the superconducting state .
the threshold corresponds to the minimal energy of the excitation needed to decay into three quasiparticles in the intermediate state . for the gap function with nodes
however it disappears at the nodal points of the spectrum .
consequently , the damping at this points appear to be gapless . taking it into account
, we restrict ourselves in treatment of damping in the vicinities of nodal points only . the straightforward calculations at @xmath36 yield @xmath37 where the indices @xmath38 run over all the nodal points .
here @xmath39 are some dimensionless positive factors .
the diagonal terms of the matrix @xmath40 ( i.e. , referring to a single nodal point ) are always numbers of order of unity while the off - diagonal ones could have additional smallness for particular fs parametrizations via their dependence on spectrum parameters .
the principal contribution to the damping at low frequencies comes from excitations with momenta lying in the small elliptic vicinities of nodal points ( see fig . [ 1 ] ) wherein linear approximation for the spectrum works well .
note that we obtain the universal cubic frequency dependence of the excitations damping at the nodal points without appealing to any fixed parametrization for spectrum . by the way , such dependence means the stability of nodes with respect to this kind of interaction . at the same time , some features of the model may , in principle , lead to different power laws . in particular , the damping would increase with frequency slower than @xmath41 if at least one of spectrum zeroes ( i ) appeared due to tangency of fs and gf node , or ( ii ) coincided with van - hove singularity in the band structure and/or ( iii ) with cusp- or discontinuity - point of the nodal line of the gap function .
all these cases are obviously quite exotic . when the frequency increases , the integration in eq .
( 5 ) can not be restricted to the nodal vicinities and such characteristics as general shape of fs , curvature of fs at the nodal points , existence of van - hove singularities in the spectrum , @xmath42 should be taken into account .
one can distinguish three qualitatively different cases : ( i ) the circular fs and also the fs of varying curvature with the nodal points located ( ii ) far from or ( iii ) at the sharp curvature maximum .
we shall illustrate the above cases by examples which would allow us to examine in detail the intermediate asymptotics and to express the corresponding crossover scales through the parameters of the models used . putting the lattice parameter to be unity , we write the circular fs as follows : @xmath43 . for definiteness
, we choose the @xmath44-wave gap function , @xmath45 $ ] .
@xcite in line with eq .
( 8) we get : @xmath46 where @xmath47 . the principal contribution to @xmath48
is obtained when all the momenta @xmath49 , @xmath50 and @xmath51 are almost parallel and belong to the vicinities of the nodal points ( see fig .
[ 1 ] ) . at higher frequencies @xmath52
we find @xmath53\end{aligned}\ ] ] to obtain eq .
( 10 ) , the curvature of fs was taken into account ( see appendix ) .
the crossover from the cubic dependence to the quadratic one appears at @xmath54 .
it can be qualitatively explained on the following way .
one can show that , when @xmath55 , the areas of integration defined by arguments of @xmath56-functions in eq .
( 5 ) are located near the nodal points ( see fig . [ 1 ] ) where the spectrum can be well approximated by ellipses . when @xmath57 , the areas of integration propagate along the whole fermi surface so that not only the vicinities of the nodal points principally contribute to @xmath58
. for @xmath59 , we obtain the quadratic intermediate asymptotic regime with unconventional logarithmic factor of the form @xmath60 which stems from integration over the fs . at frequencies
higher then the gap value @xmath61 information about the superconducting state disappears , and the conventional 2d fermi - liquid result @xcite is restored .
note that principal contributions to @xmath62 in the latter case again come from almost parallel @xmath63 , @xmath49 and @xmath50 .
now we will demonstrate how the deformation of fs affects the damping . as before we shall consider @xmath65 gap function .
let us adopt the simplest tight - banding form for the normal - state dispersion : @xmath66+\mu$ ] . here
@xmath67 is the nearest - neighbors hopping integral and @xmath68 is the chemical potential measured from the half - filling .
note that for small enough @xmath69 fermi surface has long almost flat pieces near the points of intersection with the nodal lines of the @xmath65 gap function ( see fig .
[ 2]a ) . under the above conditions one can easily rewrite the low - frequency asymptotics of damping by substituting @xmath70 in the parenthesis and @xmath71 in other parts of eq .
the range of validity for such expression is restricted by the condition @xmath72 .
next , in order to obtain expression for damping at @xmath73 , the same substitutions in eq .
( 10 ) should be done .
obviously , all the statements concerning the areas of principle contributions still hold .
finally , at @xmath74 we obtain @xmath75 thus , the additional linear - in-@xmath76 high - frequency regime appears in this case .
it occurs since at @xmath77 the areas of integration in eq .
( 5 ) engage the saddle - points of the normal - state spectrum located at @xmath78 and @xmath79 . for the normal state crossover to the linear behavior
has been demonstrated in @xcite .
it follows from the aforementioned that there is no room for linear regime of damping at @xmath80 while the crossover from the cubic dependence to the quadratic one occurs at @xmath81 .
this case can be treated as the one of slightly deformed circular fs without additional characteristic scales .
when @xmath68 decreases ( i.e. , close to the half - filling ) the long almost flat pieces of the fs appear as well as the saddle - points in the normal - state spectrum come into play . for @xmath82 there
co - exist three ranges with cubic , quadratic and linear frequency dependencies matching at @xmath83 and @xmath84 , respectively . at last , for very small @xmath85 linear approximation for spectrum works well unless the ellipses touch the saddle - points .
consequently , the range of quadratic dependence in the latter case disappears and there exist the only crossover scale @xmath86 . finally , let us consider the situation when the nodal points lie at the points of maximum of fs curvature . to illustrate this case we write the normal - state spectrum in the tight - binding form and choose the gap function as @xmath1-wave with @xmath88-symmetry , @xcite @xmath89 ( see fig . [ 2]b ) .
then the straightforward calculations at @xmath90 yield the formulas similar to those obtained in previous subsection : @xmath91 when @xmath92 we have @xmath93.\ ] ] again , we obtain the crossover from the cubic regime to the quadratic one .
we see that close to the half - filling ( at @xmath94 ) the energy scale of this crossover ( @xmath95 ) is much less than that previously considered .
it is connected with large curvature of fs near the nodal points so that the linear approximation for the spectrum breaks down much earlier .
note that this narrowing of the @xmath0 region is accompanied by the strong parametric enhancement of im@xmath22 by factor @xmath96 . at higher frequencies @xmath97 we get @xmath98 \label{14}\ ] ] comparing this formula with formula found in previous subsection we see that the additional logarithmic factor exists stemming from closeness of saddle - point to the nodal one .
note also that under @xmath99 we deal with poor - defined excitations since as it follows from our calculations @xmath100 for all @xmath76 in this case . from eq .
( 14 ) and kramers - kronig relations one can find that the quasiparticles are well - defined if @xmath101 ^ 2 \leq 1\ ] ] which is nothing but the criterion of absence of spin density wave .
we note here that the same criterion arises in the case considered in subsect .
b , as well .
the additional logarithmic enhancement stems in that case from the self - energy behavior near the `` corners '' of fs ( cf .
all the results presented in previous section may be rewritten in an universal form . to do that more compactly
let us introduce the following variables @xmath102 and @xmath103 , where @xmath104 is the curvature of fs at the nodal point .
we define also @xmath105 as the angle between @xmath106 and @xmath107
. then we may write down the following formula : @xmath108 , & \min \ , ( a , b ) \ll \omega\ll \max \left(b,{\displaystyle\frac{b^2}{a } } \right ) ; \\
\omega b^2\times\left ( 1\ \mbox{or } \
\ln { \displaystyle \frac{a\omega}{b^2 } } \right)\ , , & { \displaystyle\frac{b^2}{a}\ll\omega } , \end{array } \right.\ ] ] this result is valid until @xmath109 and @xmath110 , as it takes place in all principal cases .
the additional extrafactor @xmath111 at high frequencies @xmath112 appears only when the nodal point lies near the saddle one .
it is easy to see that all previous results are reproduced by the following substitutions ( i ) @xmath113 , @xmath114 , @xmath115 ( assuming @xmath116 ) for circular fs , ( ii ) @xmath117 , @xmath114 , @xmath118 and ( iii ) @xmath119 , @xmath120 , @xmath121 for fs of varying curvature .
we mention also that the quadratic regime disappears completely when @xmath122 and that the intermediate crossover scales differ principally as @xmath123 and @xmath124 .
note that we have evaluated the damping using spectrum expansion near specific points up to the second order terms in deviations from this points .
this is obviously valid for fermi surfaces smooth enough .
we undertake our consideration for circular and simplest tight - binding fss .
meanwhile , in the adequate theory of high - temperature superconductivity the next - nearest - neighbor hopping terms in the tight - binding scheme should be taken into account .
@xcite it is clear from our analysis how the picture will be modified in this case .
say , for @xmath44-wave pairing the range of cubic regime will be shrunken and a set of intermediate crossovers may appear in the range of the quadratic one .
however , they will differ only by the logarithmic factor .
next , for the special choice of the parameters @xmath125 and @xmath126 the saddle - points near @xmath127 and @xmath128 appear to be the `` extended '' ones ( quasi-1d ) .
@xcite here @xmath129 is the next - to - nearest neighbor hopping integral and @xmath68 controls the filling .
this case should be treated separately .
the above consideration was made for zero temperature .
it can be trivially generalized onto low finite temperatures ( less than @xmath76 and minimal frequency scale in ( 16 ) ) . in the opposite limiting case ( but not very close to @xmath130 ) eqs .
8 - 14 can be rewritten by substitution @xmath131 for @xmath76 . in particular
, it leads to the cubic @xmath131-dependence of damping for sufficiently small temperatures : @xmath132 ( the temperature dependence of the gap function yields only the corrections for this law ) . in order to apply our results to the normal state one should put simply @xmath133 . note
that divergence in the low - temperature asymptotics never exist , since always @xmath134 .
if the crossover scales @xmath135 or @xmath136 are found below @xmath130 , @xmath137 or @xmath131 regimes should appear in the superconducting phase .
when @xmath138 , the crossover from the conventional 2d fl behavior to the linear one occurs in the normal state .
our treatment has demonstrated , in particular , that the `` purely nodal '' regime for damping takes place in the quite narrow frequency ( temperature ) range .
it is restricted from above by the much less scale than the gap value as one could expect .
it seems to be not a feature of damping only , so that the same characteristic scale should appear for other quantities such as density of states , nuclear relaxation rate , penetration depth , inelastic scattered neutron intensity near incommensurate and internodal vectors , @xmath139 .
since , on the other hand , the impurities scattering restricts this range from below , one can estimate it as follows : @xmath140 ^ 2,\nonumber \\
\varepsilon_{n}&=&\left(\frac{q_{f}^{2}}{2m}\right)_{{\bf q}={\bf q_{n } } } , \end{aligned}\ ] ] where @xmath141 is the damping due to impurities .
the estimates show that the range defined by eq .
( 17 ) appear to be quite narrow for the high-@xmath142 compounds .
it may even disappear if @xmath141 will be large enough ( @xmath14310@xmath144 ) . in conclusion , we consider the damping of excitations in the 2d superconductor with nodal points in the spectrum .
we show that at @xmath27 this damping demonstrates gapless cubic - in-@xmath76 low - frequency behavior at the nodal points .
the results obtained for high frequencies depend on details of fermi surface parametrization .
we have analyzed two important particular fs parametrizations such as the models of almost free and tight - binding electrons .
we find the frequency variation of damping and the crossover scales for these cases . for smooth enough
fs they may be expressed through the model - free characteristics of the spectrum at the nodal points .
we argue that the excitations are well - defined in the wide frequency range .
we extend our treatment to finite temperatures and study the possibility of crossovers in temperature dependence of damping below the superconducting transition . here
we study eq .
( 5 ) for circular fs : @xmath145 .
we shall assume that @xmath146 is finite in the important areas of the integrand in eq .
also , let us take for a moment the combination of @xmath147 coefficients in eq .
( 5 ) to be of order of unity .
then we should analyze the following expression @xmath148 where d is defined by the condition @xmath149 it is convenient to make the substitution @xmath150 , @xmath151 in eq .
( a1 ) , then @xmath152 , @xmath153 , where @xmath154 and @xmath155 are the corresponding polar angles .
taking @xmath156 , we may represent the normal - state dispersion as @xmath157 .
besides , we expand the gap functions up to the linear in angle terms @xmath158 , @xmath159 . using such parametrization and leaving the principal terms only , we reduce eq .
( a1 ) to the following one @xmath160 where the limits of integration stem from inequality ( a2 ) .
this condition means , in particular , that at small @xmath76 momenta integrations in eq .
( a3 ) are restricted by the elliptic areas whose centers lie at the nodal points .
it is convenient to pass therefore to the elliptic coordinates in eq .
( a3 ) : @xmath161 then it transforms to @xmath162 this expression allows us to analyze all principal limiting cases .
+ ( 1 ) @xmath163 . in this case
we can neglect the second term in @xmath56-function .
the integral remained can be easily evaluated and the result is given by eq .
( 9 ) + ( 2 ) @xmath164 .
here it is convenient to rewrite @xmath56-function in eq .
( a5 ) as follows @xmath165 one can see that the integration over one of the angles gives the logarithmic divergence which is cut on the scale @xmath166 .
+ ( 3 ) @xmath167
. then we should integrate the @xmath56-function in eq .
( a5 ) with respect to the angles between the limits @xmath168 and @xmath169 .
after the trivial substitution @xmath170 @xmath171 , we have @xmath172 again , we have the logarithmic divergence , which is cut now on the scale @xmath173 .
the last two cases we may be combined as eq . (
10 ) .
see , for example : j. p. lu , phys .
rev.lett . * 67 * , 125 ( 1991 ) ; n. bulut and d. j.scalapino , _ ibid_. * 68 * , 706 ( 1992 ) ; d. s.rokhsar , _ ibid_. * 70 * , 493 ( 1993 ) ; c .- r .
hu , _ ibid_.
* 72 * , 1526 ( 1994 ) ; j. p. lu , j. mod.phys . lett .
b * 6 * , 547 ( 1992 ) ; p. monthoux _ et al_. , phys . rev .
b * 47 * , 6069 ( 1993 ) ; d. telen , d.pines , and j. p. lu , _
ibid_. 9151 ( 1993 ) ; t. p.devereaux and p. fulde , _
ibid_. 14638 ( 1993 ) ; p. j.hirshfeld and n. goldenfeld , _ ibid_. * 48 * , 4219 ( 1993 ) ; h. won and k. maki , _ ibid_. * 49 * , 15305 ( 1994 ) ; s. v. maleyev , a. g. yashenkin , and d. n.aristov , _ ibid_. * 50 * , 13825 ( 1994 ) ; a. j. leggett , physica b * 199 & 200 * , 291 ( 1994 ) ; p. monthoux and d. j. scalapino , _
ibid_. 294 ( 1994 ) ; d. z. liu , k. levin , and j. maly , preprint ; a. m. tikofsky and d. b. bailey , preprint .
j. m. valles _ et al_. , phys .
b * 44 * , 11986 ( 1991 ) ; d. a. bonn _
et al_. , _ ibid_. * 47 * , 11314 ( 1993 ) ; d. a. brawner and h. r. ott , _ ibid_. * 50 * , 6530 ( 1994 ) ; w. n. hardy _ et al_. , phys .
lett . * 70 * , 3999 ( 1993 ) ; d. a. wollman _
et al_. , _ ibid_. * 71 * , 2134 ( 1993 ) ; p.
chaudhary and s .- y .
lin , _ ibid_. * 72 * , 1084 ( 1994 ) ; c. c. tsuei _ et al_. , _ ibid_. * 73 * , 593 ( 1994 ) ; k. moler _ et al_. , _
ibid_. 2744 ( 1994 ) .
p. a. lee , phys .
* 71 * , 1887 ( 1993 ) ; a. a. nersesyan , a. m. tsvelik , and f. wenger , _ ibid_. * 72 * , 2628 ( 1994 ) ; a. v. balatsky , a.rosengren , and b. l. altshuler , _ ibid_. * 73 * , 720 ( 1994 )
. g. baskaran , z. zou and p. w. anderson , solid state commun .
* 63 * , 973 ( 1987 ) ; l. b. ioffe and a. i. larkin , phys .
b * 39 * , 8988 ( 1989 ) .
g. kotliar , phys .
b * 37 * , 3664 ( 1988 ) .
n. e. bickers , d. j. scalapino , and r. t. scalettar , int .
j. mod .
b * 1 * , 687 ( 1987 ) ; p. a. lee and n. read , phys .
lett . 2691 ( 1987 ) ; p. monthoux , a. v. balatsky , and d. pines , _
ibid_. * 67 * , 3448 ( 1991 ) .
a. v. chaplik , zh . eksp .
, 1845 ( 1971 ) [ sov .
jetp * 33 * , 997 ( 1971 ) ] ; c. hodges , h. smith , and j. wilkins , phys.rev .
b * 4 * , 302 ( 1971 ) .
d. n. aristov , s. v. maleyev , and a. g. yashenkin , phys .
b * 48 * , 3527 ( 1993 ) .
see , for example : f. wenger and s.stlund , phys .
b * 47 * , 5977 ( 1993 ) .
et al_. , phys .
b * 47 * , 9055 ( 1993 ) .
see , for example : a. a.abrikosov , j. c. campuzano and k. gofron , physica c * 214 * , 73 ( 1993 ) . | we calculate the damping of excitations due to four - fermionic interaction in the case of two - dimensional superconductor with nodes in the spectrum . at zero temperature and low frequencies
it reveals gapless @xmath0 behavior at the nodal points . with the frequency increasing the crossover to the normal - state regimes appears . at high frequencies the damping strongly
depends on details of a normal - state spectrum parametrization .
two important particular cases such as the models of almost free and tight - binding electrons are studied explicitly and the characteristic scales are expressed through the model - free parameters of the spectrum at the nodal points .
the possibility of crossover in temperature dependence of damping in the superconducting phase is discussed . |
quantum chaotic scattering is an ubiquitous phenomenon .
it occurs , for instance , in nuclear physics @xcite , in electron transport through disordered mesoscopic samples @xcite , and in microwave billiards @xcite . in all cases ,
the cross section displays random fluctuations versus energy or frequency . these are due to the random features of the underlying resonances . with @xmath1 the average resonance spacing and @xmath2 the average width , data on cross - section fluctuations exist for the entire range of the parameter @xmath3 , from the regime of isolated resonances ( @xmath4 ) to that of strongly overlapping resonances ( @xmath5 ) .
the analysis of the data focuses on the value of the average cross section and on quantities that characterize the cross - section fluctuations .
these are the variance of the cross section and certain correlation functions . for the analysis
, one needs theoretical expressions for these quantities .
these should be generic and only use a minimum of adjustable parameters .
the generic theoretical treatment of chaotic scattering employs a combination of scattering theory and random - matrix theory @xcite and uses as input the values of @xmath1 and of the energy - averaged elements @xmath6 of the scattering matrix @xmath0 .
analytical results exist for the @xmath0-matrix autocorrelation and cross - correlation functions @xcite ( including the value of the average cross section ) and for the third and fourth moments of the @xmath0-matrix @xcite . because of the complexity of the problem
, analytical results for higher moments of the cross section or for cross - section correlation functions can not be expected in the foreseeable future .
the present paper aims at filling that gap .
we combine the available analytical information @xcite , results of computer simulations , and of experimental work on microwave billiards @xcite to study the cross - section autocorrelation function for all values of @xmath3 . in particular , we address the following questions .
( i ) for which values of @xmath3 and with which accuracy can the cross - section autocorrelation function be predicted in terms of the @xmath0-matrix autocorrelation function ?
( ii ) which analytical alternatives exist should that approach fail ?
we consider chaotic scattering in a time - reversal invariant system described by a unitary and symmetric scattering matrix @xmath7 . here
@xmath8 denote the channels and @xmath9 the energy ( or , in the case of microwave billiards , the frequency ) .
the number @xmath10 of channels may range from unity to a large number , @xmath11 .
chaotic scattering is modeled by writing the @xmath0-matrix in the form @xcite @xmath12 where @xmath13 the real and symmetric hamiltonian matrix @xmath14 has dimension @xmath15 and describes the dynamics of @xmath16 resonances labeled by greek letters .
these are coupled to the channels by the real matrix elements @xmath17 .
chaos is taken into account by choosing @xmath14 as a member of the gaussian orthogonal ensemble ( goe ) of random matrices @xcite .
thus , the elements of @xmath14 are gaussian random variables with zero mean values and second moments given by @xmath18 $ ] . here and in the sequel ,
the overbar denotes the average over the ensemble .
the parameter @xmath19 determines ( or is determined by ) the average level spacing @xmath1 of the @xmath16 resonances .
it is convenient to decompose @xmath20 into an average and a fluctuating part , @xmath21 where the average @xmath0-matrix @xmath6 is assumed to be diagonal .
the values of the diagonal elements @xmath22 serve as input parameters for the statistical model and are assumed to be known .
that is the typical case : in nuclei , @xmath22 is given in terms of the optical model of elastic scattering , in microwave billiards @xmath23 is determined by the running average over a measured spectrum @xcite . in rare cases , the average @xmath0-matrix may not be diagonal . by an orthogonal transformation in channel space ,
@xmath6 can be reduced to diagonal form , see refs .
for simplicity we do not address that case . by the same transformation , the phases of the @xmath0-matrix usually appearing as factors on the right hand side of eq .
( [ 1 ] ) , can be removed . both for the s - matrix model eq .
( [ 1a ] ) considered in the present work and the experimental data the average @xmath0-matrix is real and diagonal .
starting from eq .
( [ 1a ] ) , the @xmath0-matrix autocorrelation function ( or `` two - point function '' ) @xmath24 has been calculated analytically @xcite for @xmath15 and fixed @xmath10 .
the resulting expression depends only on the difference @xmath25 of the two energy arguments , on the average level spacing @xmath1 of the system , and on the transmission coefficients @xmath26 of all channels @xmath27 defined by @xmath28 the transmission coefficients obey @xmath29 .
these coefficients measure the unitarity deficit of the average @xmath0-matrix and give the probability with which the resonances take part in the reaction .
this is seen by using the decomposition eq .
( [ 1 ] ) and the definition eq . ( [ 3 ] ) to write the unitarity condition for @xmath0 in the form @xmath30 for @xmath31 or @xmath32 , we have @xmath33 for all @xmath34 , and the resonances are not reached from channel @xmath27 .
conversely , @xmath35 is maximal for @xmath36 or @xmath37 ( complete absorption of the incident flux in channel @xmath27 by resonance formation ) . the transmission coefficients @xmath26 determine the average width @xmath2 of the resonances .
an approximation for @xmath2 is the `` weisskopf estimate '' @xmath38 the case of strongly overlapping resonances @xmath5 ( `` ericson regime '' @xcite ) occurs for @xmath39 : the number @xmath10 of channels must be large and most of the individual transmission coefficients @xmath26 must not be small .
conversely , the case of nearly isolated resonances @xmath40 is realized when @xmath10 is of order unity or when @xmath10 is large but all @xmath26 are small .
the theory developed in ref .
@xcite and used in refs .
@xcite applies to all values of @xmath3 .
equation ( [ 3b ] ) is exact in the ericson regime and fairly reliable elsewhere . under omission of kinematical factors the cross section in nuclear physics , the conductance in electron transport and the transmitted power in microwave billiards
are all given by @xmath41 or by a sum of such terms .
for brevity we refer to @xmath41 as to the cross section . the average cross section @xmath42 is given in terms of @xmath22 and of @xmath43 and is , thus , known .
fluctuations of the cross section are measured in terms of the cross - section autocorrelation function @xmath44 that function is the object of central interest in the present paper . with the help of the decomposition eq .
( [ 1 ] ) we write @xmath45 we have used that in the experiments and in the considered @xmath0-matrix model ( eq . ( [ 1a ] ) )
@xmath46 is real , that by definition @xmath47 and that @xmath48 for all @xmath49 and all @xmath50 .
the last relation holds because all poles of @xmath0 lie in the lower half of the complex energy plane . to determine @xmath51 we need to know the four - point function @xmath52 and , in the elastic case @xmath53 ,
also the three - point function @xmath54 these functions are known analytically only for @xmath55 and in the ericson regime ( @xmath5 ) , see below . to determine magnitude and @xmath25-dependence of @xmath56 , we combine analytical results with numerical and experimental evidence as follows .
( i ) analytical results : in refs .
@xcite analytic expressions are given for two functions @xmath57 and @xmath58 that look similar to but actually differ from @xmath59 and @xmath60 , respectively .
these are defined by @xmath61 ^ 2 \left[s^{\rm fl}_{a
b}(e - { \varepsilon}/2)\right]^2 } \ , \nonumber\\ f^{(3)}_{a b}({\varepsilon})&=\overline{s^{\rm fl * } _ { a b}(e+{\varepsilon}/2 ) \left[s^{\rm fl } _ { a b}(e - { \varepsilon}/2)\right]^2 } \ , .
\label{feps}\end{aligned}\ ] ] we note that in @xmath59 and in @xmath60 the elements @xmath62 and @xmath63 carry pairwise the same energy arguments .
this is not the case for @xmath57 and @xmath58 .
however , @xmath64 coincides with @xmath65 , and @xmath66 differs from @xmath67 only by the known term @xmath68 .
we are going to show that for @xmath69 it is possible to approximate @xmath59 in terms of @xmath57 , and under certain conditions @xmath60 in terms of @xmath58 .
for the convenience of the reader we , therefore , give in the appendix analytic expressions for @xmath70 for @xmath71 , where @xmath72 .
we briefly show how the threefold integrals can be evaluated numerically to avoid the apparent singularities of the integrand .
( ii ) numerical results : for the numerical simulations we use eqs .
( [ 1a ] ) and ( [ 1b ] ) and fixed values for @xmath19 , for the transmission coefficients @xmath26 , and for @xmath16 as initial values .
calculations were typically done for several 100 realizations to minimize statistical errors .
the results agree very well with the available analytical results but go beyond them .
( iii ) data : the data stem from measurements of transmission and reflection amplitudes of microwaves in a flat cylindrical resonator made of copper and coupled to two antennas , see refs .
microwave power was coupled into the resonator with the help of a vector network analyzer .
the range of the excitation frequency was chosen such that only one vertical electric field mode is excited .
then the microwave cavity simulates a two - dimensional quantum billiard @xcite .
the resonator had the shape of a tilted stadium billiard whose classical dynamics is chaotic .
transmission and reflection amplitudes correspond to complex @xmath0-matrix elements that are theoretically modeled by eqs .
( [ 1a ] ) and ( [ 1b ] ) @xcite .
the inelastic case is simpler than the elastic one because it involves only the function @xmath59 , see eq .
( [ 5 ] ) .
we begin with the ericson regime @xmath5 , see refs . @xcite . in ref . @xcite it was suggested and in ref .
@xcite it was shown that for @xmath5 the fluctuating @xmath0-matrix elements @xmath62 possess a bivariate gaussian distribution centered at zero .
that fact implies that all higher moments and correlation functions can be computed from @xmath74 by way of wick contraction .
hence @xmath75 and @xmath76 in the ericson regime , the two - point function has the value @xcite @xmath77 thus , for @xmath73 and @xmath5 , the cross - section autocorrelation function is known analytically .
it has the shape of a lorentzian with width @xmath2 as given by eq .
( [ 3b ] ) . how far can we use eq .
( [ 9 ] ) outside the ericson regime , i.e. , for smaller values of @xmath3 ?
figure [ fig1 ] shows the ratio @xmath78 versus @xmath3 as a function of @xmath3 on a semi
logarithmic plot for three cases as indicated in the figure caption . case ( i ) with unequal transmission coefficients is obtained from experimental data and suggests that eq .
( [ 9 ] ) holds approximately for @xmath79 .
the two other cases result from numerical simulations with , respectively , @xmath80 and @xmath81 equal transmission coefficients and imply , that eq .
( [ 9 ] ) holds for @xmath82 .
the ratio @xmath83 increases dramatically with decreasing @xmath84 so that eq .
( [ 9 ] ) can not be used much below @xmath85 . on a semi - logarithmic plot for three cases .
upper panel : experimental data @xcite ( filled circles ) .
lower panel : data generated numerically from eq .
( [ 1a ] ) for @xmath86 channels ( open circles ) and for @xmath87 channels ( full squares ) with identical transmission coefficients.,width=321 ] the origin of the failure of eq .
( [ 9 ] ) for small values of @xmath84 is easily understood . qualitatively speaking ,
it is intuitively clear that cross - section fluctuations ( measured in units of the average cross section ) are much larger for isolated than for overlapping resonances .
quantitatively , the assumption that underlies eq .
( [ 9 ] ) is that the distribution of the fluctuating @xmath0-matrix elements @xmath88 is gaussian .
that assumption holds only if the width @xmath89 of the distribution is sufficiently small .
indeed , the unitarity condition eq .
( [ 3a ] ) implies that for all @xmath27 and @xmath34 we must have @xmath90 , and @xmath89 must be so small that the contribution of the tails of the distribution that extend beyond the values @xmath91 , is negligible . otherwise , significant deviations from a gaussian distribution are to be expected .
to estimate @xmath89 in the ericson regime we use eq .
( [ 8 ] ) which for @xmath73 gives @xmath92^{1/2}$ ] .
thus , since @xmath93 this estimate yields @xmath94 as expected . for smaller values of @xmath3
, constraints on the distribution of @xmath62 due to the unitarity condition ( [ 3a ] ) are expected to become increasingly important as the number @xmath10 of terms in the sum in eq .
( [ 3a ] ) decreases . we have checked this explanation by investigating the distribution of @xmath95 .
a bivariate gaussian distribution for @xmath96 implies that the distribution of @xmath97 has the form @xmath98 $ ] , and that the phase of @xmath95 is distributed uniformly in the interval @xmath99 . in fig .
[ fig2a ] and fig .
[ fig2b ] and for values of @xmath3 indicated above each panel , we compare in the upper row the function @xmath100 with experimental and numerical data , respectively . in the panels in the lower row , we show the corresponding distributions of the phase . the numerical data in the three panels of fig .
[ fig2a ] are obtained by simulating absorption in the resonator in terms of a large number of fictitious channels with small transmission coefficients in each channel .
then their sum @xmath101 is the only parameter .
it was determined as described in @xcite from a fit of the experimental autocorrelation function to the analytic result given in @xcite .
the data show that agreement with the bivariate gaussian distribution is attained for @xmath102 . the data in fig .
[ fig2b ] ( generated numerically for 32 equal channels ) show agreement with the bivariate gaussian distribution only for @xmath103 .
this suggests that the limit of a bivariate gaussian distribution for @xmath104 with @xmath105 is attained for larger values of @xmath106 when all transmission coefficients are equal than when the transmission coefficients differ .
these results account for the deviations from eq .
( [ 9 ] ) displayed in fig .
[ fig1 ] . obtained from the experimental data ( histograms ) and from numerical simulations ( solid lines ) for three values of @xmath3 . in the upper row , the bivariate gaussian distribution expected in the ericson limit is shown as a dashed line .
the transmission coefficients have the following values @xcite . for @xmath107 :
@xmath108 , for @xmath109 : @xmath110 , for @xmath111 : @xmath112.,width=321 ] obtained from numerical simulations ( histograms ) for @xmath113 channels with identical transmission coefficients for three values of @xmath3 . in the upper row , the bivariate gaussian distribution expected in the ericson limit is shown as a dashed line.,width=321 ] concerning the @xmath25-dependence of @xmath114 and of @xmath115 , our data show that for those values of @xmath116 where the ratio @xmath117 , that dependence is sufficiently similar so that @xmath114 can be reliably approximated by @xmath115 .
( by that we mean that the full two point function defined in eq .
( [ 2 ] ) and not the approximate form eq .
( [ 8 ] ) has to be used ) . that leaves us with the question how to approximate @xmath114 analytically for @xmath118 .
an obvious possibility is offered by the function @xmath119 defined in the first of eqs .
( [ feps ] ) .
figure [ fig3 ] shows experimental ( upper two panels ) and numerically ( lowest panel ) generated values of @xmath120 and of @xmath114 versus @xmath25 for two different values of @xmath84 , and for @xmath113 channels with identical transmission coefficients @xmath121 and a value of @xmath3 close to the largest achieved in the experiment , respectively .
the function @xmath122 was rescaled so that at @xmath123 it agrees with @xmath124 . in the upper two panels
the experimental curves for @xmath125 ( filled circles ) and the rescaled function @xmath122 are shown together with the analytic result for @xmath122 multiplied with the same scaling factor as the experimental one . in the lowest panel
we compare the numerical result for @xmath125 with the analytic rescaled result for @xmath120 .
we note that the agreement is excellent for all three cases .
similarly good agreement was found also for @xmath126 channels and several values of @xmath3 .
we conclude that we have reached a simple and reliable prescription for approximating @xmath114 analytically for all values of @xmath3 : calculate @xmath124 and @xmath120 analytically .
( the formulas needed @xcite are given in the appendix ) .
rescale @xmath120 so that its value at @xmath123 agrees with @xmath127 .
the resulting rescaled function @xmath122 is an excellent approximation to @xmath114 for all values of @xmath3 . of @xmath114 and of the renormalized function @xmath120 for three cases with @xmath3 as indicated in the panels . in the upper two panels we show experimental curves for @xmath114 ( filled circles ) and for @xmath120 ( open squares ) , in the lowest panel a numerically generated curve for @xmath114 ( solid line ) for 32 identical channels . in all three panels ,
the analytic function @xmath120 is shown as dashed line .
both the experimental and the analytic values for @xmath120 are renormalized with a factor determined from the analytic value for @xmath127.,width=321 ]
the elastic case is analogous to the inelastic one only if @xmath37 .
then all conclusions drawn in section [ inel ] apply .
in general , that is not the case . in particular , for @xmath4 ( isolated resonances ) the last term in eq .
( [ 1a ] ) tends to zero and @xmath22 approaches unity .
the unitarity constraint on @xmath0 implies @xmath128 and the distribution of @xmath129 must then become skewed .
therefore , we have to expect that eq . ( [ 9 ] ) applies less generally in the elastic than in the inelastic case , and that @xmath130 in eq .
( [ 5 ] ) plays an important role . actually , it was pointed out in refs .
@xcite that even in the ericson limit , a profound difference between the elastic and the inelastic cases exists : for @xmath5 , @xmath131 and @xmath132 have similar values unless @xmath133 or @xmath134 .
that shows that in order to predict @xmath135 we need to know both @xmath130 and @xmath136 for all values of @xmath116 .
concerning @xmath137 , we proceed as in section [ inel ] and display in fig .
[ fig4 ] the ratio @xmath138 in the upper panel for experimental data , in the lower one for a numerical simulation with @xmath139 equal transmission coefficients , as indicated in the figure caption .
we reach the same conclusions as for @xmath73 : when the transmission coefficients differ , the ratio is close to unity for @xmath79 , when they are equal for @xmath82 .
but for @xmath140.,width=321 ] as in the inelastic case , the reason for the failure of eq .
( [ 9 ] ) is the non gaussian distribution of @xmath141 .
this is shown in figs .
[ fig5a ] , [ fig5b ] .
the deviations from a bivariate gaussian distribution now occur for larger values of @xmath3 than in the inelastic case , in keeping with the expectations formulated at the beginning of this section .
but for @xmath140.,width=321 ] but for @xmath140.,width=321 ] in fig .
[ fig6 ] we show that , as in the elastic case , the rescaled function @xmath142 agrees well with @xmath143 for all values of @xmath3 .
such agreement was likewise found for the case of @xmath126 channels with identical transmission coefficients .
figure [ fig7 ] shows that the three - point function @xmath144 is approximated quite well by the ( unrenormalized ) function @xmath145 defined in the second of eqs .
( [ feps ] ) for the experimental data with @xmath107 ( upper panel ) , where the resonances are nearly isolated , and for the numerical simulations with @xmath86 identical transmission coefficients @xmath121 ( lower panel ) .
however , considerable deviations are observed for the experimental data with @xmath146 ( middle panel ) .
thus , the conclusions drawn for the inelastic case apply similarly to the elastic one only in the regime of isolated resonances and the ericson regime , where @xmath46 is vanishingly small . to obtain an analytic expression for the cross - section autocorrelation function we now need to replace @xmath143 with the rescaled function @xmath147 and the function @xmath148 with @xmath149 .
while the former replacement yields an excellent approximation for all values of @xmath106 , this is not generally true for the latter in the intermediate regime of weakly overlapping resonances .
still , for the numerical simulations with @xmath86 equal transmission coefficients @xmath121 , which corresponds to @xmath150 , the resulting analytic expression for the cross - section autocorrelation function yields a good approximation , because there both , the values of @xmath149 and of @xmath144 , are negligibly small as compared to @xmath56 , whereas this is not true for the experimental data with @xmath111 .
but for the elastic case.,width=321 ] of @xmath151 and of @xmath144 for three cases with @xmath3 as indicated in the panels . in the upper two panels we show experimental curves for @xmath144 ( filled circles ) and for @xmath149 ( open squares ) , in the lowest panel a numerically generated curve @xmath144 for 32 identical channels ( solid line )
. the analytic function @xmath149 is shown as dashed line in all three panels .
the limited accuracy of our comparison is indicated by the difference between the dashed line and the open squares in the upper two panels.,width=321 ]
we return to the questions raised in the introduction . we have shown that eq .
( [ 9 ] ) holds essentially only in the ericson regime and , for the elastic case , even there only approximately .
this is because of the constraint imposed on the distribution of @xmath141 by unitarity .
hence , in the inelastic case the cross section autocorrelation function can be reliably predicted from eq .
( [ 9 ] ) for @xmath152 for inequivalent channels and for @xmath103 for identical channels .
even more stringent constraints exist in the elastic case .
we have demonstrated , however , that for the inelastic case for all values of @xmath153 and for the elastic case under certain conditions an excellent approximation to the cross - section autocorrelation function may be obtained by replacing in eq .
( [ 5 ] ) the three - point function @xmath154 by @xmath155 and the four - point function @xmath156 by the rescaled function @xmath157 . with these replacements ,
( [ 5 ] ) takes the form @xmath158 the input parameters for the evaluation of eq .
( [ 10 ] ) are the average @xmath0matrix elements and the associated transmission coefficients for all channels and , for the dependence on @xmath25 , the average level spacing @xmath1 of the scattering system . in terms of these parameters , the two point function @xmath74
is given in ref .
@xcite , and @xmath159 and the functions @xmath57 and @xmath58 in refs . @xcite . for the convenience of the reader ,
all relevant formulas are collected in the appendix in a form suitable for numerical implementation . for the elastic case , eq .
( [ 10 ] ) provides an excellent approximation for the cross - section autocorrelation function for all values of @xmath106 only for @xmath55 . for @xmath160 eq .
( [ 10 ] ) yields a good approximation in the regime of isolated and strongly overlapping resonances , whereas in the intermediate regime of weakly overlapping resonances this is true , if the contribution of @xmath60 or equivalently of @xmath161 to @xmath56 is negligible .
we thank t. friedrich for a discussion at an early stage of this work . the research has been supported through the sfb 634 by the dfg .
for the evaluation of @xmath162 in terms of @xmath141 and @xmath6 as given in eq .
( [ 5 ] ) we need to compute @xmath163 , @xmath164 and @xmath165 .
the autocorrelation coefficient @xmath166 , @xmath167 and @xmath168 are given in terms of a threefold integral , c.f .
@xcite , @xmath169 where the integration measure is given as @xmath170 for @xmath171 , i.e. for @xmath172 the factor @xmath173 equals @xmath174 for @xmath175 , i.e. for @xmath167 it equals @xmath176 ^ 2 \ } \right\ } \bigg ) \delta_{ab } \nonumber\end{aligned}\ ] ] and for @xmath177 , i.e. for @xmath168 we have @xmath178 ^ 2\right\ } \nonumber \\ & \quad \times \left\{{\rm trg}(\nu_b ) + \frac{1}{2}r_b\left[{\rm trg}(\mu_b)\right]^2\right\ } \nonumber \\ & \quad + \delta_{ab } \bigg ( \left[{\rm trg}(\nu_a ) \right]^2 + 4{\rm trg}\left(\nu_a^2\right ) \nonumber \\ & \quad + r_a^2{\rm trg } \left(\mu_a^2\right ) \left\ { { \rm trg } \left(\mu_a^2\right ) + \left[{\rm trg}(\mu_a ) \right]^2\right\ } \nonumber\\ & \quad + r_a \big\ { \left [ { \rm trg}(\mu_a ) \right]^2 { \rm trg}(\nu_a ) \nonumber \\ & \quad + 8\,{\rm trg}(\mu_a ) { \rm trg}(\mu_a\nu_a ) + 8\,{\rm trg } \left ( \mu_a^2\nu_a \right ) \big\ } \bigg)\ , .
\nonumber\end{aligned}\ ] ] here , @xmath179 and @xmath180 , @xmath181 are the matrices @xmath182 where @xmath183 is the @xmath184 diagonal matrix with entries @xmath185 , @xmath186 , @xmath187 and @xmath187 . for a @xmath184 matrix @xmath188 with diagonal elements @xmath189 the graded trace trg is defined as @xmath190 .
analytic expressions for the functions @xmath191 and @xmath192 defined in eqs .
( [ feps ] ) are obtained by multiplying @xmath193 in eq .
( [ s2 ] ) and @xmath194 in eq .
( [ s3 ] ) and @xmath195 in eq .
( [ s4 ] ) with the exponential @xmath196 .
\label{exp}\ ] ] the integral eq .
( [ f ] ) contains several singularities .
these can be handled by proceeding as in section 5 of ref .
we define the variable @xmath197 then the exponential in eq .
( [ exp ] ) turns into @xmath198 we distinguish the two cases @xmath199 and @xmath200 .
for @xmath199 we define two further integration variables @xmath201 for the inverse transformation that yields @xmath202 the jacobian for this transformation is equal to @xmath203 .
the threefold integral in eq .
( [ f ] ) becomes @xmath204 \left(1 - 2s^2\right)\left(1-t^2\right ) } { \sqrt{\left(1+s^2t^2 - 2s^2\right ) \left[1+p\left(1+s^2t^2 - 2s^2\right)\right ] } } \nonumber \\ & \quad \times \ , \prod_c\frac{1-t_c\ , \lambda}{\sqrt{(1+t_c\lambda_1 ) ( 1+t_c\,\lambda_2 ) } } \ , \mathcal{f}^{(n)}_{ab}(\lambda,\lambda_1,\lambda_2 ) \ , .
\nonumber\end{aligned}\ ] ] for @xmath200 we define additional integration variables @xmath205 for the inverse transformation that yields @xmath206 the jacobian equals @xmath207 and the threefold integral in eq .
( [ f ] ) becomes @xmath208 \left(p-2s^2\right)\left(1-t^2\right ) } { \sqrt{\left(p+s^2t^2 - 2s^2\right)\left(1+s^2t^2\right ) \left(1+p+s^2t^2 - 2s^2\right ) } } \nonumber \\ & \quad \times \ , \prod_c\frac{1-t_c\ , \lambda } { \sqrt{(1+t_c\lambda_1)(1+t_c\,\lambda_2 ) } } \ , \mathcal{f}^{(n)}_{ab}(\lambda,\lambda_1,\lambda_2 ) \nonumber \label{f2}\end{aligned}\ ] ] where the @xmath209-function is defined by @xmath210 | for the theoretical prediction of cross section fluctuations in chaotic scattering , the cross - section autocorrelation function is needed .
that function is not known analytically . using experimental data and numerical simulations ,
we show that an analytical approximation to the cross - section autocorrelation function can be obtained with the help of expressions first derived by davis and boos .
given the values of the average @xmath0matrix elements and the mean level density of the scattering system , one can then reliably predict cross - section fluctuations . |
for @xmath50 , define the functions @xmath51
@xmath52 @xmath53 ( note that @xmath54 depend implicitly on the length @xmath2 of the strings . )
then we have the equations @xmath55
@xmath56 @xmath57 these equations will hold even for @xmath58 if one defines trivially @xmath59 for @xmath60 .
alternatives 1,2 in the parentheses signify whether the minimum is attained by its first or second arguments , respectively .
the proof consists simply of inspecting the definitions of the functions @xmath54 and verifying that in each of the cases the required relations hold .
we omit the details , since the reader would no doubt have to go through the same thought process to verify them on her own as to check that a purported proof is correct .
depth0pt a schematic representation of the finite state machine is shown in figures 2 , 3 and 4 below .
we present several variants of the machine : the machine in figure 2 only decides between alternative 1 and alternative 2 .
the machine in figure 3 , which has auxiliary counters for the distance and for the variable @xmath2 ( so strictly speaking it is not really a finite - state automaton ) , actually computes @xmath24 .
note that this is still designed only for the case in which @xmath13 begins with the symbol @xmath33 and @xmath25 begins with @xmath32 .
figure 4 shows the complete finite - state machine which may be used to compute @xmath24 for any two states @xmath23 .
this includes an initial component that discards the first few symbols which are identical for @xmath13 and @xmath25 , and another component that permutes the symbols @xmath11 to fit the design of the basic machine in figure 2 .
( 300,230)(0,0 ) ( 30,100 ) ( 150,100 ) ( 270,100 ) ( 75,190)(30,30 ) ( 195,190)(30,30 ) ( 50,90)(1,0)80 ( 130,110)(-1,0)80 ( 170,90)(1,0)80 ( 250,110)(-1,0)80 ( 20,82)(0,55)(25,45 ) ( 40,82)(60,55)(35,45 ) ( 25,45)(30,43)(35,45 ) ( 45,77)(-1,1)7 ( 140,82)(120,55)(145,45 ) ( 160,82)(180,55)(155,45 ) ( 145,45)(150,43)(155,45 ) ( 165,77)(-1,1)7 ( 260,82)(240,55)(265,45 ) ( 280,82)(300,55)(275,45 ) ( 265,45)(270,43)(275,45 ) ( 285,77)(-1,1)7 ( 35,120)(2,3)43 ( 145,120)(-2,3)43 ( 155,120)(2,3)43 ( 265,120)(-2,3)43 ( 55,173 ) ( 47,163 ) ( 39,153 ) ( 33,143 ) ( 28,133 ) ( 22,123 ) ( 232,173 ) ( 240,163 ) ( 248,153 ) ( 253,143 ) ( 260,133 ) ( 266,123 ) ( 85,80 ) ( 70,112 ) ( 195,80 ) ( 205,112 ) ( 0,40 ) ( 25,32 ) ( 240,40 ) ( 265,32 ) ( 114,67 ) ( 114,57 ) ( 114,47 ) ( 104,146 ) ( 179,146 ) ( 19,103 ) ( 18,94 ) ( 141,94 ) ( 258,94 ) ( 81,208 ) ( 80,200 ) ( 201,208 ) ( 200,200 ) figure 2 : the finite state machine : deciding between alternative 1 and alternative 2 .
the two letters signify the two inputs from @xmath13 and @xmath25 , reading at each step the next - most - significant symbol .
the parentheses in the non - terminal states indicate that if the input terminates without a decision , then in the start state alternative 1 wins , in the rightmost state alternative 2 wins , and in the middle state there is a draw , meaning that the shortest path is not unique and both alternatives are valid .
( 300,200)(0,0 ) ( 30,100 ) ( 150,100 ) ( 270,100 ) ( 75,190)(30,30 ) ( 195,190)(30,30 ) ( 50,90)(1,0)80 ( 130,110)(-1,0)80 ( 170,90)(1,0)80 ( 250,110)(-1,0)80 ( 20,82)(0,55)(25,45 ) ( 40,82)(60,55)(35,45 ) ( 25,45)(30,43)(35,45 ) ( 45,77)(-1,1)7 ( 137,85)(114,75)(105,60 ) ( 140,82)(130,59)(115,49 ) ( 105,60)(97,40)(115,49 ) ( 137,77)(2,3)5 ( 163,85)(186,75)(195,60 ) ( 160,82)(170,59)(185,49 ) ( 195,60)(203,40)(185,49 ) ( 164,77)(-2,3)5 ( 260,82)(240,55)(265,45 ) ( 280,82)(300,55)(275,45 ) ( 265,45)(270,43)(275,45 ) ( 285,77)(-1,1)7 ( 35,120)(2,3)43 ( 145,120)(-2,3)43 ( 155,120)(2,3)43 ( 265,120)(-2,3)43 ( 55,173 ) ( 47,163 ) ( 39,153 ) ( 33,143 ) ( 28,133 ) ( 22,123 ) ( 232,173 ) ( 240,163 ) ( 248,153 ) ( 253,143 ) ( 260,133 ) ( 266,123 ) ( 85,80 ) ( 70,112 ) ( 195,80 ) ( 205,112 ) ( 0,40 ) ( 25,32 ) ( 240,40 ) ( 265,32 ) ( 85,58 ) ( 85,48 ) ( 156,53 ) ( 104,146 ) ( 179,146 ) ( 19,105 ) ( 21,98 ) ( 18,91 ) ( 141,94 ) ( 258,94 ) ( 81,208 ) ( 80,200 ) ( 201,208 ) ( 200,200 ) ( 26,47 ) ( 106,37 ) ( 184,37 ) ( 93,91 ) ( 74,100 ) ( 213,91 ) ( 194,100 ) ( 266,47 ) ( 67,163 ) ( 187,163 ) ( 300,340)(0,0 ) ( 30,100 ) ( 150,100 ) ( 270,100 ) ( 75,190)(30,30 ) ( 195,190)(30,30 ) ( 50,90)(1,0)80 ( 130,110)(-1,0)80 ( 170,90)(1,0)80 ( 250,110)(-1,0)80 ( 20,82)(0,55)(25,45 ) ( 40,82)(60,55)(35,45 ) ( 25,45)(30,43)(35,45 ) ( 45,77)(-1,1)7 ( 137,85)(114,75)(105,60 ) ( 140,82)(130,59)(115,49 ) ( 105,60)(97,40)(115,49 ) ( 137,77)(2,3)5 ( 163,85)(186,75)(195,60 ) ( 160,82)(170,59)(185,49 ) ( 195,60)(203,40)(185,49 ) ( 164,77)(-2,3)5 ( 260,82)(240,55)(265,45 ) ( 280,82)(300,55)(275,45 ) ( 265,45)(270,43)(275,45 ) ( 285,77)(-1,1)7 ( 35,120)(2,3)43 ( 145,120)(-2,3)43 ( 155,120)(2,3)43 ( 265,120)(-2,3)43 ( 55,173 ) ( 47,163 ) ( 39,153 ) ( 33,143 ) ( 28,133 ) ( 22,123 ) ( 232,173 ) ( 240,163 ) ( 248,153 ) ( 253,143 ) ( 260,133 ) ( 266,123 ) ( 85,80 ) ( 70,112 ) ( 195,80 ) ( 205,112 ) ( 0,40 ) ( 25,32 ) ( 240,40 ) ( 265,32 ) ( 85,58 ) ( 85,48 ) ( 156,53 ) ( 104,146 ) ( 179,146 ) ( 81,208 ) ( 80,200 ) ( 201,208 ) ( 200,200 ) ( 26,47 ) ( 106,37 ) ( 184,37 ) ( 93,91 ) ( 74,100 ) ( 213,91 ) ( 194,100 ) ( 266,47 ) ( 67,163 ) ( 187,163 ) ( 230,300 ) ( 220,282)(200,255)(225,245 ) ( 240,282)(260,255)(235,245 ) ( 225,245)(230,243)(235,245 ) ( 245,277)(-1,1)7 ( 219,303 ) ( 221,295 ) ( 253,268 ) ( 253,258 ) ( 253,248 ) ( 20,330 ) ( 20,320 ) ( 20,310 ) ( 75,330 ) ( 75,320 ) ( 75,310 ) ( 130,330 ) ( 130,320 ) ( 130,310 ) ( 20,280 ) ( 20,270 ) ( 20,260 ) ( 75,280 ) ( 75,270 ) ( 75,260 ) ( 130,280 ) ( 130,270 ) ( 130,260 ) ( 36,323 ) ( 91,323 ) ( 146,323 ) ( 36,273 ) ( 91,273 ) ( 146,273 ) ( 212,307)(-4,1)43 ( 212,293)(-4,-1)43 ( 215,310)(150,375)(102,346 ) ( 107,348)(-2,-1)7 ( 215,290)(150,225)(102,251 ) ( 107,249)(-2,1)7 ( 218,286)(170,233)(150,233 ) ( 150,233)(-1,0)60 ( 90,233)(50,233)(43,251 ) ( 47,245)(-2,3)4 ( 218,314)(170,367)(150,367 ) ( 150,367)(-1,0)60 ( 90,367)(50,367)(43,349 ) ( 47,355)(-2,-3)4 ( 128,309)(-1,-1)10 ( 128,309)(-1,-1)7 ( 128,289)(-1,1)10 ( 128,289)(-1,1)7 ( 73,309)(-1,-1)10 ( 73,309)(-1,-1)7 ( 73,289)(-1,1)10 ( 73,289)(-1,1)7 ( 18,309)(-1,-1)10 ( 18,309)(-1,-1)7 ( 18,289)(-1,1)10 ( 18,289)(-1,1)7 ( 118,299)(-1,0)112 ( 6,299)(0,299)(0,293 ) ( 0,293)(0,-1)187 ( 0,293)(0,-1)70 ( 6,100)(0,100)(0,106 ) ( 6,100)(1,0)5 ( 123,223 ) ( 123,368 ) ( 113,247 ) ( 113,342 ) ( 174,288 ) ( 174,304 ) ( 4,220 ) ( -52,210 ) ( -52,201 ) ( -52,192 ) ( 18,94 ) ( 141,94 ) ( 258,94 ) we now define the graph of states in the @xmath2-disc tower of hanoi puzzle , and show that it is isomorphic to @xmath9
. the isomorphism may be computed by reading sequentialy the locations of the discs , starting with the largest one ( which corresponds to the most significant digit in the sierpiski gasket labeling ) , and following a diagram of permutations translating the labels of the three pegs into the symbols @xmath11 ( another finite - state machine ! ) . together with the results of the previous section
, this will give an effective means of computing the length of the shortest path between any two vertices in the hanoi graph , and of deciding whether the largest disc will be moved once or twice in a shortest path .
after that , we describe briefly an algorithm for actually constructing the shortest path , based on the algorithm for getting to a perfect state .
label the three pegs in the tower of hanoi with the symbols @xmath64 . since in a legal state , on each of the pegs the discs are arranged in increasing size from top to bottom , a state is described uniquely by specifying , for any disc , the label of its peg .
thus , we define @xmath65 , the @xmath2-th _ hanoi graph _ , to be the graph whose vertex set is the set @xmath66 ( with the coordinates of the vectors specifying , from left to right , the labels of the pegs of the largest disc , second - largest disc , etc . ) , and where edges between states correspond to permissible moves .
figure 5 shows the graph @xmath67 .
@xmath65 and @xmath9 are isomorphic graphs .
the finite - state machine shown below translates a hanoi state @xmath68 into a sierpiski gasket labeling @xmath69 , by reading the digits from left to right and outputting the symbols @xmath11 at each step according to the identifications in its internal state , then changing the internal state according to the input .
( 300,0)(0,135 ) ( 22,70 ) ( 22,60 ) ( 22,50 ) ( 36,63 ) ( 22,86 ) ( 96,3 ) ( 82,10 ) ( 82,0 ) ( 82,-10 ) ( 54,45)(1,-1)25 ( 60,39)(-1,1)7 ( 96,123 ) ( 82,130 ) ( 82,120 ) ( 82,110 ) ( 54,81)(1,1)25 ( 60,87)(-1,-1)7 ( 170,3 ) ( 156,130 ) ( 156,120 ) ( 156,110 ) ( 170,123 ) ( 156,10 ) ( 156,0 ) ( 156,-10 ) ( 230,63 ) ( 216,70 ) ( 216,60 ) ( 216,50 ) ( 212,45)(-1,-1)25 ( 206,39)(1,1)7 ( 212,81)(-1,1)25 ( 206,87)(1,-1)7 ( 119,123)(1,0)28 ( 125,123)(-1,0)5 ( 119,3)(1,0)28 ( 125,3)(-1,0)5 ( 60,63)(1,0)146 ( 67,63)(-1,0)7 ( 106,26)(2,3)50 ( 110,32)(-2,-3)6 ( 106,100)(2,-3)50 ( 110,94)(-2,3)6 ( 60,95 ) ( 200,95 ) ( 60,24 ) ( 200,24 ) ( 90,65 ) ( 105,84 ) ( 155,84 ) ( 130,125 ) ( 130,-6 ) ( 0,-36 ) this is lemma 2 in @xcite .
there it was claimed simply that @xmath65 and @xmath9 are isomorphic , but the proof , which is by induction , actually describes how to compute the isomorphism , and this is easily seen to be equivalent to our finite - state machine formulation .
depth0pt by running the machines of figures 4 and 6 in parallel , we now have an algorithm for computing @xmath24 for two arbitrary states in the hanoi graph , and for solving the decision problem for the largest disc , i.e. to decide whether the largest disc which it is necessary to move will move once or twice . as we will show in the next section , when @xmath13 and @xmath25 are randomly chosen states , the expected stopping time of the machine is @xmath70 .
( this random variable even has an exponential tail distribution , so with very high probability only a small number of discs will need to be read to solve the decision problem . )
having solved the decision problem , the shortest path may now be computed in a straightforward manner , as described in @xcite , using the algorithm for getting to a perfect state ( use the algorithm described in @xcite , or the algorithm for the sierpiski gasket described in section 2 together with the machine of figure 6 which incidentally leads to an algorithm for getting to perfect states which we have not found in the literature ) . in this section ,
we calculate the average number of discs that must be read in order to decide whether in a shortest path the largest disc will be moved once or twice .
let @xmath71 , @xmath72 .
assume that we have already discarded the largest discs which for @xmath13 and @xmath25 were on the same peg , so that @xmath73 .
the algorithm for solving the decision problem then tells us to run the machines of figures 4 and 6 until they reach a terminal state ( or we run out of input ) .
since we have already initialized by discarding irrelevant discs , we will really be using the machine of figure 2 ( keeping track of the correct identification of the symbols @xmath74 with the pegs @xmath64 ) .
since we are dealing with random inputs , what we are really interested in is the absorption time of the markov chain whose transition matrix is @xmath75 into the terminal states 4 and 5 .
we may identify these two states to get the simpler matrix @xmath76 for @xmath77 , denote by @xmath78 the expected time to get to state @xmath79 , _ starting from state @xmath80_. then clearly we have the equations @xmath81
@xmath82 @xmath83 it may easily be verified that the solution to this system of equations is @xmath84 the value @xmath85 is our expected stopping time , since @xmath86 corresponds to the initial state .
note that this value is the limit as @xmath87 of the average number of discs that must be read ; in reality , for finite @xmath2 the value will be slightly smaller since after @xmath2 steps we run out of input and the machine terminates even if it has not reached a terminal state . to summarize : the decision problem for shortest paths can be solved in average time @xmath88 . specifically , the average number of disc pairs that our algorithm must read , once identical discs have been discarded , is bounded from above by , and converges as @xmath87 to , @xmath70 .
hinz and schief @xcite computed the average length @xmath89 of a shortest path between two random points on the _ infinite _ sierpiski gasket of unit side .
an equivalent result of hinz @xcite and of chan @xcite , in terms of the tower of hanoi , is that the average number of moves in a shortest path between two random states in the @xmath2-disc tower of hanoi , is asymptotically @xmath90 as @xmath87 . without going into too much detail ,
we show that it is possible to obtain the value of @xmath89 just by looking at the finite - state machine of figure 4 .
since we are dealing with the infinite gasket , we start with @xmath91 and , as before , decrease the value of @xmath2 after each step , so that @xmath2 will go into the negative integers .
let @xmath92 be the expected accumulated values of the variable @xmath61 if one starts the machine , with initial values @xmath93 , at either of the four non - terminal states , in order of their distance from the state start ( so @xmath94 is the total distance ; @xmath95 is the distance after discarding identical most - significant digits of @xmath13 and @xmath25 , etc . ) .
then we have the equations @xmath96
@xmath97 @xmath98 @xmath99 @xmath100 the value ( 1/2 + 2/3 ) in the second and fourth equations is the expected value of @xmath101 ( respectively @xmath102 ) , given that the first pair of inputs is one of the six values @xmath103 ( respectively @xmath104 ) . | we present efficient algorithms for constructing a shortest path between two states in the tower of hanoi graph , and for computing the length of the shortest path .
the key element is a finite - state machine which decides , after examining on the average only @xmath0 of the largest discs , whether the largest disc will be moved once or twice .
this solves a problem raised by andreas hinz , and results in a better understanding of how the shortest path is determined .
our algorithm for computing the length of the shortest path is typically about twice as fast as the existing algorithm .
we also use our results to give a new derivation of the average distance @xmath1 between two random points on the sierpiski gasket of unit side .
+ + key words : tower of hanoi , finite automata , sierpiski gasket the _ tower of hanoi _ puzzle , invented in 1883 by the french mathematician edouard lucas , has become a classic example in the analysis of algorithms and discrete mathematical structures ( see e.g. @xcite , 1.1 ) .
the puzzle consists of @xmath2 discs of different sizes , stacked on three vertical pegs , in such a way that no disc lies on top of a smaller disc .
a permissible _ move _ is to take the top disc from one of the pegs and move it to one of the other pegs , as long as it is not placed on top of a smaller disc .
the set of states of the puzzle , together with the permissible moves , thus forms a graph in a natural way .
the number of vertices in the @xmath2-disc hanoi graph is @xmath3 .
the main question of interest is to find _
shortest paths _ in the state graph , i.e. , shortest sequences of moves leading from a given initial state to a given terminal state . the simplest and most well - known case is that in which it is required to move all the discs from one of the pegs to another , i.e. where the initial and terminal states are two of the three `` perfect '' states with all the discs on the same peg .
this is very easy , and can be shown to take exactly @xmath4 moves .
more difficult is to get from a given arbitrary initial state to one of the perfect states - hinz @xcite calls this the `` p1 '' problem .
this takes @xmath4 moves in the worst case ( which is when the initial state is another perfect state ) , and on the average @xmath5 moves for a randomly chosen initial state @xcite .
moreover , there is a simple and efficient algorithm to compute the shortest path in this case . in the most general case of arbitrary initial _ and _ terminal states , however , the question of computing the shortest path and its length ( the `` p2 '' problem @xcite ) in the most efficient manner , has not been completely resolved so far . ( the worst - case behavior is still @xmath4 moves , and the average number of moves for random initial and terminal states has been shown @xcite,@xcite to be about @xmath6 . ) the main obstacle in the understanding of the behavior of the shortest path , has been the behavior of the largest disc that `` separates '' the initial and terminal states , i.e. the largest disc which is not on the same peg in both states ( trivially , any larger discs may simply be ignored ) .
it is not difficult to see @xcite that in a shortest path , this disc will be moved either once ( from the source peg to the target peg ) or twice ( from the source to the target , via the third peg ) .
the problem is to decide which of the two alternatives is the correct one .
once this is settled , the path may be constructed by two applications of the algorithm for the p1 problem .
hinz @xcite proposed an algorithm for the computation of the shortest path based on this idea .
the algorithm consists essentially of computing the length of the path for both alternatives and choosing the shorter of the two . in this paper ,
we propose a more thorough explanation of the process whereby it is decided which of the two paths is the shortest .
we show that it is possible to keep track of the relevant information using a finite - state machine , which at each step reads the locations of the next - smaller disc in the initial and terminal states , and changes its internal state accordingly . eventually , the machine reaches a terminal state , whereupon it pronounces which of the two paths is the shortest . for a random input ,
its expected stopping time is computed to be @xmath7 .
in other words , after observing on the average the locations of just the @xmath8 largest discs in the initial and terminal states , we will know which of the paths to choose , and we will be able to continue using the algorithm for the p1 problem . if one is interested just in the length of the shortest path , then our algorithm is typically about twice as fast as the algorithm proposed by hinz @xcite ( with a small constant overhead due to the initial 1.66 discs ) , since it overrides the need to compute both the distance for the path that moves the largest disc once , and the path that moves it twice . the paper is organized as follows : in the next section , we define the _ discrete sierpiski gasket _ graph , a graph which is isomorphic to the tower of hanoi state graph , but for which the labeling of the vertices is simpler to understand . in section 3 ,
we present the main ideas for this graph , and then in section 4 show how to translate the results to the hanoi graph by a re - labeling of the vertices . in section 5
we perform a probabilistic analysis of the finite - state machine , to compute the average number @xmath7 of discs that need to be read in order to decide whether the largest disc will be moved once or twice , and to give a new derivation of the asymptotic value @xmath6 for the average distance between two random states in the @xmath2-disc hanoi graph .
we now define a family of graphs called _
discrete sierpiski gaskets_. these graphs are finite versions of the famous fractal constructed by the polish mathematician waclaw sierpiski in 1915 .
the connection between the tower of hanoi problem and the sierpiski gasket was first observed by ian stewart @xcite , and was later used by andreas hinz and andreas schief @xcite in their calculation of the average distance between points on the sierpiski gasket . the @xmath2th discrete sierpiski gasket graph , which we denote by @xmath9 , consists of the vertex set @xmath10 ( the symbols @xmath11 indicate `` top '' , `` left '' and `` right '' , respectively ) , with the edges defined as follows : first , for each @xmath12 ( for reasons that will become apparent below , this will be our standard indexing of the coordinates of the vertices of @xmath9 ) we have edges connecting @xmath13 to @xmath14 second , define the _ tail _ of @xmath15 as the suffix @xmath16 of @xmath13 , where @xmath17 is maximal such that @xmath18 . if @xmath13 has a tail of length @xmath19 , then @xmath13 is of the form @xmath20 .
connect @xmath13 with an edge to the vertex @xmath21 one possible embedding of @xmath9 in the plane is illustrated in figure 1 below .
this embedding makes clear the meaning of the labeling of the vertices : the first letter ( the `` most significant digit '' ) signifies whether the vertex is in the top , left or right triangles inside the big triangle ; the next letter locates the vertex within the top , left or right thirds of that triangle , etc .
figure 1 : the graph @xmath22 it will be shown in section 4 that @xmath9 is isomorphic , in a computationally straightforward way , to the @xmath2-disc hanoi graph . ( the same was shown in @xcite , with less emphasis on explicit computation of the isomorphism . ) thus , the problem of shortest paths on the hanoi graph reduces to that of shortest paths in the discrete sierpiski gasket .
we tackle this problem in the next section . for vertices
@xmath23 , we define the distance @xmath24 to be the length of a shortest path from @xmath13 to @xmath25 .
our goal is to write down a recursion equation for this distance , which is at the heart of the finite - state machine we will construct to compute @xmath24 .
first , let us review briefly some of the known facts about @xmath24 in the simple case when @xmath25 is one of the `` perfect '' states @xmath26 . for concreteness ,
assume that @xmath27 , and let @xmath28 as before .
then it is known that @xmath29 a simple algorithm exists for computing a shortest path from @xmath13 to @xmath25 in this case . in the hanoi labeling of the graph , the algorithm is described in @xcite . in the current labeling ,
the algorithm is even simpler and is based on the binary number system : if one identifies the symbol @xmath30 with @xmath31 and the symbols @xmath32 and @xmath33 with @xmath34 , then traversing the edges of the graph becomes equivalent to the operations of subtraction or addition of 1 in binary notation .
the number of steps to reach @xmath35 is then clearly the right - hand side in the above equation . with these preparatory remarks ,
we now attack the problem of general @xmath36 .
first , observe that we may assume that @xmath37 , since otherwise we may simply consider @xmath13 and @xmath25 as vertices in the graph @xmath38 ( note the self - similar structure in the definition of the graph , also apparent in the tower of hanoi puzzle when one ignores the largest disc ) .
for concreteness , we shall analyze in detail the case where @xmath39 . referring to figure 1 for convenience
, we see that @xmath40 @xmath41 since in a shortest path from @xmath13 to @xmath25 , one must go from the top triangle to the right triangle either through the edge @xmath42 ( we call this alternative 1 , see theorem 1 below ) or through a shortest path from @xmath43 to @xmath44 ( alternative 2 ) - in the tower of hanoi language , this is an indication of the fact that in a shortest sequence of moves the largest disc must move either once or twice , see @xcite . to simplify the next few equations , introduce the following notation : if @xmath45 , let @xmath46 , and define for any @xmath47 @xmath48 then we have @xmath49 the recursion equations which will enable us to construct our finite - state machine and compute @xmath24 are now given by the following theorem : |
while positive feedback has been overwhelmingly studied in complex networks , negative feedback remains ubiquitous in nature .
there is much room for modeling network growth besides the traditional degree - based preferential attachment .
a simple twist on this seminal work is to form attachments based on the clustering coefficient . doing so naturally creates a negative feedback mechanism that leads to aging , burstiness , and the formation of community structure in networks .
the simplicity and robustness of this mechanism is encouraging and may serve as a starting point for investigating the origin of higher - order structures in growing networks , as well as evolving networks that are in equilibrium .
the emergence of communities and highly variable temporal behavior observed in many complex networks , social networks in particular , can be investigated from a ca perspective . based on our results
, it may be promising to investigate systems in which attachment propensities are determined by other centrality measures that capture a different aspect of local network properties .
it is worth considering the potential practical applications of our ca model . in a poorly understood area such as complex systems , hypothetical models such as ours
are useful for discovering potentially overlooked dynamical mechanisms and may serve to direct future empirical studies to explore such mechanisms . here
, one can imagine many systems where nodes are drawn not towards hubs , but towards densely connected groups . for example , in a social network , individuals may not want to make friends with a very popular person but , instead , with members of a small group of very closely knit friends .
such hypotheses are becoming testable thanks to the appearance of high - resolution dynamical contact networks and face - to - face proximity data @xcite .
being attracted to density may also play a role in follower - followee networks for flocking or swarming animals @xcite , where individuals may wish to belong to a small but very cohesive group instead of being part of a jumbled crowd all following a single leader ( the hub animal ) .
another area of interest may be the dynamical evolution of functional brain networks .
indeed , positive feedback is associated with neurological conditions such as epileptic seizures @xcite .
recently , it has been shown @xcite that networks derived from fmri data are better explained by a model where new connections prefer to complete triangles than by traditional preferential attachment .
this model is still quite different from our work .
it incorporates anatomical distances in its attachment mechanism , but it demonstrates that clustering can play a role in the evolution of real systems . preferential inhibition can also be used to model fads and fashions .
for example , music listeners may actively seek musicians that are not well known .
this corresponds to attachment probabilities that decrease with increasing degree , of which clustering attachment is one example .
the prevalence of community structure in social systems is not explained by degree preferential attachment alone .
likewise , social networks typically feature exponential cutoffs in the degree distribution , simply because people have limited time with which to maintain social relationships .
this may imply that * both * preferential attachment and preferential inhibition ( or , equivalently , density attachment ) mechanisms are involved .
mixing some inhibition into the system will both inject community structure and limit the formation of very high degree nodes .
practically , this means that agents in a system are simultaneously drawn towards highly connected regions and densely connected regions .
we believe that exploring these combined effects is a very intriguing direction for improving our understanding of such systems .
we thank f. simini and s. redner for many useful discussions and the volkswagen foundation for support .
35ifxundefined [ 1 ] ifx#1 ifnum [ 1 ] # 1firstoftwo secondoftwo ifx [ 1 ] # 1firstoftwo secondoftwo `` `` # 1''''@noop [ 0]secondoftwosanitize@url [ 0 ]
+ 12$12 & 12#1212_12%12@startlink[1]@endlink[0]@bib@innerbibempty @noop * * , ( ) @noop _ _ ( , ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) link:\doibase 10.1103/physrevlett.85.5234 [ * * , ( ) ] @noop _ _ ( , ) link:\doibase 10.1016/j.physrep.2012.03.001 [ ( ) , 10.1016/j.physrep.2012.03.001 ] @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) http://stacks.iop.org/1367-2630/14/i=1/a=013055 [ * * , ( ) ] @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) link:\doibase 10.1103/physreve.83.025102 [ * * , ( ) ] link:\doibase 10.1038/srep00397 [ * * , ( ) ] @noop * * , ( ) link:\doibase 10.1073/pnas.122653799 [ * * , ( ) ] @noop * * , ( ) link:\doibase 10.1103/physreve.74.036121 [ * * , ( ) ] @noop * * , ( ) link:\doibase 10.1103/physrevlett.109.098701 [ * * , ( ) ] @noop * * , ( ) link:\doibase 10.1016/j.physd.2006.09.009 [ * * , ( ) ] link:\doibase 10.1103/physreve.85.066118 [ * * , ( ) ] link:\doibase 10.1103/physreve.69.026113 [ * * , ( ) ] link:\doibase 10.1371/journal.pone.0023176 [ * * , ( ) ] link:\doibase 10.1103/physreve.83.056109 [ * * , ( ) ] @noop * * , ( ) @noop * * , ( ) link:\doibase 10.1073/pnas.1111738109 [ * * , ( ) ] | network models with preferential attachment , where new nodes are injected into the network and form links with existing nodes proportional to their current connectivity , have been well studied for some time .
extensions have been introduced where nodes attach proportionally to arbitrary fitness functions . however , in these models , attaching to a node always increases the ability of that node to gain more links in the future .
we study network growth where nodes attach proportionally to the clustering coefficients , or local densities of triangles , of existing nodes . attaching to a node
typically lowers its clustering coefficient , in contrast to preferential attachment or rich - get - richer models .
this simple modification naturally leads to a variety of rich phenomena , including aging , non - poissonian bursty dynamics , and community formation .
this theoretical model shows that complex network structure can be generated without artificially imposing multiple dynamical mechanisms and may reveal potentially overlooked mechanisms present in complex systems .
growing network models have been introduced to study the topological evolution of systems such as citations between scientific articles @xcite , protein interactions in various organisms @xcite , the world wide web @xcite , and more @xcite .
meanwhile , recent interest has been drawn towards understanding not simply the topology of these systems or how the individual system elements interact , but also the temporal nature of these interactions @xcite .
for example , studies of the burstiness of human dynamics @xcite , whether by letter writing @xcite or mobile phone usage @xcite , have advanced our knowledge of how information spreads @xcite through systems mediated by such dynamics @xcite .
one of the most successful mechanisms to model growing networks remains preferential attachment ( pa ) @xcite .
the original pa model starts from a small seed network that grows by injecting nodes one at a time , and each newly injected node connects to @xmath0 existing nodes .
each existing node @xmath1 is chosen randomly from the current network with a probability proportional to its degree : @xmath2 , where @xmath3 is the degree , or number of neighbors , of node @xmath1 .
this `` rich - get - richer ''
mechanism leads to scale - free degree distributions , @xmath4 , where the earliest nodes will , over time , emerge as the wealthiest hubs in the network , accruing far more links than those nodes injected at later times .
this strong early - mover advantage is one of the most striking features of pa .
pa alone can not account for topological and statistical features observed in real networks such as dense modular structures @xcite and high clustering ( the abundance of triangles beyond what is expected by chance ) @xcite , and its most significant feature , the scale - free degree distribution , collapses in equilibrium situations ( in which node injections are balanced by node removal ) @xcite .
however , the success of pa is the identification of a minimal set of mechanistic ingredients ( growth , degree - driven attachment , and thus positive feedback ) that are required to account for a universal feature abundant in many real systems .
pa has thus been the basic starting point for more complex models that generalize the approach to include fitness variables @xcite and temporal correlations @xcite to account for higher clustering and community structure observed in real - world scale - free networks .
here , inspired by the simplicity and generality of pa , we address the following general question : what are the dynamic and topological consequences if the attachment propensity of incoming nodes is determined by a target node s _ neighborhood _ instead of its pure degree .
although this type of modification of the original pa model is small mechanistically , we show that the dynamic consequences are substantial .
our model exhibits _ emergent _ aging and temporally correlated dynamics , and it naturally possesses negative feedback in the attachment propensity of existing nodes .
numerical investigations supported by theory show that these effects are controlled entirely by the attachment process .
no additional , artificially imposed rules are necessary .
we adapt the original preferential attachment network growth model in the following way . instead of attaching to an existing node @xmath1 with probability proportional to its degree @xmath3
, we attach proportional to its _ clustering coefficient _ ( clustering attachment , or ca ) @xmath5 is the clustering coefficient of node @xmath1 , @xmath6 is the number of links between neighbors of @xmath1 or , equivalently , the number of triangles involving node @xmath1 , @xmath7 is a constant probability for attachment ( which may be zero ) , and the exponent @xmath8 is a parameter in our model
. other aspects of network growth remain the same .
( we assume each new node attaches to @xmath9 existing nodes throughout ; the features are the same for @xmath10 , but calculations become more cumbersome . ) we investigate both growing and fixed - size evolving networks . for the latter , a random node is removed every time a new node is added . for the original pa mechanism ,
the only possible `` reaction '' upon attaching to @xmath1 is to increment its degree , i.e. , @xmath11 .
for ca , however , two reactions are possible : @xmath12 or @xmath13 . while the degree always grows , the number of triangles @xmath6 around @xmath1 depends on whether a neighbor of @xmath1 also receives a new link .
these two reactions lead to the following potential changes in the clustering coefficient of the existing node before and after the attachment : @xmath14 here , @xmath15 is the change due to connecting to @xmath1 and a neighbor of @xmath1 , while @xmath16 is the change due to connecting to @xmath1 and a non - neighbor of @xmath1 . even when a new triangle is formed , the clustering coefficient after an attachment is almost always less than it was before : an increase in @xmath17 after a new node s attachment is only possible if the existing node has degree @xmath18 .
this means that , in contrast to pa , the ca mechanism does not feature rich - get - richer effects .
instead , attaching to a node @xmath1 drives down @xmath1 s probability for further attachments . a pure ca system will not exhibit a power - law degree distribution because negative feedback prevents the emergence of hub nodes .
instead , networks grown according to ca exhibit an exponential tail in the degree distribution . forming new links based on the clustering coefficient
provides a particularly simple model of such negative feedback or preferential inhibition . yet , temporal effects play a role here as well , with the temporal sequence of node injections determining what happens to subsequent nodes .
for example , suppose a new node is injected and happens to form a triangle .
this will give that new node maximum @xmath17 ; it may become a _ hot spot _ for future attachments . in fig .
[ fig : cartoon]a we draw a single realization of the ca model with @xmath19 nodes and @xmath20 .
qualitatively , we observe that ca dynamics naturally gives rise to community structure @xcite , where the hot spot forms the seed for a new dense group to grow .
these communities tend to form sequentially : a hot spot forms and then many nodes attach to it , driving its attractiveness down until another seed appears .
this repeating process emerges naturally from the attachment mechanism ; nothing has been artificially imposed . ) .
node size is proportional to clustering and node color represents the age of the node ( time since it was injected ) .
communities emerge approximately sequentially in time . a measurement of the model s community structure using modularity @xmath21 by running a community detection algorithm @xcite while a network evolves according to ca .
raw modularity scores may be problematic since sparsity alone can potentially force @xmath21 to high values @xcite .
we instead consider @xmath21 relative to @xmath22 , the average value observed over the course of the model .
we see fluctuations in modularity over time for @xmath23 far larger than observed for purely random attachment ( @xmath24 ) .
this quantifies the successive emergence and dissolution of modular structure in the model .
these fluctuations occur for both growing and stationary networks .
the relative distributions of @xmath21 during the temporal evolution shown in ( b ) .
the random case is sharply peaked about its average value .
the clustering coefficient averaged over all nodes , whic increases significantly as @xmath8 increases .
clustering is another hallmark of community structure .
error bars denote @xmath25s.d .
[ fig : cartoon ] ] we quantify the presence and evolution of these communities by running a community detection method @xcite as a network evolves according to ca .
figure [ fig : cartoon]b depicts the optimized modularity @xmath21 of the communities found by the method .
higher values of @xmath21 can be used to indicate `` better '' communities @xcite .
however , raw values of @xmath21 should be interpreted with caution , as @xmath21 can become very large due only to sparsity in the network @xcite . instead , in fig .
[ fig : cartoon]b we plot modularity relative to its average value over the evolution of each ca realization .
we see distinct fluctuations in @xmath21 that are not present in the case of a random network ( @xmath26 ) [ fig .
[ fig : cartoon]c ] .
these fluctuations are due to the sequential growth and decay of communities : a dense community forms , boosting @xmath21 ; then it becomes sparser as more nodes attach to the community , lowering @xmath21 until a new community forms and the process repeats .
these fluctuations are present for both growing and stationary networks .
further , in fig . [ fig : cartoon]d we plot the average clustering coefficient as a function of @xmath8 .
clustering is another hallmark of modular structure , and it increases as @xmath8 is increased . taken together , we find that @xmath8 plays a significant role in the modular nature of the model .
ca thus can give rise to both correlated network structure and nontrivial temporal dynamics .
an important question , however , is if this behavior is present for the entire range of exponents @xmath8 or if a critical parameter threshold exists . to understand this and characterize the dynamics further ,
we now explore ( i ) the aging dynamics of individual nodes after injection , and ( ii ) the influence that older nodes exert on newly injected ones . for the latter ,
we fix the size of the ca networks by removing a randomly chosen node alongside each new injection , as per fig .
[ fig : cartoon]b . nodes .
each matrix element ( @xmath27 ) represents the clustering @xmath28 of node @xmath1 at time @xmath29 .
nodes are indexed from oldest ( @xmath30 ) to youngest ( @xmath31 ) . at each time
step a new node is injected and the oldest node removed such that the time course of an individual node forms a diagonal across the matrix . below each matrix is a spike train denoting injections of high - clustering nodes . as @xmath8 increases ,
the clustering coefficients of individual nodes persist for longer times and the arrivals of high - clustering nodes become increasingly temporally correlated . [
fig : matrices ] ] when a new node is injected into the system , its degree @xmath32 and clustering @xmath33 will evolve with the time since injection @xmath29 . this new node may then exert an influence on the time course of subsequent nodes . to see this qualitatively , fig .
[ fig : matrices ] depicts `` space - time '' matrices for three realizations of ca . in this matrix
, each @xmath34 column represents the clustering coefficients of the network s nodes at that time .
nodes are ordered by age .
the oldest node is removed and a new node injected such that the time course of @xmath17 for each node forms a diagonal streak across the matrix . below each matrix
a spike train is shown , highlighting the injection times of high-@xmath17 nodes .
as @xmath8 increases , the injection times of high-@xmath17 nodes become temporally correlated and the clustering coefficients of those nodes decay more slowly : both temporal correlations and individual aging effects are affected by the exponent @xmath8 of the ca mechanism .
more quantitatively , by averaging over many realizations , we measure the expected time courses @xmath35 and @xmath36 for nodes that are injected with @xmath37 , shown in fig .
[ fig : expectedtimecourses ] . these time courses exhibit approximate power - law decay ( growth ) in time for @xmath38 ( @xmath39 ) . to understand the time scaling of @xmath38 and @xmath39 ,
consider the following simple analysis : first , @xmath40 and @xmath41}^{-\alpha } \sim \delta^\alpha \kbar^{-2\alpha } $ ] . assuming the time evolution of @xmath42 is approximately constant gives @xmath43 or @xmath44 where @xmath35 follows from @xmath45 .
thus we predict , if the time evolution of @xmath42 is negligible , power - law growth in time for degree with exponent @xmath46 and power - law decay in time for clustering with exponent @xmath47 . despite the simplicity of this calculation
, we find good agreement between simulations and the predicted exponents in eq . ; see fig .
[ fig : expectedtimecourses ] . for growing networks with different attachment exponents @xmath8 .
straight lines correspond to predictions @xmath48 and @xmath49 .
we observe the same scaling for growing and stationary systems , though the latter additionally feature a system - size - dependent exponential cutoff .
[ fig : expectedtimecourses ] ] yet , knowing the expected temporal scaling of individual nodes @xmath35 and @xmath36 is insufficient to understand the emergence of the network structures that we observe .
we also need to understand the temporal nature of hot - spot injection times .
thus , we turn to the time series of _ triangle injections _ , or the times when nodes are introduced with @xmath50 .
( for @xmath51 , one can consider the times when new nodes appear with @xmath52 . )
these correspond to the injections of high - clustering nodes in fig .
[ fig : matrices ] .
if a system displays no memory such that the probability for a spike during any time interval ( @xmath53 ) depends only on @xmath54 , then the triangle injections form a poisson process and the interevent time , or the waiting time between spikes , follows an exponential distribution . yet , many systems do not follow poisson processes @xcite .
a phenomenon is considered bursty when it possesses a memory ; i.e. , the probability for a new event decays with the time since the last event giving rise to a nonexponential interevent time distribution . in fig .
[ fig : spike_times]a , we study the interevent time distribution for triangle injections during ca network evolution .
( as mentioned before , to ensure the system is stationary , for the temporal dynamics in fig .
[ fig : spike_times ] we now fix the size of the network by removing one node at each time step as well . )
when @xmath24 , there is no memory and the distribution is exponential , as expected . as @xmath8 grows , however , the interevent time distribution becomes more and more heavy tailed , indicating increased probability for a triangle to form soon after a previous triangle was introduced .
a straightforward way to study bursty dynamics is through the hazard function @xmath55 , where @xmath56 and @xmath57 are the probability and cumulative distributions of waiting time @xmath29 , respectively .
the hazard function can be interpreted as the probability rate for a new spike to occur @xmath29 timesteps following the previous spike , given that no spikes occur in the intervening time interval .
we measure the hazard functions in fig . [
fig : spike_times]b . for a poisson process
, @xmath58 is constant .
increasing @xmath8 gives increasingly non - poissonian hazard functions : the ca mechanism naturally incorporates bursty time dynamics in the sequences of triangle injections .
a typical property of bursty systems is a hazard function that behaves algebraically for early times , @xmath59 with a singularity in continuous time for @xmath60 .
the exponent @xmath61 determines the degree of burstiness of the system ( with @xmath62 corresponding to the limiting case of a poisson process ) $ ] . ] .
we now unify the bursty time dynamics for triangle formation with the aging time courses for node clustering [ eq . ] . for an active system in equilibrium , the density of spikes @xmath63 at time @xmath29 should become approximately constant ( i.e. , independent of time ) such that the expected number of spikes emitted in a time interval @xmath64 is proportional to @xmath65 .
( this is not the same as a poisson process , as the expectation is over an ensemble of ca realizations . )
suppose a spike occurred at some past time @xmath66 ( without loss of generality , we shift time so that @xmath67 )
. then , assuming spikes are rare , a point we will return to , we approximate the spike density at @xmath29 by @xmath68 in other words , a spike occurs at @xmath29 , depending on the probability for the most recent preceding spike to occur at @xmath69 ( which is itself governed by the hazard function for the spike at @xmath70 ) weighted by the clustering at time @xmath29 . given eq .
, what hazard function will give rise to a constant @xmath71 ?
if @xmath72 , we have @xmath73 where @xmath74 from eq . and the second relation follows by introducing a constant @xmath75 to ensure the initial condition @xmath76 and that the integral does not diverge . when @xmath77 , @xmath78 as @xmath79 , and thus we expect an equilibrium system to be a poisson process for @xmath80 . when @xmath81 , however , no poisson process can be in equilibrium for our expected @xmath35
. instead , a time - dependent hazard function @xmath82 ( @xmath83 ) is necessary : @xmath84 where the latter holds when @xmath81 .
therefore , the system will be in equilibrium when @xmath85 . as we mentioned , eq .
is most valid at low spike densities , where the typical time between spikes is much greater than the typical time it takes for @xmath35 to decay . for higher densities ,
the probability for a new spike to occur at time @xmath29 will depend upon a superposition of earlier spikes . yet the contributions of the earlier spikes will _ each _ be time independent when @xmath85 . thus , our derivation should hold even at higher spike densities . in summary ,
if the above arguments hold we expect an equilibrium system to exhibit a hazard function @xmath86 with @xmath87 indeed , there is good evidence for this relationship in the inset of fig .
[ fig : spike_times]b . .
when @xmath24 , we recover the constant @xmath58 ( @xmath88 ) corresponding to a poisson process .
( inset ) the observed relationship between @xmath8 and the fitted @xmath61 .
the solid line is the prediction @xmath89 of eq . .
[ fig : spike_times ] ] |
galaxies are predicted to acquire the majority of their baryons through cold , k ) is well above that of the cold material in the disk of the galaxy . ]
filamentary streams that penetrate deep inside the dark matter halos without shock - heating to the virial temperature @xcite .
this so - called cold mode accretion ( cma ) may be the primary mechanism by which galaxies acquire material needed for star formation from the intergalactic medium ( igm)@xcite . such streams are predicted to have low temperatures ( @xmath10 k ) , low metallicities ( averaging @xmath11 depending on the simulation ) , and to be predominantly ionized @xcite .
cma models predict such streams will remain cold only for galaxies below a threshold mass @xmath12 ( or @xmath13 ; * ? ? ?
* ; * ? ? ?
estimates of the covering factor of cold streams within the virial radius for such galaxies range from @xmath14 up to @xmath15 , depending on redshift and model parameters . despite the fundamental role cma may play in galaxy formation and evolution , observations have provided little direct evidence for its existence .
qso absorption lines can probe the circumgalactic medium of foreground galaxies , with intermediate h@xmath1i column density systems ( @xmath16 ) being particularly promising tracers of cold accretion streams ( e.g. , * ? ? ?
* ; * ? ? ?
* ) . among these , llss with @xmath17
are both readily identifiable and often allow for straightforward h@xmath1i column density measurements due to the flux discontinuity they cause at the lyman limit ( 912 in the rest frame ) .
the use of llss to probe infalling and outflowing matter near galaxies has the advantage that they are selected in a metallicity - independent manner .
thus , unlike searches for mg@xmath1ii and other metal line - selected absorbers , lls searches are not biased in favor of either metal - enriched winds or infalling matter in galaxy halos .
furthermore , when the strength of the lyman break is not too strong ( i.e. , @xmath18 ) , the h@xmath1i column density of the system can be measured , from which the metallicity and , ultimately , the origins of the absorbing gas can be determined .
these features circumvent some of the problems that currently cause disagreements in the interpretation of the strong mg@xmath1ii - selected absorbers , for which some studies suggest the absorbers trace outflowing material while others suggest they may represent infalling material @xcite .
recent surveys of llss have examined the statistical nature of the absorber population , delineating the redshift evolution of these absorbers @xcite .
the literature contains only a few llss for which the physical properties of the gas ( metallicity , ionization structure , kinematics ) and the host galaxy ( luminosity , metallicity , mass ) are well constrained @xcite . to understand the implications for the statistical evolution of llss in the context of galaxy evolution it is necessary to better understand the gas - galaxy relationship , including the frequency with which llss trace infall onto versus outflows from galaxies .
here we use observations from the cosmic origins spectrograph ( cos ) , with additional ground - based spectroscopic and imaging observations , to analyze a lls at @xmath19 along the sight line to the uv - bright qso pg1630 + 377 ( @xmath20 ) .
we demonstrate the lls has low metallicity ( [ sec : absorption ] ) and use large binocular telescope ( lbt ) imaging and keck spectroscopy of galaxies in the qso field to identify a near - solar metallicity , 0.3 l@xmath21 galaxy at virtually the same redshift ( [ sec : galenv ] ) .
we discuss this gas in the context of cold mode accretion models in
[ sec : origins ] and summarize our results in
[ sec - sum ] .
the uv spectra of pg1630 + 377 were obtained with cos on - board the _ hubble space telescope _ ( pid 11741 , pi tripp ) using the g130 m ( 11501450 ) and g160 m ( 14051775 ) gratings .
the exposure times for these two configurations were 23.0 and 14.3 ks , respectively , giving s / n up to 3040 per resolution element at unabsorbed wavelengths .
the excellent quality of the uv spectra is displayed in figures [ fig : big ] and [ fig : profile ] .
the data were processed using calcos ( v2.11b ) and coadded following @xcite .
the lower - right panel of figure [ fig : big ] shows a small portion of the spectrum obtained from cos , highlighting the presence of a lls at @xmath19 .
optical spectra were obtained with hires at the keck observatory on 26 march 2010 and cover mg@xmath1ii @xmath22 at @xmath19 .
we acquired 3 hires exposures totaling 2.2 ks under good conditions with the blue cross - disperser , the c1 decker ( 0.86 width giving fwhm @xmath23 km s@xmath6 ) , and the ccd mosaic .
the data were processed using hiredux in idl.xavier/hiredux/index.html ] figure [ fig : profile ] shows the normalized absorption profiles of h@xmath1i ( _ left panels _ ) and metal ions ( _ right panels _ ) as a function of velocity relative to @xmath24 , the centroid of mg@xmath1ii absorption . at @xmath25
km s@xmath6 we detect h@xmath1i , c@xmath1iii , si@xmath1iii , weak mg@xmath1ii absorption ( @xmath26 m ) , and possibly o@xmath1vi . we refer to this as the strong component due to the strength of the h@xmath1iabsorption . near @xmath27
km s@xmath6 ( @xmath28 ) we detect h@xmath1i , c@xmath1iii , si@xmath1iii , and o@xmath1vi , but no mg@xmath1ii .
we refer to this as the weak component . in figure
[ fig : nav ] we show the apparent column density profiles , @xmath29 @xcite , of selected species .
this figure demonstrates that the o@xmath1vi follows the intermediate ion si@xmath1iii and h@xmath1i in the weak component well and that all of the observed o@xmath1vi could be due to the weak component .
the column densities and limits for selected species can be found in table [ tab : stats ] .
we determine the column densities of metal ions by integrating their @xmath29 profiles . for non - detections , we quote @xmath30 limits following @xcite . from the flux decrement at the lyman limit , we measure the total optical depth of the system to be @xmath31 ( see figure [ fig : big ] ) , implying a total h@xmath1i column density @xmath32 ( see * ? ? ?
we determine the h@xmath1i column density of the weak component by integrating the @xmath29 profiles of the weak , unsaturated lyman series lines .
the top panel in figure [ fig : nav ] shows the constrast in the saturation effect between the two components : for the weak component , the weak lyman series lines provide a good estimate of the h@xmath1i column density .
we find @xmath33 for the weak component .
the column density of the strong component is the difference between the total and weak component column densities , @xmath34 . as a consistency check ,
we have also fitted the h@xmath1i absorption profiles with a two component model , wherein a model absorption profile is convolved with the cos instrumental spread function to determine the best fit values of the central velocities , column densities , and doppler parameters of the two assumed components .
the columns derived from such profile fitting models are consistent with the integrated values , while the @xmath35-values for both components are about 30 km s@xmath6 .
we emphasize , however , that the results from profile fitting depend upon the assumed component structure of the gas . while we have assumed a two component model , each is quite likely made up of several blended absorbing components . to determine the metallicity and physical conditions of the strong component , we model its ionization using cloudy ( v08.01 , * ?
we assume the gas is photoionized , modeling it as a uniform slab in thermal and ionization equilibrium .
we illuminate the slab with the @xcite background radiation field from quasars and galaxies .
we vary the ionization parameter , @xmath36 , and metallicity of the gas ( assuming solar relative abundances from @xcite ) to match the observed column density constraints ( table [ tab : stats ] ) .
we summarize the results of the cloudy simulations for the strong component in figure [ fig : cloudy ] .
the ionization parameter is very well constrained by the adjacent ionization states of si and c , while the metallicity of the gas is fixed mostly by mg@xmath1ii .
the observed column densities are reproduced for models with an ionization parameter @xmath37 , which is represented by the green shaded region in the upper panel .
this @xmath38 gives a metallicity of @xmath3=-1.71\pm 0.06 $ ] .
for this range of @xmath38 the gas is almost completely ionized , with a neutral hydrogen fraction , @xmath39 .
the particle density in these models is @xmath40 @xmath41 , the total h column is @xmath42 , the physical size of the absorbing cloud is @xmath43225 kpc , and the temperature of the gas is predicted to be @xmath44 k. this temperature is consistent with the component fitting where @xmath45 km s@xmath6 which implies a temperature of @xmath46 k. we note that the ionization state of this gas is unusually well constrained by the lack of either si@xmath1ii or si@xmath1iv absorption and the presence of strong si@xmath1iii .
the intermediate and low ions in the weak component can be described by similar photoionization models with a metallicity consistent with the strong component . however
, this does not explain the strong o@xmath1vi absorption .
we hypothesize the weak component is a multiphase structure in which low - metallicity gas is interacting directly with the gaseous corona of the host galaxy ( [ sec : galenv ] ) .
in this scenario the o@xmath1vi is produced in the interface between the low - ionization gas and the corona , analagous to the o@xmath1vi seen in the galactic high velocity clouds @xcite .
the close kinematic relationship of the intermediate and high ion profiles ( figure [ fig : nav ] ) is consistent with this hypothesis @xcite .
the close kinematic relationship of the intermediate and high ion profiles , however , could also suggest another possibility in which all the ions trace the same gas .
the ionic ratios ( c@xmath1iii / o@xmath1vi , si@xmath1iii / o@xmath1vi , si@xmath1iii
/ h@xmath1i ) can be matched by nonequilibrium models @xcite if the metallicity of the gas is near solar . in this case
, the weak component would represent radiatively - cooling material associated with the host galaxy .
the close correspondence between the o@xmath1vi and the h@xmath1i in the weak component is not unusual ; at low redshifts , o@xmath1vi and h@xmath1iabsorption profiles are often observed to be well - aligned and to have similar profile shapes @xcite .
lcc h@xmath1i & @xmath47 & @xmath48 + c@xmath1ii & @xmath49 & @xmath50 + c@xmath1iii & @xmath51 & @xmath52 + si@xmath1ii & @xmath53 & @xmath54 + si@xmath1iii & @xmath55 & @xmath56 + si@xmath1iv & @xmath57 & @xmath58 + mg@xmath1ii & @xmath59 & @xmath60 + o@xmath1vi & & @xmath61 +
the left panel of figure [ fig : big ] shows a @xmath62-band image of the field toward pg1630 + 377 obtained with the large binocular camera ( lbc ) on the lbt .
the 2.5 ks lbc image was taken with @xmath63 seeing and reaches 5@xmath64 limiting magnitude of @xmath65 mag in a @xmath66 aperture , equivalent to @xmath67 at @xmath19 .
a full description of the lbt imaging will be presented in a future paper .
we selected targets for a preliminary redshift survey based on photometric redshifts of bright galaxies from the sloan digital sky survey ( sdss ) .
for the three galaxies marked in figure [ fig : big ] , we obtained longslit spectra with the keck / lris through a @xmath68 slit using the d560 dichroic with the 600/4000 grism ( blue side ) and 600/7500 grating ( red side ) .
this gave spectral coverage between 3000 to 5500 ( blue side ) and 5600 to 8200 ( red side ) .
on the blue side , binning the data 2 @xmath69 2 resulted in a dispersion of 1.2 per pixel and a resolution of r@xmath70 ( fwhm@xmath71280 km s@xmath6 ) . on the red side , the data were binned 2 @xmath69 1 , resulting in a dispersion of 2.3 per pixel and a resolution of @xmath71200 km s@xmath6 . for strong nebular emission lines from the galaxies we achieved s / n@xmath72 pixel@xmath6 with exposure times of @xmath73 s in the blue and @xmath74 s in the red .
the data reduction and calibration ( see * ? ? ?
* ) were carried out using the lowreduxxavier / lowredux / index.html ] idl software package , which includes flat fielding to correct for pixel - to - pixel response variations and larger scale illumination variations , wavelength calibration , sky subtraction , and flux calibration using the spectrophotometric standard star g191-b2b .
we also applied a flux correction to the spectra using sdss photometry to ensure the flux calibrations of the red and blue side spectra were consistent and to correct for light losses in the 1 slit . to do this
we convolved the lris spectra with sdss @xmath75 filter response curves using the idl code `` spec2mag '' ( see * ? ? ?
we then compared the spectrally - determined apparent magnitudes with the sdss catalog apparent magnitudes ( see * ? ? ?
* ) to derive the correction .
we corrected the spectra for foreground galactic reddening using the maps of @xcite and assume an intrinsic ratio of h@xmath76 to h@xmath77 of 2.86 to correct for internal reddening , which corresponds to case b recombination at an effective temperature of 10,000 k and electron density of 100 cm @xmath78 @xcite .
we determined the galaxy redshifts by fitting template spectra to the lris spectra using a modified version of the sdss _ zfind _ code .
systematic effects ( e.g. , wavelength calibration ) dominate the errors , limiting the accuracy of our measurements to @xmath79 km s@xmath6 .
we eliminate one of the candidate hosts ( 83_19 ) due to its low redshift ( @xmath80 ) .
the other two galaxies have redshifts similar to that of the lls ( see figure [ fig : big ] ) at impact parameters km s@xmath6 mpc@xmath6 , @xmath81 , and @xmath82 . ] of @xmath83 kpc ( 296_9 ) and @xmath84 kpc ( 75_36 ) .
however , the velocity of 75_36 relative to the absorber is likely too high for it to be the associated galaxy : @xmath85 km s@xmath6 .
the velocity separation between 296_9 and the lls is @xmath86 km s@xmath6 .
we identify this galaxy as associated with the lls .
there are a number of other galaxies in the field without spectroscopic redshifts .
however , the identification of the lls with 296_9 is secure for the following reasons .
most importantly , the velocity separation between the absorber and this galaxy is extremely small .
it is very unlikely for such a small velocity offset to occur randomly .
at the very least , the absorber is associated with the environment with which 269_9 is associated .
in addition , all of the galaxies nearer to the qso sight line than the candidate host are extremely faint with very small angular diameters , consistent with their being high redshift background systems .
we use the observed h@xmath76 luminosity of 269_9 with the @xcite relationship to estimate a star formation rate of this galaxy , deriving @xmath87 m@xmath88 yr@xmath6 .
the stellar mass of 296_9 is @xmath89 , derived from sdss photometry .
we determined the oxygen abundance using the r23 metallicity indicator @xcite as calibrated by @xcite .
we derive an abundance for 296_9 of @xmath90=-0.20\pm0.15 $ ] .
this estimate places the object near the convergence of the two branches of the r23 diagnostic , thus the systematic errors due to the degeneracy in metallicities associated with the strong line method are small compared to the factor of 30 difference in metallicities seen between the lls gas and the galaxy .
we used the [ n@xmath1ii]/[o@xmath1ii ] ratio , following @xcite , to confirm the derived abundance .
the general systematic errors associated with age effects and stellar distributions can be as large as 0.25 dex @xcite , but such large effects are still not enough to modify our conclusions .
the metallicity , stellar mass , and star formation rate derived for 296_9 place this system on the mean mass - metallicity relationship of @xcite .
we have identified a low - redshift , @xmath91 , lls with a very low metallicity , @xmath3=-1.71 \pm 0.06 $ ] , associated with a near - solar metallicity galaxy .
the metallicity of the lls is much lower than the metallicity distribution of damped lyman-@xmath76 systems ( dlas ) at @xmath92 @xcite . from a sample of 11 dlas at @xmath92
( see table 5 , * ? ? ?
* ) , the mean metallicity of these systems is around @xmath93 dex ; these systems have a large spread in metallicity , but the two lowest values are @xmath94 and @xmath95 dex . in contrast , such low metallicities do not appear to be the exception for llss at @xmath96 .
a perusal of the literature finds 4 out of 8 well - studied lls at @xmath8 have metallicities @xmath97
< -1.8 $ ] to @xmath98 ( this work , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
recently , @xcite found another strong h@xmath1i absorber with @xmath99 having @xmath97<-1.5 $ ] .
preliminary results of our on - going study to expand the number of lls metallicity measurements at @xmath92 appear to confirm that the metallicity distribution of llss is different from that of dlas at similar redshifts , with a high fraction of llss having @xmath97\la -1.5 $ ] ( n. lehner et al .
2011 , in prep . ) .
@xcite and @xcite argued that a possible origin for these low metallicity llss was low metallicity dwarf galaxies .
however , the ( tentative ) high frequency of these primitive llss could favor another origin that was not really considered previously .
indeed , since only the tail of the dwarf galaxy metallicity distribution shows such very low abundances , they are unlikely to show up in significant numbers in absorption line studies .
that is , the probability a line of sight passing through gas from galaxies like the very low metallicity dwarf galaxy izw18 @xcite is very small .
furthermore , the relatively low h@xmath1i column density of the system demonstrates the sight line does not pass through a galaxy .
except for the highest - redshift llss in this sample @xcite , all of these llss show at least a sub-@xmath100 galaxy within 100 kpc consistent with being a `` host '' galaxy . for the present sight line , our deep lbc / lbt images of the field do not show other strong candidates for a galaxy associated with the lls at @xmath101 kpc , though our spectroscopic galaxy information is limited . for the present lls , we have demonstrated that the likely host galaxy has a metallicity much higher than the lls , ruling out altogether that the gas in the lls could originate from this galaxy .
we therefore propose instead that this and the low - metallicity llss from the literature trace matter infalling onto galaxies , perhaps related to the cma streams predicted by galaxy formation simulations .
the physical properties of the low - metallicity llss are similar to those predicted by cma models , including the temperature , metallicity , and h@xmath1i column as well as host galaxy properties such as velocity offset and mass . for the present lls , the temperature of this lls is constrained to be @xmath102 k from the @xmath35-value derived from our component fit to the h@xmath1iprofiles , consistent with the temperatures predicted by all cma models . more generally , these llss are well modeled by photoionization models that predict temperatures to be about a few times @xmath103 k. the low metallicity of the present and literature systems are consistent with the predictions of @xcite , who find cold streams in their simulations have a broad distribution of abundances centered on @xmath104 solar . these llss are likely to have been enriched above primordial levels by previous star formation episodes before streaming into a galaxy .
fumagalli et al .
predict the majority of cold streams should be predominantly ionized at @xmath105 , consistent with the properties of the current sample of llss . for our study
, we estimate that the stellar mass of the host galaxy is @xmath106 , well below the predicted stellar mass cutoff for cma host galaxies .
while there could always be galaxies hidden underneath the glare of the background qso that could have contributed the gas making up this lls , we note that this is also consistent with the cma simulations .
the infalling streams in such models are enriched by small galaxies that will themselves merge with the dominant host galaxy .
our work highlights the potential importance of h@xmath1i - selected llss as probes of infalling , metal - poor gas .
while observations of some llss clearly indicate they are related to galactic outflows / winds or galaxy mergers @xcite , it is also apparent that the population of llss includes very low - metallicity gas with properties very similar to those predicted by cma models . while the sample of llss is still currently small
, we will expand it at @xmath96 in the near future with the aim to better constrain the metallicity distribution of these systems .
our analysis suggests the best metal species for tracing cma may be intermediate ions such as c@xmath1iii and si@xmath1iii ( and other ions like o@xmath1ii or o@xmath1iii ) , as the low ions ( c@xmath1ii , si@xmath1ii , and mg@xmath1ii ) and high ions ( si@xmath1iv ) are supressed in such low density , low metallicity gas . @xcite
have also argued , on the basis of numerical simulations , that the majority of weak mg@xmath1ii systems ( @xmath107 m @xmath108 m ) trace material infalling onto a galaxy .
the present lls has @xmath26 m , while the metal - poor system studied by zonak et al .
( 2004 ) has @xmath109 m both consistent with this suggestion .
we have presented high - quality uv and optical spectroscopic and imaging observations from cos/_hst _ , hires / keck , lris / keck , and lbc / lbt of a lls with @xmath110 at @xmath111 along the qso pg1630 + 377 and the galaxies in its field of view in order to explore the origin(s ) of the lls at low redshift .
the main results of our analysis are as follows : 1 .
the lls shows metal line absorption from mg@xmath1ii , c@xmath1iii , si@xmath1iii , and o@xmath1vi .
the high s / n of the cos spectrum allowed us to derive stringent upper limits for several other ions ( in particular c@xmath1ii , si@xmath1ii , and si@xmath1iv ) .
these allow a secure metallicity measurement of @xmath3 = -1.71 \pm 0.06 $ ] for the stronger of the two components in this system due to the tight constraints on the ionization state of the gas . the o@xmath1vi absorption is mostly associated with the weaker component ( at @xmath112 km s@xmath6 from the stronger absorber ) and is likely a result of its interaction with a galaxy halo .
our limited redshift survey of galaxies close on the sky to pg1630 + 377 shows that there is a 0.3 @xmath4 star - forming galaxy projected 37 kpc from the qso at nearly identical redshift to the lls ( @xmath5 km s@xmath6 ) with near solar metallicity ( @xmath90={-0.20 \pm 0.15}$ ] ) .
while our spectroscopic survey is extremely limited , the deep lbt images do not suggest another plausible host galaxy at @xmath113 kpc .
the literature contains several other very low metallicity llss at @xmath96 .
while the sample is still small , we find a high frequency ( @xmath7150% ) of @xmath96 llss have @xmath97 \la -1.5 $ ] , which appears to differ from the metallicity distribution of dlas at similar redshifts and from a similar sample size .
we propose these very low metallicity llss are signatures of infall onto galaxies , perhaps similar to the cold mode accretion predicted by cosmological simulations . the ionization , metallicity , temperature , and host galaxy properties in these simulations are in good agreement with the properties derived from observations of the h@xmath1i - selected llss .
support for program number hst - go-11741 was provided by nasa through a grant from the space telescope science institute , which is operated by the association of universities for research in astronomy , incorporated , under nasa contract nas5 - 26555 .
jch and jr acknowledge support from nasa grant nnx08aj31 g .
tmt and jdm also appreciate support from nasa grant nnx08aj44 g .
this work made use of data from the large binocular telescope .
the lbt is an international collaboration among institutions in the united states , italy and germany .
the lbt corporation partners are : the university of arizona on behalf of the arizona university system ; istituto nazionale di astrofisica , italy ; lbt beteiligungsgesellschaft , germany , representing the max planck society , the astrophysical institute potsdam , and heidelberg university ; the ohio state university ; the research corporation , on behalf of the university of notre dame , university of minnesota and university of virginia . | we present uv and optical observations from the cosmic origins spectrograph on the _ hubble space telescope _ and keck of a @xmath0 lyman limit system ( lls ) seen in absorption against the qso pg1630 + 377 .
we detect h@xmath1i absorption with @xmath2 as well as mg@xmath1ii , c@xmath1iii , si@xmath1iii , and o@xmath1vi in this system .
the column densities are readily explained if this is a multi - phase system , with the intermediate and low ions arising in a very low metallicity ( @xmath3 = -1.71 \pm 0.06 $ ] ) photoionized gas .
we identify via keck spectroscopy and large binocular telescope imaging a 0.3 @xmath4 star - forming galaxy projected 37 kpc from the qso at nearly identical redshift ( @xmath5 km s@xmath6 ) with near solar metallicity ( @xmath7={-0.20 \pm 0.15}$ ] ) .
the presence of very low metallicity gas in the proximity of a near - solar metallicity , sub-@xmath4 galaxy strongly suggests that the lls probes gas infalling onto the galaxy .
a search of the literature reveals that such low metallicity llss are not uncommon .
we found that 50% ( 4/8 ) of the well - studied @xmath8 llss have metallicities similar to the present system and show sub-@xmath4 galaxies with @xmath9 kpc in those fields where redshifts have been surveyed .
we argue that the properties of these primitive llss and their host galaxies are consistent with those of cold mode accretion streams seen in galaxy simulations . |
quantum teleportation @xcite has been one of the leading discoveries followed by numerous quantum information processing technologies @xcite .
controlled quantum teleportation @xcite is a variant in which a teleportation of a quantum state is performed under the supervision of controllers .
schemes using qubits as keys distributed among controllers @xcite have been extensively studied and economization of required resources has been accomplished with respect to the number of qubits involved in an entangled qubit chain used in a scheme .
the studies also include a security discussion on players cheating controllers @xcite .
it is expected that a multifunctional quantum network is realized for a consumer market in future , presumably based on optical fibers .
so far , simple quantum cryptosystems @xcite are highly developed @xcite toward the consumer use , which are mainly used to generate classical shared cryptographic keys .
an advanced quantum network should be used not only for generating a classical key but for exchanging quantum states . a network with einstein - podolsky - rosen ( epr ) pairs as links
@xcite is a plausible form for this purpose ; optical quantum teleportation has already become a well - established subject supported by many experimental demonstrations reported in , e.g. , refs .
@xcite . considering an application for office networks
, it is demanded that data transfers can be under the control of multiple supervisors .
controlled teleportation schemes have been developed to enable this feature .
as the number of usable qubits and the quality of entanglement enhances , this function is approaching reality . a function like a vote to permit a transfer
is also considered to be of public demand .
qubit keys are commonly used in controlled quantum teleportation schemes for the basic control such that approvals of all the controllers are required to let the players transfer a quantum state .
in such a control , use of qubit keys is sufficient and extending each dimension of a quantum system for a key causes a redundancy .
qudits ( @xmath3-dimensional quantum systems ) with @xmath6 have been , in contrast , considered to be useful as keys when a vote for decision is in order .
it is easy to notice that such functions are enabled by using a threshold scheme @xcite : controlling a transfer of a quantum state with a certain threshold in the number of controllers is implemented by secret sharing schemes ( see , e.g. , a discussion in ref .
this categorizes controlled quantum teleportation as a combination of quantum secret sharing and quantum teleportation .
a well - known polynomial coding @xcite over @xmath1 with prime @xmath7 larger than the number @xmath5 of participants ( controllers in our context ) is a quick solution can be equal to @xmath5 if @xmath5 is prime in the context of controlled quantum teleportation ( see sec . [ sec3 ] ) . ] .
a classical or quantum @xmath0-threshold scheme applied to a shared state of players and controllers enables a control such that a receiver can recover an original state when and only when @xmath8 or more controllers among @xmath5 provide their keys .
there is a variant of quantum polynomial coding robust against a certain number of cheaters among participants @xcite .
a drawback to introduce a polynomial coding scheme in controlled quantum teleportation is that computation is conducted over the field @xmath9 .
thus each qudit distributed among controllers should have the dimension @xmath4 .
this also complicates a quantum circuit to compute a matrix equation for the coding . for a threshold scheme , i.e. , a secret sharing without imposing access structure ,
the dimension of each key ( share ) using best known classical protocols @xcite is @xmath10 ( namely , @xmath11 bits for each key ) for a long secret .
these protocols require the bit length of a secret @xmath11 and that of a key at least as much as the length of a secret . for a short secret ,
a best known classical protocol is shamir s one @xcite in which each key has the dimension @xmath12 .
it is thus not motivating to simply make a quantum extension of the classical protocols .
it was reported that an @xmath13-threshold secret sharing of a classical secret is achieved by a sort of key distribution without entanglement @xcite and an @xmath13-threshold controlled quantum teleportation is achieved by using classical keys @xcite .
the latter one is easily extended to a @xmath0-threshold controlling scheme .
nevertheless , schemes using qudits for controllers portion are more secure than the one using only classical keys as we will discuss in sec .
[ secccqt ] .
there is a recently - proposed graph - state formalism @xcite to produce quantum secret sharing systems using only qubits .
this approach has achieved the systems for several particular @xmath0 and not for general @xmath0 .
we will construct a scheme which is not a direct extension of these approaches .
in this paper , we introduce a @xmath0-threshold controlled quantum teleportation scheme over @xmath1 with prime @xmath2 using @xmath5 qubits for the controllers portion in addition to classical information distributed to the controllers .
this is achieved by a hybrid of classical and quantum protocols .
it enables a reduction in the dimension of each qudit to two for general @xmath0 , which is a significant improvement for realizing the scheme .
in contrast to classical systems , the dimension of a quantum system is limited and usually a large - dimensional qudit is implemented by multiple qubits .
the size of a qudit register is also very limited in the present technology .
the number of available qubits is , so far , twelve or less @xcite . to begin with , a standard controlled quantum teleportation scheme is briefly described in sec . [ sec2 ] .
on the basis of the scheme , a @xmath0-threshold controlled quantum teleportation using a polynomial coding is briefly explained in sec .
[ sec3 ] . a reduction in required resources for the threshold - control scheme
is accomplished in sec .
[ sec4 ] : first , a classical control of a quantum teleportation is discussed in sec .
[ secccqt ] .
second , a control of a quantum teleportation using @xmath5 qubits for the controllers portion is introduced in sec .
[ seceqcqt ] .
third , an economization of bob s operations is accomplished in sec .
[ secebo ] .
an explicit preparation of an initial state for this economized scheme is shown in sec .
[ sec5 ] together with the operational complexity of the whole process .
we discuss advantages and disadvantages of the scheme in sec .
section [ sec7 ] summarizes our results .
consider a controlled quantum teleportation using ( @xmath14 ) epr pairs shared by alice and bob , and a single quantum system shared by alice , bob , and @xmath5 controllers .
alice tries to send an @xmath15-qubit state @xmath16 of the system @xmath17 to bob .
a standard quantum teleportation protocol works fine for the ( @xmath14)-epr - pair channel consisting of ( @xmath14 ) pairs @xmath18 .
the remaining channel , @xmath19 , is under the control of @xmath5 controllers @xmath20 .
the setup of the quantum system is illustrated in fig.[fig1 ] . here
it should be noted that , although we consider the control of a single channel here , it is straightforward to attach controllers to each channel . thus let us limit setups to the illustrated one in the following .
the initial state of the illustrated system is given by @xmath21 with @xmath22 to be engineered for a tailored controlling scheme .
let us recall the well - known relation @xmath23 is the @xmath24th bell basis vector ( here , @xmath25 , i.e. , @xmath26 ) . with this relation , we can rewrite the initial state as @xmath27 as is usual for a standard quantum teleportation , alice makes bell measurement on each @xmath28 pair and obtains an outcome @xmath29 .
bob receives information @xmath30 from alice and applies @xmath31 to @xmath32 .
this changes the state of each @xmath33 ( of each term in the summation ) from @xmath34 to @xmath35 and hence the teleportation process for the original state of @xmath36 is completed up to @xmath37 . to complete the recovery of the original state of @xmath36 at bob s side
, he has to apply a certain operation to @xmath38 .
this requires bob to know the effect of controllers measurements on @xmath39 .
operations bob has to apply to @xmath38 depend on the controlling scheme .
a popular controlled quantum teleportation scheme is the case where @xmath40 is set to @xmath41 , i.e. , @xmath42 is set to a greenberger - horne - zeilinger ( ghz ) state , and each controller makes a measurement in the @xmath43 basis .
measuring @xmath44 in the @xmath43 basis results in the phase factor @xmath45 depending on the outcome @xmath46 ( here , @xmath47 ; the operation `` @xmath48 '' returns one if its two arguments are equal and zero otherwise ) of a qubit can be written as @xmath49 using the @xmath43-basis states @xmath50 . ] . in this case
, bob first applies @xmath51 to @xmath38 .
in addition , he applies a single @xmath52 gate to @xmath38 for recovery if the number of @xmath53 s in the controllers outcomes ( @xmath54 ) is odd .
the final state of bob after these operations becomes @xmath55 .
the teleportation is successful in this way . in this contribution
, we aim to introduce a threshold - control scheme in which the shared state @xmath56 is engineered to be different from the ghz state .
we continue to concentrate on the case where only a single qubit of bob is under the control , as illustrated in fig .
[ fig1 ] ; this is because it is straightforward to extend the scheme so that multiple qubits are under the control .
such an extension will be considered only in sec .
[ secebo ] .
the functionality of threshold control is achieved by engineering the initial setup of the shared state given by eq .
( [ eqkappa ] ) .
let us begin with a rather expensive scheme that is a straightforward extension of a classical secret sharing . in this scheme
, the state of the controllers portion is assumed to be in the following form @xmath57 where @xmath58 , @xmath59 ( @xmath60 and @xmath3 is a positive integer ) , and @xmath61 is a certain set of @xmath5-digit strings ; the computational basis for each @xmath44 is chosen arbitrarily and known by the @xmath62th controller .
there are two conditions for the state appropriate for a @xmath0-threshold scheme : + ( i ) @xmath63 can not be uniquely determined unless the sequence @xmath64 is completely specified ( i.e. , unless the variables @xmath65 are all specified ) .
+ ( ii ) a string @xmath64 is uniquely determined in @xmath61 by fixing any @xmath8 digits of @xmath65 .
+ under these conditions , @xmath8 controllers measurements result in a single surviving vector @xmath66 .
thus bob can recover the original state by the following process ( in addition to the usual process for the original quantum teleportation for @xmath67 ) .
first bob changes the phase factor @xmath68 to unity by applying @xmath69 to the qubit @xmath38 .
second bob applies @xmath51 to @xmath38 as is usual for a quantum teleportation . in these steps ,
alice s bell measurement on @xmath70 prior to controllers measurements does not affect bob s operation to recover the phase appearing as a result of controllers measurements .
this is clear from the initial state described in eq .
( [ eqrewritten ] ) ; the parameter @xmath58 is common in the states of @xmath38 and controllers portion . let us turn to a system construction for satisfying the conditions . condition ( i ) is satisfied by setting @xmath71 .
such @xmath63 can not be uniquely determined unless the variables @xmath65 are completely specified . a quantum circuit to attach the phase factors
is easily realized : from @xmath72 to @xmath5 , applying an @xmath73-controlled @xmath74-controlled phase gate with the phase @xmath75 , acting on a work qubit accomplishes this task . for the condition ( ii )
, we will find a proper set @xmath61 of @xmath5-digit vectors @xmath64 by a certain coding scheme .
this is the main concern as the resource for quantum information processing is limited in the current technologies ; a coding with a small resource is desirable .
we revisit the theory of karnin _ et al . _
@xcite , which describes a sufficient condition for coding in a threshold scheme , and evaluate a polynomial - coding scheme @xcite as a typical example .
we begin with a well - known property of a linear matrix equation . _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ proposition : _ a solvable matrix equation over @xmath1 , @xmath76 with @xmath7 a prime number , @xmath77 , @xmath78 , and @xmath79 has a unique solution @xmath80 if and only if @xmath81 and @xmath82 has a full rank . _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ the proof is similar to the case of a real number field ( see , e.g. , ref .
+ _ proof| _
( i ) first we prove that the solution @xmath80 is unique if @xmath81 and @xmath82 has a full rank .
assume that there are two solutions @xmath83 and @xmath84 .
then , @xmath85 . let us pick up @xmath86 rows of @xmath82 appropriately to generate @xmath87 so that @xmath87 has a full rank .
this is possible because otherwise the number of linearly independent row vectors of @xmath82 should be less than @xmath86 , which is a contradiction .
we have thus @xmath88 with the square full - rank matrix @xmath87 .
the matrix @xmath87 can be reduced to a diagonal matrix with nonzero diagonal elements by basic operations ; thus @xmath89 .
consequently , @xmath90 because @xmath91 exists .
this is a contradiction .
( ii ) second we prove that @xmath81 and @xmath82 has a full rank if @xmath80 is a unique solution .
the contraposition of this statement is that solution @xmath80 is not unique if @xmath92 or @xmath93 is less than @xmath94 .
this is easily shown to be true.@xmath95 on the basis of this proposition , a coding of our interest is achieved by a matrix equation for @xmath96 with a full - rank matrix @xmath82 such that striking any @xmath97 rows keeps the rank full @xcite .
given a matrix equation @xmath98 with such @xmath82 which is notified to bob , we prepare @xmath99 as @xmath100 where @xmath101 is a set of @xmath102 corresponding to @xmath103 ( hence @xmath104 ) .
suppose that at least @xmath8 controllers measure their qudits in the computational basis .
then @xmath80 is fixed uniquely because the matrix equation becomes solvable by using the rows corresponding to the fixed @xmath105 s .
this implies that the superposition is then resolved .
bob can determine the phase factor that he should modify by receiving at least @xmath8 outcomes from controllers .
we have seen a common construction of a threshold scheme . required resources for the threshold scheme
are mostly dependent on the choice of the matrix @xmath82 .
there are two practical ways among many @xcite .
one is the matrix in the form @xmath106 with @xmath107 the @xmath108 identity matrix and @xmath109 a @xmath110 strictly totally positive matrix @xcite .
any minor of @xmath109 is nonzero positive from the definition of strict total positivity ; hence any @xmath8 rows of @xmath82 build up a square matrix with nonzero determinant .
thus a matrix equation with @xmath111 can be used for the threshold scheme .
a drawback is the difficulty to find a strictly totally positive matrix @xmath109 for sufficiently small prime @xmath7 .
a known systematic construction @xcite for strictly totally positive matrices uses the largest element of @xmath109 growing exponentially in @xmath112 .
thus @xmath7 also grows exponentially if we follow the construction .
a manual optimization is indispensable .
the other is a vandermonde matrix used in the well - known shamir s scheme @xcite , @xmath113 with mutually different @xmath114 s with prime @xmath2 .
( @xmath2 is necessary to set @xmath114 s mutually different . )
striking @xmath97 rows generates a square vandermonde matrix and it is non - singular when @xmath114 s are mutually different ( see , e.g. , pages 43 and 219 of ref .
@xcite ) . hence a matrix equation with @xmath115 can be used for the threshold scheme .
this matrix has been known to be economical because @xmath7 increases linearly in @xmath5 .
each qudit distributed to a controller should have the dimension @xmath4 , consequently .
nevertheless , one may need to further reduce the dimension considering the poor resources of presently available qudit systems @xcite .
the dimension of each digit distributed to a party ( a controller in the present context ) is often evaluated by using the scale of `` bit length '' in conventional secret sharing schemes .
each key is @xmath116-bit long @xcite ( @xmath11-bit long @xcite ) in the best known protocols for a short secret with the bit length @xmath116 ( for a long secret with the bit length @xmath11 ) although this has not been taken as a drawback at all since classical bits are very cheap .
nevertheless , in quantum protocols one should not consume many qubits for individual quantum systems .
it is of our concern to find a smaller dimension for each controller s qudit facing a limited resource of a quantum system .
one way is to abandon the use of quantum systems for controllers portion and instead use a classical threshold scheme to control a quantum teleportation .
a controlled teleportation proposed by zhang and man @xcite , in the context of @xmath13 threshold , uses classical keys shared by alice and controllers for encoding alice s messages , which can be easily extended to a general threshold - control scheme . here
, we introduce a different scheme where controllers qudits are simply replaced by classical digits .
we will face the fact that classical control schemes are indeed economical but their security is based on classical keys .
using qudits is found to be more robust against bob s physical - access attack .
our interest is to find such a robust scheme with a simple quantum state for the controllers portion .
it is shown to be constructed by using the matrix equation with the vandermonde matrix ( [ vmatrix ] ) and qubit states distributed to the controllers .
we further perform an economization in the number of bob s operations , which is useful for an extension in which multiple epr channels are under controllers control .
as we mentioned , the simplest way of economization is to use classical control digits instead of quantum ones .
this is easily achieved by setting the state ( not a state , actually ) of the controllers portion to be a scalar @xmath117 with phase @xmath118 dependent on integers @xmath65 , the classical keys of a certain classical threshold scheme .
bob can modify this scalar factor by applying @xmath119 $ ] to his @xmath15th qubit @xmath38 if he can gather at least @xmath8 of the keys . the security of the scheme is dependent on the classical scheme . indeed
, classical keys can be securely distributed by using a quantum key distribution ( see ref .
@xcite and references therein ) and the risk of an interception during the key distribution is negligible .
classical keys are , however , easily copied by careless controllers .
a possible drawback of the scheme is that controllers can not stop bob from recovering alice s original state if bob manages to obtain at least @xmath8 of the keys without consent of the controllers .
in contrast , in a threshold scheme using qudits , the operations for a recovery of the original state are unfixed until @xmath8 controllers make measurements .
this fact makes the quantum one more robust : recall that the computational basis for @xmath44 in eq .
( [ eqqss ] ) is not necessarily known in public .
the correct basis for a measurement can be left unknown to bob .
then , bob can not obtain @xmath105 by a physical access to @xmath44 unless he also has an access to the information on the measurement basis .
thus the cost for bob to steal @xmath105 s in the scheme is more than that in the classically controlled one .
this logic is similar to the one utilized for keeping the security of dealer - player communications in some quantum threshold schemes @xcite .
it should be mentioned that robustness depends on the type of protocol violations .
let us discuss a different type of violation that can be made by bob .
it is a common occasion that controllers do not want bob to process further with a teleported state before they officially vote for their decision .
a violation in this regard occurs when @xmath8 or more controllers are friendly to bob and they measure their qudits or digits and send the outcomes before the voting starts officially .
this violation in the schedule of the procedure can not be prevented even if qudits are distributed . in the following ,
an economical quantum control scheme is introduced in order to reduce the resource for controllers quantum system as we have mentioned .
it is now our additional motivation to resolve the schedule violation problem due to friendly controllers .
as is discussed above , a quantum threshold scheme has a classically unachievable property , namely that the operations for a recovery of the original state are unfixed until controllers make measurements and the measurement bases can be left unnotified to bob . as we have mentioned , our aim is to achieve such a scheme using small - dimensional systems for controllers portions . here , we propose a quantum protocol for the @xmath0-threshold controlled quantum teleportation with @xmath5 qubits distributed to the controllers .
it also resolves the problem of the possible violation in the voting schedule .
it is implemented with the following state for the controllers portion : @xmath120_{{\rm c}_s } , \end{split}\ ] ] where @xmath121 and @xmath122 are the basis vectors of the @xmath62th controller s chosen basis ; @xmath123 is a logical negation of @xmath124 and @xmath105 s are the keys of the following common classical polynomial - coding threshold scheme .
the keys are generated from eq .
( [ eqaxc ] ) using a vandermonde matrix for @xmath82 and a certain fixed vector for @xmath80 [ all the matrix and vector elements are in @xmath1 ] .
the matrix @xmath82 is notified to bob and @xmath80 is hidden .
thus @xmath8 or more keys are required for bob to determine the remaining keys from eq .
( [ eqaxc ] ) .
the protocol imposed to controllers is as follows .
+ ( i ) each controller @xmath126 who agrees to allow bob to recover alice s original state measures her / his qubit in the basis @xmath127 and sends its outcome @xmath128 to bob .
+ ( ii ) @xmath126 also sends her / his key @xmath105 to bob .
+ ( iii ) each controller @xmath129 who disagrees to allow bob to recover alice s original state does not make any action until she / he receives a contact from bob .
+ ( iv ) when a solicit is sent from bob , @xmath129 must measure her / his qubit in the basis @xmath127 and send its outcome @xmath128 to bob .
the protocol imposed to bob is as follows .
+ ( i ) bob receives @xmath29 s from alice .
bob applies @xmath31 to @xmath32 .
+ ( ii ) bob waits for at least @xmath8 pairs of @xmath130 s sent from the controllers .
+ ( iii ) bob calculates the remaining @xmath105 s by substituting the obtained @xmath105 s into eq .
( [ eqaxc ] ) if at least @xmath8 pairs are obtained ; aborts otherwise .
+ ( iv ) bob sends solicits to the controllers who did not send information to him .
+ _ note : _ it is possible to count his solicits .
therefore , he can not cheat by sending more than ( @xmath131 ) solicits in this stage even when he succeeds in stealing @xmath8 or more @xmath105 s beforehand .
+ ( v ) bob receives the remaining @xmath132 s .
+ ( vi ) bob modifies the phase factor due to controllers measurements in his phase recovery process described below .
+ ( vii ) bob applies @xmath51 to @xmath38 .
[ [ bobs - phase - recovery - process - corresponding - to - the - controllers - measurements ] ] bob s phase recovery process corresponding to the controllers measurements| + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + the @xmath5 controllers measurements ( of course , after alice s bell measurements ) make the component state for @xmath133 in eq.([eqrewritten ] ) , with the controllers state ( [ econkappa ] ) in the present context , evolve into the state @xmath134
\end{split}\ ] ] the phase factor @xmath135\ ] ] should be canceled by bob before the normal recovery operation is performed in step ( vii ) .
the cancellation of the factor with probability one is possible if and only if all the controllers follow the protocol and at least @xmath8 of them send @xmath105 s to bob ( otherwise he will get an uncertain state as his operations become a guesswork ) . since @xmath105 s are classical keys of a @xmath0 threshold scheme , @xmath8 or more of them are necessary and sufficient to find out all of them .
bob can eliminate the phase by applying @xmath136\ ] ] to his @xmath15th qubit @xmath38 .
this operation is unaffected by alice s bell measurement on @xmath70 as is clear from the initial state described in eq .
( [ eqrewritten ] ) . in this way , the @xmath0-threshold control
is realized by using only qubit systems for individual portions .
one may notice that it is a hybrid scheme in the sense that the @xmath0 threshold is realized by a classical threshold scheme .
although classical keys are used , we claim that this scheme possesses a property thanks to a quantum control ; the recovery operations are unfixed unless all the controllers make measurements , and the measurement bases are unknown to bob .
the measurements are not completely performed unless there are at least @xmath8 @xmath126 s and bob sends at most ( @xmath131 ) solicits to @xmath129 s .
hence , in order to cheat under the protocol , bob needs to collect @xmath105 s , qubits , and measurement bases of at least @xmath8 controllers .
the scheme has an advantage over the previously - introduced quantum scheme in the sense that it is secure against the possible schedule violation due to friendly controllers . owing to step ( iv )
, bob has to wait for an official voting time to obtain alice s original state unless the controllers violate the voting schedule all together .
we have shown an economized quantum ( or hybrid ) controlled quantum teleportation using the state ( [ econkappa ] ) for the controllers portion , in which the dimension of each distributed qudit has been reduced to two . as two is the minimal dimension for a nontrivial quantum system ,
it is optimized with respect to the dimension of a hilbert space of each portion . here
we consider an economization of bob s operations to eliminate the phase factor coming from controllers measurements .
the number of bob s single - qubit operations for the recovery is not important as far as a single epr channel is under the controllers control because eq .
( [ eqrecov ] ) reduces to a single operation .
it is , however , not negligible in case we extend the setup to the one illustrated in fig .
[ fig2 ] , in which we may have multiple shared states .
each controller is assigned to a single @xmath137 pair and the total number of controllers is @xmath5 .
let us use the same state as in eq .
( [ econkappa ] ) , except the labels of controllers , for each of the controllers portions ( represented by dotted lines inside the square of `` controllers '' in fig
. [ fig2 ] ) .
we change the protocol for controllers in the following way .
+ ( a ) a controller who agrees to allow bob to recover the original state makes a measurement on her / his qubit in the basis @xmath138 with @xmath139 .
she / he sends bob the outcome @xmath140 together with @xmath105 .
+ ( d ) a controller who disagrees does not make any action .
however when a solicit is sent from bob , a controller who disagrees must make a measurement on her / his qubit in the basis @xmath141 and send bob the outcome @xmath142 .
+ to understand the effect of a measurement , let us decompose each component of the state ( [ econkappa ] ) in the following way : @xmath143\\ = \frac{1}{2\sqrt{2 } } \biggl [ & ( e^{i 2\pi y c_s / p}+e^{i 2\pi \neg{y } c_s / p})|{\tilde + } _ s\rangle\\ & + ( e^{i 2\pi y c_s / p}-e^{i 2\pi \neg{y } c_s / p})|{\tilde -}_s\rangle \biggr ] .
\end{split}\ ] ] first consider the case ( a ) .
+ ( a - i ) suppose that the outcome of a measurement by the @xmath62th controller is @xmath144
. then , the unnormalized phase factor owing to this measurement is @xmath145 which is a global phase uncorrelated with @xmath124 .
thus bob does not have to modify it .
+ ( a - ii ) suppose that the outcome of a measurement by the @xmath62th controller is @xmath146 .
then , the unnormalized phase factor owing to this measurement is @xmath147 which can be regarded as @xmath148 with @xmath149 .
thus bob can modify the factor by applying @xmath52 to some @xmath33 for which the controller s qubit is originally connected in fig .
[ fig2 ] . second consider the case ( d ) .
+ the phase factor due to the measurement , with the outcome @xmath142 , is @xmath150 $ ]
. this can be modified by applying @xmath151\ ] ] to a proper @xmath33 , which is possible if bob knows @xmath105 , namely , if @xmath8 or more controllers follow ( a ) .
in addition to the above recovery of the phase factors corresponding to controllers measurements , bob applies @xmath152 to each @xmath153 , as usual , to recover the phase factors corresponding to alice s bell measurements .
let us estimate the reduction in the number of operations that bob has to apply to @xmath33 s for eliminating the phase factors due to controllers measurements .
let us consider the worst case where the @xmath5 controllers are assigned to mutually different channels . in our previous protocol ,
the number of the operations is @xmath5 . in the present protocol , it is , on average , @xmath154 with @xmath155 the number of controllers who agree to allow bob to recover the original state .
( of course , bob can not recover the state when @xmath156 . )
in the previous section , a quantum ( or hybrid ) @xmath0-threshold controlled quantum teleportation using qubits ( without qudits whose dimension is more than two ) has been constructed .
we will count the number of single - qubit operations and that of two - qubit operations by following the whole process . here
, we consider the case where the controllers are attached to the @xmath15th epr channel @xmath19 and regard bob s recovery operation corresponding to each controller s measurement as a single operation , for simplicity .
the number of operations is unchanged by employing the setup illustrated in fig .
[ fig2 ] .
the process is the same as the original quantum teleportation @xcite except for the measurements and operations acting on the shared state of @xmath157 , as we have seen in sec.[sec2 ] . for the part without controllers in fig .
[ fig1 ] , there are ( @xmath14 ) epr pairs prepared between alice and bob .
the quantum circuit for preparing the epr states involves ( @xmath14 ) hadamard gates and the same number of controlled - not ( cnot ) gates .
alice makes ( @xmath14 ) bell measurements between the system @xmath158 and the system @xmath159 and sends the outcomes ( @xmath160 ) as classical information to bob .
bob applies @xmath161 single - qubit operations at most , namely , @xmath31 , according to the classical information received from alice . for the portion of the shared state in the figure , the initial state @xmath162 given by eq .
( [ econkappa ] ) should be prepared .
it is prepared as follows .
first we produce the state @xmath163 this is easily prepared by a single hadamard gate and a cnot gate acting on @xmath19 .
the state @xmath164 is a basis vector in the @xmath62th controller s favorite basis .
second , @xmath5 hadamard gates are applied to @xmath165 individually in their bases .
in addition , the operation @xmath166\ ] ] is applied to @xmath167 for all @xmath47 in the basis @xmath168 .
the desired initial state for @xmath157 is now achieved . in the teleportation stage
, alice makes a bell measurement on @xmath70 and sends bob the outcome ( @xmath169 ) .
the controllers and bob follow the protocol as described in sec .
[ seceqcqt ] or that in sec .
[ secebo ] . in this process
, bob can eliminate the phase factors due to controllers measurements if @xmath8 or more controllers agree to allow bob to obtain the original state .
finally , bob applies @xmath51 to @xmath38 .
after all these steps , he obtains the original state of @xmath170 in @xmath171 . with the above description of the process , we find that the number of single - qubit operations and that of two - qubit operations for preparing the initial state of the whole system are @xmath172 for both operations .
the teleportation process involves @xmath15 bell measurements performed by alice and @xmath5 single - qubit measurements performed by controllers .
it also involves bob s recovery operations : ( i ) at most @xmath173 single - qubit operations corresponding to alice s measurement outcomes ; ( ii ) @xmath5 single qubit operations [ on average , ( @xmath174 ) single - qubit operations ] corresponding to controllers measurement outcomes when the protocol described in sec .
[ seceqcqt ] is employed [ when that in sec . [ secebo ]
is employed ] .
in addition , as we have mentioned in sec .
[ secebo ] , the operations of ( ii ) reduce to a single operation in reality for the present setup while it does not for the setup of fig .
we have proposed an economical scheme for a @xmath0-threshold control of a quantum teleportation .
it uses a shared state of alice , bob , and controllers with the controllers portion in the state of eq.([econkappa ] ) , which consists of qubits only .
thus a drawback of a usual polynomial coding , namely , the required dimension @xmath175 for each controller s qudit , has been resolved .
in addition , it is straightforward to extend the scheme so that multiple qubits in bob s portion are under the threshold control .
our economical scheme can be seen as a hybrid of a standard @xmath13-threshold controlled quantum teleportation and a @xmath0-threshold classically - controlled quantum teleportation , as we have mentioned in sec .
[ seceqcqt ] .
it should be noted that this has been realized by a nontrivial protocol using the state ( [ econkappa ] ) .
the scheme has a good redundancy for the securement of the @xmath0 threshold : ( i ) to modify the phase factors owing to controllers measurements , @xmath8 or more classical keys are required for bob .
( ii ) to make disagreeing controllers perform measurements , at most ( @xmath131 ) solicits should be sent from bob .
in fact , the standard controlled teleportation in the context of @xmath13 threshold can be used in the context of @xmath0 threshold by stating only ( ii ) in its protocol .
the advantage of our economical scheme over this simple extension is , thus , the redundant securement .
the scheme is , however , not as economical as a classically - controlled quantum teleportation , as we have discussed in sec .
[ secccqt ] .
a quantum threshold - control scheme is certainly more expensive than a classical threshold - control scheme .
it is thus recommended to assess the trade - off between the benefit and the economicalness to choose an appropriate scheme .
the benefit to distribute qudits ( qubits in our scheme ) among controllers is to make the recovery operation of bob unfixed unless controllers make measurements .
this makes the protocol robust against bob s attack to the keys : in order for cheating , he needs both physical accesses to at least @xmath8 controllers qudits and information on their measurement bases .
one may however claim that careless controllers tend to lose both of them at once in a real world .
our economical scheme possesses a clearer advantage also : it can prevent a violation of a voting schedule , i.e. it can prevent bob from recovering the original state before an official voting time , unless controllers violate the schedule all together .
there is one drawback in our protocol . in case
we use the scheme of sec .
[ sec3 ] , controllers who disagree to the teleportation do not have to measure their systems .
in contrast , in our economical protocol , a disagreeing controller has to measure her / his qubit if a solicit is sent from bob .
our protocol can be broken by a controller who does not follow this regulation .
it seems an important drawback at a glance , but there is a quick solution : one can easily find out which controller cheats in the protocol .
any controller who does not send a measurement outcome despite a solicit sent from noncheating bob is a cheater .
a clever cheater , however , may report an opposite measurement outcome and/or a wrong key instead of being quiet .
it has been well - known that a basic secret sharing is not robust against dishonest participants who report wrong keys .
there have been several proposals to remedy this drawback in classical secret sharing schemes @xcite .
these classical schemes are easily combined into our scheme in order to find cheating in the keys @xmath105 , as is clear from the protocol . nevertheless , a protocol to find cheaters sending wrong messages as measurement outcomes should be newly constructed .
one way to achieve this task is to add a supercontroller who grasps controllers states by entangling her / his systems and their corresponding systems .
let the supercontroller know their measurement bases .
then the supercontroller can check the measurement outcomes afterward .
it is hoped that a more sophisticated way will be developed .
finally , we discuss a well - known strategy to use controlled quantum teleportation as a secret sharing to hide a quantum state as a secret @xcite .
any ( @xmath176)-threshold controlled quantum teleportation in the stage after alice s bell measurements is regarded as a ( @xmath176)-threshold quantum secret sharing : alice s original state to be recovered in bob s side is regarded as a secret and the controllers are regarded as participants sharing the secret .
a drawback of this approach is that the original state is recovered in the system of bob s side ; participants who try to cooperate for recovery should gather at this side or ask a dealer for proxy . a solution to avoid this drawback is to construct @xmath5 initially identical copies of the entire system @xmath177 for the ( @xmath176)-threshold controlled quantum teleportation , where @xmath178 . let us write a copy as @xmath179 with @xmath180 . we send the system @xmath181 , together with classical information obtained by the bell measurements on each of @xmath182 , to the @xmath183th participant ( @xmath184 ) .
the @xmath183th participant should also receive @xmath185 , namely the @xmath183th control systems of all @xmath186 s .
the @xmath183th participant can recover the original state in the system @xmath181 when she / he gets to know the measurement results on @xmath8 or more among @xmath187 , including her / his own , and corresponding classical keys if they are used in the scheme .
this approach possesses the following benefit .
the original quantum secret sharing is limited to @xmath188 due to the no - cloning theorem when an unknown quantum state is a secret @xcite . a controlled quantum teleportation with a ( @xmath176)-threshold control is not limited by the no - cloning theorem because the secret , namely , alice s original state , is teleported to bob s system .
the approach is practical when we use classical keys for controlling a quantum teleportation .
the resource required for this case is @xmath5 classical keys and @xmath5 copies of the system that consists of @xmath15 epr pairs and the @xmath15-qubit original state .
of course , the @xmath5 copies of the system are reduced to one copy in case participants may gather in a particular place or may use a trusted dealer for proxy .
there seems to be no serious drawback of using classical control because the classical keys can be securely distributed and a scheduled vote is not interposed usually for a secret sharing .
the robustness of our economical scheme is effective when an untrusted bob exists .
this is due to the fact that the measurement basis of each controller can be hidden . in this sense
, our economical scheme might be used for a secret sharing to build in robustness against physical access attacks by malicious participants who try to cheat . nevertheless , it is not attractive to spend many epr pairs despite the reduction in the resource for controllers portion .
a choice of a proper scheme is dependent on the demand of participants when a controlled quantum teleportation is applied to a secret sharing .
we have discussed the advantage and disadvantage of our scheme in which the dimension of each controller s qudit is reduced to two .
this reduction is indeed significant for physical realization of the threshold control of a quantum teleportation . nevertheless , it is controversial as to which extent a controlled quantum teleportation should be performed with quantum resources .
the answer depends on the application and as to which party is trusted , as is clear from the above discussions .
we have proposed an economical protocol for @xmath0-threshold controlled quantum teleportation .
this protocol uses qubits distributed to controllers ; hence we have achieved the reduction in the dimension of each qudit from a prime @xmath2 to two .
in addition , we have shown an economization in the number of bob s recovery operations .
a.s . and m.n .
are supported by `` open research center '' project for private universities : matching fund subsidy from mext .
r.r . is supported by the grant - in - aid from jsps ( grant no .
1907329 ) .
m.n . would like to thank a partial support of the grant - in - aid for scientific research from jsps ( grant no .
19540422 ) .
99 c. h. bennett , g. brassard , c. crpeau , r. jozsa , a. peres , and w. k. wootters , `` teleporting an unknown quantum state via dual classical and einstein - podolsky - rosen channels '' , phys . rev .
lett . , vol.70 , pp.1895 - 1899 , 1993
. j. gruska , quantum computing , mcgraw - hill , london , 1999 .
m. a. nielsen and i. l. chuang , quantum computation and quantum information , cambridge university press , cambridge , 2000 .
j. zhou , g. hou , s. wu , and y. zhang , `` controlled quantum teleportation '' , lanl arxiv : quant - ph/0006030 . n. b. an , `` teleportation of coherent - state superpositions within a network '' phys . rev . a , vol.68 , pp.022321 - 1 - 6 , 2003 .
chu , and s. han , `` efficient many - party controlled teleportation of multiqubit quantum information via entanglement '' , phys .
. a , vol.70 , pp.022329 - 1 - 8 , 2004 . f. g. deng , c .- y .
li , y .- s .
li , h .- y .
zhou , and y. wang , `` symmetric multiparty - controlled teleportation of an arbitrary two - particle entanglement '' , phys .
rev . a , vol.72 , pp.022338 - 1 - 8 , 2005 .
man , y .- j .
xia , and n. b. an , `` economical and feasible controlled teleportation of an arbitrary unknown @xmath189-qubit entangled state '' , j. phys .
b : at . mol .
vol.40 , pp.1767 - 1774 , 2007 .
d. kenigsberg and t. mor , `` secure controlled teleportation '' , lanl arxiv : quant - ph/0609028 . c. h. bennett , f. bessette , g. brassard , l. salvail , and j. smolin , `` experimental quantum cryptography '' , j. crypt . ,
vol.5 , pp.3 - 28 , 1992 .
lo , x. ma and k. chen , `` decoy state quantum key distribution '' , phys .
vol.94 , pp.230504 - 1 - 4 , 2005 . c. marand and p. townsend , `` quantum key distribution over distances as long as 30 km '' , opt .
vol.20 , pp.1695 - 1697 , 1995 .
a. muller , h. zbinden , and n. gisin , `` quantum cryptography over 23 km of installed under - lake telecom fibre '' , europhys .
lett . , vol.33 , pp.335 - 339 , 1996 . t. hasegawa , t. nishioka , h. ishizuka , j. abe , k. shimizu , m. matsui and s. takeuchi , `` an experimental realization of quantum cryptosystem '' , ieice trans .
vol.e85-a , no.1 , pp.149 - 157 , january 2002 .
d. stucki , n. gisin , o. guinnard , g. ribordy , and h. zbinden , `` quantum key distribution over 67 km with a plug&play system '' , new j. phys . , vol.4 , pp.41 - 1 - 8 , 2002 .
i. marcikic , a. lamas - linares , and c. kurtsiefer , `` free - space quantum key distribution with entangled photons '' , appl .
, vol.89 , pp.101122 - 1 - 3 , 2006 . j. f. dynes , z. l. yuan , a. w. sharpe , and a. j. shields , `` practical quantum key distribution over 60 hours at an optical fiber distance of 20 km using weak and vacuum decoy pulses for enhanced security '' , optics express , vol.15 , iss.13 , pp.8465 - 8471 , 2007 .
s. bose , v. vedral , and p. l. knight , `` multiparticle generalization of entanglement swapping '' , phys .
rev . a , vol.57 , pp.822 - 829 , 1998 .
d. bouwmeester , j .- w .
pan , k. mattle , m. eibl , h. weinfurter , and a. zeilinger , `` experimental quantum teleportation '' , nature , vol.390 , no.6660 , pp.575 - 579 , 1997 .
d. boschi , s. branca , f. de martini , l. hardy , and s. popescu , `` experimental realization of teleporting an unknown pure quantum state via dual classical and einstein - podolsky - rosen channels '' , phys
lett . , vol.80 , no.6 , pp.1121 - 1125 , 1998 . i. marcikic , h. de riedmatten , w. tittel , h. zbinden , and n. gisin , `` long - distance teleportation of qubits at telecommunication wavelengths '' , nature , vol.421 , no.6922 , pp.509 - 513 , 2003 . g. r. blakley , `` safeguarding cryptographic keys '' , in proceedings of afips 1979 national computer conference , new york , 1979 , vol.48 , pp.313 - 317 , afips press , arlington , va . ,
june 1979 .
a. shamir , `` how to share a secret '' , commun .
acm , vol.22 , no.11 , pp.612 - 613 , 1979 .
e. d. karnin , j. w. greene , and m. e. hellman , `` on secret sharing systems '' , ieee trans .
theory , vol.it-29 , no.1 , pp.35 - 41 , january 1983 .
s. c. kothari , `` generalized linear threshold scheme '' , in proceedings of crypto84 , santa barbara , 1984 , edited by g. r. blakley and d. chaum , springer - verlag , new york , 1985 , pp.231 - 241 .
r. cleve , d. gottesman , and h .- k .
lo , `` how to share a quantum secret '' , phys .
lett . , vol.83 , pp.648 - 651 , 1999 . z. zhang , `` controlled teleportation of an arbitrary @xmath15-qubit quantum information using quantum secret sharing of classical message '' , phys .
lett . a , vol.352 , pp.55 - 58 , 2006 . y. murakami , m. nakanishi , s. yamashita , and k. watanabe , `` cheater identifiable quantum secret sharing schemes '' , ipsj sig technical reports , 2005-csec-30(47 ) , pp.337 - 340 , 2005 . s. iftene , `` general secret sharing based on the chinese remainder theorem '' , cryptology eprint archive : 2006/166 , http://eprint.iacr.org/2006/166 .
h. kunii and m. tada , `` a note on information rate for fast threshold schemes '' , in proceedings of computer security symposium 2006 , kyoto , japan , october 2006 , pp.101 - 106
. j. kurihara , s. kiyomoto , k. fukushima , and t. tanaka , `` a fast ( 3,n)-threshold secret sharing scheme using exclusive - or operations '' , ieice trans .
fund . , vol.e91-a , no.1 , pp.127 - 138 , january 2008 .
guo and g .- c .
guo , `` quantum secret sharing without entanglement '' , phys .
a , vol.310 , pp.247 - 251 , 2003 .
zhang and z .- x .
man , `` many - agent controlled teleportation of multi - qubit quantum information '' , phys .
a , vol.341 , pp.55 - 59 , 2005 .
d. markham and b. c. sanders , `` graph states for quantum secret sharing '' , phys .
rev . a , vol.78 , pp.042309 - 1 - 17 , 2008 . c. negrevergne ,
t. s. mahesh , c. a. ryan , m. ditty , f. cyr - racine , w. power , n. boulant , t. havel , d. g. cory , and r. laflamme , `` benchmarking quantum control methods on a 12-qubit system '' , phys .
vol.96 , pp.170501 - 1 - 4 , 2006 .
lancaster and m. tismenetsky , the theory of matrices , 2nd ed . , academic press , san diego , 1985 .
s. karlin , total positivity , volume i , stanford university press , california , 1968 .
t. ando , `` totally positive matrices '' , linear algebra appl . , vol.90 , pp.165 - 219 , 1987 . t. craven and g. csordas , `` a sufficient condition for strict total positivity of a matrix '' , linear multilinear algebra , vol.45 , pp.19 - 34 , 1998 .
g. plya and g. szeg , problems and theorems in analysis , volume ii , springer - verlag , berlin , 1976 , 4th ed .
proceedings of the symposium `` quantum computation : are the divincenzo criteria fulfilled in 2004 ? '' , edited by m. nakahara , s. kanemitsu , m. m. salomaa , and s. takagi , published as `` physical realizations of quantum computing '' , world scientific , singapore , 2006 .
n. gisin , g. ribordy , w. tittel , and h. zbinden , `` quantum cryptography '' , rev .
vol.74 , no.1 , pp145 - 195 , 2002 . m. tompa and h. woll , `` how to share a secret with cheaters '' , j. crypt . , vol.1 , pp.133 - 138 , 1989 . j. rif - coma , `` how to avoid the cheaters succeeding in the key sharing scheme '' , designs , codes , and crypt .
, vol.3 , pp.221 - 228 , 1993 .
r. s. rees , d. r. stinson , r. wei , and g. h. j. van rees , `` an application of covering designs : determining the maximum consistent set of shares in a threshold scheme '' , ars combin .
vol.53 , pp.225 - 247 , 1999 .
r. tso , y. miao , and e. okamoto , `` a new algorithm for searching a consistent set of shares in a threshold scheme with cheaters '' , in proceedings of the 6th information security and cryptology conference , edited by j. i. lim and d. h. lee , lecture notes in computer science , vol.2971 , pp.377 - 385 , springer - verlag , berlin , 2003 . | we study a @xmath0-threshold controlling scheme for controlled quantum teleportation . a standard polynomial coding over @xmath1 with prime @xmath2 needs to distribute a @xmath3-dimensional qudit with @xmath4 to each controller for this purpose .
we propose a scheme using @xmath5 qubits ( two - dimensional qudits ) for the controllers portion , following a discussion on the benefit of a quantum control in comparison to a classical control of a quantum teleportation .
+ + keywords : quantum teleportation , threshold scheme , secret sharing + pacs : 03.67.hk , 03.67.mn |
environment impacts galaxy evolution . galaxies in groups and clusters are more likely to have suppressed ( ` quiescent ' ) star formation rates ( sfr ) and lie on the red sequence , have more elliptical / bulge - dominated morphologies , and have less atomic / molecular gas than galaxies of similar stellar mass in comparatively isolated ` field ' environments ( * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* for review ) .
while these differences are strongest for galaxies in massive clusters , they persist in groups at least down to virial masses of @xmath0 ( e.g. , * ? ? ?
* ; * ? ? ?
* ) , comparable to the milky way and m31 , whose satellites also show such environmental trends ( * ? ?
* for review ) .
several observational works have demonstrated that this environmental dependence is confined mostly to physical scales corresponding to the virial radius of a group / cluster and is driven by satellite galaxies , which are not the massive ` central ' galaxy at the core of a host halo ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
these results are physically motivated , given that a host halo s virial radius broadly corresponds to a physical transition from the low - density ` field ' environment to a high - density region where dark matter and gas are virialized . while galaxies within groups / clusters strongly exhibit the above environmental dependencies
, such trends can extend to galaxies beyond the virial radius of a group / cluster .
several authors have noted an enhanced quiescent / red fraction and reduced @xmath1 emission in galaxies out to @xmath2 virial radii around massive clusters @xcite , as well as for dwarf galaxies around lower mass groups ( y. @xcite ) , including those near the milky way and m31 @xcite .
the above observational trends are plausibly driven , at least in part , by these galaxies having passed within much smaller distances from a group / cluster .
indeed , @xcite first noted from _ n_-body simulations of clusters that particles that have passed within the virial radius can then orbit well outside of it .
satellite galaxies can do so as well , as investigated both through analytic arguments @xcite and in numerical simulations @xcite .
in general , these works found that these ` ejected ' ( or ` backsplash ' ) satellites typically orbit back out to a maximum distance of @xmath3 , such that the average interior density is @xmath4 times a reference density , @xmath5 : @xmath6 .
@xmath7 m means @xmath8 the average matter density of the universe , and @xmath7c means @xmath8 the critical density . ] beyond a host halo after passing through it , and that as many as half of all galaxies within this distance are composed of these ejected satellites , with higher fractions for less massive galaxies and around more massive host halos .
thus , ejected satellites are potentially critical for understanding the properties of galaxies near groups / clusters and obtaining a complete picture of environmental dependence . indeed
, several works suggest that ejected satellites are preferentially quiescent .
y. @xcite used the sdss group catalog of @xcite to examine the @xmath9 colors of faint ( @xmath10 ) central galaxies .
they found that the reddest population lies within @xmath11 of a nearby massive host halo and therefore argued that such central galaxies are likely ejected satellites .
relatedly , analyzing more massive galaxies near massive clusters also via the group catalog of @xcite , @xcite statistically compared the phase - space distribution ( in redshift space ) of galaxies near clusters with that of particles taken from a smooth particle hydrodynamics ( sph ) simulation , arguing that ejected satellites are quenched significantly relative to infalling galaxies and that star formation quenching occurs after a single pericentric passage .
more dramatically , @xcite found that all spectroscopically quiescent galaxies with @xmath12 lie within @xmath13 of a more massive galaxy , implying that _ all _ low - mass quiescent galaxies are either satellites or ejected satellites .
of a more massive halo , the results of @xcite , using spectroscopic measures of star formation , imply that the population of isolated , red galaxies in @xcite arises from dust reddening a significant effect for low - mass galaxies ( e.g. , * ? ? ?
* ) and not true star formation quenching . ]
indeed , @xcite argued that satellite ejection likely explains the presence of several dwarf galaxies that are beyond the virial radius of the milky way and have depleted @xmath1 masses and old stellar populations , based on these galaxies observed phase - space coordinates .
finally , @xcite and @xcite examined extended environmental processes on galaxies in sph simulations ; while they found stripping of extended gas ( beyond the disk ) in the halos of infalling galaxies from ram - pressure in filaments prior to virial crossing , the effect on star formation before virial crossing was negligible for galaxies with @xmath14 ( the limit in this paper ) . in this paper
, we explore observationally and theoretically the evolution of star formation in galaxies near ( @xmath15 ) groups and clusters , to understand in detail the physical extent of the environmental dependence of galaxy evolution and the importance of ejected satellites .
we use a galaxy group / cluster catalog constructed from the sloan digital sky survey ( sdss ) together with a cosmological _ n_-body simulation , to which we apply the same group - finding algorithm for robust comparison of model results to observations .
this combination allows us to pursue a more rigorous and statistically significant investigation , including accounting for the importance of correlated structure ( neighboring halos ) .
we test the importance of star formation quenching in ejected satellites by employing the same empirically motivated and calibrated model for satellite sfr evolution that we developed in @xcite . to summarize our main result
, we find that the quenching of star formation in ejected satellites is consistent with proceeding in the same manner as for satellites that remain within their host halo , and therefore , ejected satellites can account for essentially all of the observed environmental dependence of galaxy quenching beyond the virial radius of massive host halos .
this paper represents the fourth in a series of five . in @xcite ,
hereafter , we described our sdss galaxy sample , presented our method for observationally identifying galaxy groups / clusters , and showed that galaxy quiescent fractions are essentially independent of large - scale environment beyond their host halo . in @xcite ,
hereafter , we used our sdss group catalog to examine in detail the sfr distribution of satellite galaxies and its dependence on stellar mass , host halo mass , and halo - centric distance , finding that the sfr distribution is bimodal in all regimes . in @xcite ,
hereafter , we combined the above results with a cosmological _ n_-body simulation to constrain the star formation histories and quenching timescales of satellites , showing that their star formation must evolve in the same manner as central galaxies for @xmath16 ( depending on stellar mass ) after falling into a host halo , but that once quenching starts , it occurs rapidly , with a sfr fading e - folding time of @xmath17 . finally , in @xcite , hereafter , we will use the detailed orbital histories from our simulation to test physical mechanisms responsible for satellite quenching . before proceeding , we define some nomenclature .
we refer to galaxies with low sfr ( see ) as being ` quiescent ' , and ` quenching ' is the physical process of sfr fading rapidly from actively star - forming to quiescent , under the ansatz that once a satellite is quenched it remains so indefinitely . for our galaxy group catalog ,
` group ' means any set of galaxies that occupy a single host halo , regardless of its virial mass , and we use ` host halo ' as a more general term for group or cluster . finally , we cite all masses using @xmath18 for the hubble parameter .
here , we briefly describe our observational galaxy sample and cosmological simulation . for full details , see papers i and ii for the former and for the latter .
our galaxy sample is based on the new york university value - added galaxy catalog @xcite from data release 7 of sdss @xcite .
stellar masses are from the kcorrect code of @xcite , assuming a @xcite initial mass function ( imf ) .
we construct two volume - limited samples of galaxies with @xmath19 and @xmath20 , which go out to @xmath21 and 0.06 , from which we identify stellar mass complete limits of @xmath22 and @xmath23 , respectively .
we use specific star formation rates , @xmath24 , based on the current release of the spectral reductions of @xcite , in which @xmath25 is a galaxy s star formation rate and @xmath26 its stellar mass .
we define galaxies with @xmath27 as ` quiescent ' , based on the bimodal nature of the ssfr distribution ( see ) .
we identify groups of galaxies that occupy the same host halo using a modified implementation of the group - finding algorithm of @xcite . for our group catalog ,
we define host halos as spherical overdensities containing 200 times the average matter density of the universe ( @xmath7 m ) .
as outlined , we assign host halo virial masses , @xmath28 , to groups by matching the abundance of halos above a given dark matter mass to the abundance of groups above a given total stellar mass , @xmath29 , using the host halo mass function from @xcite based on a flat , @xmath30cdm cosmology of @xmath31 , @xmath32 , @xmath18 , @xmath33 and @xmath34 .
the center of a group is given by its most massive galaxy , which we call the ` central ' galaxy .
every group contains one central galaxy and can contain any number ( including zero ) of less massive ` satellite ' galaxies . to track the evolution of satellites from first infall to final merging / disruption across a broad range of host halo masses , we employ a dissipationless , @xmath35-body simulation , using the treepm code of @xcite with flat , @xmath30cdm cosmology of @xmath36 , @xmath37 , @xmath18 , @xmath38 and @xmath39 , evolving @xmath40 particles of mass @xmath41 with a plummer equivalent smoothing of @xmath42 in a @xmath43 box , and storing 45 outputs spaced evenly in @xmath44 from @xmath45 to 0 .
this simulation first was presented in @xcite . in the simulation , we identify ` host halos ' using the friends - of - friends ( fof ) algorithm @xcite with a linking length of @xmath46 times the average inter - particle spacing , which groups particles bounded by an isodensity contour of @xmath47 the average matter density . within host halos ,
we identify ` subhalos ' as overdensities in phase space through a 6-dimensional fof algorithm ( fof6d ) . for both host halos and subhalos ,
we keep all objects with at least 50 particles , and we define the center and velocity via the most bound particle .
we assign to each ( sub)halo a ` child ' ( sub)halo at the next simulation output , based on its 20 most bound particles .
we define a ` central ' subhalo as being the most massive subhalo in a newly formed host halo , which almost always corresponds to the subhalo at the minimum of the potential well .
a subhalo retains its central demarcation until it falls into ( is linked by the fof algorithm to ) a more massive host halo , becoming a ` satellite ' subhalo . thus , every sufficiently bound halo hosts one central subhalo and can host any number of satellite subhalos .
we assign to each subhalo a maximum mass , @xmath48 , as given by the maximum host halo mass , @xmath28 , that it ever had as a central subhalo , motivated by the expected correlation of this quantity with galaxy stellar mass ( see below ) . for a central subhalo
, @xmath48 almost always corresponds to its current halo mass , the primary exception being ejected satellites ( see below ) . for a satellite
, @xmath48 almost always corresponds to its halo mass shortly before first infall .
also , as demonstrated in @xcite , simulations at our resolution scale can resolve and track massive satellite subhalos past the point at which the galactic stellar component that they host would ( start to ) be stripped , merge with the central galaxy , or otherwise be disrupted , and we use their scheme by removing satellites with @xmath49 , in which @xmath50 is the bound subhalo mass as determined by the fof6d subhalo finder . finally , to ensure that the simulation s output time spacing of @xmath51 at @xmath52 does not bias measurements of satellite orbital properties , such as pericentric distance , we measure the orbital distance and velocity for each satellite in a continuous manner by numerically integrating its orbit between outputs , assuming energy and angular momentum conservation in a spherical nfw @xcite potential given by the host halo s mass and concentration .
we define ` first infall ' of a subhalo as when it first becomes a satellite in a more massive host halo , if it remains a satellite for at least two consecutive outputs .
the latter criterion avoids cases of temporary bridging if a subhalo briefly passes through the outskirts of a host halo . after experiencing first infall ,
if a satellite orbits outside of its host halo , becoming ( again ) a central subhalo in a distinct halo , we consider it to be an ` ejected satellite ' .
thus , an ejected satellite appears to be a central subhalo in an instantaneous sense , though as we will demonstrate , it is more appropriate to continue to consider it a satellite .
finally , if an ejected satellite grows in halo mass by @xmath53 after ejection , we define it to be a newly formed halo and discard it from the ejected population , though as we will show , this criterion affects @xmath54 of ejected satellites , regardless of satellite mass . under the ansatz that a galaxy resides at the center of each dark matter subhalo , we assign stellar mass using subhalo abundance matching ( sham ; * ? ? ?
* ; * ? ? ?
* ) , a method that assumes a one - to - one mapping that preserves rank ordering between subhalo @xmath48 and galaxy @xmath26 , such that @xmath55 . here
, one assigns @xmath26 to subhalos empirically through an observed stellar mass function ( smf ) that is recovered , by design .
sham has succeeded in reproducing many observed galaxy statistics , including spatial clustering , satellite fractions , cluster luminosity functions , and luminosity - velocity relations ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
we use the smf from @xcite , based on the same sdss nyu - vagc sample as our catalog , including the same @xmath56-correction and imf .
we apply sham to the simulation at @xmath57 , implementing a log - normal scatter of 0.15 dex in @xmath26 at fixed @xmath48 as suggested by many observations ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) . to make robust comparisons with the sdss group catalog , we produce a ` simulation mock group catalog ' by applying the same group - finding algorithm to the simulation .
while we base our models of sfr evolution on true satellite / central demarcation in the simulation , we effectively ` observe ' our model results through this group catalog , using the same host halo definition and including observational effects , such as redshift - space distortions , that affect group purity / completeness as well as satellite / central assignment .
we first examine observationally the sfrs of galaxies in and near cluster - mass host halos .
we improve our analysis from , in which we examined all galaxies regardless of their redshift at a projected distance from a host halo , by examining only galaxies that lie in close proximity to a given host halo via applying a line - of - sight velocity cut of @xmath58 , in which @xmath59 is the virial velocity .
( for reference , @xmath60 . ) shows the fraction of galaxies that are quiescent as a function of projected halo - centric distance , @xmath61 , in two different bins of stellar mass in / around cluster - mass host halos .
curves show the average quiescent fraction , stacking all host halos in the given range of @xmath62 , and error bars show 68% confidence interval on the average , as given by a beta distribution @xcite .
considering all galaxies , regardless of type ( solid black curves ) , those at @xmath63 have the highest likelihood of being quiescent , but a clear enhancement , beyond the cosmic average of all galaxies in the stellar mass bin ( dotted lines ) , persists out to @xmath64 ( see also * ? ? ?
* ; * ? ? ?
what causes this enhancement out to such large distances , well beyond the turn - around ( maximum ) distance from which matter accretes onto these clusters ? to address this , we use our group catalog to decompose these galaxies physically by type : satellite ( solid red curves ) and central ( solid blue curves ) .
( note that some central galaxies appear within @xmath65 as viewed in projection . )
this decomposition reveals two trends that cause the enhanced quiescent fraction for all galaxies out to @xmath66 .
first , structure is correlated : massive host halos are more likely to be found near each other .
is @xmath67 , 1.4 , 0.6 , respectively . in real space ,
as measured in our simulation , the median @xmath68 values are roughly twice as large . ]
thus , the likelihood that a galaxy is in another massive host halo decreases with distance from a cluster , so the fraction of galaxies that are satellites , and the average host halo mass of those that are , decreases with distance .
because the quiescent fraction is higher for satellites than central galaxies , and for satellites it is higher in more massive host halos ( ) , correlated structure can cause a gradient in the quiescent fraction with distance even if there were no gradients for central and satellite galaxies separately .
however , also shows another , more interesting trend : central galaxies , which reside in their own host halo , exhibit a strongly enhanced quiescent likelihood out to @xmath69 .
while this enhancement is strongest around cluster - mass host halos , central galaxies near lower mass host halos show similar behavior , and a non - zero enhancement beyond the cosmic average for central galaxies extends out to @xmath70 in essentially all cases ( see in ) . what causes this enhanced quiescent fraction in central galaxies near massive host halos ?
it potentially could be an artifact of interloping galaxies caused by redshift - space distortions or some other aspect of the group - finding algorithm .
however , as we will show using our simulation mock group catalog in , such observational effects can not account for this strong excess ( in agreement with y. @xcite ) .
this leaves two feasible physical causes : 1 .
strong environmental processes might extend well beyond the ( spherical ) virial radius of host halos , quenching central galaxies _ prior _ to their first crossing inside .
2 . the observed enhancement could originate from ` ordinary ' environmental quenching within a host halo s virial radius and propagate to larger distances via ` ejected ' satellites , which orbit beyond the virial radius after falling inside , becoming ( once again ) central galaxies in a distinct halo . given its simplicity
, we postulate that ( b ) most naturally explains the observed trends , as we will subsequently examine and test
. we will discuss ( a ) in .
in , we showed that a significant fraction of satellites inside a host halo spent time in an ejected phase beyond the virial radius before falling back in . in this section , we investigate in more detail the orbital histories of ejected satellites , building on previous work but focusing on orbital properties that are ( potentially ) relevant to quenching star formation .
while we examine galaxies and host halos in the same mass ranges as in our sdss catalog , in this section we use the halo catalog directly from our simulation ( not from the mock group catalog ) and measure halo - centric distance in real - space , @xmath71 , in order to highlight physical trends without projection effects or redshift - space distortions .
we first examine the importance and radial extent of ejected satellites around massive host halos
. shows what fraction of galaxies at a given ( real - space ) distance around host halos are ejected .
we show the average value from stacking galaxies around host halos of the given virial mass , and shaded regions shows the 68% confidence interval on this average .
values at @xmath72 show what fraction of all central galaxies in the distance bin are ejected satellites , @xmath73 . for reference and continuity , also shows values at @xmath74 , which indicate what fraction of all satellites experienced an ejected phase and fell back in , @xmath75 .
the solid curves show satellites that were ejected specifically from the host halos on which these profiles are centered . across all galaxy and host halo masses ,
a non - trivial population of ejected satellites extends to @xmath69 beyond their host halo , in good agreement with previous simulation results @xcite .
this maximum distance is approximately as expected based on the turn - around distance , @xmath76 , of infalling matter , if it conserves energy as it orbits back to apocenter , @xmath77 , after infall @xcite .
we do find that a small fraction ( @xmath78 ) of ejected satellites extend beyond @xmath79 , likely caused by an orbital energy boost during a multi - body encounter within the host halo @xcite .
however , this phenomenon primarily affects satellites with particularly low mass that can be boosted more easily @xcite .
the majority ( 68% ) of satellites in our mass range have @xmath80 , and the vast majority ( 90% ) have @xmath81 , implying that any such orbital energy boosts are mild for satellites that we examine .
by contrast , the dashed curves in show the ejected fraction _ regardless _ of which host halo the satellites passed through .
thus , it also includes satellites that were ejected from _ neighboring _ host halos . in examining this _ total _ fraction , a strong excess still persists out to @xmath82 , but it declines only gradually before reaching the cosmic average ( given in ) at @xmath83 .
thus , while ejected satellites are most common near massive host halos , they are cosmically ubiquitous . because separating ejected satellites based on which host halo they passed through observationally is highly non - trivial , if not impossible , one _ must _ consider this total population when comparing to observations . shows more directly how the fraction of central galaxies that are ejected satellites varies with galaxy stellar mass .
first , the solid curve shows what fraction of _ all _ central galaxies are ejected satellites .
this fraction increases at lower stellar mass , though it is always @xmath84 down to @xmath85 ( in agreement with * ? ? ?
thus , ejected satellites are of modest importance to the overall population of central galaxies . however , as showed , ejected satellites are particularly common near massive host halos , and the dashed curves in show the ejected fraction for central galaxies that are within @xmath86 of host halos of the given virial mass .
this ejected fraction strongly increases with host halo mass and decreases with galaxy mass .
most likely , these mass dependencies arise from the increased efficiency of dynamical friction ` braking ' of orbital velocity when the ratio of satellite to host halo mass is smaller @xcite .
note that 40% of all central galaxies with @xmath87 near cluster - mass host halos are ejected satellites .
we next examine how far into their host halo these ejected satellites orbited , in order to understand the feasibility of their star formation being affected .
we measure a satellite s @xmath68 at each simulation output and identify the minimum ratio experienced since first infall , using orbital integration as given in .
shows the cumulative distribution of minimum @xmath68 to which different satellite populations at @xmath88 have orbited .
solid curves show ejected satellites , which currently are beyond @xmath65 , while for reference , dashed curves show all satellites that currently are within @xmath65 . across the entire sample ,
the median pericentric passage of ejected satellites is @xmath89 , sufficiently deep into the halo s potential well to make it feasible that the majority have experienced quenching process(es ) . as shows ,
the median for all satellites within the virial radius is @xmath90 , so by comparison , ejected satellites have not orbited especially deeply .
though , it may be unfair to compare ejected satellites to all satellites , given that almost half of all satellites _ currently _ lie within @xmath91 .
if instead we compare the pericenter distribution of ejected satellites to those currently at @xmath92 , they are almost identical ( not shown ) .
thus , the majority of ejected satellites are not on orbits with particularly unusual pericenters , they simply have sufficient orbital energy to bring them back beyond the virial radius . also shows that satellites in more massive host halos typically orbited to somewhat smaller @xmath68 , with the median value for ejected satellites varying from @xmath93 across our host halo mass range .
we find no dependence of any of these distributions on satellite mass .
finally , the pericentric distances experienced by ejected satellites in are almost always ( @xmath94 ) the result of a _ single _ pericentric passage between first infall and ejection , also noted previously @xcite .
the @xmath95 of ejected satellites that experienced multiple pericentric passages typically are on highly radial orbits and experienced several passages in / out of the host halo .
our primary goal is to test whether ejected satellites experience the same quenching of star formation as ` normal ' satellites that remain within a host halo .
as showed , sfr in normal satellites evolves unaffected for @xmath16 after first infall , depending on stellar mass , after which it is quenched rapidly , with an e - folding time of @xmath96 .
thus , we now examine the orbital timescales of ejected satellites in the context of these quenching times . shows the distribution of time since first infall ( dashed ) and time since ejection ( solid ) for ejected satellites at @xmath88 .
we find little dependence of these orbital times on the virial mass of the host halo that they passed through , as shown , and no dependence on galaxy mass ( not shown ) . across all ejected satellites , the median time since ejection is @xmath97 and since first infall is @xmath98 .
the distribution of time since ejection falls off rapidly , meaning that relatively few satellites are on ( nearly ) unbound orbits that keep them beyond the virial radius for a long time . instead
, almost all ejected satellites remain bound to their host halo . for ejected satellites that are crossing back inside the virial radius at @xmath88 , we find that the median time spent in an ejected phase is @xmath99 .
we also find that the median amount of time that ejected satellites spent within their host halo is @xmath100 across our sample , again with no significant dependence on galaxy or host halo mass .
this time - as - satellite is similar to the virial crossing time ( @xmath101 ) at the typical redshift when these satellites fell in ( @xmath102 ) , as expected given that the vast majority experienced a single pericentric passage , and in good agreement with @xcite .
thus , compared to the quenching delay timescales of satellites from , the typical time - as - satellite is similar , and the typical time since first infall is considerably longer . as such , most ejected satellites
easily have the potential to be quenched in the same way as normal satellites . having shown that most ejected satellites orbited relatively deeply into their host halo and have been satellites for a sufficiently long time to be ( at least potentially ) quenched , we next explore the evolution of their ( sub)halo mass after ejection . if an ejected satellite s ( sub)halo starts to accrete mass again after ejection , in the same way as non - ejected halos , this accretion feasibly could lead to renewed star formation , even if quenching already occurred . alternately , if ejected satellites typically do not accrete mass , then they are likely to remain quenched in the same way as normal satellites .
to demonstrate the evolution of halo mass after ejection , shows the cumulative distribution of the fraction of halo mass that an ejected satellite has at @xmath88 relative to what it had immediately after ejection .
different curves show satellites ejected from host halos of different virial mass . across all host halos , only @xmath78 of ejected satellites gained halo mass since ejection . instead , ejected satellites almost always continue to lose mass , both from tidal stripping in the dynamically hot environment surrounding a massive host halo ( e.g. , * ? ? ?
* ) and from the delayed response of particles from more impulsive unbinding closer to pericenter .
shows that stronger stripping occurs near more massive host halos , with the typical mass loss since ejection being 40% near clusters ( in good agreement with * ? ? ?
* ; * ? ? ?
* ) , but only 10% near @xmath103 . of the total ( sub)halo mass loss since first infall .
] we find no significant dependence on satellite mass .
this continued halo mass loss strongly suggest that the vast majority of ejected satellites do not restart star formation from renewed gas accretion after ejection , but rather continue to be affected and quenched after ejection in the same way as satellites within a host halo .
these results also have important implications for the stellar - to - halo mass relation of ejected satellites , as we will explore in .
to summarize this section , the orbits of ejected satellites extend to @xmath69 beyond their host halo , though because of neighboring host halos and correlated structure , an enhanced total population extends out to @xmath83 .
lower mass satellites and those associated with more massive host halos are more likely to experience an ejected phase .
after first infall , these satellites almost always experience a single pericentric passage before ejection , typically to @xmath89 , which is similar to satellites that are just within @xmath65 .
thus , the typical time that ejected satellites spent within the host halo , @xmath104 , is approximately the virial crossing time , and almost all remain on bound orbits and re - enter the host halo after @xmath105 . for satellites that currently are ejected ,
their typical time since first becoming a satellite is @xmath106 , longer than the @xmath16 quenching delay time of satellites , and most ( 90% ) continue to lose halo mass after ejection .
these results strongly suggest that ejected satellites quench and remain quenched in the same way as satellites inside a host halo , which we next test directly .
in , we postulated that ejected satellites could explain the observed enhancement in the quiescent fraction for central galaxies near massive host halos .
we now test this scenario directly by examining two contrasting models for the evolution of sfr in ejected satellites : 1 . ` satellite - like ' evolution , in which the sfrs of ejected satellites are quenched in the same way as normal satellites , 2 . ` central - like ' evolution ,
in which the sfrs of ejected satellites evolve in the same way as normal ( non - ejected ) central galaxies .
to test ` satellite - like ' evolution , we apply the same physically motivated , two - stage model for the evolution of sfr in satellites from . to summarize , we start by obtaining statistically accurate initial sfrs for satellites at their time of first infall via an empirically based parametrization for the evolution of central galaxy sfrs out to @xmath107 , combining our sdss group catalog with data at higher @xmath108 from the cosmic evolution survey ( cosmos ) @xcite and the all - wavelength extended groth strip international survey ( aegis ) @xcite .
if a satellite was actively star - forming at its time of first infall , @xmath109 , obtained via the simulation , we parametrize its subsequent evolution as follows .
its sfr fades gradually , in the same manner as central galaxies , across a ` delay ' time , @xmath110 .
after this , it starts to be quenched , and its sfr fades exponentially across an e - folding time , @xmath111 : @xmath112 in which @xmath113 , and @xmath114 with @xmath115 being the time of initial formation , which we take to be @xmath116 .
we obtain @xmath117 and @xmath118 in narrow bins of stellar mass using the measured @xmath26 and @xmath25 of actively star forming central galaxies in the sdss group catalog ( assuming 40% stellar mass loss through supernovae and stellar winds by @xmath88 ) .
as showed , this empirically constrained model provides accurate sfr distributions and quiescent fractions for satellites at @xmath88 across both stellar and host halo mass , with @xmath110 and @xmath111 that depend on satellite mass but not significantly on host halo mass .
( as will show , this model also reproduces accurately the dependence of sfr versus halo - centric distance within host halos . )
we use the same quenching timescales as : @xmath119 and @xmath120 for @xmath121 , respectively .. using our ` alternate ' parametrization , along with the associated quenching timescales , does not change any results in this paper significantly . ] for a contrasting null model , we consider ` central - like ' evolution , in which ejected satellites have the same instantaneous distribution of sfrs as central galaxies .
physically , this scenario implies either that sfr in ejected satellites is not affected environmentally , so it evolves in the same way as in normal central galaxies , or that sfr in ejected satellites is quenched environmentally , but it resumes to being the same as in normal central galaxies after ejection . to implement this model ,
we proceed as above , but we re - assign sfrs to ejected satellites empirically by drawing randomly from ` isolated ' central galaxies of the same stellar mass in the sdss group catalog that do not lie within @xmath122 in projection of a more massive host halo , to minimize any possible contamination from ejected satellites .
the two models above provide sfrs at @xmath88 for all satellites in the simulation .
we assign sfrs to central galaxies by drawing randomly from isolated central galaxies of the same stellar mass in the sdss group catalog .
thus , we assign sfr to galaxies in the simulation according to their physical satellite versus central demarcation , but we then measure the results using the mock group catalog to compare robustly with sdss . shows the quiescent fraction for central - identified galaxies as a function of projected distance from host halos of the given mass in each panel , comparing the results of the two models above against the sdss group catalog . as in , points show the average observed value in the distance bin from sdss , and dotted lines show the average for all central galaxies of the given stellar mass in sdss .
note that , as compared with , extends to host halos of much lower mass , where the upturn near @xmath65 is weaker , but the excess beyond the cosmic average remains strong out to @xmath123 .
each panel shows two bins of galaxy stellar mass ; we find that galaxies of higher mass show less enhancement , as expected from .
curves show the two model results as measured through the simulation mock group catalog .
dashed curves show ` central - type ' evolution for ejected satellites , in which _ all _ central galaxies have the same sfr distribution , regardless of formation history .
we note two important points .
first , even when viewed through the mock group catalog , this null model recovers well the average quiescent fraction for central galaxies ( dotted lines ) .
some upturn occurs at @xmath65 , caused by redshift - space distortions , but it is weak compared to the observed enhancement .
thus , we conclude that observational effects related to the group finder alone can not account for the strong enhancement in the quiescent fraction for central galaxies : it is a real physical effect .
second , because this model recovers the cosmic average , it underestimates the observed quiescent fraction even out to @xmath70 .
thus , the importance of quenching in ejected satellites is not confined just to within @xmath82 of host halos ; it is critical to understanding the _ entire _ population of central galaxies .
by contrast , solid curves show ` satellite - type ' evolution for ejected satellites , in which their sfr is quenched in the same way as normal satellites .
this model provides good overall agreement for the upturn within @xmath82 .
just as importantly , it also correctly enhances the quiescent fractions at larger @xmath61 , because of satellites ejected from nearby host halos , as showed .
we emphasize that this model for ` satellite - type ' sfr evolution is simply an extension of our empirically constrained model from which provided excellent agreement for satellites that remain within host halos to the ejected satellite population .
furthermore , our hypothesis that ejected satellites are quenched ( and remain quenched ) is supported physically by their orbital histories , as showed .
thus , given the good agreement of this model with sdss in , we arrive at the following main conclusions : 1 .
sfr evolution and quenching in ejected satellites is consistent with proceeding in the same manner as in normal satellites that remain within their host halo .
this quenching of ejected satellites naturally can explain the enhanced quiescent fraction for central galaxies out to @xmath123 beyond host halos .
the success of this model leaves little - to - no room for environmental processes that quench central galaxies beyond a host halo s virial radius , that is , essentially all environmental quenching is consistent with originating within a host halo s virial radius .
in addition to star - formation quenching , in this last section we explore the implications of the mass evolution of ejected satellites .
as showed , because of the tidal stripping of ( sub)halo mass that occurs after first infall and continues after ejection , ejected satellites have significantly reduced halo masses as compared with normal ( non - ejected ) central galaxies of the same stellar mass . however , because a galaxy is smaller and more tightly bound than its ( sub)halo , ejected satellites likely do not experience significant stellar mass loss from tidal stripping during their ( typically single ) passage through the host halo @xcite .
thus , central galaxies that are ejected satellites would have a higher ratio of stellar to halo mass than normal central galaxies . to illustrate the significance of this effect for the entire population at @xmath88 , shows median @xmath125 for all central galaxies in the simulation ( dashed ) and for just central galaxies that are ejected satellites ( solid ) , as obtained through subhalo abundance matching using @xmath48 ( ) . because of mass stripping , the halo mass of an ejected satellites is typically only @xmath126 that of the whole central galaxy population of the same stellar mass .
in other words , _ ejected satellites systematically have a @xmath127 higher ratio of @xmath26 to @xmath28 than all central galaxies _ , with a somewhat stronger fractional enhancement at lower mass .
while shows the offset in halo mass if one isolates the population of ejected satellites , in reality they are spatially intermixed with normal ( first - infalling ) central galaxies . because of the intrinsic difficulty in separating these populations , observationally measuring this effect requires stacking and averaging both populations , leading to a weaker measured offset . the population of ejected satellites is most significant near the most massive host halos ( ) ,
where halo mass stripping also is strongest ( ) . additionally , ejected satellites are more likely to be quiescent than non - ejected central galaxies of the same stellar mass ( ) .
thus , quiescent , low - mass central galaxies near the most massive host halos should provide the strongest measurable reduction in average halo mass . to examine the potential observability of this effect , we use our ` satellite - like ' evolution model from to assign sfrs to all galaxies in our simulation .
we then stack all quiescent central galaxies near cluster - mass host halos and measure their average halo mass as a function of halo - centric distance .
( top ) shows this average halo mass , normalized to the average halo mass of all isolated ( do not lie within @xmath122 of a more massive host halo ) central galaxies of the same stellar mass .
we select galaxies with @xmath128 to match our sdss sample , but we note that this effect is even stronger at lower stellar mass because of a higher likelihood of being an ejected satellite .
solid blue curves show the true , underlying average as measured in real space , using the halo catalog directly from the simulation . at @xmath129 , where ejected satellites are most common
, the average halo mass of all central galaxies is 45% that of isolated central galaxies .
have average halo masses that are @xmath78 lower than isolated centrals , so this does contribute to but is not the dominant effect .
we also checked that the distribution of stellar masses within our chosen range does not vary with distance to create additional signal . ]
this reduction remains particularly strong out to @xmath69 , as expected given the spatial extent of ejected satellites from .
similar to , the average ratio does not asymptote until @xmath130 because of satellites ejected from neighboring host halos , and because ejected satellites are cosmically ubiquitous , the ratio asymptotes to @xmath131 and not 1 .
dashed red curves show the average as measured in projection using the simulation mock group catalog , imposing a line - of - sight velocity cut of @xmath132 km / s ( more conservative than before to maximize signal ) .
while measuring in projection reduces the signal somewhat , it remains strong if using a sufficiently conservative velocity cut .
these results suggest that the reduced halo masses for central galaxies near groups / clusters as caused by ejected satellites could be measured observationally via galaxy - galaxy weak gravitational lensing , a technique that can measure accurate halo masses for samples of stacked galaxies ( e.g. , * ? ? ?
. however , in applying such a technique to the sample in ( top ) , note that stacking all clusters with @xmath133 in our sdss group catalog leads to @xmath134 quiescent galaxies at @xmath128 in each bin of @xmath135 , a number that is unlikely to yield a discriminating stacked lensing signal from current or near - term surveys @xcite . however ,
if we consider instead _ all _ central galaxies near host halos with @xmath136 , the number in each such distance bin increases to @xmath137 .
( bottom ) shows average halo masses for this much larger sample .
while the average is 10 - 15 percentage points higher , the order - of - magnitude increase in galaxy count makes such a lensing measurement potentially feasible with existing sdss data , and upcoming measurements from deeper imaging surveys such as the dark energy survey ( des ) , hyper suprime - cam ( hsc ) , and the large synoptic survey telescope ( lsst ) should provide strong constraints on these halo masses of ejected satellites ( e.g. , * ? ? ?
* ) . indeed , galaxy - galaxy lensing measurements already show reduced subhalo masses for satellite galaxies _ within _ individual clusters @xcite , and @xcite measured reduced ( sub)halo masses for galaxies in high - density environments ( defined by local galaxy density ) from the canada - france - hawaii telescope lensing survey ( cfhtlens ) .
in addition to measurements of subhalo masses _ within _ groups / cluster , such as these , we emphasize the utility of measuring galaxies _ near _ groups / clusters : while the ( sub)halo masses of ejected satellites are not as tidally truncated as those of satellites at the core of their group / cluster , galaxies near groups / clusters can provide a cleaner lensing signal because their contrast with the background density field is much stronger .
overall , these results suggest promise in measuring ( sub)halo stripping in ejected satellites via galaxy - galaxy lensing , and we will pursue further investigation in future work . in summary ,
ejected satellites have systematically biased halo masses , typically only 40% as high as all central galaxies of the same stellar mass , leading to a @xmath138 higher @xmath125 ratio .
this truncation strongly manifests itself in the average halo masses of central galaxies out to @xmath69 beyond groups / clusters and should be measurable via galaxy - galaxy lensing , particularly with measurements from surveys such as des , hsc , and lsst .
using a galaxy group / cluster catalog from sdss to decompose galaxies into ` centrals ' and ` satellites ' , we examined the quiescent fraction for galaxies with @xmath128 that reside near groups / clusters with @xmath139 .
we postulated that the observed enhancement of the quiescent fraction in central galaxies beyond the virial radius of host halos is caused by ejected satellites , that is , galaxies that fell into a host halo and then orbited back out beyond the virial radius , becoming again central galaxies in their own halo .
we used a cosmological _ n_-body simulation to examine their orbital histories and mass evolution , and we examined scenarios for their sfr evolution , which we tested by applying the same group - finding algorithm to the simulation to make robust comparisons with sdss .
our primary results are as follows .
+ _ central galaxies near more massive host halos are more likely to be quiescent . _
considering all galaxies regardless of type , an enhanced quiescent fraction ( beyond the cosmic average at a given stellar mass ) extends out to @xmath64 around massive host halos .
however , this enhancement is caused partially by satellites that reside in neighboring host halos and the fact that host halos are spatially clustered .
considering only central galaxies , which reside in their own host halo , an enhanced quiescent fraction extends out to @xmath70 and is especially strong within @xmath82 . _ a large fraction of central galaxies near massive host halos
are ejected satellites . _
the orbits of ejected satellites extend to @xmath69 beyond their host halo , though because of satellites ejected from neighboring host halos , ejected satellites represent a significant fraction of all central galaxies out to much larger distances .
satellites of lower mass and those in / near more massive host halos are more likely to experience an ejected phase , a likely result of the decreased efficiency of dynamical friction braking in those regimes . in the extreme case , ejected satellites represent 40% of all central galaxies of @xmath87 within @xmath82 of clusters . _
the orbital histories of ejected satellites are consistent with satellite - like environmental quenching . _ after first infall , ejected satellites almost always ( @xmath94 ) experienced a single pericentric passage , typically to @xmath89 , spending roughly a virial crossing time , @xmath104 , within their host halo .
almost all remain on bound orbits and ( will ) re - enter the host halo @xmath105 after ejection .
their typical time since first infall is @xmath106 , longer than the satellite quenching timescale , and almost all ( 90% ) continue to lose ( sub)halo mass after ejection and do not re - accrete mass .
_ the quenching of star formation in ejected satellites is consistent with occuring in the same manner as in normal satellites , and this can explain the enhanced quiescent fraction for central galaxies near massive host halos .
_ we applied the same model for satellite quenching , based on time since first infall , that we developed for normal satellites in to ejected satellites .
this simple extension of our model naturally and accurately accounts for the enhanced quiescent fraction in central galaxies out to @xmath70 .
effectively all environmental quenching is consistent with occurring within a host halo s virial radius . _
the success of our simple and natural extension of satellite quenching timescales to ejected satellites leaves little - to - no room for additional environmental effects for quenching central galaxies prior to first virial infall .
_ ejected satellites have significantly biased halo masses , and this is potentially observable . _ the ( sub)halos of ejected satellites lose significant mass from tidal stripping after both first infall and ejection . to the extent that the stellar mass remains unstripped , ejected satellite have a @xmath140 times higher @xmath125 ratio ( or 60% lower halo mass ) than all central galaxies of the same stellar mass .
a reduction of average halo mass extends to central galaxies @xmath69 beyond massive host halos and is potentially observable via galaxy - galaxy weak lensing .
our results broadly agree with those of previous work on the sfrs of galaxies near groups / clusters @xcite , though our approach represents a significant advancement : using a larger galaxy sample and examining a broader range of galaxy and host halo masses , accounting for correlated structure and neighboring host halos through our sdss group catalog , tracking individual satellite orbital histories in simulation , applying an empirically constrained model for sfr evolution and quenching in satellites , and robustly comparing our model results with sdss through a simulation mock group catalog .
for instance , our results , based on a much larger sample of more massive galaxies , firmly support the scenario as proposed by @xcite that satellite ejection explains the presence of dwarf galaxies that are beyond the virial radius of the milky way and have depleted @xmath1 masses and old stellar populations .
our results also connect with previous work on mass stripping in ejected satellites .
@xcite examined the mass - to - light ratios of ejected ( backsplash ) versus infalling central galaxies in an sph simulation of a local group analogue .
they also found that ejected satellites have a higher ratio of stellar to halo mass than infalling central galaxies , but interestingly , they measured the mass - to - light ratio _ within _ the extent of the galactic stellar component ( farthest star particle ) .
their result suggests that , in addition to stripping halo mass , tidal forces might heat the orbits of stars in the discs of ejected satellites , which could provide another means to differentiate observationally ejected satellites from infalling centrals based on stellar concentration and/or kinematics .
we have argued that ejected satellites , which passed within the virial radius of a more massive host halo , can account for essentially all galaxy quenching beyond the virial radius .
however , we can not _ entirely _ rule out environmental processes that extend beyond the virial radius , quenching central galaxies prior to crossing inside .
our model parametrizes the quenching of star formation in ejected satellites in terms of time since first infall ( first virial radius crossing ) , with an initial ` delay ' time over which star formation is unaffected , followed by rapid quenching . from , this model successfully accounts for the sfr distribution of satellite galaxies in groups / clusters as a function of both galaxy and host halo mass , and , as will show , successfully describes the dependence of the quiescent fraction on halo - centric distance _ within
_ groups / clusters . however , it may be possible that other models , in which the quenching process starts at a farther distance , work as well . we will test and compare other physical models for satellite quenching in .
none - the - less , our model , in which the environmental quenching process starts at first virial infall , can explain the sfrs of central galaxies beyond groups / clusters , and we argue that this is the simplest and most natural explanation for the extended environmental quenching , particularly given what we already know about satellite quenching within the virial radius .
the results of @xcite and @xcite , both based on sph simulations , also broadly support this picture for sfr in galaxies in our mass range .
while they found some stripping of extended gas ( beyond the disk ) in the halos of infalling galaxies prior to virial crossing from ram - pressure in filaments , the effect on star formation before virial crossing was negligible for galaxies in our mass range , in agreement with our results .
thus , to the extent that the extension of our model to ejected satellites is correct , the agreement of our model with sdss means that any environmental effects beyond the virial radius are sub - dominant for galaxies in our mass range .
though , @xcite and @xcite did see stronger reduction of star formation prior to virial crossing for galaxies of significantly lower mass , which could be tested in the future through observations of lower mass galaxies .
we did consider two tests to isolated potential environmental effects on galaxies within our mass range before first infall .
first , we examined the relative velocities ( with respect to the host halo ) of ejected versus first - infalling halos at @xmath141 , finding that the median velocity components of ejected satellites can be lower than that of first - infalling halos ( @xmath142 for tangential , @xmath78 for radial ) , especially near massive clusters .
however , the 68% scatter in velocity of both populations is large ( factor of 3 - 4 , see @xcite ) , and when viewed in projection with redshift - space distortions , the line - of - sight velocity differences significantly wash out . as a way to test this method
, we varied our line - of - sight velocity cut ( which was @xmath58 ) , but we did not seen any significant trends to suggest that this method is particularly discriminating .
this echoes the more detailed investigation of @xcite , who showed that , while incorporating line - of - sight velocity can help in estimating time since infall , the reduction in scatter in time since infall at fixed halo - centric distance using velocity information is typically only a few percent . as a second test
, we note that because ejected satellites are significantly stripped of halo mass , they also host fewer satellites in their halo .
we find that the average number of satellites within ejected halos is up to @xmath143 lower than in first - infalling halos of the same stellar mass , especially near massive clusters .
however , while the _ relative _ reduction of satellite occupation is significant , the _ absolute _ occupation that is measurable above our mass limit remains low . for galaxies of @xmath144 , the average number of hosted satellites that are above our mass limit ( @xmath145 ) drops from 9% to 3% for first - infalling versus ejected halos .
thus , selecting galaxies that host no satellites can not cleanly demarcate the ejected population .
conversely , selecting galaxies that do host a satellite can more cleanly select the first - infalling population , but unfortunately this represents only a few percent of all such galaxies , reducing the available statistics significantly . to test this
, we examined the quiescent fraction for galaxies that host a satellite and are within @xmath141 of a more massive host halo , but the resultant uncertainties in the quiescent fraction were to large to discriminate any clear difference . incorporating catalogs with fainter galaxies , and thus larger satellite occupations , may help to improve this test , and combining this with line - of - sight velocity information may help to test star formation in ejected versus first - infalling populations more cleanly .
we leave such investigations to future work .
we note that the exact choice of virial radius is somewhat arbitrary , and the complications of ejected satellites could be ` defined ' away , to some extent , by using a more liberal virial definition to encompass all associated satellites .
however , this alternative has several drawbacks .
our virial definition of @xmath146 already is broader than many other commonly used ones , such as @xmath147 . naively extrapolating an nfw density profile , a virial definition whose radius extends to @xmath82the extent of ejected satellites
would contain an average density of only @xmath148 , a density more closely associated with filaments .
a larger virial radius also would encompass more galaxies that are not ( yet ) physically associated with the host halo and thus have not ( yet ) likely been affected environmentally .
thus , the choice of virial radius necessarily represents a trade - off between purity and completeness of the associated satellite population , and a proper treatment of ejected satellites will be important for any reasonable virial definition . finally ,
while our group - finding algorithm assumes that host halos are spherical , their density contours and virial shock fronts can be ellipsoidal , meaning that the physical extent of a halo can extend to more ( or less ) than a spherical virial radius .
we leave investigation of alternative group finders , in the context of ejected satellites , to future work .
our results highlight the importance of environmental history in understanding not only star formation but also stellar mass growth in galaxy evolution .
central galaxies that are really ejected satellites represent a challenge for the standard ` halo model ' and halo occupation distribution ( hod ) formalism ( see * ? ? ?
* for review ) .
both the sfr / color and stellar mass of ejected satellites do not depend simply on their halo mass , but also on their formation history .
as showed , ejected satellites have significantly higher stellar mass than their instantaneous halo mass would imply , and they are more likely to be quiescent / red than normal central galaxies of the same stellar mass .
thus , any modeling approach that seeks to assign stellar mass or sfr / color to halos based only on their instantaneous virial mass , as in a standard halo - model / hod approach , will underestimate both the stellar mass and the quiescent / red fraction for such central galaxies .
moreover , because less massive central galaxies are intrinsically less likely to be quiescent / red , both effects bias in the same direction .
this bias will be most important on quasi - linear scales in the vicinity of massive host halos ( ` 1-halo ' to ` 2-halo ' transition ) , where the ejected fraction is 10 - 40%
. it may be possible to modify the standard halo model to incorporate analytically an associated satellite population that extends out to 2.5 virial radii and spatially overlaps ( in a statistical sense ) with the population of distinct host halos , though such a modification would require a careful recalibration of the mass function and bias of host halos .
overall , ejected satellites represent a population for which it is particularly important to follow the formation history of a ( sub)halo to understand the galaxy inside .
relatedly , the biased @xmath124 relation for ejected satellites , if unaccounted for , could manifest itself as increased scatter in the @xmath124 relation if averaging over all central galaxies in halo - model / hod - based analyses .
while this effect is unlikely to be significant at @xmath149 , where the fraction of all central galaxies that are ejected satellites is @xmath150 ( ) , we expect it to be increasingly significant at lower @xmath26 where the overall ejected fraction is closer to 10%
. note that if 5 - 10% of galaxies at a given stellar mass have systematically 60% lower halo mass , this alone translates to a scatter of 0.09 - 0.12 dex in averaging the population .
this represents a non - trivial fraction of empirically measured values of 0.15 - 0.2 dex ( see ) .
thus , while we do not expect ejected satellites to be the dominant source of scatter in the @xmath124 relation , they may be a significant component if unaccounted for .
finally , we note the potential connection of these results to ` galactic conformity ' , that the properties of central galaxies , such as sfr / color , are correlated with those of their satellites @xcite , or more generally , with all neighboring galaxies out to @xmath151 @xcite .
as @xcite noted , this conformity signal is most pronounced for less massive galaxies that have low sfrs , and can be stronger at larger separations ( @xmath152 ) . ejected satellites provide a natural population to help explain this effect , both because ( 1 ) they are relatively low - mass , quiescent central galaxies that reside near more massive halos , which host many quiescent satellite galaxies , and ( 2 ) if ejected satellite halos contain their _ own _ satellites , then both the central and satellite galaxies in such a halo would have passed through a more massive host halo , thus both galaxy types are more likely to have environmentally quenched ( correlated ) star formation .
however , @xcite noted that a full explanation of their observed conformity effect would require more than half of all low - mass central galaxies to be ejected satellites , and our results suggest that this fraction is always @xmath84 for galaxies with @xmath153 .
none - the - less , we fully expect that ejected satellites are a population that _ should _ exhibit some conformity signal and thus would provide some contribution to the overall conformity effect
. we will investigate this issue in future work .
we thank michael blanton , david hogg , and collaborators for publicly releasing the nyu vagc , jarle brinchmann and the mpa - jhu collaboration for publicly releasing their spectral reductions , and martin white for simulation data .
we thank kathryn johnston and mark fardal for enlightening conversations , and marla geha for comments on an early draft .
we also thank the anonymous reviewer for many useful comments .
the simulation was analyzed at the national energy research scientific computing center . | galaxies that are several virial radii beyond groups / clusters show preferentially quiescent star formation rates . using a galaxy group / cluster catalog from the sloan digital sky survey , together with a cosmological _ n_-body simulation
, we examine the origin of this environmental quenching beyond the virial radius . accounting for the clustering of groups / clusters ,
we show that central galaxies show enhanced sfr quenching out to 2.5 virial radii beyond groups / clusters , and we demonstrate that this extended environmental enhancement can be explained simply by ` ejected ' satellite galaxies that orbit beyond their host halo s virial radius .
we show that ejected satellites typically orbit for several gyr beyond the virial radius before falling back in , and thus they compose up to 40% of all central galaxies near groups / clusters .
we show that a model in which ejected satellites experience the same sfr quenching as satellites within a host halo can explain essentially all environmental dependence of galaxy quenching .
furthermore , ejected satellites ( continue to ) lose significant halo mass , an effect that is potentially observable via gravitational lensing .
the sfrs / colors and stellar - to - halo masses of ejected satellites highlight the importance of environmental _ history _ and present challenges to models of galaxy occupation that ignore such history .
[ firstpage ] methods : numerical galaxies : clusters : general galaxies : evolution galaxies : groups : general galaxies : haloes galaxies : star formation . |
solar oscillations are a unique tool to infer conditions inside the sun .
they have been recorded with extreme precision since 1996 with the michelson doppler imager ( mdi , ) onboard the solar and heliospheric ( soho ) space telescope .
large - scale rotation inside the sun can be estimated by inversion of the frequencies of millions of global modes of oscillation ( _ e.g. _ , ) .
rotation is known to vary with time in the solar interior at the level of about @xmath1 ms@xmath2(_e.g .
_ , ; ; ; ; ) .
these variations , known as torsional oscillations , consist of bands of faster and slower rotation that migrate in latitude as the eleven - year solar magnetic cycle develops .
torsional oscillations may be driven by the lorentz force due to a dynamo wave @xcite .
other explanations have been proposed ( see and references therein ) , including the suggestion by that torsional oscillations are driven by horizontal pressure gradients caused by photospheric magnetic activity .
long time averages of surface doppler measurements have shown the existence of a flow from the equator to the poles with an amplitude of 1020 ms@xmath2 .
an introduction to the theory of solar meridional circulation is provided by .
temporal variations in the meridional flow have been reported by several authors . by tracking the small photospheric magnetic features , , , and found a significant change in the meridional flow near sunspot latitudes , implying a solar - cycle variation .
variations in the surface doppler meridional velocity have been detected by , in particular at latitudes above @xmath3 .
local helioseismology ( see , _
e.g. _ , ) also provides reasonable measurements of the meridional circulation for latitudes below about @xmath4 @xcite .
the time - varying component of the meridional flow with respect to a long - term average does not exceed @xmath5 ms@xmath2and is consistent with a small near - surface inflow toward active latitudes @xcite and an outflow from active latitudes at depths greater than 20 mm @xcite .
as discussed by , these variations would appear to be caused by localized flows around localized regions of magnetic activity @xcite .
the purpose of this paper is two - fold .
first , we provide independent measurements of the temporal variations of the meridional circulation near the solar surface and at a depth of about 60 mm . for the near - surface meridional flow ,
we use an original technique , which consists of measuring the advection of the supergranulation pattern @xcite .
the deeper meridional flow is calibrated from an earlier time - distance helioseismology observation by .
the meridional circulation measurements at these two depths are compared , in particular by looking at the eleven - year periodicity of the flows ( section [ sec.obs ] ) .
our study of meridional circulation confirms previous observations ( listed above ) and is complementary to the analysis of zonal flows by .
second , we wish to compare the observations with a theoretical model @xcite based on a flux - transport dynamo combined with a geostrophic flow caused by increased radiative loss in the active region belt , according to spruit s ( 2003 ) original idea .
section [ sec.model ] provides a description of the model with a focus on meridional flows , since this aspect of the model has not been discussed elsewhere in detail . to our knowledge
, this model is the only existing quantitative model of the solar - cycle dependence of internal flows : it is natural to ask whether it is consistent with the observations .
the comparison between the observations and the model ( section [ sec.discussion ] ) is encouraging , although some inconsistencies can not be ignored .
the method we employ to infer flows near the solar surface is based on the analysis of , which was originally applied to a single data set from 1996 .
here we use a series of mdi full - disk doppler images covering the period 19962002 .
each year , two to three months of continuous dopplergrams are available for analysis ( mdi dynamics runs ) . the mdi data after 2002 were not used simply because we are analyzing an existing pre - processed data set .
dopplergrams were tracked at the carrington rotational velocity to remove the main component of rotation .
f_-mode time - distance helioseismology @xcite to obtain every 12 hour a @xmath6 map of the horizontal divergence of the flow field @xmath7 mm below the photosphere .
the main component of the divergence signal is due to supergranulation . for
any given target latitude ( @xmath8 ) we considered a longitudinal section of the data @xmath9 wide in latitude . using a local plane - parallel approximation in the neighborhood of latitude @xmath8 , the divergence signal was interpolated onto a cartesian grid sampled at @xmath10 mm in the @xmath11 ( prograde ) and @xmath12 ( northward ) coordinates .
the divergence signal was decomposed into its harmonic components @xmath13 to obtain a local power spectrum @xmath14 , where @xmath15 is the horizontal wavevector and @xmath16 is the angular frequency . at fixed @xmath17 ,
we fit for two functions @xmath18 and @xmath19 and a horizontal vector @xmath20 such that @xmath21 .
this representation fits the data adequately .
as was done by , we interpret @xmath22 to be a horizontal flow causing a doppler shift @xmath23 .
this flow is likely to be an average over the supergranulation layer , which has been estimated to reach depths greater than 10 mm by . ) and ( b ) meridional flow ( @xmath24 ) near the solar surface as a function of latitude ( @xmath8 ) .
each mdi dynamics run is plotted with a different color from blue in 1996 to red in 2002 .
the rotational velocity is given with respect to the rotational velocity of the small magnetic features ( komm , howard , and harvey , 1993a ) . ]
figure [ fig.surface ] is a plot of @xmath25 and @xmath26 for each full - disk mdi run as a function of latitude in the range @xmath27 . to reduce random noise , the north
south symmetric component of @xmath28 and the antisymmetric component of @xmath24 are extracted . over the period 19962002 ,
@xmath28 varies by @xmath29 ms@xmath2peak - to - peak at the equator ( figure [ fig.surface]a ) .
the meridional flow is poleward with a mean amplitude of 10 ms@xmath2at latitude @xmath30 ( figure [ fig.surface]b ) .
the peak - to - peak variation of the meridional flow is 7 ms@xmath2at @xmath31 , _ i.e. _ a significant fraction of the time - average value .
we estimate that the standard deviation of the noise at a particular latitude ( @xmath32 bin ) for any given year is less than 1 ms@xmath2 .
the systematic errors that depend on position on the solar disk have been measured to be very low ( less than @xmath33 ms@xmath2over the @xmath6 region of analysis ) . in order to probe deeper layers into the solar convection zone , we used acoustic waves and time - distance helioseismology .
for each three - month period , travel times were measured by cross - correlation of the doppler oscillation signal recorded during the mdi structure program ( nearly continuous coverage but lower spatial resolution ) according to the procedure described by .
using a mean travel distance of @xmath34 enables us to probe layers about @xmath35 mm below the surface .
the full details of this analysis can be found in .
waves that propagate in the north south direction are used to infer the meridional flow , while waves that propagate east west are used to infer the zonal flows . in order to convert travel - time shifts into flows in units of ms@xmath2
, we use a simple calibration of @xmath28 at a depth based on the observation by ( global - mode helioseismology ) that the amplitude of the time - varying component of the zonal flow is nearly independent of depth .
we choose the near - surface zonal - flow measurements of section [ sec.obssurf ] as a reference .
the calibration of @xmath28 is then used to calibrate @xmath24 .
we find that the meridional flow at a depth of 60 mm is poleward at all latitudes and has a maximum value of 6 ms@xmath2at latitude @xmath36 .
for a particular year and at fixed latitude ( @xmath32 bin ) , the standard deviation of the noise is about 2 ms@xmath2 , significantly more than for the near - surface measurements . in order to discuss the solar - cycle dependence of the flows and to study the phase relationship between the flows measured at the two different depths ,
we extract the eleven - year periodic component from the data , as was done by and for zonal flows . at each latitude
@xmath8 and for each depth , we fit a function of the form @xmath37 to the observed velocity @xmath38 , where the index @xmath39 refers to either the @xmath11 or the @xmath12 component of the flow .
the long - term average is given by @xmath40 , while the amplitude and the phase of the time - varying component are denoted by @xmath41 and @xmath42 respectively .
we extract a 11-year periodicity from the data , since it is known from other observations that this is the dominant mode ( we do not determine the 11-year periodicity based on the dataset itself ) .
shorter and longer periodicities are certainly present in the data ; however , the length of the dataset does not allow for a determination of the full spectrum of modes . .
a positive value indicates a poleward meridional or prograde zonal flow .
the observations ( @xmath43 ) cover the first six years , while the purely sinusoidal component ( @xmath44 ) is extrapolated in time ( beyond the _ white vertical white line _ ) .
the _ black curves _ indicate the mean latitude of magnetic activity . ] ) of the eleven - year periodic component of the ( a ) meridional and ( b ) zonal flows .
the near - surface values ( _ thick solid lines _ ) are absolute measurements .
the calibration of the observations at 60 mm depth ( _ thin lines _ ) follows from the assumption that the amplitude of the zonal torsional oscillation ( _ panel b _ ) is independent of depth over the latitude range @xmath45.,width=491 ] ] between the eleven - year periodic components of the flows measured at a depth of 60 mm and near the surface .
the _ solid line _ is for the meridional flow and the _ dashed line _ is for the zonal flow .
, width=302 ] [ fig.compare_phase ] the eleven - year periodic components of the meridional and zonal flows are shown in figure [ fig.surf ] .
the torsional - oscillation pattern is clearly seen at both depths with an amplitude and a phase comparable to previous measurements ( _ e.g. _ , ) .
the meridional flow also contains a significant eleven - year periodic component . near the solar surface ,
the residuals indicate the presence of a north south inflow toward the mean latitude of activity ( _ e.g. _ , ; ) , while the data are consistent with a horizontal outflow from the mean latitude of activity deeper into the convection zone ( _ e.g. _ , ; ) .
figure [ fig.compare_ampli ] gives the amplitudes of the time - varying components of the flows ( @xmath41 ) . under the assumption ( section [ sec.obsdeep ] ) that @xmath46 does not vary appreciably with depth ( @xmath33 ms@xmath2latitudinal average )
, then the amplitude of the time - varying meridional flow ( @xmath47 ) is also found to be approximately independent of depth ( @xmath48 ms@xmath2at @xmath30 latitude ) .
the evidence that the time - varying components of the meridional flow near the surface and deeper in the interior are anti - correlated is given in figure [ fig.compare_phase ] , which shows the difference in @xmath49 at the two depths .
on the contrary , there is no significant phase variation with depth for the zonal flow .
the model results presented here are based on a non - kinematic flux - transport dynamo model developed recently by rempel .
this model combines the differential rotation and meridional flow model of with a flux - transport dynamo similar to the models of and .
we emphasize that this model is intended to give a fundamental understanding of the basic cycle properties and their relation to observable variations of zonal and meridional flows .
therefore we focus here only on axisymmetric and north south averaged quantities .
details of the model can be found in . since a detailed comparison with observed torsional oscillations can be found in and , we focus here on the meridional flow variations .
the differential rotation model utilizes a meanfield reynolds - stress approach that parametrizes the turbulent angular momentum transport (; @xmath50-effect ) leading to the observed equatorial acceleration . in this model
the tachocline is forced through a uniform rotation boundary condition at the lower boundary of the computational domain . a meridional circulation , as required for a flux - transport dynamo , follows self - consistently through the coriolis force resulting from the differential rotation . the computed differential rotation and meridional flow are used to advance the magnetic field in the flux - transport dynamo model , while the magnetic field is allowed to feed back through the meanfield lorentz - force @xmath51 ( the contribution of the fluctuating part @xmath52 is not well known and neglected here ) .
we find in our model that the lorentz - force feedback can only account for the poleward propagating branch of the torsional oscillations , while the equatorward propagating branch in latitudes beneath @xmath53 requires additional physics .
parametrizing the idea proposed by that the low - latitude torsional oscillation is a geostrophic flow caused by increased radiative loss in the active region belt ( due to small scale magnetic flux ) leads in our model to a surface oscillations pattern in good agreement with observations . in order to force a torsional oscillation with around @xmath7 nhz amplitude , a temperature variation of around @xmath54 k
is required . as a side effect
the cooling produces close to the surface ( in our model at @xmath55 ) an inflow into the active - region belt of around @xmath56 ms@xmath2 .
we incorporated this process by adding a surface - cooling term that is dependent on the toroidal - field strength at the base of the convection zone , which is assumed to be the source for active - region magnetic field ( the small scale flux required for the surface - cooling is a consequence of the decay of active - regions ) .
observations show that the low - latitude branch of torsional - oscillations starts around one to two years before the sunspots of the new cycle appear .
it is possible that magnetic flux rises towards the surface without forming sunspots in the beginning of a cycle providing enough small - scale magnetic field , this is however currently neither confirmed nor ruled out by observations .
alternative explanations for the low - latitude branch of torsional oscillations such as the models of , , and are based on the longitudinal component of the lorentz force .
recently @xcite showed that torsional - oscillations forced that way are close to the taylor - proudman state ( alignment of phase with the axis of rotation ) , which contradicts observations .
in addition , the resulting meridional surface - flow has the wrong sign ( active region belt outflow ) . despite some shortcomings ,
the model of is currently the only proposed explanation that is consistent with the observed meridional and zonal - flow variations close to the solar surface .
( b ) torsional oscillations ( _ blue _ : slower , _ red _ : faster rotation , amplitude : @xmath57 nhz ) .
( c ) meridional flow variation at @xmath55 ( _ blue _ : equatorward , _ red _ : poleward motion , amplitude : @xmath56 ms@xmath2 .
( d ) meridional flow variation at @xmath58 ( _ blue _ : equatorward , _ red _ : poleward motion , amplitude : @xmath59 ms@xmath2 ) .
the variation of the meridional flow pattern at @xmath55 is almost in anti - correlation with the flow at @xmath58 ( @xmath60 mm depth ) . in all four panels
the _ contour lines _ indicate the butterfly diagram computed from the toroidal field at the base of the convection zone . ]
figure [ fig : model ] summarizes the results of the model in latitudes below @xmath61 .
figure [ fig : model]a shows the temperature fluctuation ( color shades ) caused by increased surface cooling in the active region belt .
the contour lines indicate the magnetic butterfly diagram computed from the toroidal field at the base of the convection zone in the model .
figure [ fig : model]b shows cycle variations of the angular velocity ( torsional oscillations ) and figure [ fig : model]c shows cycle variations of the horizontal meridional flow velocity . at the equatorward side of the active region belt
( indicated by the butterfly diagram ) the rotation rate is increased , which is consistent with the increased poleward meridional flow transporting material toward the axis of rotation . on the poleward side of the active region belt the rotation rate is lower , while the meridional - flow perturbation is equatorward . at a depth of around @xmath62 mm ( figure [ fig : model]d )
the meridional - flow perturbation is almost anti - correlated to the surface flow ( active - region belt outflow ) , indicating that the surface cooling drives a flow system that closes in the upper third of the convection zone .
the flow amplitude at a depth of @xmath62 mm is around one order of magnitude lower compared to the surface flow due to the significant increase in density .
we presented a model that is qualitatively consistent with the observations , in particular the phase of the solar - cycle variations of the flows . since the surface - cooling term is parametrized , the model can only make predictions about the relative amplitude of zonal and meridional flows at different depths , and not about the absolute values . near the surface ,
the model is in agreement with the data : the relative amplitudes of the torsional oscillation and the time - varying component of the meridional flow and their relative phase are reproduced well .
deeper in the interior , it appears that the model underestimates the amplitude of the time variations of the meridional flow by an order of magnitude ; however , the flow variation is in antiphase to the surface flow as seen in the data .
the lower velocity at depth in the dynamo model is a consequence of mass conservation ( strong increase in density with depth ) .
the much larger outflow which is observed in the data can not be balanced by an inflow close to the surface unless the outflow is confined to a very narrow layer .
overall , it is fair to say that the model is encouraging .
alternative models explaining low - latitude torsional oscillations through the action of the longitudinal lorentz force tend to produce zonal flow patterns in the taylor - proudman state @xcite and meridional outflows rather than inflows close to the surface . on the other hand ,
zonal flows appear during solar minimum when no active regions are present .
this requires , in the case of thermal forcing , that a sufficient amount of magnetic flux is present in the form of small scale flux elements not evident in synoptic magnetograms ( see _ e.g. _ , section 6.3 ) .
also a local treatment of the regions of strong magnetic - field concentrations ( sunspots and active regions ) might be necessary to obtain a better match between the model and the data
. the national center for atmospheric research is sponsored by the national science foundation .
m. rempel thanks prof .
schssler and the max - planck - institut fr sonnensystemforschung for their hospitality .
gizon , l. , duvall , t.l . , larsen , r.m .
: 2001 . in : brekke , p. , fleck , b. , gurman , j.b .
( eds . ) , _ recent insights into the physics of the sun and heliosphere : highlights from soho and other space missions _ ,
iau symp .
* 203 * , astron .
pacific , san francisco , 189 .
scherrer , p.h . ,
bogart , r.s . , bush , r.i . ,
hoeksema , j.t . ,
kosovichev , a.g . ,
schou , j. , rosenberg , w. , springer , l. , tarbell , t.d . , title , a. , wolfson , c.j . ,
zayer , i. , the mdi engineering team : 1995 , _ solar phys . _ * 162 * , 129 .
schou , j. , antia , h.m . ,
basu , s. , bogart , r.s . , bush , r.i . ,
chitre , s.m . , christensen - dalsgaard , j. , dimauro , m.p . ,
dziembowski , w.a .
, eff - darwich , a. , gough , d.o . ,
haber , d.a .
, hoeksema , j.t . , howe , r. , korzennik , s.g .
, kosovichev , a.g . ,
larsen , r.m . , pijpers , f.p . , scherrer , p.h . ,
sekii , t. , tarbell , t.d . ,
title , a.m. , thompson , m.j . , toomre , j. : 1998 , _ astrophys .
j. _ * 505 * , 390 . | we present independent observations of the solar - cycle variation of flows near the solar surface and at a depth of about 60 mm , in the latitude range @xmath0 .
we show that the time - varying components of the meridional flow at these two depths have opposite sign , while the time - varying components of the zonal flow are in phase .
this is in agreement with previous results .
we then investigate whether the observations are consistent with a theoretical model of solar - cycle dependent meridional circulation based on a flux - transport dynamo combined with a geostrophic flow caused by increased radiative loss in the active region belt ( the only existing quantitative model ) .
we find that the model and the data are in qualitative agreement , although the amplitude of the solar - cycle variation of the meridional flow at 60 mm is underestimated by the model . |
the possible decay of inertially stable particles due to strong gravitational fields has been considered recently . in particular ,
the proton decay , by weak and strong interactions , in uniformly accelerated trajectories@xcite and in circular motion around compact objects@xcite , has been considered in great detail .
the astrophysical implications of these results are now under investigation .
the possible decay of accelerated protons , however , is not a new issue .
it can be traced back to the works of ginzburg and zharkov@xcite in the sixties , where processes of the type @xmath0 were considered . at the same time
, zharkov@xcite investigated the weak and strong decay of protons accelerated by an external electromagnetic field .
( see @xcite for a review . ) clearly , none of these processes would be allowed in the absence of external forces .
we notice , however , that there are subtle differences between processes involving uncharged particles where the accelerations have gravitational and electromagnetic origins , see @xcite for further details . in this paper
, we consider the decay and the lifetime of unstable particles in geodesic circular motion around spherically symmetrical compact objects .
we evaluate , in particular , decay rates and lifetime for neutrons in relativistic circular motion according to the semiclassical approach introduced in @xcite .
both the weak [ weak ] n^0 p^+e^-|_e and the ( inertially forbidden ) strong [ strong ] n^0 p^+^- channels are considered .
our results are compared to several ones obtained previously in the literature for the related processes @xmath1 and @xmath2 .
as we will see , for realistic black - holes , noticeable effects are expected to occur for circular geodesics close to the photonic orbit @xmath3 .
observers in these ( unstable ) circular orbits are necessarily in extremely relativistic motion @xmath4 , and it is well known that they indeed realize the inertial vacuum as a quasi - thermal distribution of particles characterized by a temperature @xmath5 in the range@xcite [ temp ] kt , where @xmath6 stands for the effective minkowskian centripetal acceleration for relativistic circular orbits .
our results suggest that the temperature ( [ temp ] ) have in the present case the same central role played by unruh temperature@xcite in the analysis of uniformly accelerated particles as seen from rindler observers@xcite .
( see @xcite for a recent review . )
similar conclusions hold also for other unstable particles .
the line element corresponding to a spherically symmetrical object of mass @xmath7 is given by the schwarzschild metric [ schwarz ] ds^2 = -(1- 2m / r ) dt^2 + ( 1- 2m / r ) ^-1dr^2 + r^2d^2 , where @xmath8 .
natural unities are adopted hereafter . in this coordinate system@xcite , a particle of mass @xmath9 in a circular timelike geodesic at radius @xmath10 on the equatorial plane
have energy per mass ratio given by /m
= ( 1 - 2m / r ) = ( 1 - 2m / r)/ , with the dot standing for @xmath11-derivative .
its angular momentum @xmath12 can be calculated directly from the definition of a timelike circular geodesic parameterized by the proper time @xmath11 , leading finally to the following expression for the worldline of a timelike circular equatorial geodesic in schwarzschild coordinates [ circ ] x^a(s ) = ( , r,/2,s ) .
clearly , since the trajectory ( [ circ ] ) is a geodesic , its acceleration @xmath13 calculated with respect to the metric ( [ schwarz ] ) vanishes identically .
however , we will proceed here in a different manner . in the next section
, we will consider quantum effects as realized by observers with circular trajectories as ( [ circ ] ) in the minkowski spacetime .
an observer following the worldline ( [ circ ] ) in minkowski spacetime experiences a centripetal acceleration [ accel ] a = = . on the other hand , in minkowski spacetime the worldline of a particle in a uniform circular motion on the equatorial plane with angular velocity @xmath14
is given by [ trajec ] x^a ( ) = ( t , r,/2,t ) , from which one has immediately @xmath15 and @xmath16 , where the constant @xmath17 corresponds to the lorentz factor .
the angular velocity @xmath14 is to be determined by imposing the centripetal acceleration ( [ accel ] ) for the trajectory ( [ trajec ] ) , yielding [ pseudoaccel ] = and the following lorentz factor [ pseudoaccel1 ] = .
the treatment of the quantum effects realized by observers in the circular geodesic ( [ circ ] ) of schwarzchild spacetime by means of an effective minkowskian circular trajectory is , of course , only an approximation .
it is shown in @xcite , nevertheless , that the results obtained in a semiclassical approach assuming a schwarzschild spacetime and a flat spacetime with external `` newtonian '' attraction forces such that ( [ accel ] ) and ( [ pseudoaccel ] ) hold differ by no more than 30% , if we restrict ourselves to the circular orbits with @xmath18 . since schwarzschild spacetime is asymptotically flat , it is indeed natural that the emitted powers calculated in minkowski and schwarzschild spacetime agree when one considers circular motions with large @xmath10 .
in fact , as it is shown in @xcite and @xcite , the power emitted by a particle in circular motion with radius @xmath10 in minkowski spacetime with angular velocity ( [ pseudoaccel ] ) is very close to that one emitted by a particle in a circular geodesic with the same radius @xmath10 in a schwarzschild spacetime , provided that @xmath19 .
this is in agreement with the well known fact that processes involving wavelengths with the same order of magnitude of the schwarzschild radius need necessarily to be analyzed using fully curved spacetime calculations .
moreover , the acceleration defined in ( [ accel ] ) has the additional desirable feature of being divergent at @xmath20 , in accordance with previous works on geodesic emission@xcite , which established that near the photonic orbit the emitted power diverges .
the angular velocity ( [ pseudoaccel ] ) mimics the main qualitative properties of the real schwarzschild circular geodesics , justifying our assumption of circular trajectories in a flat spacetime with centripetal acceleration ( [ accel ] ) .
the semiclassical current formalism employed in @xcite to the proton decay case consists basically in considering the proton and the neutron as distinct energy eigenstates @xmath21 and @xmath22 of a two - level system such that @xmath23 and @xmath24 , where @xmath25 is the proper hamiltonian of the system , @xmath26 and @xmath27 are , respectively , the proton and neutron masses . the weak channel ( [ weak ] ) is implemented by considering a vector current associated to the two - level system coupled to a quantized fermionic field ( corresponding to the electron @xmath28 and to the anti - neutrino @xmath29 ) by means of the effective coupling constant @xmath30 , which is about the order of the fermi coupling constant @xmath31 , whereas the strong channel ( [ strong ] ) involves a scalar current coupled to a quantized bosonic field ( the pion @xmath32 ) by means of the effective coupling constant @xmath33 , of the order of the pion - nucleon - nucleon strong coupling @xmath34@xcite .
the currents are then specialized to the case of uniform circular trajectories with radius @xmath10 and angular velocity @xmath14 ( and centripetal acceleration @xmath35 ) in minkowski spacetime . the proper decay rates corresponding to the weak ( [ weak ] ) and to the strong ( [ strong ] ) channels
are given , respectively , by @xmath36 and [ decay2 ] _
np^s = - d , where @xmath37 , @xmath38 and @xmath39 being , respectively , the electron and neutrino masses , and @xmath40 .
we assume here @xmath41 .
the rates ( [ decay1 ] ) and ( [ decay2 ] ) are obtained , respectively , from equations ( 3.31 ) and ( 3.16 ) of @xcite , where all the relevant details can be found . in particular , we have @xmath42 , where @xmath43 is a regulator . for the relativistic case ( @xmath44 ) , the terms involved in the integrations ( [ decay1 ] ) and ( [ decay2 ] ) are@xcite @xmath45 where @xmath46 , @xmath47 , see fig .
[ fig1 ] .
in contrast to the case considered in @xcite , since @xmath48 , one needs here to perform the complex integrations ( [ decay1 ] ) and ( [ decay2 ] ) along the path depicted in fig .
we get , after taking the limit @xmath49 , [ decayw ] _
np^w = _ pn^w + ( 20 + 15 ^2 + 3 ^4 -15 ^ 2 ( 1 + ^2 ) ) and [ decays ] _
np^s = _ pn^s
+ , where @xmath50 and @xmath51 correspond , respectively , to the proper decay rates associated to the inverse processes considered in @xcite , @xmath52\end{aligned}\ ] ] and [ protonrate2 ] _
pn^s = g_s^2ae^-2/(8 ) .
the approximations involved in the derivation of the expressions ( [ decay1 ] ) and ( [ decay2 ] ) require , respectively , that the centripetal acceleration @xmath6 obeys @xmath53 and @xmath54 , where @xmath55 stands for the pion @xmath32 mass .
we notice that , for accelerations @xmath56 , the no - recoil hypothesis is violated@xcite and , hence , our approximation breaks down . the neutron proper lifetime associated with the decay rates ( [ decayw ] ) and ( [ decays ] )
are given simply by @xmath57 and @xmath58 .
our results assume a particularly simple form if one considers the proton and neutron lifetime ratio , namely [ sratio ] _
p^s/_n^s = 1 + 12 e^2 , and [ wratio ] _
p^w/_n^w = 1 + e^2 , where @xmath59 and @xmath60 are polynomials easily obtained from ( [ decayw ] ) and ( [ protonrate1 ] ) .
it is clear that for large @xmath6 , both ratios obey [ asymp ] _
p/_n 1 + o(a^-1 ) .
the strong channel is expected to be the dominant decay mode for neutrons with centripetal accelerations @xmath6 such that @xmath61 .
taking into account that @xmath62 mev , @xmath63 mev and that @xmath64 mev , we have that the proton and the neutron lifetimes differ by no more than 1% to 10% in the range of accelerations where the strong channel dominates .
the weak channel , on the other hand , dominates for smaller accelerations @xmath65 .
( we remind that @xmath66 mev . ) from ( [ sratio ] ) and ( [ wratio ] ) , it is clear that for [ critical ] a a_c = m_n - m_p 1.29 mev the asymptotic expression ( [ asymp ] ) holds accurately .
the meaning of the `` critical '' acceleration @xmath67 will be discussed in the last section . here , we mention only that @xmath67 belongs to the range where the weak channel dominates .
in fact , for @xmath68 , a significative difference between the proton and the neutron lifetime is observed . a proper acceleration of the order of 1 mev is extremely high .
for sake of comparison , protons in the cern large hadron collider have proper acceleration @xmath69 mev@xcite . in order to compare @xmath67 with centripetal accelerations induced by realistic black - holes
, we cast ( [ accel ] ) in the form [ accel1 ] a 1.3410 ^ -16 ( ) , where @xmath6 is now measured in mev and the numerical constant corresponds to @xmath70 .
hence , realistic black holes @xmath71 will induce centripetal accelerations of the mev order only for those circular orbits very close to the photonic orbit @xmath72 .
smaller black holes , nevertheless , can induce considerably higher centripetal accelerations for the ( unstable ) circular orbits located between the photonic orbit and the last stable circular orbit at @xmath73 , see figure [ figaccel ] . on the other hand , the hypothesis of an extremely relativistic motion ( @xmath44 ) adopted in the approximations
( [ apro1 ] ) and ( [ apro2 ] ) requires circular trajectories close to the photonic orbit , since from ( [ pseudoaccel1 ] ) one has [ gamma ] = .
figure [ fig2 ] depicts the lifetime for a neutron due to weak decay in circular orbits with @xmath74 around small black holes . as expected , the smaller is the black hole , the larger is the reduction in the particle lifetime .
the semiclassical approach used here can be applied for other unstable particles as well .
particularly interesting is the muon weak decay @xmath75 which can also be described by a vector current based on a two - level system coupled to quantized fermions ( the neutrinos @xmath29 and @xmath76 ) by means of a coupling constant of the same order than @xmath77 .
since neutrinos have very small masses , the rate ( [ decayw ] ) for the muon weak decay is accurate for accelerations in the range @xmath78 mev .
( we remind that @xmath79 mev . )
figure [ fig3 ] depicts the lifetime of a muon in geodesic circular orbits of small black holes such that @xmath74
as figures [ fig2 ] and [ fig3 ] show , small black holes are necessary in order to induce sensitive alterations in the proper lifetime of unstable particles in circular orbits .
however , noticeable effects do occur in the vicinity of the photonic orbit for realistic black hole . despite that the analysis presented here
is restricted to the ( unstable ) circular geodesics close to @xmath20 , it can give some hints about the behavior of particles in more realistic situations .
as equations ( [ sratio ] ) and ( [ wratio ] ) reveal , free protons and neutrons in geodesic circular motion close to the photonic orbits have comparable proper lifetime .
this situation is completely different from the inertial one , and its implication to particle physics in the vicinity of black holes has not been sufficiently studied yet .
a similar conclusion holds for the muon . for a accelerations @xmath80 mev , neglecting possible effects of back - reaction@xcite , the muon and the electron , which in such a case can indeed decay by the inverse process @xmath81 have comparable lifetimes . in order to grasp the meaning of the temperature ( [ temp ] ) , let us consider the case of the weak decay of protons and neutron in uniformly accelerated motion , where the unruh temperature@xcite @xmath82 is known to play a central role@xcite .
the proton and neutron lifetime ratio for this case can be obtained from the decay rates ( 3.13 ) and ( 3.17 ) of reference @xcite , [ ee ] _
p^w/_n^w = e^2 , which also has the asymptotic form ( [ asymp ] ) for large values of @xmath6 .
notice that , in this case , we have exactly the same critical acceleration @xmath67 of ( [ critical ] ) . in the case of the uniformly accelerated motion , one can describe the decay of protons and neutrons as seen by comoving rindler observers .
the key point here is that rindler observers realize the inertial vacuum as a thermal state with temperature @xmath82 .
heuristically , one can imagine the two - level system in thermal equilibrium with the unruh radiation associated with the quantized fields in question . for our two - level system in thermal equilibrium at temperature @xmath5 , the probability of occupation of the proton @xmath21 and neutron @xmath22 states are ,
respectively n_p = , n_n = .
the ratio @xmath83 diverges for @xmath84 , indicating that for low temperatures , the system is likely to be in its fundamental state .
however , for temperatures @xmath85 the ratio tends to 1 , indicating that the system can be in the states @xmath21 or @xmath22 with equal probability . in other words , the transitions @xmath86 and
@xmath87 become equally probable for high temperatures , shedding some light in the expression ( [ ee ] ) . for linear accelerations @xmath6 such that the associate unruh temperature @xmath82 is much higher than the energy gap @xmath88
, it is natural to expect that protons and neutrons have the same lifetime , since both transitions of the two level systems are equally probable . the lifetime ratios ( [ sratio ] ) and ( [ wratio ] ) suggest something similar to the case of uniform circular trajectories .
they can be understood if one considers the two - level system in equilibrium with the quasi - thermal radiation with temperature @xmath89 associated with the quantized fields in question , confirming the view that observers in relativistic circular motion with centripetal acceleration @xmath6 do realize the inertial vacuum as a quasi - thermal state with temperature ( [ temp ] ) for the extremely relativistic case@xcite .
we stress that this state is quasi - thermal in the sense that it can be described by a temperature that varies slowly with the energy gap @xmath90 of the two - level system , monotonically from @xmath91 to @xmath92 , corresponding respectively to low and to high values of @xmath93 @xcite . for the neutron decay , for instance , @xmath94 mev , implying that for circular trajectories close to the photonic orbit @xmath95 , the quasi - thermal state is indeed characterized by a temperature close to the lower bound of ( [ temp ] ) . for the muon decay , on the other hand , the temperature is closer to the upper bound of ( [ temp ] ) . a definitive answer to this problem ,
however , must necessarily face the subtle issue of quantum thermal distributions for rotating systems@xcite .
the authors are grateful to g. matsas for enlightening discussions .
this work was supported by fapesp , cnpq , and ufabc .
is grateful to prof .
v. mukhanov for the warm hospitality at the ludwig - maximilians - universitt , munich , where part of this work was carried out .
kim , k.s .
soh , and j.h .
yee , phys .
d * 35 * , 557 ( 1987 ) . for earlier works on the subject ,
see : j.r .
letaw and j.d .
pfautsch , phys .
d * 22 * , 1345 ( 1980 ) ; j.r .
letaw , phys .
d * 23 * , 1709 ( 1981 ) ; j.r .
letaw and j.d .
pfautsch , phys .
d * 24 * , 1491 ( 1981 ) ; j. math .
* 23 * , 425 ( 1982 ) . c. w. misner , r. a. breuer , d. r. brill , p. l. chrzanowski , h. g. hughes iii and c. m. pereira , phys .
* 28 * , 998 ( 1972 ) .
r. a. breuer , p. l. chrzanowski , h. g. hughes iii and c. w. misner , phys .
d * 8 * , 4309 ( 1973 ) .
r. a. breuer , _ gravitational perturbation theory and synchrotron radiation - lecture notes in physics _ ( springer - verlag , heidelberg , 1975 ) . | we consider here the decay of unstable particles in geodesic circular motion around compact objects . for the neutron , in particular , strong and weak decay
are calculated by means of a semiclassical approach .
noticeable effects are expected to occur as one approaches the photonic circular orbit of realistic black - holes .
we argue that , in such a limit , the quasi - thermal spectrum inherent to extremely relativistic observers in circular motion plays a role similar to the unruh radiation for uniformly accelerated observers . |
transitional disks surrounding pre - main - sequence , solar - type stars are identified by their spectral energy distributions ( seds ) . at 110 microns wavelength ,
their excesses are weaker than those of classical , flat - spectrum t tauri disks , while at longer wavelengths , their fluxes are as strong as those of any classical t tauri sed .
the deficit at infrared wavelengths is consistent with transitional disks having large inner `` holes '' that are fairly transparent in the continuum .
figure 1 illustrates the situation . outside the hole radius @xmath2 ,
optically thick dust abounds .
inside @xmath2 , only trace amounts of optically thin dust are present .
some of the more well - known transitional systems are listed in table 1 ; these include gm aur ( @xmath3 au ; calvet et al .
2005 , hereafter c05 ) , tw hyd ( @xmath44 au ; calvet et al .
2002 , hereafter c02 ; hughes et al . 2007 ;
ratzka et al .
2007 ) , dm tau ( @xmath5 au ; c05 ) , and coku tau/4 ( @xmath6 au ; dalessio et al . 2005 ) .
interferometric measurements verify the existence of inner disk holes ( hughes et al . 2007 ; ratzka et al . 2007
; j. brown , personal communication ) .
transitional disks may bridge the evolutionary gap between conventional t tauri disks that do not have large inner clearings and debris disks that are entirely optically thin .
rim radii are too large to represent dust sublimation fronts , fueling speculation that inside the hole , either ( a ) grains have grown too large for the medium to remain optically thick , ( b ) planets have consumed or torqued away material , and/or ( c ) the inner disk has drained onto the central star by accretion .
this paper concerns possibility ( c)or more accurately , how accretion in transitional disks is ongoing .
transitional disk holes are not empty .
though nearly devoid of dust , the region interior to @xmath2 often contains accreting gas .
near - ultraviolet excesses , when observed , imply stellar accretion rates of @xmath7 ( muzerolle et al .
2000 ; c02 ; bergin et al . 2004 ; c05 ) ; see table 1 . we seek here to explain the origin of such disk accretion .
we propose that gas is leached from the inner disk rim by the magneto - rotational instability ( mri ) , a linear instability that amplifies magnetic fields in outwardly shearing disks and drives turbulence ( balbus & hawley 1998 ) .
x - rays emitted by the hyperactive corona of the central star irradiate and ionize the disk rim , activating the mri there .
magnetically active gas diffuses inward from the rim to the star .
the inner hole grows as the mri eats its way out .
the accretion rate interior to the ever - expanding rim is steady ( constant with radius ) and entirely set by the accretion rate established at the rim .
historically , whether the mri is relevant to t tauri disks has been a prospect fraught with doubt ( see , e.g. , hartmann et al .
2006 ) , as protoplanetary disk gas is typically too cold and/or too dusty to couple well to magnetic fields .
the following two conditions must be met for the mri to be viable .
first , magnetic fields must be frozen into whatever plasma is present . to freeze magnetic flux , the magnetic reynolds number must be sufficiently large : @xmath8 here we follow fleming et al .
( 2000 ) in using the gas sound speed @xmath9 and the vertical density scale height @xmath10 for characteristic velocity and length scales , respectively ; @xmath11 is the kepler orbital frequency ; @xmath12 is the magnetic diffusivity ; @xmath13 is the gas temperature ; @xmath14 is the disk radius ; and @xmath15 is the fractional electron density . numerical simulations by fleming et al .
( 2000 ) suggest @xmath16@xmath17 , depending on the geometry of the seed field ( but see pessah , chan , & psaltis 2007 for a critical re - analysis of results related to the zero - net flux case ; it is possible only the net vertical and net toroidal cases have their values of @xmath18@xmath19 well measured ) . a second criterion more often than not neglected in the protoplanetary disk literature requires that neutral hydrogen molecules , constituting the bulk of disk matter , be dragged inward by accreting plasma ( blaes & balbus 1994 ; mac low et al .
1995 ; hawley & stone 1998 ; kunz & balbus 2004 ) .
that is , ambipolar diffusion of the magnetized plasma out of the ( overwhelmingly ) neutral medium must not defeat the mri . to couple the neutrals to the ions
, a given h@xmath20 molecule needs to collide with enough ions within @xmath0@xmath21 , the e - folding time of the instability : @xmath22 where @xmath23 is the collisional rate coefficient for ions to share their momentum with neutrals ( draine , roberge , & dalgarno 1983 ) and @xmath24 is the number density of hydrogen molecules ( so that @xmath25 is the number density of ions ) .
linear growth rates for the mri drop dramatically with the degree to which criterion ( [ am ] ) is not satisfied ( kunz & balbus 2004 ) .
the critical number @xmath26 of neutral - ion collisions is measured in numerical simulations to be about @xmath27 ( mac low et al . 1995 ; hawley & stone 1998 ) . to our knowledge , these pioneering simulations have not been followed up or verified at higher resolution ; therefore we regard this measurement of @xmath26 , like the measurement of @xmath28 , with caution . criteria ( [ re ] ) and ( [ am ] ) are more easily met the more disk gas is ionized ; criterion ( [ re ] ) depends on the fractional ionization , whereas ( [ am ] ) depends on the total ion density .
transitional disks furnish an excellent first laboratory in which to investigate whether the mri plays a role , if any , in how t tauri disks accrete .
transitional disk holes are less plagued by sub - micron - sized dust that threatens to scour the region of plasma .
furthermore , the rim wall provides an especially simple geometry for analyzing how x - rays ionize disk material : if the wall is not severely shadowed from the star by material interior to the rim ( see [ unresolved ] ) , one need only treat the 1-d radial transfer of radiation perpendicular to the rim wall .
x - ray driven mri is not a new idea it was originally introduced by glassgold , najita , & igea ( 1997 , hereafter gni97)but our study applies it for the first time to transitional disks and to identify the correct criterion for where the mri can operate under these conditions .
we argue that the entire region interior to the rim , including the midplane , is mri - active .
this contrasts with previous studies of x - ray driven ( gni97 ; igea & glassgold 1999 , hereafter ig99 ) or cosmic - ray driven ( gammie 1996 , hereafter g96 ; sano et al .
2000 ) mri in which accretion is confined to the surface layers of a disk whose midplane properties can not be calculated from first principles .
moreover , such surface accretion is unsteady ( g96 ) .
the accretion rates that we derive here are steady and can be directly compared to observations . thus , transitional disks not only provide insight into the late stages of disk evolution , but also offer the first clean application of ideas namely , mri triggered by non - thermal ionization processes ( g96 ; gni97)pioneered for more complex systems . in [ rim ] , we construct a simple model for the rate of mass flow from the x - ray illuminated rim . we solve the equations of ionization and energy balance to decide how much of the rim satisfies criteria ( [ re ] ) and ( [ am ] ) and accretes .
in addition to accounting for the usual dissociative recombination of molecular ions with free electrons ( e.g. , gni97 ) , our treatment of ionization balance includes recombination onto dust grains ( umebayashi & nakano 1980 ) and charge transfer to free atomic metals ( oppenheimer & dalgarno 1974 , hereafter od74 ; fromang , terquem , & balbus 2002 ) .
we show how criterion ( [ am ] ) , neglected in many studies of protoplanetary disks ( e.g. , g96 ; gni97 ; ig99 ; sano et al .
2000 ; fromang et al . 2002 ; matsumura & pudritz 2003 ) , supersedes criterion ( [ re ] ) .
we establish a relation between accretion rate @xmath29 and rim radius @xmath2 and compare to observations . in [ steady ] ,
we demonstrate how the mri continues to transport gas inward at radii @xmath30 .
though the midplane of the interior disk is optically thick to incident x - rays , compton scattering permits enough x - rays to reach the midplane ( ig99 ) that the mri operates there as well .
accretion is therefore steady from the rim to the star . in
[ rad ] , we study in a preliminary way how stellar radiation pressure blows dust grains out of accreting gas and can keep the disk inside the rim relatively transparent .
we make a connection to the small but non - zero optical depth inferred for the inner disk from sed modelling ( c02 ; c05 ) to constrain the abundance of micron - sized grains in the rim .
our work leaves many issues unresolved and calls attention to outstanding questions associated with the mri .
these are discussed in [ conc ] , after a summary of our principal findings .
ionization of rim gas is maintained by stellar x - rays.@xmath31 per h@xmath20 ( spitzer 1978 , p. 116 ) .
actual cosmic ray ionization rates in t tauri disks are likely much lower because of shielding by magnetized t tauri winds .
even the much weaker solar wind modulates the cosmic ray flux at earth by as much as @xmath010% with solar cycle ( marsh & svensmark 2003 ) .
we ignore cosmic ray ionization ; it is easy to check that their contribution is negligible compared to ionization by stellar x - rays . ] as measured by _
chandra _ , the x - ray spectra of t tauri stars in orion often have separate soft and hard components , modelled as emission from thermal plasmas having characteristic energies @xmath32 and @xmath33 , where @xmath34 is boltzmann s constant ( wolk et al .
2005 , see their figure 12 ) .
the luminosities of the two components are comparable , and depending on the star , range from @xmath35 to @xmath36 .
soft , sub - kev x - rays are thought to be emitted by shock - heated gas in accretion columns ( kastner et al .
2002 ; stelzer & schmitt 2004 ) , while hard , super - kev photons may arise from magnetic flares in stellar coronae ( e.g. , wolk et al .
2005 ) . as the penetrating power of x - rays increases rapidly with photon energy , we are primarily interested in the hard , presumably coronal emission
. unfortunately , we are unable to locate detailed x - ray spectra at photon energies @xmath37 for the prototypical transitional systems discussed in this paper ( gm aur , tw hyd , dm tau , and coku tau/4 ) .
what we glean from the literature is summarized in table 1 . in the absence of detailed spectra , we assume for our computations below that each source emits @xmath38 at a characteristic energy @xmath39 .
this assumption is supported circumstantially by the aforementioned _ chandra _ observations of hard x - ray emission from orion pre - main - sequence stars .
it is , furthermore , consistent with our findings in table 1 , even for tw hyd , whose spectrum is known to be atypically soft ( stelzer & schmitt 2004 ) .
lccccccc gm aur & 24 & 1 & 2.710 & 1,2 & @xmath40 & 0.22.4 & 3 + tw hyd & 14@xmath41 & 4,5,6 & 0.30.7@xmath42 & 4,7 & @xmath43 & 0.452.25 & 8 + dm tau & 3 & 1 & 0.82 & 1,2 & @xmath44 & 0.310 & 9 + coku tau/4 & 10 & 10 & @xmath45 & 11 & na & na & 12 + stellar x - rays penetrate the exposed rim wall to a hydrogen column density @xmath46 ( see figure 1 ) .
a limited column @xmath47 will satisfy criteria ( [ re ] ) and ( [ am ] ) .
the mri - active rim contains mass @xmath48 , where @xmath49 is the vertical density scale height , @xmath50 is the gas sound speed , @xmath11 is the kepler orbital frequency , @xmath51 is the kinetic temperature of mri - active gas at the rim , and @xmath52 is the gas mean molecular weight .
this mass flows from @xmath2 to @xmath0@xmath53 over the diffusion time @xmath54 , where @xmath55 is the turbulent diffusivity : @xmath56 where @xmath57 is the gravitational constant and the extra factor of 3 follows from a more accurate derivation ( using @xmath58 , where @xmath59 is the disk surface density ; frank , king , & raine 1992 ) . as measured by numerical simulations of the mri , the transport parameter
@xmath1 might be @xmath00.010.3 , depending on the strength and geometry of the background field ( fleming et al .
2000 ; sano et al .
2004 ; pessah et al .
2007 ; see also [ unresolved ] ) . notice how @xmath29 in equation ( [ m_dot ] ) does not depend explicitly on the surface density profile @xmath60 of the disk .
given observations of @xmath2 and @xmath61 , one need only compute the rim - specific variables @xmath47 and @xmath51 , which we do below .
determining @xmath47 requires that we calculate the degree of ionization as a function of the radial column @xmath46 penetrated by x - rays at the rim . per h@xmath20 molecule ,
the ionization rate is @xmath62 where @xmath63 is the photoionization cross - section per h@xmath20 ( assuming solar abundances ) and @xmath64 accounts for the number of secondary ionizations produced per absorbed 3-kev photon ( gni97 ) .
freshly ionized h@xmath65 rapidly converts to molecular ions such as hco@xmath66 .
most of these molecules dissociate in collisions with electrons , while some transfer their charge to gas - phase atomic metals such as magnesium ( fromang et al . 2002 ; od74 ) .
the fractional number abundance @xmath67 of molecular ions relative to neutral hydrogen molecules is given by the equilibrium relation @xmath68 where @xmath69 and @xmath70 are rate coefficients for dissociation ( glassgold , lucas , & omont 1986 ) and charge transfer ( od74 ) , and @xmath71 and @xmath72 are the fractional number densities of electrons and free neutral metals , respectively . to relate the number density @xmath24 of neutral h@xmath20 molecules to @xmath46 , we spread all @xmath73 molecules comprising the x - ray penetrated rim into the volume @xmath74 interior to the rim to estimate that @xmath75
. charge transfer to free metals is important for sustaining the mri because ionized metals tend to keep their charge , neutralizing slowly either by radiative recombination with electrons or collisions with negatively charged dust grains ( od74 ; umebayashi & nakano 1980 ; glassgold et al .
1986 ) : @xmath76 where @xmath77 is the fractional abundance of metal ions , @xmath78 is the radiative recombination coefficient , and @xmath79 is the recombination coefficient , measured per h@xmath20 , for grains .
we assume the total grain surface area available for recombination is dominated by grains of radius @xmath80 containing a fraction @xmath81 of the total mass in gas and dust .
it follows that @xmath82 .
this estimate , which assumes that every grain is negatively charged , agrees with that of umebayashi & nakano ( 1980 ) , who assume @xmath83 and @xmath84 .
though unknown , the factor @xmath81 is likely to be considerably less than the maximal value permitted by solar abundance gas , @xmath85 , because of grain growth .
( which is measured per h@xmath20 ) .
] we take as standard values @xmath86 and @xmath87 . a constraint on @xmath88 from observations is contained in [ rad ] on radiation blow - out . in assuming a single grain size , we neglect effects introduced by smaller - sized grains .
the surface area for ion recombination may be dominated by very small grains .
furthermore , because grains efficiently adsorb electrons , they can , if sufficiently numerous , completely clear the gas of free electrons and catastrophically reduce the electrical conductivity . indeed , this is the case in the earth s lower atmosphere , where charges are carried by aerosols .
fortunately , as we describe below in [ acrar ] , our results enjoy some margin of safety from these effects .
the fractional electron densities that we derive , @xmath89@xmath90 ( compared , say , with terrestrial ionization fractions @xmath0@xmath91 at sea level ) , are large enough that the chemistry is arguably driven by electrons ( but see [ unresolved ] ) .
charge and number conservation read , respectively , @xmath92 and @xmath93 because free metals are depleted onto grains , @xmath94 is realistically at most @xmath0@xmath95 , corresponding to 1 out of every 10 metals in the free atomic phase ( od74 ; umebayashi & nakano 1980 ) .
we consider values for @xmath94 ranging from zero ( complete depletion ) up to @xmath96 .
we combine equations ( [ mol ] ) , ( [ met ] ) , ( [ charge ] ) , and ( [ number ] ) to derive , under certain approximations , a quartic equation for @xmath15 in terms of @xmath97 , @xmath94 , and various rate coefficients . equations ( [ mol ] ) , ( [ met ] ) , and ( [ charge ] ) yield @xmath98 the first term on the left - hand side dominates the second term when @xmath99 this is always the case for mri - active rim material , as will be evident below ( see figures 2 and 3 )
. therefore we drop the second term in ( [ rewrite ] ) to write @xmath100 combining equations ( [ mol ] ) , ( [ met ] ) , and ( [ number ] ) , we have @xmath101 \,,\end{aligned}\ ] ] which simplifies , according to the same approximation embodied in ( [ approx ] ) , to @xmath102 \ , . \label{end_part2}\ ] ] we solve ( [ end_part2 ] ) for @xmath103 and substitute into ( [ end_part1 ] ) to write @xmath104 it is safe to ignore @xmath105 in comparison to @xmath106 , so that @xmath107 this final equation is a quartic and not a cubic ( od74 ) because we allow for the possibility that nearly all gas - phase atomic metals might be ionized , via equation ( [ number ] )
. gas is heated by fast photoelectrons at a rate per h@xmath20 molecule of @xmath108 where @xmath109 is the fraction of x - ray energy deposited into heat ( glassgold , najita , & igea 2004 ) .
gas cools by ro - vibrational transitions of co at a rate per h@xmath20 molecule of @xmath110 ( glassgold et al .
2004 , their equations c7c8 ) . here
@xmath111 is the fractional abundance of co , whose level populations are assumed thermal . other cooling channels
dust - gas collisions and lyman @xmath1 emission from hydrogen are less important and ignored .
we replace @xmath24 in ( [ cool ] ) with our estimate @xmath112 ( see [ ionbal ] ) , set @xmath113 in thermal balance , and solve for @xmath13 as a function of @xmath46 .
this result is necessary for solving the quartic ( [ quart ] ) for @xmath15 because rate coefficients depend on @xmath13 . for our standard parameters for gm aur ( @xmath38 , @xmath39 , @xmath114 )
, we find that @xmath13 drops from about 340 to 200 k as @xmath46 increases from @xmath115 to @xmath116 .
results for @xmath13 are very similar for tw hyd and dm tau .
the insensitivity to input parameters is due to the exponential sensitivity of the cooling rate @xmath117 to @xmath13 . taking the run of @xmath13 with @xmath46 from our simple model of thermal balance in [ thermal ] , we solve the quartic ( [ quart ] ) numerically .
the solution @xmath71 is displayed against @xmath46 in figure 2 for parameters appropriate to gm aur , for various choices of @xmath94 .
evidently the presence of free atomic metals raises the ionization fraction by up to two orders of magnitude .
results for tw hyd and dm tau are similar to within factors of two . from @xmath118
we compute @xmath119 , as shown in figure 3 for the case of gm aur .
if @xmath120 strictly determined the viability of the mri , then we would equate @xmath47 with the maximum value of @xmath46 for which @xmath121 .
however , given uncertainties in the value of @xmath26 and in the sharpness of criterion ( [ am ] ) in predicting the non - linear , saturated outcome of the mri , only an order - of - magnitude estimate for @xmath122 ( the value for which @xmath123 peaks ) seems justifiable .
we adopt this same value of @xmath47 for tw hyd and dm tau , for which very similar results for @xmath119 obtain . for @xmath124
, we find from our thermal balance model that @xmath125 .
the ionization fractions , @xmath126@xmath89 , corresponding to @xmath127 are so large that magnetic reynolds numbers , @xmath128@xmath129 , far exceed the critical values , @xmath130@xmath17 , seemingly required for magnetic flux freezing ( fleming et al .
. therefore accretion by the mri in transitional disks is limited by ambipolar diffusion , not by ohmic dissipation .
electron densities exceed grain densities by 89 orders of magnitude , for our standard @xmath86 and @xmath87 . our neglect of a population of aerosols which can scour the gas of electrons is arguably safe .
the differential grain size distribution , if it scaled as @xmath131 , would have to extend from @xmath132 all the way down to several angstroms for the total grain density to be comparable to the electron density .
our findings are also insensitive to the uncertain grain recombination coefficient @xmath79 , because metallic ions recombine more readily with electrons than with grains . for example
, increasing @xmath79 by a factor of @xmath133 above our nominal value ( such an increase would correspond to the same grain size distribution mentioned above ) decreases the peak value of @xmath134 in figure 3 by a factor of 2 , from 140 to 70 .
armed with @xmath47 and @xmath51 , we use equation ( [ m_dot ] ) to plot @xmath29 against @xmath2 in figure 4 , adjusting @xmath1 as necessary to reproduce observed accretion rates for gm aur , tw hyd , and dm tau .
the offsets between the different lines drawn for different systems arise mostly from differences in @xmath61 and @xmath1 , since @xmath135 is held fixed and @xmath136 varies little .
the slopes of the lines ( @xmath137 , approximately ) reflect essentially the increase in the rim wall s surface area with @xmath2 .
best - fit @xmath1 s are 0.0070.035 .
such values are of order those seen in numerical simulations of the mri , depending on the strength and geometry of the background field ( fleming et al .
2000 ; sano et al . 2004
; pessah et al . 2007 ;
see also [ unresolved ] ) .
s resemble those required to fit the light curves of dwarf novae during their `` cold '' states ( @xmath1380.04 ; lasota 2001 ) .
this resemblance seems sensible insofar as white dwarf accretion disks during their quiescent phases and protoplanetary disks both contain relatively cold , poorly ionized gas .
however , gammie & menou ( 1998 ) argue that dwarf nova disks in quiescence do not support the mri .
these authors consider the relevant magnetic reynolds numbers , @xmath139 , too low .
our own view is that such reynolds numbers do not proscribe a weakened form of the mri ( see also fleming et al .
2000 ) . ]
the transitional system coku tau/4 is not detected in x - rays ( knig et al .
2001 ) and has an unmeasurably small accretion rate ( @xmath140 ; najita et al . 2007 ; dalessio et al .
2005 ) , facts that at face value are consistent with our theory .
possibly coku tau/4 has fewer free atomic metals or and this is more readily testable a softer x - ray spectrum than the other sources .
if , say , @xmath141 at its rim , then we estimate its 3-kev luminosity @xmath142 .
we expect little accreting material outside @xmath2 .
the x - ray heated rim has a vertical thickness greater than that of material immediately outside .
thus , as depicted in figure 1 , some fraction of the disk beyond the rim will dwell in the rim s x - ray shadow and be magnetically inert .
whatever distant material lies outside the rim s shadow and might be accreting from x - ray irradiation is ignored . see also [ sum ] for a reason to doubt that the outermost regions of disks accrete .
once dislodged from the rim , gas must still travel up to three decades in radius , from @xmath010 to @xmath00.01 au , to reach the stellar surface .
we now argue that x - ray - driven mri continues to provide transport at @xmath30 . at @xmath00.01
au , the mri should be well sustained by thermal ionization .
it remains to decide whether criteria ( [ re ] ) and ( [ am ] ) are satisfied at the midplane at radii of @xmath00.1 to @xmath01 au . we need to estimate the temperature and surface density of material there .
observed spectra demand that inside the rim , the disk contain so few grains as to be optically thin at mid - infrared wavelengths ( c02 ; c05 ) .
we can understand this as a consequence both of the limited number of grains contained within the mri - active mass @xmath143 , as well as stellar radiation pressure , which blows out a large fraction of grains having sizes @xmath144 ( see [ rad ] ) .
having lost its primary source of continuum opacity , the gas heats not by incident starlight , but by accretion ( ohmic dissipation ) .
cooling proceeds through line emission , at a rate that is difficult to estimate because it requires solving simultaneously for the thermal , chemical , and excitation state of the gas .
we do not attempt to construct a full - blown thermal model of electrically resistive , line - emitting gas .
instead , we find it adequate to estimate the temperature of midplane gas at @xmath30 in terms of its minimum possible value , obtained by equating the energy flux from accretion with that emitted by a blackbody : @xmath145 where @xmath146 parameterizes our uncertainty .
mass continuity then implies a vertical hydrogen column density of @xmath147 to the midplane .
we have normalized @xmath1 to the average of values obtained for gm aur , tw hyd , and dm tau ( figure 4 ) .
encouragingly , equation ( [ nperp ] ) yields surface densities that are entirely consistent with those inferred by salyk et al .
( 2007 ) for gm aur and tw hyd : at @xmath1480.5 au , these authors find @xmath149@xmath150 ( h@xmath20 surface densities of @xmath151@xmath152 assuming @xmath153 ) , based on observations of rovibrational emission from co. these same observations suggest that @xmath154 ( see their figure 2 ) , and we normalize our results below to this value . to estimate @xmath155 and @xmath134 at the midplane , we first obtain the ionization rate @xmath156 from ig99 , who calculate by monte carlo methods how x - rays incident upon disk surfaces propagate vertically and radially through protoplanetary disks .
their calculations account for multiple compton scattering off bound electrons and for a thermal spectrum of photon energies . according to their figure 3 , for ( a ) a @xmath157 plasma emitting @xmath38 , ( b ) @xmath158 as given by equation ( [ nperp ] ) , and ( c ) @xmath159 , the ionization rates at the midplane equal @xmath160 at @xmath161 au and @xmath162 at 0.1 au .
we insert these rates into the quartic equation ( [ quart ] ) for @xmath71 , setting @xmath163 because dust is largely absent in the inner disk . the quartic yields @xmath164 at 1 au and @xmath165 at 0.1 au , for @xmath166 .
these ionization fractions correspond to @xmath167 at 1 au and @xmath168 at 0.1 au , values that satisfy criterion ( [ re ] ) , albeit marginally .
finally , taking @xmath169 , we find that @xmath170 at 1 au and @xmath171 at 0.1 au .
these values for @xmath134 , which marginally satisfy criterion ( [ am ] ) , are insensitive to variations in @xmath172 , which produce changes in @xmath24 that nearly cancel changes in @xmath71 . the margins of safety for @xmath155 and @xmath134 would increase by accounting for thermal ionization .
we conclude that the mri plausibly operates everywhere interior to the rim , even at the disk midplane , carrying steadily inward all of the mass drawn from the rim .
note how the mri - active vertical columns at 0.11 au , @xmath173 , are larger than the mri - active radial column at the rim , @xmath122 .
this follows from criterion ( [ am ] ) , which assigns importance to the total density @xmath174 , not the fractional density @xmath71 .
total densities increase dramatically from the rim to the star .
gas drawn from the rim entrains dust grains . as the surface density of inspiralling gas increases towards the star , so , too , does the surface density of grains .
left unchecked , the increasing concentration of grains would render the inner disk optically thick at infrared wavelengths , violating observations . here
we explore how stellar radiation pressure can purge grains from gas ( see also eisner , chiang , & hillenbrand 2006 ) .
sub - micron - sized grains around solar - type stars feel an outward stellar radiation force that just exceeds stellar gravity .
then the time for such grains to travel from @xmath14 to @xmath0@xmath175 is @xmath176 where @xmath177 is the time for grains of radius @xmath80 and internal density @xmath178 to attain terminal velocity according to the epstein gas drag law , which applies here because the grain size is smaller than the collisional mean free path in gas ( e.g. , weidenschilling 1977 ) .
radiation blow - out is faster than aerodynamic drift ( the latter caused by radial pressure gradients in gas ; e.g. , alexander & armitage 2007 ) by @xmath179 , and so we ignore the latter .
we compare @xmath180 to @xmath181 , the time for gas to diffuse from @xmath14 to @xmath0@xmath182 . at the rim of the disk of gm aur ,
the timescales match by coincidence : @xmath183 .
this indicates that about half of the dust in @xmath143 is expelled , leaving the other half entrained with accreting gas and spread between @xmath2 and @xmath0@xmath53 .
the geometric optical depth of entrained dust , measured perpendicular to the midplane , is @xmath184 we can try to check our estimate for @xmath185 against observations of the mid - infrared sed .
because radiation at a wavelength of @xmath186 originates from grains having temperatures of @xmath0300 k , observed @xmath186 spectra of gm aur constrain the optical depth in grains at @xmath01 au but not near the rim at 24 au , where starlight - heated grains would have colder temperatures of @xmath0100 k. therefore we can not directly compare our calculated @xmath185 with @xmath186 observations for gm aur . for the case of tw hyd
, however , we can more easily make this comparison , because its rim is located at @xmath04 au . scaling to the parameters of that system , we find @xmath187 and @xmath188 , which suggests that somewhat more than half of the dust in @xmath143 is expelled .
reducing our estimate in ( [ tau ] ) by an additional factor of 2 , and accounting for the smaller aspect ratio @xmath189 , we estimate @xmath190 for tw hyd .
observationally , the vertical optical depth of dust interior to the rim of tw hydra s disk is @xmath191 at a wavelength of @xmath186 ( c02 ) .
this is essentially the same as the geometric optical depth for micron - sized grains because of the silicate resonance band at @xmath186 wavelength .
thus , our crude estimate of @xmath190 , based on assumed values for @xmath86 and @xmath87 , is within a factor of two of the observed optical depth .
such approximate agreement we consider acceptable , and helps to justify our choice earlier for @xmath192 in calculating the grain recombination coefficient @xmath79 ( [ ionbal ] ) . to satisfy the observation that the disk remain optically thin at @xmath30
, the order - half of rim dust that lies between @xmath2 and @xmath53 must fail to penetrate inside @xmath193 .
radiation pressure ensures that dust does not continue to inspiral , provided that @xmath194 decrease with decreasing @xmath14 .
this seems likely to obtain because @xmath13 should drop sharply just inside @xmath53 once midplane gas becomes too dense to be heated effectively by x - rays .
unfortunately , a conclusive statement can not be made without constructing a model that smoothly bridges conditions at @xmath195 with conditions at @xmath30 .
this would require knowing in detail how gas heats and cools at all radii .
we summarize our results in [ sum ] and discuss topics for future research in [ unresolved ] .
how do t tauri disks accrete ?
we have answered this longstanding question for the sub - class of objects known as transitional systems , disks having au - sized or larger inner holes cleared of dust .
super - kev x - rays , emitted by the young star as a consequence of its enhanced dynamo , irradiate the inner disk rim .
a portion of the rim is sufficiently ionized that it becomes unstable to the magneto - rotational instability ( mri ) and diffuses inward .
the mri , sustained even at the disk midplane by hard x - rays , transports gas steadily from the rim all the way to the stellar surface .
we infer transport parameters @xmath1 on the order of a percent ; such values are indeed yielded by the mri , according to contemporary numerical simulations .
we feel the following features of our theory deserve emphasis . 1
. _ accretion rate vs. hole size . _ all other factors ( e.g. , x - ray luminosity , stellar mass , free atomic metal abundance ) being equal , the accretion rate @xmath29 should increase nearly as the square of the rim radius @xmath2 , simply from the increased surface area of the rim wall that is exposed to stellar x - rays .
late - breaking data ( c. espaillat , personal communication ) for the transitional disks cs cha ( @xmath196 , @xmath197 ) and sz 18 ( @xmath198 , @xmath199 ) appear to support this trend of @xmath29 with @xmath2 as plotted in our figure 4 .
of course , this trend is expected to apply only for transitional disks for which @xmath200 au ( see the last issue raised in [ unresolved ] ) .
ambipolar diffusion vs. ohmic dissipation .
_ often it is assumed that ohmic dissipation controls where the mri can and can not operate .
this criterion , embodied in equation ( [ re ] ) through the magnetic reynolds number @xmath201 , assigns importance to @xmath15 , the ratio of the density of the most mobile charge carriers ( electrons ) to that of resistive species ( neutral h@xmath20 molecules ) .
large values of @xmath202 imply that magnetic flux is frozen into plasma , enabling the background shear flow to amplify magnetic fields .
+ we have shown instead that ambipolar diffusion , not ohmic dissipation , dictates the boundaries of mri - active zones .
magnetic flux can be perfectly frozen into plasma ; yet if the neutrals do not couple mechanically to the plasma , the neutrals would be immune to the mri .
protoplanetary disk gas is overwhelmingly neutral .
thus , as equation ( [ am ] ) states , a given neutral h@xmath20 molecule must collide with as many ( momentum - bearing ) ions as possible within a dynamical time for the mri to afflict protoplanetary disk gas .
ambipolar diffusion limits the accretion flow from the rim , and is only marginally defeated at the midplane far inside the rim .
+ while we have demonstrated that criterion ( [ am ] ) outweighs ( [ re ] ) only for transitional disks , we suspect the same conclusion holds for conventional t tauri disks as well .
for example , consider figure 7 of ig99 , which seeks to describe the depth of the mri - active layer in disks without holes .
at , say , @xmath203 , at the bottom of the supposedly active layer of their model , the number density is @xmath204 . by ig99 s equation ( 23 ) , the ionization fraction there is @xmath205 ( for their standard @xmath206 ) .
it follows that @xmath207 .
this falls four orders of magnitude short of the critical @xmath208 seemingly demanded by numerical simulations ( mac low et al .
1995 ; hawley & stone 1998 ) .
therefore we disagree with ig99 s determination of the depth of the active layer ; their layer does not satisfy criterion ( [ am ] ) .
+ because the ambipolar diffusion criterion ( [ am ] ) assigns importance to the total ion density rather than the fractional electron density , it is hardest to satisfy in the most rarefied regions of disks , either at large vertical heights or large radial distance .
this is just opposite to what has been concluded in the literature , that mri - active zones are restricted to the surfaces and the outermost portions of disks .
_ disk properties inside the rim are insensitive to those outside , since the mri can only draw a radial column of @xmath209 from the rim at any time . in other words , unlike the usual situation , estimating @xmath29 does not require specifying the detailed surface density profile of the disk @xmath60 ; we need only know those rim - specific variables in equation ( [ m_dot ] ) : @xmath2 , @xmath1 , @xmath210 , and @xmath211 .
our picture therefore provides a robust setting for theories of how planets form and how their orbits evolve ( e.g. , goldreich & sari 2003 ) within transitional disks .
a protoplanet lying interior to the rim will interact with gas whose density , temperature , and transport properties are definite and decoupled from uncertain initial conditions .
4 . _ how disks dissipate .
_ our study also supplies part of the answer to how disks dissipate . excepting matter that gets locked into planets , the inner disk drains from the inside out by the mri , while material beyond the rim photoevaporates by stellar ultraviolet radiation ( alexander , clarke , & pringle 2006ab ) .
indeed , the models of alexander et al .
( 2006ab ) require the disk inside a few aus to drain onto the star before ultraviolet radiation can photoevaporate the outer disk efficiently .
their analysis did not specify the means by which the inner disk evacuates .
our work identifies the mechanism : it is the mri , leaching material from the disk rim and causing the inner hole to grow .
how long the mri takes to eat its way out depends on the unknown surface density @xmath59 of the original disk . the clearing time @xmath212
if @xmath213 at @xmath214 , comparable to that of the minimum - mass solar nebula . 1 .
_ radiation blow - out .
_ we have established that a fraction on the order of 50% of the sub - micron - sized grains in the mri - active rim gets blown back into the rim by stellar radiation pressure .
the remaining fraction , which gets entrained to @xmath0@xmath53 by the accretion flow , ultimately must also be blown back to ensure that transitional disk holes stay as transparent as observed . to prove this last point requires understanding in detail how gas heats and cools at all radii from @xmath2 to the stellar surface , and checking that @xmath194 decreases with decreasing @xmath14 .
self - shadowing .
_ throughout this work we have assumed for simplicity that the accretion disk interior to the rim does not obstruct the rim wall from stellar x - rays . yet there must be some self - shadowing , which if sufficiently severe will convert the problem we have solved to one involving surface irradiation and the two - dimensional radiative transfer of x - rays ( ig99 ) . even if this were the case , however , midplane gas at the rim may still be mri - active by virtue of x - rays that compton scatter downward from the disk surface ( ig99 ; [ steady ] ) .
+ the rim might be tall enough that a good fraction of its lowermost vertical scale height escapes being shadowed . in our picture ,
rim gas is optically thin to x - rays , and is therefore well - heated by them ( see also glassgold et al .
this source of heating is not enjoyed by midplane gas in the inner disk , which is optically thick to x - rays .
thus we envision a sharp drop in temperature and a concomitant drop in disk thickness just inside the rim . as was the case for radiation blow - out , addressing the problem of self - shadowing requires a global treatment of thermal balance .
the value of @xmath1 .
_ we have estimated that @xmath1 s on the order of a percent are required to explain the accretion rates of gm aur , tw hyd , and dm tau .
such values are observed in numerical simulations of the mri under certain assumptions about the magnitude and geometry of the seed magnetic field .
are these assumptions justified ? what sets the seed field parameters ? and how far can one trust the simulations themselves ? + using vertically unstratified simulations performed by sano et al .
( 2004 ) , pessah et al .
( 2007 ) provide a remarkably simple fitting formula for @xmath1 , which states that its saturation value scales as the background @xmath215 , where @xmath216 is the mean strength of an assumed vertical magnetic field , @xmath217 is the time - averaged gas pressure , and the simulation box length @xmath218 is set equal to the pressure scale height @xmath219 .
values of @xmath220 correspond to @xmath221 , or equivalently a plasma @xmath222 .
is such a background state realistic ?
can one determine analogous requirements based on simulations that assume a net toroidal field , and ultimately on simulations that include vertical gravity ?
4 . _ the value of @xmath26 . _
is @xmath26 truly and exactly 100 ?
what is the origin of this dimensionless number ?
how sharp is criterion ( [ am ] ) in predicting the non - linear outcome of the mri ? according to our present calculations , criterion ( [ am ] ) for @xmath208 is only marginally satisfied at the rim and interior to the rim .
we need to know the transport parameter @xmath1 as a function of @xmath123 ( see also point 3 above ) to calculate more precisely the rate of gas flow from the rim .
the value of @xmath223 .
_ how can one measure the abundance of gas - phase atomic metals in transitional disks ?
our finding that the mri is viable depends on having more than 1 out of every @xmath0300 metal atoms such as magnesium remaining in the atomic gas phase .
these free metals provide the ions that collisionally couple the neutrals to the mri .
6 . _ aerosols .
_ what is the abundance of macromolecules / grains having sizes @xmath0@xmath224 to @xmath0@xmath225 and is it large enough to interfere with the standard electron - driven chemistry we have presented ?
on the optimistic side , transitional disks are observed to have much less micron - sized dust than conventional t tauri disks .
we have calculated that fractional electron densities in the rim are large enough to withstand , to some degree , the effects of electron adsorption and ion recombination onto grains . on the other hand ,
radiation pressure can not purge gas of small aerosols because of their short aerodynamic stopping times and low efficiencies for intercepting radiation .
. _ accretion rate vs. x - ray luminosity . _
any mature theory of x - ray - driven mri should predict @xmath29 as a function of @xmath226 . but without knowing @xmath227 ( see points 3 and 4 above ) , we can not derive this relation with confidence . at the very least @xmath29 must increase with @xmath226 .
is there observational evidence for such a trend ?
in fact , @xmath29 is known to increase with @xmath61 ( e.g. , hartmann et al .
2006 ) . and @xmath226 is observed to increase with @xmath61 ( guedel et al . 2007 ; telleschi et al .
but frustratingly enough , when @xmath29 is plotted against @xmath226 , whatever trend might be present is obscured by large scatter ( telleschi et al . 2007 ) .
+ a shorter - term project would be to measure the hard , @xmath37 spectra of individual sources .
the penetration depth of x - rays scales strongly with photon energy , as @xmath228 .
_ from conventional t tauri disks to transitional disks .
_ we have studied the equilibrium dynamics of transitional disks but have said little regarding origin .
presumably transitional disks evolve from conventional t tauri disks .
are transitional systems , as a group , older ?
the age of tw hyd is estimated to be approximately 10 myr , which indeed is old for a t tauri star . + how does the evolution from conventional t tauri to transition system take place ?
unfortunately , we can not answer this question by merely extrapolating our theory to earlier times .
our theory is designed to explain systems with large , well - defined holes , and would break down as @xmath2 approaches the stellar radius , if only because rim gas would be ionized thermally rather than by x - rays .
data for conventional t tauri disks clearly do not fit the relations plotted in figure 4 ; their rim radii are on the order of @xmath00.1 au but their accretion rates range up to @xmath0@xmath229
. the origin of transitional disks will likely remain a mystery until we solve the decades - old puzzle of how conventional t tauri disks , objects without large holes , accrete ( see , e.g. , hartmann et al .
the solution might involve some combination of the mri plus gravitational instability ( e.g. , g96 ) .
we thank al glassgold , david hollenbach , uma gorti , eric feigelson , martin pessah , catherine espaillat , joanna brown , and colette salyk for informative discussions .
steve balbus and an anonymous referee provided extensive and insightful criticisms that greatly improved the content of our manuscript .
this work was motivated by lectures prepared by one of us ( eic ) for the hebrew university winter school for theoretical physics , and supported by grants from the national science foundation and the alfred p. sloan foundation . | how do t tauri disks accrete ?
the magneto - rotational instability ( mri ) supplies one means , but protoplanetary disk gas is typically too poorly ionized to be magnetically active .
here we show that the mri can , in fact , explain observed accretion rates for the sub - class of t tauri disks known as transitional systems .
transitional disks are swept clean of dust inside rim radii of @xmath010 au .
stellar coronal x - rays ionize material in the disk rim , activating the mri there .
gas flows from the rim to the star , at a rate limited by the depth to which x - rays ionize the rim wall .
the wider the rim , the larger the surface area that the rim wall exposes to x - rays , and the greater the accretion rate .
interior to the rim , the mri continues to transport gas ; the mri is sustained even at the disk midplane by super - kev x - rays that compton scatter down from the disk surface .
accretion is therefore steady inside the rim . blown out by radiation pressure
, dust largely fails to accrete with gas .
contrary to what is usually assumed , ambipolar diffusion , not ohmic dissipation , limits how much gas is mri - active .
we infer values for the transport parameter @xmath1 on the order of a percent for the prototypical systems gm aur , tw hyd , and dm tau .
because the mri can only afflict a finite radial column of gas at the rim , disk properties inside the rim are insensitive to those outside .
thus our picture provides one robust setting for planet - disk interaction : a protoplanet interior to the rim will interact with gas whose density , temperature , and transport properties are definite and decoupled from uncertain initial conditions .
our study also supplies half the answer to how disks dissipate : the inner disk drains from the inside out by the mri , while the outer disk photoevaporates by stellar ultraviolet radiation . |
electromagnetic cascades , namely the evolution of @xmath0 particle numbers and energy distribution following the injection of a energetic @xmath1 or @xmath2 in a medium filled with radiation , magnetic fields and matter , is one of the physical processes most frequently encountered in astroparticle physics , in domains as disparate as high - energy gamma - ray astrophysics , ultra - high - energy cosmic ray propagation , or the physics of the early universe . in particular , the elementary theory of such a cascade onto a photon background has been well known since decades , and can be shown via a textbook derivation ( see chapter viii in @xcite , for instance ) to lead to a universal `` meta - stable '' spectrum attained on timescales much shorter than the thermodynamical equilibration scale of the form : @xmath3 in the above expression , @xmath4^{-1}$ ] is a normalization constant enforcing the condition that the total energy is equal to the injected electromagnetic energy , @xmath5 ; the characteristic energy @xmath6 denotes the effective threshold for pair - production ( @xmath7 being the highest energy of the photon background onto which pairs can be effectively created ) ; @xmath8 is the maximum energy of up - scattered inverse compton ( ic ) photons .
natural units with @xmath9 are used throughout .
a notable application of this formalism concerns the possibility of a non - thermal nucleosynthesis phase in the early universe ( for recent review on this and other aspects of primordial nucleosynthesis , or bbn , see @xcite ) .
the determination of the baryon energy density of the universe @xmath10 inferred from the cmb acoustic peaks measurements can be used in fact to turn the standard bbn into a parameter - free theory .
the resulting predictions for the deuterium abundance ( or @xmath11h , the most sensitive nuclide to @xmath10 ) are in remarkable agreement with observations , providing a tight consistency check for the standard cosmological scenario .
the @xmath12he and @xmath13he yields too are , broadly speaking , consistent with this value , although affected by larger uncertainties .
the @xmath14li prediction , however , is a factor @xmath15 above its determination in the atmosphere of metal - poor halo stars .
if this is interpreted as reflecting a cosmological value as opposed to a post - primordial astrophysical reprocessing , a question which is far from settled @xcite it requires a non - standard bbn mechanism , for which a number of possibilities have been explored @xcite . in particular , cosmological solutions based on electromagnetic cascades
have been proposed in the last decade , see for instance @xcite .
however , typically they do not appear to be viable @xcite , as confirmed also in recent investigations ( see for instance fig . 4 in @xcite , dealing with massive `` paraphotons '' ) due to the fact that whenever the cascade is efficient in destroying enough @xmath14li , the destruction of @xmath11h is too extreme , and spoils the agreement with the cmb observations mentioned above .
actually , this tension also affects some non - e.m .
non - thermal bbn models , see for instance @xcite .
this difficulty can be evaded if one exploits the property that @xmath14be ( from which most of @xmath14li come from for the currently preferred value of @xmath10 , via late electron capture decays ) has the lowest photodissociation threshold among light nuclei , of about @xmath16mev vs. @xmath17mev for next to most fragile , @xmath11h .
hence , to avoid any constraint from @xmath11h while being still able to photo - disintegrate some @xmath14be , it is sufficient to inject photons with energy @xmath18 , with a `` fine - tuned '' solution ( see e.g. the remark in @xcite or the discussion in @xcite ) .
nonetheless , it turns out to be hard or impossible to produce a sizable reduction of the final @xmath14li yield , while respecting other cosmological bounds , such as those coming from extra relativistic degrees of freedom ( @xmath19 ) or spectral distortions of the cmb .
a recent concrete example of these difficulties has been illustrated in @xcite , which tried such a fine - tuned solution by studying the effects of o(10 ) mev sterile neutrino decays . in this article , we point out that , depending on the epoch , _ at sufficiently low energies of injection the cascade develops differently and the final spectrum is significantly altered with respect to eq .
( [ eq : spectrum ] ) _ , which has been incorrectly used till recently , see e.g. @xcite . as a concrete application
, we show how this re - opens a window to a cosmological solution to the @xmath14li problem via e.m .
additionally , one expects peculiar signatures associated to such scenarios , which can be probed with cosmological observations . we will discuss this both in a proof - of - principle example and in the context of a particle physics model , involving one sterile neutrino
this was chosen for its simplicity and to allow for a direct comparison with the results of @xcite , which studied a similar model .
further considerations on some additional implications of our insight are finally outlined .
our argument is the following : let us assume that one injects photons at some time ( or corresponding plasma temperature @xmath20 ) whose energy @xmath5 is below the pair production threshold at that epoch , which can be estimated for the cmb plasma to be @xmath21mev @xcite . note that _ as long as @xmath22kev , this is compatible with the typical nuclear photo - disintegration energies relevant for bbn_. it is clear that the spectrum of eq .
( [ eq : spectrum ] ) can not stay valid in this regime : there is no pair - production cutoff , of course , but even the lower - energy part can not be correctly captured by eq .
( [ eq : spectrum ] ) . unless one considers _ other physical processes _ for the photon interactions ,
not included in the derivation of eq .
( [ eq : spectrum ] ) , there are no non - thermal electrons which can up - scatter cmb photons !
since the photon interaction probability is much smaller below pair production threshold , at leading level the injected spectrum below @xmath23 stays the same
apart for red - shifting , which happens on very long timescales with respect to particle photon interactions and hence we neglect . accounting for the finite probability for the photons to scatter via @xmath24 , via compton scattering off the background electrons , or via bethe - heitler @xmath25 production onto background protons and helium nuclei one does end up with a suppression of the injected spectrum , plus a lower energy tail due to downgraded energy @xmath1 s as well as @xmath1 s produced via ic by the secondary @xmath2 s .
the resulting secondary or tertiary photons , on the other hand , are typically at too low - energies to contribute to photo - dissociations and will be neglected . within this approximation , the boltzmann equation describing
the evolution of the distribution function @xmath26 reads : @xmath27 where @xmath28 is the source injection term , @xmath29 is the total interaction rate , and we neglected the hubble expansion rate , since interaction rates are much faster and rapidly drive @xmath26 to a quasi - static equilibrium , @xmath30 .
thus , we simply have : @xmath31 where the term @xmath32 for an exponentially decaying species with lifetime @xmath33 and density @xmath34 , whose total e.m . energy injected per particle is @xmath5 , can be written as @xmath35 with @xmath36 being the redshift at time @xmath37 , and the energy parameter @xmath38 ( conventionally used in the literature ) is simply defined in terms of the initial comoving density of the @xmath39 particle @xmath40 and the actual one of the cmb , @xmath41 , via @xmath42
. a monochromatic emission line would then correspond to @xmath43 .
for a two body decay @xmath44 into a monochromatic line plus another not better specified ( quasi)massless particle @xmath45 , one would have @xmath46 , where @xmath47 is the mass of the particle . here
, we will be interested in multi - mev values for the mass @xmath47 and at temperatures of order few kev or lower , hence the thermal broadening is negligible and a dirac delta spectrum as the one above is appropriate .
the interaction rate @xmath48 is computed by accounting for : i ) compton scattering over thermal electrons , @xmath49 , taken from @xcite ; ii ) scattering off cmb photons : @xmath50 , for which we follow @xcite ; iii ) bethe - heitler pair creation : @xmath51 , for which we use the formulae of @xcite .
note that we neglect the small effect due to the finite probability for the secondary or tertiary photons to induce some dissociations , i.e. once a photon interacts it is `` lost '' .
the results that we obtain are in this respect slightly conservative , by an amount which we estimated to be of the order of a few @xmath52 .
at temperatures of few kev or lower , the standard bbn is over , and the additional nucleosynthesis can be simply dealt with as a post - processing of the abundances computed in the standard scenario .
the non - thermal nucleosynthesis due to electromagnetic cascades can be described by a system of coupled differential equations of the type @xmath53 where : @xmath54 is the ratio of the number density of the nucleus @xmath55 to the total baryon number density @xmath56 ( this factors out the trivial evolution due to the expansion of the universe ) ; @xmath57 is the photodissociation cross sections onto the nuclei @xmath20 into the nucleus @xmath55 , i.e. the production channel for @xmath55 ; @xmath58 is the analogous destruction channel ( both cross sections are actually vanishing below the corresponding thresholds ) . in general
one also needs to follow secondary reactions of the nuclear byproducts of the photodissociation , which can spallate on or fuse with background thermalized target nuclei ( see for instance @xcite ) but none of that is relevant for the problem at hand .
if the injected energy is @xmath59 , the only open non - thermal bbn channel is @xmath60 , whose cross - section we denote with @xmath61 , there are no relevant source terms and only one evolving species ( since @xmath62 ) , thus yielding for the final ( at @xmath63 ) to initial ( at @xmath64 ) abundance ratio @xmath65 to obtain eq .
( [ result ] ) , we transformed eq .
( [ nonthbbn ] ) into redshift space , defining @xmath66 as appropriate for a universe dominated by radiation , with @xmath67 , @xmath68 and @xmath69 being the present hubble expansion rate and fractional radiation energy density , respectively . by construction , equating the suppression factor given by the rhs of the eq .
( [ result ] ) to @xmath70 provides a solution to the @xmath14li problem which is in agreement with all other constraints from bbn . in fig .
[ fig : resultsberylliummonochromatic ] , the lower band shows for each @xmath33 the range of @xmath38 corresponding to a depletion from 40% to 70% , for the case @xmath71mev . similar results would follow by varying @xmath5 by 10% about this value , i.e. provided one is not too close to the reaction threshold .
the upper band represents the analogous region if we had distributed the same injected energy according to the spectrum of eq .
( [ eq : spectrum ] ) , up to min[@xmath72 .
it is clear that in the correct treatment a large portion of this region survives other cosmological constraints , described below , while none survives in the incorrect treatment . vs. lifetime @xmath33 ,
for which the primordial lithium is depleted to 40% to 70% of its standard value , for a monochromatic photon injection with energy @xmath73 .
the upper band represents the analogous region if we had distributed the same injected energy , up to @xmath71mev , according to the erroneous spectrum of eq .
( [ eq : spectrum ] ) . above the solid blue curve , a change in entropy ( and @xmath10 ) between bbn and cmb time larger than the @xmath74 error inferred from cmb would be obtained .
the region to the right of the dot - dashed green curve is excluded by current constraints from @xmath75-distortions in the cmb spectrum @xcite according to the computation of @xcite , while the dashed cyan curve illustrates the weaker bounds that would follow from the less accurate parameterization of @xcite .
the dotted red curve is the forecasted sensitivity of the future experiment pixie , corresponding to @xmath76@xcite.[fig : resultsberylliummonochromatic],scaledwidth=40.0% ]
we mentioned the baryon abundances @xmath10 inferred from cmb and bbn ( notably @xmath11h ) probes are consistent within errors .
this implies that no major injection of entropy took place between the bbn time and the cmb epoch , otherwise the baryon - to - photon ratio ( proportional to @xmath10 ) would have changed , see for instance @xcite . in a radiation - dominated universe , the change in entropy associated to a release of energy into _ all _ e.m .
particles characterized by parameter @xmath77 and a lifetime @xmath33 can be estimated as @xmath78 for illustration , in fig .
[ fig : resultsberylliummonochromatic ] the solid blue line represent the level of entropy release associated to a variation of @xmath74 around the best - fit measured value of @xmath79 by planck , @xmath80 @xcite .
since the level of injected energy needed to solve the lithium problem via a monochromatic line is up to two orders of magnitude below the bounds , it is clear that this constraint is very weak , but for very short lifetimes of the order of @xmath81s .
another constraint comes from the level of _ spectral distortions _ in the cmb . for the relatively short lifetimes relevant for the problem ,
the compton scattering is fast enough that energy - redistribution is effective , no @xmath82-type distortion survives . on the other hand
, processes that change the number of photons are relatively rare , and a residual distortion of the @xmath75-type is possible .
this has been constrained by cobe - firas to be @xmath83 @xcite .
the level of spectral distortion produced by the decay process here has been estimated in the past ( see for instance @xcite ) , but a recent re - evaluation @xcite found significant improvements at short lifetimes , essentially due to a better treatment of the time - dependence of the visibility function . the theoretical expectation for @xmath75
can be written as @xmath84 where the function @xmath85 is taken from @xcite .
the bound excludes the region to the right of the dot - dashed , green curve in fig .
[ fig : resultsberylliummonochromatic ] . for comparison , the dashed cyan curve reports the much weaker bound that would follow from the approximations in @xcite .
we also checked that the extra constraint due to extra `` dark radiation '' parameterized by @xmath19 is irrelevant as long as the branching ratio in extra relativistic species is not larger than _ a couple of orders of magnitudes _ with respect to the photon one .
we thus conclude that there is a significant interval of lifetimes ( @xmath86 ) and corresponding energy injection parameter @xmath87 for which a perfectly viable solution is possible .
we remind once again that this possibility appeared to be closed due to the use of eq .
( [ eq : spectrum ] ) beyond its regime of applicability .
one may wonder how realistic such a situation is in a concrete particle physics model .
although we refrain here from detailed model - building considerations , it is worth showing as a proof - of - principle that models realizing the mechanism described here while fulfilling the other cosmological constraints ( as well as laboratory ones ) can be actually constructed .
let us take the simplest case of a sterile majorana neutrino with mass in the range @xmath88 , mixing with flavour @xmath89 neutrinos via an angle @xmath90 .
we also define @xmath91 .
the three main decay channels of this neutrino are ( see e.g. @xcite and refs .
therein ) : * @xmath92 , with rate @xmath93 ; * @xmath94 , with a rate depending on single @xmath95 s ; * @xmath96 , with a rate @xmath97 .
the resulting branching ratios for the masses of interest and @xmath98 are of the level of @xmath99 , respectively .
it is physically more instructive to normalize the abundance of the @xmath100 , @xmath101 , in terms of one thermalized neutrino ( plus antineutrino ) flavour species , @xmath102 . in fig .
[ fig : resultssterileneutrino ] , we show the corresponding range of parameters in the @xmath103 plane , for @xmath104mev , for which the @xmath14li problem is solved , fulfills cosmological constraints and , provided that @xmath98 , also laboratory ones @xcite .
it is worth noting that : i ) the entropy release bound is now close to the region of interest , since the decay mode @xmath94 , which is useless as far as the @xmath14be dissociation is concerned , dominates the e.m .
energy injection .
ii ) a non - negligible fraction of relativistic `` dark radiation '' is now injected , mostly via the dominant decay mode @xmath92 ; hence we added the current @xmath105 sensitivity of planck to @xmath19 @xcite , with @xmath106 computed similarly to what done in @xcite .
the needed abundance could be obtained in scenarios with low reheating temperature @xcite .
constraints for the sterile neutrino model discussed in the text .
the legend is the same as for case 1 ) .
, scaledwidth=40.0% ]
we have discussed the breaking of the universality of the photon spectrum in electromagnetic cascades , when the energy of the injected photons falls below the pair - production threshold .
this may be of interest for a number of astroparticle applications , but in the specific case of the cosmological context , this happens when @xmath107mev .
we noted that the energies concerned are of the same level of the binding energies of light nuclei .
this implies a potential large impact on non - thermal nucleosynthesis models , notably of electromagnetic type , but could be also relevant for models with late time hadronic cascades .
we provided an analytical estimate of the resulting ( much harder ) meta - stable spectrum of non - thermal photons , and showed that the impact is so large that it can potentially re - open the possibility of electromagnetic cacade solutions to the so - called `` lithium problem '' , which were thought to be excluded by other cosmological constraints .
we substantiated this point with a proof - of - principle example of a photon line injection at @xmath108mev from a particle decay , satisfying by construction all other bbn constraints but , not trivially , also all other cosmological bounds plaguing previous attempts .
although we did not indulge into particle model building , we proved that the right conditions can be actually satisfied in a simple scenario involving a @xmath109mev sterile neutrino mostly mixed with @xmath110 and/or @xmath111 with effective mixing angle @xmath112 .
the possibility to find new mechanisms to deplete the standard bbn prediction of lithium abundance in a consistent way is probably the most spectacular consequence of our investigation . in turn
, this could stimulate more specific model - building activities .
for instance , decays of relatively light new neutral fermionic particles @xmath39 for which the @xmath113 channel is the only two body standard model channel opened as it is the case for the light gravitinos in supergravity models constitute a natural class of candidates .
alternatively , one may think of decaying scenarios involving a pair of quasi degenerate mass states @xmath39 and @xmath114 , which are potentially much heavier than the mev scale .
some of these scenarios may be motivated by other astroparticle or particle physics reasons and certainly deserve further investigation .
we also showed how improvements in the determination of @xmath115type spectral distortions bounds of the cmb might be crucial to test these scenarios : testing frameworks for the particle physics solutions to the lithium problem may thus provide additional scientific motivations for future instruments like pixie @xcite .
computations of distortions corresponding to specific injection histories may also be refined : for instance , for short lifetimes relativistic corrections to the double compton and compton scattering may be important to improve the theoretical accuracy @xcite . finally , from a phenomenological perspective , an obvious spin - off of our work would be to re - compute the bbn bounds to electromagnetic decaying particles in cases where the universality of the spectrum of eq .
( [ eq : spectrum ] ) breaks down .
preliminary results indicate that bounds can be easily modified by one order of magnitude .
these results will be reported in a forthcoming publication .
f. iocco , g. mangano , g. miele , o. pisanti and p. d. serpico , phys . rept . * 472 * , 1 ( 2009 ) [ arxiv:0809.0631 [ astro - ph ] ] .
m. pospelov and j. pradler , ann .
nucl . part .
sci . * 60 * , 539 ( 2010 ) [ arxiv:1011.1054 [ hep - ph ] ] .
f. iocco , memorie della societ astronomica italiana supplementi,2012 22 , 19 [ arxiv:1206.2396 [ astro-ph.ga ] ] .
r. h. cyburt , j. r. ellis , b. d. fields and k. a. olive , phys .
d * 67 * ( 2003 ) 103521 [ astro - ph/0211258 ] .
a. fradette , m. pospelov , j. pradler and a. ritz , phys .
d * 90 * , 035022 ( 2014 ) [ arxiv:1407.0993 [ hep - ph ] ] .
m. kusakabe , m. k. cheoun and k. s. kim , phys .
d * 90 * , 045009 ( 2014 ) [ arxiv:1404.3090 [ astro-ph.co ] ] .
h. ishida , m. kusakabe and h. okada , phys .
d * 90 * , no .
8 , 083519 ( 2014 ) [ arxiv:1403.5995 [ astro-ph.co ] ] .
m. kawasaki and t. moroi , astrophys .
j. * 452 * , 506 ( 1995 ) [ astro - ph/9412055 ] .
r. svensson and a. a. zdziarski , astrophys .
j. * 349 * ( 1990 ) 415 .
r. j. protheroe , t. stanev and v. s. berezinsky , phys .
d * 51 * ( 1995 ) 4134 [ astro - ph/9409004 ] .
k. jedamzik , phys .
d * 74 * ( 2006 ) 103509 [ hep - ph/0604251 ] .
j. l. feng , a. rajaraman and f. takayama , phys .
d * 68 * ( 2003 ) 063504 [ hep - ph/0306024 ] .
p. a. r. ade _ et al .
_ [ planck collaboration ] , astron .
astrophys .
( 2014 ) [ arxiv:1303.5076 [ astro-ph.co ] ] .
d. j. fixsen , e. s. cheng , j. m. gales , j. c. mather , r. a. shafer and e. l. wright , astrophys . j. * 473 * , 576 ( 1996 ) [ astro - ph/9605054 ] .
j. chluba and r. a. sunyaev , mon . not .
. soc .
* 419 * ( 2012 ) 1294 [ arxiv:1109.6552 [ astro-ph.co ] ] .
a. kogut , d. j. fixsen , d. t. chuss , j. dotson , e. dwek , m. halpern , g. f. hinshaw and s. m. meyer _ et al .
_ , jcap * 1107 * ( 2011 ) 025 [ arxiv:1105.2044 [ astro-ph.co ] ] .
f. bezrukov , h. hettmansperger and m. lindner , phys .
d * 81 * ( 2010 ) 085032 [ arxiv:0912.4415 [ hep - ph ] ] . | the standard theory of electromagnetic cascades onto a photon background predicts a quasi - universal shape for the resulting non - thermal photon spectrum . this has been applied to very disparate fields , including non - thermal big bang nucleosynthesis ( bbn ) .
however , once the energy of the injected photons falls below the pair - production threshold the spectral shape is much harder , a fact that has been overlooked in past literature .
this loophole may have important phenomenological consequences , since it generically alters the bbn bounds on non - thermal relics : for instance it allows to re - open the possibility of purely electromagnetic solutions to the so - called `` cosmological lithium problem '' , which were thought to be excluded by other cosmological constraints .
we show this with a proof - of - principle example and a simple particle physics model , compared with previous literature . |
the su - schrieffer - heeger ( ssh ) model is a standard tight - binding model with spontaneous dimerization proposed by su , schrieffer and heeger to describe the one - dimensional ( 1d ) polyacetylene @xcite . despite of its simplicity
, it has attracted extensive studies in the past decades as it exhibits rich physical phenomena , such as topological soliton excitation , fractional charge and nontrivial edge states @xcite .
recent studies of topological insulators @xcite lead to reexamination of the ssh model , which belongs to the bdi symmetry class as the simplest example of 1d topological insulators @xcite .
it was found that various systems , including the graphene ribbon @xcite , the p - orbit optical ladder system @xcite and the off - diagonal bichromatic optical lattice @xcite , can be mapped into the ssh model .
the extended ssh models with periodic modulation of chemical potential were also used to study the topological charge pumping @xcite . on the other hand ,
the 1d superlattice models have recently attracted great attentions as the families of superlattice systems were found to be topologically nontrivial @xcite with the periodical modulation parameter providing an additional dimension .
these studies inspired the theoretical connection between the 1d superlattice system and the two - dimensional ( 2d ) quantum hall effect of hofstadter lattices @xcite .
the haldane s model @xcite is a prototype model which may realize anomalous quantum hall effect in a 2d honeycomb lattice by introducing a complex next - nearest - neighbor ( nnn ) hopping term but without any net magnetic flux through a unit cell of the system . by varying the hopping phase and the sublattice potential difference
, the system can undergo a topological phase transition from a normal insulator to a chern insulator @xcite .
however , it is hard to realize the haldane s model experimentally in ordinary condensed matter systems because of the specifically staggered magnetic flux introduced in the model @xcite .
the recent progress in simulation of topological phases in reduced dimensions raises the question whether a similar phase diagram of haldane s model is realizable in 1d systems . in this work
, we shall explore the realization of haldane s phase diagram in reduced 1d lattices by considering the extended ssh model with both the nearest - neighbor ( nn ) and nnn hopping amplitudes cyclically modulated by an additional parameter .
as the cyclical modulation parameter is continuously changed , we demonstrate that nontrivial edge states emerge due to the introduction of specific nnn hopping terms .
the additional modulation parameter and the momentum permit us to define a topological invariant in the extended 2d parameter space to distinguish different topological phases , which gives rise to a phase diagram similar to that of the 2d haldane s model . the topologically nontrivial edge states in our model may be observed in photonic crystal systems .
furthermore , we find that an extended ssh model with the sublattice potential modulated also exhibits nontrivial topological phases .
we consider an 1d lattice with one unit cell made of a and b sites described by the hamiltonian @xmath0 with @xmath1 + h.c . , \\
h_{nnn } & = & \sum\limits_{n } \left [ t_{a}c_{a , n}^{+}c_{a , n+1}+t_{b}c_{b , n}^{+}c_{b , n+1 } \right ] + h.c . ,
\end{aligned}\ ] ] where @xmath2 and @xmath3 with @xmath4 representing the dimerization strength and @xmath5 being a cyclical parameter which can vary from @xmath6 to @xmath7 continuously , @xmath8 ( or @xmath9 ) are the annihilation operators localized on site a ( or b ) of the @xmath10-th cell , and @xmath11 and @xmath12 are the nnn hopping amplitudes along sublattice a and b respectively . here
@xmath13 is set as the unit of energy .
when @xmath14 , the hamiltonian reduces to the ssh model @xcite . under the periodic boundary condition
, we can make a fourier transformation : @xmath15 , where @xmath16 is the number of the unit cells and @xmath17 takes @xmath18 or @xmath19 .
then the hamiltonian can be written in the form of @xmath20 where @xmath21 and @xmath22 alternatively , @xmath23 can be expressed in the form @xmath24 with @xmath25 the identity matrix , @xmath26 are pauli matrices acting on the pseudospin @xmath27 , @xmath28 , @xmath29 , @xmath30 and @xmath31 . diagonalizing the hamiltonian
, we get the eigenvalues @xmath32 with @xmath33
. sites and @xmath34 under obcs .
( a ) @xmath14 .
( b ) @xmath35 .
( c ) @xmath36 .
( d ) @xmath37 . ] in the absence of nnn hopping terms , the spectrum of the ssh model is given by @xmath38 , which is composed of symmetric positive and negative branches .
the two bands are separated by a band gap @xmath39 at @xmath40 which is closed at @xmath41 and @xmath42 corresponding to @xmath43 .
it is well known that the ssh model exhibits two topologically distinguishable phases protected by inversion symmetry of the hamiltonian .
the two phases can be distinguished by the presence or absence of two - fold degenerate zero - mode edge states under open boundary conditions ( obcs ) .
as shown in fig.1a for the system with @xmath34 , the phase in the parameter regime @xmath44 characterized by the presence of zero - mode edge states is topologically different from the phase in the regime @xmath45 .
if we continuously change the parameter @xmath5 , a phase transition occurs at the @xmath46 .
the existence of zero modes in the ssh model is protected by both the inversion symmetry and particle - hole symmetry @xcite .
the presence of nnn hopping terms may break the symmetry . to see how the nnn hopping terms affect the properties of the ssh model , we first consider the case with @xmath47 . as
shown in fig.1b , the symmetric nnn hopping terms break the particle - hole symmetry , but do not open the gap as the inverse symmetry is preserved .
next we consider the case with @xmath48 .
while such terms preserve the particle - hole symmetry , they break the inverse symmetry of the ssh model and thus open a gap as shown in fig.1c . for a general case with @xmath49 ,
the spectrum is split into two asymmetric bands with a gap opened . accompanying with opening of the gap ,
the edge modes split into two branches as displayed in fig.1c and fig.1d .
as we continuously change the parameter @xmath5 , the two branches of edge modes have no crossing and do not connect the separated bands , which indicates the system with the lower band filled to be a topologically trivial insulator .
as the fixed nnn hopping terms do not lead to topological nontrivial phenomena , we consider the model with both the nn and nnn hopping terms cyclically varying with the parameter @xmath5 . to give a concrete example , we consider the modulation @xmath50 here @xmath51 is a parameter to control the nnn hopping amplitude and an additional parameter @xmath52 is introduced . choosing a specific @xmath52
, we can get either symmetric or antisymmetric nnn hopping terms , e.g. , for @xmath53 we have @xmath54 , and for @xmath55 we get @xmath56 . in fig.[fig2 ] , we show the spectra of the system versus @xmath5 under obcs for several typical parameters of @xmath52 . as shown in fig.[fig2]a , the edge modes exhibit nontrivial properties as they continuously connect the bulk bands with the change of @xmath5 .
when we vary the parameter @xmath52 , we find that the edge mode spectrum exhibits similar characters except for the case of @xmath53 as shown in fig.[fig2]b , where the gap is closed and two branches of edge modes merge together .
although systems with @xmath57 and @xmath58 have identical bulk spectra , we note that the edge modes labeled by the same @xmath59 and @xmath5 locate at opposite edges as displayed in fig.[fig2]a and fig.[fig2]c , which implies the insulating states for systems with @xmath52 and @xmath60 belonging to different topological states
. sites , @xmath34 and @xmath61 under obcs .
( a ) , ( b ) and ( c ) correspond to @xmath58 , @xmath6 and @xmath62 , respectively .
insets show wavefuntions of edge modes . ]
the gapless edge states are generally associated with the nontrivial topological properties of bulk systems . for the 1d extended ssh model with cyclical modulation
, we can define a chern number @xmath63 in a 2d parameter space with @xmath5 being chosen as the second parameter besides the momentum @xmath64 , where @xmath65 and @xmath66 with @xmath67 the occupied bloch state . as a topological invariant
, the chern number can be used to characterize the topological properties of the family of the 1d system @xcite .
it is well defined only when the gap of the bulk states is fully opened throughout the brillouin zone , and can be calculated numerically in a discretized brillouin zone @xcite .
we find that the chern number of the half - filled extended ssh model with nnn hopping terms given by eq.([nnn1 ] ) is @xmath68 when @xmath69 , whereas @xmath70 when @xmath71 .
a phase transition occurs at the boundary of @xmath53 and @xmath72 .
we note that taking different values of @xmath51 shall not change the phase diagram , however one needs to choose a suitable @xmath51 to guarantee the two bands never overlapping together and separated by a finite gap .
indicates the chern number of the lower band . ]
next we consider a more general case with nnn hopping amplitudes modulated as @xmath73 where @xmath74 and @xmath75 are two constants . for convenience ,
we define @xmath76 and @xmath77 and choose @xmath78 . in order to guarantee the fermi level lying in a gap between two bands ,
we need choose suitable parameters of @xmath74 , @xmath75 and @xmath51 to avoid the overlap of the lower and upper bands . introducing two additional parameters
produces a much richer phase diagram than the case with modulation described by eq.([nnn1 ] ) . in fig.[fig3 ] , we display the phase diagram in the parameter space of @xmath52 and @xmath79 .
there are altogether three topologically different phases , which can be characterized by the topological invariant @xmath70 , @xmath68 and @xmath6 . for a topological nontrivial system ,
the chern number for each band is an invariable integer unless the bulk gap closed @xcite , therefore the phase boundary of different topological phases can be determined by the gap closing point .
for the extended ssh model with modulation of ( [ nnn2 ] ) , the eigen - energy is given by @xmath80 with @xmath81 .
the gap closes when @xmath82 , which enforces the condition @xmath83 for @xmath40 and @xmath84 . around the phase boundary
, one can analytically derive the chern number of the lower filled band given by @xmath85\ ] ] with @xmath86 .
the numerical calculation of the chern number also gives exactly the same phase diagram as shown in fig.3 with the phase boundary determined by eq.([phaseboundary ] ) .
it is interesting that the phase diagram of our 1d extended ssh model has a similar structure as that of the 2d haldane model @xcite .
the phase boundary is only associated with @xmath52 and the ratio of @xmath87 and @xmath51 , but is irrelevant to the value of @xmath88 and @xmath51 . as long as we keep @xmath52 and @xmath89 unchanged
, varying @xmath88 and @xmath51 will not affect the phase diagram . to give examples , we show spectra of systems with @xmath90 and @xmath91 but different @xmath51 and @xmath88 under obcs in fig.[fig4](a)-(c ) . despite of the shape of spectra having some minor differences , the existence of continuous edge modes connecting the lower band and upper band for all these systems indicates that they are topologically nontrivial . as a contrast
, we show the spectra of topologically trivial states in fig.[fig4](d)-(e ) , where the upper and lower branches of edge modes are separated by a gap .
sites and @xmath34 under obcs .
( a ) @xmath92 .
each of ( b)-(e ) has only one parameter different from ( a ) .
( b ) @xmath93 .
( c ) @xmath94 .
( d ) @xmath95 .
( e ) @xmath53 .
insets show wavefuntions of edge modes . ] the topological property of this model may be observed through the experimental setup composed of series of single mode waveguides arranged in a circular arc ( fig.5 ) .
as shown in fig.5a , each dot represents a waveguide vertical to the paper which corresponds to a site in the tight binding hamiltonian @xcite .
the propagating light can tunnel from each waveguide to its neighboring waveguides , and the hopping amplitude is inversely proportional to the distance between them .
the red solid dots in fig.5 represent sites a , while the green hollow ones are for sites b. each of them are arranged in a circular arc with an independent radius @xmath96 and @xmath97 .
the angle between two neighbor red ( or green ) dots @xmath98 is set to a constant , hence we can control the nnn hopping amplitudes @xmath11 and @xmath12 by adjusting @xmath96 and @xmath97 , respectively .
the distance between two neighbor dots of different circular arc is determined by @xmath17 , @xmath96 and @xmath97 all together , thus the nn hopping amplitudes @xmath99 and @xmath100 can not be adjusted independently from @xmath11 and @xmath12 . however
, the alternating relation between @xmath99 and @xmath100 can be achieved by varying @xmath17 individually .
hence we can control the hopping amplitudes @xmath101 by varying @xmath102 in fig.5a to design the waveguide system , which can be described by the hamiltonian ( 1 ) . by choosing proper parameters
, we can prepare the system in either topologically nontrivial or trivial state as shown in fig .
[ fig4](a ) or ( d ) .
if we take parameters of fig.[fig4](a ) with @xmath103 ( equivalently @xmath104 as marked in fig.[fig4](a ) ) , one can observe the edge states located in the left edge by measuring intensity distributions of injecting light as a function of the position of the waveguides @xcite .
if we change @xmath5 adiabatically in the direction of light propagation , in a topological nontrivial case , one can observe the edge state localized in one end of the lattices pumping to the other side as we vary the @xmath5 from @xmath105 to @xmath106 , while in the topological trivial case ( e.g. fig.[fig4](d ) ) the edge state will always localize in the same side . .
( b ) along the propagation direction , @xmath5 can be varied along the length of the waveguide by smoothly changing the spacing between successive waveguides .
( c ) examples for the geometrical arrangement of waveguides for several @xmath5 . ] to study the feasibility of this experimental proposal , we add a random disorder perturbation @xmath107 to each hopping amplitude to check the robustness of the topological edge states under the disorder perturbation , where @xmath108 randomly distributes in the range of @xmath109 $ ] . taking the system with parameters as given in fig.[fig4](a ) with @xmath104 , we find the edge state located in the left boundary is robust even the disorder amplitude takes @xmath110 . finally , we consider an alternative model which also exhibits the same phase diagram as model ( [ h1 ] ) with nnn hopping amplitudes given by eq.([nnn2 ] ) .
the model can be described by @xmath111 with @xmath112 the hamiltonian of ssh model and @xmath113 given by @xmath114 where the alternating on - site potentials @xmath115 and @xmath116 are given by @xmath117 the hamiltonian in momentum space can be represented as @xmath118 with @xmath119 the spectrum of the system is given by @xmath120 with @xmath121 it is easy to find the phase boundary given by @xmath122 corresponding to the close of energy gap for @xmath40 and @xmath123 .
following similar procedures for model ( [ h1 ] ) , we can find that the phase diagram for the above ssh model with alternating on - site potentials is also given by fig.[fig3 ] .
the above model is possible to be realized in cold atomic systems trapped in optical superlattices @xcite . in order to see the connection of the 1d model to the 2d haldane model , we consider the model with both nnn hopping terms and modulated on - site potentials described by the following hamiltonian @xmath124 similarly , the hamiltonian in momentum space can be represented as @xmath118 with @xmath125 diagonalizing the hamiltonian
, we get the eigenvalues @xmath126 with @xmath127 ^ 2 + t_1 ^ 2 + t_2 ^ 2 + 2 t_1 t_2 \cos k$ ] .
if both the nnn hopping amplitudes ( @xmath128 , @xmath129 ) and alternating on - site potentials ( @xmath115 , @xmath116 ) are modulated in a similar way as eq.([nnn2 ] ) or eq.([mu ] ) , topological nontrivial phases characterized by nonzero chern numbers can also be found .
for the convenience of later comparison , here we briefly review the haldane model . as discussed in details in ref.@xcite ,
the lattice tight - binding hamiltonian of the haldane model is given by @xmath130 where the summation @xmath131 is defined in the 2d honeycomb lattice , which is composed of two sublattices labeled by a " and b " , respectively . here
@xmath132 for site a and @xmath133 for site b , @xmath134 denotes the nearest - neighbor hopping amplitude and @xmath135 the nnn hopping amplitude .
the magnitude of the phase is set to be @xmath136 , and the direction of the positive phase is clockwise following haldane s work@xcite .
choosing the zigzag edges in the @xmath137 direction @xcite and using the basis of two component spinors " @xmath138 ( @xmath139 ) of bloch states constructed on the two sublattices , we can write the hamiltonian in the momentum space as : @xmath140 where @xmath141 , @xmath142 and @xmath143 . the form of this hamiltonian is very similar to eq.([both ] ) , and with similar procedure we can calculate the gap closing conditions , which give the phase transition boundaries of the haldane model determined by @xmath144 .
the chern number of the lower filled band can also be analytically given by @xmath145 $ ] , where the effective mass becomes @xmath146 .
comparing eq.([both ] ) to eq.([h - haldane ] ) , if we take @xmath147 , @xmath148 , @xmath149 and @xmath150 , we observe that the model of ( [ h - both ] ) can be mapped to the haldane model by identifying @xmath64 to @xmath151 and @xmath5 to @xmath152 .
the above mapping gives us a straightforward understanding of the connection of the 1d generalized ssh model with fine tuning parameters to the well - known 2d haldane model .
similarly , a mapping between the 1d superlattice and quasi - periodic systems and the 2d hofstadter problems has been well established @xcite .
however , we note that the model described by eq.([both ] ) with artificially fine tuning parameters is hard to be experimentally realized as both the hopping parameters and alternative on - site potentials need to be tuned simultaneously . on the other hand , models with only either the modulated nnn hopping terms or modulated on - site potentials also exhibit the nontrivial phase diagram similar to that of haldane s model , but are much easier to be realized in experiments .
in summary , we study the extended ssh model with both the nn and nnn hopping terms periodically modulated by a parameter @xmath5 .
our results show that the family of extended ssh model supports topologically nontrivial edge modes with respect to the modulation parameter @xmath5 .
the topologically nontrivial phase diagram shows some surprising connections between the 1d extended ssh models and the 2d haldane model .
we propose an experimental setup composed of photonic crystals to realize our model system .
some other models with similar phase diagrams are also discussed and their connection to the 2d haldane model is addressed .
the direction of the zigzag edges in this work is different from that of ref.@xcite , where the direction is along the y direction as shown in the fig.1(a ) therein .
replacing @xmath151 to @xmath152 and @xmath152 to @xmath151 in our work , we can get results in ref.@xcite . | we study the extended su - schrieffer - heeger model with both the nearest - neighbor and next - nearest - neighbor hopping strengths being cyclically modulated and find the family of the model system exhibiting topologically nontrivial phases , which can be characterized by a nonzero chern number defined in a two - dimensional space spanned by the momentum and modulation parameter .
it is interesting that the model has a similar phase diagram as the well - known haldane s model .
we propose to use photonic crystal systems as the idea systems to fabricate our model systems and probe their topological properties . some other models with modulated on - site potentials
are also found to exhibit similar phase diagrams and their connection to the haldane s model is revealed . |
entanglement is one important concept in quantum mechanics . in quantum teleportation ,
maximally entangled states are the pivotal resource .
it is also the source of power in quantum computation .
usually , quantum computation starts from an initial quantum state , say , @xmath0 .
quantum gate operations are unitary operations that are constructed using a finite set of basic universal quantum computing gates@xcite . at the end of a computation , a measurement is performed on the register so that the outcome is read out .
many schemes have been proposed , and considerable progress has been made over the last decade .
however , it is still a daunting task to build a practical quantum computer .
though it is difficult to build complicated quantum system and to perform complex computation operations , small quantum systems are already easily available and some basic quantum operations can be performed .
recently there have been proposals to perform complicated quantum computation with small quantum systems such as entangled pairs of qubits and simple measurement .
it has been shown that quantum computation can be performed using single qubit operations , joint two - qubit measurement and together with some simple entangled quantum states such as ghz - states@xcite . in these schemes ,
the initial state of the quantum computer may be a separable state and this spares the need to perform complicated gate operations , however at the expense of using some number of entangled states such as einstein - podolsky - rosen ( epr ) pairs .
alternatively , quantum computation can be performed by starting from a very complicated entangled states , and then proceeding to the end by merely measurements .
the cluster state quantum computation scheme@xcite is typical example of these schemes .
valence bond state , proposed by ian affleck @xmath1@xcite and used in condensed matter physics@xcite , has also been found to play an important role in quantum computation recently@xcite .
verstraete and cirac@xcite proved that the cluster state quantum computation and valance - bond - state computation are equivalent . in ref.@xcite
, it was shown that universal quantum computation can be carried out by using only pairs of qubits in singlet state and by performing joint two - qubit and three - qubit measurement .
it has the advantage of being deterministic and using a simple initial product state .
single - qubit operation can be implemented by performing a bell - basis - like measurement on the target qubit and another qubit from a singlet .
controlled - z gate is implemented by performing two ghz - like state measurements on the two qubits and three singlets system .
these two unitary gate operations are the basic building blocks for more complicated gate operations .
as this scheme is practically appealing , it is interesting to construct the elementary gate operations in this scheme . in this paper
, we will provide the constructions of the elementary gate operations for this scheme .
the paper is organized as follows . in section [ s2 ]
, we briefly summarize the construction of basic gates of ref.@xcite . in section [ s3 ] , we give the results for the well - known elementary gates mentioned in@xcite , which are the basic ingredients for quantum algorithms in quantum computation . in section [ s4 ] , we give a brief summary .
here we briefly review the basic gate operation in the scheme in ref.@xcite . in ref.@xcite ,
basic single - qubit local unitary operation on a particle can be implemented in a teleportation fashion .
suppose the state of particle @xmath2 is @xmath3 , and particles @xmath4 and @xmath5 are in an entangled state @xmath6 . by performing a joint measurement on qubits a and b in the following - basis @xmath7 where @xmath8 denote the pauli matrices @xmath9 and
@xmath10 is a maximally entangled state , and @xmath11 .
the state of particle @xmath5 becomes @xmath12 where @xmath13 is the outcome of the measurement in basis @xmath14 .
particle @xmath5 then works as the role of particle in the following process of the quantum computation .
it is worth mentioning here that the entangled state @xmath6 can be replaced by arbitrary maximally entangled states such as the bell - basis states , @xmath15 and @xmath16 .
however this substitution can not be used for the controlled - z gate .
the controlled - z gate , @xmath17 proposed in ref.@xcite involves two ghz - like state joint measurements as redrawn in fig.[f2a ] where we have given explicit labelling to the eight qubits involved . by doing three - qubit joint measurement of particles on @xmath18 , @xmath19 , @xmath20 and @xmath21 , @xmath22 , @xmath23 in the following basis , @xmath24 where @xmath25 , the controlled - z gate is implemented between qubits @xmath26 and @xmath27 .
the singlet states of the entangled pairs between @xmath19 and @xmath22 , @xmath26 and @xmath20 , @xmath27 and @xmath23 are all @xmath28 . by replacing the singlet state @xmath6 by other entangled states ,
it is also possible to construct the gate .
the detailed combination of the states used for each pair and the appropriate basis for the joint measurement are given in table [ t1 ] .
the main conclusion of this table is that the measuring - basis is determined by the state of the @xmath29 pair , the states of @xmath30 and @xmath31 pairs are irrelevant .
for example , as in fig.[f2b ] , assuming the state of the two original qubits is @xmath32 and the states of the @xmath33 , @xmath34 and @xmath35 pairs denoted by @xmath36 , @xmath37 and @xmath38 respectively , are in the bell - basis states , @xmath39 ) were used , half of the information in @xmath40 would be lost .
for instance , if @xmath41 , then the wave function of @xmath26 and @xmath27 would become @xmath42 , which have lost state information in this process due to the form of entangled state of @xmath19 and @xmath22 . to see this more clearly , we write the state of the eight - qubit system of fig.[f2b ] as @xmath43 , @xmath44 where latexmath:[\[\begin{aligned } the entangled state between @xmath19 and @xmath22 is the @xmath6 state , then the three - qubit measurement projections @xmath46 , @xmath47 , @xmath48 , @xmath49 are all none - zero .
however if we change the entangled state of @xmath19 and @xmath22 to state @xmath50 , and turn @xmath51,@xmath52 into @xmath53 toffoli gate is the further development of c - not gate .
we present the details for implementing the toffoli gate by doing joint measurement as shown in fig.[f8 ] .
explicitly , the entangled state of @xmath55 is @xmath56 and the entangled state of @xmath57 , @xmath58 are all @xmath59 the entangled state of @xmath60 is @xmath61 the joint measurement basis sets are @xmath62 where @xmath63 .
we note that the extra particles @xmath64 and @xmath65 are necessary in this process .
some extra operations should be applied to the wave function of particles @xmath66 after joint measurement too .
for instance , as shown in fig.[f8 ] , if the results of measuring - basis sets are @xmath67 , @xmath68 and @xmath69 , then an additional operation @xmath70 on particles @xmath66 is necessary .
fredkin gate is the development of swap gate in a three - qubit system .
the details of the implementation of this quantum gate is given in fig.[f9 ] . here
the entangled state of @xmath71 is @xmath72 the entangled state of @xmath73 is @xmath74 the entangled state of @xmath60 is @xmath75 the entangled state of @xmath76 : @xmath77 the measuring - basis sets are @xmath78 @xmath79 where @xmath80 .
it is worth poiting the extra particles @xmath81 , @xmath82 , @xmath83 , @xmath65 are crucial to implement the fredkin gate .
in this paper , we have constructed explicitly the elementary gates for the measurement - based quantum computation , including the generalized controlled - z gate , the phase gate and @xmath84 gate , the swap gate , the c - not gate , the fredkin gate and toffoli gate .
we have studied the relation between the form of the measuring - basis and the entangled states of the pairs .
it is found that the they have an exquisite relation among them . in some cases ,
some parts in the joint measurement basis are necessarily distinct for different qubits in a group so that they can implement the desired quantum gate .
it is interesting to mention that as shown in ref.@xcite , the teleportation - based quantum computation and the vbs quantum computation scheme are equivalent .
it is also equivalent to the cluster - state quantum computation .
in addition , we note that the patterns of entangled states in the process of constructing c - not gate and swap gate are the same and they only differ at the ways of the joint measurement . for these matters , the topological construction of quantum computation@xcite is worthwhile for further study at this point . in all these schemes , entanglement is the core of quantum computation . in the cluster - state quantum computation scheme ,
the entanglement is condensed into the initial state , while in the teleportation - based quantum computation as studied in this paper , the entanglement in injected into the quantum computing system by using entangled pairs of particles .
this work is supported by the national fundamental research program , grant no .
001cb309308 , china national natural science foundation , grant no . 60073009 , 10325521 , the srfdp program of education ministry of china .
l. k. grover,_phys .
_ * 79 * , 325 ( 1997 ) g. l. long , _
_ * a64 * , 022307 ( 2001 ) g. l. long and y. sun , _ phys . rev . _ * a64 * , 014303 ( 2001 ) k. maruyama and p. l. knight , _ phys . rev . _ * a67 * , 032303 ( 2003 ) on particles @xmath26 and @xmath27 for implementing controlled - phase gate if the result of the joint measurement is @xmath88 and @xmath89 .
it should be noted that @xmath90 is the same as in table [ t2 ] and @xmath91=@xmath92 ; @xmath93=@xmath94 besides , all the entangled pairs in this figure are in the form @xmath95,width=264 ] | in this paper , elementary quantum gate operations , such as the phase gate , the controlled - not gate , the swap and the fredkin gate are constructed using joint measurement and pairs of entangled qubit pairs .
the relation between the state of the entangled pair and the joint measurement basis is discussed .
some other generalization is also discussed . |
the idea to perform computation with the help of quantum mechanics dates back to feynman @xcite .
research in this direction got a considerable acceleration after the realisation @xcite that quantum computers could solve in a polynomial time certain problems which require nonpolynomial time on classical computers .
two main obstacles to a universal quantum computer are decoherence and scalability .
decoherence , if sufficiently weak , can be dealt with by quantum error correcting codes , see ref.@xcite for an experimental implementation .
scalability requires that a given technology can be upscaled to a quantum coherent circuit with hundreds or thousands of quantum gates , see ref.@xcite for reviews on various proposals of scalable implementations . at the present stage
solid state implementations have most to offer from the point of view of scalability but at the same time they are most likely to suffer from decoherence .
the aim of this note is to reduce decoherence in the persistent current qubit ( pcq ) proposal @xcite . two remarkable experiments @xcite took place last year where a superconducting loop broken by josephson junctions was forced into states which are quantum superpositions of two states with opposite macroscopic persistent currents : @xmath0 and @xmath1 .
the two considered combinations @xmath2 were eigenstates separated by a gap of @xmath3 so it is not surprising that each of them does not suffer from decoherence at the temperature of @xmath4 .
a more general superposition @xmath5 would quickly decohere into a mixed state @xmath6 parallel to the plane of the superconducting loop and penetrating through its thin layer , see fig.1 .
[ design ] = 6 cm @xmath7 has to be less than the critical field in a type i superconducting film .
if the @xmath8 nuclear spins are exposed to a magnetic field @xmath9 from the persistent current which is much weaker than @xmath7 , as can be characterized by a quality factor @xmath10 then the entanglement between the qubit state and the spins state is negligible and leads to negligible _ partial decoherence_. the means which treat entanglement incidentally also reduce the latter random walk problem , which can be furter eliminated by increasing the spin relaxation time @xmath11 which depends exponentially on temperature .
we begin with an elementary example where the qubit is coupled to just one environmental @xmath12 spin . for the purpose of this paper
the spin-1/2 is a sufficient representation of the nuclear spin-9/2 of niobium .
the spin is coupled to the external magnetic field @xmath7 ; its hamiltonian is @xmath13 the current of the qubit induces a magnetic field which at the location of the environmental spin has strength @xmath9 . for the time being
we assume that the field is perpendicular to the external field in , say , @xmath14-direction .
the hamiltonian of interaction between the qubit and the spin is @xmath15 the external magnetic field @xmath7 , which is in plane of the superconducting loop , does not couple to the qubit state . for the sake of simplicity we set the qubit hamiltonian @xmath16 , what does not change the general conclusion of this note .
the initial state of @xmath17 and @xmath18 is a product @xmath19 are @xmath20 eigenstates of @xmath21 . under the total hamiltonian @xmath22
the initial @xmath23 evolves into @xmath24 evolves into @xmath25 where the overlap @xmath26 is given by @xmath27 here @xmath28 .
@xmath29 is mixed when @xmath30 .
the entanglement with a single spin is reversible because there are periodic revivals at the times when @xmath31 .
we note that for @xmath32 the overlap @xmath26 remains @xmath33 and the qubit state stays pure .
even just one environmental spin can lead to irreversible full decoherence when the spin interacts with its thermal bath . in the absence of the qubit , thermalisation of the spin density matrix @xmath34 towards @xmath35
can be described by a master equation @xmath36 where @xmath37
are spin-@xmath12 lowering / raising operators and @xmath11 is the spin relaxation time of the order of minutes .
this master equation can be formally unravelled by a stochastic evolution of the spin state .
the spin state makes a down(up ) flip within time @xmath38 with a probability given by @xmath39 times the probability that the spin is in the @xmath40 state .
the master equation describes evolution of an average over such stochastic trajectories of the spin state .
let us apply this stochastic unravelling to the state of the spin and the qubit .
the @xmath41 in eq.([psit ] ) can be rewritten as @xmath42 can be described as a projection of @xmath41 on @xmath43 followed by a spin flip + @xmath44 . as a result @xmath41 jumps to @xmath45
is @xmath39 times the probability that @xmath41 is in the @xmath46 state which is @xmath47 , where we assume @xmath48 .
the probability is @xmath49 .
the down transition is a projection on @xmath50 followed by a spin flip @xmath51 .
@xmath41 jumps to @xmath52 as compared to the initial system state ( [ psi0 ] ) .
this sign flip also flips the sign of the off - diagonal elements in the reduced qubit density matrix , compare eqs.([rho0],[rhot ] ) .
the evolution of @xmath29 including spin relaxation is obtained after an average over stochastic trajectories .
the off - diagonal elements of @xmath29 average to zero on a timescale of @xmath53 .
this is the time of full decoherence due to spin relaxation .
we note that for @xmath54 the time is much longer than @xmath11 .
decoherence becomes more efficient when there are many copies of the environmental spin numbered by @xmath55 .
interaction hamiltonians are @xmath56 .
the initial product state @xmath57 evolves into a @xmath41 such that the reduced state of the qubit has the form of eq.([rhot ] ) but with @xmath58 where @xmath59 .
the early time overlap is @xmath60 , where @xmath61 .
@xmath26 decays from @xmath33 on a timescale of @xmath62 . for @xmath63 different factors in eq.([prodo ] ) oscillate with different frequencies dispersed in the range @xmath64 .
for @xmath65 the phases of oscillating terms in different factors become effectively random and the overlap averages to @xmath66 for @xmath67 the system state remains pure ; there is only a negligible partial decoherence .
fig.2 shows @xmath26 for a sample of @xmath68 s at two different values of @xmath7 .
[ overlap ] = 6 cm let us include spin relaxation .
to get full decoherence it is enough that only one out of @xmath8 spins makes the down transition so the decoherence time is @xmath8 times shorter than for a single spin , @xmath69 .
@xmath70 makes this decoherence time is much longer than @xmath11 .
it is worthwhile to compare the large @xmath17 limit with the case of @xmath71 when eq.([prodo ] ) gives @xmath72 the different factors , which oscillate with frequencies in a range proportional to @xmath9 , go out of phase after a time of @xmath73 when the overlap averages to zero .
for @xmath71 the time @xmath62 when @xmath26 decays away from @xmath33 is a time of full decoherence .
so far we considered the single spin environment and the multispin environment with a fully polarized initial state .
a thermal initial state is a weighted average over different partially polarized states , like e.g. @xmath74 .
the overlap ( [ prodo ] ) is the same for any such partially polarized state as for the fully polarized state so the estimates of decoherence time for the polarized state go through unchanged for the thermal density matrix .
the mixedness of the initial environmental state does matter when we finally take into account the @xmath75-component of the qubit magnetic field at the location of spins which gives an extra term in the interaction hamiltonian @xmath76 this hamiltonian , when applied to a partially polarized spins state , gives @xmath77 , where @xmath78 is a polarization of the @xmath79-th spin . if @xmath80 s were static , then this @xmath81 could be measured at the beginning of the quantum computation and balanced by a @xmath82-counterterm in the qubit hamiltonian , as suggested in ref.@xcite .
the problem is that @xmath80 s are not static because they flip at the rate of one spin per @xmath83 . the total polarization @xmath84 performs a random walk with one @xmath85 step every @xmath83 so that @xmath86 ^ 2}=4tn / t_r$ ] , where the overline means here an average over random walks .
the hamiltonian @xmath87 $ ] is random so the qubit states @xmath0 and @xmath1 accumulate opposite random phases @xmath88 , which grow as @xmath89 .
they lead to full dephasing after a dephasing time @xmath90 . in this discussion
we assume @xmath70 so that entanglement is negligible .
the external magnetic field @xmath7 is bounded from above by the critical field of the type i superconducting material @xmath91 .
this parameter is the highest ( @xmath92 ) for niobium . in this respect niobium
is much better than alluminium which has much lower @xmath93 .
we choose niobium and set @xmath94 .
there are materials like lead or tin that have abundant isotopes with zero nuclear spin .
we do not consider them here because niobium and alluminium microtechnology is more advanced and , what is even more important , even lead or tin or their substrates would have spin impurities requiring environment engineering . in the qubit design of ref.@xcite
the @xmath95 superconducting loop is made of an alluminium @xmath96 wire with persistent current of @xmath97 .
we suggest niobium instead of alluminium and the loop to be just @xmath98 thick .
@xmath98 is roughly twice the penetration depth so that @xmath7 can penetrate through the superconductor .
note that the critical field is higher in such thin films than in bulk superconductor @xcite .
our superconducting loop will produce a flux of @xmath99 which is an order of magnitude less than in the design of ref.@xcite .
the loop is also thinner so it contains fewer spins .
the superconductor volume of @xmath100 contains @xmath101 of niobium nuclear spins .
they are subject to an average @xmath102 .
the quality factor @xmath103 gives the overlap @xmath104 .
if the external field were @xmath71 , then full decoherence would take place after @xmath105 where we assume the nuclear spin magnetic moment of @xmath106 . for our @xmath107 after @xmath108 we get only negligible partial decoherence with @xmath104 .
the full decoherence due to spin relaxation could happen after @xmath109 where we assume @xmath11 to be of the order of minutes , but the dephasing time @xmath110 is much shorter .
even this @xmath111 time can be significantly increased by decreasing temperature because @xmath112 , where @xmath113 is the superconducting gap .
reduction of temperature from @xmath4 down to @xmath114 increases @xmath115 25 times .
maximalization of the quality factor @xmath116 so that @xmath70 makes entanglement with spin environment negligible .
the same technique can be used for any qubit design which involves superposition of states with different magnetic fields . according to our estimates , the present day persistent current qubit proposal @xcite is not far from achieving a @xmath111 dephasing time due to the finite spin relaxation time @xmath11 .
@xmath111 is sufficient for the planned quantum computation at a @xmath117 rate . to achieve this desired decoherence time
it is important to replace alluminium by niobium and make the circuit more microscopic than in ref.@xcite .
it is a common wisdom that more microscopic systems are better from the point of view of decoherence , compare e.g. our formula for @xmath108 .
we showed that in addition to this much expected behaviour , there is a threshold below which a sufficiently microscopic qubit can be made free from entanglement with the spin bath .
99 r.p.feynman , found.phys.*16 * , 507 ( 1986 ) .
p.w.shor , in _ proceedings of the 35th annual symposium on foundations of computer science _
( ieee press , los alamitos , ca , 1994 ) , pp.124 - 134 ; l.k.grover , phys.rev.lett.*79 * , 325 ( 1997 ) .
d.g.cory , m.d.price , w.maas , e.knill , r.laflamme , w.h.zurek,t.f.havel , and s.s.somaroo , phys.rev.lett .
* 81 * , 2152 ( 1998 ) . special issue fortschr.phys .
* 48 * ( 2000 ) .
j.e.mooij , t.p.orlando , l.levitov , l.tian , c.h.van der wal , and s.lloyd , science * 285 * , 1036 ( 1999 ) ; t.p.orlando , j.e.mooij , l.tian , c.h.van der wal , l.levitov , s.lloyd , and j.j.mazo , phys.rev.b * 60 * , 15398 ( 1999 ) .
j.r.friedman , v.patel , w.chen , s.k.tolpygo , and j.e.lukens , nature * 406 * , 43 ( 2000 ) ; c.h.van der wal , a.c.j.ter haar , f.k.wilhelm , r.n.schouten , c.j.p.m.harmans , t.p.orlando , s.lloyd , and j.e.mooij , science * 290 * , 773 ( 2000 ) .
l.tian , l.s.levitov , c.h.van der wal , j.e.mooij , t.p.orlando , s.lloyd , c.j.p.m.harmans , and j.j.mazo , proceedings of the nato - asi on _ quantum mesoscopic phenomena and mesoscopic devices in microelectronics _ , cond - mat/9910062 . n.v.prokofev and p.c.e.stamp , cond - mat/0006054 ; for more general theory see n.v.prokofev and p.c.e.stamp , rep.prog.phys.*63 * , 669 ( 2000 ) .
t.van duzer and c.w.turner , _ principles of superconductive devices and circuits _ , elsevier north holland inc .
, new york , 1981 . | a persistent current qubit has two quantum states with opposite currents flowing in a superconducting loop . their magnetic field couple to nuclear spins .
the qubit state is not only perturbed by the spins but it also gets entangled with the spins state on a very short timescale .
however , when the same spins are exposed to a strong but less than critical external magnetic field , then the qubit field is just a small perturbation on top of the external field and the entanglement with each spin is negligible . for a qubit which is more microscopic than certain threshold this partial entanglement results in negligible partial decoherence . |
the probe of the epoch of reionization ( eor ) remains one of the outstanding aims of modern cosmology . in the past decade ,
intriguing details have emerged about this epoch from a host of cosmological observables .
gunn - peterson ( gp ) tests on spectra of quasi - stellar objects ( qsos ) ( ( @xcite ) in the redshift range @xmath2 suggest that the universe as making a transition to full reionization during this period . on the other hand , cosmic microwave background radiation ( cmbr ) temperature and polarization anisotropy measurements ( @xcite , @xcite ) suggest that the universe might have been fully ionized in a redshift range @xmath3 ( @xcite ) .
both these observables have their strengths and weaknesses .
the gp test , based on lyman-@xmath4 absorption , is not able to distinguish between a fully neutral medium from the one ionized to one part in a thousand .
cmbr anisotropies , based on photon scattering off free electrons , are sensitive to the integrated thompson scattering optical depth and therefore can not reliably construct the tomography of the reionization epoch . in the recent past
, major experimental efforts have been undertaken to study the eor in redshifted 21-cm line emission from the epoch .
in particular , many radio interferometers in frequency range @xmath5 are currently operational that specifically aim to detect the eor , for example , low frequency array ( lofar , @xcite ) , 21 centimeter array ( 21cma , @xcite ) , giant meterwave radio telescope ( gmrt , @xcite ) , donald c. backer precision array for probing the epoch of reionization ( paper , @xcite ) , and the murchison widefield array ( mwa , @xcite ) .
even though the detection of redshifted hi line from the eor remains the most direct and , possibly the most promising , way to delineate the details of the epoch , this method is beset with its own set of issues .
first , unlike cmbr anisotropies , the theoretical modelling of the hi signal from the eor is considerably harder principally owing to uncertainty in the nature of ionizing sources and the details of their formation and evolution .
second , the signal is expected to be weak with brightness temperature @xmath6 .
many hundred hours of observation is needed to detect such a signal with current interferometers , with the attendant complication of maintaining instrumental stability for such long periods .
third , the foreground are expected to be many orders of magnitude larger than the signal ( for details on the three issues see e.g. @xcite and references therein ) .
the use of radio interferometers to estimate the underlying power spectra has been successfully employed for cmb data analysis ( @xcite ) .
this method has also been suggested as a possible probe of the intensity correlations of the redshifted hi line , including from the eor ( @xcite ) .
many different approaches have been discussed to detect the hi signal in the presence of dominant foregrounds @xcite .
they are all based on the expectation that foregrounds are smooth in frequency space as they arise from continuum emission , e.g. synchrotron radiation , in both our galaxy and extra - galactic sources . on the other hand
the hi signal has significant structure in the frequency space .
it is conceivable that all these sources , both point and diffuse , can be subtracted from the images , leaving behind the hi signal and gaussian noise , and lofar partly relies upon this technique ( @xcite ) .
another possible method is based on the isolation of foregrounds from the hi signal using power spectrum of the observed signal in conjugate space to the observed frequency ( @xcite ) .
variants of this ` delay space ' ( @xcite ) method are particularly relevant for interferometers such as mwa that have low angular resolution and have been used extensively for the analysis of paper data . in this approach the observed interferometric data visibilities for each antenna pair as a function of frequency is fourier transformed along the frequency axis .
the fourier conjugate variable effectively captures signal delay between antenna pairs , which allows one to isolate foregrounds . in the context 3d hi power spectrum
this variable can be related to cosmological distance along line of sight ( for details of this mathematical correspondence see @xcite ) . the ` delay spectrum ' constructed from ` delay space ' approach can be used to recover the cosmological 3d hi power spectrum . in this approach , one deals with visibilities directly , which are primary data products of radio interferometers . in this paper
we propose a new method based on the delay space approach to extract the power spectrum of the hi signal in the presence of noise and foregrounds .
our method is based on modelling the hi signal taking into account the impact of the @xmath0-term ( arising from non - coplanarity of the array , @xcite ) and the distortion of intensity pattern during a tracking run .
the information from the hi signal is used as weights to cross - correlate the measured visibilities . the proposed method ( `
delay spectrum with imaging arrays ( dsia ) ' ) is a general method applicable for tracking with radio interferometers with wide primary beams and arbitrary array configuration ( e.g. mwa , lofar ) and can also be applied to interferometers with redundant baselines .
we apply the proposed method to analyse 3 hours of mwa data and compare our results with noise and foreground simulations .
the outline of this paper is as follows . in the next section ,
we describe the method of visibility correlation using delay space approach in detail .
in particular , this method is applied to the hi signal . in section 3
, we describe the mwa data and the initial analysis of this data based on the publicly - available image processing software : common astronomical software applications ( casa ) . in section 4 the pipeline to extract the hi power spectrum
is discussed . in section 5 , the results from mwa data are discussed and compared with simulations of foregrounds and noise . in the final section ,
we summarize our results and indicate possible future directions . throughout this paper , we have used the planck+wp best fit values of cosmological parameters : @xmath7 , @xmath8 , @xmath9 , @xmath10 , @xmath11 , and @xmath12 @xcite .
in this section we study the hi signal using visibility correlations in delay space . using our formulation
we derive , in addition to new results , many results already known in the literature .
the main new results are : the impact of @xmath0-term ( subsection 2.1 ) and changing intensity pattern in a tracking run ( subsection 2.2 ) on the hi correlations .
our results are valid for any radio interferometer but our aim here is to underline their applicability to mwa .
the most important inputs from mwa array configuration for our study are : ( a ) the mwa primary beam and ( b ) the bandwidth .
other properties of the mwa array that have a bearing on our analysis are the non - coplanarity of the array and its baseline distribution . in delay space ,
spectrally smooth foregrounds lend themselves to ready interpretation .
it can be shown that visibilities computed in delay space allow isolation of such foregrounds from the regions dominated by the eor signal and noise ( e.g. see @xcite ) .
this can be achieved by fourier transforming the raw visibilities in frequency space . however , being 3-dimensional and statistical in nature , the properties of the hi signal can only be inferred by correlating the observed visibilities .
our approach , which is based on visibility correlations in delay space , allows us to develop a unified method to deal with both the hi signal and the foregrounds , which are discussed in appendix [ sec : foregrounds ] .
radio interferometers measure the spatial correlation of the electric fields from the sky , the visibility @xmath13 : @xmath14 here @xmath15 is the distance vector between the antennas of the interferometer ( also called baseline vector ) in units of wavelength , @xmath16 denotes the position vector on the sky plane ( which can be be expressed as a pair of direction cosines @xmath17 ) and @xmath18 is the frequency of observation .
we neglect the impact of @xmath0-term in this section in writing the relation between the visibility and specific intensity in eq .
( [ visibility ] ) .
the @xmath0-term arises from non - coplanarity of the interferometric array and changes as the interferometer tracks a region . in section [ sec : h1sigwterm ] we show that the inclusion of @xmath0-term causes an effective shrinking of the primary beam . for our study
we calculate how the hi signal is affected owing to non - zero @xmath0 as compared to @xmath19 case .
we explicitly express the frequency dependence of all the quantities .
these quantities are : the background specific intensity @xmath20 , the primary beam @xmath21 and the baseline @xmath22 . in terms of brightness
temperature @xmath23 , @xmath24 .
the sky intensity can be decomposed as : @xmath25 where @xmath26 and @xmath27 are the isotropic and fluctuating parts of the intensity distribution .
since the isotropic component does not contribute to interferometric measurement , the visibility recorded at frequency @xmath18 can be written as : @xmath28 the measured visibility receives contributions from the redshifted hi line , the foregrounds , and the noise .
for the hi signal the observed intensity fluctuations can be related to the hi perturbations in fourier space , @xmath29 , as : @xmath30 here @xmath31 specifies the three - dimensional position of the the hi emission ;
@xmath32 is the coordinate distance to the point of observation : @xmath33 with the limit of this integral extending from zero to redshift @xmath34 . @xmath29
comprises of many physical effects : density fluctuations , ionization inhomogeneity , density - ionization fraction cross - correlation , etc .
( @xcite ) . together
this can be expressed as : @xmath35 each quantity in the above equation corresponds to the fractional variation of a particular physical quantity : @xmath36 refers to fluctuation in baryonic density , @xmath37 for the ly@xmath4 coupling coefficient @xmath38 , @xmath39 for the neutral fraction , @xmath40 for @xmath41 , and @xmath42 for the line of sight peculiar velocity gradient .
@xmath43 factors denote the expansion coefficients of the corresponding quantity ( @xcite ) .
current experiments such as mwa , lofar and paper aim statistical detection of the eor signal .
the quantities of interest here are the correlation functions of the hi fluctuations .
the most important correlation function one seeks to detect in an eor experiment is the power spectrum , @xmath44 , defined as : @xmath45 the hi power spectrum can be constructed from the correlation of the observed visibilities . substituting the form of fluctuation @xmath46 from eq .
( [ deli_a ] ) in the visibility expression ( eq .
( [ vis_a ] ) ) , we get : @xmath47d^2\theta \end{aligned}\ ] ] here we have decomposed the wave vector @xmath48 as components on the plane of the sky , @xmath49 and along the line of sight , @xmath50 .
the integral over angles is the fourier transform of the primary beam @xmath51 , which allows us to re - write this equation as : @xmath52 where @xmath53d^2\theta\end{aligned}\ ] ] using eq .
( [ powspec3d ] ) , the visibility correlation function can be computed : @xmath54 here @xmath55 which for @xmath56 can be simplified to : @xmath57 , where @xmath58 .
( [ viscorrfreq ] ) gives the correlation of the hi signal in three dimensions in which the two coordinates @xmath59 correspond to fourier components of the hi signal while the third @xmath18 refers to the coordinate of the fluctuation in position , @xmath32 , space ( @xcite ) . to isolate the impact of foregrounds and
obtain regions dominated by the hi signal and the noise ( ` eor window ' ) , we compute the the visibilities in delay space ( @xcite ) : @xmath60 here @xmath61 , the conjugate variable of @xmath18 , defines the relevant variable in delay space .
the delay space approach can be applied to data to isolate spectrally smooth foregrounds ; we discuss the delay space approach as applied to such foregrounds in appendix a. in eq .
( [ deftau ] ) , we have suppressed the frequency dependence of the baseline on the lhs as the frequency dependence of all the quantities has been integrated . the baseline vector can be expressed as : @xmath62,where @xmath63 is some fixed frequency that lies within the bandwidth . on the lhs of eq .
( [ deftau ] ) , the frequency independent baselines , @xmath64 . throughout this paper ,
we assume : @xmath65 , the central frequency of the bandwidth we use for mwa data analysis .
the autocorrelation of @xmath66 can be written as : @xmath67 to make further progress , frequency dependent quantities are taylor expanded . for baselines ,
this is a straightforward re - expression of the baseline vector as the vector is linear in frequency : @xmath68 where @xmath69 .
it should be noted that @xmath70 is the physical baseline length measured in the units of time .
after the taylor expansion of relevant quantities , @xmath32 and @xmath22 , we obtain : @xmath71 \label{etacorrela12}\end{aligned}\ ] ] here @xmath69 ; all the quantities in eq .
( [ etacorrela12 ] ) have been written as explicit functions of @xmath18 and @xmath72 .
this allows us to simplify the integral further by making the coordinate transform @xmath73 and @xmath74 ; the jacobian of this transformation is unity .
we can make further simplification by using @xmath75 .
this is justified for our case as we assume the bandwidth to be @xmath76 around a central frequency of @xmath77 .
all the frequency dependent variables change by less than @xmath78 for this case .
this reduces eq .
( [ etacorrela12 ] ) to : @xmath79 \label{etacorrela1}\end{aligned}\ ] ] given the hi power spectrum @xmath80 this integral could be computed numerically .
however , it is possible to determine the correlation scales in both the transverse and line of sight directions by carefully examining eq .
( [ etacorrela1 ] ) .
the integral over angles shows that the dominant contribution comes from wavenumbers such that @xmath81 .
this relation allows us to simplify the integrals over @xmath82 , @xmath83 and @xmath84 .
in particular , different terms in the exponent containing @xmath61 can be estimated .
using @xmath85 , the last two terms in the exponents are on the order of @xmath86 ( the term containing @xmath87 is slightly smaller because @xmath88 ) . for mwa primary beam , @xmath89 , and for mwa
baseline distribution , the term @xmath90 generally dominates over these terms , especially in the regions dominated by eor . for all our calculations we use parameters specific to mwa , in particular , the primary beam of mwa .
however , the formulation presented here is general enough to be applicable to other arrays . by dropping the last two terms , which are subdominant , in the exponent containing @xmath61 , we can separate the integrals over @xmath84 and angles , this gives us : @xmath91 \label{etacorrela2}\end{aligned}\ ] ] the integral over @xmath84
can readily be carried out now . the dominant contribution to
the integral comes from @xmath92 , which establishes the correlation scale in the direction along the line of sight .
the variation of frequency dependence of integrals over @xmath82 and @xmath83 is expected to be small for the bandwidth of mwa and therefore these integrals can be computed at some frequency that lies within the bandwidth . in this paper we assume these values to be fixed at the central frequency @xmath93 and use them to compute the relations in eq ( [ ukrelations ] ) . if the frequency dependence of the primary beam and the background intensity are neglected , the integral over @xmath18 in trivial . as noted above , this is a good approximation for mwa . the power spectrum of the hi signal , based on eq .
( [ etacorrela2 ] ) , is shown in figure [ signal_ps ] . in this calculation
, we use the hi power spectrum @xmath80 given by the simulation of @xcite .
where p is in units @xmath94.,scaledwidth=50.0% ] eqs .
( [ etacorrela2 ] ) and ( [ viscorrfreq ] ) can be used to infer many important properties of the hi signal : * there exists a near one - to - one relation between * u * , @xmath61 and the fourier components of the hi power spectrum ( @xcite ) : @xmath95 where @xmath96 is the rest frame frequency of the 21 cm line , @xmath32 is the transverse comoving distance , @xmath97 and z is the redshift corresponding to the observed frequency @xmath18 .
the relation between @xmath50 and @xmath61 follows from the relation : @xmath98 ( eq . ( [ etacorrela2 ] ) ) . as noted above , all the frequency dependent quantities in eq .
( [ ukrelations ] ) are computed as a fixed frequency @xmath65 . *
the correlations in the sky plane and along the line of sight are nearly separable .
this allows us to compute weights in the plane of the sky owing to @xmath0-term and the distortion of intensity pattern in a tracking run ( the next two subsections ) without the additional complication owing to frequency dependence of these quantities . *
( [ etacorrela2 ] ) and ( [ ukrelations ] ) allow us to simplify the relation between visibility correlation and the hi power spectrum .
( [ etacorrela2 ] ) can be solved in the limit defined by eq .
( [ ukrelations ] ) to give ( e.g. @xcite ) : @xmath99 here the mwa primary beam solid angle @xmath100 . for mwa @xmath101 at 150 mhz ( @xcite ) .
@xmath102 is the total band width we use in this work .
the mean specific intensity @xmath103 . this allows us to express the hi signal as the square of the product of mean brightness temperature and the hi power spectrum @xmath104 in the units @xmath105 .
it should be emphasized that eq .
( [ normali ] ) provides the suitable normalization only when @xmath106 , @xmath19 , as has been assumed throughout this section , and the impact of sky intensity distortion while tracking a region is not considered .
all these effects act to lower the rhs of eq .
( [ normali ] ) , the measured visibility correlation , for a fixed signal @xmath104 .
this is accounted for by appropriate weights we discuss in the next two sections . from eqs .
( [ etacorrela1 ] ) and ( [ etacorrela2 ] ) , we can gauge the impact of the @xmath0-term . these equations and
the discussion following them shows that the angular integrals depend only weakly on the line of sight variables .
the main effect of the @xmath0-term is to alter the integrals over angles which we study here . for a given baseline @xmath107 : @xmath108 , where @xmath109 is the phase center at any time . as a region
is tracked , the @xmath0-term changes owing to the drift of the phase center . for a tracking run , @xmath110 is left invariant at any frequency ; this result simply follows from the fact the the baseline length is fixed .
after the inclusion of the @xmath0-term , the measured visibility for a given intensity distribution is given as : @xmath111 } dldm \label{visw}\end{aligned}\ ] ] here we have replaced @xmath16 with its components ( l , m ) and also made the approximation:@xmath112 , which is valid for mwa primary beam . after substituting eq .
( [ deli_a ] ) into eq .
( [ visw ] ) gives us : @xmath113dldm \label{viswithw}\end{aligned}\ ] ] each mwa tile being approximately a square aperture , the primary beam @xmath114 can be written as : @xmath115 here @xmath116 and @xmath117 are dimensionless .
they correspond to the ratio of the length of the tile along x- and y - axis to the wavelength . for central wavelength of the observation @xmath118 .
eqs ( [ visw ] ) and ( [ pribeam ] ) show that integrals over @xmath119 and @xmath120 are separable and identical .
these integrals can not be done analytically but under certain approximations meaningful analytic expressions can be found .
let us define : @xmath121 \frac{\sin(\pi l_x l)}{\pi l_x
l } \label{intprib}\end{aligned}\ ] ] @xmath122 is a function of @xmath123 and is parametrized by @xmath124 and @xmath125 .
first we consider , @xmath19 . in this case
, it can be shown that if the limits of the integral are allowed to go from minus infinity to plus infinity , we obtain , @xmath126 we notice that the approximation used is good because the function has a compact support provided by the primary beam .
as @xmath127 where @xmath128 is the extent of the primary beam , this result means that , for a given @xmath124 , the wavenumbers that contribute to the integral are the ones that are bounded by the extent of the primary beam .
this result is already implied by eq .
( [ etacorrela1 ] ) .
( [ intprib ] ) can not be analytically approximated so readily for non - zero @xmath0 .
we use the stationary phase approximation to analytically evaluate the integral . for this assumption to hold , the phase of the exponent should be much larger than the slow variation of the primary beam
. this would be the case if @xmath129 is large . in this approximation
, we obtain : @xmath130 \label{wtermpb}\end{aligned}\ ] ] the main impact of the inclusion of the @xmath0-term can be discerned from this expression . in the limit of large @xmath0 , the impact of the @xmath0-term is to shrink the mwa beam and the primary beam tends to @xmath131 ( @xcite).-term can be more readily computed if the beam is gaussian ( e.g. appendix b of @xcite ) in this case , the primary beam approaches @xmath132 for non - zero @xmath0 .
] however , this also means that the spread of @xmath123 for which the integral is non - zero also increases , as seen in the terms involving the @xmath133 function . if the decrease of primary beam results in a loss of signal - to - noise , an increase in the correlation length @xmath134 gains signal - to - noise .
we can write visibility correlation for pairs of @xmath135 and @xmath136 as : @xmath137 eq .
( [ wtermvisvis ] ) can be computed numerically . in figure
[ fig : wterm1 ] , we show how the hi correlation function is affected in the presence of @xmath0-term .
these expressions are also valid for diffuse foregrounds which have a different 2-dimensional power spectrum and frequency dependence , e.g. optically thin synchrotron radiation for which the angular and frequency dependence is separable . as eq .
( [ wtermvisvis ] ) can be used to compute the impact of @xmath0-term at any frequency , it can readily be generalized to study diffuse foregrounds . for different values of w.
the impact of w - term is to decrease the overall signal owing to shrinking of the primary beam and increase the correlation length @xmath134.,scaledwidth=50.0% ] in a tracking interferometric observation , a phase center is tracked and snapshots are taken at regular intervals with short duration .
each of these snapshots can be imaged and the images added if the successive fields of view can be assumed to be coplanar .
this approximation breaks down for wide field - of - view instruments such as mwa .
one manifestation of the wide field - of - view is the @xmath0-term whose impact was studied in the previous sub - section . in this section
we generalize the discussion of the last sub - section to take into account the time dependence of the non - coplanarity of the tracked region ( @xcite ) . as the region
is tracked , the relation between the image and astronomical coordinates changes which distorts the intensity pattern with respect to the phase center being tracked .
more specifically , this effect arises from the projection of a non - coplanar array on a plane , which is necessary to perform the fourier transform for imaging .
it is best illustrated with a set of point sources .
these sources appear to move with respect to the phase center ( e.g. fig .
19 - 9 in @xcite ) .
the distortion of the intensity pattern corresponds to non - uniform stretching and it increases for sources further away from the phase center .
thus this effect can not be corrected by a standard shift of coordinate .
the non - uniform stretching makes the situation complex , and the standard grid approach is difficult to implement in this case . for a set of point sources , the correction for this effect
could be applied iteratively in the image plane ( chapter 19 , @xcite ) . for a small field - of - view ,
this effect can be neglected and a unique coordinate system ( e.g. time independent direction cosines @xmath17 ) can be used to relate the image coordinates with the sky intensity pattern for a long tracking run .
however , it is not possible to define such a coordinate system when either the field of view is large or the tracking period is long .
our aim here is not to correct for this effect but rather to estimate its impact on the correlation of visibilities at two different times during a tracking run : suppose we measure visibilities within a small cell in the @xmath138@xmath139 plane ( the size of the cell will be discussed in a later section ) centered around a baseline @xmath140 at @xmath141 . at a later time
@xmath142 another baseline might enter this cell . from the discussion in the previous subsections ( e.g. figure [ fig : wterm1 ] ) the two visibilities are expected to correlate strongly with each other ( even if the values of @xmath0 differ significantly for these two sets this statement is generally true ) .
however , visibilities measured at two different times do not correspond to the same intensity pattern .
our aim here is to estimate the level of de - correlation caused by the distortion of intensity pattern during a tracking run . in this paper , we construct a time - dependent coordinate system which allows us to analyse this distortion of intensity pattern .
we assess the impact of this effect when a region is tracked using the mwa primary beam .
in particular , we consider this effect on the visibilities produced by the eor hi signal .
we start by recalling the definition of direction cosines for a point on the sky whose coordinates , declination and hour angle @xmath143 , are : written as ( @xcite ) : @xmath144 it can be shown that @xmath145 .
the phase center is always defined as @xmath146 , @xmath147 , @xmath148 ; for the coordinates defined above it is : @xmath149 and @xmath150 . as a phase center
is tracked owing to the rotation of the earth , @xmath151 remains fixed but the hour angle changes . for a wide field of view , this can result in distortion of the intensity pattern of the sky . to take into account this effect
, we can define a time - dependent coordinate system : @xmath152 here @xmath153 and @xmath154 define the phase center for @xmath155 ; @xmath156 defines the flow of time .
@xmath120 can be similarly defined and @xmath157 can be computed from @xmath119 and @xmath120 .
this definition gives a time dependent coordinate system where the coordinates are always defined with respect to the phase center .
it is easy to verify that for small field of view and for small tracking times , which corresponds to cases when higher order terms in @xmath158 , @xmath156 and @xmath151 can be dropped , @xmath119 is independent of time which means that the distance of a point from the phase center is left invariant under tracking . in such cases
, the intensity pattern on the sky corresponds to the unique intensity pattern defined by sky coordinates @xmath151 and @xmath158 and remains unchanged as the phase center is tracked .
however , when this approximation breaks down , @xmath119 becomes a function of time and it is impossible to define a unique relation between direction cosines and sky coordinates .
this means that any quantities defined with respect sky coordinate ( e.g. intensity pattern ) become time dependent .
the visibility for the hi signal is given by : @xmath159\sin\delta d\delta dh \label{viswithw}\end{aligned}\ ] ] the direction cosines @xmath119 and @xmath120 are now functions of time . the angular integral is carried out over @xmath151 and @xmath158 . unlike the earlier case ( fixed grid ) this is not a product of two one - dimensional integrals .
the correlation of the visibilities @xmath160 can be computed using the same methods as outlined in the previous sections . in figure [ wterm2 ]
we show the results when the effect of the time dependent coordinate system ( moving grid ) is included . the initial phase center ( @xmath161 ) is chosen to be @xmath162 and @xmath150 .
the results are shown for two different values of @xmath156 and a range of @xmath0 values .
we only show the auto - correlation function for a given value of @xmath156 .
but the results shown in figure [ wterm2 ] can be used to assess the cross - correlation of visibilities measured at two different times . for our case the value of this cross - correlation lies between the auto - correlations of visibilities measured at the same time .
the moving grid does nt introduce another scale in the problem and the results in this case are not very different from the case for a fixed grid . in both cases
the dominant correlations occur for @xmath163 .
figures [ fig : wterm1 ] and [ wterm2 ] are based on mwa primary beam . however , it is possible to glean generic information applicable for other primary beams from them .
first , the decorrelation length @xmath164 scales as the inverse of the primary beam ( e.g. @xcite ) .
so for a smaller beam , the decorrelation seen in the figures as a function of @xmath164 would be shallower .
the impact of the @xmath0-term for a different primary beam can be partially gauged from eq .
( [ wtermpb ] ) , which is valid for large values of @xmath0 . in this limit ,
the primary beam tends to @xmath131 , irrespective of the primary beam of the telescope .
it is difficult to analytically estimate the impact of the @xmath0-term when this limit does not hold .
but it can be shown that the impact of @xmath0-term diminishes for a smaller primary beam , e.g. a gaussian beam for which the primary beam tends to @xmath132 for non - zero @xmath0 ( e.g. @xcite ) . as noted above
, the distortion of intensity pattern during a tracking run is a wide field effect . for a smaller primary beam
, the level of decorrelation seen in figure [ wterm2 ] would be smaller but it is difficult to analytically estimate it . for different values of w and @xmath156 .
the initial phase center is chosen to be @xmath162 and @xmath150 .
, scaledwidth=50.0% ] eq .
( [ viswithw ] ) can be used to compute the counterpart of eq .
( [ etacorrela1 ] ) which takes into account the impact of non - zero @xmath0-term and the distortion of intensity pattern .
we compute this expression for the visibility correlation in delay space numerically . in this formulation ,
the measured visibility is a function of five parameters : @xmath165 . here , as noted above , @xmath59 and @xmath0 are the values of these variables at a fixed frequency which we choose to be @xmath65 .
we define the weight on a given cross - correlation as : @xmath166 the weights are defined with respect to the hi cross - correlation computed in eq .
( [ etacorrela1 ] ) for @xmath167 , @xmath168 , @xmath142 and @xmath169 .
we only consider the case @xmath170 for the computation of weights .
using eq .
( [ weightsdef ] ) allows one to recover the hi power spectrum for a fixed wave number @xmath171 from visibility cross - correlations .
the mwa is a low frequency radio interferometer array located in western australia .
it consists of 128 antenna tiles with each tile comprising of 16 crossed dipole antennas over a metal ground screen in 4 x 4 grid .
mwa bandwidth is 30.72 mhz , divided into 24 coarse channels of width 1.28 mhz each .
the total bandwidth is divided into 768 fine channels . with the use of an analog beamformer
appropriate phase delays are introduced in each individual dipole antenna to track the pointing center of the beam across the sky . for more information on mwa
please see @xcite . to minimize the effect of galactic synchrotron emission ,
the mwa eor community has chosen three fields on the sky away from the galactic plane .
these fields have been named as eor0 , eor1 and eor2 and are shown in figure [ eor_fields ] . in this paper
we present 3 hours of tracking analysis of the eor1 field centered at ra @xmath172 4h , dec @xmath173 .
many research groups are currently developing pipelines to extract statistical information from radio interferometric data , with an aim to detect the hi signal from eor ( @xcite ) .
these can be divided broadly into two categories : image based and visibility based pipelines . for foreground subtraction and imaging these pipelines
use the following imaging algorithms : real time system ( rts ; @xcite ) and fast holographic deconvolution ( fhd , @xcite ) .
the image based pipelines ( @xcite ) use source catalog created through the deconvolution of the data which is subtracted to obtain a residual image cube .
the fourier transform of this image cube with some further processing yields the power spectra .
on the other hand , the visibility based pipelines ( @xcite ) use the data in visibility domain for power spectra estimation , after the initial processing in the image domain for obtaining the foreground model .
the detailed comparison of the outputs from all the methods described above is provided in @xcite .
@xcite describe the impact of wide field of view in power spectra estimation .
a special variant among the visibility based estimators is ` delay spectrum ' ( @xcite ) , which directly fourier transforms each calibrated visibility along its frequency channels .
the radio interferometer paper uses this approach extensively ; it relies upon redundant baselines to calibrate the interferometer and east - west and near east - west baselines for power spectrum estimation ( @xcite ) .
this particular scheme has been discussed and implemented only for redundant drift scan observations . in this paper
we explore the possibility of applying this approach for non - redundant imaging arrays and tracking measurements . in this and the next section
, we discuss in detail our method of mwa data analysis and power spectrum estimation from the data .
we summarize below the major ingredients of the method and then describe each of the stpng in detail in subsequent sections : * casa ( @xcite ) is used for initial processing of the data to calibrate raw visibility measurements .
this is followed by the creation of a model sky image from clean components .
this model is then subtracted in the visibility domain to obtain residual visibilities .
we use both the calibrated and residual visibilities for computing the power spectrum . *
each visibility is then fourier transformed in frequency space ( eq . ( [ visetaspace ] ) ) .
this process is needed for isolation of foregrounds in the @xmath174 plane .
we note the our method utilizes both the subtraction of foregrounds and their isolation .
but it does not employ an external point source catalog . * the procedure outlined above yields complex visibilities as a function of five variables : @xmath175 . for computing the power spectrum we cross - correlate these visibilities for @xmath176 to remove the noise bias .
to weigh each cross - correlation we assume that there exist regions in @xmath174 plane which are dominated by only noise and the hi signal .
this allows us to compute a weight for each cross - correlation based on the expected hi signal . for computing these weights
we take into account the impact of @xmath0-term and the distortion of intensity pattern in a tracking scan .
the relevant method is elaborated in detail in sections [ sec : h1signal ] , [ sec : h1sigwterm ] , [ sec : movinggrid ] , and [ sec : weightscross ] and summarized in section [ sec : weights ] . * in section [ sec : powspecest ] , we describe the power spectrum estimator , taking into account weights given by the expected hi signal , in 3- , 2- and 1-dimension .
we also discuss our method to compute the errors on the estimated power spectrum .
mwa data were collected at 2-minute intervals with a time resolution of 0.5 seconds and frequency resolution of 40 khz .
the central frequency of these observations is 154.24 mhz . for preprocessing
we have used the cotter pipeline ( @xcite ) to average to 10 seconds of integration ; we have not performed any averaging over the frequency channels .
cotter also uses the in - built aoflagger to flag and remove radio frequency interference .
the edge channels of each coarse band are flagged with cotter due to aliasing effects .
after this preprocessing the cotter pipeline delivers the data in the casa readable ` measurement set ( ms ) ' format for further processing .
once the ` ms ' files are produced for each 2-minute data set , we process each of these 2-minute data in casa to produce an image .
the hydra a source is used to calculate the bandpass solutions which are applied to the uncalibrated data . we next construct a sky model from these data so that we could subtract it to obtain the residual visibility . after the bandpass calibration the first round of ` clean ' is applied on each 2-minute data set .
the multi - scale multi - frequency synthesis algorithm ( @xcite ) has been used for imaging .
we have created images of size 3072 x 3072 pixels with 1 arc - minute cell size using the cotton - schwab clean ( @xcite ) with uniform weighting scheme .
after the first round of clean we have performed self calibration ( both phase and amplitude+phase ) and apply the clean loop until the rms value of the residual image converges .
the threshold limits for the clean stpng were chosen to be 5 sigma . the w - projection algorithm ( @xcite )
was also used to correct for the errors arising due to non - coplanarity of baselines .
once we obtain the best model of the sky for each 2 minute observation , the model visibilities are then subtracted from the calibrated data using the uvsub algorithm in casa to obtain the residual data .
this process is followed for both xx & yy polarizations separately .
a flow chart of the data pipeline is shown in figure [ pipeline ] . in figure
[ eor1_image ] , we present a sample image of 2 minute deconvolution .
+ as noted above we process the data for only 2 minutes to ensure the primary beam does nt substantially change during the run . for a 2-minute scan
we obtain an rms of nearly 40 mjy / beam .
the residual visibility @xmath177 is a function of five variables .
we compute the discrete fourier transform of the residual visibilities in the frequency space weighted by the blackman - nuttall ( @xcite ) window @xmath178 to suppress leakage into the eor window ( @xcite ) : @xmath179 notice that in eq .
( [ visetaspace ] ) the frequency dependence of the baseline vector @xmath180 is integrated over . therefore , the labels @xmath181 on the lhs of eq .
( [ visetaspace ] ) need further explanation . as noted above ( the discussion following eq .
( [ deftau ] ) ) they can be chosen to denote a given baseline vector at a fixed frequency , @xmath63 .
we choose this frequency to be the central frequency of the band @xmath65 .
@xcite provide detail implications of the frequency dependence of the baseline vector . here
@xmath182 and 256 channels are used for our study , which correspond to total bandwidth 10.24 mhz in the frequency range 149.09 mhz to 159.34 mhz .
the visibilities ( eq . [ visetaspace ] ) are cross - correlated with weights determined from the hi signal ( section [ sec : weightscross ] ) to estimate the power spectrum . for each pair of parameters , e.g. @xmath183 ,
the weights are generally different .
it is computationally prohibitive to deal with weights for all cross - correlations .
we make several simplifying assumptions to make the problem tractable based on the properties of the hi signal . in sections [ sec : h1signal ] , [ sec : h1sigwterm ] , and [ sec : movinggrid ] we discuss in detail the hi signal and how it is affected by the inclusion of the @xmath0-term and the additional complication arising from distortion of the field of view as a region is tracked for mwa .
we summarize the main results of these sections as applied to the data : * in section [ sec : h1signal ] the hi signal and its correlations are discussed in detail .
( [ etacorrela2 ] ) shows that correlations in sky plane are nearly independent of correlations along the line of sight .
this allows us to compute weights for correlations in the plane of the sky independent of the third axis .
( [ etacorrela2 ] ) allows us to derive a relation between the measured correlation and the inferred hi power spectrum ( eq .
[ normali ] ) .
( [ normali ] ) defines the scale of cross - correlation weights .
the weight function @xmath184 is unity when @xmath167 , @xmath168 and @xmath185 .
( [ normali ] ) refers to this case . * in section [ sec : h1sigwterm ] , the impact of @xmath0-term on the hi signal is computed .
( [ wtermpb ] ) and figure [ fig : wterm1 ] capture the effect of non - zero @xmath0 on hi correlations .
the @xmath0-term diminishes the signal by shrinking the effective primary beam and increases the correlation length scale @xmath134 .
we use the analytic expression based on eq .
( [ wtermpb ] ) for computing weights for @xmath186 . * in section [ sec : movinggrid ] , we attempt to assess the impact of time - dependent distortion of intensity pattern in a tracking run for mwa .
figure [ wterm2 ] shows the combined effect of moving grid and @xmath0-term .
the distortion of intensity pattern generally acts to enhance decorrelation but is found to be not significant and does nt alter the main features of the signal . for our computation , we only update the weights after every 10 minutes to account for this effect . as shown above each correlation
receives a different weight depending on the values of @xmath187 of the baselines being correlated . as noted above
, we define the weights @xmath188 such that they approach unity when @xmath189 , @xmath190 , @xmath168 and @xmath185 such that the effect of the moving grid is not important ( @xmath191 for all correlations to remove noise bias ) .
the hi power spectrum is a function of @xmath192 ; eq . ( [ ukrelations ] )
gives the relation between the fourier components of the hi signal and @xmath193 .
all cross correlations for which the wave vector lies in some range @xmath171 and @xmath194 can be used to construct the unbiased hi signal : @xmath195 ; here @xmath196 is the number of all pairs for which @xmath171 lies in the range specified above .
however , this estimator , though unbiased for the hi signal , could be dominated by small values of weights @xmath197 , which does nt make it the lowest noise ( or optimal ) estimator .
as the observed signal is dominated by noise , we consider an optimal estimator for our study : @xmath198 where @xmath199 . to avoid noise bias , @xmath200 for all cross - correlation . for a given @xmath201 , eq .
( [ powspecdef ] ) allows us to compute the power spectrum by optimally weighing over all the cross correlations .
however , as figures [ fig : wterm1 ] and [ wterm2 ] show the correlations fall substantially for @xmath202 ( see also @xcite and references therein ) . this motivates us to pixelize the @xmath59-plane and consider only those visibility pairs for which the correlations are significant .
we consider cells of different sizes and present results here for @xmath203 .
the number of visibility measurements in a cell vary depending on the ( u , v ) values .
the shortest baselines have higher population as expected for mwa . for 3 hours of analysis and @xmath204 , the number of visibilities in a cell lie in the range @xmath205 where each visibility has a time resolution of @xmath206 sec .
all the cross - correlation within a cell are computed using eq .
( [ powspecdef ] ) . for averaging over different cells ,
each cell is assigned an average weight corresponding to the rms of the power spectrum for a cell , @xmath207 .
these weights are then used for optimally averaging the power spectrum ( eq .
( [ powspecdef ] ) ) over other cells ( for details see appendix b ) .
note that this procedure allows us to separate large correlations of the hi signal , the ones for which @xmath197 is close to unity , from the ones which are expected to be incoherent because @xmath197 is small .
the schematic of the two processes the computation of power spectrum in 3- and 2-dimensions is displayed in figure ( [ schem ] ) : the top panel delineates the process of computing cross - correlations within each cell and the bottom panel depicts how azimuthal average for a fixed baseline length @xmath208 is computed . for mwa data , @xmath209 , which means the value of @xmath171
is dominated by the value of @xmath50 .
this suggests the following method for computing the 1-dimensional power spectrum , which we adopt : all the cells for a given @xmath61 are optimally averaged using the method described above .
this procedure yields a complex number . in the figures that display 2- and 1-d power spectra we plot the absolute value of the estimated power spectrum .
+ + the error on power spectrum in 1-dimension is computed by first estimating the rms for each cell , @xmath207 .
@xmath207 are then used as weights for optimal averaging over all the cells for a fixed @xmath61 .
the resultant rms after averaging over the cells approaches @xmath210 if the power spectrum across cells is uncorrelated .
this holds for noise but , as noted above , is an approximation for the hi signal .
we expect this assumption to be valid in our case as the observed signal is dominated by noise ( for detailed explanation see appendix b ) .
in figure [ ps ] , the power spectra computed from 3 hours of data are shown in the @xmath174 plane . the power spectra for both xx and yy polarization
are shown for calibrated and residual visibilities .
+ + + we first discuss discernible features in the power spectra : * in fourier space the foreground contributions occupy a wedge shaped region ( also called ` foreground wedge ' ) owing to the smooth spectral characteristics of foreground sources ( @xcite ) .
the region beyond the foreground wedge is expected to be free from the foreground contamination and dominated by thermal noise and expected hi signal .
this foreground isolation approach is particularly useful for the mwa as it has low angular resolution .
the strongest hi signals lie in the shortest baselines ( low @xmath211 values ) and it decreases rapidly with increasing @xmath211 values .
+ the upper and lower panels of figure 8 display the dirty ( calibrated with no foreground subtraction ) and the residual ( clean components subtracted ) power spectra , respectively .
the figure [ ps ] bear out the assumption that foregrounds have smooth spectral characteristics as they are seen to form the ` foreground wedge ' , this separation is in good agreement with the expectation from foreground simulations ( figure [ forenoise ] ) .
the first few @xmath50 modes exhibit maximum foreground contributions , the @xmath212 mode being the strongest .
the amplitude at this mode is roughly @xmath213 which is in good agreement with the results of other mwa eor pipelines ( @xcite ) .
a clear decrement in power in the ` foreground wedge ' is visible in residual power spectra as compared to the dirty one . *
as described in the previous section , mwa has missing channels on either side of coarse bands of width 1.28 mhz .
this leads to a periodicity of missing data across the frequency axis in visibility , the effect of which is reflected in the fourier - transformed power spectra as the horizontal bright lines at fixed @xmath214 .
figure [ 1d_power ] shows the 1-dimensional power spectra ; the 1-dimensional power spectrum is obtained from regions that exclude the foreground wedge and bright coarse bands in figure [ ps ] . for computing the 1-d power spectrum , the foreground wedge and the bright coarse horizontal bands of the 2-d power spectrum ( figure [ ps ] )
are rejected .
more specifically , the channels corresponding to @xmath215 and @xmath216 are not considered . for each coarse band , the central brightest channel along with two channels on either side
are excluded .
the remaining contiguous regions are used in estimation of the 1-d power spectrum .
for instance , for a given @xmath50 that meets the criterion outlined above , all the cells that correspond to @xmath217 are used for the computation of 1-d power spectrum .
the error on the binned power spectra are computed using a scheme outlined in appendix [ sec : powspecest1 ] .
the 2-d power spectra obtained from data ( figure [ ps ] ) can be compared with figure [ forenoise ] that shows the expected power spectrum based on simulations of foregrounds and noise . in particular
, this comparison allows us to assess the structure of coarse channels and the foreground wedge .
it also indicates the range of scales of the 2-d power spectrum .
however , figure [ forenoise ] is based on a single realization of noise and a model of foregrounds based on random distribution of point sources , and therefore a more detailed comparison between the data and simulations is not possible .
we shall return to this detailed comparison in future work .
in this paper , we propose a new method to extract the hi power spectrum from mwa visibility data in delay space .
the proposed method is applicable when a region is tracked using imaging radio interferometers .
one of the crucial factors in power spectrum estimation is how the @xmath0-term is dealt with within the pipeline .
our findings are that the @xmath0-term causes an effective shrinking of primary beam which reduces the contribution of the hi signal .
we carefully model the hi signal by taking the @xmath0-term into account , the weights calculated are then applied to cross - correlate the measured visibilities . moreover , the cross - correlation approach is particularly useful to minimize various systematics in the system .
we also model and account for the impact of changing intensity pattern in a tracking run .
we find this effect to be sub - dominant to the @xmath0-term correction .
we analyse three hours of mwa data from the eor1 field , one of the field identified by the mwa community for eor science .
casa has been used for calibration and to create foreground model using the clean components .
both the dirty ( calibrated with no foreground subtraction ) & residual ( foreground model subtracted ) power spectrum in delay space are presented .
our results ( figures [ ps ] and [ 1d_power ] ) are in good agreement with the analyses of other mwa eor pipelines ( @xcite ) . in the future
we plan to apply the method proposed here for more integration time and , in particular , to a longer single tracking run .
the decorrelation caused by the @xmath0-term and the changing intensity pattern would be more dominant in the latter case .
this will allow us to test the efficacy of our method for more extreme cases and might indicate the best possible way of detecting the hi signal from the epoch of reionization .
we thank the referee for useful comments which helped us to improve the paper .
this scientific work makes use of the murchison radio - astronomy observatory , operated by csiro .
we acknowledge the wajarri yamatji people as the traditional owners of the observatory site .
support for the operation of the mwa is provided by the australian government department of industry and science and department of education ( national collaborative research infrastructure strategy : ncris ) , under a contract to curtin university administered by astronomy australia limited .
we acknowledge the ivec petabyte data store and the initiative in innovative computing and the cuda center for excellence sponsored by nvidia at harvard university .
the primary contribution to foregrounds come from spectrally smooth point and diffuse sources .
they differ from the hi signal in both spatial and spectral behaviour . however , it is the latter difference that allows us to potentially isolate foregrounds from the hi signal in the power spectrum estimation . to understand the impact of foregrounds in the data , we model them as a set of point sources .
we note that if both the point and the diffuse sources have smooth spectra across the instrumental bandwidth , their impact on the power spectra are similar and therefore point sources allow us to capture adequately our ability to isolate foregrounds from the signal . in this section , for analytic work , we assume @xmath218 .
we note without further proof that this assumption does nt alter our main inferences . for a set of point sources ,
the intensity distribution is given by : @xmath219 here @xmath220 and @xmath221 are the source fluxes and positions , respectively .
this allows us to compute the visibility for a given baseline @xmath22 and frequency @xmath18 .
@xmath222 here @xmath223 is the primary beam .
as discussed earlier , we also define a visibility in the conjugate space by taking the fourier transform with respect to @xmath18 ( eq . ( [ deftau ] ) ) : @xmath66 .
our aim is to compute the correlation of this visibility : @xmath224 \right ) a_\nu(\vec\theta_i ) a_{\nu'}(\vec\theta_j ) \label{viscorpsou}\ ] ] both the source flux @xmath225 and primary beam @xmath21 are functions of frequency so it is difficult to analytically compute this expression .
however , assuming smooth and small variation of both of these quantities across the bandwidth , we can make meaningful analytic estimates ; we verify this assumption from detailed simulations and the analysis of the data .
the main frequency variation in this case comes from the phase of the integral ( the terms in the exponent ) and in particular from the change in the baseline length as frequency changes .
we note here that multiple correlations are available to us for this analysis for different pairs of @xmath226 and @xmath227 . here
we assume @xmath169 .
_ delay space
foreground wedge _
: here we expand the same baseline in frequency space : @xmath228 . in this case ,
@xmath229 , where @xmath63 is some fixed frequency . making the simplifying assumption that both point source fluxes and the primary beam are independent of frequency , eq . ( [ viscorpsou ] ) can be analytically integrated .
we further make coordinate transformation : @xmath230 and @xmath231 and assuming @xmath232 in all the quantities except those in the exponent containing their difference , which allows us to use @xmath75 : @xmath233
\right ) a_\nu(\vec\theta_i ) a_\nu(\vec\theta_j ) \label{viscorpsou1}\ ] ] integrals over @xmath84 and @xmath234 are now separated which gives us : @xmath235\ ] ] as noted above , @xmath236 , or it is independent of frequency .
the integral in the equation is insignificant only when @xmath237 .
this linear relation between @xmath61 and the baseline @xmath59 gives a region bounded by a ` wedge ' in the @xmath61@xmath238 space for a spatial distribution of point sources ( e.g. see @xcite ) .
another possible way to understand the nature of spectrally smooth foregrounds is to first compute the correlation in the frequency space . using eq .
( [ vispoints ] ) , this gives us : @xmath239 \right ) a_\nu(\vec\theta_i ) a_{\nu'}(\vec\theta_j ) \label{viscorpsou2}\ ] ] using @xmath240 and substituting into eq .
( [ viscorpsou2 ] ) , and performing a single fourier transform with respect to @xmath72 , we recover the main expected feature of the foreground ` wedge ' described above .
computationally , if the variation of other quantities with frequency , primary beam and source fluxes , is neglected , this method is completely equivalent to the one based on eq .
( [ viscorpsou1 ] ) .
even though we used a set of point sources , the main inferences of the analysis also follow for diffuse sources .
in particular , the frequency space integrals used to prove our case are exactly the same . for our simulations , we assume a set of point sources isotropically distributed with fluxes above 1 jy at 150 mhz .
we construct this flux distribution from radio source count at 1.4 ghz , which is given by ( @xcite ) : @xmath241= \sum_{i=0}^6 a_i \left [ \log \left(\frac { s}{mjy}\right ) \right]^i\ ] ] for flux range 0.05 mjy @xmath242 s @xmath242 1000 mjy .
the constants are @xmath243 .
we simulate sources over the entire hemisphere ( nearly 15000 sources ) to suitably take into account the contribution from mwa primary beam sidelobes .
we extrapolate the distribution to the frequencies of interest to us by assuming a spectral index @xmath244 .
thermal noise is independent of the baseline and depends on three parameters : system temperature , integration time and the channel width .
the rms of thermal noise associated with a visibility measurement for channel width @xmath72 and integration time @xmath245 is : @xmath246 here @xmath247 and @xmath248 denote the system temperature and antenna gain respectively . for mwa ,
@xmath249 with @xmath101 for mwa at @xmath250 ( @xcite ) . in our analysis we choose @xmath251 khz , @xmath252 seconds are very small compared to the frequency and time coherence of the signal ( @xcite ) .
the system temperature has two components : sky temperature ( dominant source of noise at low frequency ) and receiver temperature .
we consider @xmath253 for a single polarization which is consistent with the reported system temperature at 154.24 mhz for the mwa pointing we consider in this paper .
it is fair to assume that the thermal noise for a radio interferometer follows a gaussian statistics with zero mean . in our simulation
, we follow the same pipeline used for analysing the real data to estimate the thermal noise power .
we use the baseline distribution from the observation with @xmath254 .
for every ( u , v ) point the noise is drawn from a gaussian distribution with zero mean and the rms given by eq .
( [ sigma_th ] ) .
as discussed in section [ sec : powspecest ] , the power spectrum from the data is computed in two stages . first the power spectrum and its rms is computed for a single cell in which the hi signal is expected to be near coherent and then an average is obtained across cells assuming the hi signal to be incoherent for different cells ( figure [ schem ] ) . as noted in the text ,
the hi signal can be recovered from a visibility cross - correlation by inverse weighing with @xmath197 .
we denote such a cross correlation : @xmath255 ; @xmath256 is generally a complex number . for optimal averaging to get the lowest noise estimator , one needs to sum over these cross - correlations by inverse weighing with the square of the rms of each cross - correlation @xmath257 . for pure noise , @xmath258 and one can obtain eq .
( [ powspecdef ] ) .
notice that this estimator is invariant under an overall scaling of @xmath257 .
the error on power spectrum for each cell @xmath207 is : @xmath259 , where the average is obtained optimally from the data for all the cross - correlations .
it can be shown that if each cross - correlation is assumed to be uncorrelated , as would be the case for pure noise , @xmath260 .
notice that if the rms for all the cross - correlations is the same , as would be the case if all measurements are equally weighted , then this expression reduces to @xmath261 , where @xmath262 is the number of all the cross - correlations within a cell . for averaging over cells , we repeat the procedure described above by taking the estimated power spectrum for a cell as the signal and @xmath207 as the weights .
this allows us to estimate 2- and 1-dimensional power spectrum and its rms . for pure noise , the final error on the power spectrum
is expected to approach : @xmath263 , where @xmath196 is the number of cells used for obtaining the average .
we briefly discuss some shortfalls of such a procedure .
first , we do not construct the covariance matrix of the power spectrum estimator .
we only estimate its diagonal terms , @xmath264 and @xmath265 .
this means that we are not able to assess the extent of cross - correlation between two neighbouring bins in figure [ 1d_power ] .
such cross - correlation might contain important information about systematic errors , foreground leakage , and hi signal and noise cross correlation .
second , we do not include the hi signal in our estimation procedure .
this is justified for the present work as the observed signal is clearly dominated by noise and foreground residuals ( figure [ 1d_power ] ) .
we briefly assess the impact of the hi signal for computing the error on the power spectrum .
we assume the following estimator for computing the power spectrum for a cell and consider the contribution of only the hi signal : @xmath266 as noted above , this estimator allows us to recover the hi signal .
the subscripts @xmath267 correspond to a pair for visibilities and the sum is carried over all the cross - correlations . after further computation , we obtain the error on the signal : @xmath268\ ] ] if all the weights are unity this reduces to the usual cosmic variance expression : @xmath269 . even though this term is negligible for our purposes , this would need to be included for longer integration times .
mcmullin , j. p. , waters , b. , schiebel , d. , young , w. , & golap , k. 2007 , astronomical data analysis software and systems xvi ( asp conf .
376 ) , ed .
r. a. shaw , f. hill , & d. j. bell ( san francisco , ca : asp ) , 127 | the detection of redshifted 21 cm emission from the epoch of reionization ( eor ) is a challenging task owing to strong foregrounds that dominate the signal . in this paper
, we propose a general method , based on the delay spectrum approach , to extract hi power spectra that is applicable to tracking observations using an imaging radio interferometer ( delay spectrum with imaging arrays ( dsia ) ) .
our method is based on modelling the hi signal taking into account the impact of wide field effects such as the @xmath0-term which are then used as appropriate weights in cross - correlating the measured visibilities .
our method is applicable to any radio interferometer that tracks a phase center and could be utilized for arrays such as mwa , lofar , gmrt , paper and hera . in the literature
the delay spectrum approach has been implemented for near - redundant baselines using drift scan observations . in this paper
we explore the scheme for non - redundant tracking arrays , and this is the first application of delay spectrum methodology to such data to extract the hi signal .
we analyze 3 hours of mwa tracking data on the eor1 field .
we present both 2-dimensional ( @xmath1 ) and 1-dimensional ( k ) power spectra from the analysis .
our results are in agreement with the findings of other pipelines developed to analyse the mwa eor data . |
it is well known that the galactic center ( gc ) region has a very complex astrophysical enviroment and is rich in various kinds of objects , and is a good library for the study of astrophysical phonomena .
a supermassive black hole with mass @xmath1 m@xmath2 , sgr a@xmath3 , lies in the gc .
although the gc is rather quiet nowadays , frequent flares were observed in the x - ray as well as near infrared ( nir ) bands @xcite , which means the existence of continuous weak activities of the black hole . as a consequence high energy particles
could be accelerated during such kinds of activities and imprinted in the @xmath0-ray sky .
the very high energy ( vhe ) @xmath0-rays from the gc have been observed by several atmospheric cerenkov telescopes such as cangaroo @xcite , veritas @xcite , hess , and magic @xcite .
recently gev @xmath0-ray emission from the gc was also revealed by the spatial detector fermi large area telescope ( fermi - lat , @xcite ) .
the observations in both tev and gev bands showed that the @xmath0-ray emission tended to be stable without significant variability .
serveral models have been proposed to explain the @xmath0-ray emission observed at the gc , including hadronic models @xcite and leptonic ones @xcite .
an issue prevents a simple explanation to the gev - tev spectra of the gc is that the gev @xmath0-ray emission is not consistent with the direct extrapolation of tev @xmath0-rays to the low energy range .
chernyakova et al .
@xcite proposed that the strong energy dependence of the diffusion coefficient would seperate the low energy and high energy particles into two different diffusion regimes and the gev - tev spectral shape of the @xmath0-rays could be reproduced . in @xcite
the stochastic acceleration of particles in a two - phase interstellar medium ( ism ) was employed to generate two distinct populations of protons to explain the data . in @xcite the fermi - lat @xmath0-ray emission
was suggested to originate from a population of electrons ics off the soft background photons and the vhe @xmath0-ray emission was thought to be from different sources instead of sgr a@xmath3 .
the morphology study of the vhe @xmath0-ray source , hess j1745 - 290 , by hess showed no spatial extension and the upper limit of the angular size of @xmath4 was set @xcite .
note there is another diffuse component of the @xmath0-ray emission along the ridge @xcite , which is not discussed in this work .
it may imply that the high energy particles are confined in several parsec regions around the gc @xcite or the high concentration of the target gas in the inner region @xcite . in this work ,
we propose a hybrid model to explain the gev and tev @xmath0-ray emission .
it is natural to expect that both protons and electrons can be accelerated during the gc activity .
therefore without finely tuning the environmental parameters , we should expect the existence of two components of the @xmath0-ray emission , which is just the case shown by the gev - tev observations .
a simple expectation is that high energy protons interact with the ism is responsible to the vhe @xmath0-rays , and the electrons , which may cool down in the gc radiation field and/or magnetic field , produce the gev @xmath0-rays through inverse compton scattering ( ics ) off the background photons . in sec .
2 , we present the detailed modeling of this picture .
3 is the discussion and the main conclusions are summarized in sec .
although the overall behavior of sgr a@xmath3 is quite silent , the observations in x - ray and infrared bands indicate that it still has continuous weak activities @xcite . during such kinds of avtivities ,
the accretion of stars and gas by the supermassive black hole could be effictive to accelerate particles .
we assume that both the protons and electrons were accelerated instantaneously during such flare events .
then these particles would diffuse away from the acceleration site , and radiate during the propagation . the propagation equation for both protons and electrons
can be written as @xmath5+n(e)\delta(t)\delta({\bf r}),\ ] ] where @xmath6 is the propagated flux as a function of space , energy and time , @xmath7 is the diffusion coefficient , @xmath8 is the injection spectrum of particles at @xmath9 and @xmath10 , and @xmath11 is the energy loss rate , which is important for electrons but negligible for protons .
the energy dependence of diffusion coefficient is assumed to be power - law of rigidity @xmath12 , @xmath13 , in which @xmath14 is the spectral index and @xmath15 is the particle velocity . in this work
we assume @xmath16 , which is close to a kolmogorov type of the ism turbulence .
the solution of the above equation for electrons is @xcite @xmath17 where @xmath18 is the initial energy of the electron whose energy is cooled down to @xmath19 at time @xmath20 , and @xmath21 , in which @xmath22 is the effective diffusion length when the particle energy decreases from @xmath18 to @xmath19 .
for protons we neglect the energy loss term , and the solution of the propagation equation is even simpler @xmath23 where @xmath24 is the effective diffusion length within @xmath20 .
the energy loss of electrons includes ics off the soft background photons , synchrotron radiation in the magnetic field and bremsstrahlung radiation in the ism .
the soft photons consist with infrared from dust and optical from stars . in @xcite
the measurement of infrared emission in the central @xmath25 pc was presented . according to the data ,
the expected optical emission was proposed based on assumptions of the optical photon energy density .
taking the uncertainties of the expected optical emission into account , we adopt the two models given in in this work .
the intensity of the soft photons is shown in figure [ bkg_photon_filed ] .
the magnetic field strength varies with the radius away from the gc @xcite . for the very small region ( @xmath26 pc )
the average magnetic field strength is adopted to be @xmath27 g @xcite . as for the ism density
, there is large uncertainty in the gc region .
typically the adopted ism density within @xmath28pc region is of the order @xmath29
@xmath30 @xcite .
compared with the ics energy loss in the strong radiation field , the bremsstrahlung energy loss is negligible @xcite .
we calculate the cooling time of the electrons , @xmath31 , in the above background photons field and magnetic field , as shown in figure [ energy_loss ] . here
the optical emission is adopted to be the one of kusunose & takahara ( 2012 ) @xcite .
it is shown that the energy loss of electrons is dominated by the ics up to @xmath32 tev .
g. the ism density is adopted to be @xmath29 @xmath30 to calculate the bremsstrahlung energy loss.,scaledwidth=80.0% ] and @xmath33 ( from top to bottom ) years respectively
. the energy loss of electrons adopted in this calculation is shown in figure [ energy_loss].,scaledwidth=80.0% ] in figure [ spctgreenall ] we show the integral spectra of both electrons and protons after time @xmath20 of the injection .
the integral radius is taken to be @xmath25 pc for electrons and @xmath34 pc for protons respectively .
for electrons we integrate the particles within radius @xmath35 pc since the high background photon density is expected to exist in that region , and most of the radiation should come from such a region . however , for protons , we adopt a some larger integral radius , @xmath36 pc , according to the spatial extention upper limit of hess j1745 - 290 derived by hess @xcite . the diffusion coefficient is @xmath37 @xmath38 s@xmath39 , and the injection spectra of both electrons and protons are adopted as @xmath40 .
we can see from this figure that the cooling of electrons will result in a cutoff of the spectrum , above which the electrons can not survive from the energy loss .
the cutoff energy depends on the time .
the younger the injection is , the higher energy of electrons we can observe .
the spectral shape also has an evolution with the time . for shorter time , the relative fraction of low energy particles is smaller due to slower propagation of low energy particles , which results in harder spectra . for protons
there is also a time evolution of the spectra .
it reflects the fact that for longer time , more high energy particles will diffuse out of the integral region and result in a soft spectrum
. then we discuss the @xmath0-ray emission of those electrons and protons in the background radiation field and the interstellar medium . for electrons
we consider only the ics emission and neglect the bremsstrahlung radiation .
the klein - nishina cross section of ics is employed . for the @xmath0-ray emission from @xmath41 collisions , we use the parameterization given in @xcite .
-ray spectra of the hybrid model , compared with the observational data by fermi - lat @xcite and hess .
different lines show the effect of observational time @xmath20 ( 100 , 200 and 300 yr from top to bottom ) .
the diffusion coefficient is @xmath37 @xmath38 s@xmath39 .
the soft background photon model is kusunose & takahara ( 2012 ) @xcite .
, scaledwidth=80.0% ] figure [ spctkusunose ] presents the results of the calculated @xmath0-ray spectra and the comparison with the observational data .
the injection spectra are adopted to be @xmath42 , and the diffusion coefficient is adopted to be @xmath37 @xmath38 s@xmath39 . the background photon model is adopted to be kusunose & takahara ( 2012 ) @xcite .
three different observational time @xmath43 , @xmath44 and @xmath33 yr are shown .
the observational time determines the cutoff energy of the electron spectrum ( see figure [ spctgreenall ] ) .
we find that for the kusunose & takahara ( 2012 ) photon field @xmath45 yr can match the data well .
another effect of the time @xmath20 , is the overall normalization .
if @xmath20 is larger , more particles diffuse out of the @xmath28pc region , and we expect a decrease of the overall normalization of the integral particle spectra . for @xmath46
yr we find the total energy of electrons above @xmath47 gev is @xmath48 erg to match the fermi - lat data .
the electron - to - proton ratio @xmath49 is about @xmath50 in this case .
note , however , a significant part of high energy protons ( @xmath51 tev ) will diffuse out of the @xmath34 pc region in this case , as can be seen from figure [ spctgreenall ] .
therefore to be consistent with the morphology of the source , we may require the gas density is higher in inner @xmath34 pc region than outside , as the scenario of @xcite .
but for smaller diffusion coefficient @xmath52 @xmath38 s@xmath39.,scaledwidth=80.0% ] to be consistent with the cutoff behavior of vhe @xmath0-rays observed by hess , we employ an exponential cutoff of the proton spectrum with cutoff energy @xmath53 tev .
it is possible that the pair production of vhe photons in the background radiation field could also be responsible to the cutoff of the @xmath0-ray spectra .
however , our calculation shows that for the adopted soft background photon field in @xmath28pc region , the pair production attenuation is negligible .
thus the cutoff should be understood as the acceleration limit of the flaring event .
the maximum energy protons can achieve for diffusive shock acceleration is @xcite @xmath54 where @xmath55 is the magnetic field and @xmath12 is the size of the acceleration region . as in @xcite , we assume the acceleration takes place within @xmath56 schwarzschild radii ( @xmath57 cm ) of the black hole . to accelerate protons to above @xmath58 tev
requires a magnetic field @xmath28 g in the acceleration region .
such a condition could be reached in the very central region of the gc @xcite . on the other hand ,
if the acceleration takes place in larger regions , the required magnetic field could be smaller .
figure [ spctkusunosed0 ] shows the results for even smaller diffusion coefficient @xmath52 @xmath38 s@xmath39 . in this case
most of the protons up to @xmath58 tev may still be confined in several pc region around the gc . as a simple estimate
, the diffusion length for protons is about @xmath59 pc .
the accumulative spectrum of protons is similar with the injection spectrum .
therefore we adopt the injection spectrum @xmath60 to be consistent with the hess data .
the results show a good fit to the gev - tev observational data for @xmath46 yr too . it is shown that for protons different time @xmath20 leads to small differences of the results , which means that indeed most of the protons are confined .
we also note that the differences of the ics spectra with respect to @xmath20 are larger than that of @xmath61 decay spectra .
this is simply because the integral radius of electrons is smaller than that of protons .
the total energy of electrons above @xmath47 gev is @xmath62 erg in this case , and the electron - to - proton ratio @xmath49 is @xmath63
. finally , the results for mezger et al .
( 1996 ) soft background photon model are shown in figure [ spctmezger ] .
other conditions are the same as figure [ spctkusunose ] .
compared with kusunose & takahara ( 2012 ) photon field , the optical emission is stronger and the cooling time for electrons is shorter .
thus for @xmath64 yr the cutoff energy of electrons is proper to fit the fermi - lat data .
however , the energy spectrum of ics emission does not differ significantly from that shown in figure [ spctkusunose ] .
this is because in stronger background radiation field the cooling is more significant , and a larger cooling is just canceled out when calculating the ics @xmath0-ray spectrum in such a radiation field .
the total energy of electrons above @xmath47 gev is @xmath65 erg , and the electron - to - proton ratio @xmath49 is about @xmath66 .
but for the mezger et al .
( 1996 ) soft background photon model .
, scaledwidth=80.0% ]
the diffusion coefficient adopted in this work is @xmath67 @xmath38 s@xmath39 , which is much smaller than that in the disk and halo ( @xmath68 as induced from the cosmic ray transportation ) .
this might be due to the higher magnetic field and more turbulent ism in the gc region .
the lower limit of the diffusion coefficient should be the bohm limit , which is @xmath69 @xmath38 s@xmath39 @xcite .
the diffusion coefficient adopted in this work is well larger than the bohm limit in the whole energy range discussed . in the calculation of proton propagation
, the energy loss of protons via interaction with the ism is neglected . as a rough estimate ,
the average collision probability of one projectile proton is about @xmath70 .
thus for the typical parameters adopted in this work , the collisional energy loss of protons is indeed negilible .
the secondary electrons / positrons from charged pion decay due to @xmath41 collision could also contribute to the @xmath0-ray emission . as a very rough estimate , for one @xmath41 collision
, @xmath0-ray photons will take @xmath71 of the energy of all the pions , and @xmath72 will take @xmath73 of the energy ( assuming energy equipartition between @xmath72 and neutrinos ) .
therefore the total energy of secondary @xmath72 should be lower than that of @xmath61 decaying @xmath0-rays .
furthermore , the soft radiation field is restricted in 1.2 pc region around the black hole , while the @xmath41 collision occurs in @xmath36 pc region .
therefore the contribution of the secondary @xmath72 to the @xmath0-rays through ics emission should be much smaller than that of neutral pion decay .
the electron component will also produce synchrotron radiation in the magnetic field , and there might be constraints from the radio to x - ray data .
as shown in @xcite the synchrotron emission in @xmath74 @xmath75 g magnetic field may exceed the radio measurement of sgr a@xmath3 for the quiescent state .
however , the spatial scale of the present study is arc - minutes ( pc ) , instead of arc - seconds as the radio observations show .
the expected synchrotron emission is consistent with the data - based result of pc scale radio emission as given in .
early in 2012 , one bright flare of gc is observed by hetgs onboard of the chandra x - ray observatory @xcite .
the total energy in @xmath76 kev band was approximately @xmath77 erg @xcite . assuming the total energy of this accretion event is about @xmath78 orders of magnitude higher , @xmath79 erg , and @xmath80 of it converts to acceleration of cosmic rays , we give the expected @xmath0-ray emission of such an event within the current framework of the model , as shown in figure [ 2012flare ] .
the solid lines are same as that in figure [ spctkusunose ] .
the dashed line is the expected flux of the ics emission , with the same parameters of the diffusion coefficients , injection spectral index and electron - to - proton ratio as that of figure [ spctkusunose ] .
the @xmath0-ray emission at this stage is dominated by the ics emission and the hadronic component is much lower .
we can see that the total flux of @xmath0-rays is too low to be able to be detected by the current vhe @xmath0-ray detectors .
we should keep in mind that this estimate is very rough and suffers from large uncertainties of the assumption of total energy output of the flare and the energy fraction goes into particle acceleration .
this gives us an impression that how large a flare is needed in order to produce the observed @xmath0-rays .
-ray emission for the flare event of 2012 .
we assume the total power of the flare event is @xmath78 orders of magnitide high than the @xmath76 kev energy.,scaledwidth=80.0% ]
in this work , a hybrid model is proposed to explain the gev @xmath0-ray emission observed by fermi - lat and tev @xmath0-ray emission observed by hess telescope , of the gc .
the current @xmath0-ray emission could be originated from one flare with total energy @xmath81 ergs at @xmath82 years ago .
both the protons and electrons were accelerated to very high energies in this flaring event .
furthermore the flaring event produced a thermal bath of soft photons in the gc region , with typical scale @xmath28pc .
high energy particles could then diffuse out of the acceleration site and interact with the ambient medium and radiation fiels .
the hadronic collisions between protons and ism gives rise to the tev @xmath0-rays . at the same time
, electrons would lose energy through ics and synchrotron radiation .
the ics photons could be responsible for the gev @xmath0-ray emission observed by fermi - lat .
this work is supported by the ministry of science and technology of china , natural sciences foundation of china ( nos .
10725524 , 10773011 , 11135010 and 11105155 ) , the chinese academy of sciences ( nos . kjcx2-yw - n13 , kjcx3-syw - n2 , gjhz1004 ) and natural science foundation of shandong province of china ( zr2009am003 ) .
qy acknowledges the support from the key laboratory of dark matter and space astronomy of chinese academy of sciences .
10 dodds - eden , k. , et al . , 2009 ,
apj , 698 , 676 dodds - eden , k. , et al . , 2011 , apj , 728 , 37 nowak , m. a. , et al .
, 2012 , apj , 759 , 95 tsuchiya , k. , et al .
, 2004 , apj , 606 , 115 kosack , k. , et al . , 2004 ,
apj , 608 , 97 aharonian , f. , et al . , 2004 , a@xmath83 a , 425 , 13 aharonian , f. , et al . , 2006 ,
, 97 , 9901 aharonian , f. , et al . , 2008 ,
a@xmath83a , 492 , 25 albert , j. , 2006 , apj , 638 , 101 chernyakova , m. , et al . , 2011 , apj , 726 , 60 atoyan , a. @xmath83 dermer , c. d. 2004 , apj , 617 , l123 liu , s. , et al . , 2006 , apj , 647 , 1099 ballantyne , d. r. , et al . , 2007 ,
657 , l13 linden , t. , et al . , 2012 ,
apj , 753 , 41 fatuzzo , m. & melia , f. , 2012 , apj , 757 , l16 hinton , j. a. @xmath83 aharonian , f. a. , 2007 , apj , 657 , 302 kusunose , k. @xmath83 takahara , f. , 2012 , apj , 748 , 34 acero , f. , et al . , 2010 , mnras , 402 , 1877a aharonian , f. , et al . , 2006 , nature , 439 , 695 yusef - zadeh , f. , et al .
, 2012 , arxiv:1206.6882 becklin , e. e. , gatley , i. @xmath83 werner , m. w. , 1982 , apj , 258 , 135 davidson , j. a. , 1992 , apj , 387 , 189 atoyan , a. m. , aharonian , f. a. @xmath83 volk , h. j. , 1995 , phys . rev .
d , 52 , 3265 mezger , p. g. , duschl , w. j. @xmath83 zylka , r. , 1996 , a@xmath83ar , 7 , 289 bertone , g. et al .
, 2009 , jcap , 03 , 009 strong , a. w. & moskalenko , i. v. , 1998 , apj , 509 , 212 kamae , t. , et al . , 2006 , apj , 647 , 692 aharonian , f. , et al . , 2009 , a@xmath83a , 503 , 817 zhang , j .- l . , bi , x .- j . &
hu , h .- b . , 2006 , a@xmath83a , 449 , 641 moskalenko , i. v. , porter , t. a. & strong , a. w. , 2006 , apj , 640 , 155 aharonian , f. & neronov , a. , apj , 2005 , 619 , 306 ginzburg , v. l. , & ptuskin , v. s. , 1981 , icrc , 2 , 336 drury , l , oc . , 1983 , rep . , prog .
, phys . , 46 , 973 heavens , a. , f. , 1984 , mnras , 207,1 melia , f. & falcke , h. , 2001 , ara&a , 39 , 309 | the observations of high energy @xmath0-ray emission from the galactic center ( gc ) by hess , and recently by fermi , suggest the cosmic ray acceleration in the gc and possibly around the supermassive black hole . in this work
we propose a lepton - hadron hybrid model to explain simultaneously the gev - tev @xmath0-ray emission .
both electrons and hadronic cosmic rays were accelerated during the past activity of the gc .
then these particles would diffuse outwards and interact with the interstellar gas and background radiation field .
the collisions between hadronic cosmic rays with gas is responsible to the tev @xmath0-ray emission detected by hess . with fast cooling in the strong radiation field ,
the electrons would cool down and radiate gev photons through inverse compton scattering off the soft background photons .
this scenario provides a natural explanation of the observed gev - tev spectral shape of @xmath0-rays . |
quantum spin chains have been the subject of intensive study during the past seven decades @xcite .
experimentally , their properties are investigated via inelastic neutron scattering on ( anti)ferromagnetic quasi one - dimensional compounds @xcite .
one important quantity that holds much of the information related to such compounds is the dynamical correlation function ( dcf ) @xmath5 of two local spin operators .
indeed , the neutron scattering intensity is directly proportional to it , see for example @xcite .
an important feature of these chains is that some of them , like the heisenberg model , are amenable to exact theoretical treatment while still describing nontrivial interactions @xcite , see also @xcite .
this is because they incorporate in them a rich mathematical structure : the quantum affine algebra . the early work on such models consisted in determining exactly their static thermodynamic properties , see @xcite , whereas , more recently , a fuller exploitation of the quantum symmetry using bosonization techniques has allowed for a more systematic description of their dynamical properties @xcite .
a systematic account of this work is given in @xcite . however , one aspect of these new techniques is that the exact correlation functions are usually obtained in the form of quite complicated contour integrals and this renders their manipulation somewhat cumbersome .
but on the other hand , before this exact treatment , the approach to these dynamical quantities was only approximate . indeed , if one considers for instance the dcf in the antiferromagnetic heisenberg model , the focus has for a long time been only on what we now know to be its two - spinon contribution @xmath4 .
first there has been the anderson ( semi - classical ) spin - wave theory @xcite , an approach based on an expansion in powers of @xmath6 , where @xmath7 is the spin of the system and hence , is exact only in the classical limit @xmath8 .
it can describe with some satisfaction compounds with higher spins @xcite , but fails in the quantum limit @xmath9 .
then there has been the so - called mller ansatz @xcite , which gives an approximate expression for @xmath4 that can account for many aspects of the phenomenology for @xmath9 compounds .
only very recently could we get for this system a final exact expression for @xmath4 @xcite , and that gave a better account of the data @xcite .
now for the spin-@xmath10 antiferromagnetic heisenberg model , there is a need to go beyond the two - spinon contribution .
the need is two - fold .
first , it is important to see if one is able to get useful information from these complicated and compact expressions we alluded to above .
second and more important perhaps is the fact that , though the exact two - spinon contribution accounts for much of the phenomenology as we said , about 70% , in a sense that will become clear later in section 3 , it still does nt account for _ all _ of it .
this point is demonstrated in particular in @xcite .
the natural step forward is to look into the exact four - spinon contribution . to the best of our knowledge ,
refs @xcite and @xcite constitute the first direct attempt in this direction . a general expression for the @xmath11-spinon contribution to the dcf in the anisotropic heisenberg model is given in @xcite and ( a still compact one ) for the isotropic limit in @xcite . in ref @xcite
, we specialize to the four - spinon case and give a discussion of some of its properties in the isotropic and ising limits . in particular
, we show that in the isotropic limit , the one of interest here , the exact four - spinon contribution is safe of any potential divergences . in this work , we further the description of the four - spinon contribution @xmath12 .
we determine the region in which @xmath0 is not identically zero and show that it is larger than that of @xmath4 . also , we show that the behavior of @xmath12 as a function of the neutron momentum transfer @xmath2 is similar in its overall shape to that of the corresponding @xmath4 .
this result is to be contrasted with the fact that the expression of @xmath0 is a lot more involved than that of @xmath4 , see ( [ s4 ] ) and ( [ s2ex ] ) respectively .
this paper is organized as follows . in the next section ,
we briefly discuss the antiferromagnetic spin 1/2 heisenberg model .
we describe the spinon hilbert - space structure and define the dynamical correlation function . in section 3
, we succinctly review the properties of the two - spinon contribution .
we give the exact expression of @xmath4 and that of the mller ansatz , and briefly compare their main features . in section 4 ,
we give the exact expression of @xmath0 and discuss in detail its properties we mentioned above .
the last section comprises concluding remarks .
the antiferromagnetic @xmath9 @xmath1-heisenberg model is defined as the isotropic limit of the @xmath13-anisotropic heisenberg hamiltonian : @xmath14 where @xmath15 is the anisotropy parameter .
the isotropic antiferromagnetic limit is obtained via the limit @xmath16 or equivalently @xmath17 . here
@xmath18 are the usual pauli matrices acting at the site @xmath19 the exact diagonalization directly in the thermodynamic limit of the hamiltonian in ( [ ham ] ) is possible using the @xmath20 quantum group symmetry present in the model @xcite .
the resulting hilbert space @xmath21 consists of @xmath22-spinon energy eigenstates @xmath23 built on the two vacuum states @xmath24 , @xmath25 such that : @xmath26 where @xmath27 is the energy of spinon @xmath28 and @xmath29 is a spectral parameter living on the unit circle . in the above equation , @xmath30 and the index
@xmath31 refers to the boundary condition on the spin chain , see @xcite .
the translation operator @xmath32 which shifts the spin chain by one site acts on the spinon eigenstates in the following manner : @xmath33 where @xmath34 and @xmath35 is the lattice momentum of spinon @xmath28 .
the expressions of the spinon energy and lattice momentum in terms of the spectral parameter are quite cumbersome in the anisotropic case , see @xcite , but simplify considerably in the isotropic limit , see eq ( [ disrel ] ) below .
the completeness relation in @xmath21 reads : @xmath36 the two - point dcf is the fourier transform of the zero - temperature vacuum - to - vacuum two - point function , i.e. , it is defined by : @xmath37 where @xmath38 and @xmath2 are the neutron energy and momentum transfer respectively and @xmath39 denotes @xmath40 .
the dcf is such that : @xmath41 the two relations that express the reflection and periodicity symmetries on the linear chain . inserting the completeness relation and using the heisenberg relation : @xmath42
we can write the transverse dcf as the sum of @xmath11-spinon contributions : @xmath43 where the @xmath11-spinon dcf @xmath44 is given by : @xmath45 relation in which @xmath46 denotes the form factor : @xmath47 note that each @xmath44 must satisfy relations ( [ symdcf ] ) .
the exact expression of this form factor has been determined in @xcite . using this form factor
, we can give an exact expression for the @xmath11-spinon dcf in the anisotropic case , see @xcite , and determine exactly its isotropic limit , see @xcite .
this limit is obtained via the replacement @xcite : @xmath48 where @xmath49 is the spectral parameter suited for this limit .
the expressions of the energy @xmath50 and momentum @xmath51 in terms of @xmath49 then read : @xmath52 the transverse two - spinon dcf @xmath4 does not involve a contour integration and has been given in @xcite .
the four - spinon one @xmath0 involves only one contour integration and its expression is given in @xcite . in the next section , we review the properties of @xmath4 and in the one that follows it , we discuss those of @xmath0 .
the exact expression of the two - spinon dcf of the spin @xmath10 @xmath1-model is given in @xcite and reads : @xmath53 where @xmath54 is the heaviside step function and the function @xmath55 is given by : @xmath56 @xmath57 is the upper ( lower ) bound of the two - spinon excitation energies called the des cloizeaux and pearson @xcite upper ( lower ) bound or limit .
they read : @xmath58 the quantity @xmath59 and is related to @xmath38 and @xmath2 by the relation : @xmath60 a relation obtained using eq ( [ disrel ] ) and the energy - momentum conservation laws : @xmath61 note that the explicit expression of @xmath4 given in eq ( [ s2ex ] ) satisfies the reflection and periodicity symmetries expressed in ( [ symdcf ] ) . the properties of @xmath4 have been discussed in @xcite .
there , a comparison with the mller ansatz @xcite is given .
this latter was derived from the properties of some solutions to the bethe - ansatz equations , from numerical calculations on finite spin chains and from an analysis of phenomenological results .
it reads : @xmath62 where @xmath63 is a constant determined in such a way to fit best the phenomenology @xcite .
there are two main differences between the exact expression ( [ s2ex ] ) and the approximate one ( [ s2mul ] ) , @xcite . first , the two - spinon threshold at @xmath64 is more singular in ( [ s2ex ] ) than in ( [ s2mul ] ) .
second , at the upper two - spinon boundary @xmath65 , @xmath4 vanishes smoothly whereas @xmath66 has a sharp cut - off .
but if one defines the frequency moments of the dcf : @xmath67 one shows that as @xmath68 the moment of @xmath4 vanishes as : @xmath69 and the same holds for the mller ansatz @xmath66 .
actually , the frequency moments ( [ fremom ] ) are particular cases of a set of sum rules the dcf is known to satisfy exactly . for exemple , we know that the first moment is exactly equal to @xcite : @xmath70 it turns out that the same frequency moment for @xmath4 is such that @xcite : @xmath71 this means that , according to this sum rule , @xmath4 is way off the total dcf @xmath5 by roughly 30% .
in fact , other exact sum rules confirm this trend , see @xcite .
of course , those remaining 30% are `` filled '' , so to speak , by the @xmath72-spinon contributions .
the natural question that comes to mind is : how much @xmath0 takes up from these 30% ?
obviously , one has first to study the behavior of this contribution before trying to answer this question , and this we do in the remainder of this paper .
the analytic expression of the four - spinon dcf has been worked out in @xcite and for @xmath73 it reads : @xmath74 where @xmath75 is a numerical constant and the integrand @xmath76 is given by : @xmath77 the different quantities involved in this expression are : @xmath78 the function @xmath79 is given by : @xmath80 \ ; , \label{f}\ ] ] where @xmath81 and the function @xmath82 reads : @xmath83 where @xmath54 is the heaviside step function . in ( [ fs4 ] ) , the sum @xmath84 is over the two pairs @xmath85 and @xmath86 solutions of the energy - momentum conservation laws : @xmath87 they read : @xmath88 note that the solution in ( [ scl ] ) is allowed as long as @xmath89 where @xmath90 is given in ( [ not ] ) and : @xmath91 we henceforth put ourselves in the interval @xmath92 . to get the behavior of @xmath0 for the values of @xmath2 outside this interval , one uses the symmetry relations ( [ symdcf ] ) given in section 2 . in the work @xcite
, we have discussed the behavior of the function @xmath76 given in ( [ fs4 ] ) .
we have shown that the series @xmath82 is convergent .
we have also shown that in the region where two @xmath93 s or more get equal , the function @xmath82 is finite .
the function @xmath79 going to zero in these same regions @xcite , this means the integrand @xmath76 of @xmath0 has a nice regular behavior there .
furthermore , we have shown that @xmath76 is exponentially convergent when one of the @xmath93 s goes to infinity , which means the two integrals over @xmath94 and @xmath95 in ( [ s4 ] ) do not yield infinities .
all these analytic results pave the way to `` safe '' numerical manipulations .
the first thing we wish to discuss in this work is the extent of the region in the @xmath96-plane in which @xmath0 is not identically zero , and compare it to that of @xmath4 .
remember that from ( [ s2ex ] ) , @xmath4 is zero identically outside the spin - wave continuum @xmath97 , where @xmath98 are given in ( [ dcp ] ) . from the condition @xmath99 discussed after eq ( [ scl ] ) ,
one infers that in order for @xmath0 to be nonzero identically , one has to have @xmath100 , where : @xmath101 all these branches are plotted in fig . 1 .
the first thing we immediately see is that the @xmath0-region , i.e. , the region in which @xmath0 is not identically zero , is not confined to the spin - wave continuum delimited by the dcp branches @xmath102 given in ( [ dcp ] ) .
this means that , a fortiori , the full @xmath5 is also not confined to the @xmath4-region .
this fact is confirmed by early finite chain numerical calculations @xcite and the phenomenology @xcite .
furthermore , fig .
1 shows that for @xmath103 , the @xmath0-region is entirely beyond the @xmath4-region .
this means that for this interval , we may expect @xmath4 to be dominant in @xmath5 within the spin - wave continuum .
however , for @xmath104 , there is overlap between the two regions such that the @xmath4-region is more or less within the @xmath0-region .
we may therefore expect here the contribution of @xmath0 to play a rle , and hence , we expect @xmath4 to be a little less dominant within the spin - wave continuum .
the next feature we discuss in this work is the behavior of @xmath0 as a function of @xmath2 for fixed @xmath38 .
2a , 3a and 4a show the behavior of @xmath0 for @xmath105 and @xmath106 respectively .
note that we have scaled @xmath0 to appropriate units .
2b , 3b and 4b show the behavior of @xmath4 as a function of @xmath2 for the same values of @xmath38 .
let us for example discuss the case @xmath107 .
we see from fig . 3b
that the function @xmath4 vanishes for ( roughly ) @xmath108 .
looking back into fig . 1 , this corresponds indeed to the region outside the spin - wave continum for @xmath109 , i.e. , @xmath110 .
the function @xmath4 starts at @xmath111 with a large value and goes to a minimum at @xmath112 1 .
3a shows that the function @xmath0 has a somewhat similar behavior . from the figure
, we read that @xmath0 too is not vanishing for ( roughly ) @xmath113 . from fig . 1 , that corresponds within the @xmath0-region to @xmath114 . but fig .
1 shows also that for @xmath115 , @xmath0 may be non - vanishing in the interval @xmath116 .
that contribution does nt appear on fig .
3a , presumably because @xmath0 there is negligeable .
this is confirmed in the case @xmath117 which is close to the case @xmath118 : there we see @xmath0 having a very small contribution in the corresponding interval . from fig .
3a , we see that @xmath0 , like @xmath4 , starts from a large value at its lower boundary @xmath119 and decreases while moving to @xmath120 .
the case @xmath121 is practically the same .
4b shows @xmath4 starting from zero at roughly @xmath122 , getting to a maximum and sharply dropping to zero a little further .
it then starts sharply again from a large value a little after @xmath123 and decreases to a minimum at @xmath120 .
this is also consistent with fig . 1 :
the function @xmath4 starts to be nonvanishing for @xmath121 at @xmath124 .
it stays nonvanishing until @xmath125 reaches the value @xmath126 it remains identically zero until @xmath127 reaches the value @xmath128 at which we enter back into the @xmath4-region .
as we said , @xmath0 has in this case too the same overall behavior as that of @xmath4 . in fig . 4a
, we see that @xmath0 starts to increase from the value zero at @xmath129 .
it goes quickly to a maximum before dropping sharply to zero a little before @xmath130 .
it stays at zero till a little after @xmath123 and increases sharply
. then it decreases while wiggling to @xmath120 .
this overall behavior is also consistent with fig . 1 .
indeed , for @xmath121 , the @xmath0-region starts at @xmath129 and extends first to @xmath131
. then we get outside this region from this value of @xmath2 till @xmath132 .
then @xmath0 is no more identically zero beyond this point until we reach the point @xmath133 .
as we said , this is quite consistent with fig .
in this work , we have discussed the behavior of the exact four - spinon contribution @xmath0 to the dynamical correlation function @xmath5 of the @xmath134 antiferromagnetic heisenberg model and compared it to the one of the exact two - spinon contribution @xmath4 .
we first reviewed the model and the spinon structure of the corresponding hilbert space .
we then gave a brief account of the results concerning @xmath4 and its comparison to the mller ansatz .
the first thing concerning @xmath0 we discussed is the region in the @xmath96-plane in which @xmath0 is not identically zero .
we found it to be different and larger than that of @xmath4 , i.e. , the spin - wave continuum .
both regions are drawn in fig . 1 .
we then discussed the behavior of @xmath0 as a function of @xmath2 for fixed values of @xmath38 , and compared it to the one of @xmath4 .
these behaviors are plotted in figs .
2,3 and 4 .
we have found that the overall shape of @xmath0 is more or less the same as that of @xmath4 .
the first thing to emphasize we think is the overall similarity between the shapes of @xmath0 and @xmath4 .
this is not at all expected from the outset , given the more complicated expression of @xmath0 , see ( [ s4 ] ) , as compared to that of @xmath4 , see ( [ s2ex ] ) .
does this mean that the shape of the other @xmath72-spinon contributions and hence of the total @xmath5 is already more or less `` traced '' by that of @xmath4 ?
we think that at this stage , it is too early for such an inference : this is a preliminary investigation into the behavior of @xmath0 and clearly more work is needed . in any case
, it would have been interesting to measure @xmath0 for other ( larger ) values of @xmath3 , especially in regions where @xmath4 is identically zero whereas @xmath0 is not , see fig . 1 .
but as @xmath3 increases , the structure of the function @xmath135 of eq([fs4 ] ) in the @xmath136-plane gets `` richer '' , which means numerically harder to handle .
to illustrate this point , we have plotted for the reader in figs .
5 the function @xmath76 for @xmath137 and @xmath138 ( 5a ) , @xmath139 ( 5b ) .
one can see that as @xmath3 increases , the function @xmath76 gets distributed nontrivially in larger areas in the @xmath136-plane , with an increasing more involved structure .
this is the main reason why we preferred to defer the discussion of these regions to future work .
also , for the same reason , we have deferred the systematic discussion of @xmath0 as a function of @xmath38 for fixed values @xmath2 .
one other interesting question we havent touched on in this work but merely alluded to at the end of section 3 is the following : how much @xmath0 accounts for in the total @xmath5 ?
in other words , is @xmath140 better an approximation to the total @xmath5 than @xmath4 alone , and if yes , by how much ?
as we said , we know that @xmath4 accounts for about 70% of the total @xmath5 , which means that roughly 30% are left for the @xmath72-spinon contributions . to tackle this question as it should , one has to rely on a certain number of sum rules @xmath5 is known to satisfy exactly .
then one compares the contribution of @xmath140 to the exact result and carries a discussion thereon .
the other interesting question one may ask is the physical interpretation and implications of the behavior of the four - spinon dcf we have described in this work . what would certainly be interesting is to be able to systematically measure @xmath5 outside the spin - wave continuum so that we are assured of having the effects of the two - spinon contribution eliminated .
some ( not _ all _ ) of the literature related to the heisenberg model , our interest in this work , comprises w. heisenberg , z. phys .
49 ( 1928 ) 619 ; h. bethe , z. phys . 71
( 1931 ) 205 ; l. hulthn , arkiv .
fysik a 1126 ( 1938 ) 1 ; e.h .
lieb and d.c .
mattis , j. math .
phys . 3 ( 1962 ) 749 ; j. des cloizeaux and j.j .
pearson , phys .
128 ( 1962 ) 2131 ; r.b .
griffiths , phys .
133 ( 1964 ) a 768 ; c.n .
yang and c.p.yang , phys .
150 ( 1966 ) 321 ; 150 ( 1966 ) 327 ; 151 ( 1966 ) 258 ; th .
niemeijer , physica 36 ( 1967 ) 377 ; e. barouch , b.m .
mccoy and d.b .
abraham , phys . rev .
* a*4 ( 1971 ) 2331 ; m. gaudin , phys . rev .
( 1971 ) 1301 ; m. takahashi , prog .
46 ( 1971 ) 401 ; l.a .
takhtajan and l.d .
faddeev , russ .
surveys 34 ( 1979 ) 11 ; b.m .
mccoy , j.h.h .
perk and r.e .
shrock , nucl . phys . * b*220 ( 1983 ) 35 ; o. babelon , h.j . de vega and c.m .
viallet , nucl . phys . * b*220 ( 1983 ) 13 ; g. mller and r.e .
shrock , phys . rev .
* b*29 ( 1984 ) 288 ; j.m.r .
roldan , b.m .
mccoy and j.h.h .
perk , physica 136a ( 1986 ) 255 ; v.e .
korepin , a.g .
izergin and n.m .
bogoliubov , ` _ the quantum inverse scattering method and correlation functions _ ' , cambridge university press , 1993 ; f.h.l .
essler , h. frahm , a.g .
izergin and v.e .
korepin , comm . math . phys . 174 ( 1994 ) 191 ; v.e .
korepin , a.g .
izergin , f.h.l .
essler and d. uglov , phys . lett . * a*190 ( 1994 ) 182 . d.a .
tennant , r.a .
cowley , s.e . nagler and a.m. tsvelik ,
* b*52 ( 1995 ) 13368 ; d.a .
tennant , s.e .
nagler , d. weltz , g. shirane and k. yamada , phys .
* b*52 ( 1995 ) 13381 ; d.a .
tennant , t.g .
perring , r.a .
cowley and s.e .
nagler , phys .
70 ( 1993 ) 4003 ; s.e .
nagler , d.a .
tennant , r.a .
cowley , t.g . perring and s.k .
satija , phys . rev . * b*44 ( 1991 ) 12361 .
o. davies , o. foda , m. jimbo , t. miwa and a. nakayashiki , comm .
( 1993 ) 89 ; for the techniques of bosonization on quantum affine algebras , see for example i.b .
frenkel and n.h .
jing , proc .
85 ( 1988 ) 9373 ; a. abada , a.h .
bougourzi and m.a .
el gradechi , mod .
a8 ( 1993 ) 715 ; a.h .
bougourzi , nucl . phys . * b*404 ( 1993 ) 457 , and more recently a.h .
bougourzi , _ bosonization of quantum affine groups and its application to the higher spin heisenberg model _ , itp - sb-97 - 29 , q - alg/9706015 .
a. fledderjohann , k.h .
mtter and m. karbach and g. mller , _ significance of the non-2-spinon part of _
@xmath142 _ in the one dimensional _ @xmath141 _
heisenberg antiferromagnet at zero - temperature _ , cond - mat/9607102
. figs .
24 : @xmath0 ( a ) and @xmath4 ( b ) as functions of @xmath2 for fixed @xmath38 .
the values of @xmath38 are indicated . note that for the sake of illustration , @xmath0 is plotted to appropriate units . | we discuss some properties of the exact four - spinon dynamical correlation function @xmath0 in the antiferromagnetic spin 1/2 @xmath1-model , the expression of which we derived recently .
we show that the region in which it is not identically zero is different from and larger than the spin - wave continuum .
we describe its behavior as a function of the neutron momentum transfer @xmath2 for fixed values of the neutron energy @xmath3 and compare it to the one corresponding to the exact two - spinon dynamical correlation function @xmath4 .
we show that the overall shapes are quite similar , even though the expression of @xmath0 is much more involved than that of @xmath4 .
we finish with concluding remarks .
# 1([#1 ] ) -0.5 cm |
we write the hamiltonian eq .
( [ hii ] ) in the form @xmath121 , where @xmath26 is the site ( bond ) index and indices @xmath122 denote different types of ( local ) operators .
the parameter @xmath123 ( resp .
@xmath124 ) assumes values from @xmath125 to @xmath126 and refers to the change of the phonon number on site @xmath26 ( resp .
@xmath127 ) resulting from the action of the operator @xmath128 . in the sequel
we will refer to @xmath123 or @xmath124 as phonon parameters .
the index @xmath129 refers to the action on the spin degrees of freedom .
for example , the operators for different @xmath129 and @xmath130 read @xmath131 following ref . , we expand the boltzmann weight in terms of the local operators , i.e. , we sample the operator strings @xmath132 according to their weight in the boltzmann factor ( @xmath133 are states in the @xmath134-phonon occupation - number basis and @xmath135 is the local operator @xmath128 which appears on the @xmath136th position of the string ) .
the spin degrees of freedom can be updated using the very efficient scheme of operator loops .
@xcite during this update the parameters @xmath123 and @xmath124 remain unchanged . all that remains to be done is to devise an additional update for the phonon parameters .
this is achieved as follows : as the operators @xmath10 and @xmath137 appear always pairwise in the string , we have to increase the phonon parameter at one position ( of the string ) and decrease it at another position . in principle the two positions can be chosen at random , but here we want to point out that this can be done more efficiently : we choose an operator @xmath128 in the string .
next , we decrease ( resp . increase , depending on an appropriate probability ) one of its phonon parameters ( associated with a site which we will call @xmath28 ) . for the previous ( resp .
next ) operator in the string we choose with a certain probability
one of three options : either ( i ) we update the phonon number on site @xmath28 by @xmath138 or ( ii ) we cancel the operation and restore the initial configuration , or ( iii ) we increase ( resp .
decrease ) the phonon parameter related with site @xmath28 . in case ( ii ) or
( iii ) the procedure stops , while in the first case it is iterated until we finally choose one of the other options .
even though one might suspect that because of option ( ii ) the algorithm is inefficient since it is reminiscent of the bounce process in ref .
we find that our new update scheme reduces the autocorrelation time significantly .
the reason for the latter may be seen in the fact that we choose only one position of the two phonon operators at random , the other is in some optimized way chosen by the mc algorithm .
the simplest method to determine @xmath41 with monte carlo is measuring the order parameter . in our case
we have two options for the order parameter : one is @xmath139 the staggered bond length,@xcite another is the staggered kinetic energy@xcite @xmath140 since these operators are linear combinations of terms appearing in the hamiltonian , there is a simple method to evaluate the expectation values with the sse.@xcite to this end we define @xmath141 ( and @xmath142 ) which are the sum of the numbers of appearances of the operators @xmath143 and @xmath144 ( @xmath145 and @xmath146 ) in the string of operators created and updated in the simulation process .
the expectation values of the order parameters are related with the mean values of the numbers @xmath141 and @xmath142 ; the relation being @xmath147 and @xmath148 since the contributions to the order parameters may be positive or negative , it is difficult to detect a finite dimerization especially , when we are close to the critical coupling .
hence , it is difficult to decide whether we are in the gapped phase or not .
we therefore calculate the following expectation values @xmath149 and @xmath150 ( one should note that in general @xmath151 where @xmath152 . )
these have non - zero values , but in the gapless phase they should go to zero in the thermodynamic limit .
we note that while @xmath153 is independent of @xmath154 the expectation value @xmath155 _ does _
depend on the shift @xmath154 .
because of this and the fact that @xmath156 ( which suggests that @xmath157 decays faster with @xmath158 ) we prefer the second order parameter to the first and will only discuss @xmath157
. results for @xmath157 , the staggered kinetic energy , are presented in figs .
[ gapj0 ] and [ gapj1 ] . for the @xmath0 model ( fig .
[ gapj0 ] ) the order parameter @xmath157 decreases with system size as long as @xmath159 . for @xmath160
it seems to increase with system size rather than decrease .
we conclude that @xmath161 .
a correct assessment of the critical coupling remains difficult . a rough estimate for @xmath41
is obtained by assuming that @xmath41 lies in the interval @xmath162 where the behavior of @xmath157 changes .
this yields @xmath163 . in principle
, this estimate gives only an upper bound for @xmath41 , but when we apply the same method to the spin - rotationally invariant case , we find good agreement with the value for @xmath41 from ref .
( see fig .
[ gapj1 ] ) .
to better understand the physics of the model a mean field analysis is helpful .
the main idea is to replace the displacements @xmath164 by classical variables @xmath4 .
a common but general ansatz for these variables is @xmath165 . substituting this ansatz into the hamiltonian eq .
( [ hii ] ) we obtain , one sees that the spin - peierls model eq .
( [ hi ] ) does not depend on @xmath167 because of @xmath38 . at the mean field theory level we arrive at two conclusions , ( i )
there is a shift in @xmath46 proportional to @xmath168 and ( ii ) there is a dimerization transition where the dimerization is proportional to @xmath169 . in the spin - peierls model ( @xmath38 )
the shift in @xmath46 is absent , whereas in a toy model with @xmath170 no dimerization should be found . in our simulations ( @xmath171 )
we always have a shift but , since the conductance at @xmath172 does not depend on @xmath46 , we expect that @xmath29 is independent of @xmath99 and we arrive at the scaling law @xmath173 aa s.a .
wolf et al . ,
science * 294 * , 1488 ( 2001 ) .
m.c . cross and d.s .
fisher , phys .
b*19 * , 402 ( 1979 ) .
sandvik , r.p.r .
singh , and d.k .
campbell , phys .
b*56 * , 14510 ( 1997 ) .
sologubenko , k. giann , h.r .
ott , a. vietkine and a. revcolevschi , phys .
rev . b*64 * , 054412 ( 2001 ) .
c. hess , c. baumann , u. ammerahl , b. bchner , f. heidrich - meissner , w. brenig and a. revcolevschi , phys .
b*64 * , 184305 ( 2001 ) .
x. zotos and p. prelovek , in _ interacting electrons in low dimensions _
, kluwer academic publishers ( 2003 ) ; arxiv : cond - mat/0304630 .
alvarez , c. gros , phys .
lett . * 88 * , 077203 ( 2002 ) ; phys .
rev . b*66 * , 094403 ( 2002 ) .
f. heidrich - meisner , a. honecker , d.c .
cabra , and w. brenig , phys .
rev . b*68 * , 134436 ( 2003 ) .
k. louis and c. gros , phys .
b*68 * , 184424 ( 2003 ) ; k. louis , ph.d .
thesis , university of saarbrcken , ( 2004 ) .
f. meier and d. loss , phys .
lett . * 90 * , 167204 ( 2003 ) .
a.w . sandvik and d. campbell , phys .
lett . * 83 * , 195 ( 1999 ) .
khne and u. lw , phys .
b*60 * , 12125 ( 1999 ) .
sandvik , phys .
rev . b*59 * , r14157 ( 1999 ) .
o.f . syljusen and a.w .
sandvik , phys .
e*66 * , 046701 ( 2002 ) .
bursill , r.h .
mckenzie , and c.j .
hamer , phys .
* 83 * , 408 ( 1999 ) .
schulz , g. cuniberti , and p. pieri in _ field theories for low - dimensional condensed matter systems _
, g. morandi et al
springer ( 2000 ) .
k. kuboki and h. fukuyama , j. phys .
jpn . * 56 * , 3126 ( 19987 ) ; g.s .
uhrig , phys .
b * 57 * , r14004 ( 1998 ) .
l.g.caron and s.moukouri , phys .
* 76 * , 4050 ( 1996 ) . i. safi and h.j .
schulz , phys .
b * 52 * , r17040 ( 1995 ) .
maslov and m. stone , phys .
b * 52 * , r5539 ( 1995 ) . c. raas , u. lw , g.s .
uhrig , and r.w .
khne , phys .
b * 65 * , 144438 ( 2002 ) .
p. sun , d. schmeltzer , and a.r .
bishop , phys .
b * 62 * , 11308 ( 2000 ) . | we present results for the spin conductance of the one dimensional spin-1/2 heisenberg and @xmath0 model coupled to phonons .
we apply an approach based on the stochastic series expansion ( quantum monte carlo ) method to evaluate the conductance for a variety of phonon dispersions and values of spin - phonon coupling . from our numerical simulations and analytical arguments ,
we derive several scaling laws for the conductance .
_ introduction _ controlling and understanding the dynamics and transport of spins is an experimental and theoretical challenge aiming at the development of mesoscopic systems for spintronics applications .
in particular , the interest is recently focusing on the effect of interactions and the coupling of spin systems to the environment ; several proposals for the generation and detection of spin currents exist .
@xcite as a key quantity characterizing ballistic mesoscopic spin transport is the ( spin ) conductance , it is important to understand its behavior in prototype interacting spin systems coupled to lattice vibrations phonons .
related to this problem , since the 60 s , the spin - peierls transition , the dimerization of a lattice due to the spin - phonon interaction in the bulk , has been discussed for quasi - one dimensional compounds as ttf , tcnq , cugeo@xmath1 . a variety of analytical and numerical methods are still employed in the study of the dependence of the transition on the phonon spectrum and spin - phonon coupling .
@xcite finally , in close relation to this work , spurred by the experimental finding of an exceptionally high thermal conductivity @xcite in quasi - one dimensional transition metal compounds , there is recently considerable experimental and theoretical research interest in the finite temperature transport properties of spin models .
@xcite in particular , this activity leads to the development of novel numerical simulation techniques @xcite for the calculation of zero and finite temperature transport properties in one dimensional electronic / spin models . in this paper
we address the fundamental question of the spin conductance of the spin-1/2 heisenberg model and the @xmath0 model coupled to phonons .
thus this study , on the one hand , sheds light from a new perspective on the old question of the spin - peierls transition and on the other hand addresses a prototype question in mesoscopic physics ; it is of direct experimental relevance , as a detailed experimental set - up for the observation of the spin conductance was recently proposed .
@xcite quantum monte carlo ( qmc ) methods have already been successfully applied to the evaluation of thermodynamic quantities in spin - phonon systems .
@xcite we will use here the stochastic series expansion ( sse ) which allows the calculation of the conductance of interacting spin systems ; @xcite in this original application we extend it to coupled spin - phonon systems . in the following
we will first discuss the model and the method we are using , then we will present data for the conductance in the heisenberg model for optical and acoustic phonon dispersions , and derive several scaling laws . finally , we will show results for the @xmath0 model coupled to optical phonons . _ model and method _
the nominal spin - peierls hamiltonian describing the coupling of the one dimensional heisenberg model to lattice vibrations is given by @xmath2({\bf s}_n\cdot { \bf s}_{n+1})\nonumber \\
& + & \sum_n\omega_0/2\left[2{p_n^2}+(x_n - x_{n+1})^2\right ] , \label{hi}\end{aligned}\ ] ] where @xmath3 are spin-1/2 operators and @xmath4 , @xmath5 the displacement and momentum operators of the ions ( @xmath6 is the characteristic phonon frequency and here we set @xmath7 ) . because of the minus sign problem , this system eludes a direct analysis using the qmc method that we intend to employ here .
hence , we resort to the following generalized hamiltonian which is an extension of the hamiltonian discussed in refs . ,
@xmath8 , \label{hii}\end{aligned}\ ] ] where @xmath9 and @xmath10 ( @xmath11 ) are phonon annihilation ( creation ) operators . from eq .
( [ hii ] ) one recovers the spin - peierls model eq .
( [ hi ] ) if one sets @xmath12 and @xmath13 , @xmath14 . however , as a finite @xmath15 is required for the implementation of the sse @xcite we shift ( following ref . ) the boson operators @xmath10 and @xmath16 , @xmath17 so that @xmath18 . still , to sample the hamiltonian eq .
( [ hii ] ) with the sse we have to meet the requirement that all matrix elements of @xmath19 are negative , @xcite implying @xmath20 , or explicitly ( @xmath21 ) : @xmath22 the conditions in eq .
( [ cond ] ) thus forbid a direct analysis of the spin - peierls hamiltonian eq .
( [ hi ] ) as may be seen from eq .
( [ itoii ] ) .
however , as we will see we find two scaling laws which enable us to make predictions for the spin - peierls model eq .
( [ hi ] ) .
we define the operators @xmath23 and @xmath24 , @xmath25 which are the current operator for magnetic moments at site @xmath26 ( we denote the `` charge '' quantum by @xmath27 in analogy to the electric current ) and the potential operator of a local `` voltage drop '' at site @xmath28 , respectively .
the spin conductance @xmath29 is the linear response of the operator @xmath23 to the perturbation @xmath24 , @xmath30\rangle \,dt.\ ] ] it is independent of @xmath26 or @xmath28 and will be evaluated according to ref . while the sse method for the hamiltonian eq .
( [ hii ] ) is implemented following refs . .
note that we are considering only low temperatures and the effect of leads on the conductance is not included . for the update of the spin degrees of freedom we use the very efficient scheme of `` operator - loop updates '' .
@xcite the simplest version of an update for the phonons is to suggest a specific change in the configuration and accept / reject it with a certain probability .
here we want to emphasize that we obtain smaller autocorrelation times by using an appropriate cluster update scheme , reminiscent of the loop update ( see appendix , section [ update ] ) .
in our simulations we use a system size @xmath31 , a temperature @xmath32 and typically perform @xmath33 mc sweeps .
it should also be mentioned that in our simulations there is no need to introduce a bound for the phonon occupation number , but for technical reasons we impose a restriction of maximum @xmath34 phonons per site ; since it is much larger than the phonon numbers created during the simulation this introduces no further error . _ optical phonons _ the two parameters @xmath35 and @xmath36 lead to a nontrivial dispersion relation of the phonon modes .
for the moment we will set them to zero and consider phonons of constant frequency @xmath37 .
this ( einstein ) model was discussed for @xmath38 using the dmrg method @xcite and for @xmath39 with the sse .
@xcite both find that if @xmath40 is larger than a critical coupling @xmath41 the system undergoes a dimerization transition . in ref .
the authors analyze different values for @xmath37 finding that @xmath41 is approximately linear in @xmath37 if @xmath42 .
the authors in ref . discuss only one phonon frequency , namely @xmath43 .
they find that @xmath44 .
( data in our notation . )
we will use this result since we focus on @xmath45 .
the reason for this is that in many materials of actual interest @xmath46 is significantly larger than the debye temperature ; we also find that our qmc method converges better for small @xmath37 . for the conductance @xmath47 we expect the form of a step function : above the critical coupling @xmath48 it is zero due to the opening of the gap ; for @xmath49 the system
is supposed to be in a luttinger phase , where the conductance is given by the luttinger parameter @xmath50 . since the hamiltonian eq .
( [ hii ] ) is spin - rotationally invariant , the latter is fixed to the value of @xmath51 ( see ref . ) .
the fact that @xmath29 should be constant for @xmath52 is consistent with an effective hamiltonian that we can derive using second order perturbation theory ( cf .
refs . )
@xmath53 indeed , for @xmath39 we find no change in the conductance to order @xmath54 since @xmath29 does not depend on the bandwidth @xmath46 . in fig .
[ gphowi ] , we present results for the spin conductance @xmath29 for @xmath55 and @xmath39 . within error bars we confirm the expectation that the conductance remains at @xmath51 until the critical coupling is reached . in the gapped phase the conductance goes rather smoothly to zero instead of step - like which is due to the finite temperature @xmath56 we are using .
the point where the conductance has decayed to zero gives a rough estimate for the coupling @xmath57 where the gap satisfies @xmath58 .
( this can be tested using data from ref .
, if we assume that @xmath59which is a natural assumption since this is the scaling relation of @xmath41 to be derived below , cf .
eq .
( [ scale3 ] ) . for @xmath60
we find @xmath61 , i.e. , @xmath62 to be compared with fig .
[ gphocc3 ] below . )
subsequent calculations we performed for @xmath63 find an increased statistical error , but reveal no temperature dependence which implies that @xmath64 and @xmath65are very close to each other .
note that the gap has a typical kosterlitz - thouless form:@xcite first it opens very slowly , but then it grows rather rapidly . since @xmath47 is a step function which is fully described by the critical coupling where the conductance drops to zero
it inherits the scaling law of @xmath41 : @xmath66 to analyze the behavior of @xmath29 in the gapped region more closely , it is instructive to plot @xmath29 versus @xmath67 .
this is done in fig .
[ gphocc3 ] .
we find good agreement with another scaling relation
different from eq .
( [ scale1])specific to the gapped phase , namely , @xmath68 \label{scale2g}.\ ] ] where @xmath69 .
it is natural to assume that in the gapped region @xmath29 depends on the parameters of the system only through @xmath70 .
this would in turn imply that the scaling relation eq .
( [ scale2 g ] ) holds for the gap @xmath71 .
note that this relation is not consistent with the exponential behavior of @xmath71 typical for a kosterlitz - thouless transition .
however , we should expect eq .
( [ scale2 g ] ) to be approximately valid for not too large @xmath72 and @xmath37 .
we note that the data for @xmath73 from ref .
show also good agreement with eq .
( [ scale2 g ] ) when @xmath74 .
_ acoustical phonons _
now we will consider nontrivial phonon dispersions , @xmath75 . performing a canonical bogoliubov transformation in momentum ( k- ) space
, we obtain the phonon frequency dispersion @xmath76 ^ 2 -[\omega^ + \cos(k)]^2}\ , .
\label{omegk}\ ] ] we see that if @xmath77 , as , e.g. in the spin - peierls model eq .
( [ hi ] ) , @xmath78 , whereas if @xmath79 there is a gap in the spectrum .
this dispersion is illustrated in fig .
[ omphonk ] .
after diagonalizing the phonon part of the hamiltonian eq .
( [ hii ] ) we proceed similarly with the spin - phonon coupling term . to this end , adopting for convenience a fermion representation ( @xmath80 ) by the jordan and wigner transformation , the phonon - fermion scattering term @xmath81 is written in k - space as @xmath82}}c_k^\dag c_{q } ( a_{k - q}+a_{q - k}^\dag).\ ] ] since we are here interested only in the low temperature regime we may assume that the variables @xmath83 lie at the fermi momenta @xmath84 .
the process responsible for the decrease of the conductance should then be the scattering between two fermi points by a phonon with momentum @xmath85 and the only relevant frequency should be @xmath86 .
thus , for a general dispersion relation , we should recover the same result for @xmath29 as for the optical phonons of the previous section , by mapping @xmath87 , @xmath88 .
we expect then the conductance @xmath29 to depend only on @xmath89 with @xmath90 ; to verify this expectation we show @xmath29 in fig .
[ akpho2 ] for a set of parameters @xmath37 , @xmath35 , @xmath36 compatible with eq .
( [ cond ] ) .
the figure indeed suggests that @xmath91 and together with eq .
( [ scale1 ] ) we arrive at the desired scaling relation @xmath92 _ scaling with @xmath40 and @xmath93 _ let us again consider the hamiltonian in k - space as in the previous section .
if we have @xmath40 and @xmath93 different from zero , the coupling terms read @xmath94 here @xmath95 is a prefactor whose precise form is for the following discussion irrelevant . at low temperatures we have again @xmath96 which implies that @xmath97 .
hence only two phonon modes are important : @xmath98 and @xmath86 .
the @xmath98 mode describes scattering on one fermi point which means that it probably rescales the parameters of the system , but has otherwise no influence on the transition . for the phase transition
only the @xmath86 mode is important . as we see from eq .
( [ lowenergy ] ) the mode with @xmath98 couples to the spin degrees of freedom via @xmath99 and the mode @xmath86 via @xmath100 . as the decay of the conductance is due to the opening of the gap , we are led to assume that @xmath29 depends on @xmath100 , but not on @xmath99 , i.e. , @xmath101 ( alternatively , one can motivate this scaling law with a mean field theory ansatz which is outlined in the appendix , section [ meanfield ] . ) in fig .
[ gaaic ] we confirm this scaling by the overlap of curves as a function of @xmath102 , while in the inset we indeed find that the dimerization transition is delayed by increasing @xmath103 , till @xmath104 where it is absent as predicted by eq .
( [ scale3 ] ) . a further test of eq .
( [ scale3 ] ) is provided by comparing data for @xmath41 . for @xmath43 and @xmath39 ref .
finds @xmath105 ( with @xmath106 uncertainty ) , whereas according to ref . for the same @xmath37 but @xmath38
the transition occurs when @xmath107 .
the difference of the two values is @xmath108 which means reasonable agreement . _
@xmath0 model_ in this section we turn to the @xmath0 model , that is we replace @xmath109 in the hamiltonian eq .
( [ hii ] ) by @xmath110 .
the model was studied by caron and moukouri @xcite for @xmath38 .
they found a phase transition similar to the one in the spin - peierls model but here @xmath111 .
we computed the conductance for two phonon frequencies .
the results are shown in fig.[gpho0 ] .
we find that the zero temperature conductance obeys the scaling law @xmath112 it is important to point out that the @xmath0 model is not spin - rotationally invariant and therefore the value of the conductance is not fixed in the luttinger phase , but rather decreases monotonously , as the fig .
[ gpho0 ] shows . in the gapped phase
the conductance decays again quite smoothly to zero , such that it is barely possible to decide where exactly the transition takes place .
we therefore employed here two different methods for calculating @xmath41 .
first qmc simulations for the staggered kinetic energy which give a rigorous upper bound ( details are deferred to the appendix , section [ critical ] ) , second we used the data for @xmath41 from ref . while assuming that the scaling law eq .
( [ scale3 ] ) , @xmath113 , holds the two results for @xmath41 are indicated by lines in fig .
[ gpho0 ] for comparison .
the difference between the two predictions for @xmath41 is about @xmath114 .
one should emphasize again that both methods are not numerically exact : the qmc yields only an upper bound for @xmath41 , and the other method relies on the validity of eq .
( [ scale3 ] ) which is examined in fig .
[ gph0c ] by considering nonzero @xmath93 . for the parameter
accessible to us eq .
( [ scale3 ] ) seems to be approximately , but _ not exactly _ fulfilled . in the heisenberg model the dependence on @xmath115 and @xmath37
was described by two separate scaling laws eqs .
( [ scale1 ] ) and ( [ scale3 ] ) . for the @xmath0 model
the situation is more complicated , as we found that for @xmath116 the scaling law eq .
( [ scale0 ] ) is violated .
this implies that a generalization of eq .
( [ scale3 ] ) which holds for the @xmath0 model must depend on @xmath37 .
we will not try to find such a generalization here , but want to point out that for the special case of @xmath43 one can obtain a much better fit for our data ( for @xmath117 ) than eq .
( [ scale3 ] ) by using the following modified scaling ansatz ( see fig . [ gph0c ] ) @xmath118.\ ] ] _ conclusions _ we have shown that the known scaling laws for @xmath41 in the heisenberg and the @xmath0 model , i.e. , eqs .
( [ scale1 ] ) & ( [ scale0 ] ) also hold for the conductance . for the spin - rotationally invariant case we found a generalization for acoustic phonons
eq .
( [ scale2 ] ) as well as a scaling relation for the gap eq .
( [ scale2 g ] ) .
we also discussed the difference between two possible couplings to phonons ( @xmath39 & @xmath119 ) .
for the spin - rotationally invariant case the two ways of coupling are related by a simple scaling law eq .
( [ scale3 ] ) , whereas for the @xmath0 chain the situation appears to be more complicated .
assuming that our scaling laws are not restricted we may make a prediction for the conductance of the spin - peierls model eq .
( [ hi ] ) by putting together eqs .
( [ itoii ] ) , ( [ scale2 ] ) and ( [ scale3 ] ) as well as data for the critical coupling from ref . .
we expect a step - like conductance which drops to zero when @xmath120 . one final comment should be made . over the last few years , it has generally agreed upon that the conductance measured in experiments corresponds to the conductance of a system coupled to noninteracting leads .
the leads - which were not considered here - may strongly affect the conductance .
@xcite we can argue however that the dimerization transition we observed will drive the conductance to zero even if it occurs only locally , i.e. in the interacting part system and not in the leads .
hence , we think that our results have a clear experimental consequence .
we would like to acknowledge useful discussions with d. baeriswyl and financial support by the e.u .
grant mirg - ct-2004 - 510543 . |
in early september 2001 one of the authors ( uf ) attended the zakopane meeting on tubes , sheets and singularities @xcite which was also attended by richard pelz .
there were many discussions about the issue of finite - time blowup for 3d incompressible euler flow .
richard , who had studied a flow introduced by @xcite , had evidence in favor of blowup but one could not rule out that the highly special structure of this flow would lead to quasi - singular intermediate asymptotics .
the three authors of this paper then decided to embark in a long - term project aimed at getting strong evidence for or against blowup for a wide class of flows encompassing the taylor green flow @xcite and the kida
pelz vortex @xcite , namely space - periodic flow with or without symmetry having initially only a few fourier harmonics .
such initial flow is not only analytic but also entire : there is no singularity at finite distance in the whole complex spatial domain .
as has been known since the mathematical work of @xcite , any real finite - time singularity is preceded by complex - space singularities approaching the real domain and which can be detected and traced using fourier methods @xcite .
this method is traditionally carried out by spectral simulations which run out of steam when the distance @xmath1 from the real domain to the nearest complex - space singularity is about two meshes .
we pointed out in a recent paper ( * ? ? ?
* henceforth referred to as fmb ) , which also reviews the issue of blowup , that it may be possible to extend the method of tracing of complex singularities by performing a holomorphic transformation mapping singularities away from the real domain and , perhaps doing this recursively .
this is the basic idea of the _ spectral adaptive _ method which aims at combining the extreme accuracy of spectral methods with the local mesh refinement permitted by adaptive methods . in one dimension
the complex - space singularities of pde s are isolated points , at least in the simplest cases , as for the burgers equation . in higher dimension
they are extended objects , such as complex manifolds .
understanding the nature and the geometry of such singularities is a prerequisite for mapping them away .
many aspects can already be investigated in the two - dimensional case for which we know not only that blowup is ruled out , but we also know that the flow stays a lot more regular than predicted by rigorous lower bounds ( basically @xmath1 seems to decrease exponentially whereas the bound is a double exponential in time ) .
in fmb we gave some evidence that in 2d the complex singularities are on a smooth manifold , but the nature of the singularities was not very clear and in particular the issue of finiteness vs. blowup of the complex vorticity was moot .
in fmb we also pointed out that the issue of singularities can be simplified if we limit ourselves to short times .
let us briefly recall the setting .
we start with the 2-d euler equation written in stream function formulation @xmath3 where @xmath4 . as in fmb
, we focus on the two - mode initial condition @xmath5 one of the simplest initial condition having nontrivial eulerian dynamics . the solution @xmath6 , obtained by analytic continuation to complex locations @xmath7 , is expected to have singularities at large imaginary values when @xmath8 is small .
if one focuses on the quadrant @xmath9 and @xmath10 , an asymptotic argument given in fmb suggests looking at solutions satisfying the _ similarity ansatz _
@xmath11 substitution in gives the _ similarity equation _ @xmath12 where the overscript tilde means that the partial derivatives are taken with respect to the new variables .
the initial condition becomes a boundary condition @xmath13 note that has no time variable .
its solution can be shown to be analytic for sufficiently negative @xmath14 and @xmath15 .
if it has singularities in the real or complex domain then , by , the solution of the original euler equation should have short - time singularities at a distance @xmath16 .
the outline of the present paper is as follows . in section [ s : numshort ] we check the logarithmic law for @xmath1 and thus the validity of the similarity ansatz . in section [ s : solving ] we develop a new technique for solving the similarity equation in both the real and complex domains . in section [ s : results ] we present numerical results on the nature and the geometry of the singularities . in section [ s : singular ] we show how to actually construct the singular manifold from the fourier transform of the solution . in section [ s : conclusion ] we make some concluding remarks .
the standard way of measuring the distance @xmath17 of the nearest complex - space singularity ( also called the width of the analyticity strip ) is to use the method of tracing @xcite .
indeed , the spatial fourier transform @xmath18 of a periodic function @xmath19 has its modulus decreasing at large wavenumbers @xmath20 as @xmath21 in both one and several space dimensions .
more precisely , for each direction @xmath22 , the modulus of the fourier transform decreases as @xmath23 ; the width of the analyticity strip is then the minimum over all directions @xmath24 of @xmath25 which by a steepest descent argument also controls the high-@xmath26 decrease of the angle average of the modulus of the fourier transform .
is related to the singular manifold in the two - dimensional case , see section [ s : singular ] . ]
accurate measurement of the width of the analyticity strip @xmath17 gets difficult when it becomes smaller than a few meshes , so that there is not enough resolution to see long exponential tails @xcite .
a different difficulty appears when @xmath17 is very large and the fourier transform decreases so rapidly that it gets lost in roundoff noise at rather small @xmath26 s . to check on the logarithmic law of variation of @xmath1 at short times for the full time - dependent euler equation
, we have to face the latter difficulty . to overcome it , we employed a 90-digit multiprecision spectral calculation with the number of grid points ranging from @xmath27 to @xmath28 the temporal scheme is fourth order runge - kutta .
[ f : shorttimespectra ] shows the @xmath26 dependence at various short times of the angular averages .
] of the modulus of the fourier transform of the velocity @xmath29 , where @xmath30 and @xmath31 are the derivatives with respect to @xmath32 and @xmath33 .
note that there is a strong odd - even wavenumber oscillation .
this has to do with the interference of two complex singularities separated by @xmath34 in the @xmath33 direction ( a consequence of a symmetry of the initial condition ) .
60 10 short - time spectra ( fig.6 report ) a very clean exponential decrease is observed for even wavenumbers .
this allows the measurement of @xmath1 for hundreds of values of @xmath8 covering a very wide range , as shown in fig .
[ f : deltat ] in log - lin coordinates .
60 10 logarithmic law it is seen that , up to approximately @xmath35 , the logarithmic law is satisfied over eight decades , a range which could not be achieved without the use of very high - precision computations .
we are interested in space - periodic solutions to satisfying the boundary condition .
this problem can be solved exactly using fourier series @xmath36 note that we have dropped all tildes on the space variables since from now on we will work exclusively with the similarity variables .
obviously , the boundary condition allows the presence only of fourier harmonics with wavevectors in the quadrant @xmath37 .
after is rewritten in terms of the fourier coefficients of @xmath38 , the multiplications appearing in the jacobian go over into simple convolutions with only finitely many terms , because all wavevectors involved must have non - positive components .
it follows that , for integer @xmath39 , all fourier coefficients for wavevectors which are on the line @xmath40 ( see fig .
[ f : quadrant ] ) 60 10 quadrant are expressible in terms of the finitely many fourier coefficients with wavevectors lying above this line .
more precisely , for any point p on this line , the region of dependence is the grey - shaded rectangle .
specifically , we define the selective fourier sum over this line @xmath41 so that @xmath42 from we then obtain the recursion relations @xmath43 with @xmath44 equation can then be rewritten as a set of algebraic equations for the fourier coefficients : @xmath45 where @xmath46 .
one could solve the poisson equations which recursively define the @xmath47 s using fft methods as in @xcite . alternatively and this is the method used here
equations - can be used to calculate recursively the _ exact _ expressions of all the fourier coefficients .
the first two @xmath47 s are @xmath48 we have numerically determined the @xmath47 s for @xmath49 up to @xmath50 , using quadruple - precision ( 35-digit ) accuracy . as we shall see in the next section , lower accuracy would give spurious results . from and ,
the @xmath51 are obviously real .
furthermore , we find numerically that their signs alternate : specifically @xmath52 , for all @xmath53 except @xmath54 .
we shall work here mostly with shell - summed ( fourier ) amplitudes .
for a given periodic function @xmath55 , we define its shell - summed amplitude as @xmath56 where the @xmath57 are the fourier coefficients of @xmath55 .
the shell - summed amplitudes for the solution @xmath38 to the similarity equation can be calculated , in principle exactly , for all wavenumbers @xmath58 such that @xmath59 where @xmath60 , this being the radius of the largest disk centered at the origin and contained in the triangular region of fig .
[ f : quadrant ] .
[ f : ssaf ] shows , as a function of @xmath58 , the shell - summed amplitude of @xmath38 for @xmath61 . as seen in fig .
[ f : ssaf](a ) , the results obtained with 15 and 35 digit precisions differ markedly beyond wavenumber 800 .
even the 35-digit calculation becomes unreliable beyond wavenumber 1300 . in fmb , when we discussed results about singularities without resorting to short - time asymptotics , we reported various difficulties : the need to perform a kind of krasny filtering @xcite and our failure to improve the range of scaling by going to resolutions higher than @xmath62
. it is now clear that the key to clean results is to use high precision .
we fit the shell - summed amplitude of @xmath38 by a function of the form @xmath63 and find the best fit to be @xmath64 , as shown in fig .
[ f : ssaf](b ) .
the fit is done by least squares in the lin - log representation over the interval @xmath65 .
60 10 ssa for f with loglog as inset hence , the solution to the similarity equation has complex - space singularities , the closest one being within @xmath66 .
this relatively small value , for an equation in which all the coefficients are order unity , is accidental and can be changed by slightly modifying the coefficients in front of the two harmonics in the initial condition and in , as will be shown later in this section .
the angular dependence of the mode amplitude at high wavenumbers can also be obtained .
[ f : thetadepampl ] shows that @xmath67 decreases exponentially with @xmath68 for a given direction @xmath69 .
is within @xmath70 of the direction @xmath71 . ]
we see that the logarithmic decrement , denoted by @xmath72 , varies strongly with @xmath71 .
its variation is shown in fig .
[ f : deltaoftheta ] for @xmath73 , that is over the third angular quadrant where @xmath74 and @xmath75 are negative . becomes infinite ( w. pauls , private communication ) , but this is hidden by the slight angular averaging .
] obviously , the width of the analyticity strip is @xmath76 the minimum being achieved in the _ most singular direction _
@xmath77 , such that @xmath78 .
the function @xmath72 can be related to the singular manifold ( see section [ s : singular ] ) .
60 10 lin - log of ampl . for various theta 60 10 deltaoftheta as in one dimension , the prefactor of the exponential in the shell - averaged amplitude @xmath79 contains information about the nature of the singularities . in the inset of fig .
[ f : ssaf ] we see about one decade of power - law scaling before the exponential falloff
. this range can be increased by moving closer to the singularity , more precisely to the point @xmath80 on the singular manifold closest to the real domain which has imaginary part @xmath81 with @xmath82 and @xmath83 .
such an imaginary shift produces a function @xmath84 whose fourier coefficients are obtained by multiplying the fourier coefficients of @xmath38 by @xmath85 , where @xmath86 .
amounts to changing the initial condition into @xmath87 within terms irrelevant for @xmath88 and @xmath10 .
[ f : ssashifted ] shows shell - summed amplitudes of @xmath84 for four values of @xmath89 . for @xmath90
more than two decades of power - law scaling is seen with an exponent somewhere between @xmath91 and @xmath92 .
this clean scaling evidence has an important consequence for the vorticity on the singular manifold ( see below ) .
60 10 loglog of shifted to gain additional insight , we now show results in physical space in terms of the real and imaginary parts of the vorticity @xmath93 .
there are actually kramers kronig relations between the real and imaginary parts . ] for this we limit ourselves to @xmath94 . as a consequence , @xmath74 runs from @xmath95 to 0 , while @xmath75 varies from @xmath96 to 0 .
we use an fft program with @xmath97 grid points .
[ f : contours ] shows contours of the real and imaginary parts 60 10 of the vorticity .
the symmetries seen are a consequence of dynamically preserved symmetries in the initial condition . near the center @xmath98
there is a highly anisotropic large - amplitude sheet - like structure which can be interpreted as the manifestation of a smooth singular manifold which gets within roughly four meshes of the real domain .. ] by far , the fastest vorticity variation is obtained perpendicularly to the sheet - like structure , in the most singular direction @xmath99 .
[ f : singcuts ] shows the variation of the vorticity along a cut through @xmath98 in the most singular direction .
it is seen that the vorticity becomes rather large ( around 40 for the real part ) .
the behavior of the real part near the peak , as a function of the distance @xmath100 to the peak is very roughly as @xmath101 as seen in fig .
[ f:1overxfit ] .
60 10 4 cuts in the most singular direction 60 10 1overx fit for vorticity it is also of interest to show the variation of the vorticity in the @xmath102-plane , that is @xmath103 .
symmetry implies that this is a real quantity which , by , can be written as @xmath104 k^2&= & k_1 ^ 2 + k_2 ^ 2 .
\nonumber\end{aligned}\ ] ] the latter can be viewed as a double power series in the variables @xmath105 and @xmath106 .
the alternating sign property of the @xmath51 mentioned at the end of section [ s : solving ] implies that all the coefficients @xmath107 are nonnegative for @xmath108 .
it is well known that if a power series in one variable with nonnegative coefficients defines an analytic function with singularities , then the nearest singularity to the origin is on the real positive axis ( vivanti s theorem , see , e.g. , * ? ? ?
here we expect that there will be a whole singular curve in the @xmath102-plane .
the fourier series can be used to calculate @xmath109 , at least within the disk @xmath110 .
[ f : straight ] shows contours 60 10 straight of @xmath111 in and around this disk .
the contours are almost straight lines perpendicular to the most singular direction , because @xmath17 is very small .
the variation of @xmath111 along the the most singular direction as a function of the distance @xmath112 to the origin is shown in fig .
[ f : threequarters ] .
when plotted as a function of the distance @xmath113 to the nearest singularity @xmath114 , it follows roughly a power law with an exponent close to @xmath115 over almost two decades .
we observe that the vorticity reaches values around 200 .
peak values between 100 and 200 and a scaling exponent close to @xmath115 are also obtained for the behavior of @xmath111 near other points of the singular manifold ( to be defined precisely in the next section ) .
we observe that there may be an inconsistency between the exponent observed in the real direction which is roughly @xmath116 ( fig .
[ f:1overxfit ] ) and the exponent @xmath115 observed in the imaginary direction ( fig .
[ f : threequarters ] ) since the vorticity , being analytic , should have the same scaling .
the singularities being contained in the imaginary plane above @xmath117 , at least a few meshes away from the real domain , the scaling in the imaginary direction is expected to be more reliable .
actually , there is strong evidence for the vorticity being infinite at @xmath118 coming from the behavior of the shell - summed amplitudes .
indeed , we know ( i ) that all the terms in the double sum are nonnegative and ( ii ) that the shell - summed stream function amplitude decreases nearly as @xmath119 over two decades or more ( fig .
[ f : ssashifted ] ) .
it follows that the shell - sums for the vorticity at @xmath118 are almost @xmath26-independent and thus , when we sum over @xmath26 , we get an infinite value . 60 10 threequarters
in one dimension it is well known that the high - wavenumber asymptotics of the fourier transform of an analytic periodic function is governed by the singularities nearest to the real domain ( see , e.g. , * ? ? ?
* ; * ? ? ?
this has a non - trivial extension to more than one dimension when the function has singularities on a complex manifold @xmath120 determined by an equation @xmath121 ; let us sketch this in the two - dimensional case .
consider a periodic analytic function @xmath122 given by the fourier series .
the fourier coefficients are given by the double integral over the real periodicity domain @xmath123 let us set @xmath124 and @xmath125 .
we are interested in the behavior of @xmath126 when @xmath127 for a given direction defined by the angle @xmath71 .
for this we change coordinates from @xmath128 to @xmath129 chosen , respectively , parallel and perpendicular to @xmath130 .
note that the exponential in involves only @xmath131 . hence can be rewritten as a one - dimensional fourier transform @xmath132 where @xmath133 and @xmath134 is @xmath24 rotated by @xmath135 .
it follows from that , for @xmath136 , @xmath137 where @xmath138 is the singularity of @xmath139 in the complex @xmath140 plane nearest to the real domain and with negative imaginary part .
hence @xmath141 where @xmath142 .
how do we obtain the singularities of @xmath139 , given by the integral ?
for some ( complex ) @xmath140 there may be singularities of @xmath143 for real @xmath144 .
they can however be avoided by shifting the contour of integration away from the real @xmath144 axis .
if we change @xmath140 , this will work as long as the contour does not get pinched between two ( or more ) coalescing singularities ( the pinching , generically , does not take place on the real @xmath144 axis ) .
assuming that the singular manifold @xmath145 can be represented by @xmath146 in the @xmath140 and @xmath144 variables , a necessary condition for this pinching is a double root in @xmath144 , i.e. @xmath147 and @xmath148 .
this system of two equations has discrete solutions @xmath149 , one of which will control the asymptotics . in an earlier version of this paper ( available at http://arxiv.org/pdf/nlin.cd/0310044v1 ) we stated that the pinching argument may already be known .
recently w. pauls pointed out to us that it has already been used in a special case by henri poincar ( see sections 9496 of * ? ? ?
* ) and has later been generalized by @xcite .
all this simplifies considerably for the function @xmath38 , the solution of the similarity equation .
indeed , all the relevant singularities have real part at the point @xmath150 and we can restrict everything to the pure imaginary plane @xmath102 passing through this point . let the restriction - plane .
] of the singular manifold to this plane have the parametric representation @xmath151 , where @xmath152 and @xmath153 are differentiable functions of the parameter @xmath71 , chosen in such a way that the angle between the @xmath152 axis and the tangent at the singular manifold is @xmath154 .
we obviously have @xmath155 this representation is convenient because the aformentioned condition of having a double root is easily shown to express the tangency to the singular manifold of the straight line perpendicular to the direction @xmath156 at a distance @xmath72 from the origin , as illustrated in fig . [
f : podaire ] . is because the wavector has negative components and thus lies in the third quadrant @xmath157 . ] 60 10 delta of theta from sing .
manifold 60 10 analyticity disks the equation for the tangent reads @xmath158 the system - is easily inverted ( provided the singular manifold restricted to the @xmath102-plane is convex ) to give @xmath159 where @xmath160 .
these equations allow in principle the construction of the singular manifold from the knowledge of the angular dependence of the logarithmic decrement @xmath72 , as given in fig .
[ f : deltaoftheta ] . in practice , because @xmath72 changes very quickly near its minimum , this is not a very well conditioned procedure .
an alternative construction by envelope of analyticity disks is now proposed .
we observe that , when we replace the real plane by a parareal plane shifted by an imaginary vector @xmath161 , the location on the singular manifold nearest to this parareal plane is generically different from the one nearest to the original real plane . for a general 2d problem without special symmetry ,
the singular manifold is a one - dimensional complex manifold , and thus may be parametrized with two real variables , e.g. @xmath162 and @xmath163 .
for the special case considered here , we can work in the @xmath102 plane and draw around each point @xmath161 an analyticity disk of radius @xmath164 .
the latter is determined as the logarithmic decrement of the shell - summed average of @xmath165 , where the exponential factor is the consequence in fourier space of the imaginary translation @xmath166 .
it is easily shown that @xmath167 where @xmath72 is the logarithmic decrement in the direction @xmath71 , defined in section [ s : results ] .
since @xmath72 is rather poorly determined , it is better to measure @xmath164 directly .
we perform this calculation for 305 choices of the pair @xmath161 , taken on a regular grid .
the shell - summed averages are fitted to an exponential with an algebraic prefactor . only points displaying decreasing exponentials , i.e. , located below the singular manifold are kept .
the result is shown in fig .
[ f : envelope ] , where the singular manifold emerges as the envelope of all the analyticity disks .
let us finally observe that the method of analyticity disks can be generalized to problems without any special symmetry ; this will be discussed elsewhere .
let us now summarize what we have learned about singularities for the similarity equation governing the short - time asymptotics .
first , we remind the reader that for one - dimensional problems , complex - space singularities can be at isolated points but need not : for nonintegrable ode s they often form fractal natural boundaries @xcite . in two dimensions or higher ,
isolated singularities are ruled out : the singular set is a complex manifold ( or worse ) . in the present case ,
the evidence is that the singularities are on a smooth and probably analytic one - dimensional complex manifold .
this was already conjectured in fmb on the basis of numerical results for the full time - dependent euler equation .
now , we have an explicit construction of the singular manifold by the method of envelope of analyticity disks ( section [ s : singular ] ) , which gives strong numerical evidence for smoothness . the very clean scaling we have observed for the fourier transform of the solution near the singularity ( fig .
[ f : ssashifted ] ) implies that the vorticity is infinite on the singular manifold since in 2d the vorticity is conserved along lagrangian fluid trajectories ( both in the real and the complex domain ) , this result strongly suggests that the singular manifold is mapped to complex infinity by the inverse lagrangian map .
we have studied here a special flow , whose two - mode initial condition given by has a center of symmetry at @xmath168 .
such symmetries are hard to avoid when using a minimal number of fourier modes . at short times
, the symmetry constrains the complex - space singularity nearest to the real domain to have the real part of its location at @xmath169 . in fmb
we have shown that this ceases to be the case at later times .
we do of course hope that there is something universal in the nature of the singularities , be it only their very existence .
we must however stress that in the present case we have a singular manifold with a continuous piece and not just isolated points when @xmath170 .
this is definitely non - generic , since a manifold with complex dimension 1 may be viewed as a two - dimensional real manifold in four - dimensional real space .
when higher - order harmonics are added to the initial condition , the situation becomes more complicated and the short - time asymptotic equation will in general depend on the ratio @xmath171 , that is on the direction in which the imaginary coordinates become large .
let us finally point out that one particularly challenging problem is to actually prove that there are singularities in the complex domain : at the moment we are not aware of a single instance of a solution to the incompressible euler equation with smooth initial data for which the existence of a singularity ( real or complex ) is demonstrated . in the present case , the proof may be facilitated by the observation made in section [ s : results ] that , in the @xmath102-plane the vorticity is real and can be represented as a double power series with ( apparently ) nonnegative coefficients . a lower bound on these coefficient which are obtained from the solutions of the recurrence relations
could lead to such a proof . *
acknowledgments * + we are grateful to walter pauls , tetsuo ueda and vladislav zheligovsky for useful remarks .
computational resources were provided by the yukawa institute ( kyoto ) .
this research was supported by the european union under contract hprn - ct-2000 - 00162 and by the indo - french centre for the promotion of advanced research ( ifcpar 2404 - 2 ) .
tm was supported by the grant - in - aid for young scientists [ ( b ) , 15740237 , 2003 ] and the grant - in - aid for the 21st century coe `` center for diversity and universality in physics '' from the japanese ministry of education and received also partial support from the french ministry of education .
jb acknowledges support from the national science foundation under agreement no .
dms-9729992 .
99 bailey , d.h .
, 1995 . a fortran-90 based multiprecision system , rnr technical report rnr-94 - 013 .
see also bardos , c. , benachour , s. , zerner , m. , 1976 .
analyticit des solutions periodiques de lquation deuler en deux dimensions , c. r. acad .
paris 282 a , 995998 .
bajer , k. , moffatt , h.k .
( ed ) , 2003 .
tubes , sheets and singularities in fluid dynamics : proceedings of the nato arw , 27 september 2001 , zakopane , poland , kluwer academic publishers , dordrecht .
brachet , m .- e . , meiron , d.i . ,
orszag , s.a . , nickel , b.g . , morf , r.h . ,
frisch , u. , 1983 . small - scale structure of the taylor - green vortex , j. fluid mech .
130 411452 .
caflisch , r.e . ,
hou , t.y . ,
lowengrub , j. , 1999 . almost optimal convergence of the point vortex method for vortex sheets using numerical filtering , math . comput .
68 , 14651496 .
carrier , g.f . , krook , m. , pearson , c.e .
functions of a complex variable : theory and technique , mcgraw - hill , new york .
chang , y.f .
, tabor , m. , weiss , j. , 1982 . analytic structure of the hnon heiles hamiltonian in integrable and nonintegrable regimes , j. math . phys .
23 , 531538 .
cramr , h. , 1938 .
sur un nouveau thorme - limite de la thorie des probabilits , actualits scientifiques et industrielles , 736 , 523 dienes , p. , 1931 .
the taylor series , an introduction to the theory of functions of a complex variable , oxford university press .
frisch , u. , 1984 .
the analytic structure of turbulent flows , in proceed .
chaos and statistical methods , sept .
1983 , kyoto , y. kuramoto , ed .
pp . 211220 , springer .
frisch , u. , matsumoto , t. , bec , j. , 2003 .
singularities of euler flow ?
not out of the blue !
j. stat .
113 , 761781 .
frisch , u. , morf , r. , 1981 .
intermittency in nonlinear dynamics and singularities at complex times , phys .
rev . a 23 , 26732705 .
kida , s. , 1985 .
three - dimensional periodic flows with high - symmetry , j. phys .
japan 54 , 21322136 .
krasny , r. , 1986 .
a study of singularity formation in a vortex sheet by the point - vortex approximation , j. fluid mech .
167 , 6593 .
majda , a.j . ,
bertozzi , a.l . , 2001 .
vorticity and incompressible flow ( section 9.4 ) , cambridge texts in applied mathematics , cambridge university press , cambridge .
pauls , w. , matsumoto , t. , 2005 .
lagrangian singularities of steady two - dimensional flow , geophys .
fluid dynam .
99 , 6175 .
pelz , r.b . , 1997 .
locally self - similar , finite - time collapse in a high - symmetry vortex filament model , phys .
e 55 , 16171626 .
pelz , r.b . , 2001 .
symmetry and the hydrodynamic blow - up problem , j. fluid mech .
444 , 299320 .
pelz , r.b . ,
gulak , y. , 1997 .
evidence for a real - time singularity in hydrodynamics from time series analysis , phys .
79 , 49985001 .
poincar , h. , 1899 .
les mthodes nouvelles de la mcanique cleste , gauthier - villars , paris reprinted by dover in 1957 .
sulem , c. , sulem , p .-
l . , frisch , h. , 1983 . tracing complex singularities with spectral methods , j. comput .
50 , 138161 .
tsikh , a. , 1993 .
conditions for absolute convergence of series of taylor coefficients of meromorphic functions of two variables , math .
ussr sbornik 74 , 336360 . | using a very high precision spectral calculation applied to the incompressible and inviscid flow with initial condition @xmath0 , we find that the width @xmath1 of its analyticity strip follows a @xmath2 law at short times over eight decades . the asymptotic equation governing the structure of spatial complex - space singularities at short times ( frisch , matsumoto and bec 2003 , j. stat . phys . 113 , 761781 ) is solved by a high - precision expansion method . strong numerical evidence is obtained that singularities have infinite vorticity and lie on a complex manifold which is constructed explicitly as an envelope of analyticity disks
. _ fluid dyn .
res . _ * 36 * , 221237 ( 2005 ) .
47.11.+j 02.40.xx 0230.fn 02.70.hm + euler equation ; asymptotics ; complex singularities . |
planetary nebulae ( pne ) display a wide variety of shapes ranging from round , which can be simply understood in terms of the symmetric interacting stellar winds ( isw ) model @xcite , through complex symmetrical shapes , such as hour - glasses and butterflies , to shapes which have only rotational point - symmetry .
many theories have been introduced to explain these complex shapes from adding an asymmetric slow wind to the isw model @xcite to involving the effects of binary central stars @xcite and magnetic fields @xcite .
observations of pne have shown several cases where only the outer shell shows a departure from symmetry . in these cases ,
the cause of the asymmetries has been postulated to be an interaction with the interstellar medium ( ism ) .
interaction with the ism by pne was first discussed by @xcite with an early theoretical study by @xcite employing the snow - plough model of @xcite .
smith concluded that a nebula will fade away before any disruption of the nebular shell becomes noticable .
@xcite used the same approximation with higher relative velocities to the ism and ism densities and concluded similarly .
in contrast , @xcite found that many pne with large angular extent show signs of pn
ism interaction and that all nebulae containing central stars with a proper motion greater than 0.015 arcsecyr@xmath0 do so .
@xcite hydrodynamically modelled the interaction revealing that the pn shell is first compressed in the direction of motion and then in later stages this part of the shell is significantly decelerated with respect to the central star .
both conclude that the interaction with the ism becomes dominant when the density of the nebular shell has dropped below a certain critical limit , typically @xmath1@xmath2 for a pn in the galactic plane .
these low densities require large , evolved nebulae , in agreement with the observational result of @xcite .
they noted that their simple picture breaks down for high velocity pne in a low density environment . here , a rayleigh taylor ( rt ) instability develops , leading to shell fragmentation .
their 2d hydrodynamic simulations started with the nebula shell already formed but above their upper density limit for ism interaction to become apparent .
the fragmentation predicted by @xcite was discussed in more depth by @xcite and then @xcite . in these papers , dgani and soker applied theoretical results of hydrodynamic instabilities and found that the rt instability can play an important role in the shaping of the outer regions of a pn .
they suggested these rt instabilities can cause fragmentation of the bow shock with kelvin helmholtz instabilities also playing a part .
any fragmentation caused by these instabilities would only be present if the relative velocity to the ism of the central star was greater than 100 .
@xcite ( hereafter referred to as vgm ) pointed out that the pn
ism interaction had previously been studied by considering the interaction after the nebular shell had formed .
pne are formed when a slow wind ( @xmath3 ) ejected during the preceding asymptotic giant branch ( agb ) phase of evolution is swept up into a dense shell by a fast wind ( @xmath4 ) from the exposed , ionizing hot white dwarf core .
vgm performed 2d hydrodynamic simulations and found that crucially the interaction begins during the agb phase where the slow wind is shaped by the ism . the pn forms in this pre - shaped environment .
choosing a conservative relative velocity of the central star to the ism of 20 and a low density of the surrounding ism of @xmath5 @xmath2 , they discovered that the pn is brightened on the upstream side of the nebular shell .
they concluded that pn
ism interaction provides an adequate mechanism to explain the high rate of observed asymmetries in the external shells of pne .
further , stripping of mass downstream during the agb phase provides a possible solution to the problem of missing mass in pn whereby only a small fraction of the mass ejected during the agb phase is inferred to be present during the post agb phase .
observational evidence for the effect of the ism on agb wind stuctures , supporting vgm s findings was found by @xcite .
vgm found that simple hydrodynamic simulations can reveal much information regarding the pn
ism interaction . in order to investigate the interaction further
, we have developed a triple - wind model including an initial slow agb wind , a subsequent fast post agb wind , and a third continuous wind reflecting the movement through the ism . employing a parallel 3d hydrodynamic scheme developed by @xcite ,
we have understood the formation of the extreme pn sh 2 - 188 @xcite and the structure around the agb star r hya @xcite .
sh 2 - 188 was thought to be a bright one - sided arc - like pn when new observations taken as part of the isaac newton group photometric h@xmath6 survey of the northern galactic plane ( iphas ) @xcite revealed a faint ring - like completion of the arc and a tail stretching away in opposition to the bright arc .
our model revealed this pn to be a strong pn
ism interaction where the central star is moving at 125 in the direction of the bright arc relative to the ism and the nebular shell is interacting with a bow shock formed during the agb phase between the slow wind and the ism .
recent ir observations of the agb star r hya as part of the miriad programme @xcite have revealed the arc - like structure to the north west of the star to be a bow shock ahead of the star @xcite .
the existence of this agb wind bow shock has confirmed the hypothesis that the major shaping effect for the pn
ism interaction occurs during the agb phase of evolution . using our model , we have now run a comprehensive set of 92 simulations equivalent to over 5 years of single cpu computation time investigating the pn
ism interaction . in this paper , we discuss a representative set of these simulations and generalise the interaction into four distinct stages , indicating its effect on shaping and other pn characteristics .
we apply our generalisation to a selection of pne from the iac morphological catalog of northern galactic planetary nebulae @xcite and find indications of interaction with the ism in approximately 20 per cent of objects .
the numerical scheme , cubempi , used in our simulations to solve the hydrodynamics equations employs a second - order godunov solution due to @xcite . in recent years
, variants of cubempi have been used to shed light on nova explosions @xcite , extragalactic jet - cloud interactions @xcite and most recently pn @xcite .
the version of the scheme used here is posed in 3d cartesian coordinates , fully parallel and includes the effect of radiative cooling due to the cooling curves of @xcite above 10@xmath7 k as the cooling curves extend no further .
the parallelisation was developed using the mpi library and involves slicing the numerical domain along an axis and communicating relevant boundary data at the correct points during a computational timestep .
the parallelisation has been successfully tested for efficiency and scalability .
the scheme itself has also been tested on a number of standard computational fluid dynamics problems and performed well on all tests .
it has been further tested using astrophysical problems which have highlighted its capability for shock capturing , an important requirement for the modelling of pne .
the parallelisation and testing are detailed in @xcite the numerical domain consists of a cubic grid , 200 cells along each axis , containing a total of @xmath8 uniformly - spaced cells .
the grid is sliced ten times for parallelisation purposes and distributed across ten processors of the cobra beowulf - type supercomputer at jodrell bank .
this level of parallel distribution is a comprimise between computational efficiency and availability of computational resources . in the triple - wind model
, the simulation is performed in the frame of reference of the star , which is placed at cell coordinates ( @xmath9 ) .
mass loss is effected by setting the values of hydrodynamic variables ( density , momentum density and energy density ) in a volume - weighted spherical region of radius 5@xmath10 cells centred on the star .
the radius of this source volume has been chosen by an experimental process balancing production of the most spherical pn when modelling the stationary isw model with spherically symmetric winds ( i.e. reducing the pixelation of the cartesian grid ) and avoidance of interference with results of the simulation .
the conditions within the source volume are reset at the beginning of every computational timestep to drive the ejection of the wind .
the wind has been modelled with a spherically symmetric constant mass - loss rate @xmath11 with constant velocity @xmath12 and temperature @xmath13 .
density in the source volume has been defined by @xmath14 where @xmath15 is the physical radial distance from the central star .
the other hydrodynamic variables are set accordingly .
simulation of movement through the ism is achieved by flowing ism material in at the ( @xmath16 ) boundary with a velocity vector v@xmath17 , v@xmath18 , v@xmath19 = ( @xmath20,0,0 ) .
the ism density and temperature are constant .
all other numerical boundaries have conditions allowing material to flow out of the domain freely . gas pressures in the model are calculated assuming an ideal gas equation of state .
in our simulations , we have held the following parameter values constant : for the slow agb wind a velocity of @xmath21 and a temperature of @xmath22 k as the cooling curves extend no further ; for the fast post agb wind a mass - loss rate of @xmath23 , a velocity of @xmath24 and a temperature of @xmath25 k ; and for the ism a temperature of @xmath26 k , characteristic of the warm intercloud medium @xcite .
these parameter values are typical of observations of pn conditions . in the model ,
the switch between the agb wind and the post agb wind is instantaneous and occurs after @xmath27 years of agb evolution , typical of the duration of this phase . in view of the still considerable uncertainties on the detailed properties and evolution of these winds
, more detailed temporal variations have not been modelled .
we have discussed this further in later sections . in our simulations , we have varied three parameters : relative velocity of the star to the surrounding ism , slow wind mass - loss rate @xmath28 and ism density @xmath29 .
@xcite discussed of stars in the galactic plane : typical thin disk stars resulting in a pn appear to have a transverse variation on their galactic rotation velocity of 35 50 ; thick disk stars , rarer in the galactic plane , have a typical range of 50 75 ; halo objects can have a typical variation greater than 100 .
older stellar groups , such as pne , are characterized by larger velocity dispersions and asymmetric drift velocities than are younger stellar groups .
the average transverse motions of stars in the solar neighbourhood have been found to be in the range of 2040 with the tail of the distribution up to 130 @xcite . for white dwarfs
an average velocity of @xmath30 has been found by @xcite . in terms of galactic disk pne , @xcite state 60 as an average velocity . for galactic bulge pne
, velocities up to 200 could be reasonably expected as these stars are dynamically a much hotter population .
therefore , we have considered a range of from 0 , testing the implementation of the triple - wind model and its ability to simulate a spherical nebula , up to 200 , in 25 steps in order to fully cover the range of velocities of pn - forming stars in the galaxy .
the typical density of the ism in the galactic plane is approximately @xmath31 = @xmath32 @xmath2 up to a scale height of 100 pc above the plane where it then begins to drops off exponentially .
we have used three constant values of ism density @xmath33 = 2 , 0.1 & 0.01 @xmath2 to investigate this range .
these densities are comparable to that of the warm intercloud medium @xcite .
mass - loss rates during the agb phase of evolution vary between 10@xmath34
10@xmath35 with brief periods of enhanced mass - loss up to 10@xmath36 .
as we are using a constant mass - loss rate , we have not modelled mass - loss variations .
we have used values of 10@xmath34 , @xmath37 , 10@xmath35 & @xmath38 up to = 75 . above this velocity
, we have not used a mass - loss rate of 10@xmath35 due to time and computational constraints .
in our initial test cases is set to zero .
the forward shock driven by the agb wind is spherical and expands into a homogeneous ism . the shocked ism material behind the forward shock is of relatively low density with the contact discontinuity and the reverse shock expanding spherically symmetrically outward .
the unshocked agb wind behind the reverse shock is also spherically symmetric around the central star with an approximate 1/r@xmath39 distribution .
after 500000 years , the fast post agb wind is introduced into this distribution driving a shock which sweeps up the agb wind material into an expanding shell .
the shell has a relatively high density and a temperature on the scale of 1020000 k. the stromgren sphere around the central star encompasses the shell indicating the material in the shell would be ionized and observable as a pn .
this method of pn formation is the result of interacting stellar winds and is the premise of the isw model .
next , we added the ism velocity .
we display and discuss a representative set of five simulations covering various parameter values which serve to illustrate the range of pn structures formed in all our simulations .
table [ parameters ] shows the parameter values for these five simulations .
the resulting pn structure is discussed for each case . in our online supplementary appendix ,
we have also included snapshots at the end of the agb phase of evolution and parameter values for our full set of simulations .
we have performed calculations to find the extent of the stromgren sphere during the pn phase and find that in all cases it extends beyond the simulation domain .
.input parameters for the five selected simulations discussed in section 4 .
column a ) gives the simulation reference ; column b ) ; column c ) the density of the surrounding ism in @xmath40 ; column d ) the constant mass - loss rate during the agb phase of evolution ; and column e ) the physical size of the grid along one dimension of the numerical domain . [ cols="<,^,^,^,^ " , ] | interaction with the interstellar medium ( ism ) can not be ignored in understanding planetary nebula ( pn ) evolution and shaping . in an effort to understand the range of shapes observed in the outer envelopes of pne
, we have run a comprehensive set of three - dimensional hydrodynamic simulations , from the beginning of the asymptotic giant branch ( agb ) superwind phase until the end of the post
agb / pn phase .
a triple - wind model is used , including a slow agb wind , fast post agb wind and third wind reflecting the linear movement through the ism .
a wide range of stellar velocities , mass - loss rates and ism densities have been considered .
we find ism interaction strongly affects outer pn structures , with the dominant shaping occuring during the agb phase .
the simulations predict four stages of pn
ism interaction whereby the pn is initially unaffected ( 1 ) , then limb - brightened in the direction of motion ( 2 ) , then distorted with the star moving away from the geometric centre ( 3 ) and finally so distorted that the object is no longer recognisable as a pn and may not be classed as such ( 4 ) .
parsec - size shells around pn are predicted to be common .
the structure and brightness of ancient pne is largely determined by the ism interaction , caused by rebrightening during the second stage ; this effect may address the current discrepancies in galactic pn abundance .
the majority of pne will have tail structures .
evidence for strong interaction is found for all known planetary nebulae in globular clusters .
[ firstpage ] hydrodynamics planetary nebulae : general stars : agb and post - agb ism : structure stars : mass - loss . |
at quark matter 2015 , star has presented results from its recently installed specialized detectors : 1 ) the heavy flavor tracker ( hft ) , which facilitates the topological reconstruction of heavy flavor hadrons @xcite and 2 ) the muon telescope detector ( mtd ) which enables the identification of muons and consequently the study of quarkonia via the di - muon channel .
the hft has enabled , for the first time , the measurement of @xmath2 azimuthal anisotropy at rhic , a measurement that will help constrain the qgp transport coefficients .
the first quarknonium measurements via the di - muon channel using the mtd have also been presented .
the rhic beam energy scan phase - i has recently been completed with au+au data taken at @xmath3 gev which filled the previous 100 mev gap in baryon chemical potential ( bes - i covers @xmath4 mev ) .
directed , elliptic and triangular flow together with spectra , nuclear modification factors and higher - moments measurements from the full datasets of bes - i have been shown at this conference .
the large masses of heavy quarks ( charm and bottom ) endow them with unique properties that make them a calibrated probe of the qgp medium created in heavy - ion collisions .
their masses are larger than all the relevant medium and qcd scales ( @xmath6 , @xmath7 , t ) , furthermore , their masses are mostly external to qcd , thus are not modified by their presence in the medium in a scenario where the qcd condensate melts @xcite .
their large masses also make them amenable to brownian motion theoretical - treatment in the qgp medium @xcite .
@xmath5 @xmath8 and @xmath9 are especially sensitive to the medium dynamics and thus have the potential to experimentally constrain the qgp transport coefficients @xcite .
( red open symbols are previous results without hft @xcite , black filled symbols are the new measurements using hft ) and pions ( blue open symbols ) @xcite in central au+au collisions at @xmath10 gev .
( middle ) @xmath9 of @xmath5 ( filled symbols ) compared to that of @xmath11 ( open symbols ) @xcite in minimum - bias collisions at @xmath10gev .
( right ) @xmath9 of @xmath5 compared to tamu model calculations with ( blue ) and without ( green ) charm diffusion @xcite.,title="fig:",scaledwidth=32.0% ] ( red open symbols are previous results without hft @xcite , black filled symbols are the new measurements using hft ) and pions ( blue open symbols ) @xcite in central au+au collisions at @xmath10 gev .
( middle ) @xmath9 of @xmath5 ( filled symbols ) compared to that of @xmath11 ( open symbols ) @xcite in minimum - bias collisions at @xmath10gev .
( right ) @xmath9 of @xmath5 compared to tamu model calculations with ( blue ) and without ( green ) charm diffusion @xcite.,title="fig:",scaledwidth=32.0% ] ( red open symbols are previous results without hft @xcite , black filled symbols are the new measurements using hft ) and pions ( blue open symbols ) @xcite in central au+au collisions at @xmath10 gev .
( middle ) @xmath9 of @xmath5 ( filled symbols ) compared to that of @xmath11 ( open symbols ) @xcite in minimum - bias collisions at @xmath10gev .
( right ) @xmath9 of @xmath5 compared to tamu model calculations with ( blue ) and without ( green ) charm diffusion @xcite.,title="fig:",scaledwidth=32.0% ] figure [ fig : d0_data ] left shows star measurements of @xmath2 @xmath8 @xcite , the open red circles show the measurement before the completion of the hft @xcite , solid black circles show the recent measurement with the hft .
the hft clearly helped reduce the uncertainties in au+au data especially in the transition region between the enhancement region at low-@xmath12 and the suppression region at high-@xmath12 .
the current uncertainties from the @xmath13+@xmath13 baseline are expected to be greatly reduced with the star data collected in year 2015 with the hft in place .
the enhancement in the low-@xmath12 region is understood as a manifestation of charm coalescence with a flowing medium while the suppression at high-@xmath12 is , according to model calculations , an interplay of elastic collisional and radiative charm quark energy loss in the medium .
while the suppression of @xmath2 is , within the current experimental uncertainties , similar to that of light hadrons at high-@xmath12 , radiative energy loss alone seems to be sufficient to explain the energy loss of light quarks @xcite .
figure [ fig : d0_data ] middle shows the first measurement of @xmath2 @xmath9 at rhic top energy @xcite .
finite @xmath9 is observed for @xmath14 gev / c that is systematically lower than those of lighter hadrons ( @xmath15 gev / c ) .
figure [ fig : d0_data ] right shows a comparison of @xmath2 @xmath9 to tamu calculation with and without charm diffusion @xcite , the data are more compatible with charm diffusion in the medium , establishing a strong experimental evidence for charm quark exhibition of collectivity with the medium at rhic top energy .
the ordering , which is not explained by hydro mass ordering , might be hinting at an incomplete thermalization of charm quarks with the medium .
higher precision data from rhic run 2016 will help us investigate this .
figure [ fig : d0_theory ] left and middle panels show a simultaneous comparison of @xmath2 @xmath8 and @xmath9 to models that incorporate charm diffusion in the medium @xcite . apart from the small tension in the duke model with the @xmath9 data ,
the models show a good description of star @xmath2 data , the third panel shows calculations of charm spatial diffusion coefficient , as indicated by the yellow band .
star data are compatible with models that use @xmath16 of @xmath17 demonstrating a strong charm medium interaction that could possibly be at the quantum limit .
the inferred range is also consistent with lattice qcd calculations .
while this comparison demonstrates the constraining power of the data and the current status of theoretical models , more systematics studies are imperative to constrain the evolving bulk parameters .
variations in these parameters have been shown to generate changes by a factor as large as @xmath18 in the heavy quark observables @xcite .
[ sec : d0 ] nuclear modification factor ( left ) and @xmath9 ( middle ) to model calculations @xcite .
( right ) spatial diffusion coefficient of charm quarks ( @xmath19 ) from various calculations and d mesons ( @xmath20 ) in hadronic matter from hadron resonance gas model , in units of the thermal wavelength @xcite .
the right box shows the range of charm diffusion coefficient with which the models have an adequate description of star s
@xmath5 @xmath8 and @xmath21 data.,title="fig:",scaledwidth=31.0% ] nuclear modification factor ( left ) and @xmath9 ( middle ) to model calculations @xcite .
( right ) spatial diffusion coefficient of charm quarks ( @xmath19 ) from various calculations and d mesons ( @xmath20 ) in hadronic matter from hadron resonance gas model , in units of the thermal wavelength @xcite .
the right box shows the range of charm diffusion coefficient with which the models have an adequate description of star s
@xmath5 @xmath8 and @xmath21 data.,title="fig:",scaledwidth=31.0% ] nuclear modification factor ( left ) and @xmath9 ( middle ) to model calculations @xcite .
( right ) spatial diffusion coefficient of charm quarks ( @xmath19 ) from various calculations and d mesons ( @xmath20 ) in hadronic matter from hadron resonance gas model , in units of the thermal wavelength @xcite .
the right box shows the range of charm diffusion coefficient with which the models have an adequate description of star s
@xmath5 @xmath8 and @xmath21 data.,title="fig:",scaledwidth=33.0% ] there are several @xmath22 production models ( eg : color octet model , color evaporation model , non - relativistic qcd ) that correctly describe the production yields in elementary particle collisions . however , the actual production mechanism is still unknown .
star continues to carry differential measurements to help distinguish between these models @xcite .
figure [ fig : jpsi_upsilon ] left shows @xmath22 polarization parameter , @xmath23 , as a function of @xmath24 in the helicity frame , at different collision energies .
common trend towards strong negative values with increasing @xmath25 values is observed .
figure [ fig : jpsi_upsilon ] middle shows the @xmath8 of @xmath22 in au+au collisions at @xmath10 gev via the di - electron @xcite and the di - muon channels .
the suppression ( @xmath26 gev / c ) and the rising trend with @xmath12 are understood to be an interplay between dissociation and recombination effects at low @xmath12 and the longer formation at high @xmath12 @xcite .
the mtd has been built to study and quantify the sequential melting of @xmath27 states in heavy - ion collisions .
figure [ fig : jpsi_upsilon ] right shows the projection for the statistical uncertainty on @xmath28 using the 14.2 @xmath29 which were sampled with the mtd in rhic year 2014 run . a proof of principle analysis from 30% of the collected statistics
is shown as a black star on the figure .
the full data production is underway .
polarization parameter , @xmath23 , vs @xmath24 in the helicity frame .
( middle ) j/@xmath30 nuclear modification factor via di - electron ( open symbols ) and di - muon ( filled symbols ) channels as a function of @xmath12 .
( right ) measurement of @xmath28 in different collision systems at different energies @xcite , also shown ( red star ) is the projection for measurement at star using 14.2 @xmath29 au+au data collected with the mtd in year 2014.,title="fig:",scaledwidth=35.0% ] polarization parameter , @xmath23 , vs @xmath24 in the helicity frame .
( middle ) j/@xmath30 nuclear modification factor via di - electron ( open symbols ) and di - muon ( filled symbols ) channels as a function of @xmath12 . ( right ) measurement of @xmath28 in different collision systems at different energies @xcite , also shown ( red star ) is the projection for measurement at star using 14.2 @xmath29 au+au data collected with the mtd in year 2014.,title="fig:",scaledwidth=27.0% ] polarization parameter , @xmath23 , vs @xmath24 in the helicity frame .
( middle ) j/@xmath30 nuclear modification factor via di - electron ( open symbols ) and di - muon ( filled symbols ) channels as a function of @xmath12 .
( right ) measurement of @xmath28 in different collision systems at different energies @xcite , also shown ( red star ) is the projection for measurement at star using 14.2 @xmath29 au+au data collected with the mtd in year 2014.,title="fig:",scaledwidth=35.0% ] fully reconstructed jets are one of the main tools to constrain the medium transport parameters and it is , thusly , an active area of research for current and planned future experiments .
full jet reconstruction in au+au collisions at @xmath10 gev have been carried out at star using the anti-@xmath31 algorithm ( r = 0.2 to 0.5 ) with a low ir - cutoff ( @xmath32 gev / c ) @xcite . to address the challenge posed by the large soft and inhomogeneous underlying event background in measuring jet yields star measures the semi - inclusive distributions of reconstructed charged particle jets recoiling from a high @xmath33 trigger hadron @xmath34 where @xmath35 is the cross - section to generate a trigger hadron
, @xmath36 is the cross - section for coincidence production of a trigger hadron and a recoil jet .
a mixed - event technique has been developed and used to treat the uncorrelated background .
figure [ fig : jets ] , upper panels , show the fully corrected semi - inclusive recoil jet distributions for central and peripheral events after unfolding @xmath12-smearing and instrumental effects .
results for cone radii r = 0.3 and 0.5 are shown for @xmath37 gev / c .
the ratios of recoil jet distributions in central to that in peripheral collisions , @xmath38 , in lower panels show a clear suppression of yields in central events for @xmath39 gev / c for both cone radii .
the horizontal shifts in the range where the @xmath38 is flat ( @xmath40 ) are @xmath41 gev / c and @xmath42 gev / c for r = 0.3 and r = 0.5 , respectively .
the suppression and shift , and their reduction for larger cones might indicate an out - of - cone energy transport .
an open question is whether at jet-@xmath43 scale the medium is effectively continuous or composed of quasi - particles . for this
, star has also studied the azimuthal distribution of jet axes with respect to the recoil trigger hadron in search for evidence of large angle scattering , molire scattering @xcite , whose presence would indicate hard scattering off quasi - particles .
the study shows so far no evidence of molre scattering with the current statistical precision . , the ratio of central to peripheral yields .
( left ) cone size @xmath44 .
( right ) cone size @xmath45.,title="fig:",scaledwidth=30.0% ] , the ratio of central to peripheral yields .
( left ) cone size @xmath44 .
( right ) cone size @xmath45.,title="fig:",scaledwidth=30.0% ]
the excess in the yields of di - leptons at low mass ( @xmath46 gev/@xmath47 ) has been long linked to chiral dynamics in heavy - ion collisions .
star has recently published its acceptance - corrected measurement of di - electron production in au+au collisions at @xmath10 and @xmath48 gev @xcite .
new measurement of di - electron production in u+u at @xmath49 gev , a system with an expected 20% increase in energy density @xcite and consequently a longer lifetime , have been presented at this conference @xcite .
figure [ fig : dielectrons ] left shows the acceptance - corrected excess di - electron invariant - mass spectra normalized by @xmath50 in minimum - bias au+au collisions at @xmath51 and @xmath52 gev @xcite and u+u collisions at @xmath49 gev .
models with collisional broadening of @xmath53-mesons @xcite describe the data across all energies , centralities , and @xmath33 @xcite . figure [ fig : dielectrons ] right shows the integrated di - electron excess yields ( @xmath54 gev / c@xmath55 ) normalized by and as function of @xmath50 .
the yields show an increasing trend from peripheral to central collisions and from lower to higher energies , which is predicted to be the case if the emission is correlated with the medium lifetime .
the measurements are also compared to theoretical calculations of medium lifetime @xcite . in minimum - bias au+au collisions at @xmath51 and @xmath52 gev @xcite and u+u collisions at @xmath49 gev .
model calculations ( red solid ) with a broadened @xmath53 spectral function which is the sum of qgp radiation contribution and in hadronic medium broadening @xcite is shown .
( right ) integrated di - electron excess invariant - mass spectra ( @xmath54 gev / c ) normalized by @xmath50 in different collision species and at different collision energies @xcite compared to the theoretical calculations of the fireball lifetimes @xcite.,title="fig:",scaledwidth=40.0% ] in minimum - bias au+au collisions at @xmath51 and @xmath52 gev @xcite and u+u collisions at @xmath49 gev .
model calculations ( red solid ) with a broadened @xmath53 spectral function which is the sum of qgp radiation contribution and in hadronic medium broadening @xcite is shown .
( right ) integrated di - electron excess invariant - mass spectra ( @xmath54 gev / c ) normalized by @xmath50 in different collision species and at different collision energies @xcite compared to the theoretical calculations of the fireball lifetimes @xcite.,title="fig:",scaledwidth=40.0% ] the geometry of non - central collisions is expected to create strong magnetic fields at the early stages of heavy - ion collisions which if coupled to a local chirality imbalance gives rise to electric charge separation , a phenomenon known as chiral magnetic effect ( cme ) .
the same geometrical configuration also creates an angular momentum that gives rise to fluid vorticity which , in turn , if coupled to local chirality imbalance creates baryonic charge separation which is known as chiral vortical effect ( cve ) .
star uses a three point correlator ( @xmath56 ) to search for signatures of cve and cme .
correlations between different particle species have different sensitivities to these effects ( see fig .
[ fig : cve ] legend ) .
figure [ fig : cve ] left shows the difference of the correlator between opposite sign and same sign particles @xcite .
a clear hierarchy is observed , while this hierarchical structure and its ordering meets the expectation of these chiral effects more studies of possible experimental backgrounds are needed and underway .
one such study has been presented at this conference @xcite .
figure[fig : cve ] right shows the charge separation parameter for non - central events selected with observed @xmath57 where the @xmath9 related background is eliminated @xcite .
the charge separation is slightly positive on average , but exhibits an oscillating dependence on the third harmonic plane relative to the second harmonic plane .
this suggests additional physics background related to the event shape and further investigations are underway . .
( right ) charge separation parameter for non - central events with @xmath57.,title="fig:",scaledwidth=34.0% ] .
( right ) charge separation parameter for non - central events with @xmath57.,title="fig:",scaledwidth=27.0% ]
high-@xmath12 hadron suppression is considered to be a key evidence of the formation of sqgp in heavy - ion collisions .
it is only natural to study this observable in the bes to look for the onset energy at which the strong suppression starts to appear .
figure [ fig : rcp ] left shows charged hadron @xmath58 at different bes - i energies @xcite .
@xmath58 exhibits a smooth transition from strong suppression at high energies to enhancement at lower beam energies .
it is well known , however , that the so - called cronin effect , which gives rise to @xmath33 elongated production enhancement , is more prominent at lower beam energies @xcite , which complicates the interpretation of this observation .
the complication can perhaps be resolved with a p+a bes to disentangle the cnm enhancement and qgp suppression effects .
the enhancement and suppression effects are proportional to centrality , so a study of the centrality dependence of the charged hadron yields is expected to bring insight into the interplay of these effects .
figure [ fig : rcp ] right shows integrated yields ( @xmath59 gev ) per number of binary collisions as a function of number of participants at different collision energies .
the yields are normalized such that the yield in most peripheral bin is unity .
it is observed that at high beam energies suppression effects quickly dominate . at lowest beam energies ,
however , the yields in most central collisions are highly enhanced compared to most peripheral collisions . at @xmath3 gev an approximate flatness across centralities is observed , an apparent equilibration of enhancement and suppression effects at this energy and this @xmath12 region .
theoretical calculations of the beam energy and impact parameter dependence could help understand the dynamics at play here and possibly uncover the suppression effects from data .
in au+au collisions at different energies .
( right ) integrated yields ( @xmath60 gev ) per number of binary collisions as a function of number of participants at different collision energies .
the yields are normalized such that the yield in most peripheral bin is unity.,scaledwidth=70.0% ] triangular flow is argued to be almost directly proportional to the duration of the low - viscosity phase in heavy ion collisions @xcite .
figure [ fig : v3 ] left shows star measurements of charged hadron @xmath61 in au+au collisions at @xmath62 gev @xcite together with alice measurement in pb+pb collisions at @xmath63 tev @xcite .
the data show that sizable @xmath64 in central to mid - central collisions persists all the way down to the lowest beam energy inviting the question about the existence of a qgp phase even at those low energies .
peripheral collisions @xmath64 is consistent with zero for @xmath65 gev .
figure [ fig : v3 ] right shows @xmath61 scaled by the multiplicity per participant pair ( @xmath66 ) as a function of collision energy , an observable that is proportional to the system energy density . the data exhibit a flat trend for energies @xmath67 gev in most central collisions
, it is another open question whether this could be related to a softening of the equation of state . in au+au collisions from star and in pb+pb collisions from alice @xcite as a function of collision energy .
( right ) @xmath68 normalized by the multiplicity per participant pair ( @xmath66 as a function of collision energy.,title="fig:",scaledwidth=40.0% ] in au+au collisions from star and in pb+pb collisions from alice @xcite as a function of collision energy .
( right ) @xmath68 normalized by the multiplicity per participant pair ( @xmath66 as a function of collision energy.,title="fig:",scaledwidth=40.0% ] the dip in star measurement of net - proton slope of directed flow at midrapidity @xcite , @xmath69 , is due to an interplay between hydro , baryon transport dynamics and baryon / anti - baryon annihilation , so more understanding of these phenomena is needed as to whether a first order phase transition is indicated by the data .
so far all models failed to reproduce this dip structure .
star has recently measured directed flow of @xmath13 , @xmath70 , @xmath71 , @xmath72 , @xmath73 , @xmath74 and @xmath75 in au+au collisions at the bes - i energies @xcite .
figure [ fig : v1 ] left shows net - proton and net - kaon @xmath69 .
it is evident that the data for net - kaon and net - proton coincide from @xmath10 gev all the way down to @xmath76 gev and then starkly split at the lowest two energies .
more theoretical insight is needed to understand if this observation is related to the nature of the phase transition at these low energies . in au+au collisions at different collision energies .
( right ) net - protons @xmath77 in au+au collisions as a function of collision energy compared to urqmd calculations ( hashed area ) and poisson predictions ( dashed line ) in 0 - 5% ( stars ) and 70 - 80% ( circles ) centralities.,title="fig:",scaledwidth=40.0% ] in au+au collisions at different collision energies .
( right ) net - protons @xmath77 in au+au collisions as a function of collision energy compared to urqmd calculations ( hashed area ) and poisson predictions ( dashed line ) in 0 - 5% ( stars ) and 70 - 80% ( circles ) centralities.,title="fig:",scaledwidth=46.0% ] higher moments of conserved quantum numbers ( charge , strangeness and baryon number ) are expected to be sensitive to the proximity of the system to a critical point ( see @xcite and references therein ) .
it is also expected that higher order moments are more sensitive to criticality .
figure [ fig : v3 ] right shows net - proton @xmath77 in most central and peripheral collisions compared to urqmd and poisson expectations .
the measurement phase space has been recently extended from @xmath78 to @xmath79 gev / c .
the data in central collisions clearly show a non - monotonic behavior deviating from the poisson expectation of unity and that is absent from both peripheral and urqmd expectations . more work is being carried out to check if this experimental observable is sufficient to establish the existence of a critical point , the role of the finite volume size in heavy - ion collisions and the random system evolution path in the phase diagram @xcite .
it is currently agreed upon that more systematic studies of higher moments and their phase space evolution are needed from the experiment side .
this is one of the driving measurements behind rhic plan for bes - ii ( 2019 - 2020 ) .
star inner tpc upgrade will extend its coverage from @xmath80 to @xmath81 .
in addition , an event plane detector ( epd ) upgrade at forward rapidities will provide more control on the measurement systematics by giving a tpc - independent measurement of centrality and reaction plane .
star has recently boosted its heavy flavor program with its high resolution heavy flavor tracker ( hft ) and muon telescope detector ( mtd ) . @xmath5
@xmath9 measurement with the hft indicates that charm flows at rhic top energy .
first results of quarkonium suppression from mtd have been shown at this conference and more data on @xmath27 states melting at rhic top energy from full mtd datasets production are underway .
star bes - i data exhibit non - monotonicity in net - proton @xmath77 , baryon @xmath82 and charged hadron scaled-@xmath61 at the same energy range , trends that could possibly be connected to the nature of phase transition at lower beam energies and the location of the critical point .
00 g. contin , nucl .
a , these proceedings .
b. mller , nucl .
a750 ( 2005 ) 84 .
r. rapp , h. van hees , r. c. hwa , x .-
n . wang ( ed . ) , quark gluon plasma 4 , world scientific , p. 111
a. andronic et al . , arxiv:1506.03981 .
l. adamczyk et al . ,
star collaboration , phys .
( 2014 ) 142301 .
abelev et al . ,
star collaboration , phys .
lett . b655 ( 2007 ) 104 .
abelev et al . ,
star collaboration , phys .
c77 ( 2008 ) 54901 .
m. he , r. j. fries , r. rapp , phys .
c86 ( 2012 ) 014903 , and private communications .
g. xie , for the star collaboration , nucl .
a , these proceedings .
m. lomnitz , for the star collaboration , nucl .
a , these proceedings .
m. nahrgang , j. aichelin , s. bass , p. b. gossiaux , k. werner , phys .
c91 ( 2015 ) 014904 , and private communications .
s. cao , g .- y .
qin , s. a. bass , phys .
c88 ( 2013 ) 044907 , and private communication .
p. b. gossiaux , s. vogel , h. van hees , j. aichelin , r. rapp , m. he , m. bluhm , arxiv:1102.1114 .
m. he , r. j. fries , r. rapp , phys .
( 2013 ) 112301 .
h. ding , a. francis , o. kaczmarek , f. karsch , h. satz , et al .
d86 ( 2012 ) 014509 .
d. banerjee , s. datta , r. gavai , and p. majumdar , phys .
d85 ( 2012 ) 014510 .
o. kaczmarek , nucl .
a931 ( 2014 ) 633 .
f. riek and r. rapp , phys .
c82 ( 2010 ) 035201 .
k. huggins and r. rapp , nucl .
a896 ( 2012 ) 24 .
b. svetitsky , phys .
d37 ( 1988 ) 2484 .
r. ma , for the star collaboration , nucl . phys .
a , these proceedings .
s. chatrchyan et al .
, cms collaboration , phys .
lett . 109 ( 2012 ) 222301 .
l. adamczyk et al . ,
star collaboration , phys .
b735 ( 2014 ) 127 .
s. yang , for the star collaboration , nucl .
a , these proceedings .
d. kikoa , g. odyniec , r. vogt , phys .
c84 ( 2011 ) 054907 .
l. adamczyk et al . ,
star collaboration , phys .
b750 ( 2015 ) 64 and phys .
c92 ( 2015 ) 024912 .
r. rapp , phys .
c63 ( 2001 ) 054907 and r. rapp , h. van hees , arxiv:1411.4612 .
h. van hees , r. rapp , nucl .
a806 ( 2008 ) 229 .
r. arnaldi et al .
, na60 collaboration , eur .
j. c59 ( 2009 ) 607 .
i. vitev , m. gyulassy , phys .
( 2002 ) 252301 .
k. aamodt et al .
, alice collaboration , phys .
( 2011 ) 032301 .
p. jacobs , a. schmah , for the star collaboration , nucl .
a , these proceedings .
b. trzeciak , for the star collaboration , nucl .
a , these proceedings .
l. adamczyk et al . ,
star collaboration , phys .
b722 ( 2013 ) 55 and phys .
c90 ( 2014 ) 024906 .
l. wen , for the star collaboration , poster ( 948 ) at qm15 .
b. tu , n. n. ajitanand , for the star collaboration , poster ( 123 ) at qm15 .
l. adamczyk et al . ,
star collaboration , phys .
c89 ( 2014 ) 044908 .
s. horvat , for the star collaboration , nucl .
a , these proceedings .
j. aunvine , h. petersen , phys .
c88 ( 2013 ) 064908 . l. song , for the star collaboration , nucl .
a , these proceedings .
l. adamczyk , star collaboration , phys .
( 2014 ) 162301 .
p. shanmuganathan , for the star collaboration , nucl .
a , these proceedings .
j. thder , for the star collaboration , nucl .
a , these proceedings . | the solenoidal tracker at rhic ( star ) experiment utilizes its excellent mid - rapidity tracking and particle identification capabilities to study the emergent properties of quantum chromodynamics ( qcd ) .
the star heavy - ion program at vanishingly small baryon density is aimed to address questions about the quantitative properties of the strongly - interacting quark gluon plasma ( qgp ) matter created in high energy collisions ( @xmath0 , @xmath1 , chirality , transport parameters , heavy quark diffusion coefficients ... ) . at finite baryon density ,
the questions concern the phases of nuclear matter ( the qcd phase diagram ) and the nature of the phase transition , namely : what is the onset collision energy for the formation of qgp ?
what is the nature of phase transition in heavy - ion collisions ?
are there two phase transition regions ?
if yes , where is the critical point situated ? at quark matter 2015 , the star collaboration has presented a wealth of new experimental results which address these questions . in these proceedings
i highlight a few of those results .
quark gluon plasma ( qgp ) , heavy flavor , heavy quark diffusion , elliptic flow , nuclear modification factor , silicon pixel detector , jets , phase transition , critical point , chirality |
the gonihedric ising model was defined by savvidy and wegner @xcite as a cubic lattice transcription of an earlier gonihedric random surface model developed by savvidy @xcite which was intended to be a discretization of string theory . in the case of the gonihedric ising model ,
the spin cluster boundaries were used to model a gas of surfaces and the spin couplings were chosen so that edges and intersections of plaquettes rather than their surface areas were weighted , which was also the defining feature of the gonihedric random surface model .
such a weighting can be arranged using a one - parameter family of generalized @xmath0 ising models with fine - tuned nearest neighbour @xmath5 , next to nearest neighbour @xmath6 and plaquette interactions @xmath7 $ ] .
the hamiltonians in this family are given by @xmath8}\sigma_{i } \sigma_{j}\sigma_{k } \sigma_{l } \ ; , \ ] ] where the parameter @xmath9 gives the relative weight of two plaquettes meeting at right angles and four - plaquette intersections in the spin cluster boundaries .
one member of this family , the purely plaquette action with @xmath10 @xmath11}\sigma_{i } \sigma_{j}\sigma_{k } \sigma_{l}\ ] ] has merited particular attention due to its interesting properties , both static and dynamical @xcite .
it has a strong first order transition surrounded by a region of metastability and shows evidence of glass - like behaviour at lower temperatures in spite of having no quenched disorder .
the symmetry properties of both the ground states and low temperature phase of @xmath12 are unusual . the standard ising model with only nearest neighbour interactions @xmath13 displays a twofold symmetry in its ground state , whereas a @xmath0 @xmath14 gauge theory ( where the spins live on plaquette edges rather than vertices as in @xmath12 ) @xmath15 } u_{ij } u_{jk } u_{kl } u_{li}\ ] ] has the local gauge symmetry @xmath16 , with @xmath17 .
the symmetries of plaquette gonihedric model of equ.([e2k ] ) lie somewhere between these since it is possible to flip arbitrary , possibly intersecting , planes of spins in the ground state at zero energy cost . both monte - carlo simulations and low temperature expansions
@xcite confirm that this remains a symmetry of the low temperature phase , which is thus highly degenerate .
a consequence of this symmetry is that the standard magnetization is zero in the low temperature phase .
the nature of the magnetic order has thus remained unclear . in this paper
we discuss the recent suggestion by hashizume and suzuki @xcite that a suitable order parameter for @xmath12 might be constructed using nearest - neighbour spin - spin correlations @xmath3 , which was prompted by suzuki s earlier study of an _ anisotropic _ gonihedric model @xcite .
we conduct measurements of this candidate order parameter on a toy ensemble of ground state configurations constructed by flipping planes of spins starting from a ferromagnetic starting configuration to mimic the ground state structure of the gonihedric model , before moving on to simulations of both the anisotropic and isotropic gonihedric models themselves .
the study of a strongly anisotropic variant of the plaquette gonihedric hamiltonian by suzuki @xcite pre - dates the work of savvidy and wegner by some twenty years . for reasons that will become apparent , suzuki dubbed this the `` fuki - nuke '' or `` no roof '' model and showed that it was ( non - trivially ) equivalent to a stack of @xmath2 ising models .
this was later rediscovered by jonsson and savvidy @xcite using a transfer matrix approach and discussed by castelnovo _ et.al .
_ @xcite using methods more akin to suzuki s original approach . in suzuki s
work the principal concern was to obtain solvable @xmath0 spin models , and the fuki - nuke model was constructed as a specific example .
the starting point was an anisotropically coupled variant of @xmath12 h_aniso = - j_x _ [ ijkl]_i _ j_k _ l - j_y _ [ ijkl]_i _ j_k _ l - j_z _ [ ijkl]_i _ j_k _ l where @xmath18 were the couplings for plaquettes perpendicular to the @xmath19 axes respectively . setting @xmath20 and
@xmath21 gave a model where the horizontal `` roof '' plaquettes had zero coupling , hence the sobriquet of `` fuki - nuke '' . to show that this is equivalent to a stack of @xmath2 ising models requires a little further work , which can be done succinctly by using the non - linear @xmath22-@xmath23 transformation , following suzuki @xcite and castelnovo _ et.al .
we first define bond spin variables @xmath24 that can be thought of as living at the centre of ( only ) the vertical lattice bonds .
since we have condensed the site labels on both sets of spins , we write @xmath25 and @xmath26 to represent any of the spins in the @xmath27 horizontal @xmath2 layer . with this notation the direct transformation for a spin in the @xmath27 layer is _
k = _ 1 _ 2 _ 3 _ k and the inverse transformation is given by _ 1 = _ 1 , _ 2 = _ 1 _ 2 , , _
n = _ n-1 _ n in order to obtain a one - to - one correspondence between @xmath22 and @xmath23 spin configurations we need to specify the value of the @xmath28 spins on a given ( first in this case ) horizontal plane , but in all the other planes a product of two @xmath22 spins at the end of each vertical bond gives the @xmath23 spin .
the resulting hamiltonian is then h = - _ i ( _ i _ i + e_x + _ i _ i + e_y ) where @xmath29 are unit vectors in the horizontal directions .
we have not written the contribution of the fixed plane , which will vanish in the thermodynamic limit .
the @xmath30 spins are coupled to their horizontal nearest neighbours in the @xmath31 and @xmath32 directions and there are no vertical inter - plane couplings .
we have thus shown that the fuki - nuke model is equivalent to a stack of uncoupled @xmath2 ising models arranged horizontally one above each other .
the magnetic order parameter for the @xmath33th @xmath2 ising layer is given by the standard expression [ mone ] m_2d , i = which may be rewritten in terms of the original @xmath34 spins as m_2d , i = .
where @xmath35 is the unit vector in the vertical direction .
the nearest neighbour @xmath36 correlator is measured on all the vertical bonds bisected by the horizontal plane of the corresponding @xmath37 spins
. the unusual appearance of a two - spin correlator as an order parameter is simply the result of expressing the magnetization in terms of the original @xmath34 variables rather than the bond spins @xmath30 .
remembering that @xmath38 and using the standard results for the @xmath2 ising model magnetization we have @xcite [ m2d ] m_2d , i = = ( 1 - ^-4 ( ) ) ^1/8 which will behave as @xmath39 near the critical point @xmath40 .
more generally , since @xmath41 and the different @xmath37 layers are decoupled , m_2d , i , n = = ( m_2d , i ) ^n .
the in - plane horizontal correlations between `` dimers '' of @xmath34 spins , where the @xmath42 spins now lie on the same plane , = are also , of course , ising - like .
since the various @xmath2 ising layers magnetize independently we could construct a quasi-@xmath0 order parameter to be of the form [ mabs ] m_abs = _ xy planes | _ single plane _
i + e_z | or [ msq ] m_sq = _ xy planes ( _ single plane _
i + e_z ) ^2 to avoid inter - plane cancellations .
we have explicitly retained the various normalizing factors in equs.([mabs ] , [ msq ] ) for a cubic lattice with @xmath43 sites .
we could also define similar quantities for non - nearest neighbour correlators [ minabsxyz ] m_abs , n^ = _ xy planes | _ single plane _
i + n e_z | or [ minsqxyz ] m^_sq , n = _ xy planes ( _ single plane _
i + n e_z ) ^2 with @xmath44 .
since each plane magnetizes at the same point the resulting expressions are given by @xmath45 as the various other variants of the magnetization are given in terms of @xmath46 this is probably the most suitable choice of `` three - dimensional '' order parameter for the fuki - nuke model .
in addition , it emphasizes that the critical behaviour of the fuki - nuke model is really that of the @xmath2 ising model in each layer . in the anisotropic model
the orientation of the planes in the summation is fixed by the choice of zero coupling , which we have taken to be horizontal planes in our case .
the suggestion by hashizume and suzuki in @xcite is that a similar order parameter measuring coplanar , nearest neighbour spin correlations could still serve for the _ isotropic _ gonihedric model .
they considered a mean - field approach to the model and correlation function inequalities in order to corroborate this and found supporting evidence from both , as well as extending the discussion to disordered couplings .
here , we investigate the order parameter in the isotropic @xmath0 gonihedric model directly using monte - carlo simulations , restricting discussion to the purely ferromagnetic case for simplicity . in equs.([mabs],[msq ] ) we have introduced either a modulus or square for the spin correlators in order to avoid inter - plane cancellations , whereas suzuki s approach is tantamount to considering the direct equivalent of equ.([mone ] ) [ msuz ] m = _ planes _ single plane _
i + e_z in the presence of an external field ( later taken to zero ) which picks out one of the @xmath2 magnetizations on each plane .
the possibility of using an order parameter akin to the @xmath47 in equ.([msq ] ) has been mentioned previously by lipowski @xcite , who noted in a simulation of @xmath12 that it appeared to possess the correct behaviour .
he also remarked on the possibility of using non - nearest neighbour correlators as the order parameter(s ) , which we shall discuss further below . in the remainder of the paper
we compare measurements of the candidate order parameters on toy ground state configurations , the fuki - nuke model and the isotropic @xmath0 gonihedric model itself .
ground states of the gonihedric hamiltonian @xmath12 may be obtained by flipping some set of planes of spins in a reference ferromagnetic configuration which has all spins up or down .
a simple test for any candidate order parameter is then to generate an ensemble of such configurations and measure its expectation value .
a typical configuration on a finite cubic lattice generated in this manner would look something like that in fig.(1 ) .
[ cube_stripe ] we can also measure proposed order parameters on purely random ( i.e. infinite temperature ) configurations .
any that fail to distinguish between the ground states and random configurations would not , at first sight , appear to be suitable for use in a simulation of the full hamiltonian .
if the suggestion of @xcite is correct , we would expect that the order in the low temperature phase of the isotropic @xmath0 gonihedric model could be discerned using an appropriate generalization of either the @xmath48 and @xmath47 in equs.([mabs],[msq ] ) . for isotropic configurations such as that in fig.(1 ) the system should be agnostic to the orientation of the planes used in calculating the correlators , so any of the three choices of orientation for reference planes in [ mabsxyz ] m^x , y , z_abs & = & _ yz|xz|xy planes | _ single plane _
i + e_x , y , z | or [ msqxyz ] m^x , y , z_sq & = & _ yz|xz|xy planes ( _ single plane _ i _
i + e_x , y , z ) ^2 where @xmath49 are the unit vectors in the @xmath19 directions respectively , should be able to serve as order parameters . in all cases
the correlators are nearest neighbour correlators for the @xmath34 spins on bonds perpendicularly bisected by the reference planes , as in the fuki - nuke model .
reassuringly , we find that the various @xmath50 from equ.([mabsxyz ] ) and the @xmath51 from equ.([msqxyz ] ) all give @xmath52 when measured on an ensemble of flipped states such as that in fig.(1 ) so they do , indeed , pick up the planar order that is characteristic of the gonihedric models .
in addition , the @xmath50 and @xmath51 all give zero when measured on purely random configurations , so they are capable of distinguishing this order from the complete disorder at infinite temperature
. the choice of nearest neighbour correlators in defining the order parameter does not appear to be obligatory , though given the line of reasoning leading to it from the fuki - nuke model it is probably the most natural . if we measure quantities such as [ mnabsxyz ] m_abs , n^x , y , z = _ yz|xz|xy planes | _ single plane _
i + n e_x , y , z | or [ mnsqxyz ] m^x , y , z_sq , n = _ yz|xz|xy planes ( _ single plane _
i + n e_x , y , z ) ^2 with @xmath44 on the ensemble of flipped ground state configurations these also give @xmath52 ( and zero on random configurations ) .
that this should be so is clear from starting with configurations in which planes of one orientation only , say horizontal , are flipped .
for these , plane - to - plane correlations at any separation must be @xmath53 . taking a modulus or square for each planar sum
then makes all of these @xmath52 .
flipping planes of spins in the two remaining directions perpendicular to this will change the signs of lines of spins on both the horizontal planes contributing to the correlator sums and therefore not change the values of the contributing correlators . when the standard magnetization is zero m
= _ i _ i = 0 the magnetic susceptibility @xmath54 is given by [ chi ] = _ i _ j _ i _ j = _ i _ n _ i _ i+n which is related to the class of order parameters discussed here , since the sums may be arranged in a similar manner to those in @xmath47 and @xmath48 .
however , the expression for @xmath54 includes a summation over distances greater than one and also retains the signs of the various different contributions in the sum .
consequently , the susceptibility does not distinguish between configurations such as those in fig.(1 ) and random configurations , giving @xmath52 in both cases , but as we shall see below it _ does _ appear to display the characteristics of an order parameter in a simulation of the full isotropic gonihedric model .
since the anisotropic gonihedric hamiltonian , or fuki - nuke model , originally inspired the definition of the candidate order parameters it too can serve as a test case for simulations .
we know that @xmath2 ising criticality is expected in each horizontal layer , so the transition point will be seen in the continuum limit at the @xmath2 ising value of @xmath55 . in the fuki - nuke model the orientation of the the @xmath2 planes used to measure the nearest neighbour spin correlations
is fixed by the choice of the zero couplings . in our simulations
we take @xmath21 , which gives horizontal planes in the sums and means that @xmath56 should be the correct choice for the order parameter .
we simulate the fuki - nuke model on a @xmath57 lattice with periodic boundary conditions using @xmath58 metropolis monte - carlo measurement sweeps with a hot start and an equilibration time of @xmath59 sweeps as a test case . in fig.(2 ) we can see that both @xmath48 and @xmath47 give a clear signal for the pseudo - critical point , which can be observed close to the continuum value of @xmath60 already on the @xmath57 lattice via , for example , the peak in the specific heat measurements . as a consequence of cancellations between differently signed @xmath2 layers of ising spins in the magnetized phase the standard magnetization @xmath61 remains zero for all @xmath62 in the fuki - nuke model and does not give an obvious signal for the transition . and @xmath47 on a @xmath57 lattice for the fuki - nuke model.,title="fig:",height=226 ] [ aniso20 ] the preceding measurements reported for the toy ground state model and the fuki - nuke model suggest that any of @xmath50 from equ.([mabsxyz ] ) , the @xmath51 from equ.([msqxyz ] ) or even the @xmath63 and @xmath64 of equs.([mnabsxyz],[mnsqxyz ] ) might be viable candidates for an order parameter in the isotropic @xmath0 gonihedric model itself . to clarify this
, we carried out simulations of the isotropic gonihedric model on @xmath65 and @xmath66 lattices using @xmath58 metropolis monte - carlo measurement sweeps , following a hot start and an equilibration time of @xmath59 sweeps .
the various @xmath50 and @xmath51 along with @xmath63 and @xmath64 for @xmath44 were measured on both lattices . in fig.(3 )
we plot the measurements of @xmath67 and @xmath68 on the @xmath65 lattice and in fig.(4 ) on the @xmath66 lattice .
the measurements of the @xmath61 s for the different @xmath19 orientations of the reference planes are indistinguishable from each other and the observed jump in the various @xmath61 s coincides with the pseudo - critical values of @xmath62 determined from the jump in the energy observed in simulations ( @xmath69 on the @xmath65 lattice and @xmath70 on the @xmath66 lattice with `` hot '' starts ) .
these results confirm that the orientation of the reference planes is , indeed , irrelevant in measuring these magnetizations and that all three choices are equivalent .
they also show clearly that the correlators distinguish between the disordered and ordered phases of the isotropic model as suggested by the arguments in @xcite .
( upper ) and @xmath67 ( lower ) on a @xmath65 lattice for the isotropic gonihedric model .
lines are drawn to guide the eye.,title="fig:",height=226 ] [ m_10 ] ( upper ) and @xmath67 ( lower ) on a @xmath66 lattice for the isotropic gonihedric model .
again , lines are drawn to guide the eye.,title="fig:",height=226 ] [ msq_20 ] to guard against the possibility of discarding too much information by considering a simple global order parameter we can measure the distribution @xmath71 of the coplanar @xmath2 spin correlations . in figs.(5,6 ) we have combined the three orientations of reference planes to plot a @xmath72 in the high temperature and low temperature phases respectively .
the three individual orientations contributing to this give identical distributions . in fig.(5 ) @xmath72 is gaussian ( narrowing with system size and increasing temperature ) and centred at @xmath73 , whereas in the low temperature phase in fig.(6 ) at @xmath74 there are two sharp peaks at @xmath75 .
we would expect the slight asymmetry between the peaks in the plot to disappear with better statistics and increasing lattice size . the observed behaviour strongly supports the hypothesis of fuki - nuke type order , where coplanar correlators `` magnetize '' independently of the other planes , in the isotropic model too .
the plots also demonstrate that a @xmath76 $ ] type order parameter is sufficient for the isotropic gonihedric model , since there is no non - trivial structure in @xmath72 such as that seen in the overlap distribution @xmath77 of ( mean - field ) spin glasses . at @xmath78 in the high temperature phase on a @xmath65 lattice , title="fig:",height=188 ] [ h_10 ] at @xmath74 in the low temperature phase on a @xmath65 lattice , title="fig:",height=188 ] [ h_58 ]
we also find , as with the toy ground state ensemble , that @xmath79 and @xmath80 for @xmath44 show very similar behaviour to @xmath68 and @xmath67 . in fig.(7 )
@xmath81 on a @xmath66 lattice is plotted , along with @xmath82 for comparison .
@xmath81 is typical of the various @xmath83 ( and @xmath84 ) for other @xmath85 .
( upper ) and @xmath86 ( lower ) on a @xmath66 lattice for the isotropic gonihedric model.,title="fig:",height=226 ] [ mabs_15_06 ] as might be expected the magnetization constructed from the more widely spaced correlators in @xmath86 is lower in the disordered phase , but it too jumps sharply to one at the transition point just as does @xmath87 . since we employ periodic boundary conditions the measured magnetizations repeat from @xmath88 on the @xmath66 lattice as the distances in the correlators wrap around the lattice . with the isotropic model
there is no longer a mapping to the @xmath2 ising model , unlike the fuki - nuke model , so we do not have an exact expression for the planar average of nearest neighbour correlators as in equ.([m2d ] ) , nor do we have any assurance that different ( bi-)layers of spins decouple , which allows us to express everything in terms of the nearest neighbour correlators alone in the fuki - nuke case .
nonetheless , it would appear that the most fundamental correlations in the isotropic case are also those between nearest neighbour spins .
the observed signal for the transition in the higher correlators are a consequence of that in the nearest neighbour correlations , just as with the toy ground states and the fuki - nuke model .
a curious feature of the isotropic gonihedric model simulations that should be highlighted is that the standard magnetic susceptibility , as defined in equ.([chi ] ) : @xmath89 , also behaves as an effective order parameter , unlike the toy ground state ensemble .
this can be seen in fig.(8 ) where @xmath54 is @xmath52 at high temperatures and drops sharply to zero at the observed pseudo - critical points . on both the @xmath65 and @xmath66 lattices
is one at high temperatures and drops sharply to zero at the observed pseudo - critical points.,title="fig:",height=226 ] [ susp_10_20 ] we have already seen in our discussion of the toy model for ground states that @xmath54 does _ not _ distinguish between random , infinite temperature configurations and the flipped ground - state configurations , which means that the observed behaviour is not a direct consequence of @xmath54 detecting the coplanar order of the low temperature phase as with @xmath90 .
it appears rather to be a dynamical ergodicity breaking effect . at high temperatures
the flip symmetry of the hamiltonian is not broken , so @xmath54 must be one , whereas the observed value of zero for @xmath54 throughout the low temperature ( high @xmath62 ) phase is due to the system `` freezing '' into one configuration as the metropolis spin flip acceptance drops rapidly to zero at @xmath91
. it would be interesting to explore the behaviour of @xmath54 in a multicanonical simulation such as that carried out in @xcite . in this case
the system would still be free to explore multiple ground states , so the behaviour of @xmath54 would presumably be more akin to that in the toy ground state ensemble ( and less like an order parameter ) .
it is also worth noting that the strong hysteresis associated with the first order transition means that a more accurate estimate of the continuum transition point @xmath92 is best done with such a multi - canonical simulation , or similar . in @xcite
the value of @xmath93 obtained from the finite size scaling of the specific heat maxima is a little higher than that obtained from taking careful account of scaling corrections arising from the boundary conditions , @xmath94 , using a metropolis simulation @xcite .
the various pseudo - critical temperatures observed on differently sized lattices in a multi - canonical simulation fall between those observed for hot and cold starts in metropolis simulations such as those performed here on the same lattices , as one might expect .
as suzuki remarked many years ago @xcite the anisotropic gonihedric ising ( fuki - nuke ) model may be reformulated in terms of new spin variables as a stack of decoupled @xmath2 ising models which magnetize independently . when translated back into the original spin variables the magnetization of each @xmath2 ising model
is expressed in terms of a two spin nearest neighbour correlator @xmath3 . to take account of the independent ordering of planes a quasi-@xmath0 order parameter
may then be constructed by taking the modulus , @xmath48 of nearest neighbour correlators summed over each plane .
we investigated numerically the suitability of using this and related quantities as an order parameter for the isotropic @xmath0 gonihedric hamiltonian , @xmath12 , following the suggestion of @xcite .
the transition in this case is now first order and hence no longer in the universality class of of the @xmath2 ising model , but the distinctive planar flip symmetry remains .
we concluded that the @xmath50 of equ.([mabsxyz ] ) might serve as an order parameter for the isotropic model and found that the choice of reference direction @xmath19 was immaterial in this case .
the absence of non - trivial structure in the distribution @xmath72 of the coplanar magnetizations suggested that such a global parameter was sufficient to capture the order .
somewhat unexpectedly , the standard magnetic susceptibility @xmath54 also shows the characteristics of an order parameter in the simulations of the isotropic gonihedric model since it was measured to be one at high temperatures and dropped sharply to zero at the pseudo - critical point .
however , @xmath54 failed to distinguish between an ensemble of flipped toy ground states and purely random configurations .
we suggested that the behaviour in the full gonihedric simulations was a dynamical consequence of freezing and that a multicanonical simulation might help to clarify this . regarding similar order parameters in other models
, the dual gonihedric model may be written as strongly anisotropic ashkin - teller model and also possesses similar flip symmetries to the purely plaquette model @xcite .
understanding the low temperature order in this dual model , whose standard magnetization and polarization are zero at all @xmath62 , might proceed along similar lines to those considered here .
d. a. johnston would like to thank martin weigel , wolfhard janke and adam lipowski for helpful discussions .
g. k. savvidy and f.j .
wegner , nucl .
b413 * ( 1994 ) 605 .
+ g. k. savvidy and k.g .
savvidy , phys . lett .
* b324 * ( 1994 ) 72 .
+ g. k. savvidy and k.g .
savvidy , phys . lett.*b337 * ( 1994 ) 333 .
+ g.k.bathas , e.floratos , g.k.savvidy and k.g.savvidy , mod .
. lett . * a10 * ( 1995 ) 2695 .
+ g. k. savvidy , k.g .
savvidy and f.j .
wegner , nucl .
* b443 * ( 1995 ) 565 .
+ g. k. savvidy and k.g .
savvidy , mod .
* a11 * ( 1996 ) 1379 .
+ g. k. savvidy , k.g .
savvidy and p.g .
savvidy , phys.lett . *
a221 * ( 1996 ) 233 . + d. johnston and r.k.p.c .
malmini , phys .
b378 * ( 1996 ) 87 .
+ g. koutsoumbas , g. k. savvidy and k. g. savvidy , phys.lett . *
b410 * ( 1997 ) 241 .
+ j.ambjrn , g.koutsoumbas , g.k.savvidy , europhys.lett .
* 46 * ( 1999 ) 319 .
+ g.koutsoumbas and g.k.savvidy , mod.phys.lett .
* a17 * ( 2002 ) 751 .
ambartzumian , g.s .
sukiasian , g. k. savvidy and k.g .
savvidy , phys . lett .
* b275 * ( 1992 ) 99 .
+ g. k. savvidy and k.g .
savvidy , int . j. mod
* a8 * ( 1993 ) 3393 .
+ g. k. savvidy and k.g .
savvidy , mod .
. lett . * a8 * ( 1993 ) 2963 .
+ j. ambjrn , g.k .
savvidy and k.g .
savvidy , nucl.phys . *
b486 * ( 1997 ) 390 .
m. baig , d. espriu , d. johnston and r.k.p.c .
malmini , j. phys .
* a30 * ( 1997 ) 405 .
+ a. lipowski and d. johnston , j. phys . *
a33 * ( 2000 ) 4451
. + a.lipowski and d.johnston , phys.rev . *
e61 * ( 2000 ) 6375 .
+ a. lipowski , d. johnston and d. espriu , phys rev . * e62 * ( 2000 ) 3404 .
+ m. swift , h. bokil , r. travasso and a. bray , phys . rev . *
b62 * ( 2000 ) 11494 .
+ s. davatolhagh , d. dariush and l. separdar , phys rev . * e81 * ( 2010 ) 031501 . | a @xmath0 ising model with a purely plaquette , @xmath1-spin interaction displays a planar flip symmetry intermediate between a global and a gauge symmetry and as a consequence has a highly degenerate low temperature phase and no standard magnetic order parameter .
this plaquette hamiltonian is a particular case of a family of @xmath0 gonihedric ising models defined by savvidy and wegner .
an anisotropic variant of the the purely plaquette gonihedric model , originally discussed as the `` fuki - nuke '' model by suzuki , is non - trivially equivalent to a stack of @xmath2 ising models , each of which can magnetize independently at the phase transition point .
consideration of this anisotropic model suggests that a suitable order parameter in the isotropic case may also be constructed using a form of planar magnetization , in which nearest neighbour correlators @xmath3 summed over planes replace a sum over spin values @xmath4 .
we conduct monte - carlo simulations to investigate this and related candidate order parameters in a toy model of gonihedric ground states , the fuki - nuke model and the isotropic plaquette gonihedric model itself . |
the @xmath10 interaction below the pion - production threshold is fairly weak and featureless , as anticipated from an ` exotic ' channel corresponding to quark content @xmath11 , where @xmath12 denotes a light nonstrange quark .
this merit has motivated past suggestions to probe nuclear in - medium effects by studying scattering and reaction processes with @xmath0 beams below 800 mev / c ; see ref.@xcite for an early review .
limited total cross - section data @xcite on carbon , and elastic and inelastic differential cross section data @xcite on carbon and calcium , drew theoretical attention already in the 1980s to the insufficiency of the impulse - approximation @xmath1 form of the @xmath0 - nucleus optical potential , where @xmath13 is the free - space @xmath10 @xmath13 matrix , particularly with respect to its reaction content ( ` reactivity ' below ) . in order to account for the increased reactivity in @xmath0 - nucleus interactions , siegel _
et al._@xcite suggested that nucleons ` swell ' in the nuclear medium , primarily by increasing the dominant hard - core @xmath14 phase shift .
et al._@xcite suggested that the extra reactivity was due to the reduced in - medium masses of exchanged vector mesons , and this was subsequently worked out in detail in ref.@xcite .
another source for increased reactivity in @xmath0 - nucleus interactions was discussed in the 1990s and is due to meson exchange - current effects @xcite .
some further experimental progress was made during the early 1990s , consisting mostly of measuring attenuation cross sections in @xmath0 transmission experiments at the bnl - ags on deuterium and several other nuclear targets in the momentum range @xmath15 mev / c @xcite and of measuring @xmath0 quasielastic scattering on several targets at 705 mev / c @xcite .
new measurements of @xmath0 elastic and inelastic differential cross sections on c and ca at 715 mev / c were reported in ref.@xcite and analyzed in ref.@xcite , and self - consistent final values of @xmath0 integral ( reaction and total ) cross sections on @xmath16li , @xmath17c , @xmath18si and @xmath19ca were published in refs.@xcite . by the late 1990s , experimentation in @xmath0 - nuclear physics has subsided , and with it died out also theoretical interest .
the subject was reviewed last time in hyp97 , concluding that every experiment of @xmath0 mesons with complex nuclei finds cross sections larger than those predicted " @xcite ; and based on analyses of @xmath0 - nuclear integral cross sections @xcite , it was concluded that at present theory misses some unconventional in - medium effects " @xcite .
the @xmath5(1540 ) exotic baryon @xcite provides a new mode of reactivity to @xmath0 - nuclear interactions .
the @xmath5 couples directly to nucleons , @xmath20 , but its small width of order 1 mev or less @xcite indicates that this coupling is weak .
the @xmath21 phase - shift input to the @xmath1 optical potential in the vicinity of the @xmath5 mass ( corresponding to @xmath22 mev / c ) is derived from @xmath23 scattering data across the @xmath5 resonance energy . therefore , whatever the @xmath24 coupling is , its effect is already included at least implicitly in the @xmath1 impulse approximation .
a deuteron target also allows for a two - body production reaction @xmath25 with threshold at @xmath26 mev / c @xcite , but the available @xmath23 scattering data do not give any evidence for the opening of such a production channel .
it is plausible that _ denser _ nuclear targets would help to fuse the @xmath0 s @xmath27 quark together with two @xmath28 diquarks , each of which belongs to a distinct nucleon , into the @xmath29 configuration according to the jaffe - wilczek model of @xmath5 @xcite , and similarly for the diquark - triquark karliner - lipkin model @xcite .
therefore , one should look for traces of @xmath0 _ absorption _ on two nucleons , @xmath30 , in normal nuclei . in the present work
we show that traces of such two - body @xmath0 absorption may be identified in @xmath0 - nucleus dynamics and help resolve the long - standing anomalies in the @xmath0 - nucleus integral cross sections .
we determine the @xmath0 absorption cross section on nuclei at @xmath31 mev / c which is the closest momentum to the @xmath5 rest mass , where good data are available .
our results provide the first concrete demonstration that nuclear targets are potentially useful in the study of exotic baryons .
in the calculations presented below the klein gordon equation is solved , using the simplest possible @xmath1 form for the optical potential @xmath32 where @xmath33 is the reduced energy in the cm system , @xmath34 is a kinematical factor resulting from the transformation of amplitudes between the @xmath35 and the @xmath0 - nucleus cm systems and @xmath36 is the ( complex ) value of the isospin - averaged @xmath35 scattering amplitude in the forward direction .
the coulomb potential due to the charge distribution of the nucleus is included .
this form of the potential takes into account @xmath37 corrections , an important issue when handling as light a nucleus as @xmath16li . using this approach friedman _
et al._@xcite showed that no _ effective _ value for @xmath36 could be found that fits satisfactorily the reaction and total cross sections derived from the bnl - ags transmission measurements at @xmath38 mev / c on @xmath16li , @xmath17c , @xmath18si , @xmath19ca .
this is demonstrated in the upper part of fig .
[ kplusfig1 ] for the reaction cross sections per nucleon @xmath39 at 488 mev / c , where the calculated cross sections using a best - fit @xmath1 optical potential ( dashed line ) are compared with the experimental values listed in ref .
the best - fit values of re@xmath36 and im@xmath36 which specify this @xmath1 potential are given in the first row of table [ tab : fga04 ] where im@xmath36 represents @xmath40 increase with respect to the free - space value .
the @xmath41/n of this density - independent fit is very high .
its failure is due to the impossibility to reconcile the @xmath16li data ( which for the total cross sections are consistent with the @xmath23 ` elementary ' cross sections ) with the data on the other , denser nuclei , as is clearly exhibited in fig .
[ kplusfig1 ] for the best - fit @xmath1 dashed line .
if @xmath16li were removed out of the data base , then it would have become possible to fit reasonably well the rest of the nuclei , but the rise in im@xmath36 would then be substantially higher than that for the @xmath1 potential used here .
such fits , excluding @xmath16li , are less successful at the higher energies .
reaction cross sections per nucleon ( @xmath39 ) at @xmath31 mev / c are shown in the upper part .
calculated @xmath0 absorption cross sections per nucleon ( @xmath42 ) are shown in the lower part , see text.,height=257 ] to incorporate @xmath30 two - nucleon absorption into the impulse - approximation motivated @xmath43 , eq.([equ : vopt ] ) , we add a @xmath44 piece , as successfully practised in pionic atoms to account for @xmath45 absorption on two nucleons : @xmath46 where the parameter @xmath47 reperesents the effect of @xmath0 nuclear absorption into exotic @xmath48 baryonic channels . using this potential
we have repeated fits to all 32 data points for the reaction and total cross sections . as there were correlations between the parameters of re@xmath49 we subsequently kept the parameter re@xmath36 fixed at its free kaon - nucleon value .
the results are summarized in table [ tab : fga04 ] ( marked as ` eq.([equ : dd1 ] ) ' ) and the substantial improvement compared to the @xmath1 potential is self evident .
however , the fits at the higher momenta are not as successful as the fit at 488 mev / c , suggesting that one needs a more effective way to distinguish between @xmath16li and the denser nuclear targets .
in fact , it was shown _ empirically _
@xcite that the average nuclear density @xmath50 provides such discrimination and is instrumental in achieving good agreement with experiment .
therefore , we replace eq.([equ : dd1 ] ) by the simplest ansatz @xmath51 the added piece is a functional of the density which to lowest order reduces to a @xmath52 form . to justify it
theoretically would require to understand quantitatively the production mechanisms of the @xmath5 pentaquark in dense nuclear matter .
the resulting fits are shown in table [ tab : fga04 ] , marked as ` eq.([equ : dd2 ] ) ' .
the superiority of this @xmath53 version compared to the @xmath54 one is very clearly observed .
this conclusion remains valid also when the data for the elastic scattering of 715 mev / c @xmath0 by @xmath16li and @xmath17c @xcite are included in the analysis .
the calculated reaction cross sections at 488 mev / c , using eq.([equ : dd2 ] ) , are shown by the solid line marked @xmath55 in the upper part of fig .
[ kplusfig1 ] , where @xmath56 is the added piece of @xmath49 due to a nonzero value of @xmath47 .
clearly , it is a very good fit .
inspecting the results in table [ tab : fga04 ] one notes that the splitting of im@xmath49 into its two reactive components im@xmath36 and im@xmath47 appears well determined by the data at all energies , with very accurate values of im@xmath36 thus derived .
these values of im@xmath36 are close to , but somewhat below the corresponding free - space values .
the two - nucleon absorption coefficient im@xmath47 rises slowly with energy as appropriate to the increased phase space available to the underlying two - nucleon absorption process @xmath30 .
its values in this energy range are roughly independent of the form of @xmath56 , the more conservative eq.([equ : dd1 ] ) or the more effective eq.([equ : dd2 ] ) , used to derive these values from the data .
regarding re@xmath49 , and recalling that @xmath57 for the dense nuclear targets , it is clear that re@xmath58 at the two higher momenta , illustrating the inadequacy of the @xmath1 model in which the best - fit solution flips from repulsion at the two lower momenta ( consistently with the impulse approximation ) to attraction of a similar order of magnitude at the two higher momenta .
we note that our @xmath30 absorption reaction is related to the mechanism proposed recently in ref . @xcite as causing strong @xmath5 - nuclear attraction , based on @xmath59 two - meson cloud contributions to the self energy of @xmath5 in nuclear matter .
however , it would appear difficult to reconcile as strong @xmath5 - nuclear attraction as proposed there with the magnitude of re@xmath47 derived in the present work .
.fits to the eight @xmath0-nuclear integral cross sections at four laboratory momenta @xmath60 ( in mev / c ) , using different potentials .
[ cols="^,^,^,^,^,^,^",options="header " , ] the @xmath0 nuclear absorption cross section @xmath61 due to the availability of @xmath5 - nuclear final states is driven by im(@xmath56 ) .
we approximate it by @xmath62 are evaluated in two different ways in order to assess the theoretical uncertainty .
calculated absorption cross sections _ per target nucleon _ at @xmath31 mev / c are shown in the lower part of fig .
[ kplusfig1 ] for the fit using eq .
( [ equ : dd2 ] ) for @xmath49 in table [ tab : fga04 ] .
the triangles are for the case where @xmath56 does not enter the evaluation of the distorted waves @xmath63 , as appropriate to the dwia approximation , and the solid circles are for the case where @xmath63 are the fully distorted waves , including the effect of @xmath64 .
the error bars plotted are due to the uncertainty in the parameter im@xmath47 .
it is seen that these calculated absorption cross sections , for the relatively dense targets of c , si and ca , are proportional to the mass number @xmath65 , and the cross section per target nucleon due to im@xmath66 is estimated as close to 3.5 mb .
although the less successful eq.([equ : dd1 ] ) gives cross sections larger by @xmath67 , this value should be regarded an upper limit , since the best - fit density - dependent potentials of refs.@xcite yield values smaller than 3.5 mb by a similar amount . the experience gained from studying @xmath68-nuclear absorption @xcite leads to the conclusion that @xmath69 is smaller than the extrapolation of @xmath70 in fig .
[ kplusfig1 ] to @xmath71 , and since the @xmath35 interaction is weaker than the @xmath72 interaction one expects a reduction of roughly @xmath73 , so that @xmath74 mb .
we note in fig.[kplusfig1 ] the considerably smaller absorption cross section per nucleon calculated for @xmath16li which , considering its low density , suggests a cross section of order fraction of millibarn for @xmath75 , well below the order 1 mb which as gibbs has argued recently @xcite could indicate traces of @xmath5 in @xmath76 total cross sections near @xmath22 mev / c . to be definite , we assume that @xmath77 mb .
it is worth noting that this _ two - nucleon _ cross section is considerably larger than what a _
one - nucleon _ production process @xmath78 would induce on a deuteron target .
such a one - step process at the @xmath5 mass ( @xmath79 mev / c ) @xmath80 may be compared to pion absorption near the ( 3,3 ) resonance energy @xcite , where the primary two - nucleon absorption mechanism is through a direct @xmath81 production @xmath82 with @xmath83 mb . scaling this value by the ratio of coupling constants squared @xmath84 , assuming @xmath85 and @xmath86 mev , we estimate a cross section level of 0.03 mb for the one - step production process at the @xmath5 resonance energy . [ assuming @xmath87 , the one - step production cross section is lower by at least another order of magnitude . ]
the one - step cross section affordable by the neutron fermi motion at @xmath31 mev / c would be considerably smaller than this estimate @xcite .
in contrast , the two - nucleon reaction need not involve the suppressed @xmath24 coupling and its cross section , @xmath77 mb at @xmath31 mev / c , should depend smoothly on energy .
in conclusion , we have demonstrated a very good agreement between experiment and calculation for all the available integral cross - section data by adding to the @xmath1 optical potential a density - dependent term which simulates absorption channels .
we have identified these absorption channels , as exhibited by the anomalous reactivity established systematically in past phenomenological analyses of @xmath0 - nuclear interactions , with the @xmath5 production reaction @xmath4 with threshold at @xmath26 mev / c .
the analysis of these data is consistent with an upper limit of about 3.5 mb on the @xmath0 absorption cross section per nucleon , for @xmath5 production on the denser nuclei of c , si , ca , and indicates a sub - millibarn cross section for @xmath5 production on deuterium .
we urge experimenters to look for the @xmath75 two - body production reaction @xcite which requires experimental accuracies of 0.1 mb in cross section measurements . given the magnitude of the @xmath0 nuclear absorption cross sections , as derived in the present work , we also urge doing ( @xmath88 ) experiments on nuclear targets .
this reaction which has a ` magic momentum ' about @xmath89 mev / c , where the @xmath5 is produced at rest , is particularly suited to study bound or continuum states in _ hyponuclei _ @xcite .
it might prove more useful than the large momentum transfer ( @xmath90 ) reaction proposed in this context @xcite .
undoubtedly , precise @xmath0 - nuclear scattering and reaction data would be extremely useful to obtain further , more direct evidence for the presence of the @xmath5 exotic baryon and its effects in the nuclear medium
. in particular , @xmath23 and @xmath0 - nuclear data in the range @xmath91 mev / c would be very helpful to study the onset of strange - pentaquark dynamics in the nuclear medium .
the term ` hyponucleus ' was suggested by alfred s. goldhaber , as early as in 1982 , in the proceedings of the second lampf ii workshop , _ eds . _
thiessen , t.s .
bhatia , r.d .
carlini , and n. hintz , la-9572-c , vol .
i ( 1982 ) 171 , for a positive strangeness nucleus , if any exist " , adding that such objects are not expected from the known interactions of kaons and nucleons , which are repulsive at low energies , and therefore if found would be a sign of new degrees of freedom . " | ' '' '' long - standing anomalies in @xmath0 - nucleus integral cross sections could be resolved by extending the impulse - approximation @xmath1 optical - potential framework to incorporate @xmath0 absorption on pairs of nucleons .
substantially improved fits to the data at @xmath2 mev / c are obtained .
an upper bound on the absorption cross section per nucleon is derived , @xmath3 mb .
we conjecture that the underlying microscopic absorption process is @xmath4 , where @xmath5(1540 ) is the newly discovered exotic @xmath6=2 , @xmath7=0 , @xmath8=1 pentaquark baryon , and estimate that @xmath9 is a fraction of millibarn .
comments are made on @xmath5 production reactions on nuclei . |
quantum phase transitions in ultracold atoms provide important insights into the behaviour of matter at very low temperatures @xcite .
an important example of this class of transitions is the transition from a superfluid(sf ) to a mott insulator(mi ) which has been observed in cold bosonic atoms in @xmath3 optical lattices @xcite as predicted by jaksch _
subsequently , this transition has been observed in a @xmath4 optical lattice @xcite .
detailed theoretical studies of this transition have been carried out by pai _
et al _ @xcite .
an important question to address is how do the characteristics of this transition alter in going from one to two dimensions . in coupled bosonic chains
, competition exists between the ratio of atomic interactions to the intra - chain hopping and the inter - chain hopping . a large value of the former favors a mott insulator state overcoming the effect of the inter - chain hopping while if the latter dominates it would tend to delocalize the bosons and drive the system to a superfluid state .
it is not practical to perform numerical studies for the above mentioned transitions in a very large number of coupled chains , so one must restrict to a finite number of such chains .
the aim of the present work is to study the effect of the inter - chain hopping on the sf - mi transition for a bosonic ladder consisting of two coupled chains .
although a substantial amount of theoretical and numerical work has been done in this direction for the case of spin - less fermionic ladders and spin ladders @xcite , no work has been done to our knowledge for the bosonic ladders except a recent work using the bosonization method .
@xcite .
the hamiltonian of the bosonic ladder(as shown in fig.[fig : fig1])is given by @xmath5 in this model ( [ eq : ham ] ) @xmath6 ( @xmath7 ) represents bosonic creation ( annihilation ) operator for the site @xmath8 of the chain with index @xmath9 .
@xmath10 and @xmath11 are the intra - chain hopping amplitude between the nearest neighboring sites of chain @xmath12 and the on - site interaction between the bosons respectively .
the last term in this model ( [ eq : ham ] ) represents inter - chain hopping with an amplitude @xmath1 between corresponding sites on the two chains .
we set our energy scale by taking @xmath13 .
this model has been studied using the bosonization technique @xcite at or close to commensurate filling of one boson per site .
this study predicts a transition from a mott insulator to a superfluid phase when the inter - chain hopping is increased and it is in the beresenskii - kosterlitz - thouless ( bkt ) universality class at commensurate filling . in the present work we verify these predictions and thus complement the earlier analytical results . for this purpose , we study the variation of critical on - site interaction @xmath0 for the superfluid phase to the mott insulator phase transition with the change in the inter - chain hopping amplitude @xmath1 using the finite size density matrix renormalization group method ( fsdmrg )
@xciteand obtain the phase diagram in the @xmath14 plane .
to the best of our knowledge , the present work is the first application of fsdmrg method to bosonic ladders .
the remaining part of the paper is organized in the following manner.the fsdmrg method in the context of bosonic ladders is briefly described in section [ sec : fsdmrg ] .
our results are described and discussed in section [ sec : results ] and our conclusions are stated in section [ conclusion ] .
we use the fsdmrg technique to obtain the energies and the correlation functions of the ground state .
this method is very efficient and has proven to give accurate results for 1d quantum lattice systems and has been applied to low - dimensional strongly correlated fermionic and bosonic systems @xcite .
we give below some pertinent details of this method adapted to the system of two coupled chains that we have considered .
we begin with a super - block configuration @xmath15 of @xmath16 rungs as shown in fig.[fig : fig2 ] .
the left @xmath17 and the right block @xmath18 have @xmath19 rungs each and the @xmath20 represents one rung of two sites , one from each chain .
thus in every iteration , the new left and right blocks are @xmath21 and @xmath22 respectively .
we increase the size of the system in every iteration by adding two rungs which increases the number of lattice sites by @xmath23 . to keep the density @xmath24 fixed
, we also increase the number of bosons in the system by @xmath23 .
the truncation of states of left ( right ) block in each iterations corresponds to choosing @xmath25 highest weighted states out of @xmath26 of the left ( right ) density matrix . here
@xmath27 is the number of states kept at each site , which is in general infinity , but we truncate it for a feasible numerical calculation .
we keep @xmath28 in this calculation which is found to be sufficient for the values of @xmath11 considered here @xcite .
the value of @xmath25 is chosen such that the truncation error in our calculation is always less than @xmath29 .
the mott insulator phase has a finite gap in its energy spectrum .
the single particle gap is defined as @xmath30 where @xmath31 is the ground - state energy of two leg bosonic ladder with length @xmath16 having @xmath32 bosons .
the mi phase is signaled by the opening up of the gap @xmath33 .
however , @xmath34 is finite for finite systems and we must extrapolate to the @xmath35 limit , which is best done by using finite - size scaling @xcite . in the critical region ,
i.e. , sf region , the gap @xmath36 where the scaling function @xmath37 and @xmath38 is the correlation length .
@xmath39 in the sf region .
thus plots of @xmath40 versus @xmath11 , for different system sizes @xmath16 , consist of curves that intersect at the critical point at which the correlation length for @xmath41 diverges and gap @xmath42 vanishes . the phase diagram , as discussed below , is obtained from such plots .
it is now well known that the single chain bose - hubbard model with density @xmath24 shows a sf - mi transition with the critical on - site interaction @xmath43@xcite . in order to understand the effect of the inter - chain hopping on this transition , we varied @xmath1 from @xmath44 to @xmath45 and obtained the corresponding critical on - site interaction @xmath46 for the sf - mi transition .
we found that @xmath0 increases with @xmath1 and saturates in the limit @xmath47 .
these results are highlighted by the plots of scaling of gap @xmath40 versus @xmath11 for different values of @xmath1 and lengths @xmath16 .
for example , in the fig.[fig : fig3 ] we plot @xmath40 versus @xmath11 for @xmath48 .
the coalescence of @xmath40 curves for different values of @xmath16 below @xmath49 demonstrates the sf - mi transition with @xmath50 which is much larger than the corresponding value for the single chain @xmath51 .
fig.[fig : fig4 ] represents similar plots for @xmath52 . for this case
the critical on - site interaction increases further to @xmath53 . from similar plots of @xmath40 versus @xmath11
, we obtain the phase diagram for model ( [ eq : ham ] ) in the @xmath54 plane and it can be seen in fig.[fig : fig5 ] .
@xmath0 for the sf - mi transition initially increases sharply as the inter - chain hopping @xmath1 increases .
this phase diagram verifies the prediction of mi - sf transition with respect to increase in the inter - chain hopping @xmath1@xcite . for higher values of @xmath1
, @xmath0 tends to saturate . for @xmath55 ,
each rung has two one particle states : corresponding to bonding or anti - bonding .
as predicted in the bosonization @xcite study , this problem then maps onto a single chain bose hubbard model with commensurate density @xmath56 and on - site interaction @xmath57 . to confirm this prediction we plot the variation of @xmath0 with respect to @xmath1 in fig.[fig : fig6 ] .
critical on - site interaction @xmath0 for large @xmath1 converges to a value equal to @xmath58 .
plotting @xmath40 versus @xmath11 for single chain bose - hubbard model for @xmath56 in fig.[fig : fig7 ] we find that @xmath59 , which is one half the converged value of @xmath60 for the bosonic ladder confirming the prediction made in bosonization study@xcite .
the mott insulator to superfluid transition is found to be beresinskii - kosterlitz - thouless ( kt ) universality class at commensurate filling@xcite .
the correlation function that characterizes the superfluid phase is given by @xmath61 which decays as a power law in the limit @xmath62 . here the expectation value is taken with respect to the ground state .
however , in the mott insulator phase it has an exponential decay due to the finite gap in the energy spectrum . the power law decay of this correlation function has been obtained using the bosonization method @xcite and it is predicted to go as @xmath63 with the luttinger liquid parameter @xmath64 stated to be @xmath65 at the superfluid to mott insulator transition point . in order to obtain the luttinger liquid parameter @xmath64 we fix @xmath66 and vary the inter - chain hopping @xmath1 and obtain the mott insulator to superfluid transition .
the scaling of gap @xmath40 as a function of @xmath1 is given in fig.[fig : fig8 ] .
the critical inter - chain hopping @xmath67 for the mi to sf transition .
the correlation functions @xmath68 for @xmath66 different values of @xmath1 are given in fig.[fig : fig9 ] .
the luttinger liquid parameter @xmath64 which is obtained by fitting @xmath68 with the expression given in eq.[eq : corr ] is plotted as a function of @xmath69 in fig.[fig : fig10 ] . from these values
the critical @xmath70 for which @xmath71 is given by @xmath72 which is consistent with values obtained from the scaling of the gap .
the results we have obtained could have experimental implications .
it is now possible to prepare bosonic ladders by growing optical superlattices in the form of double well potential along one direction@xcite .
the tunneling between the double well potential will control the inter - chain hopping . by changing dynamically the optical lattice parameters
one can control all the interaction and hopping parameters of the model ( [ eq : ham ] ) .
we have studied the ground state properties of a two chain bosonic ladder with commensurate filling of one boson per site using the finite size density matrix renormalization group method .
the critical on - site interaction for sf - mi phase transition increases sharply for small values of inter - chain hoping amplitude @xmath1 .
however , it saturates in the limit @xmath73 .
thus in the presence of large inter - chain hopping , the system continues to be in the superfluid state even though the single chain is a mott insulator .
thus we confirm the prediction of mott insulator to superfluid transition as a function of @xmath1 .
we have obtained the luttinger liquid parameter and compared it with the analytical results .
in addition to verifying and complementing the predictions made by the bosonization technique , we have pointed out the possible experimental verification of our results . we hope that our analysis of the sf - mi transition in bosonic ladders will stimulate experimental studies in this direction .
one of us ( ml ) thanks the indian institute of astrophysics , bangalore where this work was done during her visit . this work was supported by dst , india ( grants no . sr / s2/cmp-0014/2007 ) .
99 i. bloch , j. dalibard , and w. zwerger , arxiv:@xmath74 .
s. l. sondhi , s. m. grivin , j. p. carini and d. shahar , rev .
phys . * 69 * , 315 ( 1997 ) .
s. sachdev , quantum phase transitions , cambridge university press ( 1999 )
. m. lewenstein , l. santos , m. a. baranov and h. fehtmann , phys .
lett . * 92 * , 050401 ( 2004 ) .
m. greiner , o. mandel , t. esslinger , t. w. haensch , and i. bloch , nature * 415 * , 39 ( 2002 ) .
t. stoeferle , h. moritz , c. schori , m. koehl , and t. esslinger , phys .
lett . * 92 * , 130403 ( 2004 ) .
d. jaksch , c. bruder , j. i. cirac , c. w. gardiner , and p. zoller , phys .
lett . * 81 * , 3108 ( 1998 ) .
r. v. pai , r. pandit , h. r. krishnamurthy , and s. ramasesha , phys .
lett * 76*,2937(1996 ) .
r. v. pai and r. pandit , phys .
b * 71 * , 104508 ( 2005 ) and references there in .
p. donohue , m. tsuchiizu , t. giamarchi , and y. suzumura phys .
b * 63 * , 045121(2001 ) .
s. t. carr , b. n. narozhny and a. a. nersesyan , phys .
b , * 73 * , 195114 ( 2006 ) .
y. park , s. liang and t. k. lee , phys .
b. * 59 * , 2587 ( 1999 ) .
p. donohue and t. giamarchi , phys .
b , * 63 * , 180508 ( 2001 ) . e. orignac and t. giamarchi , phys . rev.b
* 57 * , 11713 ( 1998 ) .
s. r. white , phys .
lett . * 69 * , 2863 ( 1992 ) ; phys .
b * 48 * , 10345 ( 1993 ) .
u. schollwck , rev .
phys . * 77 * , 259 ( 2005 ) .
m. tezuka and m. ueda , arxiv:@xmath75 .
a. m. rey , v. gritsev , i. bloch , e. demler , and m. d. lukin phys .
lett * 99 * , 140601 ( 2007 ) . | we study a bosonic ladder with two coupled chains using the finite size density matrix renormalisation group method .
we show that in a commensurate bosonic ladder the critical on - site interaction ( @xmath0 ) for the superfluid to mott insulator transition becomes larger as the inter - chain hopping ( @xmath1)increases .
we analyze this quantum phase transition and obtain the phase diagram in the @xmath2 plane . |
the hard probes 2012 conference was the first in this series since the start of the ion program at the cern lhc about 1.5 years ago .
it saw a wealth of new and high precision data from lhc , including first results from the 2011 high luminosity ion run and pp comparison data from different centre - of - mass energies , as well as plenty of fresh and significant results from rhic , in particular on heavy flavours .
this experimental summary will mention briefly a few of the highlights and the reader is referred to the individual experimental contributions in these proceedings for details and all original plots and results ( all contributions to these proceedings are referred to in the text by listing speaker / experiment ) .
recent reviews of ultra - relativistic heavy ion physics can be found in @xcite .
many heavy ion results rely on a direct comparison to pp and pa data ; high precision comparison data is therefore crucial , if not always flashy and glamorous .
besides its prime purpose of providing the baseline against which to quantify the modifications from either cold ( nuclear ) or hot ( qgp , hadron gas ) matter , comparison data is needed to tune event generators , help in establishing the validity range and accuracy of pqcd calculations , and last but not least , offers an important cross- and consistency check for detector performance and analysis techniques .
the good news at this conference was therefore that plenty of pp comparison data was presented , including open charm cross sections from lhc ( c. geuna / alice ) and rhic ( d. tlusty / star ) ; hidden charm ( j/@xmath1 , @xmath2 c. geuna /alice , c. lourenco / cms , m. calderon de la barca sanchez / star ) and hidden beauty ( @xmath3 family , c. lourenco / cms , a. kesich / star , s. whitaker / phenix ) ; jet cross sections and jet fragmentation ( r. ma / alice , d. perepelitsa / phenix ) .
the bad news was that the lhc pp data at the correct comparison energy of 2.76 tev was taken in only a couple of days during 2010 at rather low luminosity and therefore has mostly much larger statistical errors than the corresponding nuclear data !
this is , at the very least , unfortunate , and a few more days can hopefully be found in the lhc schedule during the ongoingupsilon pp run to remedy the situation .
proton - nucleus data is of equal importance for comparison to the nuclear data and interesting in its own right to study shadowing and saturation physics , e.g. in the theoretical framework of the colour glass condensate model .
the first long pa run at lhc will only happen later in early 2013 , but new d - au data are available from rhic and show a surprisingly large suppression of jets up to 30 gev relative to the scaled pp expectations ( d. perepelitsa / phenix ) ; naively one would have expected shadowing effects to be much smaller at the corresponding large values of @xmath4 .
the upsilon at forward rapidity looks suppressed as much if not more than the j/@xmath1 ( a. frawley / phenix ) .
also this is contrary to expectation , based on fixed target pa results , that cold nuclear matter absorption should be smaller for the @xmath3 than for the j/@xmath1 . while the statistical significance of this result is not very high at this moment , it serves as a useful reminder that a measurement is always preferable to even a well - motivated expectation .
the energy lost by energetic partons which traverse the hot and dense matter ( jet quenching ) measures the stopping power of the qgp , quantified by the transport coefficient @xmath5 , as well as its characteristic dependence on parton type ( quark vs. gluon ) , parton mass @xmath6 ( light vs. heavy quark ) , path length inside the medium @xmath7 , and parton energy @xmath8 . besides the obvious question of how much energy is lost , usually addressed by measuring the energy balance between di - jets or gamma - jet pairs
, one has also to verify the scaling with @xmath9 and @xmath10 , for which different models make different predictions .
finally , in order to learn about the dynamics of the energy loss process ( e.g. few hard gluon versus multiple soft gluon radiation , contributions of collisional and radiative energy loss ) , one has to measure the energy and angular distribution of the lost energy , which reappears typically in the form of very soft particles ( @xmath11 few gev/@xmath12 ) and at large angles to the initial parton direction .
a number of new results , and addressing all of the above questions , have been shown at this conference in a variety of different observables , from inclusive single particle spectra ( unidentified / identified / heavy flavour ) , two and three particle correlations on both near and away side , inclusive single jets ( @xmath13 , longitudinal and transverse fragmentation functions ) , and finally jet correlations ( di - jet , @xmath14-jet ) . only by looking at
many observables one can tie together and characterise the different facets of jet - quenching ( p.
jacobs / alice ) . suppression of charged hadrons , the signal where jet quenching was discovered at rhic , is still a very useful and sensitive observable even in the era of fully reconstructed jets , and severely constrains the various models of jet quenching at lhc ( m. floris / alice ) .
the suppression is found to be independent of particle species above some 8 - 10 gev/@xmath12 ( p.
christiansen / alice ) , and the 2011 high luminosity run has significantly decreased the statistical errors at high pt ( k. krajczar / cms ) , where the suppression remains at about a factor of two up to at least 100 gev/@xmath12 .
suppression of calorimetric ( a. angerami / atlas ) or charged ( m. verweij / alice ) jets is also strong out to the highest measured jet energy of about 200 gev and of the same order ( @xmath15 as the suppression of single high @xmath16 particles .
this implies that very little or none of the lost energy is collected within the typically used jet - cones ( @xmath17 , @xmath16 @xmath18 gev/@xmath12 ) .
new high statistics data on di - jets ( c. roland / cms ) confirm that the back - to - back correlation does not change with increasing energy loss from its value measured in pp , implying that the medium induced transverse momentum kick is ( much ) smaller than the natural broadening from initial and final state vacuum radiation .
the average energy imbalance between di - jets is about 10% higher in central pb - pb than in pp , independent of jet energy out to more than 300 gev .
together with the very first results on energy balance in the particularly clean @xmath19-jet channel ( y. lai / cms ) , these data can be used to extract , via model comparisons , both the amount of energy loss ( or @xmath5 ) as well as its energy dependence .
also the first @xmath19-hadron correlations measured at rhic ( j. frantz / phenix ) show the expected medium effect on fragmentation functions .
electroweak probes are excellent calibration candles for jet - quenching measurements , as both production and propagation should not be effected by the medium . and indeed , as was amply demonstrated at this conference , it is nt ( p. steinberg / atlas , z. citron / atlas , a. milov / atlas , b. de la cruz / cms , y.j .
lee / cms , l. benhabib / cms ) .
the nuclear modification factor @xmath20 is one , within current systematic and statistical errors of order 10 - 20% , for all measured centralities , transverse momenta , and probes ( @xmath19 , w , z ) .
these results confirm that the collision geometry and assorted derived parameters like impact parameter , number of participants and collisions , are under good experimental control ( shadowing effects at the relevant @xmath4 are expected to be below the 5 - 10% level ) .
the fate of heavy ( c , b ) quarks in the qgp is of interest in several respects .
they are tracers created essentially only during the initial collision in hard scattering processes and their energy loss should be smaller than the one of gluons ( different colour charge ) or of light quarks ( mass dependent reduction of energy loss or deadcone effect ) . in a very strongly interacting plasma , they may eventually become thermalized and take part in the hydrodynamic collective flow of the bulk matter .
heavy flavour results from lhc and rhic include fully reconstructed d - mesons ( z. conesa del valle / alice , m. calderon de la barca sanchez / star ) , heavy flavour muons ( y. chen / atlas , d. stocco / alice , k. read / phenix ) and electrons ( m. kweon / alice , o. hajkova / star , s. tarafdar / phenix ) , as well as b - mesons identified via non - prompt j/@xmath1 decays ( t. dahms / cms ) .
overall indications are roughly consistent with the expected hierarchy of suppression , @xmath20 ( @xmath21 ) @xmath22 @xmath20 ( d ) @xmath11 @xmath20 ( b ) ; however statistical and systematic errors have to be improved and potential shadowing corrections need to be quantified with pa data before definite conclusions can be drawn .
the lhc data however do confirm with fully reconstructed charm mesons ( s. masciocchi / alice ) what has been seen previously with heavy flavour ( non - photonic ) leptons at rhic , namely that the suppression of charm is very similar to that of other hadrons above 5 - 8 gev/@xmath12 and larger only , if at all , for non - relativistic momenta ( @xmath23 of order one ) .
indications , again at the edge of significance , have been shown for charm elliptic flow from both lhc and rhic ( g. ortona / alice , o. hajkova / star , s. tarafdar / phenix ) for momenta above 2 gev/@xmath12 , of a magnitude comparable to the one of charged particles .
this could be a sign of charm quark thermalisation at the lower momenta and/or path length dependent energy loss at the higher momenta ; smaller error bars and an extension towards higher momenta at rhic and lower momenta at lhc would be very helpful for a quantitative interpretation . in order to investigate the energy lost by the jet - quenching process , experimentalists have to dig into the soft part of the hadron spectrum , typically by looking at correlations between a trigger particle ( as a proxy for the jet ) and soft associated particles of a few gev . trigger bias , which in this method enriches the sample with unmodified jets from the surface , as well as correlations introduced by hydrodynamic flow , have to be taken into account in analysis and interpretation .
data on two particle correlations from lhc ( j.f grosse oetringhaus / alice ) and rhic , ( a. ohlsen / star , j. franz / phenix ) cover the interesting momentum range from around 1 gev/@xmath12 to above 10 gev/@xmath12 , i.e. the region where one expects the lost energy to reappear as well as any medium response to the large local energy deposited by jet quenching . indeed , tell - tale signatures of increased soft particle correlations are seen in the energy balance ( @xmath24 , a. ohlsen / star ) and the azimuthal correlation ( t. todoroki / phenix ) at the away side at rhic . at lhc , the near - side shows a curious modification ( a. morsch / alice ) , which affects the width of the rapidity correlation but not ( or much less ) the width of the azimuthal correlation between trigger and associated particles .
whether this is a sign of medium modified jet fragments ( radiated jet energy ) or jet modified medium fragments ( response of the medium to local energy deposit ) remains to be seen ; interestingly enough the ampt monte carlo shows quantitative agreement with the data without , so far at least , revealing its underlying cause .
further hints about the nature of jet quenching come from correlations with identified particles , where it seems that the anomalous large proton to pion ratio typical of heavy ion collisions is not present in jet fragments on the near side ( m. veldhoen / alice ) and is significantly reduced on the away side ( a. davila / star ) . while the former could be caused , at least in part , by the previously mentioned surface bias , the latter clearly is not .
these results may therefore indicate that jet fragments are less subject to bulk effects like radial flow and coalescence , both of which are thought to be the cause of the baryon enhancement in bulk matter .
the melting of hidden heavy quark mesons ( the j/@xmath1 and @xmath3 families ) is arguably the most prominent and sought after signal of deconfinement in the qgp . notwithstanding its 1986 declaration as an unambiguous signature of qgp formation , and the first observation of j/@xmath1 suppression
the same year by the na38 collaboration at the cern sps , j/@xmath1 production in nuclear collisions has surprised and confounded the heavy ion community for more than 25 years : from the first suppression measurement with light ions , later attributed to hadronic final state absorption in cold nuclear matter , the anomalous j/@xmath1 suppression seen later in central pb - pb collisions , the indecisive results with lighter indium ions , to the initially very surprising observation that j/@xmath1 suppression as function of centrality at the sps is essentially identical to the one measured at more than ten times higher energy at rhic . for the latter observation ,
two broad classes of explanations were put forward . in the sequential suppression scenario , sps and lhc results
are equal by design , because the temperature at both machines remains below the critical temperature needed to melt the directly produced j/@xmath1 , but is ( in both machines ) above the melting temperature of @xmath1 and @xmath2 , which together contribute somewhat less than 1/2 of the total j/@xmath1 yield . in the recombination picture ,
the suppression caused by melting ( deconfinement ) of all members of the j/@xmath1 family increases with temperature from sps to rhic , but is approximately cancelled by secondary j/@xmath1 production via coalescence between independently created charm quarks ; in this scenario sps and rhic are equal by accident. a decisive answer was ( yet again ) expected by going to even higher energy , the lhc , because melting and recombination in general should depend differently on @xmath25 .
the sequential suppression picture predicted an equal or stronger suppression at lhc , the recombination picture a weaker one ( or even an enhancement ) .
and indeed , the lhc lived up to expectation .
high statistics results shown for the first time at this conference ( c. suire / alice ) establish that the j/@xmath1 is overall _ less _ suppressed at lhc than at rhic , but that , unlike at rhic , the suppression is _ stronger _ at high @xmath16 than at low @xmath16 .
in addition , indications are that the j/@xmath1 shows elliptic flow at lhc ( l. massacrier / alice ) , but not at rhic ( m. calderon de la barca sanchez /star ) .
this pattern is qualitatively and , within uncertainties , quantitatively consistent with the coalescence prediction that the abundantly produced individual charm quarks at lhc thermalize and recombine , and preferentially so at low relative momenta .
while at first sight regeneration via charm quark coalescence may seem to be yet another dirty effect which obscures the deconfinement signal , it actually is a consequence of deconfinement itself and therefore a very relevant observation ! unlike the light quarks , charm quarks are produced at the very beginning of the collisions in independent ( and distant ) hard interactions ; they therefore have to roam freely and over large distances to occasionally find a coalescence partner to bind to ; in other words , they are deconfined during the system evolution .
the story seems to be more complicated for the @xmath1 , which was supposed to have melted away completely already at rhic . in a totally unexpected and bizarre twist ,
the @xmath1 was found at lhc to be _ enhanced _ with respect to the j/@xmath1 at low @xmath16 ( t. dahms / cms ) ! if this anomalous @xmath1 un - suppression , which as of now has limited significance due to the lack of 2.76 tev pp comparison data , stands the test of time , it will have totally unpredictable , but definitely profound consequences for the quarkonia suppression !
recombination is thought to be less of an issue for beauty quarks , and therefore @xmath3 suppression should reflect the resonance melting more directly .
the pattern observed at lhc ( c. mironov / cms ) is the following : the tightly bound @xmath3(1s ) is suppressed by slightly more than a factor of two , the @xmath3(2s ) by about a factor of ten , and the most loosely bound state , the @xmath3(3s ) , is no longer visible within statistics .
this is qualitatively consistent with a sequential melting scenario , where most of the higher mass resonances have partially or completely disappeared whereas the directly produced @xmath3(1s ) may actually be not suppressed at all ( as about 50% of the observed @xmath3(1s ) state results from feed down from @xmath26 and other higher resonance states ) .
crucial support for a sequential melting scenario would come from observing a change of the suppression pattern with temperature ( i.e. beam energy ) .
first results on the sum of the three s - wave @xmath3 states have been shown from rhic ( a. kesich / star , s. whitaker / phenix ) ; the suppression is comparable to the one at lhc , maybe a bit smaller ( which would be in line with a melting scenario ) , but errors are too large at the moment and this critical comparison has to await more statistics from rhic . finally , and for the first time since many years , the quarkonia results seem qualitatively consistent with the presence of a deconfining medium , which implies both melting as well as regeneration of some of the heavy quark bound states .
however , even leaving aside for the time being the @xmath1 , a quantitative understanding needs better statistics ( in particular for the @xmath3 ) , correcting for possible cold nuclear matter effects ( absorption / shadowing ) at lhc with data from the upcoming ppb run , and quantitatively taking into account ( if possible , even measuring ) the contributions of feed down from the various higher lying quarkonia resonance states .
while not a traditional focus at the hard probes conferences , a number of significant new results were reported on flow observables , mostly from the intermediate semi - hard momentum range .
azimuthal anisotropies ( v@xmath27 ) become independent of particle species ( p. christiansen / alice ) , but only above some 10 gev/@xmath12 ; presumably at these high momenta they are no longer caused by hydrodynamic flow .
the anisotropy extends out to an amazing 50 gev/@xmath12 ( c. maguire / cms ) , with a magnitude roughly in line with the predictions from path length dependent jet - quenching calculations .
elliptic flow fluctuations have been measured for the first time as a function of transverse momentum ( j.f grosse oetringhaus / alice ) , and are found to be amazingly independent of it , from the lowest momenta where hydro flow dominates ( @xmath11 3 gev/@xmath12 ) out to the region presumably governed by the completely different mechanism of jet - quenching ( @xmath28 6 gev/@xmath12 ) .
a first measurement of the rapidity - even directed flow ( v@xmath29 ) has been reported ( s. mohapatra / atlas ) , taking into account momentum conservation terms , as well as a remarkable variety of higher order flow plane correlations ( j. jia / atlas ) , some of which seem caused by subtle correlations in the initial geometry , others by non - linear evolution in the hydro expansion .
a significant excess is seen in the low mass electron pair continuum ( @xmath30 200 - 800 mev/@xmath31 ) at rhic ( l. ruan / star , a. drees / phenix ) .
this most important signal , of relevance for both thermal radiation from the qgp as well as chiral symmetry restoration at the phase boundary , is also one of the most difficult signals to observe in heavy ion collisions , because of the very small signal - to - background ratio .
while both experiments report a significant excess above standard resonance contributions , the two results are unfortunately also known to be inconsistent with each other , with the larger value ( from phenix ) well above any current estimate of thermal radiation and/or rho - meson broadening .
a resolution of this discrepancy is eagerly awaited , as are first results from lhc , which however will be even more difficult to extract than at rhic .
less than 18 months into heavy ion operation at lhc , and with a steady stream of new and very competitive results from rhic , the hard probes 2012 conference has seen impressive progress for the field . on the topic of parton energy loss / jet - quenching , some of the expected scaling laws mass , energy , path length ,
colour charge are being put to the test , with passing grades so far ( some only barely ) ; first hints are found of the lost energy in modified fragmentation functions and two particle correlations ; and the holy grail signal of clean @xmath14-jet pairs is starting to appear , begging for more statistics and inviting direct comparisons with theory .
light is visible at the end of the quarkonia tunnel with the tentative picture taking hold of both sequential quarkonia melting as well as quarkonia recombination , both direct consequences of deconfinement , but also a potential new obstacle in the form of a ghostly resurrected @xmath32 . in the bulk matter below 2 - 3 gev/@xmath12 , hydro reigns supreme not only to leading ( elliptic and radial flow ) and next - to - leading ( higher harmonic flow and fluctuations ) order , but even to nnlo , e.g. explaining some of the non - trivial correlations between different order reaction planes as consequences of non - linear hydrodynamic evolution .
the intermediate momentum region however , joining the soft and the hard regimes between a few and some ten gev/@xmath12 , remains the least explored and the least understood , and it is to be hoped that in a future hard probes conference also this region will be coming more into focus from both experiment and theory .
00 b. muller and j. l. nagle , ann .
nucl . part .
* 56 * ( 2006 ) 93 [ nucl - th/0602029 ] .
b. v. jacak and b. muller , science * 337 * ( 2012 ) 310 .
b. muller , j. schukraft and b. wyslouch , arxiv:1202.3233 [ hep - ex ] . | the _ @xmath0 international conference on hard and electromagnetic probes in high - energy nuclear collisions _ was held in may 2012 in cagliari , italy .
this contribution summarises some of the experimental highlights presented at the meeting , concentrating on new results from lhc and rhic on parton energy loss ( jet - quenching ) and heavy quark meson production ( quarkonia suppression ) .
heavy ion physics , quark - gluon plasma , lhc , rhic |
independent observational data such as sn ia @xcite , cmb @xcite , and bao @xcite suggest that about 70% of the energy density today consists of dark energy responsible for cosmic acceleration .
for the constant dark energy equation of state @xmath0 the recent joint data analysis based on sn ia , cmb , bao , and the hubble constant measurement shows that @xmath0 is constrained to be @xmath8 at 68 % cl @xcite .
if we use the time - dependent parametrization @xmath9 , where @xmath10 is the scale factor normalized as @xmath11 today , the two parameters @xmath3 and @xmath12 are constrained to be @xmath13 and @xmath14 @xcite .
one of the simplest candidates of dark energy is the cosmological constant characterized by the equation of state @xmath6 , which is consistent with the current observational data .
however , if the cosmological constant originates from the vacuum energy associated with particle physics , there is a huge gap between the theoretical and observed values @xcite . instead , alternative dark energy models with dynamically changing @xmath0 such as quintessence @xcite and k - essence @xcite
have been proposed ( see refs .
@xcite for reviews ) .
quintessence is described by a canonical scalar field @xmath15 with a potential @xmath16 . in the framework of particle physics
it is generally difficult to accommodate a very light scalar field with a mass of the order of the hubble parameter @xmath17 ev today @xcite .
however there have been theoretical attempts to construct viable quintessence models in particle physics especially in supersymmetric theories @xcite .
for example , the pseudo - nambu - goldstone - boson ( pngb ) @xcite or axions @xcite have the potential of the form @xmath18 $ ] with suppressed quantum corrections .
caldwell and linder @xcite classified quintessence models into two classes , depending on the evolution of @xmath0 .
the first class corresponds to thawing models , in which the field is nearly frozen by a hubble friction during the early cosmological epoch and it starts to evolve once the field mass @xmath19 drops below the hubble rate @xmath20 . in this case
the evolution of @xmath0 is characterized by the growth from @xmath2 .
the representative potential of this class is the hilltop potential such as @xmath21 $ ] .
the second class consists of freezing models , in which the evolution of the field gradually slows down because of the shallowness of the potential at late times . for the inverse power - law potential @xmath22 ( @xmath23 ) @xcite there is a so - called tracker solution characterized by a nearly constant field equation of state @xmath24 during the matter era @xcite . in this case
the solutions with different initial conditions approach a common trajectory ( tracker ) first , which is followed by the decrease of @xmath0 toward @xmath2 .
in addition to tracking freezing models there is another sub - class of freezing models associated with scaling solutions @xcite . in this case the field equation of state scales as that of the background fluid during most of the matter era ( @xmath25 ) .
the representative potential of this class is @xmath26 , where @xmath27 is the reduced planck mass , @xmath28 and @xmath29 are constants with @xmath30 and @xmath31 @xcite . in the early matter era
the potential is approximated as @xmath32 , which gives rise to the scaling solution characterized by @xmath33 with the field density parameter @xmath34 . at late times
the dominance of the potential @xmath35 leads to the rapid decrease of @xmath0 relative to the tracker case mentioned above .
note that the potential @xmath36 $ ] proposed in ref .
@xcite also exhibits a similar property to that of the double exponential potential ( see also ref .
@xcite ) .
in this paper we place observational constraints on three types of quintessence models : ( i ) thawing models , ( ii ) tracking freezing models , and ( iii ) scaling freezing models ( see refs .
@xcite for related works ) .
this analysis covers most of quintessence potentials proposed in literature . for thawing models we employ the analytic expression of @xmath0 derived in ref .
@xcite under the approximation that @xmath37 ( see also ref .
the likelihood analysis with the sn ia and bao data was carried out in ref .
@xcite ( see also ref .
we update the analysis by using the latest sn ia data ( union 2.1 dataset @xcite ) and by adding the data of cmb shift parameters measured by wmap7 @xcite .
we also take into account the recent bao data of the boss experiment @xcite .
forest appeared in the redshift range @xmath38 @xcite .
we do not take into account this new data in our likelihood analysis . ]
note that the observational constraints on thawing models were carried out by using a multi - parameter extension of the exponential potential @xcite and by introducing a statefinder hierarchy @xcite .
our study based on the analytic solution of @xmath0 is more convenient in that it covers any quintessence potential having thawing properties and that @xmath0 is expressed in terms of three parameters without the need of introducing more free parameters . for tracking freezing models one of the present authors obtained the approximate analytic formula of @xmath0 expressed in terms of two free parameters @xcite ( see also ref .
@xcite ) . the likelihood analysis based on the sn ia and
bao data was performed in ref .
we show that adding the cmb and boss data further strengthens the constraints on the tracker equation of state @xmath4 .
@xcite placed observational bounds on a number of quintessence potentials having tracker properties .
our study based on the analytic formula of @xmath0 is general enough to cover such potentials .
moreover we show that inclusion of the boss bao data further strengthens the bounds on @xmath4 previously derived in the literature . for scaling freezing models
it is difficult to derive an analytic expression of @xmath0 , so we resort to numerical simulations to find a viable parameter space .
this paper is organized as follows . in sec . [
para ] we briefly review the procedure to derive approximate analytic expressions of @xmath0 in thawing and tracking freezing models .
the accuracy of those approximations is also discussed by solving the equations of motion numerically . for scaling freezing models
we show that in some cases it is possible to fit the evolution of @xmath0 by using a specific parametrization . in sec .
[ obsersec ] we first explain the method of our likelihood analysis based on the sn ia , cmb , and bao data and then we proceed to observational constraints on three classes of quintessence models . sec .
[ consec ] is devoted to conclusions .
quintessence @xcite is described by a minimally coupled scalar field @xmath15 with a potential @xmath16 .
in addition to the field @xmath15 we take into account non - relativistic matter with an energy density @xmath39 .
the action in such a system is given by s = d^4 x + s_m , where @xmath40 is the determinant of the metric @xmath41 , @xmath42 is the ricci scalar , and @xmath43 is the action for non - relativistic matter . in the flat friedmann - lematre - robertson - walker background with the scale factor @xmath44 , the dynamical equations of motion are & & 3h^2 m_pl^2=_+_m , [ be1 ] + & & + 3h + v_,=0 , [ be2 ] + & & _
m+3h _ m=0 , [ be3 ] where @xmath45 is the hubble parameter , a dot represents a derivative with respect to cosmic time @xmath46 , @xmath47 , and @xmath48 .
the pressure of the field is given by @xmath49 .
we introduce the equation of state @xmath50 and the density parameter @xmath51 of dark energy . from eqs .
( [ be1])-([be3 ] ) we obtain the following equations for @xmath0 and @xmath52 @xcite ( see also refs .
@xcite ) : w & = & ( 1-w ) , [ weq ] + _ & = & -3w _ ( 1- _ ) , [ omeeq ] + & = & - ( -1 ) ^2 . [ lameq2 ] where @xmath53 , @xmath54 , and a prime represents a derivative with respect to @xmath55 . depending on the field potential and the initial conditions ,
there are several different cases for the evolution of @xmath0 @xcite . in the following we shall discuss possible analytic solutions of @xmath0 for three different cases : ( i ) thawing , ( ii ) tracking freezing , and ( iii ) scaling freezing models .
the presence of analytic solutions is useful in that some general properties of physical parameters can be extracted without studying a host of quintessence potentials separately .
moreover , if @xmath0 can be analytically expressed in terms of the redshift @xmath56 with several free parameters , we do not need to integrate the background equations of motion with arbitrary initial conditions .
this greatly simplifies the likelihood analysis carried out in sec .
[ obsersec ] . for thawing models of quintessence the field @xmath15
is nearly frozen in the early matter era because of the hubble friction , so that @xmath0 is close to @xmath2 .
one can regard @xmath6 as the fixed point of eq .
( [ weq ] ) . for @xmath57
such a point is not stable , and hence @xmath0 starts to grow at the late cosmological epoch .
if we assume that @xmath58 is nearly constant , one can express @xmath0 in terms of @xmath52 by using eqs .
( [ weq ] ) and ( [ omeeq ] ) under the approximation @xmath59 @xcite .
this neglects the effect of the field mass squared @xmath60 , but it is possible to derive a more elaborate form of @xmath0 with the mass term taken into account @xcite . in doing so , the potential is expanded around the initial field value @xmath61 up to second order , i.e. , @xmath62
. provided that @xmath37 , the evolution of the scale factor can be approximated by that of the @xmath7cdm model , i.e. , @xmath63^{1/3 } \sinh^{2/3 } ( t / t_{\lambda})$ ] , where @xmath64 and @xmath65 is the today s density parameter of quintessence . on using this solution , eq .
( [ be2 ] ) is integrated to give the field in terms of the function of @xmath46 and hence the field equation of state @xmath66 is known .
this process leads to the following analytic expression of @xmath0 @xcite : w(a)=-1+(1+w_0 ) a^3(k-1 ) ^2 , [ wtha ] where @xmath3 is the value of @xmath0 today , and k & = & , [ kdef ] + f(a ) & = & .
the solution ( [ wtha ] ) is valid for @xmath67 .
the equation of state ( [ wtha ] ) is expressed in terms of the three parameters @xmath3 , @xmath65 , and @xmath68 .
the quintessence equation of state @xmath0 versus @xmath10 for the potential ( [ pngbpo ] ) with ( a ) @xmath69 , @xmath70 ( @xmath71 ) , ( b ) @xmath72 , @xmath73 ( @xmath74 ) , and ( c ) @xmath75 , @xmath76 ( @xmath77 ) .
these cases correspond to @xmath78 , so that @xmath79 .
the solid curves show numerical solutions , whereas the bald dashed curves describe the results derived from the parametrization ( [ wtha ] ) with @xmath80.,width=326,height=307 ] as a concrete example , let us consider the hilltop potential v()=^4 .
[ pngbpo ] in this case the parameter ( [ kdef ] ) is given by k=^1/2 .
if @xmath81 and @xmath82 , one has @xmath79 and @xmath83 respectively . in the former case the potential is approximately given by @xmath84 $ ] around @xmath85 , whereas in the latter case it is approximated as @xmath86 around @xmath87 . in fig .
[ wthaw1 ] we plot the numerical evolution of @xmath0 versus @xmath10 for @xmath79 with several different values of @xmath88 and @xmath61 .
the bald dashed curves correspond to the results derived by the analytic expression ( [ wtha ] ) , which show good agreement with the numerically integrated solutions for @xmath89 . for @xmath68 larger than 10
, the initial displacement of the field is required to be close to @xmath90 to avoid its rapid roll down . in such cases
the field mass is largely negative , which leads to the tachyonic instability of field perturbations .
if the field reaches the potential minimum by today and it starts to oscillate , numerical simulations show that eq .
( [ wtha ] ) is no longer reliable .
we set the prior @xmath91 in the likelihood analysis of sec .
[ thawconsec ] . the analytic estimation ( [ wtha ] ) starts to lose accuracy for @xmath68 smaller than 1 .
this reflects the fact that the field is initially located away from the potential maximum .
then the taylor expansion around @xmath92 tends to be more inaccurate because of the rapid variation of the field .
numerically we find that the analytic solution ( [ wtha ] ) is reliable for @xmath93 and @xmath94 . in eq .
( [ weq ] ) there is another fixed point given by _
= , [ ometracker ] along which @xmath0 is constant .
this corresponds to the tracker that attracts the solutions with different initial conditions to a common trajectory .
the condition under which the tracking occurs is @xcite with @xmath95 is given by @xmath96 @xcite . ]
> 1 . in this case
the variable @xmath58 approaches 0 . from ( [ ometracker ] )
it follows that @xmath97 . using this relation with eqs .
( [ omeeq ] ) and ( [ lameq2 ] ) in the regime @xmath98 , the field equation of state along the tracker is given by w = w_(0 ) - .
[ 0th ] for the potential @xmath99 ( @xmath23 ) one has @xmath100 and hence @xmath101 @xcite .
the result ( [ 0th ] ) was derived by neglecting the contribution of @xmath52 , but its effect can be accommodated by dealing with @xmath52 as a perturbation to the 0-th order solution ( [ 0th ] ) @xcite .
we consider the first - order perturbation @xmath102 around @xmath4 and then approximate @xmath103 by the 0-th order solution _
( a)=. [ omepa ] for the models in which @xmath104 is nearly constant , we obtain the following analytic solution @xcite : w(a ) & = & w_(0)+ _ n=1^ ( ) ^n + & = & w_(0)+ _
( a)+ _ ( a)^2 + & & + _ ( a)^3 + , [ wtrack ] where in the second and third lines we carried out the expansion @xmath105 .
the equation of state ( [ wtrack ] ) is expressed in terms of the two parameters @xmath4 and @xmath65 .
numerically we confirm that the approximated formula ( [ wtrack ] ) tends to approach the full numerical solution by adding higher - order terms of @xmath106 .
the scaling solution @xcite is a special case of a tracker along which @xmath52 in eq .
( [ ometracker ] ) is constant with @xmath107 . from eq .
( [ omeeq ] ) it then follows that @xmath33 and hence @xmath108 during the matter era . since @xmath58 is constant , one has @xmath109 from eq .
( [ lameq2 ] ) .
this corresponds to the exponential potential @xmath110 , where @xmath111 is a constant ( where @xmath112 ) the condition for the existence of scaling solutions restricts the lagrangian in the form @xmath113 , where @xmath40 is an arbitrary function in terms of @xmath114 @xcite .
quintessence with the exponential potential corresponds to the choice @xmath115 . ] . in this case , however , the field equation of state is the same as that of the background fluid ( @xmath33 ) , so that the system does not enter the phase of cosmic acceleration .
this problem can be circumvented for the following model @xcite v()=v_1 e^-_1 /m_pl+ v_2 e^-_2 /m_pl , [ doublepo ] where @xmath116 and @xmath117 ( @xmath118 ) are constants . if @xmath30 and @xmath31 then the solution first enters the scaling regime characterized by @xmath119 , where @xmath120 is the equation of state of the background fluid . from the bound coming from big bang nucleosynthesis @xcite
there is a constraint @xmath121 ( 95% cl ) during the radiation era ( @xmath122 ) , which translates into the condition @xmath123 .
the scaling matter era ( @xmath34 , @xmath33 ) is followed by the dark energy dominated epoch driven by the presence of the potential @xmath35 .
if @xmath124 , the solution approaches another attractor characterized by @xmath125 and @xmath126 @xcite .
the cosmic acceleration occurs for @xmath127 .
the onset of the transition from the scaling matter era to the dark energy dominated epoch depends on the parameters @xmath28 , @xmath29 , and @xmath128 .
numerically we find that the transition redshift is not very sensitive to the choice of @xmath128 , so we study the case @xmath129 . in fig .
[ wscaling ] we plot the numerical evolution of @xmath0 for @xmath130 with three different values of @xmath28 . for larger @xmath28 the transition to the dark energy dominated epoch occurs earlier . the quintessence equation of state @xmath0 versus @xmath10 for the potential ( [ doublepo ] ) with ( a ) @xmath131 , @xmath130 , ( b ) @xmath132 , @xmath130 , and ( c ) @xmath133 , @xmath130 .
the solid curves show the numerical solutions , whereas the dashed curves represent the results derived from the parametrization ( [ lhpara ] ) with @xmath134 and @xmath135 .
each dashed curve corresponds to ( a ) @xmath136 , @xmath137 , ( b ) @xmath138 , @xmath137 , and ( c ) @xmath139 , @xmath140 . ,
width=326,height=307 ] it is possible to accommodate the above variation of @xmath0 analytically by using the parametrization proposed by linder and huterer @xcite ( see also refs .
@xcite ) : w(a)=w_f+ , [ lhpara ] where @xmath141 and @xmath142 are asymptotic values of @xmath0 in the past and future respectively , @xmath143 is the scale factor at the transition , and @xmath144 describes the width of the transition .
the scaling solution during the matter era corresponds to @xmath134 . for @xmath130 one
has @xmath135 , so that the parametrization ( [ lhpara ] ) reduces to @xmath145^{-1}$ ] .
figure [ wscaling ] shows that the parametrization ( [ lhpara ] ) can fit the numerical evolution of @xmath0 very well for appropriate choices of @xmath143 and @xmath144 . for @xmath130
the transition width is around @xmath146 , while @xmath143 depends on the values of @xmath28 . in this case
one can carry out the likelihood analysis by fixing @xmath137 and find the constraints on @xmath143 ( see ref .
@xcite for a related work ) . the quintessence equation of state @xmath0 versus @xmath10 for the potential ( [ doublepo ] ) with @xmath147 and @xmath148.,width=326,height=307 ] if @xmath149 , then the field equation of state finally approaches the value @xmath126 .
numerically we find that @xmath0 tends to have a minimum for larger @xmath29 before the solutions reach the attractor .
if @xmath147 , for example , the minimum appears for @xmath150 ( see fig .
[ wscaling2 ] ) . in order to place observational bounds on @xmath29 in such cases ,
we need to resort to numerical simulations without using the parametrization ( [ lhpara ] ) .
in this section we place observational constraints on the three types of quintessence models separately .
we use the recent sn ia data called union 2.1 dataset @xcite , shift parameters provided by wmap7 @xcite , and the bao distance measured by sdss7 @xcite and by boss @xcite . in order to make the analysis simpler , we fix , for all the models under consideration , the today s radiation density parameter to be equal to the one of the @xmath7cdm model . in sn
ia observations the luminosity distance @xmath151 is measured by the difference ( distance modulus ) of the apparent magnitude @xmath152 and the absolute magnitude @xmath153 , as ( z ) m(z)-m=5 _ 10 . for the observed distance modulus @xmath154 with the errors @xmath155
, the chi square of the sn ia measurement is given by _
sn ia^2=_i , where @xmath156 is the theoretical value of @xmath157 known for a given dark energy model . the position of the cmb acoustic peaks is determined by the following parameter @xcite l_a= , where @xmath158 is the redshift at the decoupling epoch , @xmath159 is the comoving angular diameter distance to the last scattering surface ( @xmath160 is the matter density parameter today ) , and = _ 0^z _ * .
the sound horizon @xmath161 is defined by r_s(z_*)=_z_*^. here @xmath162 and @xmath163 are the today s density parameters of baryons and photons , respectively . for the redshift
@xmath158 we use the fitting formula of hu and sugiyama @xcite .
, two parameters @xmath164 and @xmath158 depend primarily on @xmath165 and @xmath162 , respectively .
while @xmath166 characterizes the position of the cmb acoustic peaks , the parameters @xmath165 and @xmath162 are mostly related to the amplitudes of the peaks @xcite .
it is possible to employ the parameter sets ( @xmath167 ) in the likelihood analysis ( as in ref .
[ 44 ] ) , but we use the parameter sets ( @xmath168 ) because those are the parameters that the wmap team provides an approximate covariance matrix for . ]
the chi square associated with the wmap7 measurement is _ cmb^2=x_cmb^t _
cmb^-1x_cmb , where @xmath169 , and the covariance matrix is given by @xcite _ cmb= ( ccc 0.58269 & 0.00274801 & 0.318613 + 0.00274801 & 0.000338358 & 0.0122901 + 0.318613 & 0.0122901 & 0.824753 ) , and the inverse covariance matrix is _
cmb^-1= ( ccc 2.305 & 29.698 & -1.333 + 29.698 & 6825.27 & -113.18 + -1.333 & -113.18 & 3.414 ) . in bao observations
the ratio @xmath170 is measured , where @xmath171 is the sound horizon at which the baryons are released from the compton drag of photons and @xmath172 is the effective bao distance defined by @xmath173^{1/3}$ ] @xcite . for the redshift
@xmath174 we use the fitting formula of eisenstein and hu @xcite .
the chi square of the sdss7 measurement is given by _
bao , sdss7 ^ 2=x_bao^t _ bao^-1x_bao , where @xmath175 .
the covariance matrix is given by @xcite _ bao= ( cc 3.743610 ^ -5 & 7.414810 ^ -6 + 7.414810 ^ -6 & 1.296610 ^ -5 ) , and the inverse covariance matrix is _ bao^-1= ( cc 30124 & -17227 + -17227 & 86977 ) .
we also use the bao data from the wigglez and 6dfgs surveys , for which @xmath176 @xcite , and @xmath177 @xcite , where @xmath178 is defined as @xmath179 .
finally , we use the latest and most precise data point from the boss experiment @xcite , for which 1/r_bao(z=0.57 ) = 13.67 0.22 .
[ bossdata ] note that the error bar of this data is less than 1.7 % .
this is the most precise distance measurement ever constrained from a galaxy survey .
moreover , as we will see below , this data puts a severe upper bound of @xmath0 close to @xmath2 at @xmath180 .
the total chi - square from the three datasets is ^2=_snia^2 + _ cmb^2+_bao^2 , [ chis ] where the best - fit corresponds to the lowest value of @xmath181 .
let us study observational constraints on thawing models given by the equation of state ( [ wtha ] ) . since @xmath68 is weakly constrained in the three - parameter analysis , we first study this model by fixing the value of @xmath68 and vary the two parameters @xmath3 and @xmath182 .
this is the approach taken by dutta and scherrer @xcite .
however , because of a mild dependence ( i.e. degeneracy ) of the @xmath181 as a function of the parameter @xmath68 , we will also perform the data analysis by marginalizing the @xmath181 over @xmath68 itself .
this last procedure allows us , by hiding the information of @xmath68 , to understand the measured parameters @xmath65 and @xmath3 easily .
although quintessence corresponds to the case @xmath183 , we also extend to the regime @xmath184 in the likelihood analysis .
in fact it was shown in ref .
@xcite that the equation of state of a phantom scalar field can be accommodated by the analytic formula ( [ wtha ] ) .
the likelihood results are quite different depending on whether the recent bao data from boss @xcite are included or not , so we present two constraints with / without the boss data .
let us first focus our analysis on some particular values of @xmath68 , which allow us to make comparison with the results given in @xcite .
our numerical analysis gives that , in the absence of the boss bao data , the dark energy equation of state today is constrained to be @xmath185 ( at 95 % cl for @xmath186 ) .
dutta and scherrer @xcite showed that even the value @xmath187 is allowed from the sn ia data alone . adding the wmap7 and sdss7 data gives rise to much tighter bounds on @xmath3 .
this is the case for @xmath65 as well .
we obtained the bound @xmath188 ( 95 % cl for @xmath186 ) , whereas the values @xmath189 are allowed in ref .
@xcite .
we also study the same case _ with _ the latest boss data .
having so far fixed @xmath68 , by studying the @xmath181 over the parameter space of the remaining two parameters , we find that @xmath3 and @xmath65 are constrained to be @xmath190 and @xmath191 , respectively , at 95 % cl for @xmath186 .
rather surprisingly , the allowed parameter space in the regime @xmath192 is very narrow .
in particular the @xmath7cdm model , which corresponds to @xmath193 , is outside the @xmath194 observational contour .
this comes from the fact that the boss data ( [ bossdata ] ) does not allow a large parameter space with @xmath195 at the redshift @xmath180 .
we have also studied the case @xmath196 and derived bounds for @xmath3 and @xmath65 . without the boss data we obtain the bounds @xmath197 and @xmath198 ( 95 % cl , @xmath196 ) , whereas with the boss data these parameters are constrained to be @xmath199 and @xmath200 ( 95 % cl , @xmath196 ) respectively . in the former case
the upper bound of @xmath3 gets larger than that for @xmath186 .
this reflects the fact that @xmath0 changes more rapidly at late times for larger @xmath68 . in the presence of the boss data ,
however , the allowed parameter space in the regime @xmath183 is still very tiny .
this same qualitative behavour holds for all the different values of @xmath68 considered in our analysis , as we have found that the @xmath181 does not vary significantly as a function of @xmath68 .
@xmath194 ( red ) and @xmath201 ( green ) observational contours in the ( @xmath202 ) plane marginalized over @xmath65 .
we set the prior @xmath203 .
the dot - dashed and dotted curves correspond to the @xmath194 and @xmath201 constraints without the boss data , respectively.,width=307,height=288 ] @xmath194 ( red ) and @xmath201 ( green ) observational contours in the ( @xmath204 ) plane marginalized over @xmath68 .
we set the prior @xmath203 .
the dot - dashed and dotted curves correspond to the @xmath194 and @xmath201 constraints without the boss data , respectively.,width=307,height=288 ] we also vary the three parameters @xmath3 , @xmath65 , and @xmath68 in the likelihood analysis with the prior @xmath203 updating the analysis made in ref .
we have fixed this prior for @xmath68 because , for @xmath205 , the analytic expression ( [ wtha ] ) is not completely reliable because of the rapid rolling down of the field along the potential with carefully chosen initial conditions . with the boss data taken into account ,
the best - fit model parameters are found to be @xmath206 , @xmath207 , and @xmath208 with @xmath209 . under the prior @xmath192
the best - fit parameters reduce to those in the @xmath7cdm model as @xmath181 has a minimum at @xmath193 . although the case @xmath184 is plagued by a ghost problem , the dynamics of @xmath0 given phenomenologically by eq .
( [ wtha ] ) is able to fit the data quite well for @xmath210 .
in fact , with two parameters more than those in the @xmath7cdm , according to the akaike information criterion ( aic ) ( where we should add twice the number of free parameters @xmath211 to the original @xmath181 ) @xcite ) , aic should be modified to the sugiura s criterion : @xmath212 , where @xmath211 is the number of free parameters @xcite . since in our case
@xmath213 is large ( @xmath214 ) and @xmath211 is of the order of 1 , the difference between two criteria is very small . ]
, the best - fit corresponds to @xmath215 , whereas in the @xmath7cdm @xmath216 .
therefore the model with @xmath184 , even with three parameters , can compete with the @xmath7cdm . in fig .
[ thawconk ] we show observational constraints in the ( @xmath202 ) plane marginalized over @xmath182 . in the regime
@xmath217 the constraints on @xmath3 are practically independent of @xmath68 , i.e. , @xmath218 ( 95 % cl ) with the boss data , which should be compared with the previous result of ref . @xcite ( slightly updating the result in ref .
@xcite ) , @xmath219 for @xmath220 . in the presence of the boss data the allowed region shifts toward @xmath0 less than @xmath2 , as we see in fig .
[ thawconk ] .
for @xmath79 the lower bound on @xmath0 gets smaller with increasing @xmath68 , whereas the upper bound on @xmath0 is practically unchanged .
if @xmath221 , for example , @xmath3 is constrained to be @xmath222 ( 95 % cl ) .
for @xmath68 larger than the order of 1 the field equation of state can rapidly increase in low redshifts , but such rapid growth of @xmath0 is strongly disfavored from the boss data .
if we do not take into account the boss data , the growth of @xmath0 away from @xmath193 can be still allowed . for @xmath205 the analytic expression ( [ wtha ] )
is not completely reliable because of the rapid rolling down of the field along the potential with carefully chosen initial conditions .
finally , in fig . [ thawcon ] we plot observational constraints in the ( @xmath223 ) plane marginalized over @xmath68 with the prior @xmath203 .
also in this case , we find the same trend already mentioned above , namely , in the presence of the boss data , the allowed region shifts toward the values of @xmath0 less than @xmath2 , as we see in fig .
[ thawcon ] . after the marginalization over @xmath68
we obtain the bounds @xmath224 and @xmath225 ( 95 % cl ) . if we put the prior @xmath183 , we find that @xmath3 is constrained to be @xmath226 ( 68 % cl ) and @xmath227 ( 95 % cl ) .
let us proceed to observational constraints on tracker solutions whose equation of state is given by eq .
( [ wtrack ] ) .
although @xmath4 is theoretically larger than @xmath2 for quintessence , we do not put prior @xmath228 in the actual likelihood analysis . in fig .
[ trackercon ] we show @xmath194 and @xmath201 observational contours in the @xmath229 plane .
@xmath194 ( red ) and @xmath201 ( green ) observational contours on tracking freezing models in the ( @xmath230 ) plane .
the dot - dashed and dotted curves correspond to the @xmath194 and @xmath201 constraints without the boss data , respectively.,width=316,height=288 ] without the bao data , the tracker equation of state is constrained to be @xmath231 ( 95 % cl ) .
meanwhile the analysis of ref .
@xcite based on the constitution sn ia and the sdss bao data gives the bound @xmath232 ( 95 % cl ) .
hence the upper bound of @xmath4 becomes tighter by including the wmap7 data of cmb shift parameters .
for the potential @xmath99 ( @xmath23 ) the @xmath201 constraint @xmath5 translates into @xmath233 .
if we take into account the boss data in the analysis , the best - fit model parameters are found to be @xmath234 , and @xmath235 with @xmath236 . in this case
the akaike criterion gives @xmath237 , which is smaller than the @xmath7cdm value @xmath238 with the difference more than 2 .
the @xmath201 observational bounds are found to be @xmath239 and @xmath240 ( 95 % cl ) .
the upper bound of @xmath4 is very close to @xmath2 , which shows that the tracking quintessence away from @xmath2 is strongly disfavored from the data . if we put the prior @xmath228 in the analysis with the boss data , we find that the best - fit is obtained for @xmath241 and the model coincides with the standard @xmath7cdm case .
we then obtain the upper bound @xmath242 ( 95 % cl ) .
@xmath194 ( red ) and @xmath201 ( green ) observational contours on @xmath143 and @xmath243 for the parametrization ( [ lhpara ] ) of the scaling solution with @xmath134 , @xmath135 , and @xmath137 .
the dot - dashed and dotted curves correspond to the @xmath194 and @xmath201 constraints without the boss data , respectively .
, width=307,height=278 ] @xmath194 ( red ) and @xmath201 ( green ) observational contours on @xmath28 and @xmath243 for the potential ( [ doublepo ] ) with @xmath130 .
the dot - dashed and dotted curves correspond to the @xmath194 and @xmath201 constraints without the boss data , respectively.,width=307,height=278 ] for the scaling models characterized by the potential ( [ doublepo ] ) with @xmath130 we already showed in sec .
[ scalingfresec ] that the evolution of @xmath0 can be well approximated by eq .
( [ lhpara ] ) .
this parametrization admits an exact solution of @xmath244 in terms of @xmath10 and the other free four parameters , which is numerically convenient for analyzing the data @xcite . in fig .
[ scalingcon1 ] we plot observational bounds in the ( @xmath245 ) plane derived by using the parametrization ( [ lhpara ] ) with @xmath246 , and @xmath137 .
if the boss data are taken into account in the analysis , the transition redshift is constrained to be @xmath247 ( 95%cl ) .
the minimum of @xmath181 is found to be @xmath248 with @xmath249 , i.e. , the @xmath7cdm limit .
the case ( a ) shown in fig .
[ wscaling ] corresponds to the marginal one in which the model is within the @xmath201 observational contour .
this means that @xmath0 needs to approach @xmath2 in an early cosmological epoch ( @xmath250 for the redshift @xmath251 ) . even without the boss data the upper bound on @xmath143 is practically unchanged , which reflects the fact that @xmath0 is close to @xmath2 at low redshifts . for @xmath130
we also carry out the likelihood analysis without resorting to the parametrization ( [ lhpara ] ) .
numerically we solve the background equations of motion by tuning initial conditions to find the evolution which gives the desired values of @xmath160 and @xmath252 ( radiation density parameter ) for the potential @xmath253 .
we put the prior @xmath123 coming from the constraint of big bang nucleosynthesis .
as long as @xmath254 the solutions approach the scaling fixed point @xmath255 during the matter era , so the initial conditions of @xmath256 and @xmath257 are irrelevant to the likelihood analysis . in fig .
[ scalingcon2 ] we show observational bounds on the parameters @xmath28 and @xmath160 .
the parameter @xmath28 is constrained to be @xmath258 ( 95%cl ) , which is consistent with the results presented in fig .
[ scalingcon1 ] . for larger @xmath28
the transition from @xmath33 to @xmath6 occurs earlier , so that the models are favored from the data .
@xmath194 ( red ) and @xmath201 ( green ) observational contours on @xmath28 and @xmath29 for the potential ( [ doublepo ] ) with @xmath259 .
the dot - dashed and dotted curves correspond to the @xmath194 and @xmath201 constraints without the boss data , respectively.,width=307,height=278 ] for non - zero values of @xmath29 we do not have an analytic expression of @xmath0 , so we solve the background equations of motion numerically with the priors @xmath123 and @xmath260 . we set the latter prior because the @xmath130 case was already discussed above . varying the three parameters @xmath28 , @xmath29 , and @xmath160 , we find that the best - fit model parameters are @xmath261 , @xmath262 , and @xmath263 with @xmath264
. then the akaike criterion gives @xmath265 , which is larger than the @xmath7cdm value @xmath238 with the difference more than 4 .
we also obtain the following bounds @xmath266 figure [ scalingcon3 ] shows the observational constraints in the @xmath267 plane for @xmath263 . if the three parameters @xmath28 , @xmath29 , @xmath160 are varied in the likelihood analysis , it is difficult to marginalize over @xmath160 in the range @xmath268 because the solutions are prone to numerical instabilities around the tail regions of @xmath160 .
hence we use the fixed density parameter @xmath263 , which corresponds to the best - fit value when the three parameters are varied . if @xmath269 , the model is excluded at 95 % cl .
this is associated with the fact that @xmath0 possesses a minimum for larger @xmath29 ( see fig . [ wscaling2 ] ) .
therefore , in the context of this model , the expansion of the universe accelerates forever .
if @xmath147 , for example , the parameter @xmath29 is constrained to be @xmath270 ( 95 % cl ) .
in fact the minimum of @xmath0 appears for @xmath150 , which leads to the deviation of @xmath0 from @xmath2 today . in summary ,
the model is within the @xmath201 observational contour provided that @xmath271 and @xmath30 .
relative deviation of the hubble parameter , @xmath272 , for the best - fit cases of : ( i ) scaling ( dot - dashed line ) , ( ii ) thawing ( dotted line ) , ( iii ) tracker ( solid line ) models .
the relative deviations from the @xmath7cdm model are less than 2.5% for @xmath273 .
the difference in the evolution among these models during dust domination , is due to the fact that different model parameters lead to different values of @xmath160 for the best fit ( in the @xmath7cdm we have @xmath274 , whereas the @xmath160 s for the other models exceed this value by about 4%).,width=307,height=278 ] in fig .
[ ploth ] we plot the evolution of the relative deviation of the hubble parameter from the @xmath7cdm as a function of @xmath10 for three best - fit models studied in this paper .
the deviation @xmath275 of the best - fit scaling model is very small , which reflects the fact that @xmath0 quickly approaches @xmath2 after the early transition from the scaling regime .
meanwhile , in the best - fit thawing and tracker models , the relative deviations from the @xmath7cdm can reach the level of 2% .
in this paper we placed observational bounds on three types of quintessence models : ( i ) thawing , ( ii ) tracker , and ( iii ) scaling models .
we used the recent data of sn ia from union 2.1 , the cmb shift parameters from wmap7 , and bao from sdss7 and boss , by which the background cosmic expansion history is constrained from the distance measurements . in thawing models where the field starts to evolve at the late cosmological epoch ,
the dark energy equation of state can be expressed as eq .
( [ wtha ] ) with the three parameters @xmath3 , @xmath182 , and @xmath68 .
the parameter @xmath68 is related to the field mass at the initial stage .
we put observational bounds in the @xmath276 after marginalizing over @xmath68 ( see fig .
[ thawcon ] ) .
if we take into account the boss data ( [ bossdata ] ) in the likelihood analysis , the upper bounds on @xmath277 are very close to @xmath2 independent of the values of @xmath68 ranging in the region @xmath203 ( under which the analytic formula ( [ wtha ] ) is trustable ) . without
the boss data the deviation of @xmath0 away from @xmath2 today can be still allowed , as seen in fig .
[ thawcon ] . in tracking freezing models where @xmath0 is nearly constant ( @xmath278 ) during the matter era
, there is the analytic formula ( [ wtrack ] ) derived by considering a homogeneous perturbation around the tracker . without
the boss data the tracker equation of state is constrained to be @xmath231 ( 95 % cl ) , whose upper bound is tighter than @xmath279 derived by using sn ia and sdss7 data @xcite .
inclusion of the boss data gives the upper bounds of @xmath4 close to @xmath2 .
we find that without the prior on @xmath4 the constraint is @xmath239 ( 95 % cl ) and with the quintessence prior @xmath280 the upper bound is @xmath242 ( 95 % cl ) . for the potential ( [ doublepo ] ) with @xmath30 and @xmath281 , @xmath0 is close to 0 during the deep matter era because of the dominance of the steep potential @xmath282 .
the field equation of state starts to decrease after the potential @xmath35 dominates over @xmath282 . for larger @xmath28
the exit from the scaling regime ( @xmath33 ) occurs earlier . when @xmath130 we found that the evolution of @xmath0 can be approximated by the parametrization ( [ lhpara ] ) with @xmath134 , @xmath135 , and @xmath283 , where the transition scale factor @xmath143 depends on @xmath28 . using this parametrization we derived the bound @xmath247 ( 95%cl ) , which translates into the constraint @xmath258
this is consistent with the bound shown in fig .
[ scalingcon1 ] derived by solving the background equations of motion numerically
. for the potential ( [ doublepo ] ) with non - zero values of @xmath29 , the field equation of state tends to have a minimum for larger @xmath29 . as we see in fig .
[ scalingcon3 ] , the models with @xmath269 is excluded at 95%cl . the parameters @xmath271 and @xmath284 are allowed from the data , which shows that the early transition from the scaling regime to the regime close to @xmath6 is favored . for the scaling models the observational constraints on @xmath28 and @xmath29 are not very sensitive to inclusion of the boss data , because @xmath0 evolves toward @xmath2 at the late cosmological epoch . as we see in fig .
[ ploth ] , the difference of @xmath20 between the three best - fit quintessence models and the @xmath7cdm model is only less than 2.5 % in the past . in current observations
there is no strong evidence that quintessence is favored over the @xmath7cdm from the statistical point of view .
this property is especially significant by including the boss bao data .
if we extend the analysis in the regime @xmath285 , we showed that some of the models studied in this paper can compete with the @xmath7cdm model according to the akaike information criterion .
we note that in many modified gravity models such as @xmath286 gravity @xcite , ( extended ) galileons @xcite , and lorentz violating theories @xcite it is possible to realize @xmath285 without having ghosts and instabilities .
it remains to see how much extent future high - precision observations can constrain dark energy models from the background expansion history as well as from the cosmic growth of density perturbations .
ad thanks jsps for financial support to visit tokyo university of science ( no . s12135 ) .
st thanks warm hospitalities during his stays in weihai , observatorio nacional in rio de janeiro , passa quatro , szczecin , and university of heidelberg . this work was supported in part by the grant - in - aid for scientific research fund of the jsps no .
24540287 ( tc ) and no .
24540286 ( st ) and scientific research on innovative areas ( no .
21111006 ) ( st ) and in part by nihon university ( tc ) .
y. fujii , phys .
d * 26 * , 2580 ( 1982 ) ; l. h. ford , phys . rev .
d * 35 * , 2339 ( 1987 ) ; c. wetterich , nucl .
phys b. * 302 * , 668 ( 1988 ) ; t. chiba , n. sugiyama and t. nakamura , mon . not .
soc . * 289 * , l5 ( 1997 ) ; p. g. ferreira and m. joyce , phys .
lett . * 79 * , 4740 ( 1997 ) ; r. r. caldwell , r. dave and p. j. steinhardt , phys .
lett . *
80 * , 1582 ( 1998 ) .
v. sahni and a. a. starobinsky , int .
j. mod .
d * 9 * , 373 ( 2000 ) ; s. m. carroll , living rev .
rel . * 4 * , 1 ( 2001 ) ; t. padmanabhan , phys .
* 380 * , 235 ( 2003 ) ; p. j. e. peebles and b. ratra , rev .
* 75 * , 559 ( 2003 ) ; e. j. copeland , m. sami and s. tsujikawa , int .
j. mod .
d * 15 * , 1753 ( 2006 ) ; t. p.
sotiriou and v. faraoni , rev .
* 82 * , 451 ( 2010 ) ; a. de felice and s. tsujikawa , living rev .
* 13 * , 3 ( 2010 ) ; s. tsujikawa , arxiv:1004.1493 [ astro-ph.co ] . p. binetruy , phys .
d * 60 * , 063502 ( 1999 ) ; p. brax and j. martin , phys .
b * 468 * , 40 ( 1999 ) ; e. j. copeland , n. j. nunes and f. rosati , phys .
d * 62 * , 123503 ( 2000 ) ; p. k. townsend , jhep * 0111 * , 042 ( 2001 ) ; s. hellerman , n. kaloper and l. susskind , jhep * 0106 * , 003 ( 2001 ) ; r. kallosh , a. d. linde , s. prokushkin and m. shmakova , phys .
d * 65 * , 105016 ( 2002 ) ; m. gasperini , f. piazza and g. veneziano , phys
. rev .
d * 65 * , 023508 ( 2002 ) ; a. j. albrecht , c. p. burgess , f. ravndal and c. skordis , phys .
d * 65 * , 123507 ( 2002 ) .
y. nomura , t. watari and t. yanagida , phys .
b * 484 * , 103 ( 2000 ) ; k. choi , phys .
d * 62 * , 043509 ( 2000 ) ; j. e. kim and h. p.
nilles , phys .
b * 553 * , 1 ( 2003 ) ; l. j. hall , y. nomura and s. j. oliver , phys .
rev . lett .
* 95 * , 141302 ( 2005 ) ; s. panda , y. sumitomo and s. p. trivedi , phys .
d * 83 * , 083506 ( 2011 ) .
t. d. saini , s. raychaudhury , v. sahni and a. a. starobinsky , phys .
* 85 * , 1162 ( 2000 ) ; z. haiman , j. j. mohr and g. p. holder , astrophys .
j. * 553 * , 545 ( 2000 ) ; c. baccigalupi , a. balbi , s. matarrese , f. perrotta and n. vittorio , phys . rev .
d * 65 * , 063520 ( 2002 ) ; m. yahiro , g. j. mathews , k. ichiki , t. kajino and m. orito , phys .
d * 65 * , 063502 ( 2002 ) ; u. alam , v. sahni , t. d. saini and a. a. starobinsky , mon . not .
soc . * 344 * , 1057 ( 2003 ) ; r. r. caldwell and m. doran , phys .
d * 69 * , 103517 ( 2004 ) ; l. perivolaropoulos , phys .
d * 71 * , 063503 ( 2005 ) s. nesseris and l. perivolaropoulos , phys .
d * 72 * , 123519 ( 2005 ) ; d. huterer and h. v. peiris , phys .
d * 75 * , 083503 ( 2007 ) ; m. sahlen , a. r. liddle and d. parkinson , phys .
d * 75 * , 023502 ( 2007 ) ; s. sen , a. a. sen and m. sami , phys .
b * 686 * , 1 ( 2010 ) ; s. thakur , a. nautiyal , a. a. sen and t. r. seshadri , arxiv:1204.2617 [ astro-ph.co ] . b. a. bassett , m. kunz , j. silk and c. ungarelli , mon . not .
soc . * 336 * , 1217 ( 2002 ) ; p. s. corasaniti and e. j. copeland , phys .
d * 67 * , 063521 ( 2003 ) ; b. a. bassett , p. s. corasaniti and m. kunz , astrophys .
j. * 617 * , l1 ( 2004 ) .
w. hu and i. sawicki , phys .
d * 76 * , 064004 ( 2007 ) ; l. amendola and s. tsujikawa , phys .
b * 660 * , 125 ( 2008 ) ; h. motohashi , a. a. starobinsky and j. i .
yokoyama , prog .
phys . * 123 * , 887 ( 2010 ) . | for two types of quintessence models having thawing and tracking properties , there exist analytic solutions for the dark energy equation of state @xmath0 expressed in terms of several free parameters .
we put observational bounds on the parameters in such scenarios by using the recent data of supernovae type ia ( sn ia ) , cosmic microwave background ( cmb ) , and baryon acoustic oscillations ( bao ) .
the observational constraints are quite different depending on whether or not the recent bao data from boss are taken into account . with the boss data the upper bounds of today s values of @xmath0 ( @xmath1 ) in thawing models
is very close to @xmath2 , whereas without this data the values of @xmath3 away from @xmath2 can be still allowed .
the tracker equation of state @xmath4 during the matter era is constrained to be @xmath5 at 95 % confidence level ( cl ) even without the boss data , so that the tracker models with @xmath0 away from @xmath2 are severely disfavored .
we also study observational constraints on scaling models in which @xmath0 starts to evolve from 0 in the deep matter era and show that the transition to the equation of state close to @xmath6 needs to occur at an early cosmological epoch . in the three classes of quintessence models studied in this paper ,
the past evolution of the hubble parameters in the best - fit models shows only less than the 2.5 % difference compared to the @xmath7cdm . |
understanding the non valence quark content of hadrons remains , despite a history spanning over two decades , a major theoretical challenge .
the interest in non valence physics derives mainly from the unique opportunities which it provides for new insights into quantum aspects of hadron structure beyond the naive and spectroscopically very successful quark model .
important questions in this realm include modifications of the qcd vacuum inside hadrons @xcite , the mechanism of flavor mixing and the origin of the ozi
rule @xcite , the structure of constituent quarks @xcite , and the role of gluons in the dynamics of the ( isoscalar ) strange quark sea in the nucleon @xcite . the strangeness distribution inside of the nucleon @xcite represents the most intensely studied example of a hadronic non valence quark effect .
currently the vector channel of this distribution , as described by the strange vector form factors , is a focus of experimental @xcite and theoretical @xcite research . since systematic and model independent approaches have still little predictive power for the nucleon s strangeness content ( as exemplified by studies in chiral perturbation theory @xcite and on the lattice @xcite ) , most previous and current theoretical analyses of the strangeness form factors were model based . among the first and most transparent models for the vector form factors were those which implement a kaon cloud of the nucleon @xcite and thus complement pole dominance approaches @xcite . in kaon
cloud models the nucleon s strangeness distribution is generated by fluctuations of the `` bare '' ( i.e. nonstrange ) nucleon into kaon
hyperon intermediate states which are described by the corresponding one loop feynman graphs @xcite .
the two crucial assumptions underlying the loop model are 1 ) that the lightest valence strangeness carrying intermediate states generate the dominant contribution to the strangeness content and hence give at least a rough estimate of its size , and 2 ) that rescattering ( i.e. multi loop ) contributions are suppressed ( despite large couplings ) . both of these assumptions have recently been challenged .
a dispersive analysis on the basis of analytically continued @xmath3 scattering data demonstrated that rescattering corrections are important even at low momentum transfers , both to restore unitarity and to build up resonance strength in the @xmath4 meson region @xcite .
furthermore , a study in an `` unquenched '' quark model found the contributions from higher lying intermediate states ( up to surprisingly large invariant masses ) indispensable for the calculation of the strange quark distribution @xcite and prompted our collaborators and us to investigate these issues in a complementary hadronic one loop model @xcite .
in addition to the original @xmath5 loops ( @xmath6 ) , we included the next higher lying intermediate states , i.e. the @xmath7 pairs . in the first part of this
study the corresponding loops were evaluated using bonn
jlich @xmath8 form factors ( see below ) , as in the original @xmath5 model .
the results were discouraging : the @xmath0 contributions were found to be larger than those from the kaon loop , and their dispersive analysis indicated strong unitarity violations .
the anomalously large and apparently unrealistic @xmath0 contributions could be traced to the large @xmath0 tensor couplings and , in particular , to the very large cutoff parameter of the @xmath9 form factor taken from the bonn
jlich potential model @xcite .
since both the nijmegen potential @xcite and the forthcoming update of the bonn jlich @xmath10 potential @xcite find substantially smaller values for this cutoff , we feel that a detailed and quantitative analysis of the cutoff ( and coupling ) dependence of the @xmath0 contributions ( covering the whole range of so far proposed values ) would be a useful contribution to the ongoing discussion @xcite of hadron loop model applications to nucleon strangeness .
such an analysis is one of the main objectives of the present paper .
furthermore , we will calculate and discuss the momentum dependence of the strange vector form factors in the loop model at low momentum transfers ( @xmath11 gev ) relevant for the present and planned measurements at mit - bates , tjnaf and mami .
we begin by recapitulating the definition of the strange vector form factors , the pertinent features of the hadronic loop model @xcite , and its extension to additional intermediate states containing @xmath0 mesons @xcite .
the focus of our investigations will be on the nucleon matrix element of the strangeness current , which is parametrized by two invariant amplitudes , the dirac and pauli strangeness form factors @xmath12 , @xmath13 u(p)\ .\ ] ] here @xmath14 denotes the nucleon spinor and @xmath15 , due to the absence of an overall strangeness charge of the nucleon .
the leading nonvanishing moments of these form factors are the dirac and sachs ( square)strangeness radii @xmath16 as well as the strangeness magnetic moment @xmath17 ( the sachs radius is obtained from the electric sachs form factor @xmath18 and related to the dirac radius by @xmath19 . )
as mentioned above , we have chosen a hadronic one loop model containing @xmath20 and @xmath0 mesons as the dynamical framework for the calculation of these moments .
this model is based on the meson
baryon effective lagrangians @xmath21 \ ; , \label{la}\end{aligned}\ ] ] where @xmath22 ( @xmath23 ) , @xmath20 , and @xmath24 are the baryon , kaon , and @xmath0 vector - meson fields , respectively , @xmath25 mev is the nucleon mass and @xmath26 is the ratio of tensor to vector coupling , @xmath27 . in order to account for the finite extent of the above vertices , the model includes form factors from the bonn
jlich @xmath28 potential @xcite at the hadronic @xmath29 and @xmath8 ( @xmath30 ) vertices , which have the monopole form @xmath31 with meson momenta @xmath32 and the physical meson masses @xmath33 mev and @xmath34 mev @xcite .
these form factors render all encountered loop integrals finite and reproduce the on shell values of the mesonic couplings .
the range of currently favored values for the couplings @xmath35 and cutoff parameters @xmath36 and @xmath37 , as well as their impact on the strangeness observables , will be discussed below .
since the non - locality of the meson - baryon form factors ( [ ff ] ) gives rise to vertex currents , gauge invariance was maintained in @xcite by introducing the photon field via minimal substitution in the momentum variable @xmath32 @xcite .
( consequences of the non uniqueness of this prescription are discussed in refs .
the resulting nonlocal seagull vertices are given explicitly in @xcite .
the diagonal couplings of @xmath38 to the strange mesons and baryons in the intermediate states are straightforwardly determined by current conservation , i.e. they are given by the net strangeness charge of the corresponding hadron .
the situation is more complex for the non diagonal ( i.e. spin flipping ) coupling @xmath39 of the strange current to @xmath20 and @xmath0 , which is defined by the transition matrix element @xmath40 ( where @xmath41 and @xmath42 are isospin indices and @xmath43 is the polarization vector of the @xmath0 ) .
this coupling was estimated in @xcite on the basis of the vector meson dominance model of ref .
@xcite , with the result @xmath44 .
the relevant one loop feynman graphs of the model are completely determined by the above vertices and the standard meson and baryon propagators . the diagrams containing strange mesons fall into three categories , corresponding to fluctuations of the nucleon into either @xmath5 or @xmath7 pairs , or to the strangeness current induced spin flip transition from a @xmath5 to a @xmath7 intermediate state .
the explicit expressions for the corresponding loop amplitudes are collected in appendix a. in the following calculations we will concentrate on the @xmath45 contributions and omit the comparatively negligible @xmath46 contribution @xcite .
in this section we investigate the dependence of the loop - model results on the couplings and cutoffs which parametrize the internal @xmath47 vertices . in our previous analysis
@xcite we have used the ( @xmath48 based ) couplings @xmath49 , @xmath50 , @xmath51 and the cutoff parameter values @xmath52 ( 2.1 ) , @xmath53 gev of the bonn
jlich @xmath10 potential @xcite .
the cutoffs were determined from hyperon
nucleon scattering data , with the numbers in parenthesis denoting values obtained in an alternative model for the baryon - baryon interaction .
the numerical results of the loop model @xcite are summarized in table i. a glance at these numbers shows that the magnitude of the @xmath0 contributions exceeds those of the @xmath20 contributions by factors of @xmath54 .
as already mentioned , the main reason for these unrealistically large contributions can been traced to the unusually large @xmath9 cutoff parameter @xmath55 gev found in @xcite .
it has twice the size of the typical hadronic scale @xmath56 gev around which such cutoff parameters lie normally .
a substantially larger value ( which in our case also exceeds the largest hadron masses in the loops by a factor of two ) must be considered suspect in any model with hadronic degrees of freedom since one expects their quark
gluon substructure to become relevant at such scales .
indeed , the appearance of anomalously large cutoffs in a potential model suggests that those cutoffs are burdened with short distance physics ( not directly related to the @xmath0 sector ) which would otherwise remain unaccounted for . literally taking such effects over to the loop
model estimates of the nucleon s strangeness content , by fully associating them with the physical @xmath57 would therefore very likely be misleading .
a hint that the @xmath0 sector of the original bonn
jlich potential might indeed be overburdened can be obtained from a comparison with the conceptually similar nijmegen @xmath10 potential @xcite .
the nijmegen potential contains more degrees of freedom in the scalar meson sector ( including a pomeron ) and finds a much smaller @xmath0 cutoff @xmath58 gev .
morevoer , a smaller @xmath0 cutoff is also favored in the forthcoming update of the bonn
jlich potential @xcite .
the @xmath0 cutoff of this model is expected to lie around @xmath59 gev @xcite .
hence , nowadays smaller cutoffs seem to be consistently favored by @xmath10 potential models .
this motivated our reanalysis of the loop
model results of ref .
@xcite for the nucleon s strangeness observables ( as also suggested in @xcite ) which we will discuss below . to get a qualitative idea of the range of coupling and cutoff values to be considered , and to maintain the underlying philosophy of the loop model
, we will orient ourselves at the values used in the existing potential models .
of course , these values depend to some extent on the particular dynamics and particle content of a given model , and since the various potential models and also our model differ in this respect , the corresponding parameter sets can not be directly compared or strictly related to each other . however , we expect the parameter ranges considered below , in particular for the cutoffs , to cover most of the physically reasonable parameter space of our model . we begin our discussion of the parameter dependence by documenting the sensitivity of the results to variations in the cutoff . to this end , we display the cutoff dependence in the range @xmath60 , which covers all so far proposed values for @xmath37 .
the first two figures show the behavior of the dirac strangeness radius ( fig .
1a ) and the strangeness magnetic moment ( fig .
1b ) as a function of @xmath61 , with the @xmath0 couplings fixed at the old , @xmath48based bonn jlich values given above . the full curves show the total results , the dashed ones represent the @xmath62 contributions , the dash dotted ones the @xmath63 contributions , and the dotted curve corresponds to the result from the @xmath20 loop .
clearly , the cutoff dependence is very pronounced , reducing e.g. the value of ref .
@xcite for the magnetic moment by almost an order of magnitude for @xmath64 gev .
the reduction is even considerably stronger for the nijmegen value @xmath65 gev .
although a strong cutoff dependence was to be expected ( especially due to the enhanced degree of divergence of @xmath66the @xmath0 loops with derivative couplings ) , its actual magnitude is still surprising .
if cutoff sizes of the order of that used in the nijmegen potential were to be the most realistic , some of the conclusions reached in @xcite would have to be revised .
in particular , for cutoffs of the nijmegen size some sort of `` convergence '' of the intermediate state sum ( to one loop ) could not anymore be excluded . in this case
the kaon cloud contribution might well be sufficient for a first orientation about the overall size of the nucleon s strangeness content in hadronic one loop models , as advocated in @xcite and in contrast to the findings of ref .
@xcite in the quark model .
other problematic aspects of hadron loop models , such as the expected importance of rescattering corrections and unitarity violations @xcite , are of course not affected by these arguments .
the strong cutoff dependence of the strangeness observables exposes another , both conceptual and practical problem of hadronic loop models .
although the cutoff is principally a physical parameter ( which indicates up to which resolution a purely hadronic description might be adequate ) , the foundations of the approach are not solid enough to give it a precise and quantitative meaning . moreover
, the available @xmath10 scattering data do not allow an accurate determination of the @xmath8 form factors even in the framework of a specific potential model , not to mention other sources of uncertainty as , for example , the largely uncontrolled off shell ambiguities incurred by transplanting such form factors into another model context . as a consequence , the numerical value of the @xmath0 cutoff can not be accurately determined , and the corresponding uncertainty propagates , amplified by a hightened sensitivity , into the strangeness observables .
these facts already imply that at most semi quantitative predictions can be expected from the hadron loop approach .
the existing @xmath10 potential models differ not only in the momentum dependence of the @xmath8 form factors , but also in the values of the corresponding @xmath0 couplings .
the @xmath0 vector ( dirac ) coupling of the nijmegen potential is , for example , considerably smaller than the one of ref .
@xcite which we used above . in order to illustrate the dependence of the strange form factor moments on these couplings , we compare in fig .
2 the @xmath0 contributions to the strangeness radius ( fig . 2a ) and magnetic moment ( fig .
2b ) for the five pairs ( @xmath67 ) , ( @xmath68 ) , ( @xmath69 ) , ( @xmath70 ) , ( @xmath71 ) of coupling values ( @xmath72 ) .
the first of these pairs corresponds to the values of holzenkamp et al . while the others were chosen to encompass the range of values which appear in other potential models .
2 a , b demonstrate that the variations of the strangeness observables due to different choices for the couplings can be quite substantial for large values of the @xmath0 cutoff .
the sensitivity to the couplings remains fairly small , however , for @xmath61 of the order of @xmath73 gev , in particular for the strange magnetic moment and for couplings in the range between the bonn - jlich ( @xmath67 ) and nijmegen @xcite ( @xmath74 ) values . throughout all of the above calculations we have kept the values of ref .
@xcite for the cutoff and coupling of the kaon fixed .
the nijmegen potential uses a pseudovector coupling which can not be related without off shell ambiguities to the pseudoscalar coupling of the bonn
jlich potential . on
shell , the equivalent pseudoscalar coupling of the nijmegen potential ( @xmath75 ) differs by less than 5% from the bonn
jlich coupling used here , and the nijmegen cutoff parameter @xmath76 gev @xcite is similarly close to the one we use . in any case , the cutoff and coupling values from potential models should not be taken too literally since , as discussed above , the limited available data and the implicit model assumptions do not allow their precise ( and unique ) determination .
we close this section with a remark on alternative choices for the internal form factors .
the bonn monopole form factors ( [ ff ] ) have two well - known limitations : an artificial zero for cutoffs of the size of the meson mass ( due to on - shell normalization ) on which we will comment in the next section , and an unphysical singularity at time - like momenta @xmath77 .
although the singularity can be tamed by the usual causal boundary condition , it will still affect the values of the loop amplitudes to a certain extent .
this problem could be avoided , although at the price of additional model dependence , by choosing alternative , singularity - free extrapolations of the bonn form factors into the time - like region .
one such extrapolation was proposed in ref .
all of the currently running or planned experiments measure the strange vector form factors at @xmath78 .
therefore , their data need to be extrapolated in order to obtain the leading form - factor moments .
this extrapolation requires information on the momentum dependence of the form factors ( at low energies ) and thus on the spatial strangeness distribution inside the nucleon . in the following section we will evaluate the loop - model predictions for the momentum dependence by extending the above calculations to momentum transfers @xmath79 1 gev@xmath80 both strange vector form factors will be measured in about the same @xmath81-range by the planned spectrometer experiment g0 at jefferson lab @xcite . in the loop model , the coupling of the strange current to hyperon
, meson , @xmath82 vertex or @xmath83 transition vertex inside the loop is described by corresponding vertex functions , which we list in the appendix .
these vertex functions contain the full momentum dependence of the form factors and can be evaluated numerically .
fixing the cutoffs for the @xmath0 and the kaon at @xmath84 gev and using the nijmegen values for the coupling , we show in fig .
3 ( solid line ) the resulting strange magnetic form factor @xmath85 together with the data point measured by the sample experiment at @xmath86 gev@xmath87 @xcite .
( for comparison we have also included the older data point from the first runs only . )
figure 4 ( solid line ) shows the loop - model prediction for the combination @xmath88 measured by the happex collaboration , together with their data point at @xmath89 gev@xmath87 @xcite .
the corresponding loop - model results for the dirac strangeness radius and magnetic moment are given in table i. at first sight , the results are in disagreement with both the sample and the happex data .
the measured form factor combinations are both positive ( at their respective momentum transfers ) while the loop model leads to negative form factors of smaller slopes and magnitudes . in particular , the loop model predicts - like the majority of the other so far investigated models - a small and negative value for the strangeness magnetic moment . with regard to the sample result , however , there is a caveat .
the derivation of the strangeness form factor from the measured asymmetry requires knowledge of the neutral weak axial form factor @xmath90 of the nucleon and , in particular , its isovector radiative correction @xcite .
the newest published value @xmath91 gev@xmath92 is based on the theoretical estimate @xcite for this radiative correction , which carries a substantial amount of uncertainty .
the sample collaboration is currently measuring with a deuterium target @xcite in order to pin down the axial weak form factor contribution separately . only after taking these data or more accurate and reliable theoretical estimates into account ( which could substantially change the quoted value for @xmath91 gev@xmath93 both in sign an magnitude ) can the strange magnetic moment be extracted unambigously and compared to the small value of the loop - model prediction . furthermore , it is interesting to note that the results of the loop model could be made consistent with the happex data by allowing for smaller cutoff values , as suggested by recent applications of the meson cloud model to deeply inelastic scattering @xcite . in order to illustrate this point
, we show in figs . 3 and 4 ( dashed lines ) the results obtained by using @xmath94 gev , which also brings the magnetic form factor somewhat closer to the sample data point . moreover ,
this cutoff value is very close to the @xmath0 mass and therefore effectively switches off the internal @xmath0 form factors ( cf .
( [ ff ] ) ) . as a consequence ,
the contributions from the @xmath0 and the @xmath83 transition are completely negligible relative to the kaon contribution .
this shows that the @xmath0 contributions worsen the agreement of the loop model predictions with the current experimental results .
however , the zeros of the bonn form factors have to be regarded as artefacts of the on - shell normalization and the results for the small cutoffs should therefore be viewed with caution .
the latter reservations are enhanced by the difficulties with the physical interpretation of cutoffs of the same size as particle masses in a loop . in figs . 5 and 6 we show the @xmath20 , @xmath0 and @xmath83 transition contributions to the dirac ( @xmath95 ) and pauli ( @xmath96 ) form factors separately ( again for @xmath84 gev and nijmegen couplings ) .
these figures demonstrate that even with the new values for cutoffs and couplings adopted in the present paper , the @xmath0 and @xmath83 contributions are still of the same order of magnitude as the kaon contributions .
this is in contrast to the much smaller size of these contributions to the strangeness magnetic moment in the chiral quark model @xcite . in the latter
the tensor coupling of the @xmath0 to the quarks is small , while the contributions from the vector coupling partially cancel each other and become proportional to the difference between the light and strange quark constituent masses , and therefore also small .
finally , we present the loop - model result for the form - factor combination @xmath97 which the forthcoming mami a4 experiment @xcite plans to measure at @xmath98 gev@xmath87 . we find @xmath99 where the same reservations as above apply to the smaller cutoff value .
to summarize , we have analyzed the cutoff , coupling , and momentum dependence of the @xmath0 contributions to the nucleon s vector strangeness content in the hadronic one loop model of ref .
it turns out that the softer @xmath8 form factors now generally favored by @xmath10 potential models have some welcome consequences for such models .
first , they can reduce the @xmath0 contributions to the strangeness radius and magnetic moment by over an order of magnitude , thereby indicating that the contributions from the lightest @xmath100 intermediate states might be sufficient for rough estimates of the strangeness content in one loop models .
although the @xmath0 contributions remain non negligible ( in particular towards larger momentum transfers ) and worsen the agreement with the data of the happex experiment , they cease to be unrealistically large and they do not anymore exclude some sort of ( slow ) `` convergence '' of the intermediate state sum .
the very sensitive dependence of the results on the @xmath0 cutoff emphasizes , however , the limited and mostly qualitative character of hadron loop model predicitions .
h.f . would like to thank the national institute for nuclear theory in seattle for hospitality during the 1998 strangeness program , where the plan for this work originated .
he would also like to thank wally melnitchouk for useful correspondence on the bonn
jlich potential model , e. kolomeitsev for pointing out ref .
@xcite , and the deutsche forschungsgemeinschaft for support under habilitation grant fo 156/21 .
f.s.n . and m.n .
would like to thank fapesp and cnpq , brazil , for support .
in the loop model of section [ mod ] the strangeness form factors receive four distinct types of contributions .
three of them correspond to amplitudes associated with processes in which the current couples either to the hyperon line ( y ) , the meson line ( m ) , or the meson - baryon vertex ( v ) in the loop .
the fourth contribution corresponds to the amplitude associated with the strangeness - current induced spin - flip transition form @xmath20 to @xmath0 .
the corresponding vertex functions are @xmath101 & & \left.\left(\gamma_\beta - i{\frac{\kappa}{2m_n}}\sigma_{\beta\gamma } k^\gamma\right)-g_{ps}^2(f_k(k^2))^2\delta(k^2)\gamma_5 s(p^\prime - k ) \gamma_\mu s(p - k)\gamma_5\right]\ ; , \label{1b}\end{aligned}\ ] ] @xmath102 & + & \left.i{\frac{\kappa}{2m_n } } \sigma_{\alpha\nu}(k+q)^\nu\right ) [ ( 2k+q)_\mu \ , g_{\sigma\lambda}-(k+q)_\sigma g_{\lambda\mu}-k_\lambda g_{\sigma\mu}]\times \nonumber \\ * [ 7.2pt ] & & s(p - k)\left(\gamma_\beta - i{\frac{\kappa}{2m_n}}\sigma_{\beta\gamma } k^\gamma\right)+g_{ps}^2f_k((k+q)^2)f_k(k^2 ) \times \nonumber \\ * [ 7.2pt ] & & \left.\delta((k+q)^2)\delta(k^2)(2k+q)_\mu\gamma_5s(p - k)\gamma_5 \right]\ ; , \label{1m}\end{aligned}\ ] ] @xmath103 & & \left(\gamma_\alpha + i{\frac{\kappa}{2m_n}}\sigma_{\alpha\nu}k^\nu\right ) s(p - k)\left(\gamma_\beta - i{\frac{\kappa}{2m_n}}\sigma_{\beta\gamma}k^\gamma\right ) - \frac { ( q-2k)_\mu } { ( q - k)^2-k^2 } ( f(k^2)+ \nonumber \\ * [ 7.2pt ] & -&\left.f((k - q)^2))\left(\gamma_\alpha + i{\frac{\kappa}{2m_n}}\sigma_{\alpha\nu } k^\nu\right ) s(p^\prime - k)\left(\gamma_\beta - i{\frac{\kappa}{2m_n}}\sigma_{\beta\gamma}k^\gamma \right ) \right ) \ ; + \nonumber \\ * [ 7.2pt ] & + & \;{\frac{\kappa}{2m_n}}\left(f((k+q)^2)\sigma_{\alpha\mu } s(p - k)\left(\gamma_\beta - i{\frac{\kappa}{2m_n}}\sigma_{\beta\gamma}k^\gamma \right)\right .
+ \nonumber \\ * [ 7.2pt ] & -&\left.\left.f((k - q)^2)\left(\gamma_\alpha + i{\frac{\kappa}{2m_n } } \sigma_{\alpha\nu}k^\nu \right)s(p^\prime - k)\sigma_{\beta\mu}\right)\right]-ig_{ps}^2 f_k(k^2 ) \delta(k^2)\times \nonumber \\ * [ 7.2pt ] & & \left[\frac { ( q+2k)_\mu } { ( q+k)^2-k^2 } \left(f_k(k^2)\ , - f_k((k+q)^2)\right)\gamma_5 s(p - k)\gamma_5\right .
\nonumber \\ * [ 7.2pt ] & - & \left.\left.\frac { ( q-2k)_\mu } { ( q - k)^2-k^2 } \left(f_k(k^2)- f_k((k - q)^2)\right ) \gamma_5s(p^\prime - k)\gamma_5\right]\right\ } \ ; , \label{1v}\end{aligned}\ ] ] @xmath104 & & \delta(k^2)(k+q)^\nu k^\lambda \left(\gamma_\beta + i{\frac{\kappa}{2m_n } } \sigma_{\beta\delta}(k+q)^\delta\right)s(p - k)\gamma_5 + \nonumber \\ * [ 7.2pt ] & + & f(k^2)f_k((k+q)^2)d^{\alpha\beta}(k)\delta((k+q)^2)k^\nu ( k+q)^\lambda \gamma_5 \times \nonumber \\ * [ 7.2pt ] & & \left.s(p - k)\left(\gamma_\beta -i{\frac{\kappa}{2m_n}}\sigma_{\beta\delta } k^\delta\right)\right\}\ ; .
\label{1kk}\end{aligned}\ ] ] in the above equations we define @xmath105 and use the notation @xmath106 for the @xmath107 propagator , @xmath108 for the kaon propagator , @xmath109 for the hyperon , @xmath110 , propagator with mass @xmath111 mev .
the strangeness charges are @xmath112 and @xmath113 and @xmath114 , @xmath115 refer respectively to the @xmath107 and kaon form factors , both given by eq.([ff ] ) .
aniol et al .
( happex collaboration ) , phys .
82 , 1096 ( 1999 ) ; tjnaf report no .
pr-91 - 017 , d.h .
beck , spokesperson ; tjnaf report no . pr-91 - 004 , e.j .
beise , spokesperson ; tjnaf report no .
pr-91 - 010 , j.m .
finn and p.a .
souder , spokespersons .
dong , j .- f .
lagae , and k .- f .
liu , phys . rev
75 , 2096 ( 1995 ) ; phys . rev .
d 54 , 5496 ( 1996 ) ; d.b .
leinweber , phys . rev .
d 53 , 5115 ( 1996 ) ; s.j . dong , k.f .
liu , and a.g .
williams , phys . rev .
d 58 , 074504 ( 1998 ) .
w. koepf and e.m .
henley , phys .
c 49 , 2219 ( 1994 ) ; w. koepf , s.j .
pollock and e.m .
henley , phys .
b 288 , 11 ( 1992 ) ; t.d .
cohen , h. forkel and m. nielsen , phys .
b 316 , 1 ( 1993 ) ; m.j . musolf and m. burkardt , z. phys .
c 61 , 433 ( 1994 ) ; h. forkel , m. nielsen , x. jin and t.d .
cohen , phys .
c 50 , 3108 ( 1994 ) .
r. l. jaffe , phys .
b 299 , 275 ( 1989 ) ; h. forkel , prog .
36 , 229 ( 1996 ) ; h .- w .
hammer , u .- g .
meiner and d. drechsel , phys .
b 367 , 323 ( 1996 ) ; h. forkel , phys . rev .
c 56 , 510 ( 1997 ) ; u .-
meiner , v. mull , j. speth and j.w .
van orden , phys .
b 408 , 381 ( 1997 ) ; m. j. musolf , eleventh student workshop on electromagnetic interactions , bosen , germany , 1994 ( unpublished ) . | a crucial input for recent meson hyperon cloud model estimates of the nucleon matrix element of the strangeness current are the nucleon hyperon@xmath0 ( @xmath1 ) form factors which regularize some of the arising loops . prompted by new and forthcoming information on these form factors from hyperon nucleon potential models
, we analyze the dependence of the loop model results for the strange quark observables on the @xmath1 form factors and couplings .
we find , in particular , that the now generally favored soft @xmath2 form factors can reduce the magnitude of the @xmath0 contributions in such models by more than an order of magnitude , compared to previous results with hard form factors .
we also discuss some general implications of our results for hadronic loop models . /#1/#1
# 1januaryfebruarymarchaprilmayjunejulyaugustseptemberoctobernovemberdecember # 1_phys .
rev . _ * # 1 * # 1_phys .
rev . _
* c#1 * # 1_phys .
rev .
_ * d#1 * # 1_phys .
rev .
lett . _ * # 1 * # 1_nucl .
phys . _ * a#1 * # 1_nucl .
phys . _ * b#1 * # 1_ann . of phys .
_ * # 1 * # 1_phys
. reports _ * # 1 * # 1_phys . lett .
_ * b#1 * # 1_z .
fr phys .
_ * a#1 * # 1_z .
fr phys .
_ * c#1 * |
the use of quantum cluster expansions was introduced in 1938 by kahn and uhlenbeck @xcite , who generalized to quantum statistical mechanics the ursell functions @xmath1 defined by this author in 1927 @xcite .
the major virtue of cluster expansions is that they provide directly density expansions for systems where the interaction potential is not a necessarily small perturbation ; in fact , it may even diverge at short relative distances ( hard cores for instance ) while usual perturbations theories generate power series in the interaction potential , where each term becomes infinite for hard core potentials .
starting from a quantum cluster analysis , the beth uhlenbeck formula @xcite gives an explicit expression of the first terms of a fugacity expansion ( or virial expansion ) for the grand potential of a quantum gas .
the expression is valid for any potential , the latter being characterized by its phase shifts in a completely general way .
one should nevertheless keep in mind that the words `` density expansion '' have a double meaning in this context . in a dilute gas ,
there are actually two dimensionless parameters which characterize `` diluteness '' : the product @xmath2 , where @xmath3 is the number density of the gas and @xmath4 is a length characterizing the potential range ( diameter of hard cores for instance ) , and the product @xmath5 , where @xmath6 is the quantum thermal wavelength of the particles .
the former parameter is small if , classically , a snapshot of the system shows particles among which almost all are moving freely , while the few that interact are engaged in binary collisions only ; the latter , purely quantum in nature , is sometimes called the quantum degeneracy parameter , and remains small provided there is little overlap of the quantum wave packets .
the validity of the beth uhlenbeck formula , as all fugacity expansions , therefore requires two independent parameters to be small . in a previous article @xcite
, we discuss one method which conveniently treats the two parameters separately , and allows one to include the effects of statistics by exact summations while limiting the expansion to the lowest orders in @xmath2 .
the technique is based on the use of ursell operators @xmath7 generalizing the ursell functions ( for a system of distinguishable particles ) , coupled with the exact calculation of the effect of exchange cycles @xmath8 of arbitrary length @xmath9 ; for short we call it the technique of u - c diagrams .
the result is another expression of the grand potential , which is no longer a fugacity expansion since it includes a summation over all sizes of exchange cycles so that statistical effects are included to all orders . truncating the expansion to its first terms ( lowest @xmath10 values and/or low order in a given @xmath7 ) gives results which remain valid for `` dilute degenerate systems '' @xcite , that is for all systems where the potential range is sufficiently small , but where the degeneracy of the system may become significant . in this article
we will start from an expression of the grand potential which is limited to the first order correction in the second ursell operator @xmath0 , obtained in @xcite as a trace over two particles of a product of operators .
we reduce the trace to an explicit integral where the effects of the interactions are contained in a simple matrix element .
the range of validity of our result is actually similar to that of the calculations based on the use of pseudopotentials @xcite , another approach where the final results automatically include the summation of an infinite perturbation series in terms of the initial potential .
the two methods are comparable , but we think that the u - c diagram method provides more general and more precise results , basically because it includes the short range correlations between the particles , and because all scattering channels with given angular momentum as well as their exact energy dependence are included instead of only one constant scattering length .
another point of comparison is the class of methods , for instance discussed in @xcite or @xcite , where a renormalization procedure is used in order to obtain expansions in terms of the scattering @xmath11 matrix instead of the interaction potential itself @xmath12 ; in the calculations discussed in the present article , no renormalization of this kind is needed since , roughly speaking , it is already included in the @xmath7 s , which are the building blocks of our method .
nevertheless , as we will see , our method is no longer valid when the gaseous system is brought too close to a phase transition ( superfluid transition of single particles for bosons , of pairs for fermions ) .
we begin this article with a study of the expression of the grand potential , and show how it can be expressed as an expression that is similar to the well known beth uhlenbeck formula ; actually it can be obtained from it by two simple substitutions .
we then discuss the physics contained in this general result , as well as the changes introduced by the possible occurrence of bound states .
in particular we consider the effects of exchange between bound and unbound particles , an effect which is not contained in the usual beth uhlenbeck formalism ; this kind of exchange may play some role in clouds of laser cooled alkali atoms @xcite for which the potential is sufficiently attractive to sustain a large number of bound states . in section [ spin1/2 ]
we apply these results to spin 1/2 particles . finally , in section [ elements ] , we discuss the appropriate quantities in terms of which one should describe the effects of the potential on the physical properties of the system , and introduce for this purpose the so called `` ursell length '' , which plays a role similar to the scattering length @xmath13 . in the theoretical study of quantum gases , and as already mentioned , one frequently used method is to replace the real interaction potential between the particles by a `` pseudopotential '' that has no range ( a dirac delta function of the space variables ) , and to treat this potential to first perturbation order ; in other words one ignores the distortion of the many - body wave functions at short relative distances and the associated effects of the inter particle correlations .
the justification of this approach is based on the physical expectation that , for a dilute gas , all the effects of the potential should be contained in the binary collision phase shifts associated with the potential , which can easily be reproduced to first order by a pseudopotential ; meanwhile all detailed information on the behavior of the wave function at short relative distances can safely be discarded .
our formalism allows one to explicitly distinguish between short range effects ( `` in potential effects '' ) and asymptotic effects ( out of the potential ) , which naturally leads to a discussion of this ansatz .
the interactions appear in terms of a matrix element of an ursell operator , which depends on the potential but does not reduce to it ; for fermions at low temperatures , the matrix elements contain physical effects which are not included in usual treatments of normal fermi gases .
the basic object in terms of which most physical quantities will be written in this article is the second ursell operator @xmath14 , defined by : @xmath15 } \right ] \label{u2}\ ] ] where @xmath16 and @xmath17 are single particle hamiltonians , containing the kinetic energy of the particle and , if necessary , its coupling to an external potential , and where : @xmath18 is the hamiltionan of two particles , including the mutual interaction potential @xmath19 . depending of the context , it may be more convenient to use the symmetrized operator @xmath20 : @xmath21 } { 2}=\frac{\left [ 1+\eta p_{ex}\right ] } { 2}u_{2}(1,2 ) \label{u2sa}\ ] ] where @xmath22 is the exchange operator between particles 1 and 2 and @xmath23 has the value + 1 for bosons , @xmath24 for fermions .
moreover , the `` interaction representation version '' of either @xmath0 and @xmath20 , obtained by multiplying these operators by @xmath25 , will also be useful ; we denote them with an additional bar over the operator , for instance : @xmath26 this operator and its symmetrized version @xmath27 act only in the space of relative motion of the two particles ; they have no action at all on the variables of the center of mass .
finally , if the particles are not submitted to an external potential ( @xmath28 contains only the kinetic energy ) , it is convenient to introduce the momentum @xmath29 of the center of mass of the two particles as well as the hamiltonian of the relative motion : @xmath30 ( @xmath31 is the mass of the particles ) ; then , @xmath32 can be written in the form of a product : @xmath33 _ { rel } \label{u2bis}\ ] ] with : @xmath34 _ { rel}=\left [ \text{e}^{-\beta h_{rel}}-\text{e}^{-\beta \frac{{\bf p}^2}m}\right ] \label{u2rel}\ ] ] we now start from relation ( 46 ) of @xcite which gives the grand potential ( multiplied by @xmath35 ) in the form : @xmath36 _ { ig}+\left [ \limfunc{log}z\right ] _ { int } \label{1}\ ] ] where @xmath37 _ { ig}$ ] is the well known value of @xmath38 for the ideal gas : @xmath39 _ { ig } & = -\eta \text{tr}\left\ { \limfunc{log}\left [ 1-\eta z\text{e}^{-\beta h_{1}}\right ] \right\ } \\ & = \eta \text{tr}\left\ { \limfunc{log}\left [ 1+\eta f\right ] \right\ } \end{array } \label{1prime}\ ] ] and where the correction introduced by the interactions is : @xmath40 _ { int}=z^{2}\,\text{tr}_{1,2}\left\ { u_{2}^{s , a}(1,2)\,\left [ 1+\eta f(1)\right ] \,\left [ 1+\eta f(2)\right ] \right\ } \label{1bis}\ ] ] in these equations , @xmath41 is the fugacity , @xmath42 the inverse temperature and @xmath43 the chemical potential , while @xmath44 is defined as the operator : @xmath45 similar results can be found in the work of lee and yang , see formulas ( ii.8 ) and ( ii.23 ) of ref . @xcite . as mentioned in the introduction , equation ( [ 1bis ] )
gives the correction introduced by the interactions to the lowest order approximation in @xmath0 ; see @xcite for a discussion of the higher order corrections . using the definition of @xmath46 : @xmath47 \label{u2sabarre}\ ] ] as well as the relation @xmath48e@xmath49 = f$ ] , we can rewrite ( [ 1bis ] ) in the form : @xmath40 _ { int}=\,\text{tr}_{1,2}\left\ { \overline{u}_{2}^{s , a}(1,2)\,f(1)f(2)\right\ } \label{logz - int}\ ] ] which expresses the correction as the average of the operator @xmath27 over unperturbed distributions functions @xmath44 s . for spinless particles , by making the trace in ( [ logz - int ] ) explicit ,
we obtain : @xmath40 _ { int}=-\frac{\lambda _ { t}^{2}{\mbox{$\cal v$}}}{\left ( 2\pi \right ) ^{6}}\,\int d^{3}k_{1}\int d^{3}k_{2}\,f({\bf k}_{1})f({\bf k}_{2})\,a_{u}^{s , a}(\left| { \bf k}_{1}-{\bf k}_{2}\right| ) \label{eq - base}\ ] ] where @xmath50 is the volume of the system , @xmath6 the thermal wavelength : @xmath51 and where @xmath52 is defined by : @xmath53 _ { rel}\mid { \bf k } > \label{sigmaprime}\ ] ] here : @xmath54 is the appropriate variable since @xmath27 does not have any action in the space of states associated to the center of mass of the two particles ; @xmath52 is a microscopic length , independent of @xmath50 for large systems has a microscopic range and since the factor @xmath50 in ( [ sigmaprime ] ) makes up for the normalization factor of the plane waves that occur in the matrix element ( note that all plane wave kets in our formulas are normalized in a finite volume ; hence the absence of dirac delta functions of momenta differences in ( [ sigmaprime ] ) . ] , which we will call the `` ursell length '' - see section [ ur - length ] for a more detailed discussion and a justification of the numerical factors that we have introduced .
the correction written in ( [ eq - base ] ) is analogous to a first order energy correction due to binary interactions , while @xmath52 plays the role of some effective interaction ( within a numerical factor ) .
for instance , if we assume that we treat to first order in perturbation a pseudopotential[multiblock footnote omitted ] of the form : @xmath55 where @xmath13 is a scattering length ( or the diameter of hard cores ) , we easily obtain : @xmath56 \label{elementsde}\ ] ] so that ( [ sigmaprime ] ) shows that , in this approximation , the ursell length becomes independent of the wave number @xmath57 . for bosons , the correction to the grand potential
then becomes : @xmath40 _ { int}\simeq -2a\frac{{\mbox{$\cal v$}}}{\lambda _ { t}}\,\times g_{3/2}(z ) \label{variation}\ ] ] with the usual notation : @xmath58 while , for fermions , no first order correction is obtained .
these results coincide with the first order terms of the well - known results of lee and yang ; see formulas ( 1 ) and ( 4 ) for @xmath59 of ref .
@xcite.nevertheless , in section [ elements ] , we discuss the validity of this first order approximation in the calculation of the matrix elements of @xmath60 and conclude that , for bosons , ( [ elementsde ] ) is a good approximation while it is not necessarily the case for fermions . at this point
, it is convenient to introduce the free spherical waves @xmath61 associated with the relative motion of the particles , with wave functions vectors and kronecker delta s of @xmath9 and @xmath31 ) , it would be necessary to multiply all the @xmath62 s by factors @xmath63 ; this operation is not necessary here .
] : @xmath64 ( with standard notation ; @xmath65 is a spherical bessel function , @xmath66 a spherical harmonics of the angular variables of @xmath67 ) ; if @xmath68 is a plane wave normalized in a volume @xmath50 : @xmath69 ^{*}\mid j_{k , l , m}^{(0 ) } > \label{spherical}\ ] ] if we insert this equality into ( [ sigmaprime ] ) and take into account the well - known relation : @xmath70 we readily obtain the result are diagonal in @xmath9 and @xmath31 . ] : @xmath71 \times a_{u}^{(l)}(k ) \label{sigmaprim}\ ] ] with : @xmath72 _ { rel}\mid j_{klm}^{(0 ) } > \label{uldef}\ ] ] where @xmath73 _ { rel}$ ] has been defined in ( [ u2rel ] ) ; rotational invariance ensures that the right hand side of ( [ uldef ] ) is independent of @xmath31 .
these results can be inserted into the integral appearing in ( [ eq - base ] ) and provide an expression of the correction to the grand potential which is a direct generalization of the usual beth uhlenbeck formula to gases having a significant degree of degeneracy : @xmath40 _ { int}=-\frac{\lambda _ { t}^{2}{\mbox{$\cal v$}}}{(2\pi ) ^{6}}\sum_{l}(2l+1)\left [ 1+\eta ( -1)^{l}\right ] \times \int d^{3}k_{1}\int d^{3}k_{2}\,f({\bf k}_{1})f({\bf k}_{2})\times a_{u}^{(l)}(k)\ , \label{bill}\ ] ] for comparison , we recall the explicit expression of this formula : @xmath40 _ { int}^{b.u.}=\frac{2^{3/2}}{2\pi } z^{2}\frac{{\mbox{$\cal v$}}}{\lambda _ { t}}\sum_{l}(2l+1)\left [ 1+\eta ( -1)^{l}\right ] \int dk\text{e}^{-\beta \hbar ^{2}k^{2}/m}k\delta _ { l}(k ) \label{bu}\ ] ] ( we temporarily ignore possible bound states ) , where @xmath6 is defined in ( [ lambda ] ) . as already discussed in @xcite , if we replace each of the two @xmath44 s in ( [ bill ] ) by their low density limit @xmath74 , the integration over @xmath75 of a gaussian function introduces a factor @xmath76 ^{3}$ ] and , after some algebra , we recover ( [ bu ] ) . in other words ,
the following substitutions are necessary to obtain ( [ bill ] ) from the usual beth uhlenbeck formula:[multiblock footnote omitted ] @xmath77 /2m}\rightarrow
z^{-2}f({\bf k}_{1})f({\bf k}_{2 } ) \\ \displaystyle \delta _
{ l}(k)\rightarrow -ka_{u}^{(l)}(k ) \end{array } \right .
\label{subst}\ ] ] in equation ( [ 1bis ] ) , the first of these substitutions amounts to adding the terms in @xmath78 . the operator @xmath79 _ { rel}^{s , a}$ ] may be written as the sum of the contributions of bound states and of the continuum : @xmath80 _ { rel}^{s , a}=\sum_{n}\mid \phi _
{ n}\rangle \langle \phi _ { n}\mid \text{e}^{-\beta e_{n}}+\text{continuum } \label{decomposition}\ ] ] where the @xmath81 s are the kets associated with the eigenstate of the relative motion of the particles with ( negative ) energy @xmath82 ( with appropriate symmetry for the statistics of the particles ) .
it is therefore not difficult make the contribution of bound states in ( [ eq - base ] ) explicit , which provides the following term : @xmath83 _ { int}^{bound } & = \displaystyle(2\pi ) ^{-6}{\mbox{$\cal v$}}^{2}\int d^{3}k\int d^{3}q\,\,f(\frac{{\bf k}}{2}+{\bf q)}\,f(\frac{{\bf k}}{2}-{\bf q)\times } \\ & \displaystyle\,\,\,\,\,\,\,\,\;\;\;\;\;\;\times \,\,\qatopd . .
{ \qatopd . .
{ } { } } { } \sum_{n}^{\qatopd . .
{ \qatopd . .
{ } { } } { } } \mid \langle { \bf q}\mid \phi _ { n}\rangle \mid ^{2}\text{e}^{\beta \left ( e_{n}+\hbar ^{2}q^{2}/m\right ) } \end{array } \label{int - bound}\ ] ] where ; because these kets have a finite range , and because the plane waves @xmath84 are normalized in a macroscopic volume @xmath50 , the product @xmath85 is independent of @xmath50 in the thermodynamic limit , as necessary to obtain an extensive correction to the grand potential .
if we rewrite the integral of ( [ int - bound ] ) in the form : @xmath86 \times \left [ 1+\eta f(\frac{{\bf k}}{2}-{\bf q)}\right ] \sum_{n}\mid \langle { \bf q}\mid \phi _ { n}\rangle \mid ^{2}\text{e}^{\beta e_{n } } \label{integr}\ ] ] we may distinguish between two contributions in the correction : @xmath40 _ { int}^{bound}=\left [ \limfunc{log}z\right ] _ { b.u.}^{bound}+\left [ \limfunc{log}z\right ] _ { stat.}^{bound } \label{somme}\ ] ] the first contribution is obtained by ignoring in ( [ integr ] ) the @xmath78 s inside the brackets , which allows one to integrate over @xmath75 ; using the closure relations over the plane waves @xmath87 and the normalization of the bound states @xmath81 then provides the following result for this first contribution of the bound states : @xmath40 _ { b.u.}^{bound}=2^{3/2}\left [ \lambda _ { t}\right ] ^{-3}{\mbox{$\cal v$}}\sum_{n}\,\text{e}^{\beta e_{n } } \label{bound - bu}\ ] ] which is identical to the term corresponding to bound states in the usual beth uhlenbeck formula .
the second contribution arises from the rest of the product of the brackets and is equal to : @xmath88 _ { stat.}^{bound } & \displaystyle=(2\pi ) ^{-6}{\mbox{$\cal v$}}^{2}\int d^{3}k\int d^{3}q\,\text{e}^{-\beta \hbar ^{2}{\bf k}^{2}/4m}\left [ f(\frac{{\bf k}}{2}+{\bf q)}f(\frac{{\bf k}}{2}-{\bf q)}+\right . \\ & \displaystyle\;\;\;\left .
+ \eta f(\frac{{\bf k}}{2}+{\bf q)+}\eta f(\frac{{\bf k}}{2}-{\bf q)}\right ] \sum_{n}\mid \langle { \bf q}\mid \phi _ { n}\rangle \mid ^{2}\text{e}^{\beta e_{n } } \end{array } \label{bound - stat}\ ] ] it accounts for exchange effects between bound states and continuum states . ] , which explains the appearance of the scalar product @xmath89 of a free plane wave and a bound state wave function ; the factors @xmath90 correspond physically to the boltzmann distribution of the populations of the bound states .
if we assume that the momenta in the continuum , which have values that are limited by the presence of the @xmath44 s under the integral , are much smaller than those contained in the bound states @xmath81 s , we can replace the product @xmath89 by @xmath91 ; this shows that the effect of statistics is more important for bound states with a wave function with a significant integral over space , as for instance the ground state wave function which has no node ; states with rapidly oscillating wave functions give almost no correction to the usual beth uhlenbeck formula . generally speaking , for bosons ,
the effect of statistics is always to increase the beth uhlenbeck term ( [ bound - bu ] ) . for fermions ,
the situation is more complicated : if the gas is only slightly degenerate one has @xmath92 so that the terms in @xmath23 dominate in ( [ bound - stat ] ) , leading to a decrease of the effects of bound states ; but if the gas is strongly degenerate , there seems to be no general rule , and exchange of bound states with the continuum may either enhance or reduce their contribution .
an easy generalization is to include spins ; the only difference is that all the traces must now also include spin states .
it is then convenient to replace the functions @xmath93 of momentum by spin operators @xmath94 ( corresponding to two by two matrices for spin 1/2 particles ) which are functions of @xmath95 and are defined as : @xmath96 where @xmath97 is the one particle hamiltonian , including kinetic energy as well as coupling of the spins to the magnetic field ( if the particles carry magnetic moments ) .
similarly , because now the exchange of particles must also include their spin states , @xmath52 becomes an operator @xmath98 which acts in the space of the states of two spins : @xmath99 \label{sigma - spins}\ ] ] where @xmath100 is the exchange operators of two spins .
this leads to the following generalization of ( [ eq - base ] ) : @xmath40 _ { int}=-\frac{\lambda _ { t}^{2}{\mbox{$\cal v$}}}{\left ( 2\pi \right ) ^{6}}\int d^{3}k_{1}\int d^{3}k_{2}\;\text{tr}_{s_{1},s_{2}}\left\ { f_{s}({\bf k}_{1})f_{s}({\bf k}_{2})\;\,\sigma _ { s}(k)\right\ } \label{z - spins}\ ] ]
we now apply the preceding calculation to the study of the magnetic susceptibility of a dilute gas of fermions or bosons with spin 1/2 ( spin polarized atomic hydrogen provides an example of spin 1/2 bosons @xcite ) .
we assume that the one particle hamiltonian is : @xmath101 where @xmath102 is the larmor frequency in the ( homogeneous ) magnetic field and @xmath103 the pauli matrix ( operator ) associated with the component of the spin along the field ; we set : @xmath104 which gives the following values for the matrix elements s are not the populations of the one body density operators , but differ from them by density corrections @xcite .
] of @xmath94 defined in ( [ rho ] ) : @xmath105 for short we will write : @xmath106 with this notation we have : @xmath107 _ { ig } & \displaystyle=-\eta \frac{{\mbox{$\cal v$}}}{8\pi ^{3}}\int d^{3}k_{1}\limfunc{log}\left\ { \stackrel{}{\left [ 1-\eta z_{+}\text{e}^{-\beta \hbar ^{2}k_{1}^{2}/2m}\right ] } \left [ 1-\eta z_{-}\text{e}^{-\beta \hbar ^{2}k_{1}^{2}/2m}\right ] \right\ } \\ & \displaystyle=\eta \frac{{\mbox{$\cal v$}}}{8\pi ^{3}}\int d^{3}k_{1}\limfunc{log}\left\ { \stackrel{}{\left [ 1+\eta f_{+}(1)\right ] } \left [ 1+\eta f_{-}(1)\right ] \right\ } \end{array } \label{logzig}\ ] ] for a gas at equilibrium the operators @xmath94 are diagonal in the basis corresponding to a quantization axis parallel to the magnetic field ; we then have : @xmath108 } \left [ f_{+}(2)+f_{-}(2)\right ] \end{array } \label{trs1s2}\ ] ] as well as : @xmath109 ( the latter result arises because the trace gets non zero contributions only from the two spin states @xmath110 and @xmath111 , which are invariant under the effect of @xmath100 ) .
we therefore have : @xmath107 _ { int } & \displaystyle=-\frac{\lambda _ { t}^{2}{\mbox{$\cal v$}}}{\left ( 2\pi \right ) ^{6}}\int d^{3}k_{1}\int d^{3}k_{2}\,\left\ { a_{u}^{s , a}(k)\,\stackrel{\stackrel{}{}}{\left [ f_{+}(1)f_{+}(2)+f_{-}(1)f_{-}(2)\right ] } \right . \\ & \displaystyle\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\left .
+ 2a_{u}(k)\;f_{+}(1)f_{-}\stackrel{\stackrel{}{}}{(2)}\right\ } \end{array } \label{logzint}\ ] ] where : @xmath112 while @xmath52 is defined by ( [ sigmaprime ] ) - or equivalently ( [ sigmaprim ] ) - while @xmath113 is the un - symmetrized version ) ; we choose this convention since the same factor 1/2 appears in the definition ( [ u2sa ] ) of the symmetrized version of @xmath0 ; in this way , if exchange effects are ignored ( high temperature limit for instance ) , the various ursell lengths become equal .
] of the ursell length : @xmath114 ( it would correspond to distinguishable particles ) .
if , as in the beginning of section [ rotation ] , we treat the pseudopotential ( [ veff ] ) to first order , for fermions we obtain @xmath115 and @xmath116 ; the first order correction is now given by : @xmath40 _ { int}\simeq -2\frac{a{\mbox{$\cal v$}}}{\lambda _ { t}}\,\times g_{3/2}(-z ) \label{corrferm}\ ] ] which coincides with the result of lee and yang ( equation ( 1 ) of @xcite for @xmath117 ) ; but , again , we note that a critical discussion of this first order calculation is made in section [ elements ] which shows that , for fermions , the results of the ursell approach may be different from those of a pseudopotential theory .
the number density of the gas is obtained from the relation : @xmath118 \label{n}\ ] ] similarly , then density of internal energy is given by : @xmath119 \label{u}\ ] ] while the `` magnetization '' of all atoms ; in other words , the maximum value of @xmath120 ( complete spin polarization , all spins parallel ) is equal to the total number of particles . ]
is equal to : @xmath121 \label{m}\ ] ] we therefore have to vary either @xmath48 , or @xmath122 , or @xmath123 in formulas ( [ logzig ] ) and ( [ logzint ] ) ; we can then use the simple relations : @xmath124 and : @xmath125 \label{derf}\ ] ] as well as : @xmath126 \left [ -\frac{\hbar ^{2}k^{2}}{2m}\pm \frac{\hbar \omega _ { 0}}{2}\right ] \label{dfdbeta}\ ] ] we then obtain , by making use of the symmetry in the indices @xmath127 and @xmath128 : @xmath129 + \\ & \displaystyle-\frac{\lambda _ { t}^{2}}{\left ( 2\pi \right ) ^{6}}\int d^{3}k_{1}\int d^{3}k_{2}\,\left\ { 2a_{u}^{s , a}(k)f_{+}(1)f_{+}(2)\stackrel{\stackrel{}{}}{\left [ 1+\eta f_{+}(1)\right ] } + id.\left ( f_{+}\leftrightarrow f_{-}\right ) + \right . \\ & \,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\displaystyle\,\,\,\,\,\left .
+ 2a_{u}(k)f_{+}(1)f_{-}(2)\stackrel{\stackrel{}{}}{\left [ 2+\eta f_{+}(1)+\eta f_{-}(2)\right ] } \right\ } \end{array } \label{n2}\ ] ] where @xmath130 symbolizes the same expression where @xmath131 and @xmath132 are interchanged .
the calculation of @xmath133 is almost the same , except that now , because of ( [ dz ] ) , @xmath131 and @xmath132 introduce different signs ; the result is : @xmath134 + \\ & \displaystyle-\frac{\lambda _ { t}^{2}}{\left ( 2\pi \right ) ^{6}}\int d^{3}k_{1}\int d^{3}k_{2}\,\left\ { 2a_{u}^{s , a}(k)f_{+}(1)f_{+}(2)\stackrel{\stackrel{}{}}{\left [ 1+\eta f_{+}(1)\right ] } -id.\left ( f_{+}\leftrightarrow f_{-}\right ) + \right . \\ & \,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\displaystyle\,\,\,\,\,\left .
+ 2\eta a_{u}(k)f_{+}(1)f_{-}(2)\stackrel{\stackrel{}{}}{\left [ f_{+}(1)-f_{-}(2)\right ] } \right\ } \end{array } \label{m}\ ] ] finally , the calculation of the internal energy @xmath135 provides the result : @xmath136 where @xmath137 is the density of energy associated with the external variables of the particles ( kinetic and potential energy ) : @xmath138 \frac{\hbar ^{2}k_{1}^{2}}{2m}+ \\ & \displaystyle-\frac{\lambda _ { t}^{2}}{\left ( 2\pi \right ) ^{6}}\int d^{3}k_{1}\int d^{3}k_{2}\,\left\ { 2a_{u}^{s , a}(k)f_{+}(1)f_{+}(2)\left [ 1+\eta f_{+}(1)\right ] \frac{\hbar ^{2}k_{1}^{2}}{2m}+id.\left ( f_{+}\leftrightarrow f_{-}\right ) + \right . \\ & \,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\displaystyle\,\,\,\,\,\left .
+ 2a_{u}(k)f_{+}(1)f_{-}(2)\left [ \left [ 1+\eta f_{+}(1)\right ]
\frac{\hbar ^{2}k_{1}^{2}}{2m}+\left [ 1+\eta f_{-}(2)\right ] \frac{\hbar ^{2}k_{2}^{2}}{2m}\right ] \right\ } \end{array } \label{uk}\ ] ] while @xmath139 is the density of magnetic energy : @xmath140 we now assume that the magnetic field is sufficiently low so that the preceding expressions may be limited to their first order expansion in @xmath141 .
we then use the relations : @xmath142 \right ] \label{df}\ ] ] which provide : @xmath143 } \left [ a_{u}^{s , a}(k)+a_{u}(k)\right ] \end{array } \label{nchampnul}\ ] ] and for the zero field magnetic susceptibility of the gas : @xmath144 -\frac{\lambda _ { t}^{2}}{\left ( 2\pi \right ) ^{6}}\beta \hbar \int d^{3}k_{1}\int d^{3}k_{2}f(1)f(2)\times \\ & \displaystyle\times \left\ { 2\,a_{u}^{s , a}(k)\,\stackrel{\stackrel{}{}}{\left [ 1+\eta f(1)\right ] } \left [ 2 + 2\eta f(1)+\eta f(2)\right ] \right . + \\ & \,\,\,\,\,\,\displaystyle\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,+\left .
a_{u}(k)\stackrel{}{\left [ \left [ f(1)-f(2)\right ] ^{2}+\,\,2\eta f(1)\left [ 1+\eta f(1)\right ] \right ] } \right\ } \end{array } \label{chi}\ ] ] equations ( [ n2 ] ) and ( [ m ] ) show how quantum statistical effects modify the equation of state and the magnetic susceptibility of a gas .
what determines the properties of the dilute system is not directly the matrix elements of the potential @xmath145 , but those of the operator @xmath146 .
of course , if the potential is weak , the calculation may be limited to first order perturbation theory , so that @xmath146 may be replaced by @xmath147 ; the two operators then become equivalent.but , in more realistic situations , for instance in atomic systems where the interaction potential at short relative distances becomes very large , the effects of the potential can not be treated to first order .
an obvious difference between @xmath146 and @xmath145 is that the former depends on one more characteristic length than the potential , namely the thermal wavelength @xmath6 , so that one can expect that its range will depend explicitly on the temperature .
it is sometimes argued that a description of the interactions in terms of the scattering length is sufficient at very low temperatures , because the range of the potential @xmath145 remains always much smaller than the de broglie wavelengths of the particles .
as far as the matrix elements of @xmath146 are concerned , this is of course true if the interaction potential is treated to first order only but , precisely at very low temperatures , the higher order terms become significant and @xmath146 may acquire a range that exceeds by far that of the potential itself .
we will see examples of this phenomenon in section [ elements ] .
we now discuss the limit of validity of the results obtained in the preceding section or , since they are equivalent , of equations ( [ logz - int ] ) and ( [ z - spins ] ) .
if the particles are bosons , our theory is limited to gases which are not too close to bose einstein condensation - while of course the degeneracy may be much more pronounced than for the beth uhlenbeck formula since @xmath48 does not have to be small .
the reason is that , when a system of bosons approaches the region of quantum condensation , the distribution function @xmath44 starts to build up a singularity at low energies .
one can then see that , when @xmath148 , the partition function becomes more and more sensitive to terms of higher and higher order in @xmath146 , in @xmath149 , @xmath150etc .
, since they contain larger and larger number of functions @xmath44 ( or of factors @xmath151
. in other words , terms which normally remain small corrections become dominant when the point of bose einstein condensation is approached ; see the discussion already given in @xcite , where it is emphasized should be positive ( dominant character of the attractive interactions in the matrix element ) . ]
that a theory limited to first order in @xmath146 would predict the disappearance of the bose einstein condensation phenomenon and its replacement by a simple crossover between two regimes . for fermions ,
the discussion is different since no special phenomenon takes place when @xmath48 reaches one ; when the gas is cooled at constant density , this merely corresponds to a cross over region where the gas is becoming degenerate .
when the temperature is decreased even more , a stronger degeneracy builds up while @xmath152e@xmath153 becomes larger and larger ( @xmath122 increases while @xmath43 remains almost constant if the density is fixed ) .
it is therefore clear that @xmath48 itself can not be an expansion parameter in this region ; but our perturbation series is not a @xmath48 expansion and it may still converge for degenerate systems , provided the matrix elements of @xmath146 are sufficiently small . in the next section , we will see that these matrix elements are equal to the product @xmath154 , where @xmath6 is the thermal wave length and @xmath155 is some microscopic length ( the ursell length ) that we will define more precisely later .
let us for instance discuss ( [ n2 ] ) where the interaction corrections involve a double integral over two momenta .
the first introduces the number density @xmath3 ; the second contains a function @xmath156 , which in the limit of low temperatures is non zero only in an energy slice of width @xmath157 at the surface of the fermi sphere ; the corresponding width @xmath158 in terms of momentum is given by : @xmath159 ( where @xmath160 is the momentum at the fermi surface ) so that the result of these integrations is : @xmath161 the small parameter of the expansion plays an essential role in this argument .
this factor occurs in all expressions , for instance , in ( [ m ] ) , the bracket @xmath162 $ ] may be written as @xmath163 $ ] .
if we had ignored this factor , we would have found @xmath164 as the expansion parameter , which is a temperature dependent factor ( as in the usual beth uhlenbeck formula ) . ]
is therefore the product @xmath165 , which is not very different from @xmath166 for degenerate gases , and indeed remains small if the average distance between the particles is much larger than @xmath155 .
this result is not surprising : indeed , in the usual theory of dilute non ideal fermi gases @xcite , the expansion parameter is independent of the temperature , which plays no particular role as long as the system is strongly degenerate .
one might be tempted at this point to conclude that the generalized beth uhlenbeck formula is valid for arbitrary degeneracy of a fermionic system ; nevertheless , in the next section , we will see that the factor @xmath155 itself may become very large at very low temperatures , which in turn increases the expansion parameter and , automatically , limits the validity of the expansion at some point .
we will assume that this somewhat unexpected phenomenon is a precursor of bcs pair condensation ;
if this is the case , the low temperature limit of the validity of the generalized beth uhlenbeck is that the gas should remain a normal fermi gas and , even , not to be too close to condensation . in general , for bosons as well as for fermions , we can therefore conclude that our calculations remain valid as long as bose einstein or bcs condensation is not too close . .
we now study in more detail the values of the basic ingredient that we use to describe the effects of the interactions on the thermodynamic properties of a quantum gas , namely the diagonal matrix element of @xmath46 together with those of the un - symmetrized operator if the particles have spins , see section [ spin1/2 ] .
a natural question then is the following : to determine these coefficients , is it sufficient to characterize the potential in terms of its binary collision phase shifts , which determine the asymptotic behavior of interacting wave functions , or is it also necessary to include some information on the behavior of the wave functions at short relative distances , inside the potential ? for bosons at very low temperatures ,
is it possible , even more simply , to reason in terms of the scattering length only ?
we have already mentioned in the introduction that , in the literature on low temperature dilute systems , it is often considered as physically obvious that the short range properties of the potential play no role in the thermodynamics ; it is then possible to replace the real potential by an effective potential ( or pseudopotential ) which has the expression already given in ( [ veff ] ) in terms of the scattering length @xmath13 . with a potential having zero range , it becomes of course meaningless to take into account the distortions of the wave functions inside the potential , in other words the contribution of particles `` in the middle of a collision '' . but with real potentials , it is not obvious that these contributions are non existent , in particular since statistical effects arising from particle exchange are expected to be more important when the particles are close .
we discuss here this question in the context of the generalized beth uhlenbeck formula that we have obtained , with the help of simple examples such as a step like potential mimicking a real interaction potential , or even hard spheres .
we first discuss the ursell length @xmath113 , which we have already introduced above , and which depends on the temperature while @xmath13 does not .
if the interaction potential @xmath145 were sufficiently weak , we could use a first order perturbation expression of the ursell operator : @xmath168 which would lead to the following expression : @xmath169 inserting ( [ veff ] ) into this result provides : @xmath170 where @xmath50 is the macroscopic volume and @xmath171 the thermal wavelength ( [ lambda ] ) . by analogy with this first order calculation , it is convenient to characterize the diagonal matrix elements in terms of a length , @xmath113 , which we call the ursell length and which we have defined in ( [ sigma ] ) as : @xmath172 this very definition ensures that , within the theory of the pseudopotential , @xmath113 is exactly equal to the scattering length @xmath13 .
this length therefore provides a convenient tool for a discussion of the validity of this approximation : as long as @xmath113 remains very close to @xmath13 for all relevant values of @xmath95 , the theory of pseudopotentials and ours provide strictly equivalent results . for identical particles , what is needed is the symmetrized ursell length , already defined in ( [ sigmaprime ] ) and ( [ sigmaprim ] ) , which allows us to write the correction to the grand potential for spinless particles in the form ( [ eq - base ] ) .
this equation ( [ eq - base ] ) expresses that the symmetrized ursell length gives , within the numerical factor in front of this integral , the crossed contribution of two velocity classes @xmath173 and @xmath174 to the grand potential of the system ( its pressure ) .
if the particles have spins , we have to use ( [ z - spins ] ) and ( [ sigma - spins ] ) . in section [ rotation ] , we introduced spherical orbital variables from the free spherical waves @xmath175 . at this stage , it is convenient to use spherical waves that are normalized in a sphere of volume @xmath176 .
we therefore introduce a new notation , @xmath177 , but these kets are simply proportional to the @xmath178 : @xmath179 where @xmath180 is a normalization coefficient which in the limit of large values of the product @xmath181 is equal to : @xmath182 introducing in the same way the normalized kets @xmath183 in the presence of the interaction potential , we can write : @xmath184 ^{-2}\left [ \sum_{k^{^{\prime } } } \left| < \varphi _ { k , l , m}^{(0)}\mid \varphi _ { k^{^{\prime } } , l , m}>\right| ^{2}e^{-\beta \left [ e(k^{^{\prime } } ) -e(k)\right ] } -1\right ] \label{toto}\ ] ] where @xmath185 and @xmath186 are the energies of the free and interacting states . in what follows we discuss the values of the ursell length and its @xmath95 dependence with the help of a few examples . in order to simplify our discussion , we now consider an interaction potential made of a hard core of diameter @xmath187 ( with @xmath188 ) and of an attractive part ( @xmath189 ) from relative distance @xmath190 to @xmath191 ; see figure 1 .
our discussion is in fact more general , but this kind of simplified potential is a convenient way to mimic the effects of a more realistic interaction potential , containing strong repulsion at short distances and van der waals attraction at large distances .
for instance , it is not difficult to find a relation between @xmath192 , @xmath193 and @xmath4 which ensures that the scattering length of this potential vanishes ( compensation of the effects of attraction and repulsion at low energies ) .
does this imply that the matrix elements of @xmath0 , that is the ursell length @xmath113 , also vanish ?
not in general , since the scalar products @xmath194 in ( [ toto ] ) are not only sensitive to the changes of the interacting wave functions outside of the potential ( which do not occur if the scattering length vanishes ) but also to their values inside the potential . in other words , for a degenerate dilute gas , all the effects of the potential are not necessarily contained in the scattering phase shifts .
we nevertheless note that , when the range @xmath4 of the potential tends to zero , the corrections to the scalar products are necessarily of third order in @xmath4 ( at least ) , while first order terms occur in the phase shifts ( for instance , at very low temperatures , the beth uhlenbeck formula predicts a correction which is proportional to @xmath195 ) .
therefore , if the potential range is sufficiently small ( compared to the two other microscopic distances , the average distance between particles @xmath196 and @xmath197 ) , one may limit the calculation to first order in @xmath4 so that the contribution of the distortion of the wave functions inside the potential may be ignored . within this approximation ,
all effects of the potential on the thermodynamic properties are indeed contained in the collision phase shifts . a brief similar discussion in terms of `` shape dependent terms '' ( as opposed to phase shift dependent terms ) appears in the two last paragraphs of section 5 of ref .
@xcite consequently , in all calculations of the thermodynamic quantities which are limited to first order in the potential range , it is sufficient to characterize the potential by its long distance effects on the wave function ( phase shifts ) only , while short range effects are irrelevant ; if all collisions take place at very low energies , all effects are then contained in the scattering length only . on the other hand
, this is not necessarily the case if higher order terms in the potential range are included , and corrections which originate `` inside the potential '' , corresponding to the contribution of particles `` in the middle of a collision '' , may appear ; see also @xcite .
actually , the question remains open as to whether is would be possible , by some mathematical transformation , to express these `` in potential effects '' in terms of the phase shifts only ; this is for instance possible in the absence of the phase occupation factors , since one then gets the usual beth uhlenbeck formula for which such a transformation is known . in the presence of these factors
, we have made efforts to investigate the possibility that a similar simplification takes place , but we have not been able to prove it .
a possible conjecture is that short range and long range effects are in general independent from each other , but this remains to be proved by a precise example . for hard spheres
, the interacting wave function does not penetrate into the potential ; this is a special case where all physical effects of the potential are necessarily contained in the collision phase shifts .
the potential range @xmath4 coincides in this case with the scattering length @xmath13.but this is not sufficient to ensure that the theory of pseudopotentials should be equivalent to our results , and we now discuss this question . in order to obtain the correction to the thermodynamic potential for hard spheres ,
we now perform a calculation of the scalar products which appear in ( [ toto ] ) . [
[ s - waves ] ] s waves + + + + + + + for free waves , the quantification for the wave numbers is given by : @xmath198 while , for hard spheres of diameter @xmath13 , the interacting waves satisfy the relation : @xmath199 where @xmath3 and @xmath200 are integer numbers .
the scalar products of ( [ toto ] ) then become functions of these numbers : @xmath201 which , after a simple calculation , leads to : @xmath202 } { \left [ n^{^{\prime } } -n\left ( 1-a / r\right ) \right ] ^2\left [ n^{^{\prime } } + n\left ( 1-a / r\right ) \right ] ^2 } \label{pscalbis}\ ] ] this result , when inserted into ( [ toto ] ) for @xmath203 , provides the contributions of s waves to the ursell length .
equivalent results for infinite volume ( only s wave ) are given by lee and yang @xcite ; we have checked that our numerical results are compatible with those of these authors . since @xmath204 , equation ( [ pscalbis ] ) shows that the scalar product is peaked around a value of @xmath200 given by : @xmath205 where @xmath206 is the integer value ; for low energies , @xmath207 and the preceding equation reduces to @xmath208 . assume for instance that we are interested in calculating the ursell length to first order in @xmath13 , and in the s wave channel only .
equation ( [ pscalbis ] ) shows that only the term @xmath208 contributes and that the corresponding value is : @xmath209 } { \left [ \frac{n\pi a}{r}\right ] ^{2}n^{2}\left [ 2-\frac{a}{r}\right ] ^{2}}\simeq 1 \label{un}\ ] ] so that ( [ toto ] ) becomes to this order : @xmath210 = a \label{a}\ ] ] the first order value is therefore merely equal to @xmath13 ; but we will see that @xmath211 may strongly differ from this value if higher orders effects in the potential range are included . [ [ larger - l - values ] ] larger @xmath9 values + + + + + + + + + + + + + + + + + + + + + + + + + + + + + formula ( [ pscal ] ) can be generalized to non - zero angular momentum @xmath9 .
the sine functions of the free and of the hard - sphere wave - functions have to be replaced by a spherical bessel function and a linear combination of this function and a neumann function of order @xmath9 respectively .
after normalization we obtain for @xmath212 : @xmath213 } } \left [ j_{l}(k^{^{\prime } } a)n_{l}(k^{^{\prime } } r)-n_{l}(k^{^{\prime } } a)j_{l}(k^{^{\prime } } r)\right ] \,y_{l}^{0}(\hat{r } ) \label{hseigenf}\ ] ] @xmath214 stands for `` modulus''@xcite : @xmath215 the quantification conditions have to be modified ; ( [ enprime ] ) now becomes implicit : @xmath216 where @xmath217 stands for the @xmath9-wave phase shift for wave number @xmath218 , which is given by : @xmath219 in this definition we limit the values of the inverse tangent to the interval @xmath220 and @xmath221 counts the number of times that by increasing @xmath222 from zero to the final value the value of inverse tangent jumps from @xmath223 to @xmath224 . in this way
we obtain @xmath225 as a continuous function of @xmath218 . for large values of @xmath222 ,
the effects of the centrifugal barrier become negligible and the quantification condition becomes independent of @xmath9 so that it reduces to the @xmath226-wave expression ( [ enprime ] ) .
the normalized free wave function is given by ( [ besel ] ) , so that we arrive at the following expression for the square of the scalar product : @xmath227 } \right ] ^2\,\frac 1{(k^{\prime } a)^2m_l^2(k^{\prime } a)-a / r}\,\,j_l^2(ka ) \label{hsscprod}\ ] ] inserting ( [ pscalbis ] ) and ( [ hsscprod ] ) into ( [ toto ] ) and then into ( [ sigma ] ) - or ( [ sigmaprim ] ) - provides the ursell length , which in turn determines the diagonal matrix element of @xmath146 - or @xmath46 for identical particles . for brevity , we just give the results of our numerical calculations of @xmath113 ; in other words we only discuss interactions between particles in different spin states if the particles are indistinguishable .
but there is no difficulty in treating the general case , since the calculation of @xmath52 is very similar ; moreover , as soon as the wave number @xmath57 is sufficiently small , one simply has @xmath228 and @xmath229 . the results concerning @xmath113 are shown in figures 2 and 3 ( more details about the calculations can be found in @xcite ) .
there are two parameters in the problem , @xmath13 and @xmath6 . as convenient dimensionless variables we choose , either the product @xmath230 , or @xmath231 ;
the ratio @xmath232 is then kept as a fixed parameter , which has small values either when the potential range is very small or when the temperature is low .
figure 2 shows that , when @xmath233 , and as long as @xmath230 remains smaller than 1 , the ursell length is indeed equal to @xmath13 ( with a good accuracy ) , so that the theory of the pseudopotential is fully justified . for higher temperatures
, we note that the low energy values of @xmath113 become noticeably different from @xmath13 .
this is not surprising since , when @xmath11 increases , we progressively reach a classical regime of small wave packets ; they no longer have a much larger spatial extent than the potential , while this is necessary for the approximation of the pseudopotential to apply .
we therefore concentrate on low values of the ratio @xmath232 . if the system is made of bosons , since the distributions @xmath44 s contained in ( [ eq - base ] ) have at most the same width as gaussian thermal exponentials , the values of @xmath230 that are relevant in the integral are comparable to 1 , or smaller ; figure 2 then shows that the use of our theory or of the theory of pseudopotentials leads to the same results ; see the discussion of section [ rotation ] , in particular formula ( [ variation ] ) . but assume now that the system is made of fermions , and that the temperature is progressively lowered at constant density .
if the system is degenerate , the width of the functions @xmath234 s is determined by the fermi momentum @xmath235 , which in turn depends on the density ; in other words the width is practically independent of the temperature . meanwhile , if the temperature is lowered more and more , @xmath6 increases so that the product @xmath230 can take arbitrarily large values inside the integral .
on the other hand , @xmath231 remains limited to values smaller than 1 since , for degenerate gases where the distance between the particles is comparable to the inverse fermi momentum in a degenerate gas , the diluteness condition @xmath236 yields : @xmath237 the departure from 1 of the curves of figure 2 , which fall well below this value when @xmath230 increases , shows that significant discrepancies from a pseudopotential theory may indeed be obtained .
this is even more visible in figure 3 , which uses a different variable , the product @xmath231 : even for small values of the ratio @xmath232 , significant departures of the ursell length from @xmath13 are obtained for small values of @xmath231 ; in other words , ( [ corrferm ] ) is no longer a good approximation of the generalized beth uhlenbeck formula .
actually , the smaller @xmath232 , the more pronounced these departures are ; we have an illustration of the consequences of the temperature dependence of @xmath146 , where the effects of the potential are more and more delocalized by thermal effects while @xmath6 increases ; consequently , variations of the matrix element occur even if the range of the potential is very small and even if @xmath207 . in figure 4
, we plot the variations of the diagonal element @xmath238 _ { rel}\mid { \bf r}>$ ] as a function of the relative position @xmath67 ; the results show clearly how the second ursell operator acquires a longer and longer range at decreasing temperatures .
a striking feature of figure 3 is the change of sign of the ursell length which takes place when @xmath231 increases from zero .
the origin of this change is understandable from ( [ toto ] ) , from which one can convince oneself that the contribution of low values of @xmath218 becomes dominant as soon as @xmath57 is sufficiently large ; this is because , while the scalar product @xmath239 with @xmath240 decreases relatively slowly when @xmath57 increases ( as @xmath241 ) , the exponential @xmath242 varies much more rapidly and so that it makes small values of @xmath243 dominate the sum . in other words ,
what determines the diagonal matrix element of @xmath0 is the contribution of interacting states that have a very small relative energy ; because these states evolve more slowly in time than the free wave packets of energy @xmath244 , the net differential result is equivalent to an attraction .
we therefore come to the conclusion that even hard cores can result in an effective attraction between fermions of opposite spins at the surface of a fermi sphere , provided that the temperature is sufficiently low .
but other interesting features also appear ; for instance there is a strong dependence of the matrix element on the relative momentum and , for some value of @xmath57 , the ursell length ( the effective interaction ) vanishes ; probably more important is the fact that the effective interaction increases almost exponentially as a function of @xmath245 , which is nothing but the square of the relative momentum of the interacting particles .
this shows that the correction to the partition function is dominated by processes that take place preferably between particles having almost opposite momenta on the surface of the fermi sphere ( assuming that they have opposite spins ) .
we now come back to the discussion made at the end of section [ physdisc ] concerning the validity of the ursell expansion for a dilute gas of fermions .
the result of this discussion was that the expansion parameter is the product @xmath166 .
this parameter would remain a constant as a function of temperature for a gas of constant density if the ursell length did not vary too much as a function of temperature , so that the situation would be simple .
but in fact , we have actually found for fermi gases that the maximum value of @xmath155 becomes larger and larger at low temperatures.this automatically limits the range of validity of the first order @xmath0 theory to temperatures at which the relevant values of the ursell length are not too large . to determine a limit temperature
, we will look for an asymptotic expression of the ursell length valid for wave vectors of the magnitude of the fermi wave vector @xmath246 . in this case , the terms corresponding to low values of @xmath247 in the sum over the states of interacting pairs of ( [ toto ] ) are dominant .
when @xmath248 the square of the scalar product ( [ hsscprod ] ) reduces to : @xmath249 we replace the square of the scalar product in ( [ toto ] ) by this expression and transform the sum into an integral : @xmath250 where @xmath251 is the fermi energy ( proportional to @xmath252 for a strongly degenerate gas ) .
we are interested in the limit @xmath253or @xmath254 ; we can thus replace the sine by its first order approximation , which yields : @xmath255 the validity criterion of the ursell expansion then becomes : @xmath256 which shows that the theory is valid only when the temperature satisfies the approximate condition : @xmath257 } \label{validity}\ ] ] inside the logarithmic function in the denominator , both the factor @xmath258 and @xmath259 are larger than one , especially the latter for a very dilute gas ; but since the logarithm has slow variations , whatever parameters we choose , in practice the limit temperature fixed by this condition is not lower than is about a tenth of fermi temperature , a rather a high temperature compared to the transition temperature for cooper pairing .
our conclusion is that , for any type of interaction , and even if the density of the gas is fixed at a very low value , at sufficiently low temperature the ursell length becomes larger and larger so that the ursell expansion is no longer a good expansion .
the maximum value of the ursell length is obtained for pairs of fermions with opposite spins and maximum relative momenta , which corresponds to two fermions having opposite momenta on the surface of the fermi sphere ; this is reminiscent of the bcs pairing phenomenon , while the phenomenon takes place at much higher temperatures ( the right hand side of ( [ validity ] ) does not coincide with the standard expression of the critical temperature ) so that stricto sensu it can not be called a precursor of this transition . whether of not it is related to this transition , the change of sign of the ursell length will have a strong effect on the two body correlation function in the system ( we have seen in @xcite that the second ursell operator is closely related to the short range properties of the two body density operator ) .
these results are also reminiscent of the well known work of luttinger and kohn @xcite who predict the occurrence of superconductivity in purely repulsive systems , as well of the more recent work of kagan and chubukov @xcite who also predict in this case @xmath260 wave superfluidity in a dilute fermi gas .
we are planning to investigate this connection in more detail in a future article . to summarize , for both fermionic of bosonic systems , the validity of the ursell expansion is limited , even for a very dilute gas , to temperatures sufficiently above any transition point .
mathematically , in the case of bosons the divergence of the series arises from the distribution function ( or operator ) @xmath44 , while in the case of fermions its origin is the increasing value of the matrix elements of @xmath0 themselves .
our formalism provides a systematic treatment of the interactions in a dilute gas where the basic objects are not the matrix elements of the potential itself but those of temperature dependent operators . in @xcite , we have shown how microscopic , short range correlations between particles could explicitly be taken into account and calculated . in the present article , we investigate the macroscopic properties of the gas by basing the discussion on expression ( [ logz - int ] ) , which resembles a first order perturbative expression of an energy correction , while it actually is rather different .
this is mostly because the matrix element which appears in the expression is not the matrix element of the potential itself , or of some variety of pseudopotential , but the matrix element of the second ursell operator @xmath0 , which corresponds physically to a local boltzmannian equilibrium . except of course in trivial cases where the interaction potential is indeed weak for all values of the relative distances of the particles , which allows for a first order treatment of the potential ,
this introduces significant differences ; the major reason is the temperature dependence of the matrix elements of @xmath0 , which is in general more complex than being merely proportional to @xmath122 ( as would be the case in usual perturbation theory ) .
this matrix element is conveniently described in terms of the ursell length .
we have seen in particular that , for fermions at very low temperatures , effective attractions at the surface of a fermi sphere may take place , independently of the repulsive or attractive character of the potential itself .
moreover , our formalism contains naturally effects such as the statistical exchange between bound molecules and free particles , which may play some role in experiments with alkali atoms at very low temperatures @xcite . many useful and friendly discussions with roger balian took place on the subject of this article .
a.m. and w.m . would like to thank the kastler brossel group for hospitality during stays in paris .
we are grateful for the following financial supports : nsf grant dmr-9412769 ( a.m. ) and int-9015836 in conjunction with a cnrs grant ( f.l . and a.m. ) , and natootan crg 930644 ( f.l . and w.m . )
after completion of this manuscript , we received a preprint from h. stein et al.@xcite who , in the case of bosons , find that the many body scattering length undergoes a divergence when the superfluid transition is approached , an effect similar to what we obtain for fermions .
the significance of this similarity has not been examined in detail yet .
figure 1 : steplike potential used to discuss compensation effects between the attractive and the repulsive part on the scattering length ; @xmath192 is the depth of the attractive potential , @xmath4 its range ; a hard core of range @xmath187 is assumed ( @xmath261 ) . while a relation between these three parameters can be found to make the scattering length @xmath13 vanish , the effect of the potential on the wave function inside its range remains significant .
figure 2 : variations of the ratio between the ursell length @xmath262 and the range @xmath13 of the hard core potential as a function of the dimensionless variable @xmath263 .
the ratio @xmath264 , with values shown in the upper right , is a parameter which takes small values if either the temperature is sufficiently low or the radius of the hard cores sufficiently small .
figure 3 : variations of the same quantity as in figure 2 , but as a function of the dimensionless variable @xmath231 ; this representation is convenient for the discussion of fermions ( where it characterizes interactions between particles in opposite spin states ) . figure 4 : variations of the diagonal matrix element @xmath265 _ { rel}\mid { \bf r}>$ ] as a function of @xmath67 , for various values of the parameter @xmath232 ; at low temperatures , the thermal increase of the range of the operator is clearly visible . | we study in more detail the properties of the generalized beth uhlenbeck formula obtained in a preceding article .
this formula leads to a simple integral expression of the grand potential of any dilute system , where the interaction potential appears only through the matrix elements of the second order ursell operator @xmath0 .
our results remain valid for significant degree of degeneracy of the gas , but not when bose einstein ( or bcs ) condensation is reached , or even too close to this transition point .
we apply them to the study of the thermodynamic properties of degenerate quantum gases : equation of state , magnetic susceptibility , effects of exchange between bound states and free particles , etc .
we compare our predictions to those obtained within other approaches , especially the `` pseudo potential '' approximation , where the real potential is replaced by a potential with zero range ( dirac delta function ) .
this comparison is conveniently made in terms of a temperature dependent quantity , the `` ursell length '' , which we define in the text .
this length plays a role which is analogous to the scattering length for pseudopotentials , but it is temperature dependent and may include more physical effects than just binary collision effects ; for instance , for fermions at very low temperatures , it may change sign or increase almost exponentially . as an illustration ,
numerical results for quantum hard spheres are given . |
the study of electron transfer comprises a rich variety of systems in many different areas such as chemistry , biology , and life sciences @xcite . although electron transfer processes are mainly attributed to electrochemical applications , they are conceptually related to molecular electronics @xcite and electron transport in low dimensional materials in solid - state physics . in that context ,
semiconductor heterostructures represent a popular physical system for the investigation of mesoscopic transport @xcite and tunnelling phenomena @xcite .
the main reason for this is the high mobility and the rather long mean free path of the charge carriers populating them .
standard beam epitaxy techniques make the accurate growth of alloys of such materials on substrates possible , and the nearly identical lattice parameters , together with the possibility of controlling the band gap , turns the combination gaas / al@xmath0ga@xmath1as into an ideal candidate for building complex low dimensional structures with quantum wells and tunnel barriers . moreover , these setups open various ways to study tunnelling in time - dependent systems @xcite .
a straightforward possibility for introducing a time - dependence is the application of an ac transport voltage which only modulates the energies of the electrons in the leads while the potential inside the mesoscopic region remains time - independent .
this kind of driving allows for a description within tien - gordon theory @xcite which expresses the dc current in terms of the static transmission and an effective distribution function for the lead electrons . if the time - dependence enters via an external microwave field or an ac gate voltage , however , such an approach is generally insufficient @xcite .
a remarkable difference with respect to the static situation is the emergence of inelastic transport channels stemming from the emission or absorption of quanta of the driving field . for a periodically time - dependent transport situation
, however , we expect the transmission probabilities and , consequently , the resulting current to be time - dependent as well .
this follows indeed from a recently presented floquet theory for the transport through driven tight - binding systems @xcite .
for the computation of the dc current , this approach justifies the applicability of a landauer - like current formula where the static transmission is replaced by the time - averaged transmission of the time - dependent system .
the transmission of both the elastic and the inelastic transport channels can depend sensitively on the driving parameters ; the contribution of certain channels can even vanish .
for the transport across two barriers which enclose an oscillating potential well , wagner @xcite showed that it is possible to suppress the contribution of individual inelastic scattering channels .
the total current , however , is given by the sum over all channels , and thus it is not possible to isolate the contribution of a single channel in a current measurement .
by contrast , in the case of transport through a two - level system with attached leads , it has been found that driving with a dipole field has directly observable consequences .
there , the driving not only affects the contribution of individual transport channels , but the dc current can be suppressed almost entirely @xcite .
therefore , for the appearance of this _ coherent current suppression _ , it is essential that the central region consists of at least two weakly coupled wells which oscillate relative to each other @xcite . ,
@xmath2 and the widths @xmath3 , @xmath4 .
the dotted lines mark the energy of a metastable tunnel doublet with splitting energy @xmath5 .
the left well is subject to an electric dipole field generated by an alternating gate voltage with amplitude @xmath6.,width=283 ] in this work , we explore the possibility of coherent current suppression in double - well heterostructures .
thereby , we compare two theoretical approaches to describe coherent transport in quantum - well structures : the transfer - matrix method and a tight - binding approach . as a model we consider the triple - barrier structure sketched in fig .
[ fig : hetero - model ] where the driving enters via an oscillating gate voltage which modulates the bottom of the left well .
the applied transport voltage is assumed to shift the fermi energy of the left lead by @xmath7 with @xmath8 being the electron charge .
we note that since the time - dependent gate voltage affects only one well , the structure depicted in fig . [
fig : hetero - model ] is sufficiently asymmetric to also act as an electron pump , i.e. to induce a non - zero current for @xmath9 @xcite . in this work , however , we focus on the transport properties in the presence of a finite bias voltage .
for the exact numerical computation of the transmission probabilities , we employ the transfer - matrix method developed by wagner @xcite , which is reviewed in section [ sec : transfer - matrix ] . in section [ sec : tight - binding ] , we introduce the related tight - binding system for which the transport properties can be calculated analytically within a high - frequency approximation scheme @xcite .
the predictions from the perturbative approach are compared to the exact solution in section [ sec : results ] .
following landauer @xcite , we consider the coherent mesoscopic transport as a quantum mechanical scattering process .
the central idea of this approach is the assumption that sufficiently far from the scattering region , the electronic single - particle states are plane waves and that their occupation probability is given by the fermi function with the chemical potential depending on the applied voltage .
the unitarity of evolution under coherent ac driving allows us to write the resulting currents as @xcite @xmath10 , \ ] ] where @xmath11 denotes the total transmission probability i.e. summed over transverse modes and outgoing inelastic channels of an electron with energy @xmath12 from the left lead to the right lead while @xmath13 describes the respective scattering from the right to the left lead . for time - independent conductors ,
the time - reversal symmetry of the quantum mechanical scattering process together with the energy conservation ensures @xmath14 such that , in the absence of a transport voltage , the current vanishes .
this is not the case for a general time - dependent structure @xcite . when the total hamiltonian is time - periodic due to an external driving field , @xmath15
, one can apply floquet theory @xcite .
it states that the corresponding time - dependent schrdinger equation has a complete set of solutions of the form @xmath16 where @xmath17 denotes the so - called floquet states , and @xmath18 the so - called quasienergies in analogy to the quasimomenta of bloch theory .
owing to their time - periodicity , we can decompose the floquet states into a fourier series @xmath19 the form of the floquet states suggests that during the scattering , an electron with initial energy @xmath12 evolves into a _
coherent _ superposition of states with energies @xmath20 .
the arbitrary integer @xmath21 is referred to as the sideband index ; the inelastic channels are called sidebands .
we emphasise that , despite the existence of a band bottom , the summation over @xmath21 in eq .
is unrestricted .
a proper calculation of the dc current through a time - dependent scatterer must now include these inelastic channels , and the landauer formula of eq .
has to be conveniently generalised to take them into account .
since the floquet scattering states can be thought of as having been created from the orthogonal dc states by adiabatically switching on the driving , they too must be orthogonal , so that it is sufficient to sum over channels incident from both leads @xcite
. in the transfer - matrix method described below , the floquet states are decomposed into plane waves throughout the driven heterostructure , so that they can be appropriately matched with the scattering channels in the leads .
this decomposition allows us to separate the time- and the space - dependent parts of the wave function , thus obtaining directly the time - independent probabilities that go into eq . .
for a spatially constant potential with a time - dependent gate voltage @xmath22 , the solution of the schrdinger equation is readily obtained to read @xmath23 with the accumulated phase @xmath24 its time - periodicity follows from the zero time - average of the gate voltage .
neighbouring layers of a heterostructure may have different ac voltages applied in addition to different band - edges . as a consequence , the wave functions in eq . which solve the schrdinger equation in each layer do not coincide in the general case .
the solution for the complete system has to be constructed by matching the corresponding wave functions at the interfaces between layers . with this goal in mind , we assume that the wave function is a solution of the schrdinger equation for the time - dependent hamiltonian @xmath25 and that , moreover
, @xmath26 is an eigenfunction of the time - independent hamiltonian @xmath27 with the spatially piecewise constant effective mass @xmath28 , and has the general form @xmath29 the wave vector @xmath30^{1/2}\ ] ] describes travelling as well as decaying waves ( bound states ) for complex and real values of @xmath31 , respectively .
the matching conditions at an interface follow from the fact that both the wave function and the flux have to be continuous , i.e. at @xmath32 @xmath33 this yields an infinite system of algebraic equations for the coefficients @xmath34 and @xmath35 in each layer . inserting the fourier expansion of the phase in eq .
, @xmath36 where @xmath37 is the @xmath38-th order bessel function of the first kind , allows one to recast these equations for an interface between layers i and ii at @xmath39 in matrix form : @xmath40 the matrices @xmath41 and @xmath42 with elements @xmath43 and @xmath44 , respectively , are of infinite dimension , and contain the coefficients of all possible scattering channels , i.e. photon exchanges between the incoming electron and the driving field , at either side of the interface .
their precise form depends on whether or not a layer is affected by the time - dependent gate voltage .
the transfer matrix @xmath45 between two sides of a layer of width @xmath46 is then defined as @xcite @xmath47 this definition in terms of layers , rather than using a similar one across interfaces , is physically more sensible , since the matrices depend on the properties of just one layer .
they are also numerically easier to implement , as there are fewer qualitatively different matrices to deal with .
to calculate the total transfer matrix across the structure , we have to multiply the matrices across all different layers and obtain @xmath48 where @xmath49 and @xmath50 represent the initial and final matrices at the ends of the heterostructure . with the elements @xmath51 of the total transfer matrix we can find the probability that an electron with energy @xmath52 in lead @xmath53 is scattered into a channel with energy @xmath54 in lead @xmath55 , with integer @xmath56 .
the diagonal elements @xmath57 are closely related to the ( static ) transmission probability @xmath58 , while the off - diagonal elements @xmath59 describe the effects of the absorption or emission of @xmath60 photons on the transmission probability of the electron . for flat conduction bands on both sides of the heterostructure , the wave functions in the contacts are plane waves , and in this case the proper boundary conditions to describe an electron incident from , say , the left - hand side at energy @xmath12 are @xmath61 and @xmath62 .
the transmission probability in sideband @xmath21 is then defined as @xmath63 where @xmath64 and @xmath65 represent the wave vectors on the right- and left - hand side in sidebands @xmath21 and 0 , respectively . in a numerical implementation of the transfer - matrix technique , it is necessary to truncate the infinite matrices .
thereby for consistency , a proper cut - off has to be so large that unitarity of the scattering process is preserved , i.e. @xmath66 where @xmath67 represents the reflection probability of an electron to be reflected from energy @xmath12 into a sideband at energy @xmath68 on the same side .
the number of sidebands that need to be taken into account to meet a given accuracy ( which in our calculations was set to @xmath69 ) depends essentially on the ratio @xmath70 , as this is the argument of the bessel function @xmath71 that determines the weight of each sideband @xmath21 . to proceed ,
one starts at the initial value of @xmath72 with a tentative number of sidebands , and increases it for growing driving amplitudes if a check with eq .
suggests that unitarity is breaking down .
when particle number conservation is restored one can go to a higher @xmath6 .
transfer matrices such as those employed here have the advantage of being easily scalable to arbitrarily complex structures .
the combination of flexibility in structural properties and numerical accuracy makes this method well - suited to the study of strongly driven semiconductor heterostructures .
and each well couples with @xmath73 and @xmath74 to the associated lead .
the state @xmath75 in the left well is driven by a oscillating gate voltage with amplitude @xmath6 .
in addition , an external bias @xmath76 is applied.,width=283 ] a different approach to study resonant tunnelling in a driven double - well structure is based on the adoption of a tight - binding ( tb ) picture .
figure [ fig : tb - model ] depicts the tb configuration corresponding to the heterostructure introduced above .
then , the hamiltonian of the system is given by @xmath77 where the individual terms describe the driven quantum wells , the electron reservoirs in the leads , and the coupling of the left and the right well to the respective neighbouring lead .
for simplicity , we neglect the electron spin . within the framework of the tb approximation
, the time - dependent quantum - well hamiltonian reads @xmath78 an electron can be localised in the left or right well , whereupon the fermion operator @xmath79 ( @xmath80 ) annihilates ( creates ) an electron in the respective well ( @xmath81 ) .
these localised states @xmath75 and @xmath82 are coupled by the tunnelling matrix element @xmath83 . for convenience ,
the energy scale is set such that the on - site energies of the two resonant tb levels are zero and lie exactly halfway in between the transport bias window defined by the chemical potentials @xmath84 and @xmath85 .
the second term of the hamiltonian accounts for the harmonic driving of the traversing electrons in the left well via an oscillating gate voltage with amplitude @xmath6 and period @xmath86 .
the leads are modelled as ideal fermi gases with the hamiltonian @xmath87 where @xmath88 ( @xmath89 ) annihilates ( creates ) an electron in the lead with energy @xmath90 with @xmath81 . as an initial condition
, we employ the grand - canonical ensembles of electrons in the leads at inverse temperature @xmath91 .
therefore , the lead electrons are characterised by the equilibrium fermi distribution @xmath92\}^{-1}$ ] . the localised state in each well couples via the tunnelling matrix element @xmath93 to the state @xmath94 in the respective lead .
the hamiltonian which describes this interaction has the form @xmath95 the lead
well coupling is entirely specified by the spectral density @xmath96 . since , for the system at hand , the bandwidth of the conduction band of the leads is much larger than the energy regime where transport happens , the spectral densities are practically constant , i.e. @xmath97 , which defines the so - called wide - band limit . in order to determine the time - averaged dc current which matches eq . for the tb approximation
, we employ a generalised floquet approach to solve the corresponding heisenberg equations of motion and derive an expression for the retarded green s function in terms of floquet states .
this result is applied to study transport by evaluating the operator @xmath98 $ ] for the time - dependent current through contact @xmath99 . in the next subsection
we will also derive a analytic expression for the current valid in the high frequency limit with respect to the driving .
we start out by stating the solution of the heisenberg equations of motion for the lead operators , which is given for the left lead by @xcite @xmath100 for the corresponding solution of @xmath101 , @xmath53 has to be substituted by @xmath55 .
now inserting these solutions into the heisenberg equations of motion for the quantum wells yields the two coupled linear differential equations @xmath102 here , within the wide - band limit , the coupling to the leads has been eliminated in favour of the spectral density @xmath103 and the fermionic fluctuation operator @xmath104 assuming that the initial conditions are those of the grand canonical ensemble , this gaussian noise operator satisfies @xmath105 then the operator for the time - dependent current through the left lead becomes @xcite @xmath106\ ] ] with a corresponding expression for @xmath107 .
here we made use of the wide - band limit . to evaluate the time - dependent current
, we thus have to find the solution for the inhomogeneous set of the quantum langevin eqs . and of the quantum - well operators , which is formally given by @xmath108\ ] ] in the stationary limit @xmath109 with @xmath81 .
what remains is to determine the retarded green s function @xmath110 .
this is where floquet theory comes into play by making use of the @xmath111-periodicity of the driving . solving the floquet eigenvalue equation @xmath112 of the physical problem at hand , where @xmath113 , we get the floquet states @xmath114 and the complex - valued quasienergies @xmath115 .
note that the prior equation is , in contrast to the usual floquet equation , non - hermitian .
this is due to the presence of the self - energy @xmath116 @xcite . with the corresponding expression for the propagator @xmath117
, the retarded green s function assumes the form @xmath118 where @xmath119 is the heaviside step function .
the dc current is now obtained by calculating the expectation value @xmath120 and averaging over one driving period .
this time - averaging will cancel those terms of @xmath121 , which are responsible for a @xmath122-periodic charging of the wire . after eliminating backscattering terms @xcite ,
we arrive at the very compact form for the final result @xmath123,\ ] ] where the total transmission probabilities are given by @xmath124 and @xmath125 , which resemble the fisher - lee relation @xcite .
the above equation also holds for the current through the right contact owing to charge conservation .
the green s function @xmath126 is the @xmath127-averaged fourier transformed of the propagator .
physically , it describes the propagation of a transmitted electron with initial energy @xmath128 from one lead to the other lead undergoing scattering events with emission ( @xmath129 ) or absorption ( @xmath130 ) of @xmath131 photons , or being transmitted elastically ( @xmath132 ) .
the floquet treatment of the present transport problem allows for the implementation of a stationary perturbation scheme for driving frequencies much larger than all other frequency scales of the system @xcite .
this approach has recently been extended to transport situations which are characterised by the presence of leads @xcite ; here we only outline the derivation and refer the reader to ref .
a particular benefit of this perturbation scheme is the achievement of a physical understanding of the transport processes by approximately mapping the time - dependent problem to a static one with renormalised parameters .
for the static situation , in turn , the current is well known by eq . , where the transmission in the wide - band limit reads @xmath133 .
this looks similar to the driven case but with @xmath132 and therefore we have in contrast to the driven system @xmath134 . for the static system given by setting @xmath72
, we obtain for the transmission @xmath135 assuming equal coupling to the leads ( @xmath136 ) .
the starting point of the approximation scheme is the unitary transformation @xmath137 which is first applied to the quantum - well hamiltonian . for sufficiently large driving frequencies @xmath138 , a separation of time scales is performed by this transformation .
thereby , fast oscillations of the transformed hamiltonian are neglected by averaging over a driving period @xcite . finally , we arrive at the effective hamiltonian for the quantum wells @xmath139 which is of the same form as in the static case but with the effective tunnelling matrix element @xmath140 @xcite .
@xmath141 is the zeroth order bessel function of the first kind .
therefore , the resulting effective transmission @xmath142 with the substitution @xmath143 in eq .
is controllable via the driving parameters and even vanishes at zeros of the bessel function .
the transformation also affects the lead well coupling .
if we apply @xmath144 also to @xmath145 and solve the heisenberg equations for the lead and quantum - well operators in the wide - band limit , we can eventually extract the new fluctuation operator .
for the left lead one finds @xmath146 whereas @xmath147 remains unaffected . now calculating the correlation function and time - averaging it over one driving period to neglect the @xmath122-periodic contributions
, the resulting expression assumes the form but with the fermi function of the left lead replaced by the effective electron distribution @xmath148 the squares of the @xmath21th - order bessel function of the first kind in this expression weight those processes where an electron with energy @xmath128 is transmitted from the left lead to the double - well system under the emission ( @xmath129 ) or absorption ( @xmath130 ) of @xmath131 photons . the effective electron distribution exhibits steps at the energies @xmath149 and is constant elsewhere . with the effective quantities
@xmath142 and @xmath150 the driven problem is ascribed for fast driving to a static one . since @xmath142 is sharply peaked around @xmath151 and @xmath152 and @xmath153 are constant for finite voltage , the current in the high - frequency approximation results in @xmath154\ ] ] applying the effective parameters to the current formula . here
@xmath155 is a shorthand notation for the integer part of @xmath156 . in order to compare the transfer - matrix and the tight - binding approach
, we have to ensure that the same physical situation is addressed . as a matching condition we compare the transmission @xmath157 in the time - independent case ( @xmath72 ) .
the level splitting energy @xmath5 due to the central tunnel barrier is extracted from the resonance peaks of the doublet states computed within the transfer - matrix method .
solving for @xmath158 in eq .
with @xmath151 and @xmath159 and @xmath83 taken from the previous calculation , the corresponding lead
well coupling for the tight - binding system is determined .
we now turn our attention to the coherent control of current .
tunnelling suppression in a closed , driven system is known for more than a decade .
for example for a driven bistable potential , tunnelling breaks down at exact crossings of the quasi - energy spectrum , i.e. one observes the so - called _ coherent destruction of tunnelling _
tunnelling suppression has been studied in a number of cases @xcite , but the investigation in a transport context , i.e. in an open system where an appropriate treatment of the leads is crucial , has received attention only recently @xcite . .
the chosen parameters are @xmath160 , @xmath161 , @xmath162 and @xmath163 .
the corresponding parameters for the barriers are the same as those of fig .
[ fig : hetero - model].,width=283 ] surveying the time - averaged current calculated numerically from the transfer - matrix and the tight - binding method plotted in fig .
[ fig : tmvstb ] , we observe current minima for distinct values of @xmath164 for frequencies in the microwave regime .
the reason for the current suppressions becomes apparent by comparison with the high - frequency approximation , which exhibits minima close to those of the transfer - matrix and tight - binding curves .
the current vanishes whenever the ratio @xmath165 assumes a zero of the bessel function @xmath166 , i.e. for the values 2.405 , 5.520 , 8.654 , , since then @xmath167 . by varying the ratio between driving amplitude and frequency
, we can thus tune the tunnelling between the two wells and thereby control the current @xcite . for a frequency @xmath168
, the analytical expression shows a remarkable agreement with the exact tight - binding result for @xmath169 .
the inset of fig .
[ fig : tmvstb ] shows the minimum current at the first suppression decays as a function of the driving frequency @xmath170 .
this is expected from the good agreement between the numerical results and the high - frequency approximation , because the approximation accounts for the first order term in a perturbative scheme in @xmath171 @xcite .
higher order contributions are included in a numerically exact calculation , which results in a non - vanishing current at the minima .
a similar @xmath170-dependence is observed also for the transfer - matrix formalism . , @xmath172 , for different barrier widths and heights . the parameters for the first three data sets are @xmath173 nm , @xmath174mev and @xmath175mev , whereas for the last one ( @xmath176 )
, we chose @xmath177mev and @xmath178mev.,width=283 ] while the general shape and magnitude of the current are very similar for both models , there still appears a small difference in the location of the minima for the relatively low barriers chosen in fig .
[ fig : tmvstb ] . for a continuous potential ,
the current assumes minima at values of @xmath164 higher than those predicted by the tight - binding description .
we can understand this shift by analysing , for given @xmath170 , the deviation @xmath179 of the driving amplitude @xmath180 for which the current exhibits its first minimum .
the amplitude @xmath181 corresponds to the first zero of @xmath141 . in fig .
[ fig : wkb ] we plot @xmath182 as a function of @xmath183 , where @xmath184 and @xmath185^{1/2}$ ] , i.e. @xmath183 is the instanton action in units of @xmath186 and @xmath187 is the wkb transmission probability of the outer barriers in fig .
[ fig : hetero - model ] . here
@xmath188 is the corresponding barrier height and @xmath189 denotes the average chemical potential representing approximately the mean energy of the resonance doublet .
if the width of the outer barriers is kept fixed , @xmath182 decreases for growing @xmath183 because then the resonance energies are further away from the barrier edge .
therefore , the wave functions of the well states become more localised .
this situation corresponds in the tight - binding picture to a lead
well coupling @xmath158 that is almost energy independent and thus reproduces the wide - band limit .
furthermore , this argument is used to explain the smaller deviation observed with thinner barriers for the same @xmath183 , since @xmath190 is much larger in that case . as can be seen by comparing data sets for different central barrier heights in fig .
[ fig : wkb ] , an increase of the height of the central barrier @xmath191 reduces the level splitting @xmath5 , that is , the overlap between the localised states in the left and right well in a tight - binding description . thus the tight - binding and the transfer - matrix results converge as a function of the barrier height .
finally , it is important to note that varying any of the barriers affects the transmission properties of the whole heterostructure , in contrast to the tight - binding model , where the different coupling parameters can be controlled independently .
we have demonstrated that the dc current across a double - well heterostructure can be suppressed by the purely coherent influence of an oscillating gate voltage .
we have used a transfer - matrix method as an exact approach to compute tunnelling currents through such a system . we compared these results to that obtained from a tight - binding floquet description .
in particular , we find that the current suppression is controlled by the ratio of the driving frequency and amplitude . this can be understood by exploring the high - frequency limit within the tight - binding formalism . in this perturbative scheme ,
the time - dependent system is mapped onto a static one with renormalised parameters , that is , with an effective hopping matrix element accounting for inter - well coupling and with an effective electron distribution for the attached left lead . since the effective inter - well coupling depends on the ratio between driving amplitude and frequency , the transport properties of the double well can be adjusted through the driving parameters , with the effective behaviour ranging from transport through an almost open channel to a regime of rare tunnel events .
the results presented in this work strongly support the idea that transfer - matrix and tight - binding descriptions of quantum transport are equivalent provided the barriers are sufficiently high . in this case
the lowest resonance states in the wells are rather localised and consequently the tight - binding description becomes accurate . for a proper choice of parameters , we find a good agreement between the exact transfer - matrix calculation and the results obtained within the tight - binding formalism .
the study presented here shows that the coherent control of time - dependent electron transport can be investigated with current semiconductor nanotechnology .
we thank sbastien camalet , gert - ludwig ingold , and jrg lehmann for helpful discussions .
two of us ( m.r . and f.s . )
wish to thank mathias wagner for introducing us to a numerical code based on ref .
this work has been supported through a joint germany - spain accin integrada ( no .
ha2003 - 0091 ) .
financial support is also acknowledged from mec ( spain ) , grant no .
bfm2001 - 0172 , fundacin ramn areces , and dfg ( germany ) , graduiertenkolleg 283 and sonderforschungsbereich 486 .
we also thank the centro de computacin cienti="10 fica ( uam ) for technical and computational support . | we explore the possibility of inducing in heterostructures driven by an ac gate voltage the coherent current suppression recently found for nanoscale conductors in oscillating fields .
the destruction of current is fairly independent of the transport voltage , but can be controlled by the driving amplitude and frequency . within a tight - binding approximation
, we obtain analytical results for the average current in the presence of driving .
these results are compared against an exact numerical treatment based on a transfer - matrix approach . , , , , and quantum transport , driven systems , heterostructures , tunnelling 05.60.gg , 85.65.+h , 72.40.+w |
pre - protostellar cores ( ppcs or starless cores ) are thought to represent the initial stages of low - mass ( m @xmath15 few ) star formation .
ppcs are identified with dense molecular cores ( @xmath16 ) which lack evidence for an internal heating source .
their column density is traced by dust continuum emission ( ward - thompson et al .
1994 ; ward - thompson , motte , & andr 1999 ; shirley et al .
2000 ; tafalla et al .
2002 ) and nitrogen - bearing molecular tracers such as nh@xmath5 and n@xmath1h@xmath2 ( myers & benson 1983 , jijina et al . 1999 ,
caselli et al .
2002 , crapsi et al .
it is important to characterize their physical structure ( density , temperature , kinematical , and chemical ) to constrain the initial conditions of low - mass star formation .
it is also important to understand the evolution of the physical structure of ppcs to trace the evolution of a nascent protostar .
lynds 1498 is a nearby molecular core ( @xmath17 pc ) located in the west - central portion of the taurus molecular cloud ( see cambrsy 1999 ) .
l1498 was classified as a pre - protostellar core by its submillimeter dust emission and lack of an iras detection at 100 ( ward thompson et al .
the relatively weak submillimeter dust emission indicated that l1498 was a low density core .
l1498 has been extensively studied with molecular line observations and using theoretical chemical modeling ( rawlings et al 1992 , taylor et al 1996 ) .
l1498 was initially identified kinematically as a core potentially on the verge of protostellar collapse ( zhou et al .
subsequent molecular line observations have shown dramatic evidence for the freezing of gas phase molecules onto dust grains ( e.g. , h@xmath1co , cs , ccs , c@xmath18o ; wang 1994 ; lemme et al . 1995 ; kuiper , langer , & velusamy 1996 ; willacy , langer , & velusamy 1998 , tafalla et al .
2004 ) ; as a result , many molecular species can not reliably trace the densities in the center of the core .
prominent exceptions are the nitrogen - bearing molecules and n@xmath1h@xmath2 which resist depletion ( flower , pineau des forts , & walmsley 2005 ) .
even though and n@xmath1h@xmath2 remain in the gas phase in cold , dense molecular cores , the optical depth , gas temperature , and abundance profile must be reliably determined in order to accurately calculate the density structure of the core . optically thin dust emission at submillimeter wavelengths is a good tracer of temperature and density since the specific intensity is proportional to the integral of density , opacity , and planck function along a line - of - sight @xmath19 ( adams 1991 , shirley et al .
knowledge of the temperature and density structure is essential to a correct interpretation of chemical and kinematic models of l1498 .
previous dust continuum studies of ppcs used radiative transfer models to constrain the density and temperature structure of the cores ( e.g. evans et al .
2001 , zucconi et al .
2001 , stamatellos & whitworth 2004 ) .
the density structures are well characterized by bonner - ebert spheres ( hereafter , be ) : pressure bounded , isothermal solutions to the equations of hydrostatic equilibrium that do not include the effects of magnetic fields or turbulence .
a general picture of the initial stages of low - mass star formation involves the evolution of a be sphere from low central density to high central density , perhaps retarded by magnetic pressure , until a central hydrostatic core develops . if we wish to understand the beginning of this process , we must identify and study low density ppcs .
in this paper we present deep submillimeter maps of l1498 ( 2.1 ) with radiative transfer modeling which includes a more sophisticated treatment of the heating of the ppc ( 3 ) .
the results of dust continuum modeling are discussed in 4 .
a comparison between the derived properties from dust continuum observations of ppcs that were observed with iso and scuba is discussed in 4.1.2 .
we also present the highest resolution single - dish n@xmath1h@xmath2 and c@xmath5s maps of l1498 ( 2.2 ,
2.3 ) with analysis of variations in the column density , abundance , and velocity ( 4.2 , 4.3 ) .
l1498 was observed simultaneously at 850 and 450 with scuba on the 15 m james clerk maxwell telescope during the nights of august 29 and 30 , 1998 .
a total of fifty jiggle maps were made towards the source , each map consisting of four 64-point jiggle maps with 256s of on - source integration time .
the total on - source integration time was 3.55 hours .
previous spectral line maps showed that l1498 is extended in a se to nw direction ; therefore , the chop angle was set to a constant position angle of 20 with a 120 chop throw to chop perpendicular to the major axis of the core . non - azimuthal chopping results in a slight reduction in signal - to - noise in the map since a symmetrical chopping pattern does not lie at constant airmass . a preliminary map made in april
, 1998 showed that the source was detected to the edge of a single 64-point jiggle map field - of - view ( @xmath20 ) .
therefore , the map was extended using three offset 5-pointing maps , each with 30 spacing ( see figure 1a ) .
the five - pointing maps were centered at ( 0,0 ) , ( @xmath2130 , @xmath2230 ) , and ( @xmath2230,@xmath2130 ) from the central coordinates of j@xmath23 @xmath24 @xmath25 @xmath26 @xmath27 , @xmath28 @xmath29 @xmath30 @xmath31 .
the final maps spans @xmath32 in right ascension and declination .
each map was reduced using the standard surf ( scuba user reduction facility ) reduction routines ( jenness & lightfoot 1997 ) .
each 64-point jiggle map was flat - fielded and corrected for chop throw , extinction , and sky noise .
the 450 images were reduced using a set intensity scale to ensure consistent identification of improper sky noise subtraction .
the telescope pointing was checked every hour .
the largest pointing shift was @xmath33 ; therefore , the standard pointing offsets from the five - pointing maps were used to shift and add together the individual maps .
skydips at 850 and 450 were performed every hour during both nights .
these skydips are compared to the opacity measured every 10 minutes at 225 ghz ( @xmath34 ) and 350 from tippers located at the caltech submillimeter observatory .
the @xmath34 data was only available for the night of august 30 . using the relationship derived by archibald et al .
( 2002 ) between @xmath34 and the skydip - determined opacity at 850 ( @xmath35 ) , @xmath36 , we found excellent agreement between our skydips and the scaled 850 opacity ( figure 2 ) .
therefore , we use the the scaled @xmath34 values to correct for the 850 opacity in individual jiggle maps on the night of august 30 .
we linearly interpolated between 850 skydips for the night of august 29 .
since the sky opacity at 350 and 450 is very sensitive to short term variations in atmospheric precipitable water vapor , we used the tipper opacity at 350 to monitor the variability of @xmath37 between the hourly skydips .
we found an average ratio of @xmath37 skydips to @xmath38 of @xmath39 for both nights .
the 450 opacity was corrected for each jiggle map using the scaled @xmath38 opacity interpolated from the tipper measurements bracketing each jiggle map .
the reduced maps are shown in figure 1 .
l1498 was detected at both wavelengths with a peak signal - to - noise of @xmath40 and @xmath41 for the 850 and 450 maps , respectively .
the contours are spaced by @xmath42 for both images .
the submillimeter images are characterized by an amazingly flat intensity plateau .
a 1200 continuum map , smoothed to 20 resolution ( tafalla et al .
2004 ) , is also shown for comparison .
the fwhm intensity contour of the 1200 , 850 , and 450 maps are very similar and well fitted by an ellipse with a major axis ( @xmath43 ) of 197 , minor axis ( @xmath44 ) of 108 , and position angle of 122 .
the 850 and 450 ellipses have similar centroids ( @xmath45 , @xmath46 ) with an uncertainty of @xmath47 with respect to the pointing center ( 2.1.1 )
. the hwhm radius of the core , defined by the geometric mean of the major and minor axes ( @xmath48 ) , is 73 .
while this radius may be used an a representative size of the core , it is important to realize that the fwhm contour is asymmetric with an aspect ratio ( @xmath49 ) of 1.8 .
furthermore , the 1200 and 850 maps are non - axisymmetric with a sharper gradient in the intensity along the northeast edge compared to the southwest edge . the iso map at 170 is also shown in figure 1b ( ward - thompson et al .
this is the shortest wavelength at which the far - infrared emission has been observed to peak on the core ; iso 90 observations detect a more diffuse emission peak approximately 3 south of the submillimeter centroid .
the scale of the 170 map is @xmath50 , twice the scale in the ( sub)-millimeter maps .
the iso beam was very large ( @xmath51 ) at 170 ; therefore , the map does not show the non - axisymmetric structure observed on smaller scales at 850 and 1200 .
the core is elongated , but with a slightly different position angle on larger scales ( pa @xmath52 ) .
far - infrared emission is detected on very large scales , up to 300 ( 42,000 au ) from the centroid of the submillimeter emission .
azimuthally - averaged , normalized radial profiles of the 850 , 450 , and 1200 maps were calculated .
the images were re - binned to pixels with half beam spacing ( 7 at 850 microns and 3.5 at 450 ) corresponding to the nyquist sampling limit in the map .
the centroid was chosen to be the average of the 850 and 450 centroids ( @xmath45 , @xmath46 ) .
the 1200 profile was binned at 10 since the image was smoothed by a 20 gaussian ( tafalla et al .
the flat intensity plateau has the same extent in all of the profiles .
since l1498 is a very elongated core , azimuthally averaged profiles will be weighted towards a flatter intensity profile ( see jrgensen et al .
this effect may be quantified by considering the normalized profile within sectors centered on the major and minor axes . if the sector centered on the major axis is chosen with the opening angle given by , @xmath53 where @xmath54 is the ratio of the minor to major axes ( @xmath55 ) , then the areas of sectors are equal .
the azimuthally averaged and sector - averaged profile will be modeled separately in 3 .
uranus and afgl618 were observed on august 29 and 30 as flux calibrators and for beam profiles at 850 and 450 .
the total flux of uranus was 74.0 jy at 850 and 195.5 jy at 450 .
the apparent diameter of uranus was 37 effectively making uranus a point source at both wavelengths ( see shirley et al .
2000 for an analysis of uranus finite size on the beam size ) .
the flux of afgl618 was assumed to be @xmath56 jy / beam at 850 and @xmath57 jy / beam at 450 ( scuba secondary calibrator webpage ) . since we are interested in the total flux observed towards l1498 , we calibrated the flux in apertures of 40 and 120 . flux densities are calculated for an aperture diameter , @xmath58 , using @xmath59 , where @xmath60 was the voltage measured at wavelength @xmath61 in an aperture of diameter @xmath58 at airmass @xmath62 and @xmath63 is the calibration factor . the fluxes for l1498 are included in the source properties in table 1 .
the uncertainty on the calibration factor was calculated by propagating the errors using , @xmath64\ ] ] where is the flux density ( jy ) , @xmath65 is the calibration factor ( jy / v ) , @xmath66 is the total voltage observed within the aperture ( v ) , and @xmath67 is the area of a pixel included within the aperture .
the scuba beam profiles were determined using observations of uranus at 850 and 450 .
uranus was observable only at the beginning of our observing shift and only at high airmass ( @xmath68 ) .
the uranus profile can not be traced beyond 60 at 850 and beyond 50 at 450 . unfortunately , afgl618 is not strong enough to be used for extended beam profiles ( @xmath69 ) .
the uranus profiles are very similar to the average uranus beam profile determined from maps made in april , 1998 ( shirley et al . 2000 ) .
since all of our observation were made during the morning observing shift ( 1:30am hst - 9:30am hst ) , the dish should have cooled sufficiently such that the beam shape is stable .
no significant trends were seen in the april 1998 beam maps observed during the morning observing shift ; therefore , we feel justified in using the uranus beam profile for all second shift observations .
the sed was constructed from our scuba observations and a search of the literature ( table 1 ) .
the sed includes previous ( sub)-millimeter observations by ward - thompson et al .
( 1994 ) as well as iso far - infrared observations at 170 and 200 ( ward - thompson et al .
the inclusion of the far - infrared fluxes is extremely important for bracketing the peak of the sed .
lynds 1498 was mapped with bears , a 25-beam double - sideband sis receiver array that operates at 3 mm ( sunada et al . 2000 ) , on april 26 , 2004 at the nobeyama radio observatory 45 m telescope .
the average beamsize within the array is 178 at 93.17 ghz ( tatematsu et al .
2004 ) ; however , each beam in the array is separated by slightly more than @xmath70 ( 411 ) .
we observed the source using a dithered @xmath71 position - switched grid on the sky spaced by 2055 .
the array was rotated by @xmath72 after each dithered map to change relative beam positions with respect to the source .
a digital autocorrelator backend with @xmath73 khz resolution ( @xmath74 km / s ) was used .
bears was tuned to the frequency of the strongest hyperfine component ( @xmath75 ghz @xmath76 , see appendix a ) .
the resulting n@xmath1h@xmath2 map has 100 spectra separated at nearly full beam resolution and represents the highest resolution n@xmath1h@xmath2 spectrum obtained towards l1498 .
bears tuning was consistent across the array with an average @xmath77 of 220 k. the intensity was calibrated for the receiver sideband ratio by observing the bright high - mass star - forming core , g40.50 + 2.54 , with every beam in the bears array and the nobeyama facility single - sideband s100 receiver .
spectra from the s100 receiver were calibrated on the @xmath78 scale using the standard chopper - wheel method ( penzias & burrus 1973 ) .
the calibration source was observed twice with the array rotated by @xmath79 to minimize systematic calibration errors .
the resulting calibration is good to approximately @xmath80% .
the spectra were converted to the scale by assuming the bears sidelobes are negligible and using @xmath81 ( tatematsu et al .
2004 ) .
pointing observations were performed every hour using the sio @xmath82 , @xmath83 maser transition towards nml tau .
the largest pointing shifts were @xmath84 with average corrections of @xmath85 and @xmath86 in azimuth and elevation respectively .
the weather was clear with winds typically below @xmath87 m / s during l1498 observations .
conditions were good for n@xmath1h@xmath2 observations and the absolute pointing errors are small compared to the bears beamsize .
the data were baselined , summed , calibrated , and interpolated onto a regular spatial grid using the _ _ newstar _ _ software package of the nobeyama radio observatory .
the reduced maps were written out to the group fits format ( greisen & harten 1981 ) and read into _ aips++_. fitting of the hyperfine components were performed with _ glish _ scripts in _
aips++_. the group fits data cube was also written to a series of ascii files for calculation of baseline rms and the integrated intensity .
the n@xmath1h@xmath2 @xmath88 transition was observed in position - switched mode towards the centroid continuum position of lynds 1498 with the caltech submillimeter observatory on the night of july 24 , 2004 .
the 230 ghz double sideband sis receiver ( kooi 1992 ) with a 50 mhz aos backend were used .
the average @xmath77 was approximately @xmath89 k. the spectral resolution was 147 khz ( @xmath90 km / s ) after smoothing to the actual resolution of the spectrometer ( 3 channels ) .
the receiver was tuned to the jpl line catalog frequency , @xmath91 ghz , for n@xmath1h@xmath2 @xmath88 ; this frequency is within 160 khz of a blend containing the strongest hyperfine component ( @xmath92 ghz , @xmath93 ; luca dore , priv .
comm . ) . pointing and main beam calibration were performed on venus .
the main beam efficiency was determined to be @xmath94 at 279 ghz for a @xmath95 beam .
the pointing was consistent with a largest pointing shift of less than @xmath96 in both azimuth and zenith angle during the observations .
the spectra were smoothed , baselined , and summed using the _ class _ software package ( buisson et al .
l1498 was observed for a total of @xmath97 minutes on - source .
the final spectrum has a baseline rms of @xmath98 mk .
the c@xmath5s @xmath99 transition at 23.1229820 ghzsand its isotopologues are from the cologne database for molecular spectroscopy ( mller et al .
2001 ) : http://www.ph1.uni-koeln.de/vorhersagen/ ] was mapped using the 100 m robert c. byrd green bank telescope on the nights of march 14 and 15 , 2005 .
the spectrometer was setup to observe 4 ifs with 2 polarizations at 6.125 khz resolution ( 0.079 km / s ) using a single beam of the k - band receiver .
the four ifs were centered on the c@xmath5s c@xmath5@xmath100s , @xmath101cccs , and c@xmath101ccs @xmath99 transitions .
the observations were frequency switched by @xmath102 mhz at a frequency of 4 hz .
initially , a pointed map was made using a @xmath103 grid , with 60s integration time per pointing , spaced at 33 ( full beam spacing ) centered on the scuba dust continuum centroid position .
three smaller @xmath104 maps with 165 spacing were centered at the offsets ( 0,0 ) , ( @xmath105,@xmath106 ) , and ( @xmath107,@xmath108 ) with respect to the dust continuum centroid . the raw frequency - switched spectra were folded and calibrated on the @xmath109 scale using the _ dish _
software package within _
the data were then written to ascii files where opacity corrections were applied ( @xmath110 ) and rr and ll polarizations were summed with @xmath111 weighting .
the opacity at @xmath112 ghz was determined using the weather model of r. maddalena ( private communication ) which averages the precipitable h@xmath1o measurements from three nearby towns ( elkins , hot springs , and lewisburg , west virginia ) .
the average zenith opacity was @xmath113 on march 14 and @xmath114 on march 15 during the observations .
final conversion to the @xmath115 scale was achieved using position - switched scans of the quasars 3c48 and 3c286 .
unfortunately , the aperture efficiency was not properly determined at @xmath112 ghz on march 14 ; therefore , we have linearly interpolated from measurements made at @xmath30 ghz ( langston et al .
2004 ) and @xmath116 ghz ( march 18 , 2005 ) to find @xmath117 .
the gbt has an active surface which mitigates variations in the aperture efficiency with elevation and temperature .
we have applied a @xmath118 systematic uncertainty to account for these effects .
the main beam efficiency is then @xmath119 .
the @xmath115 temperature scale may be compared with observations on the scale if the sidelobe pickup is negligible ( see chapter 8.2.5 of rohlfs & wilson 2000 ) ; this is a reasonable assumption for the k - band receiver on the gbt .
pointing and focusing were performed once per hour .
the sky was clear and the wind was less than @xmath120 m / s on both march 14 and march 15 during the observations . since the observations were performed at nighttime , the temperature variation was small ( @xmath121c ) .
the pointing and focus were very stable with a largest pointing shift of @xmath122 .
the dust continuum observations are modeled using the one - dimensional radiative transfer code _ csdust3 _ ( egan , leung , & spagna 1988 ) that self - consistently calculates the temperature distribution for a given density distribution , dust opacity , and isrf .
_ csdust3 _ simulates anisotropic scattering which may affect the temperature distribution in the outer regions of the core , a feature that is not present in the radiative transfer code of ivezi , nenkova , & elitzur ( 1999 ) .
once the temperature distribution has been calculated , the intensity profiles at 850 and 450 are created by convolving the model intensity profiles with the measured beamshape and simulating chopping .
the model sed is also calculated and compared to the observed fluxes .
simultaneous modeling of the intensity profiles and sed result in nearly orthogonal constraints on the shape of the density profile and the total mass for a given opacity choice ( see shirley et al .
the modeling procedure for ppcs is described in more detail in evans et al .
( 2001 ) .
we shall use a family of be spheres ( ebert 1955 , bonnor 1956 ) to attempt to fit the l1498 observations since be spheres were used by evans et al .
( 2001 ) to successfully fit submm continuum observations of other ppcs .
be sphere are solutions to the pressure bounded equations of hydrostatic equilibrium that ignore the effects of magnetic fields and turbulence .
the be density profile is flat in the inner regions and asymptotically approaches @xmath123 in the outer regions .
the size of the flat density plateau decreases as the central density of the be sphere increases ( density fwhm @xmath124 , chandrasekar & wares 1949 ) .
the temperature structures are characterized by decreasing dust temperature from outside to inside due to extinction of heating radiation from the isrf .
the temperature gradient from outside to inside increases for more centrally condensed be spheres .
isothermal be spheres may be parametrized by a single variable , the central density ( @xmath125 ) for a fixed temperature ( 10 k ) .
evans et al .
( 2001 ) showed that non - isothermal correction to the shape of the be profile was negligible for profiles used to describe ppcs .
isothermal be spheres are calculated with central densities ranging from @xmath126 to @xmath127 .
since l1498 is embedded within a larger molecular cloud with @xmath128 mag ( cambrsy 1999 ) , the power - law portion of the be profile is not continued below @xmath129 .
since ppcs lack an internal source , the heating is from the isrf .
the spectrum of the isrf is derived from _ cobe _
data ( black 1994 ) with ultraviolet wavelengths constructed using an empirical description of the radiation field that reproduces the isrf of draine ( 1978 ; see van dishoeck 1988 ) .
this spectrum is the same as used in the models of evans et al .
the shape of the isrf spectrum may be modified by two effects : the intensity of the isrf ( uv to far - infrared ) varies at different locations throughout the galaxy ; and the short wavelength intensity of the isrf ( uv to near - infrared ) may be modified by dust extinction ( see figure 3 ) .
the scaling of the intensity is parametrized by while the amount of dust extinction at the core s outer radius is parametrized by .
we use dust opacities appropriate for the general ism ( graphite and silicate based grains , draine & lee 1984 ) to modify the isrf spectrum .
the combination of scaling the isrf and extinction at the outer core radius is a more realistic description of the incident intensity than has been used in previous models of ppcs .
the reprocessing of the isrf is strongly affected by the dust opacity within the core .
we model the emission using eight different opacities models ( see table 2 and figure 3 ) .
the opacities of ossenkopf & henning ( 1994 ) are derived from models of grains that have coagulated for @xmath130 years and that have acquired varied depths of ice via gas adsorption : oh2 no ice mantles ( column 2 table 5 of ossenkopf & henning 1994 ) , oh5 thin ice mantles ( column 5 , table 5 ) , and oh8 thick ice mantles ( column 8 , table 5 ) .
the opacities of mathis , mezger , & panagia ( 1983 ) are derived from an empirical fit to dust properties .
we also test the opacities of weingartner & draine ( 2001 ) which describe silicate and graphite grains , including a population of small grains and pahs , appropriate for the ism .
the opacities of pollack ( 1994 ) describe an alternative grain composition based on silicate grains ( olivine and orthopyroxene ) , iron compounds ( troilite and metallic iron ) , and various organic c compounds .
the mass opacity at 850 , @xmath131 , varies by more than one order of magnitude between the opacity models ( table 2 ) . a constant opacity with radius ( @xmath132 $ ]
, @xmath133 is the outer core radius ) is assumed throughout the core .
a total of @xmath134 radiative transfer models were run that vary @xmath125 , @xmath133 , , , and @xmath135 . for each model ,
the reduced @xmath136 , given by @xmath137 was calculated for the 850 and 450 normalized intensity profiles and for the sed .
a @xmath138 is also calculated for the flux in a @xmath139 aperture at 850 .
the best - fitted models are determined by finding the intersection of the sets of @xmath136 for each quantity within the typical range @xmath140 $ ] .
since the errorbars of the intensity profiles include azimuthal variations as well as intensity uncertainties , and since the errorbars of the sed include an estimate of systematic uncertainty , it is possible for @xmath141 .
the first grid of models varies @xmath125 and @xmath133 for @xmath142 , @xmath142 , and oh5 opacities ( see top row of figure 4 ) .
the best - fitted model to the azimuthally averaged , normalized intensity profiles is @xmath143 .
if we re - calculate the @xmath144 using the sector - averaged , normalized intensity profiles , then a wider range of central densities fit , @xmath145 ( see dashed profiles in figure 5d ) .
the best fitted central density does not strongly depend on the size of the outer radius ; radii from @xmath146 au to @xmath147 au are well fit .
since emission is clearly detected @xmath148 from the center of the scuba maps , @xmath133 must be greater than @xmath149 au .
furthermore , the iso 170 image ( ward - thompson et al .
2002 ) detects dust emission at radii greater than @xmath150 ( @xmath151 au ) from the center ; therefore , the best - fitted model @xmath133 are consistent with submillimeter and far - infrared observations . since the scuba continuum centroid and 1.2 mm
continuum peaks are different by @xmath152 , we have also compared the dust model intensity profiles to azimuthally averaged intensity profiles centered at the 1.2 mm peak .
the best - fitted model are @xmath153 bonnor - ebert spheres .
this agreement with the best - fitted models toward the scuba centroid position is a result of l1498 s very flat submillimeter intensity profile that does not vary strongly with different central positions within @xmath154 of the scuba continuum centroid ( see figure 1 ) .
the second grid of models varies , and for the @xmath143 be sphere , @xmath155 , @xmath156 , and @xmath147 au , and the different opacity models .
the and are varied on non - linear grids with values between @xmath157 $ ] and @xmath158 $ ] respectively . the middle and bottom rows of figure 4 display @xmath159 and @xmath160 for oh5 dust opacities . changing and has a negligible effect on the best fitted normalized intensity profiles ; the @xmath143 be sphere
is the best fit for all opacity models .
previous dust continuum models relied only upon submillimeter fluxes to constrain the heating properties of the isrf .
however , @xmath160 alone do no strongly constrain and ( bottom row , figure 5 ) ; a wide range of values provide good fits . when we include the complete sed ,
the is constrained to be @xmath161 for the best - fitted models
( middle row , figure 5 ) .
the only dust opacity models that provide satisfactory fits to @xmath162 , @xmath163 , and the sed are the ossenkopf & henning opacities .
none of the model @xmath159 for wd3 , wd4 , pollack1 , or draine - lee opacities are below @xmath87 .
these models underestimate the fluxes observed at far - infrared through millimeter wavelengths for the @xmath164 be sphere .
this is not surprising since the mass opacity at 850 for these models is @xmath96 to @xmath165 times smaller than the mass opacity of oh8 dust .
furthermore , ossenkopf & henning opacities for grains with ice mantles ( oh5 and oh8 ) provide better fits to @xmath163 and the sed than grains without ice mantles ( oh2 ) .
the oh2 opacities tend to overestimate @xmath163 while underestimating the far - infrared fluxes .
it is encouraging that the dust opacities for grains with ice mantles provide the best fits since these models are qualitatively consistent with a physical model of l1498 that accounts for the observed gas depletion of species such as co , h@xmath1co , cs , and ccs .
the best - fitted models from all of the model grids are listed in table 3 and shown in figure 5 .
the models fall into two categories : oh5 opacities with @xmath166 $ ] and @xmath167 $ ] ; and oh8 and oh5 opacities with @xmath168 $ ] and @xmath169 $ ] .
the fluxes and intensity profiles are well fit at all wavelengths except @xmath170 mm for all of the models all of the best - fitted models have an outer radius of @xmath147 au ( 0.29 pc ) .
since a large outer radius is preferred , it seems unlikely that the high models are physical .
an optical extinction map of the taurus molecular cloud limits the large - scale extinction around l1498 to @xmath171 mag ( cambrsy 1999 ) .
indeed , for @xmath172 , the column density of material outside @xmath133 would be @xmath173 @xmath174 ; this value is larger than the column density derived from a @xmath164 be sphere with @xmath175 au .
we prefer the low models as the best , most physical fit .
all of the best - fitted models are characterized by dust temperatures between @xmath97 and @xmath176 k ( figure 5a ) .
the low oh5 and oh8 models have temperature gradients of @xmath177 } < 0.5 $ ] k. these are nearly isothermal temperature profiles compared to previous models of ppcs ( @xmath178 } > 1 $ ] k , e.g. evans et al .
2001 ) . the low density of l1498 combined with the inclusion of dust attenuation at the outer radius results in the nearly isothermal temperature profiles
the bolometric luminosity was calculated by integrating under the spectral energy distribution of l1498 ( table 1 ) and is @xmath179 , low for a ppc , but not the lowest reported luminosity ( see table 4 ) .
the bolometric temperature , the temperature of a blackbody with the same mean frequency as the observed sed ( chen et al .
1995 ) , is @xmath180 k. this is one of the lowest bolometric temperatures reported for a ppc ( see table 4 ) .
we calculate the mass from the flux at 850 in a 120aperture and assuming a constant density , isothermal dust temperature , and optically thin emission using , @xmath181\ ] ] based on 1d dust radiative transfer models , a dust temperature of @xmath182 k is used for l1498 . the resulting dust mass , @xmath183 for an average of oh5 and oh8 opacities ( @xmath184 @xmath185 g@xmath186 ) .
the errors in @xmath187 are propagated from @xmath188 , @xmath189 , and @xmath190 .
l1498 s dust mass is consistent with and slightly lower than previously observed pre - protostellar cores ( @xmath191 , table 4 ) .
we also calculate the aperture - averaged column density from the 850 fluxes using @xmath192 \ ; .\ ] ] the column density at the scuba continuum centroid is @xmath193 in a @xmath194 aperture , where errors are propagated in @xmath188 , @xmath189 , and @xmath195 .
this aperture size matches the bears beamsize allowing a direct comparison between dust and n@xmath1h@xmath2 column density ( 4.1.2 ) .
the column density corresponds to a visual extinction of @xmath196 mag ( whittet 2003 ) .
this extinction is comparable to the peak extinction observed using near - infrared extinction mapping towards the low - density ppc , coalsack g2 ( @xmath197 mag , lada et al .
modeling of the coalsack g2 extinction profile indicates a peak density of @xmath198 , corroborating the low density derived for l1498 from our radiative transfer models .
dust opacities tend to follow a single power - law with frequency at submillimeter wavelengths ( @xmath199 , @xmath200 $ ] ) .
the observed dust emissivity index , @xmath201 , for a core observed at 850 and 450 is given by , @xmath202 - \left\ { 3 + \frac{\log\left [ \exp(h\nu_{850}/kt_d ) - 1 \over \exp(h\nu_{450}/kt_d ) - 1 \right ] } { \log\left({850 \over 450}\right ) } \right\ } \ ; , \ ] ] where the term in square brackets is the spectral index between 450 and 850 , @xmath203 , and the term in curly braces is the correction for failure of the rayleigh - jeans approximation . assuming a dust temperature of @xmath182 k , we find @xmath204 and @xmath205 . the spectral index is higher than all of the ppcs observed by shirley et al .
( 2000 ) except for the two cores associated with l1689a .
the @xmath201 is larger than the theoretical @xmath201 for dust opacities used in modeling but is consistent , within errors , with all of the dust opacity models except oh2 and wd4 ( see table 2 ) .
while oh5 and oh8 opacities give good fits , within errorbars , to the sed , they consistently underestimate the flux at 450 while matching the 850 flux .
submillimeter measurements of @xmath206 and @xmath201 do not discriminate well between opacity models ; radiative transfer modeling of the complete sed is a better discriminator .
the properties of l1498 may be placed in context by comparing to other ppcs that have been observed with iso and scuba . table 4 lists observed and derived properties of 10 nearby ( @xmath207 pc ) sources . including the 170 isophot flux
gives at least one flux point on the wien side of the sed , resulting in a more accurate calculation of .
each source sed contains at least 3 flux points ( 170 , 200 , and 850 ) .
we also analyzed the scuba 850 images for each source . all of the scuba observations are published except for three cores , l183 , l1709a , and l63 which were obtained from the cadc scuba archive .
the reduction procedure for these cores is identical to that described in 2 .
we have calculated the size , aspect ratio , bolometric luminosity , standard evolutionary indicators ( and ) , and the dust - determined mass from the 850 flux .
this is a small sample but probably representative of nearby ppcs .
the cadc scuba archive contains many more observations of ppcs ; however , far - infrared observations are needed to calculate .
l1498 is the largest nearby ppc in the sample by 50% and nearly twice as large as the average core size .
l1498 also has one of the lowest masses ( @xmath208 ) of the sample .
this is consistent with the interpretation that l1498 is a low density core with a uniquely flat intensity profile .
in contrast , the average nearby ppc is a core with @xmath209 , @xmath210 , and elongated ( @xmath211 ) with a mean radius of 6000 au .
ppcs are clearly separated from class 0 sources on a vs. diagram ( figure 6 ) .
all of the ppcs have low @xmath212 k and @xmath213 whereas the observed class 0 sources all have @xmath214 k. and for ppcs do not correlate .
this is not surprising since these evolutionary indicators are strongly influenced by the strength of the isrf ; this is an environmental effect not directly related to the evolutionary state of the core .
a better scheme for characterizing the evolutionary state of ppcs would be the comparison of the dynamical maturity determined from radiative transfer models ( @xmath125 ) and the chemical state of molecules that are abundant at different times .
for instance , comparisons of [ ccs]/[n@xmath1h@xmath2 ] , comparisons of [ n@xmath1d@xmath2]/[n@xmath1h@xmath2 ] , and the depletion fraction of , @xmath216 , of c@xmath18o and h@xmath1co are promising candidates .
the evolutionary state of l1498 is discussed in 4.3 .
the seven hyperfine lines of n@xmath1h@xmath2 @xmath83 were well detected over an extent of @xmath217 in the map ( figure 1f ) .
the peak @xmath78 in the map is coincident with the continuum centroid ( figure 7a ) .
we can estimate the effective density needed to excite an easily observable 1 k n@xmath1h@xmath2 line ( see table 1 of evans 1999 ) .
the critical density of the @xmath83 is @xmath218 ( collision rates determined from turner 1995 ) . by analogy with hco@xmath2 , the resulting effective density ( corrected for optically thin hyperfine structure )
is @xmath219 , consistent with the derived central density .
the integrated intensity is calculated by integrating over the seven hyperfine components .
the @xmath220 at the continuum centroid is also the peak in the n@xmath1h@xmath2 map , @xmath221 k km / s .
the fwhm integrated intensity contour agrees very well with the fwhm continuum contour in size and position angle .
the n@xmath1h@xmath2 emission appears to trace the general shape of the continuum emission , but does not peak as sharply along the north - eastern ridge as is seen in the 850 and , especially , in the 1200 images .
the hyperfine structure was fitted with seven gaussian functions , constrained by the hyperfine transition velocities ( appendix a ) , using the profile fitter task in _ aips++_. the average linewidth was calculated to be @xmath222 km / s over the map using the isolated hyperfine component ( @xmath223 ) .
this linewidth is slightly smaller than the linewidth of @xmath224 km / s found by caselli et al .
( 2002 ) using the fcrao .
the non - thermal contribution to the linewidth is @xmath225 km / s .
a plot of the linewidth versus radius ( figure 7b ) indicates that the linewidth does not vary significantly out to @xmath149 au ( @xmath226 ) .
therefore , the average gas motions in l1498 have low turbulence and are similar throughout the core .
this result agrees with the @xmath227-@xmath228 relationship towards other ppcs in taurus observed with bears ( tatematsu et al .
2004 ) .
we calculate the virial mass using the doppler linewidth and assuming an isothermal dust temperature , @xmath229 \ ; .\ ] ] for a more accurate calculation , we split the radial profile into two sections .
the first corresponds to @xmath230 ( the radius where @xmath231 ) which is approximated by a constant density . for a @xmath164 be sphere , @xmath232 is @xmath233 au . within the second section , @xmath234 , the density can be approximated by an inverse power law with @xmath123 ( @xmath235 au , 2.1.2 ) .
the term in square brackets converts the observed n@xmath1h@xmath2 linewidth to a 3d velocity dispersion ( shirley et al .
2002 ) . using the @xmath227 determined above
, we find @xmath236 with errors propagated in @xmath237 , @xmath238 , and @xmath227 .
this mass agrees with the mass determined from dust continuum emission ( 4.1.1 ) indicating that l1498 is likely a bound core and that the best - fitted ossenkopf & henning dust opacities are appropriate .
the n@xmath1h@xmath2 @xmath83 transition is observed to have moderate optical depths towards low - mass cores ( e.g. tatematsu et al .
2004 , crapsi et al .
2005 ) ; therefore , we must calculate the optical depth at each position in the map to derive the column density .
for a position - switched observations , theoretically we observe the quantity @xmath239 $ ] , where @xmath240 $ ] .
this @xmath241 is related to the actual observed quantity , @xmath78 , through source - beam coupling and through the efficiency of the antenna ( including spillover and atmospheric attenuation , see kutner & ulich 1981 ) .
since the hyperfine lines lie close together in frequency ( @xmath242 mhz ) , we assume that the proportionalities for each line are constant ( keto et al . 2004 ) .
therefore , we find the optical depth in the strongest hyperfine component ( @xmath243 , @xmath244 ) by taking ratios of the observed antenna temperatures of the @xmath245 hyperfine component to the strongest hyperfine component @xmath246 where @xmath247 is the ratio of relative strengths of the hyperfine lines to the strongest line .
@xmath247 is calculated in appendix a. the average optical depth in the strongest hyperfine line , @xmath248 , was calculated using four of the six observed line ratios and equation 9 .
two of the lines ( @xmath249 and @xmath250 ) consistently gave lower @xmath248 and higher @xmath248 by factors of 0.4 and 2.0 respectively .
this may be due to anomalous hyperfine excitation , an effect that has been observed towards other ppcs ( caselli et al .
1995 , turner 2001 ) .
an optical depth was determined for 47 positions with suitable signal - to - noise .
the optical depth is not affected by line blending since the linewidth is @xmath251 km / s ( see appendix a ) . since , each hyperfine line has a different optical depth , then correcting the column density using only @xmath243 would overestimate the column density .
we define the average tau by averaging the theoretical hyperfine ratios for the seven observed transitions @xmath252 .
the average optical depth in the map is @xmath253 , indicating that the n@xmath1h@xmath2 @xmath83 transition is moderately optically thick over most of the map . as an example
, the scuba continuum centroid position has an optical depth of @xmath254 , resulting in a correction for optical depth to the column density of @xmath255 .
the total column density for n@xmath1h@xmath2 is ( see goldsmith & langer 1999 ) ( e.g. caselli et al .
] @xmath256 where @xmath257 is the rotational partition function and @xmath258 d ( green et al . 1974 ) is the dipole moment .
we assume an excitation temperature of @xmath259 k. errors in @xmath260 are propagated for @xmath261 , @xmath262 , and @xmath220 .
the peak column density is @xmath263 @xmath174 with an average column density of @xmath264 @xmath174 .
the peak column density agrees remarkably well with the calculation of caselli et al .
( 2002a ) for lower resolution fcrao 15 m n@xmath1h@xmath2 observations towards l1498 ( @xmath265 @xmath174 ) and the intermediate resolution iram 30 m observations of crapsi et al al .
( 2005 ; @xmath266 @xmath174 ) .
we calculate the abundance , @xmath267 , by comparing the n@xmath1h@xmath2 column density to the beam averaged column density determined from the 850 dust continuum map ( equation 6 ) .
the average column density in the map is @xmath268 relative to h@xmath1 .
this column density agrees very well with the early - time abundances of the coupled dynamical - chemical models of lee et al .
( 2004 ) for quasi - statically evolving ppcs ( 4.3 ) .
the abundance show no significant variations with radius out to @xmath149 au , although the errors on the abundance are sizable ( figure 7d ) .
radiative transfer models of the n@xmath1h@xmath2 emission should assume a constant abundance of @xmath269 .
the n@xmath1h@xmath2 @xmath270 transition was not detected towards the continuum centroid position with a @xmath271 baseline rms of @xmath272 mk ( figure 7c ) ; however , crapsi et al .
( 2005 ) report a detection with the iram 30 m toward the 1.2 mm continuum peak position ( @xmath78 @xmath273 mk ; see inset in figure 7c ) .
this is puzzling since the @xmath274 bears map shows no enhancement toward that position .
it is possible that the 1.2 mm continuum peak represents the true density peak of l1498 . a high - resolution , oversampled map of the n@xmath1h@xmath2 @xmath270 emission is needed to resolve this discrepancy . the exact location of the density peak will be important for molecular line radiative transfer modeling , but does not affect the conclusions from dust continuum modeling ( 3.2 ) .
the l1498 n@xmath1h@xmath2 observations will be modeled , along with other ppcs , in a future paper .
the c@xmath5s @xmath99 line was strongly detected toward l1498 ( figure 8) and the emission follows the same double - peaked pattern as previous observations of cs and ccs ( figure 1 of willacy , langer , & velusamy 1998 ) .
the line intensity is strongest in the southeastern peak ( @xmath275 k ) at j@xmath23 @xmath24 @xmath25 @xmath26 @xmath276 , @xmath28 @xmath29 @xmath30 @xmath277 .
since the c@xmath5s emission has an intensity saddle at the scuba dust continuum centroid , it is another excellent example of strong chemical differentiation in l1498 .
the southeastern c@xmath5s peak does not exactly correspond to the willacy , langer , & velusamy cs or ccs peak , but instead lies between them and closer to the cs peak .
combination of the gbt observations with higher resolution observations ( e.g. , vla ) would improve the peak position and may confirm slight chemical differentiation between all of the observed sulfur - carbon chain molecules .
the c@xmath5s @xmath99 linewidth was measured using the gaussian fitting routine is _
aips++ _ toward 42 positions in the map with detections greater than @xmath278 .
the average linewidth is @xmath279 km / s , consistent with the n@xmath1h@xmath2 linewidth .
no significant variation of @xmath227 is also seen for c@xmath5swith radius ( figure 8b ) .
the column density of c@xmath5s was estimated using the same method as for n@xmath1h@xmath2 ( equation 10 ) .
we estimate the optical depth of the c@xmath5s @xmath99 with 10 minute observations at the southeastern peak position of three isotopologues .
the detection of c@xmath5@xmath100s @xmath99 ( figure 8c ; @xmath280c@xmath5s@xmath281c@xmath5@xmath100s@xmath282 ) and lack of a significant detection ( @xmath283 ) of @xmath101cccs or c@xmath101ccs @xmath99 , indicates , for the standard ism isotope ratios of [ @xmath284s]/[@xmath100s ] @xmath285 and [ @xmath286c]/[@xmath101c ] @xmath287 ( wilson & rood 1994 ) , that the c@xmath5s emission is consistent with being optically thin .
the peak column density is @xmath288 @xmath174 and is less than a factor of 2 less than the peak ccs column density determined by wolkovitch et al .
( 1997 ) ( @xmath289 @xmath174 ) .
the abundance of c@xmath5s was calculated by comparing the c@xmath5s column density to the column density derived from the 850 map in 33 apertures .
since the c@xmath5s emission is very asymmetrical , we plot the abundance versus radius in the major - axis sector ( 2.1.2 ) in figure 8d .
negative radii refer to positions se of the dust continuum centroid .
the depletion of c@xmath5s is clearly visible with enhanced abundances by factors of @xmath290 and @xmath20 at the two c@xmath5s intensity peaks .
the abundance is largest in the se peak ( @xmath291 ) .
the abundance peaks are symmetrically located at a radius of 9500 au from the dust continuum centroid position .
this radius set a limit for the depletion radius of c@xmath5s , corresponding to a density of @xmath292 for the best - fitted be spheres . since the abundance pattern of c@xmath5s is similar to smaller sulfur - carbon chain molecules ( e.g. , cs and c@xmath1s ) , the southeastern c@xmath5s peak of l1498 would be an excellent site to search for larger sulfur - carbon chain molecules such as c@xmath293s and c@xmath294s ( gordon et al .
2001 ) .
radiative transfer modeling of submillimeter images combined with far - infrared and millimeter photometry indicate that l1498 is a low density , nearly isothermal ppc , with a low strength of the isrf and dust opacities appropriate for coagulated , icy grains .
the central density of @xmath295 is the among lowest central densities reported for ppcs , similar to coalsack g2 ( lada et al .
this density range agrees with the densities determined by langer & willacy ( 2001 ) from analysis of iso observations of l1498 ( @xmath296 with lower densities more appropriate for larger core radii ) . however , our modeling results differ substantially from the analytical model of tafalla et al .
( 2004 ) which fits the radial profile of l1498 at 1.2 mm with a function of the form @xmath297 $ ] and assumes an isothermal dust temperature profile of @xmath298 k. tafalla et al .
find a central density of @xmath299 , three times larger than the maximal be central density in our models .
there are a few discrepancies that may account for this difference .
tafalla assume a dust opacity of @xmath300 @xmath185g@xmath186 at 1.2 mm that is a factor of @xmath301 smaller than oh5 opacities ( @xmath302 @xmath185g@xmath186 ) .
since the column density is inversely proportional to the opacity , the tafalla et al .
opacity would result in a factor of 2 larger central density .
this does not account for the full difference between the results .
the tafalla et al .
centroid position is located at the peak of the non - axisymmetry which is much more pronounced at 1.2 mm than at submillimeter wavelengths .
tafalla et al .
calculate a column density of @xmath96 - @xmath303 @xmath174 , a factor of 3 higher than our column density determined at 850 .
for all of the best - fitted radiative transfer models , the 1.2 mm flux is always underestimated while the other six fluxes ( 170 , 200 , 450 , 850 , 1100 , and 1300 ) are fit within the errorbars .
the 1.2 mm flux point appears to be anomalously high .
it is possible that calibration uncertainties may account for the remaining difference .
the low density implies that l1498 is dynamically young .
indeed , l1498 may be in a stable hydrostatic state .
one measure of stability of be spheres is the density gradient between inside and outside of the core . if the density contrast exceeds @xmath304 , then the be sphere is in an unstable equilibrium ( see foster & chevalier 1993 ) .
a lower density limit of @xmath129 was assumed for the core in the radiative transfer models .
this lower limit was chosen to be larger than the average density in a molecular cloud ( @xmath305 ) since l1498 is situated within an extended condensation of @xmath306 to @xmath96 in the taurus - auriga molecular cloud ( see cambrsy 1999 ) .
the average solution ( @xmath164 ) is stable , but this results depends sensitively on the density at the outer radius . for the maximal central density ( @xmath307 ) , the density contrast is @xmath308 and the core is unstable .
unfortunately , we can not strongly constrain the density contrast since the submillimeter observations are insensitive to structure outside the chop distance ( 120 or 16,800 au ) . a more quantitative way to test the stability is to directly compare the jeans mass with the mass of the be sphere . the jeans
mass for @xmath309 k and @xmath310 is @xmath311 ( spitzer 1978 ) , while the @xmath164 be sphere has a mass of @xmath312 ( bonnor 1956 ) . the be mass is within a factor of 2 of the jeans mass , but @xmath313 .
if we use the @xmath307 be sphere , then the mass is comparable to the jeans mass @xmath314 .
this mass comparison is consistent with l1498 being a core that is marginally stable to gravitational collapse .
while l1498 may be dynamically unevolved , it appears to be chemically evolved .
species such as h@xmath1co , cs , ccs , and c@xmath5s ( e.g. , willacy et al . 1998 , tafalla et al . 2004 , young et al .
2004 , 4.3 ) are depleted towards the center of the core .
we can estimate the depletion timescale from the rate equation given in rawlings et al .
( 1992 ) . for a depletion fraction , @xmath315 , the adsorption timescale of a neutral molecular species , @xmath316 ,
is given by , @xmath317 where @xmath318 is the molecular mass in a.m.u .
and @xmath319 is the sticking coefficient ( also see charnley , rodgers , & ehrenfreund 2001 ) .
for example , h@xmath1co has been observed to be severely depleted towards l1498 over a large spatial extent ( wang 1994 ; young et al .
the adsorption timescale is @xmath320 years for @xmath321 , @xmath322 , and @xmath309 k in the absence of other formation or destruction mechanism . in reality
, many chemical effects may affect this timescale .
two of the first molecules to be depleted are ccs and c@xmath5s .
comparisons with detailed coupled dynamical - chemical models indicate that the ccs becomes depleted on timescales of a few @xmath130 years ( e.g. , li et al .
2002 , aikawa et al .
2003 , lee et al .
2004 ) .
a chemical age of a few @xmath130 years is consistent with the observed abundance of n@xmath1h@xmath2 ( aikawa et al .
2003 , lee et al .
n@xmath1h@xmath2 is considered a `` late time '' species since it is primarily destroyed in the gas phase by co and its abundance increases after co is significantly adsorbed ( although dissociative recombination may also be important at low densities , see geppert et al .
detailed chemical models indicate that this abundance increase occurs after @xmath323 years ( e.g. aikawa et al .
clearly , l1498 must be have been in a stable or very slowly evolving configuration for at least @xmath130 years to display the extreme molecular depletion and modest n@xmath1h@xmath2 abundance that is observed .
the dust opacities that best - fitted the l1498 sed were ossenkopf & henning opacities ( 1994 ) for grains that had coagulated for @xmath130 years and accreted ice mantles ( oh5 & oh8 ) .
these opacities have also successfully fit the seds of low - mass embedded protostars ( e.g. , shirley et al .
2002 ) and high - mass embedded protostars ( e.g. , mueller et al . 2002 ) .
the oh opacities are qualitatively consistent with the observed molecular depletion and the chemical timescale for l1498 ; however , the coagulation simulation of oh5 and oh8 assumes a density of n@xmath324 sustained for @xmath130 years ( see ossenkopf 1993 ) .
this density is substantially higher than the density derived from dust continuum models ; therefore , the oh5 and oh8 opacities may not be based on an accurate representation of fluffy aggregation of grains in low density ppcs .
how do you create a dynamically young yet chemically evolved core ? if l1498 is in an unstable equilibrium , it can not have been in that state for a long period of time since the free - fall time for a @xmath8 be sphere , @xmath325 ( spitzer 1978 ) , is approximately @xmath326 years . clearly , l1498 has not yet collapsed and displays chemical differentiation that requires a similar timescale .
l1498 must have been static or slowly collapsing for more than @xmath130 years . while magnetic fields may also play a significant role in cloud support and evolution for dynamically nascent ppcs ( e.g. mouschovias & spitzer 1976 , li et al .
2002 ) , in the case of l1498 , they are not necessary for support of the cloud ( since @xmath327 ) .
nevertheless , we can estimate the strength of the b - field required for equipartition of magnetic , gravitational , and kinetic energy .
the strength of this b - field is given by @xmath328 @xmath7 g ( lada et al .
this is a small magnetic field compared with zeeman measurements of the ccs @xmath329 transition which provide a rough estimate of the line - of - sight field strength of @xmath330 @xmath7 g ( levin et al .
if l1498 becomes massive enough to be unstable to gravitational collapse , then the magnetic field appears to be strong enough to support l1498 , in which case , the timescale for collapse is controlled by the ambi - polar diffusion timescale until the core becomes magnetically super - critical .
recent molecular line studies of ppcs in the taurus - auriga molecular cloud have identified cores ( e.g. l1521b and l1521e ) that appear to be dynamically young and chemically young ( hirota et al .
2002 ; tafalla & santiago 2004 ; hirota , maezawa , & yamamoto 2004 ) . the signature of chemical youth is strongly peaked sulfur - bearing carbon chain molecules ( e.g. , ccs and c@xmath5s ) , weak and n@xmath1h@xmath2 emission , and low molecular depletion of c@xmath18o .
confirmation of the chemical and dynamical youth of these cores requires a detailed comparison between submillimeter dust continuum emission and high - resolution observations of ccs or c@xmath5s .
these objects , along with l1498 , l1512 ( @xmath331 ) , and l1544 ( @xmath332 ) may represent an evolutionary sequence in taurus - auriga where chemical maturity is reached first and then dynamical evolution occurs ( see figure 9 ) .
an important caveat is that our 1d modeling has ignored the observed non - axisymmetry in l1498 .
the dust continuum emission peaks along the north - east ridge and not at the fwhm centroid .
this effect is most pronounced in the 1.2 mm but is also detected at 850 . other ppcs ( e.g. , l1544 and l63 ) also display clear non - axisymmetries .
the spherical be model is limited and unable to describe this structure .
a full 3d radiative transfer modeling is needed to probe the structure of the non - axisymmetry ( e.g. doty et al .
also , since the 90 iso image ( ward - thompson et al . 2002 ) shows a clear gradient across the image , there is evidence that the isrf asymmetrically heats l1498 ( also see langer & willacy 2001 ) .
3d models could explore triaxial , be - like model with a spatially varying isrf compared to fully non - axisymmetric density distributions ( see gonalves , galli , & walmsley 2004 ) .
encouragingly , a comparison between 3d models and 1d models of l1544 indicate that the 1d models are an acceptable fit to the average physical structure ( doty et al .
we have presented deep scuba observations , bears and cso n@xmath1h@xmath2 observations , and gbt c@xmath5s observations of the pre - protostellar core , lynds 1498 .
radiative transfer modeling of the submillimeter intensity profiles and the complete sed were performed .
our main conclusions are as follows .
l1498 is characterized by a low density bonnor - ebert sphere with central density of @xmath8 and outer radius greater than @xmath156 au .
the sed is best - fitted with ossenkopf & henning opacities for coagulated grains with thick and thin ice mantles ( oh8 and oh5 opacities ) .
a more realistic treatment of the isrf was used that includes variations in the strength of the incident specific intensity and extinction due to dust grains at the outer radius of the core .
l1498 is a potentially stable ( @xmath313 ) , magnetically sub - critical core that appears to be dynamically young .
these results do not change if the 1.2 mm continuum peak is used for the radial intensity profiles instead of the scuba continuum centroid .
observations of the depletion of species such as c@xmath5s and h@xmath1co indicate l1498 is chemically evolved ( t @xmath333 years ) .
the modest n@xmath1h@xmath2 abundance is also consistent with a chemically evolved source .
the best - fitted ossenkopf & henning opacities are consistent with this timescale , although the ossenkopf coagulation models assume a higher density than is observed toward l1498 .
the n@xmath1h@xmath2 and c@xmath5s linewidth do not vary significantly with radius out to 15,000 au .
the n@xmath1h@xmath2 abundance also does not vary significantly with radius , while the c@xmath5s abundance indicates significant depletion toward the center of l1498 .
comparisons of l1498 with nearby ppcs that have also been observed at far - infrared and submillimeter wavelengths indicate that l1498 is larger than average , less luminous than average , and less massive than average .
the standard evolutionary indicators used for low - mass protostars , and , do not correlate for ppcs .
all of the ppcs in this sample are characterized by @xmath212 k and @xmath213 .
the temperature and density structure derived from 1d models may now be used in more realistic models of molecular line emission .
future studies of ppcs require high resolution far - infrared observations coupled with 3d radiative transfer modeling of multiple wavelengths .
instruments such as hawc on sofia and the launch of herschel will provide the needed high resolution far - infrared observations .
since every nucleus in the n@xmath1h@xmath2 molecules has non - zero nuclear spin , n@xmath1h@xmath2 has a very rich hyperfine spectrum that can be exploited to probe the physical conditions in star - forming regions . in this appendix , we summarize the hyperfine splitting of the ground state rotational transition of n@xmath1h@xmath2 .
the hyperfine splitting ( in order of importance ) is due to : ( 1 ) an electric quadrupole interaction due to the outer nitrogen ; ( 2 ) an electric quadrupole interaction due to the inner nitrogen ; ( 3 ) a magnetic dipole interaction due to the outer nitrogen ; ( 4 ) a magnetic dipole interaction due to the inner nitrogen ; ( 5 ) a magnetic dipole interaction due to the hydrogen nucleus ; and ( 6 ) spin - spin interactions between the nuclei .
the final two effects are ignored in calculations of hyperfine splitting for astronomical sources since the observed linewidth is @xmath334 km / s ( @xmath335 khz ) and the hyperfine coupling constants are small ( see caselli , myers , & thaddeus 1995 ) .
there are 15 electric dipole allowed transitions for n@xmath1h@xmath2 @xmath83 in the @xmath336 basis ( @xmath337 , @xmath338 , @xmath339 , @xmath340 ) .
the 15 transitions are listed in table 5 ; however , since in practice the @xmath341 level is not split , degenerate transitions are collected together into the 7 observed transitions .
we adopt the standard notation from caselli , myers , & thaddeus for the transition labels ( 1995 ) .
it is convenient to calculate the ratio of relative strengths , @xmath247 , of the hyperfine levels for calculations of optical depth and column density ( 4.2 ) .
the relative strengths , @xmath342 , are determined using irreducible tensor methods ( gordy & cook 1984 , chapter 15 ) and are defined such that the sum of the relative strength of the electric dipole allowed transitions are equal to one ( see rudolph 1968 ) .
the ratio of relative strengths for the n@xmath1h@xmath2 @xmath83 transitions in the @xmath336 basis are given by @xmath343 where the 6-j symbols needed in equation a1 may be found in table 5 of edmonds ( 1974 ) .
the ratios must be summed over the degenerate transitions from the same @xmath344 level ( table 5 ) .
it is necessary to calculate the dependence of @xmath247 on the linewidth for closely spaced hyperfine lines ( see turner 2001 ) .
we have performed this calculation for the n@xmath1h@xmath2 @xmath83 transitions assuming each hyperfine component is gaussian , @xmath345 } { \sum_{j=1}^7 s_j \exp \left ( -4\ln 2 \frac{v_j^2}{\delta v^2 } \right ) } \;\ ; .\ ] ] the results are shown in figure 10 . for sources with @xmath346 km / s , the theoretical hyperfine ratios are sufficient .
this is the case for l1498 .
however , for n@xmath1h@xmath2 observations of sources with broader linewidths , the linewidth - corrected @xmath247 should be used .
we wish to thank the staffs of the nro , jcmt , cso , and gbt for their excellent assistance .
we are very grateful to mario tafalla and derek ward - thompson for providing us 1.2 mm and 170 images of l1498 , to luca dore for providing us his calculation of the n@xmath1h@xmath2 @xmath88 hyperfine frequencies , and to antonio crapsi for providing iram 30 m spectra of the n@xmath1h@xmath2 @xmath88 transition .
we especially thank ron maddalena for assistance with gbt calibration .
george moellenbrock , kumar golap , and joe mcmullan provided excellent assistance with glish scripting in aips++ .
finally , we thank the referee for a very detailed and helpful review .
nje acknowledges support from grant ast-030725 from the national science foundation and nasa grants nag5 - 10488 and nng04gg24 g .
adams , f. c. 1991 , , 382 , 544 aikawa , y. , ohashi , n. , & herbst , e. 2003 , , 593 , 906 alves , j. , lada , c. j. , & lada , e. a. 2001 , nature , 409 , 159 archibald , e.n . , et al .
2002 , mnras , 336 , 1 bianci , s. , gonalves , j. , albrecht , m. , caselli , p. , chini , r. , galli , d. , & walmsley , m. 2003 , a&a , 399 , l43 black , j. h. 1994 , asp conf
58 , the first symposium on the infrared cirrus and diffuse interstellar clouds , ed . r.m . cutri & w.b .
latter ( san fransico : asp),355 bonnor , w. b. 1956 , , 116 351 buisson , g. , desbats , l. , duvert , g. , forveille , t. , gras , r. , guilloteau , s. , lucas , r. , & valiron , p. 2002 , continuum and line analysis single - dish system manual cambrsy , l. 1999 , a&a , 345 , 965 caselli , p. , myers , p. c. , & thaddeus , p. 1995
, , 455 , l77 caselli , p. , benson , p.j . , myers , p.c . , & tafalla , m. 2002a , apj , 572 , 238 caselli , p. , walmsley , c. m. , zucconi , a. , tafalla , m. , dore , l. , & myers , p. c. 2002b , , 565 , 344 chen , h. myers , p. c. , ladd , e. f. , & wood , d. o. s. 1995 , , 445 , 377 chandrasekar , s. , & wares , g. w. 1949 , , 109 , 551 charnley , s. b. , rodgers , s. d. , & ehrenfreund , p. 2001
, a & a , 378 1024 crapsi , a. , caselli , p. , walmsley , c. m. , myers , p. c. , tafalla , m. , lee , c. w. , & bourke , t. l. 2005 , , 619 , 379 doty , s. d. , everett , s. e. , shirley , y. l. , evans , n. j. , ii , palotti , m. l. 2004 , submitted draine , b. t. 1978 , , 36 , 595 draine .
b. t. , & lee h. m. 1984 , , 285 , 89 ebert , r. 1955 , zap , 37 , 217 edmonds , a.r .
1974 , angular momentum in quantum mechanics , 3rd ed .
, ( princeton university press : princeton , new jersey ) egan , m. p. , leung , c. m. , & spagna , g. f. , 1988 , comput .
physics comm . , 48 , 271 elias , j. 1978 , , 224 , 453 evans , n. j. , ii 1999 , ara & a , 37 , 311 evans , n.j . ,
ii , rawlings , j.m.c . , shirley , y.l . , & mundy , l.g .
2001 , 557 , 193 flower , d. r. , pineau des forts , g. , & walmsley , c. m. 2005 , a & a , in press ( astro - ph/0503501 ) foster , p. n. , chevalier , r. a. 1993 , , 416 , 303 hirota , t. , ito , t. , & yamamoto , s. 2002 , , 565 , 359 hirota , t. , maezawa , h. , & yamamoto , s. 2004 , , in press ( astro - ph/0409105 ) geppert et al .
2004 , , 609 , 459 goldsmith , p. f. , & langer , w. d. 1999 , , 517 , 209 gonalves , j. , galli , d. , & walmsley , m. 2004 , a & a , 415 , 617 gordon , v. d. , mccarthy , m. c. , apponi , a. j. , & thaddeus , p. 2001
, , 131 , 311 gordy , w. , & cook , r.l .
1984 , microwave molecular spectroscopy 3rd ed . , in techniques of chemistry vol.17 , ( john wiley & sons : new york ) green , s. , montgomery jr . , j. a. , & thaddeus , p. 1974 , , 193 , l89 greisen , e. & harten 1981 , a&as , 44 , 371 ivezi , z. , nenkova , m. & elitzur m. 1999 , user manual for _ dusty _ , university of kentucky internal report jenness , t. , & lightfoot , j.f .
1997 , surf - scuba user reduction facility ver . 1.1 user s manual , starlink user note 216 ( hilo : joint astronomy centre ) jijina , j. , myers , p. c. , & adams 1999 , , 125 , 161 jrgensen , j. k. , schier , f. l. , & van dishoeck , e. f. 2002 , , 389 , 908 keto , e. rybicki , g. b. , bergin , e. a. , & plume , r. 2004 , , in press ( astro - ph/0407433 ) kooi , j. w. , chan , m. , phillips , t. g. , bumble , b. , & leduc , h. g. 1992 , ieee trans .
microwaves theory and techniques , 40 , 812 kuiper , t.b.h . ,
langer , w.d . , &
velusamy , t. 1996 , apj , 468 , 761 kutner , m. l. , & ulich , b. l. 1981 , , 250 , 341 lada , c. j. , huard , t. l. , crews , l. j. , & alves , j. f. 2004 , , 610 , 303 lai , s .-
p . , velusamy , t. , langer , w. d. , & kupier , t. b. 2003 , , 126 , 311 langer , w. d. , & willacy k. 2001 , , 557 , 714 langston , g. , braatz , j. , ghigo , f. , maddalena , r. , minter , t. , & oneal , k. 2004 , gbt commissioning memo # 24 lee , j .- e . , bergin , e. a. , evans , n. j. , ii 2004 , , in press ( astro - ph/0408091 ) lemme , c. , walmsley , c. m. , wilson , t. l. , & muders , d. 1995 , a & a , 302 , 509 levin , s. m. , langer , w. d. , velusamy , t. , kuiper , t. b. h. , & crutcher , r. m. 2001 , , 555 , 850 li , z .- y . , shematovich , v. i. , wiebe , d. s. , & shustov , b. m. 2002 , , 569 , 792 mathis , j. s. , mezger , p. g. , & panagia , n. 1983 , a & a , 128 , 212 mouschovias , t. c. , & spitzer , l. , jr .
1976 , , 210 326 myers , p.c . , & benson , p.j . 1983 , apj , 266 , 309 mueller , k. e. , shirley , y. l. , evans , n. j. , ii , & jacobson , h. r. 2002 , , 143 , 469 mller , h. s. p. , thorwirth , s. , roth , d. a. , & winnewisser , g. 2001 , a & a , 370 , l49 ossenkopf , v. 1993 , a & a , 280 , 617 ossenkopf , v. , & henning , th .
1994 , a & a , 291 , 943 penzias , a. a. , & burrus , c. a. 1973 , ara & a , 11 , 51 pollack , j. b. , hollenbach , d. , beckwith , s. , simonelli , d. p. , roush , t. , & fong , w. 1994 , , 421 , 615 rawlings , j.m.c . ,
hartquist , t.w . ,
menten , k.m . , & williams , d.a .
1992 , mnras , 255 , 471 rohlfs , k. , & wilson t. l. 2000 , tools of radio astronomy ( springer verlag : berlin ) 194 - 198 rudolph , h. d. 1968 , z. naturforsch , 23a , 540 taylor , s.d . , morata , o. , & williams , d.a .
1996 , a&a , 313 , 269 tafalla , m. , myers , p. c. , caselli , p. , walmsley , c. m. , & comito , c. 2002 , , 569 , 819 tafalla , m. , myers , p. c. , caselli , p. , & walmsley , c. m. 2004 , a&a , 416 , 191 tafalla , m. , & santiago , j. 2004 , a&a , 414 , l53 turner , b. e. 1995 , , 449 , 635 turner , b. e. 2001 , , 136 , 579 shirley , y. l. , evans , n.j , rawling , j.m.c .
, & gregersen , e.m .
2000 , apjs , 131 , 249 shirley , y. l. , evans , n.j , & rawling , j.m.c . 2002b , apj , 575 , 337 shirley y. l. , mueller , k. e. , young , c. h. , evans , n. j. , ii 2003 , in galactic star formation across the stellar mass spectrum , ed .
j. m. de buizer , asp conference series 287 , 298 spitzer , l. , jr .
1978 , physical processes in the interstellar medium , ( john wiley & sons : new york ) 287 stamatellos , d. , & whitworth , a. p. 2004 , ( /astro - ph/0406549 ) sunada , k. yamaguchi , n. , kuno , s. , okumura , s. , nakai , n. , & ukita , n. 2000 , in imaging at radio through submillimeter wavelengths , asp conference series , 217 , 19 tatematsu , k. umemoto , t. , kandori , r. , & sekimoto , y. 2004 , , 606 , 333 van dishoeck , e. f. 1988 , rate coefficients in astrochemistry , ed .
millar & d.a .
williams ( astrophys . &
space sci .
library;dordrecht : kluwer),49 wang , y. 1994 , ph.d .
thesis , the university of texas at austin ward - thompson , d. , scott , p. f. , hills , r. e. , & andr , p. 1994 , mnras , 26 8 , 276 ward - thompson , d. , motte , f. , & andr , p. 1999
, , 305 , 143 weingartner , j. c. , & draine , b. t. 2001 , , 548 , 296 whittet , d. c. b. 2003 , dust in the galactic environment , second ed .
( iop publishing : philadelphia ) , 11 willacy , k. , langer , w.d . , & velusamy , t. 1998 , apjl , 507 , 171 wilson , t. l. , & rood r. 1994 , ara & a , 32 , 191 wolkovitch , d .. langer , w. d. , goldsmith , p. f. , & heyer , m. 1997 , , 477 , 241 langer , w. d. , & willacy , k. 2001 , , 557 , 714 young , c. h. , shirley , y. l. , evans , n. j. , ii , & rawlings , j. m. c. 2003 , , 145 , 111 young , k. e. , lee , j .- e . , evans ii , n. j. , goldmsith , p. f. , & doty , s. 2004 , , in press zhou , s. , evans ii , n. j. , wang , y. , peng , r. , & lo , k. y. 1994 , , 433 , 131 zucconi , a. , walmsley , c. m. , & galli , d. 2001 , a&a , 376 , 650 lccl 90 & @xmath347 1.0 & 150 & 1 + 170 & 10.4 ( ... ) & 150 & 1 + 200 & 15.2 ( ... ) & 150 & 1 + 450 & 0.700 ( 0.080 ) & 18 & 2 + 450 & 2.69 ( 0.59 ) & 40 & 3 + 450 & 14.70 ( 2.42 ) & 120 & 3 + 800 & 0.120 ( 0.018 ) & 18 & 2 + 850 & 0.376 ( 0.021 ) & 40 & 3 + 850 & 2.30 ( 0.08 ) & 120 & 3 + 1100 & 0.035 ( 0.006 ) & 18 & 2 + 1200 & 0.016 ( 0.004 ) & 12 & 4 + 1300 & 0.010 ( 0.003 ) & 12 & 2 + llccc draine - lee & silicates + carbonaceous grains & 2.0 & 0.003 & 1 + wd3 & silicate + carbonaceous + pah ism fit & 1.8 & 0.005 & 2 + wd4 & silicate + carbonaceous + pah ism fit & 1.6 & 0.006 & 2 + pollack1 & silicates + organic c & 2.2 & 0.004 & 3 + mmp & empirical fit to ism & 2.3 & 0.030 & 4 + oh2 & coagulated for @xmath130 yrs .
no ice mantles & 1.3 & 0.034 & 5 + oh5 & coagulated for @xmath130 yrs .
thin ice mantles & 1.8 & 0.018 & 5 + oh8 & coagulated for @xmath130 yrs .
thick ice mantles & 1.9 & 0.022 & 5 + lcccccccc 4.0 & 60 & 0.5 & 1.0 & oh8 & 0.67 & 0.12 & 0.84 + 4.0 & 60 & 2.0 & 4.0 & oh5 & 0.52 & 0.10 & 0.90 + 4.0 & 60 & 2.0 & 6.0 & oh5 & 0.52 & 0.26 & 0.84 + 4.0 & 60 & 0.5 & 2.0 & oh8 & 0.66 & 0.28 & 0.88 + 4.0 & 60 & 1.0 & 1.0 & oh5 & 0.58 & 0.35 & 0.90 + 4.0 & 60 & 2.0 & 8.0 & oh5 & 0.52 & 0.46 & 0.97 + lccccccccccc 90 l1498 & 04 10 53.3 & @xmath2125 09 39 & 1 & 140 & 27600 @xmath348 15000 & 10170 & 1.8 & 0.12 ( 0.02 ) & 5 ( 2 ) & 13.9 ( 3.2 ) & 0.67 + l1517b & 04 55 18.1 & @xmath2130 37 48 & 3 & 140 & 11600 @xmath348 9700 & 5300 & 1.2 & 0.06 ( 0.01 ) & 18 ( 9 ) & 15.2 ( 3.5 ) & 0.67 + l1512 & 05 04 08.2 & @xmath2132 43 20 & 2 & 140 & 19000 @xmath348 8500 & 6350 & 2.2 & 0.10 ( 0.02 ) & 7 ( 4 ) & 15.2 ( 4.0 ) & 0.53 + l1544 & 05 04 17.1 & @xmath2125 10 48 & 2 & 140 & 10900 @xmath348 5700 & 3940 & 1.9 & 0.15 ( 0.02 ) & 5 ( 2 ) & 13.9 ( 3.2 ) & 1.06 + l183 & 15 54 08.5 & @xmath2202 52 32 & 3 & 150 & 11400 @xmath348 6200 & 4200 & 1.8 & 0.09 ( 0.01 ) & 13 ( 3 ) & 15.6 ( 2.5 ) & 0.89 + l1709a & 16 30 50.4 & @xmath2223 42 05 & 3 & 125 & 15800 @xmath348 10500 & 6440 & 1.5 & 0.09 ( 0.02 ) & ... & 15.4 ( 4.9 ) & 0.98 + l1689b & 16 34 49.2 & @xmath2224 38 07 & 2 & 125 & 9600 @xmath348 7700 & 4300 & 1.3 & 0.13 ( 0.02 ) & 7 ( 1 ) & 15.9 ( 3.7 ) & 0.74 + l63 & 16 50 14.4 & @xmath2218 06 17 & 3 & 160 & 12400 @xmath348 7800 & 4920 & 1.6 & 0.20 ( 0.04 ) & 21 ( 2 ) & 16.4 ( 5.3 ) & 1.15 + b68 & 17 22 39.0 & @xmath2223 49 57 & 4 & 95 & 9500 @xmath348 7500 & 5530 & 1.3 & 0.05 ( 0.02 ) & 5 ( 2 ) & 14.5 ( 3.5 ) & 0.50 + b133 & 19 06 08.0 & @xmath2206 52 52 & 2 & 200 & 14900 @xmath348 12100 & 6710 & 1.2 & 0.37 ( 0.06 ) & 9 ( 2 ) & 15.9 ( 4.0 ) & 1.22 + @xmath349 & & & & 142 & & 5790 & 1.6 & 0.14 & 9.9 & 15.2 & 0.8 + @xmath350 & & & & 140 & & 5420 & 1.55 & 0.11 & 7.4 & 15.3 & 0.9 + @xmath351 & & & & 27 & & 1830 & 0.4 & 0.09 & 6.0 & 0.9 & 0.3 + lclccc 1 1 0 @xmath352 0 1 1 & @xmath353 & 1 1 0 @xmath352 0 1 1 & 93.1716200 & @xmath216.944 & @xmath354 + 1 1 2 @xmath352 0 1 2 & @xmath355 & 1 1 2 @xmath352 0 1 2 & 93.1719168 & @xmath215.988 & @xmath356 + 1 1 2 @xmath352 0 1 1 & @xmath357 & + 1 1 1 @xmath352 0 1 2 & @xmath357 & + 1 1 1 @xmath352 0 1 1 & @xmath358 & + 1 1 1 @xmath352 0 1 0 & @xmath353 & 1 1 1 @xmath352 0 1 0 & 93.1720533 & @xmath215.549 & @xmath359 + 1 2 2 @xmath352 0 1 2 & @xmath357 & + 1 2 2 @xmath352 0 1 1 & @xmath355 & 1 2 2 @xmath352 0 1 1 & 93.1734796 & @xmath210.956 & @xmath356 + 1 2 3 @xmath352 0 1 2 & @xmath360 & 1 2 3 @xmath352 0 1 2 & 93.1737767 & @xmath210.000 & @xmath306 + 1 2 1 @xmath352 0 1 2 & @xmath361 + 1 2 1 @xmath352 0 1 1 & @xmath357 & 1 2 1 @xmath352 0 1 1 & 93.1739666 & @xmath220.611 & @xmath359 + 1 2 1 @xmath352 0 1 0 & @xmath362 + 1 0 1 @xmath352 0 1 2 & @xmath362 & 1 0 1 @xmath352 0 1 2 & 93.1762650 & @xmath228.011 & @xmath359 + 1 0 1 @xmath352 0 1 1 & @xmath353 & + 1 0 1 @xmath352 0 1 0 & @xmath363 & + | pre - protostellar cores likely represent the incipient stages of low - mass ( @xmath0 ) star formation .
lynds 1498 is a pre - protostellar core ( ppc ) and was one of the initial objects toward which molecular depletion and differentiation was detected . despite the considerable scrutiny of l1498 , there has not been a extensive study of the density and temperature structure as derived from radiative transfer modeling of dust continuum observations .
we present deep scuba observations of l1498 at 850 and 450 , high resolution bears maps of the n@xmath1h@xmath2 @xmath3 transition , cso observations of the n@xmath1h@xmath2 @xmath4 transition , and gbt observations of the c@xmath5s @xmath6 transition .
we also present a comparison of derived properties between l1498 and nearby ppcs that have been observed at far - infrared and submillimeter wavelengths .
the l1498 continuum emission is modeled using a one - dimensional radiative transfer code that self - consistently calculates the temperature distribution and calculates the sed and intensity profiles at 850 and 450 @xmath7 m .
we present a more realistic treatment of ppc heating which varies the strength of the isrf , , and includes attenuation of the isrf due to dust grains at the outer radius of the core , .
the best - fitted model consists of a bonner - ebert sphere with a central density of @xmath8 @xmath9 , @xmath10 pc , @xmath11 , a@xmath12 mag , and a nearly isothermal temperature profile of @xmath13 k for oh8 opacities .
c@xmath5s
emission shows a central depletion hole while n@xmath1h@xmath2 emission is centrally peaked .
we derive a mean n@xmath1h@xmath2 abundance of @xmath14 relative to h@xmath1 that is consistent with chemical models for a dynamically young yet chemically evolved source .
the observed depletions of c@xmath5s and h@xmath1co , the modest n@xmath1h@xmath2 abundance , and a central density that is an order of magnitude lower than other modeled ppcs suggests that l1498 may be a forming ppc .
our derived temperature and density profile will improve modeling of molecular line observations that will explicate the core s kinematical and chemical state . |
in the recent years the na49 experiment has collected data on pb+pb collisions at beam energies between 20 to 158with the objective to cover the critical region of energy densities where the expected phase transition to a deconfined phase might occur in the early stage of the reactions . in this contribution the energy dependence of @xmath0- and rapidity - distributions , as well as the production rates of strange particles are reviewed .
na49 is a fixed target experiment at the cern sps and consists of a large acceptance magnetic spectrometer equipped with four tpcs as tracking devices and a forward calorimeter for centrality selection .
details on the experimental setup can be found in @xcite .
, @xmath1 , @xmath2 ) and 10% ( @xmath3 , @xmath4 , @xmath5 ) at 158 ) .
the closed symbols indicate measured points , open points are reflected with respect to mid - rapidity .
the solid lines represent fits with a single gaussian or the sum of two gaussians.,width=415 ] the large acceptance of the na49 spectrometer allows to measure particle spectra over a wide range of the longitudinal phase space .
[ rapall ] shows a compilation of the rapidity distributions of @xmath6 , @xmath1 , @xmath2 , @xmath3 , @xmath4 , and @xmath5 .
the shapes of these distributions are generally well described by a gaussian . only the @xmath5-distributions exhibit a strong variation of the shape .
while at 30 they are still gaussian - shaped , a clear plateau develops with increasing beam energy . , @xmath7 , @xmath3 , and @xmath4 in central pb+pb ( au+au ) collisions as a function of @xmath8 .
ags data are taken from @xcite .
the solid line is a linear fit to the pion data .
the dashed lines have the same slope , but shifted to match the other particle species.,height=245 ] an increase of the rms - widths of the rapidity spectra , calculated from the fits shown in fig .
[ rapall ] , with beam energy can be observed which for the pions exhibits to a good approximation a linear dependence on the beam rapidity in the center - of - mass system @xmath8 over the whole energy range covered by the ags and sps ( see fig .
[ rmsybeam ] ) . between 20and 158
this is also true for the other particle types having a gaussian - like distribution , with a clear hierarchy in the widths : @xmath9 .
however , this seems to break down at lower energies , where the widths of the kaons apparently approach the ones of the pions . the dependence of the rms - widths on the particle mass at sps energies is shown in fig . [ rmsmass ] .
an approximately linear mass dependence is observed , which appears to have the same slope for all sps - energies .
this mass dependence can be interpreted as a result of the thermal spectrum of hadrons superimposed on the longitudinal collective expansion .
for pions , kaons , and protons at mid - rapidity for 5 - 10% most central pb+pb / au+au reactions .
open symbols represent negatively charged particles . ,
width=453 ] the increase with energy of the inverse slope parameter @xmath10 of the kaon @xmath0-spectra , as derived from an exponential fit , exhibits a sharp change to a plateau around 30 @xcite .
since the kaon @xmath0-spectra in contrast to the ones of the lighter pions or the heavier protons have to a good approximation an exponential shape , the inverse slope parameter provides in this case a good characterization of the spectra . for other particle species , however , the local slope of the spectra depends on @xmath0 .
instead , the first moment of the @xmath0-spectra can be used to study their energy dependence .
the dependence of @xmath11 on the center of mass energy @xmath12 is summarized in fig .
[ meanmt ] .
the change of the energy dependence around around a beam energy of 20 30 .
is clearly visible for pions and kaons .
while @xmath11 rises steeply in the ags energy range , the rise is much weaker from the low sps energies on . to a lesser extend this change
is also seen for protons .
-yields of strange hadrons , normalized to the pion yields , in central pb+pb / au+au collisions .
the data are compared to string hadronic models @xcite ( urqmd 2.0 : dotted lines , hsd : dashed - dotted lines ) and statistical hadron gas models @xcite
( with strangeness under - saturation : dashed line , assuming full equilibrium : solid line).,height=453 ] -yields of strange hadrons , normalized to the pion yields , in central pb+pb / au+au collisions . the data are compared to string hadronic models @xcite ( urqmd 2.0 : dotted lines , hsd : dashed - dotted lines ) and statistical hadron gas models @xcite ( with strangeness under - saturation : dashed line , assuming full equilibrium : solid line).,height=453 ] by integrating a measured rapidity distribution , as shown in fig .
[ rapall ] , the total multiplicity of a given particle type can be determined . in fig .
[ ratios ] the energy dependence of the total multiplicities for a variety of strange hadrons , normalized to the total pion yield , is summarized and compared to model predictions .
a detailed discussion can be found in @xcite .
generally , it can be stated that string hadronic models urqmd and hsd @xcite do not provide a good description of the data points .
especially the @xmath13 and @xmath14 production is substantially underestimated and the maximum in the @xmath2/@xmath6ratio is not reproduced .
the statistical hadron gas models @xcite , on the other hand , provide a better overall description of the measurements .
however , the introduction of an energy dependent strangeness under - saturation factor @xmath15 is needed @xcite , in order to capture the structures in the energy dependence of most particle species ( @xmath1 , @xmath2 , @xmath3 , @xmath13 ) . .
the solid line represents the prediction of the statistical hadron gas model with full equilibration of strangeness @xcite.,height=226 ] .
the solid line represents the prediction of the statistical hadron gas model with full equilibration of strangeness @xcite.,height=226 ] from the measured total yields of the strange particles the energy dependence of the number of produced strange quarks and anti - quarks is constructed ( see left panel of fig . [ ssbar ] ) .
the strange quark carriers which are taken into account are @xmath2 , @xmath16 , @xmath5 ( including @xmath17 ) , @xmath18 , @xmath19 , and @xmath20 . for the strange anti - quark
these are @xmath1 , @xmath21 , @xmath4 ( including @xmath22 ) , @xmath23 , @xmath24 , and @xmath25 contribution is calculated using isospin symmetry ( @xmath26 , @xmath27 ) . if no measurement is available , the values for the @xmath13 and @xmath14 yields were taken from statistical model fits @xcite .
the @xmath20 contribution is estimated based on the empirical factor @xmath28 @xcite .
note that the strange quarks from the @xmath3 and @xmath29 are not included . ]
the @xmath30- and @xmath31-yields derived from the na49 measurements agree at all energies , illustrating the consistency of the analysis . a departure from the energy dependence of strangeness production as observed at lower energies ( indicated by the straight line ) is observed around 30 . if divided by the total number of pions a clear maximum of the relative strangeness production at the same energy can be seen ( right panel of fig . [ ssbar ] ) .
while the statistical hadron gas model with full strangeness equilibration @xcite matches the data below this maximum quite well , it over - predicts the measurements at higher energies .
is the endpoint of the first order transition line ( thin line on the right side of @xmath32 ) @xcite . on the left side of @xmath32
a smooth cross over is expected .
the points are the chemical freeze - out points as derived from a fit with a statistical hadron gas model @xcite .
the open symbols schematically indicate possible initial parameters of the reaction systems , which then might evolve along paths as depicted by the vertical lines.,height=283 ] the results of the na49 energy scan program have revealed a variety of interesting features . among these
are a clear change in the energy dependence of the @xmath0-spectra and a maximum in the strangeness to pion ratio around 30 .
both are difficult to explain in a hadronic scenario , but can be understood as a reflection of a phase transition @xcite .
one of the most remarkable features of the phase diagram , shown in fig .
[ phasediag ] , is the fact that the line of the first order phase transition ends in a second order critical point @xmath32 ( for a review of the current theoretical situation see @xcite ) .
its position is subject to large theoretical uncertainties , but recent lattice calculations @xcite with physical quark masses indicate that it might be around @xmath33 mev .
if this estimate holds , it might be possible to access the critical endpoint with heavy ion reactions at sps energies .
since the above described results on the energy dependence of hadronic observables might indicate that the phase transition is already reached at beam energies of 30 , where the chemical freeze - out happens at @xmath34 mev .
this would imply that at this energy the first order transition line is crossed ( see vertical lines in fig .
[ phasediag ] ) , while at the top sps energy the evolution passes through the rapid crossover region .
therefore , a careful scan of this region of beam energy , accompanied by an additional variation of the system size , might allow to experimentally identify the position of the endpoint .
an indication for the endpoint would possibly be a strong increase of fluctuations in e.g. transverse momentum , particle ratios or multiplicities .
so far , only limited attempts of systematically studying these effects in the interesting energy range have been made .
significant non - statistical fluctuations of the kaon to pion ratio at low sps energies are reported in @xcite . also , large multiplicity and @xmath35fluctutions in collisions of small systems at 158have been observed @xcite
. however , these effects are still far from having a clear connection to the critical point .
future measurements that would scan the interesting energy range and the system size more closely , might have a chance of seeing stronger evidence for its existence .
this work was supported by the us department of energy grant de - fg03- + 97er41020/a000 , the bundesministerium fur bildung und forschung , germany , the polish state committee for scientific research ( 2 p03b 130 23 , spb / cern / p-03/dz 446/2002 - 2004 , 2 p03b 04123 ) , the hungarian scientific research foundation ( t032648 , t032293 , t043514 ) , the hungarian national science foundation , otka , ( f034707 ) , the polish - german foundation , and the korea research foundation grant ( krf-2003 - 070-c00015 ) .
* the na49 collaboration : + c. alt@xmath36 , t. anticic@xmath37 , b. baatar@xmath38,d .
barna@xmath39 , j. bartke@xmath40 , l. betev@xmath41 , h. biakowska@xmath42 , a. billmeier@xmath36 , c. blume@xmath36 , b. boimska@xmath42 , m. botje@xmath43 , j. bracinik@xmath44 , r. bramm@xmath36 , r. brun@xmath45 , p. buni@xmath41 , v. cerny@xmath44 , p. christakoglou@xmath46 , o. chvala@xmath47 , j.g .
cramer@xmath48 , p. csat@xmath39 , n. darmenov@xmath49 , a. dimitrov@xmath49 , p. dinkelaker@xmath36 , v. eckardt@xmath50 , g. farantatos@xmath46 , d. flierl@xmath36 , z. fodor@xmath39 , p. foka@xmath51 , p. freund@xmath50 , v. friese@xmath51 , j. gl@xmath39 , m. gadzicki@xmath36 , g. georgopoulos@xmath46 , e. gadysz@xmath40 , k. grebieszkow@xmath52 , s. hegyi@xmath39 , c. hhne@xmath53 , k. kadija@xmath37 , a. karev@xmath50 , m. kliemant@xmath36 , s. kniege@xmath36 , v.i .
kolesnikov@xmath38 , t. kollegger@xmath36 , e. kornas@xmath40 , r. korus@xmath54 , m. kowalski@xmath40 , i. kraus@xmath51 , m. kreps@xmath44 , m. van leeuwen@xmath43 , p. lvai@xmath39 , l. litov@xmath49 , b. lungwitz@xmath36 , m. makariev@xmath49 , a.i .
malakhov@xmath38 , c. markert@xmath51 , m. mateev@xmath49 , b.w .
mayes@xmath55 , g.l .
melkumov@xmath38 , c. meurer@xmath36 , a. mischke@xmath51 , m. mitrovski@xmath36 , j. molnr@xmath39 , s. mrwczyski@xmath54 , g. plla@xmath39 , a.d .
panagiotou@xmath46 , d. panayotov@xmath49 , a. petridis@xmath46 , m. pikna@xmath44 , l. pinsky@xmath55 , f. phlhofer@xmath53 , j.g .
reid@xmath48 , r. renfordt@xmath36 , a. richard@xmath36 , c. roland@xmath56 , g. roland@xmath56 , m. rybczyski@xmath54 , a. rybicki@xmath57 , a. sandoval@xmath51 , h. sann@xmath51 , n. schmitz@xmath50 , p. seyboth@xmath50 , f. siklr@xmath39 , b. sitar@xmath44 , e. skrzypczak@xmath52 , g. stefanek@xmath54 , r. stock@xmath36 , h. strbele@xmath36 , t. susa@xmath37 , i. szentptery@xmath39 , j. sziklai@xmath39 , t.a .
trainor@xmath48 , v. trubnikov@xmath52 , d. varga@xmath39 , m. vassiliou@xmath46 , g.i .
veres@xmath58 , g. vesztergombi@xmath39 , d. vrani@xmath51 , a. wetzler@xmath36 , z. wodarczyk@xmath54 , i.k .
yoo@xmath59 , j. zaranek@xmath36 , j. zimnyi@xmath39 + @xmath43nikhef , amsterdam , netherlands .
@xmath46department of physics , university of athens , athens , greece .
@xmath44comenius university , bratislava , slovakia .
@xmath39kfki research institute for particle and nuclear physics , budapest , hungary .
@xmath56mit , cambridge , usa .
@xmath40institute of nuclear physics , cracow , poland .
@xmath51gesellschaft fr schwerionenforschung ( gsi ) , darmstadt , germany .
@xmath38joint institute for nuclear research , dubna , russia .
@xmath36fachbereich physik der universitt , frankfurt , germany .
@xmath45cern , geneva , switzerland .
@xmath55university of houston , houston , tx , usa . @xmath54institute of physics wietokrzyska academy , kielce , poland .
@xmath53fachbereich physik der universitt , marburg , germany .
@xmath50max - planck - institut fr physik , munich , germany .
@xmath47institute of particle and nuclear physics , charles university , prague , czech republic .
@xmath59department of physics , pusan national university , pusan , republic of korea .
@xmath48nuclear physics laboratory , university of washington , seattle , wa , usa .
@xmath49atomic physics department , sofia university st .
kliment ohridski , sofia , bulgaria .
@xmath42institute for nuclear studies , warsaw , poland .
@xmath52institute for experimental physics , university of warsaw , warsaw , poland .
@xmath37rudjer boskovic institute , zagreb , croatia . | new results of the na49 collaboration on strange particle production are presented .
rapidity and transverse mass spectra as well as total multiplicities are discussed .
the study of their evolution from ags over sps to the highest rhic energy reveals a couple of interesting features .
these include a sudden change in the energy dependence of the @xmath0-spectra and of the yields of strange hadrons around 30 .
both are found to be difficult to be reproduced in a hadronic scenario , but might be an indication for a phase transition to a quark gluon plasma . |
the fact that the speed of physical interactions is finite , as for light or sound , has well known crucial consequences in all physics , and it is the ultimate reason for the existence of waves .
for example , if we artificially assume the speed of light as infinite in the maxwell equations , the magnetic field would be zero and we would not have electromagnetic waves . the situation is quite different when the equations contain non linear terms , which can result in erratic or even chaotic trajectories .
notice that , in the absence of mechanical kinetic terms , which is our framework , the presence of non linear terms is exclusively due to the presence of non bilinear interactions among the components of the system .
for example , in the kuramoto model @xcite-@xcite , sinusoidal interactions lead to chaos @xcite-@xcite if the coupling constant @xmath0 is larger than some threshold @xmath1 . it should be emphasized that the presence of non linear terms is not always sufficient for the emergence of chaos .
however , it has been observed that the introduction of time - delays among the components of a system augments the chance that the system develops chaos @xcite-@xcite .
the idea at the base of such a mechanism is that time - delays among the system components makes it difficult for the system to set in a steady state . in particular , for linear flows , a fixed delay among the system components corresponds to a finite speed of interactions , which in turn leads to oscillations or waves . for
non linear flows a delay can have more dramatic consequences , with chaos scenarios more probable . in @xcite , we have introduced a simple discrete - time two components dynamical model , characterized by local - time updates .
local time updates means that the each component updates its status only at certain times @xmath2 . in the specific model analyzed in @xcite ,
a component updates ( sleeps ) its status only at odd times @xmath3 , ( even times @xmath4 ) , while the other component updates ( sleeps ) its status only at even times @xmath4 ( odd times @xmath3 ) .
remarkably , as we have shown in @xcite , for certain coupling values , this simple system exhibits chaos even if it is characterized only by bilinear interactions .
in fact , when the couplings are in the steady state regime , the system can be described via a hamiltonian which is quadratic in the system components .
we are not aware of a similar case in the literature .
our understanding here is that there are three factors generates chaos : _ i ) _ the presence of unfriendly couplings , _ ii ) _ the discrete - time nature of the dynamics , and _
iii ) _ the existence of at least two interacting communities which rearrange their configurations at alternate times , _
i.e. _ via local - time updates .
notice that _
( iii ) _ is not just a delay ; it also implies that each component has a sleeping period .
it is during such a sleeping period that the component accumulates a tension with respect to the other component when the reciprocal coupling is unfriendly .
we are in fact dealing with a sort of dynamical frustration .
in this paper we review some aspects of the model which were only partially treated in @xcite . in particular , we compare the analogous flow model with the present map and show that the above conditions _ ( i)_-_(iii ) _ are in fact necessary , and calculate the maximal lyapunov exponent .
the reader not interested in the derivation of the dynamic equations may skip the next section .
here we consider a two components system .
generalization to an arbitrary number of components will be reported elsewhere .
we consider two communities @xcite of agents @xmath5 and @xmath6 , with cardinalities @xmath7 and @xmath8 .
the agent with index @xmath9 can be in two possible status , @xmath10 , with @xmath11 . according to the sign of the couplings , friendly or unfriendly
, each agent @xmath9 tends to follow or anti - follow its neighbors , by minimizing or maximizing the term @xmath12 , where @xmath13 runs over the set of neighbors of @xmath9 .
it is necessary to distinguish between intra- and inter - couplings , therefore we introduce the @xmath14 matrix @xmath15 : @xmath16 and @xmath17 are the intra - couplings , and @xmath18 the inter - coupling . in the most general formulation
we should introduce also the @xmath14 matrix @xmath19 defined as the set of coupled spins @xmath20 within the same community ( intra ) , or between the two communities ( inter ) . finally , we introduce a global factor @xmath21 that rescales all the couplings in @xmath22 .
let us indicate the time by @xmath2 , which is supposed to range in the set of natural numbers @xmath23 for the map , while it can take any non negative real number for the flow .
inspired by the glauber dynamics @xcite , for flow and map we introduce the following transition rate probabilities for the spin with state @xmath24 to jump to the state @xmath25 @xmath26 let @xmath27 be any periodic partition of @xmath23 , _
i.e. _ , @xmath28 , @xmath29 , and @xmath30 and @xmath31 contain both infinite elements of @xmath23 .
for instance , @xmath30 and @xmath31 can be the set of even and odd numbers , respectively , and we will assume this choice in all the next examples .
we now introduce the following local - time transition rate probabilities for the spin with state @xmath24 to jump to the state @xmath25 @xmath32\neq 0 , \\ \\
\frac{1+\sigma_i'\tanh\left(\beta j^{(1,1)}\sum_{j:(i , j)\in \gamma^{(1,1)}}\sigma_j+ \beta j^{(1,2)}\sum_{j:(i , j)\in \gamma^{(1,2)}}\sigma_j\right)}{2 } , \quad \\ \mathrm{for~ } i\in\mathcal{n}^{(1 ) } , \quad \mathrm{if~}\mod[t,2]=0,\\ \\ \delta_{\sigma_i,\sigma_i ' } \quad i\in\mathcal{n}^{(2 ) } , \quad \mathrm{if~}\mod[t,2]\neq 0,\\ \\
\frac{1+\sigma_i'\tanh\left(\beta j^{(2,2)}\sum_{j:(i , j)\in \gamma^{(2,2)}}\sigma_j+ \beta j^{(2,1)}\sum_{j:(i , j)\in \gamma^{(2,1)}}\sigma_j\right)}{2 } , \quad \\ \mathrm{for~ } i\in\mathcal{n}^{(2 ) } , \quad \mathrm{if~}\mod[t,2]=0 , \end{array } \right .
\end{aligned}\ ] ] @xmath33 from the viewpoint of modeling , eqs .
( [ rate0 ] ) , ( [ rate ] ) , or ( [ rate1 ] ) , are justified as they make each spin to follow the majority of its intra- and inter - neighbors and , thanks to the presence of the functions @xmath34 , the rates are non negative and normalized at any time @xmath35(time unit ) . from a deeper viewpoint ,
( [ rate0 ] ) , ( [ rate ] ) , or ( [ rate1 ] ) , are based on the fact that , as we will see soon , in the case of positive couplings they lead to boltzmann equilibrium governed by only quadratic interactions .
more precisely , in the case of positive couplings these forms guarantee that at equilibrium the system satisfies the principle of detailed balance and the principle of maximal entropy for any quadratic interactions .
we formalize the discrete - time probabilistic dynamics induced by eqs .
( [ rate0 ] ) , ( [ rate ] ) , or ( [ rate1 ] ) , or as follows .
let @xmath36 .
let us introduce the spin vector @xmath37 , and the associated probability vector @xmath38 , _ i.e. _ , the probability that the system is in the configuration @xmath39 at time @xmath40 .
the master equation for flow and map ( local - time or not ) reads @xmath41 @xmath42 where we have introduced the global transition rates@xmath43 where @xmath44 can be of the type eqs .
( [ rate0 ] ) , ( [ rate ] ) , or ( [ rate1 ] ) , and where @xmath45 may be interpreted as the rate at which , due to the interaction with an environment , a free spin ( @xmath46 ) makes transitions from either state to the other . as we have proved in @xcite , it is necessary to impose the bound @xmath47 for @xmath38 to be a probability at any time @xmath2 . by using eqs .
( [ rate])-([w ] ) it is easy to check that the stationary solutions @xmath48 of eq .
( [ master1 ] ) are given by the boltzmann distribution @xmath49 $ ] , where @xmath50 of course , the existence of a stationary solution @xmath48 does not represent a sufficient condition for equilibrium .
in fact , asymptotically the system can reach non - point - like attractors , and even aperiodic or chaotic regimes .
however , as we have discussed in the introduction , the fact that for certain values of couplings the system can be described via the hamiltonian ( [ h ] ) , means that we are dealing with only bilinear interactions . here , the non linearities that lead to chaos are caused by the local time updates .
from now on , we will omit the suffixes ( flow ) , ( map ) and ( local - time )
. it will be clear from the context what we are referring to .
( [ rate])-([w ] ) define the microscopic dynamics from which one can derive the macroscopic dynamics , _
i.e. _ , the dynamics for the order parameters @xmath51
the mean - field limit is defined by the settings @xmath52 , @xmath53 , @xmath54 , and the replacements @xmath55 , @xmath56 , @xmath57 .
we parametrize the size of the two communities as @xmath58 let us introduce the matrix @xmath59 @xmath60 as shown in @xcite , for @xmath61 we obtain the following deterministic evolution eqs . for the order parameters ( [ mm1])-([mm2 ] ) @xmath62 @xmath63 @xmath64 @xmath65 @xmath66\neq 0 , \\ \\
\tanh\left(\beta \tilde{j}^{(1,1)}x^{(1)}(t)+\beta \tilde{j}^{(1,2)}x^{(2)}(t)\right)-x^{(1)}(t),~\mathrm{if~}\mod[t,2]=0 , \end{array } \right.\end{aligned}\ ] ] @xmath67\neq 0,\\ \\ \tanh\left(\beta \tilde{j}^{(2,2)}x^{(2)}(t)+\beta \tilde{j}^{(2,1)}x^{(1)}(t)\right)-x^{(2)}(t),~\mathrm{if~}\mod[t,2]=0 .
\end{array } \right.\end{aligned}\ ] ] @xmath68 @xmath69 when the couplings are positive , there is little difference among the four dynamics , and they all asymptotically reach equilibrium according to the boltzmann distribution @xmath70 $ ] , where @xmath71 $ ] is the mean - field analogous of eq .
( [ h ] ) , and @xmath72 and @xmath73 tend , for @xmath74 , to the stationary solutions of eqs .
( [ mm1])-([ffl2 ] ) , _ i.e. _ , @xmath75 eqs .
( [ mstat ] ) represent a particular case of the general result derived in @xcite valid for @xmath76 interacting communities at equilibrium . in particular
, one can check that eqs .
( [ mstat ] ) give rise to second order phase transitions whose critical surface is determined by the condition @xmath77 in general , @xmath78 is stable when the eigenvalues of @xmath79 are inside the interval @xmath80 ( disordered phase , or no consensus ) , otherwise the system reaches a spontaneous ordered phase @xmath81 ( frozen phase , or consensus ) .
however , eq . ( [ crit ] ) is exact only for the flow and the map . in general ,
due to the sleeping time , the critical value of @xmath21 can differ from the one given by eq .
( [ crit ] ) .
this was erroneously not stressed in @xcite .
the correct way to to derive the exact critical value of @xmath21 must use the full expression of the jacobian , which is the subject of section 5 .
from now on we shall always assume @xmath82 and @xmath83 . in this section
we compare the four dynamics numerically . in figs ( 1a)-(2b )
we compare the normal flow and map . in figs .
( 1a ) and ( 1b ) we consider a situation with positive couplings at small and large values of @xmath21 ; in figs .
( 2a ) and ( 2b ) we consider a situation with positive and negative couplings at small and large values of @xmath21 , respectively .
it is evident that , if the couplings are positive , the flow and the map remain roughly close at any time , and tend to coincide for @xmath74 to the same stationary solution provided by eq .
( [ mstat ] ) . when instead some of the couplings are negative , at small @xmath21 the map tends to make damping oscillations around the flow ( which for the present case is null ) , whereas , at large @xmath21 , the map tends to make regular oscillations around the flow solution ( which for the present case is null ) . and @xmath73 , solution of the flow , eqs .
( [ ff1])-([ff2 ] ) , and the map , eqs .
( [ mm1])-([mm2 ] ) ( here with @xmath82 ) , for positive couplings ; below ( panel a ) and above ( panel b ) the threshold @xmath84 .
initial conditions are the same in all cases
. the lines of the map are guides for the eyes .
[ fig1 ] , title="fig:",height=268 ] + and @xmath73 , solution of the flow , eqs .
( [ ff1])-([ff2 ] ) , and the map , eqs .
( [ mm1])-([mm2 ] ) ( here with @xmath82 ) , for a case with positive and negative couplings ; below ( panel a ) and above ( panel b ) the threshold @xmath84 .
initial conditions are the same in all cases
. the lines of the map are guides for the eyes .
[ fig2 ] , title="fig:",height=268 ] + in figs .
( 3a)-(3c ) we compare the flow and map with local - time updates for a case with positive and negative couplings and three values of @xmath21 , respectively .
( 3a ) provides the behavior for @xmath85 where the map follows damping oscillations and both the map and the flow tend to the same stationary value provided by eq .
( [ mstat ] ) .
( 3b ) provides the behavior for a relatively larger value of @xmath21 , @xmath86 , where the map follows regular oscillations of period 2 , while the flow tends to a stationary value .
( 3c ) provides the behavior for @xmath87 , where the map follows regular oscillations of period 4 , while the flow tends to a stationary value .
increasing further @xmath21 leads to larger and larger periods ( not shown ) .
+ + ( red dots ) , solution of the local - time map , eqs .
( [ mml1])-([mml2 ] ) ( here with @xmath82 ) as a function of @xmath21 , with @xmath88 high enough to remove temporary transients . in this case @xmath89 .
we plot also the maximum lyapunov exponent @xmath90 ( black highly irregular line ) calculated from eqs .
( [ glauberm2])-([q ] ) .
each value of @xmath21 is in correspondence with a random initial condition for a total of @xmath91 samples .
panel ( a ) is a case with bifurcation cascade , panel ( b ) is a case with marginal chaos .
[ fig4 ] , title="fig:",height=268 ] +
the case analyzed in figs . ( 3 ) shows that , for large values of @xmath21 , the local - time map can lead to erratic and , possibly , chaotic trajectories . in general
, for each dynamics , the major quantity of interest is the jacobian @xmath92 , where @xmath93 : @xmath94\beta j^{(l , m)},\end{aligned}\ ] ] @xmath95\beta j^{(l , m)},\end{aligned}\ ] ] @xmath96\neq 0 , \\ \\
\left[1-\left(x^{(l)}_{t}\right)^2\right]\beta j^{(l , m ) } , \quad \mod[t+l-1,2]=0 , \end{array } \right.\end{aligned}\ ] ] @xmath97\neq 0 , \\ \\
\left[1-\left(x^{(l)}_{t}\right)^2\right]\beta j^{(l , m ) } , \quad \mod[t+l-1,2]=0 .
\end{array } \right.\end{aligned}\ ] ] the knowledge of @xmath92 allows to calculate the maximum lyapunov exponent @xmath98 .
it turns out that in the map case ( local - time or not ) , due to the fact that - via recurrence - we have the explicit solution of the eqs .
, the evaluation of @xmath98 is easier if compared to the flow case .
since the flow does not show any signs of chaos or erratic trajectories , we can limit the evaluation of the lyapunov exponent to the map , where we can use @xmath99 where @xmath100 is the largest eigenvalue of the following matrix @xmath101 given a set of couplings , we let the system to evolve toward high enough values of @xmath102 in order to remove temporary transients and repeat the numerical experiment for several values of @xmath21 , each @xmath21 being associated to a random initial condition .
hence , we plot @xmath103 as functions of @xmath21 , and we indicate these functions as @xmath104 . in our examples and for
the range of @xmath21 considered , @xmath105 turns out to be enough high .
in fact , the variable @xmath21 plays the role of a time - scaling : the higher @xmath21 , the higher the necessary transient @xmath88 . in general , the functions @xmath104 look multi - valued functions due to the existence of bifurcation points .
there exist two kinds of bifurcations : doubling period and phase transition .
however , it is clear that in either case , doubling period or phase transition , the presence of bifurcations increases the chance to develop chaos . therefore , for our aims here , it is more interesting , as well as highly more efficient , to show the plots that include all the bifurcation points . figs .
( 4a)-(4b ) show two different scenarios .
the case of panel ( a ) is a `` classical '' bifurcation cascade @xcite , where , after a threshold @xmath106 , bifurcation takes place increasingly , up to windows of stability . in the case of panel ( b )
we observe a different situation which we call `` marginal chaos '' . in this case , after a threshold @xmath107 , the dynamics is characterized by totally erratic trajectories with no period , up to windows of stability . strictly speaking , even out of the windows of stability , the system is not chaotic , since the maximum lyapunov exponent is @xmath108 .
however , in such regions , the system turns out to be critical over continuous intervals ; a situation which is not typical in models characterized by discrete symmetries like the ising model .
we have analyzed and compared four kinds of dynamics of a simple two component ising - like system : flow , map , local - time flow , and local - time map .
remarkably , even if only bilinear interactions are present , the local - time map gives rise to erratic trajectories and , depending on the set of couplings , two chaotic scenarios take place : bifurcation cascades or `` marginal chaos '' , _
i.e. _ , criticality extended over continuous intervals of @xmath21 , the time - scale parameter .
the analogous local - time flow does not present any of such behaviors , not even oscillations , confirming all the scenarios we have already discussed in @xcite .
we stress that this local - time update is not just a time - delay : here we have also a sleeping time during which , in the case of some unfriendly couplings , there is an accumulation of frustration among the system components , giving rise to instability and chaos , even for a system characterized by only bilinear interactions . as we have discussed in @xcite , we believe that maps with local - time updates are a quite common feature in complex systems .
urgent issues will be to understand how this scenario generalizes to large systems with many , say @xmath109 components , how the probability to have chaos changes with @xmath109 , and if the mean - field ( or more precisely fully connected ) picture remains robust .
work supported by cnpq grant pds 150934/2013 - 0 .
w. f. also acknowledges the brazilian agencies cnpq and capes .
9 m. r. may , _ simple mathematical models with very complicated dynamics _ , nature * 261 * , 459 - 467 ( 1976 )
. s. h. strogatz , `` nonlinear dynamics and chaos , with applications to physics , biology , chemistry , and engineering '' , reading , ma : addison - wesley ( 1994 ) . | we consider a simple model of communities interacting via bilinear terms . after analyzing the thermal equilibrium case , which can be described by an hamiltonian
, we introduce the dynamics that , for ising - like variables , reduces to a glauber - like dynamics .
we analyze and compare four different versions of the dynamics : flow ( differential equations ) , map ( discrete - time dynamics ) , local - time update flow , and local - time update map .
the presence of only bilinear interactions prevent the flow cases to develop any dynamical instability , the system converging always to the thermal equilibrium .
the situation is different for the map when unfriendly couplings are involved , where period - two oscillations arise . in the case of the map with local - time updates , oscillations of any period and chaos
can arise as a consequence of the reciprocal `` tension '' accumulated among the communities during their sleeping time interval .
the resulting chaos can be of two kinds : true chaos characterized by positive lyapunov exponent and bifurcation cascades , or marginal chaos characterized by zero lyapunov exponent and critical continuous regions . |
through a strong coupling expansion of a wilson loop in a space - time ( euclidean ) lattice , wilson @xcite showed evidence for confinement in a pure _ compact _ @xmath3 gauge theory , marking the beginning of a new method for nonperturbative investigation of quantum field theories .
ironically , a similar calculation is equally applicable to pure @xmath0 gauge theory that shows nontrivial properties at strong gauge coupling ( the compact formulation allows self - interaction for all powers of the abelian gauge fields on the discrete lattice ) .
the hallmark of wilson s approach is that gauge invariance is manifest at all stages of the calculation and gauge - fixing is not required .
the theory is rigorously defined through a functional integral with a gauge - invariant ( haar ) measure with group - valued gauge fields .
the algebra - valued gauge fields become noncompact , smooth and dimensionful only in the continuum limit . because of known physics from weak - coupling quantum electrodynamics , a @xmath0 gauge theory , it was expected that at an intermediate gauge coupling , there would be a phase transition from the strong coupling to a familiar weak coupling phase with free massless photons in the continuum limit for the pure gauge theory . indeed monte carlo simulations found this transition , which was later confirmed to be first order @xcite .
absence of a diverging correlation length meant that no quantum continuum limit could be taken in this @xmath0 theory .
we shall now take a short detour to lattice formulations of chiral gauge theories to understand why in certain situations there is a need to control the longitudinal modes of lattice gauge fields .
fermions on the discrete lattice necessarily break chiral symmetry @xcite . for chiral gauge theories ,
obviously the gauge symmetry is then explicitly broken on the lattice .
lack of gauge invariance in the wilson framework ( without gauge - fixing ) necessarily means strong coupling between the physical degrees of freedom and the longitudinal gauge degrees of freedom ( _ lgdof_s ) .
this is explained in the following .
because of the haar measure , the functional integral is over all gauge configurations , including the ones related to each other by gauge transformations . as a result , after a gauge transformation , the _ lgdof_s become explicitly present in the action and interact with the physical degrees of freedom .
this interaction is strong because there is no gauge - fixing and any point on the gauge orbit is as likely as any other , essentially making the gauge fields very rough .
the rough gauge problem was the main reason of failure of a full class of lattice chiral gauge theories @xcite .
these failures gave rise to the understanding that controlling the dynamics of _ lgdof_s in these theories ( in other words , gauge - fixing ) is essential to avoid undesirable results .
however , the becchi - rouet - stora - tyutin ( brst ) scheme , a standard mechanism for taking care of the redundancy related to the _ lgdof_s , can not be used in this general nonperturbative case with compact gauge fields , because of a theorem , due to neuberger , that proves that the partition function and the unnormalized expectation value of a gauge - invariant operator are each zero in presence of a brst symmetry @xcite . this is presumably due to the cancelling contributions from a bunch of gribov copies signalling multiple solutions of the gauge - fixing condition . for the general non - abelian case
, the above theorem can be evaded by employing an equivariant brst ( ebrst ) formalism @xcite where gauge - fixing is done only in the coset space , leaving , for example , an abelian subgroup gauge - invariant .
this may be taken as a viable alternate nonperturbative scheme for defining a non - abelian gauge theory , a proposal worthy of investigation by itself .
however , for a chiral gauge theory , the residual abelian gauge symmetry needs to be fixed in an appropriate manner . failing to do so leads to a strongly interacting sector of _
lgdof_s which is undesirable , as explained above .
in fact , because of the no - go theorem mentioned above , any brst - type symmetry can not be entertained for the abelian theory either . a naive lattice transcription of a covariant gauge fixing term results disastrously in a dense set of lattice gribov copies @xcite . to overcome this issue , shamir and
golterman @xcite proposed to add , to the standard wilson lattice gauge action for the compact @xmath0 pure gauge case , a higher - derivative ( hd ) term ( involving physical fields only ) , breaking gauge invariance explicitly .
this term , as a first requirement , leads to a covariant gauge fixing term in the naive continuum limit , and , at the same time , is designed to ensure , in the weak gauge coupling limit , a unique absolute minimum for the effective potential , thus avoiding the problem of the gribov copies and enabling weak - coupling perturbation theory ( wcpt ) around the unique vacuum .
counterterms are possible to construct because of the emergence of a renormalizable gauge , and are required to restore gauge symmetry .
wcpt analysis and numerical investigations performed earlier @xcite , only in the weak gauge coupling region of the above compact abelian pure gauge theory , confirmed the existence of a new continuous phase transition between a regular ordered phase and a spatially modulated ordered phase , for sufficiently large value of the coefficient of the hd term . at this phase transition ,
gauge symmetry is restored and the scalar fields ( _ lgdof_s ) decouple , leading to the desired emergence of massless free photons only , in the continuum limit taken from the regular broken phase . in this paper
, we explore the phase diagram of the above compact @xmath0 pure gauge theory with the hd gauge - fixing term and a suitable counterterm , _ in the strong gauge coupling region _ and present only the key findings regarding the nature of the possible continuum limits in that region .
details of our investigation will be available in @xcite .
our work is important from several points of view .
first , for both abelian and non - abelian chiral gauge theories in the nonperturbative gauge - fixing approach , it is important to know , for a large range of the gauge couplings ( including strong gauge couplings ) , that a correct continuum limit ( with the _ lgdof_s decoupled and massless free photons ) is achievable in the pure abelian gauge theory . theory .
] second , given that the pure compact @xmath0 gauge theory with the hd term , in the weak gauge coupling region , has produced a correct quantum continuum limit with free massless photons , it is important to explore a broader region of the coupling parameter space and ask what happens for strong gauge couplings .
obviously this question is linked with the issue of short distance behavior of @xmath0 gauge theory and possibility of nontrivial physics .
third , hd actions are a challenge for algorithms , especially the so - called local update algorithms .
the theory provides a good opportunity to evaluate aspects of different algorithms with large coefficients of the hd term .
lastly , because of the presence of tricritical points and critical endpoints in the phase diagram as we shall see , and phase transitions that restore a local symmetry , the investigated theory is interesting also from the point of view of statistical mechanics and critical phenomena
. the paper is organized as follows . in sec.[action ] , the lattice action under investigation for the compact @xmath0 gauge theory is presented .
the action contains a hd gauge fixing term which includes irrelevant pieces , and also a counter - term required for reviving the gauge symmetry .
section [ action ] also summarises the main results from previous investigations of the theory at weak gauge couplings .
section [ num ] presents some of the details of our numerical investigations in the parameter space of the action , including the algorithms used in our investigation .
the results of our investigation are presented in sec.[results ] which identifies the tricritical point and determines the possible continuum limits . finally , in sec.[conclu ] , we present our conclusions on the possible continuum physics from the theory investigated at strong gauge couplings .
the investigated lattice action is given by : @xmath4 where , @xmath5 is the ( gauge - invariant ) wilson term containing a summation over all gauge plaquettes @xmath6 , the plaquette being the smallest wilson loop on a @xmath7 plane .
the second term in ( [ s ] ) is the golterman - shamir hd gauge - fixing term , given by @xmath8 with the covariant laplacian @xmath9 and , @xmath10 the third term @xmath11 represents possible counterterms , and we use @xmath12 which is a dimension-2 mass counterterm , as apparent from expanding the lattice gauge field @xmath13 for small lattice spacing @xmath14 .
also possible are a host of marginal counterterms that can be treated perturbatively @xcite .
the dimension-2 counterterm is enough to give rise to a new universality class , as we shall see .
the presence of the hd term ensures a unique absolute minimum for the action at @xmath15 , validating wcpt around @xmath16 or @xmath17 and leads to the familiar covariant gauge fixing term @xmath18 in the naive continuum limit with @xmath19 .
plane at a given weak gauge coupling ( @xmath20 ) .
, height=170 ] plane at gauge coupling @xmath21 on @xmath22 lattice . ]
the action contains only physical fields ( and does not include , for example , ghosts which are expected to decouple only in the continuum limit ) and as such is not brst - invariant .
the relevant symmetry for the action ( [ s ] ) is the gauge symmetry , and @xmath23 $ ] and @xmath24 $ ] are not gauge - invariant . under a gauge transformation @xmath25 ,
these terms pick up the _ lgdof_s , and the theory becomes a scalar - gauge system with @xmath26 $ ] where mass of the scalar fields @xmath27 with @xmath28 may scale in an appropriate continuous phase transition of the lattice theory .
however , the goal here is not to have a gauge - higgs theory in the continuum , it is rather to decouple the scalar fields ( _ lgdof_s ) at a continuum limit .
the two extra terms ( with coefficients @xmath29 and @xmath30 ) ensure that in the neighborhood of the perturbative point ( i.e. , for small @xmath31 and large @xmath30 ) the _ lgdof_s are weakly coupled , and indeed numerical simulations confirm that the _ lgdof_s decouple at a new phase transition separating the regular ordered phase ( to be called fm in the following ) from a so - called spatially modulated ordered phase ( fmd ) @xcite .
numerical simulations in @xcite , were done at weak couplings ( @xmath20 ) in the so - called vector picture ( action ( [ s ] ) where no scalars appear explicitly ) and in the so - called higgs picture ( action with both scalars and gauge fields , obtained after a gauge transformation ) . for weak couplings ,
these studies confirmed a phase diagram with generic features as given in fig.[weakphasedia ] .
the nomenclature of the phases in this theory has been taken as per the phases in the so - called reduced model @xcite .
the reduced model is obtained by putting @xmath15 ( corresponding to the trivial orbit ) in the higgs picture of the theory .
the regular broken phase , fm ( with ferromagnetic order ) is characterized by a massive photon and a massive scalar , the pm ( for paramagnetic ) phase is the disordered ( symmetric ) phase having massless photons , and finally the new fmd ( ferromagnetic - directional ) phase is the spatially modulated ordered phase that breaks euclidean rotational symmetry with a nonzero vector condensate ( @xmath32 ) ( there is also an antiferromagnetic or am phase with staggered order , not to be discussed further in this study ) .
photon and scalar masses scale by approaching the continuous fm - pm transition from the fm phase , leading to a continuum gauge - higgs theory .
a sufficiently large @xmath30 ( and small @xmath31 ) ensures a satisfactory continuum limit with only the photon mass scaling ( thereby recovering gauge symmetry and decoupling the scalars ) at the fm - fmd phase transition by tuning a single parameter @xmath29 from the fm side .
given the above that this new formulation of a compact @xmath0 gauge theory on lattice produces a correct continuum limit for weak gauge couplings , it is certainly worthwhile to ask about the nature of a continuum limit , if at all , for strong gauge couplings and also explore the possibilities of a nontrivial theory .
the strong coupling region was first explored in @xcite with speculations of a few novel features . in this paper ,
a completely independent and new investigation , a more careful and precise exercise has been carried out employing new methods ( see sec.[num ] below ) . as a result ,
a clear picture of the phase diagram of the theory at strong gauge couplings has emerged . in the following , we present
some of the key findings , principal among them is the existence of fm - fmd transition even at strong gauge couplings , and tricritical points on this transition . versus @xmath29 plot on three different lattice volumes at @xmath21 and @xmath33 , showing a first order fm - fmd transition .
inset shows a histogram with a double peak structure at @xmath34 . ] versus @xmath29 plot on three different lattice volumes at @xmath21 and @xmath35 , showing a continuous fm - fmd transition .
inset shows @xmath36 versus @xmath29 plots for the same parameters . ]
multihit metropolis , a local update algorithm that was used in all previous investigations of the theory ( e.g. @xcite ) , was discarded for the current work because at large @xmath31 it produced results unstable against variation of the number of hits and also the particular order the lattice was swept .
this is understandable , since with such a hd action density , spread over quite a few lattice sites , a local algorithm is bound to struggle , especially at large @xmath30 ( at strong gauge couplings , the fm - fmd transition is obtained at larger @xmath30 ) . in this paper
, we present results of numerical simulation done with hybrid monte carlo ( hmc ) , a global algorithm , and this marks a major difference with our previous work @xcite and produces new , reliable and numerically stable results at strong gauge couplings so that we now have a better understanding of the possible continuum limit at the fm - fmd transition at strong gauge couplings .
numerical simulation was done at gauge couplings @xmath37 and also at 0.6 and 0.8 ( for comparing with available results in literature ) at a variety of lattice volumes @xmath38 and @xmath39 to determine the phase diagrams in the @xmath40 plane at each fixed gauge coupling with @xmath29-scans and @xmath30-scans having intervals as fine as @xmath41 around the interesting phase transition regions .
each run in the scans typically has 5000 hmc trajectories for thermalization , and 10000 - 30000 hmc trajectories for measurement .
integrated autocorrelation times were measured and taken into account for error estimates .
error bars of all data points , wherever not shown explicitly , are smaller than the symbols .
only a small fraction of our results are produced here , more details will be made available in @xcite .
measurements were made on lattice volume of @xmath42 ( or @xmath43 , @xmath44 for propagators ) for the plaquette energy @xmath45 the photon mass term @xmath46 the lattice version @xmath36 of the vector condensate @xmath47 ( expectation value of the modulus of @xmath48 averaged over all the directions ) , the photon propagator and also the chiral condensate with quenched kogut - susskind ( ks ) fermions .
the vector condensate @xmath36 is the order parameter for the fm - fmd transition .
in fig.[strongphasedia ] we show the phase diagram at a fixed strong coupling @xmath21 in the @xmath40 plane ( with a certain criterion to determine the transitions on a finite lattice ) .
similar phase diagrams were also obtained for @xmath49 .
contrast the above with that of fig.[weakphasedia ] .
the fm - fmd phase transition in fig.[weakphasedia ] for @xmath20 is entirely a continuous transition @xcite , while for strong @xmath31 ( fig.[strongphasedia ] ) there is a tricritical point separating a first order fm - fmd transition from a continuous transition .
figure [ strongphasedia ] shows the tricritical point for @xmath21 at @xmath50 and @xmath51 for the lattice size @xmath22 .
the location changes slightly with lattice volumes bigger than @xmath52 .
the pm - fmd transition is found to be strongly first order , and ends at a critical endpoint @xcite where the continuous fm - pm transition terminates at first order transitions .
the first order fm - fmd transition weakens gradually as @xmath30 is increased till the tricritical point where it becomes continuous .
the location of the triciritical point in the @xmath40 plane shifts to more negative @xmath29 and also to larger @xmath30 with increasing @xmath31 .
it appears from our simulation that , at a particular gauge coupling @xmath53 between @xmath54 and 1.1 , the tricritical point tends to approach the critical endpoint . for @xmath55 ,
the fm - fmd transition is fully continuous .
the rest of the plots in this paper are all at @xmath21 .
figures [ firstorder ] and [ secondorder ] plot @xmath56 , an observable similar to the entropy for the @xmath29 scan . with increasing volume , fig.[firstorder
] shows a more distinct gap at @xmath57 for @xmath33 , while fig.[secondorder ] shows no discontinuity at @xmath58 for @xmath35 .
the inset of fig.[firstorder ] shows a double peaked histogram at the critical @xmath29 , confirming the transition to be first order .
the inset of fig.[secondorder ] shows the corresponding @xmath36 versus @xmath29 plot illustrating the fm - fmd transition at @xmath59 for @xmath35 .
( first order fm - fmd ) , and consistent with zero at @xmath35 ( continuous fm - fmd ) at @xmath21 . ] and at three values of @xmath30 for which the transition is continuous .
inset shows scaling of the photon mass as fm - fmd transition is approached from fm side . ] to understand the properties of the fm phase around the tricritical point , the region was further probed with quenched ks fermions .
figure [ chiralcond ] shows , for two lattice volumes , quenched chiral condensates in the fm phase near the fm - fmd transition at @xmath60 ( where the transition is first order ) and at @xmath61 ( where the transition is continuous ) .
noisy estimator method was employed using 4 noise vectors with conjugate gradient inverter ( bi - cgstab , a more modern inverter , was also tried without any gain ) .
extrapolation to zero bare fermion mass @xmath62 was done with a phenomenological polynomial ansatz ( keeping up to quadratic terms with five lowest masses fitted ) and shows a condensate consistent with zero on the continuous side of the fm - fmd transition while clearly there is a nonzero condensate where the transition is first order .
there is a hint of nonzero chiral condensate as the tricritical point is approached from the fm side .
however , confirmation on larger volumes is required .
the inverse of photon propagator ( 2-point correlator of @xmath63 ) in momentum space was also measured and is plotted against the square of lattice momentum @xmath64 ( discrete on a finite box ) in fig.[photonprop ] for the continuous part of the fm - fmd transition , staying in the fm phase . at three @xmath30 are each at a value of @xmath29 which is at a fixed small distance away ( @xmath65 ) from the fm - fmd transition .
] inset shows a gradually vanishing photon mass ( @xmath66-intercept ) , as @xmath29 approaches @xmath67 ( @xmath68 ) for the given fixed @xmath30 ( 0.6 ) , suggesting an expected scaling of the photon mass at the transition and recovery of gauge symmetry .
the slope of the fitted straight lines , in the main figure , suggests a field renormalization constant @xmath69 that is not unity .
however , the figure shows that the slope increases with increasing @xmath30 .
it seems reasonable to expect the slope to approach unity at large @xmath30 , consistent with wcpt at @xmath16 or @xmath17 .
in addition , the continuous fm - fmd transition line at all strong @xmath31 is found in our simulations to be below but roughly parallel to the transition ( the dotted line in fig.[strongphasedia ] ) , obtained from 1-loop wcpt in @xcite .
of course , the gap between them decreases as the coupling gets smaller .
the phase diagram of the compact @xmath0 pure gauge theory in the nonperturbative gauge - fixing approach including the hd term and a mass counterterm has turned out to be somewhat more complex in the strong gauge coupling region ( @xmath70 ) .
numerical simulation is also more difficult in this region , and local update algorithms struggle , forcing us to use global update algorithms with some care .
however , after the algorithmic issues were sorted out , a clear picture of the possible continuum limits has emerged which is very relevant for all the major important issues discussed in sec.[intro ] , e.g. , both abelian and non - abelian lattice chiral gauge theories and short distance behavior of @xmath0 gauge theories .
existence of the fm - fmd transition at strong gauge couplings is confirmed .
the continuous part of this transition , away from the tricritical point , appears to produce familiar physics with free massless photons ( and _ lgdof_s decoupled ) and zero chiral condensate .
the possibility for a nontrivial continuum limit in this pure compact @xmath0 lattice gauge theory at strong gauge couplings rests on the tricritical points with a new universality class .
details of our study including results at other gauge couplings that help develop the overall picture in the strong gauge coupling region , will appear in @xcite .
based on evidences so far , the gauge - fixing scheme appears as a valid method to define a gauge theory nonperturbatively . for the non - abelian case , this involves ebrst ( gauge - fixing the coset ) and preparations
are underway to study it as well .
the work is fully supported by the department of atomic energy ( dae ) , government of india .
the authors acknowledge the computing facilities of the theory division , saha institute of nuclear physics under dae . | pure _ compact _ @xmath0 lattice gauge theory exhibits a phase transition at gauge coupling @xmath1 separating a familiar weak coupling coulomb phase , having free massless photons , from a strong coupling phase .
however , the phase transition was found to be of first order , ruling out any nontrivial theory resulting from a continuum limit from the strong coupling side . in this work ,
a compact @xmath0 lattice gauge theory is studied with addition of a dimension - two mass counterterm and a higher derivative ( hd ) term that ensures a unique vacuum and produces a covariant gauge - fixing term in the naive continuum limit . for a reasonably large coefficient of the hd term , now there exists a continuous transition from a regular ordered phase to a spatially modulated ordered phase . for weak gauge couplings , a continuum limit from the regular ordered phase results in a familiar theory consisting of free massless photons . for strong gauge couplings with @xmath2 ,
this transition changes from first order to continuous as the coefficient of the hd term is increased , resulting in tricritical points which appear to be a candidate in this theory for a possible nontrivial continuum limit . |
it has been shown that site selective fragmentation of c - h , n - h and o - h bonds in molecules are possible using electron energy as a control parameter @xcite .
this functional group dependent site selective dissociation has enormous implications from the point of view of chemical control . in order to characterize the dissociation dynamics of the negative ion resonant state and understand the similarities / dissimilarities in various molecules with the same functional group , kinetic energy and angular distribution data are needed .
having made measurements on ions from the dissociation of oh , nh and ch bond in water , ammonia and methane , we now know the dissociation characteristics in these molecular anions . to compare the dissociation dynamics in these precursor molecules with larger molecules containing similar functional groups , we looked at ions from formic acid ( - molecule with ch and oh bonds ) and propyl amine ( - molecule with ch and nh bonds ) .
the choice of these molecules is significant considering their role in the formation of biologically relevant molecules facilitated by the formation of highly reactive radicals due to radiation / charged particle interactions , in particular the resonant processes that occur with high cross section .
formic acid has been observed in several astronomical sources / objects , especially as ices .
the abundance of ices rather than gaseous hcooh is attributed to its low photo - stability @xcite .
therefore , it is important that we understand the gas phase photophysical properties / dynamics of hcooh and subsequent chemical reactions via intermediate radicals / ions producing larger molecules .
similarly , studying dea dynamics in propyl amine would be useful in understanding resonant interactions in amino acids .
we describe here our finding of similarities / dissimilarities in the dea dynamics vis - a - vis the precursor molecules of the functional groups .
the ground state geometry of neutral formic acid is planar and belongs to the symmetry group .
the electronic configuration of hcooh , in the usually accepted hartree - fock approximation is .... the two lowest unoccupied molecular orbitals are , which is a @xmath0 molecular orbital and which is the antibonding in oh analogue of orbital .
studies of excited states of using vuv absorption spectra and photoelectron spectra @xcite show the lowest energy valence transition , corresponding to a transition and observed between 5 and 7 ev .
the second valence transition , corresponds to the removal of the electron and experimentally observed close to 8.1 ev .
the third valence transition is the transition whose energy is calculated to be fairly close to the previous transition .
thus , there are three transitions very closed spaced in the 6 to 9 ev region .
one of the characteristics observed in these excitations is the lengthening of the oh bond and reduction in their stretching frequency @xmath1 .
the ion yield curve of from ( figure [ fig6.1 ] ) on dissociative electron attachment shows an asymmetric resonance structure peaking at about 7.5 ev . from the velocity images , we find three distinct scattering patterns - observed at electron energies ( i ) below 7 ev ( ii ) between 7 and 8 ev and ( iii ) above 8 ev .
we analyze the / velocity images in these three energy intervals and compare them with earlier measurements done using the same setup @xcite on deuterated acetic acid ( ) and with water and methane ( precursor molecules for oh and ch bonds ) .
ions at electron energies below 7 ev are mostly scattered in perpendicular direction to the electron beam as seen from the velocity images at 6.2 and 6.7 ev respectively ( figures [ fig6.2](a ) and ( b ) ) .
the kinetic energy distribution at these energies is shown in figure [ fig6.2](e ) .
the maximum ke observed is about 2.0 ev . the observed ions could emanate from the dissociation of either the ch bond or the oh bond in hcooh .
the bond dissociation energy of h - cooh and hcoo - h bonds are 3.43ev and 4.54 ev respectively .
electron affinity of h atom being 0.75 ev , the appearance energy for the and channels are 2.68 ev and 3.75 ev respectively .
correspondingly , the maximum kinetic energy of ions produced via these channels at incident electron energy of 6.2 ev will be close to 3.4 ev and 2.3 ev respectively .
the ke deduced from the velocity images of / show a maximum value of about 2 ev which is close to the maximum ke estimated for h- ions from the oh site i.e. 2.3 ev . the broad ke distribution shows internal excitation of the hcoo fragment .
thus , the ions observed below 7 ev are attributed to the dissociation of the oh bond in formic acid .
+ further , there exists a distinct similarity in the angular distribution pattern and velocity images of the hydride ions from formic acid , acetic acid and water at the resonance situated at 6.5 ev . in an earlier thesis work @xcite ,
measurements on deuterated acetic acid ( ) using the same setup showed similarity between the velocity images of / and /.
in all these molecules , the ions ( in the case of ) are scattered in the perpendicular direction with respect to the electron beam .
figure [ fig6.2](c ) and ( d ) shows the images of from and from at 6.5 ev and it can be seen that the angular distributions are quite similar ( figure [ fig6.2](f ) ) .
similar results were obtained from measurements on from also @xcite .
the similarity in the angular distributions indicates that there is localization of the dea process to the hydroxyl site and that the oh(od ) bond is perpendicular to the incident electron beam at the instant of attachment giving rise to the angular distribution peaking around @xmath2 .
the molecule does not seem to undergo substantial rotation in the dissociation timescale ( which is of the order of vibrational time scale ) thereby making axial recoil approximation valid at this resonance .
thus , it is the specific orientation of the oh bond undergoing dissociation rather than the whole molecule that determines the observed anisotropy in the angular distribution of ions .
this reaffirms our observation of bond orientation specific dea process as deduced from the earlier measurements on acetic acid and methanol @xcite for the resonance process at electron energies close to 6.5 ev . comparing with the vuv absorption spectrum of @xcite , the resonance process
below 7 ev appears to occur via the projectile electron attaching to the excited state giving rise to configuration of the with symmetry .
we compare the resonance process in formic acid at these electron energies with the second resonance in water and acetic acid to look for similarities in the dissociation dynamics .
the velocity images of ions at electron energies 7.2 and 7.7 ev show almost an isotropic distribution ( see figures [ fig6.3](a ) and ( b ) ) .
there is a central blob becoming more prominent at 7.7 ev .
the maximum ke is seen to be 2.5 ev . for ions coming from the ch site ,
the maximum ke expected at 7.2 ev is approximately 4.4 ev and that from the oh site will be 3.3 ev .
as the ke estimated from the dissociation of the oh bond is closer to the observed value of 2.5 ev , it may be the case again that the ions seen are coming from the oh site with vibrational excitation of the neutral hcoo fragment .
+ figures [ fig6.3](c ) and ( d ) show the velocity images of from at 7.7 ev and from at 8.5 ev and the corresponding angular plots compared with from at 7.2 ev and 7.7 ev in figure [ fig6.3](f ) . on comparing the angular distribution with and data , we find that the scattering distributions are not similar to what we see in / velocity images as an isotropic distribution .
clearly , they are not consistent .
the ke distribution of from formic acid ( figure [ fig6.3](e ) ) suggests the dissociation of the oh site .
the immediate question is whether the bond orientation specificity properties seen in the case of second resonance in water show up for hcooh too .
however , as the angular distribution of ions ( isotropic distribution ) following the dissociation of is seen to be different from that in acetic acid and water ( perpendicular distribution ) , we believe that the isotropic distribution of ions may be a result of a negative ion resonance exclusive to formic acid .
it is possible that this resonance state occurs when a projectile electron with energies in the range 7 to 8 ev attaches to the state caused by the excitation of the electron .
thus at this resonance , the orientation specificity based dissociation dynamics as seen in the first resonance process ( below 7 ev ) across the various molecules is absent .
+ + the velocity images of ions at electron energies 8.3 ev , 9.0 ev and 9.6 ev are shown in figure [ fig6.4](a ) , ( b ) and ( c ) .
there are two distinct structures seen in these images - one is the outer ring with ions scattered in the @xmath2 direction and the other one is the central blob .
the kinetic energy distribution of ions in these three images is plotted in figure [ fig6.4](g ) .
the maximum kinetic energy is found to be about 3.5 ev . at 9 ev ,
the ions from the oh and ch sites are estimated to have a maximum ke of 5.2 and 6.2 ev approximately .
thus , there is a large difference between the measured and the estimated ke values and it is not conclusive as to which bond is being dissociated even while attributing this difference to excitation of various vibration modes .
it is also possible that new dissociation channels ( other than or ) resulting via a three - body breakup or in excited electronic state appear .
thus , ke energy distribution can not single out a particular bond and suggests the possibility of both oh and ch bonds being dissociated . we compare the data with measurements done earlier on , and at similar energies and deduce the similarities / dissimilarities . figures [ fig6.4](d ) and ( e ) show the velocity images of and ions from and ( f ) and ( g ) show the velocity images from methane at 10 ev and water at 8.5 ev . comparing the images of these molecules , we see that there are two dissociation limits for the ions - the outer group and the inner group .
the outer ring seen in the images from dea to hcooh is very similar to the ions from at 9.1 ev with an angular distribution peaking at @xmath2 direction indicating that the outer ring of ions emanate from the dissociation of the oh bond .
measurements on at similar electron energies ( close to 9 ev ) @xcite also showed similar behaviour , where the constituted the outer ring and the ions formed the inner blob .
ions from at 8.5 ev have a predominantly @xmath2 distribution similar to the outer ring of ions from formic acid .
however , ions emanating from methane via the channel also have an angular distribution similar to the outer ring . although the measurements in , and at similar energies seem to show the dissociation of the hydroxyl site , the dea data in methane at similar electron energy leaves room to suspect the ch site causing the outer ring distribution of ions .
measurements on deuterated sample of formic acid ( or ) are likely to resolve the question of the bond dissociating to give the outer ring .
coming to the inner blob seen in the case , we see that such a blob is seen in the images from and both .
such a central blob was also seen in the images of ions from .
this similarity in the images leads us to conclude that the central blob is due to the ions coming from the ch site in all these molecules .
thus , electron attachment to formic acid at energies above 8 ev has two distinct dissociation limits with different angular distributions . while the outer distribution ( or higher ke ) of ions seem to involve the dissociation of h - cooh or hcoo - h bonds , the inner blob of ions are attributed to the cleavage of the c - h bond .
propyl amine ( pa ) - is a molecule of symmetry and contains a propyl group and an amino group .
while the three carbon and nitrogen atoms are in a plane , the hydrogen atoms are out of the plane . using the gaussian 03 @xcite program
, we find the ground state electronic configuration of neutral pa in hartree fock approximation to be ... and the two lowest unoccupied orbitals are and .
the molecular orbital is dominantly the lone pair on the nitrogen atom of the amino group .
we measure the ke and angular distribution of the ions produced from dea to pa and look at the dissociation dynamics vis - a - vis measurements on ammonia and methane at similar energies .
there is no literature available on the vuv absorption or on photoelectron spectra of pa to make correlations / analogies of the resonance features in our measurements with the excited states of the neutral / cation species .
ion yield on electron attachment to pa shows two structures as in figure [ fig6.5 ] .
the first one is seen with a sharp peak centered at 5.7 ev and the other one is a broad structure in the 8 to 11 ev region .
the details are described for the two resonance structures .
figures [ fig6.6](a ) , ( b ) and ( c ) show the velocity images of ions scattered at electron energies 4.7 ev , 5.7 ev and 6.7 ev across the resonance along with from at 5.5 ev ( figure [ fig6.6](d ) ) and are seen to be scattered mostly around the @xmath3 direction with finite intensity at all other angles .
the velocity image at 6.7 ev shows a central blob with very little kinetic energy and becomes dominant at higher electron energies while the outer ring diminishes .
the kinetic energy distribution of ions at the three electron energies are plotted in figure [ fig6.6](e ) and is seen to range from 0 to about 2 ev at 5.7 ev . while data on exact bond dissociation energies of pa is unavailable to deduce the maximum ke of ions from the nh and ch sites in the molecule , it is seen that the kinetic energy distribution is very similar to from dea to at 5.5 ev .
the resonance at 5.5 ev in ammonia has been described in detail in paper - iii .
the ke distribution of at 5.5 ev ( see figure 5(a ) in paper - iii ) from ranges from 0 to 1.8 ev with a peak at 1.2 ev .
this similarity in the kinetic energy distribution of ions from pa and ammonia suggests ions emanating from the dissociation of the nh bond in pa .
this is further substantiated by the comparison of angular distribution of the ions at these energies with that of from ammonia at 5.5 ev ( figure [ fig6.6](d ) ) .
figure [ fig6.6](f ) shows the angular distribution of ions from pa at 4.7 , 5.7 and 6.7 ev along with that of ions from at 5.5 ev .
we see the angular distribution curves very similar to each other thereby suggesting that the dissociation dynamics in pa and ammonia are due to similar resonance behaviour in the 4.5 to 6.5 ev region .
thus , based on this similarity in angular distribution , it is also inferred that the orientation of the nh bond is almost perpendicular to the incoming electron beam . as in the case of ammonia
, it appears that the excitation of the lone pair on the n atom in the amino group is what causes the resonance process about 5.7 ev .
since , the lone pair on n atom has a dominant contribution to the highest occupied molecular orbital - , the first resonance process can be attributed to the excitation of the electron to the lowest unoccupied orbital followed by electron attachment in this first excited state resulting in similar dissociation dynamics as in ammonia .
+ the velocity images of ceh- ions at electron energies 7.7 , 8.7 and 9.7 ev are shown in figures [ fig6.7](a ) - ( c ) along with the velocity image of from at 10 ev ( d ) .
these images are characterized by a central blob along with a faint outer ring that is isotropic .
the kinetic energy distribution of these ions is shown in figure [ fig6.7](e ) where the maximum ke observed is lower than 4 ev and the most probable energies are in the range 0 to 1.5 ev .
we do not have exact bond dissociation energy data on pa to determine the kinetic energies .
it is possible that the central blob observed could be emanating from a possible three body breakup scheme .
assuming average bond dissociation energy of 4 ev , the dissociation of either ch or nh bonds in the molecule is expected to give a maximum ke close to 5 ev .
it is also possible that the excess energy goes into the excitation of the neutral fragment causing further dissociation .
in such a case , lower values of the kinetic energy are observed .
an instantaneous 3 body breakup can also account for the low ke energies .
thus , the ke plot in figure [ fig6.7](e ) suggests occurrence of a three body fragmentation along with two body breakup process if any .
+ having argued in favour of a three body fragmentation scheme , it appears that the propyl group is fragmented as against the amino group .
measurements on ammonia at similar energies show ions produced with high kinetic energies from a two body breakup channel ( ) and with a strong backward scattering . whereas , dissociation of methane negative ion close to 10 ev occurs via a two body as well as a three body channel . while in methane
, this central blob resolves with increase in electron energy giving a forward - backward distribution , such a feature is not seen in pa .
the fact that the central blob does nt resolve into a characteristic anisotropic distribution could be because of strong internal excitation of the propyl group where more vibrational modes are available for soaking up the excess energy .
thus , a particular dissociation limit may not be favoured over the many available ones . alternatively
, the dissociation could be a two step process where the propyl group attaches an electron and is in an excited state which then undergoes dissociation again giving rise to low kinetic energies of and also the isotropic distribution .
thus , it is inferred that electron attachment to pa in the 8 - 10 ev region arises from fragmentation of the ch sites with isotropic distribution of the scattered ions . while the electron attachment is understood to be from the ch site
, the isotropic distribution of ions does nt give information on the bond orientation at the instant of electron attachment .
strong internal excitations could change the geometry of the molecular anion prior to dissociation and this renders it difficult to infer the bond orientation .
in conclusion , the measurements across the two resonances in pa indicate bond orientation specific dissociation dynamics from the first resonance process peaking at 5.5 ev akin to that in ammonia as deduced from the close similarity in the velocity images . while the data from second resonance process around 9 ev appear to have features similar to what was seen methane .
the very low kinetic energy of the fragments does not allow us to identify signatures of bond orientation specificity in the latter case .
1 . momentum distribution from formic acid and propyl amine are similar to precursor molecules ( water and methane ) of the constituent functional groups . 2 .
dissociation dynamics independent of overall symmetry of the molecule - local symmetry of functional group / bond orientation determining factor .
functional group dependence manifests in angular distribution too as seen above . *
oh bond close to 7 ev * ch bond close to 10 ev * nh bond close to 5.5 ev 4 .
bond orientation rather than the orientation of the whole molecule significant in determining the electron attachment and the angular distribution even in molecule of lower symmetry .
w. a. schutte , a. c. a. boogert , a. g. g. m. tielens , d. c. b. whittet , p. a. gerakines , j. e. char , p. ehrenfreund , j. m. greenberg , e. f. van dishoek , th
. de graauw , _ astron .
_ , * 343 * , 966 ( 1999 ) . | in this paper , we discuss the dissociative electron attachment process in formic acid and propyl amine .
these are molecules containing more than one functional group and have low symmetry ( group ) .
we measured the kinetic energy and angular distributions of fragment ions from the resonances observed in these molecules and compared with that in the precursor molecules , namely - water , ammonia and methane .
measurements suggest that the dissociation dynamics in bigger molecules are independent of overall symmetry of the molecule , rather depend only on the local symmetry of functional group and bond orientation factors . |
although core - collapse supernovae ( sne ) present a wide range of spectral and photometric properties , there is growing consensus that much of this variety is due to the state of the progenitor star s hydrogen and helium envelopes at the time of explosion . those stars with massive , intact envelopes produce sne of type ii - plateau ( sne ii - p ; ii due to the presence of h in the spectrum , and `` plateau '' due to the characteristic shape of the light curve ) , those that have lost their entire hydrogen envelope ( perhaps through stellar winds or mass transfer to a companion ) result in sne ib ( type i because of the absence of obvious hydrogen in their spectra ) , and those that have been stripped of both hydrogen and most ( or all ) of their helium produce sne ic ; see filippenko ( 1997 ) for a general review .
recently , a new subclass of objects has emerged whose members generically resemble sne ic ( no hydrogen or obvious helium spectral features ) , but , unlike traditional sne ic , have spectra characterized by unusually broad features at early times , indicating velocities in excess of @xmath8 km s@xmath6 .
a few also possess inferred kinetic energies exceeding that of `` normal '' core - collapse sne by more than a factor of 10 ( see , e.g. , nomoto et al .
these objects are colloquially referred to as `` hypernovae , '' although not all of them are clearly more luminous or energetic than normal sne ic .
there are currently 5 generally accepted members of this rare class : sn 1997dq , sn 1997ef , sn 1998bw , and , most recently , sn 2002ap and sn 2002bl ( see , e.g. , iwamoto et al .
2000 ; matheson et al . 2000b ; filippenko , leonard , & moran 2002 ; mazzali et al .
2002 ; kawabata et al .
. a related subclass of sne exhibits many of the characteristics of these objects , but with hydrogen present in the spectra ; the clearest examples are sn 1997cy and sn 1999e ( see filippenko 2001 , and references therein ) , and they , too , are sometimes called hypernovae .
intense interest in hypernovae has been sparked not only by their peculiar spectral features , but also by the strong spatial and temporal association between the brightest and most energetic of these events , sn 1998bw , and the @xmath9-ray burst ( grb ) 980425 ( e.g. , galama et al .
this potential association has fueled the proposition that some ( or , perhaps all ) core - collapse sne explode due to the action of a `` bipolar '' jet of material ( wheeler , meier , & wilson 2002 ; khokhlov et al .
1999 ; macfadyen & woosley 1999 ) , as opposed to the conventional neutrino - driven mechanism ( colgate & white 1966 ; burrows et al . 2000 , and references therein ) . under this paradigm ,
a grb is only produced by those few events in which the progenitor has lost all of its outer envelope material ( i.e. , it is a `` bare core '' collapsing ) , and is only observed if the jet is closely aligned with our line - of - sight ( l - o - s ) .
with such an explosion mechanism , one can well imagine that severe distortions from spherical symmetry may exist in the ejecta .
since all hot , young sn atmospheres are dominated by electron scattering ( wagoner 1981 ) , which is an inherently polarizing process , spectropolarimetry provides a powerful probe of young sn morphology .
the basic question is whether the sn is round : by symmetry , if the atmosphere is spherical the directional components of the electric vectors cancel to produce zero net polarization .
in contrast , an asymmetric atmosphere will yield a nonzero polarization due to incomplete cancellation . from recent polarimetry of core - collapse events ,
a tentative trend has been identified : the degree of polarization ( and , hence , asymmetry ) increases with decreasing progenitor envelope mass ( wheeler 2000 ) . indeed , while the intrinsic polarizations of `` normal '' sne ii - p are generally below @xmath10 ( leonard et al .
2001 ; leonard & filippenko 2001 ) , a polarization of over 4% has been reported for an sn ic , suggesting an axial asymmetry of more than 3:1 in this event ( wang et al .
the number of events investigated in detail with spectropolarimetry , however , remains small .
an additional technique that has been used to infer explosion asymmetry is the analysis of nebular line profiles .
recent simulations involving explosive nucleosynthesis in aspherical , jet - induced sn explosions predict that intermediate - mass and heavy elements such as iron are ejected ( at high velocity ) primarily along the poles whereas elements synthesized in the progenitor ( e.g. , he , c , ca , o ) are preferentially located at lower velocities near the equatorial plane in the expanding ejecta ( maeda et al .
2002 ; khokhlov & hflich 2001 ; hflich et al .
maeda et al .
( 2002 ) model the effect that explosion asymmetry has on the spectra of hypernova explosions , and compare the results to the observed nebular line profiles of sn 1998bw .
the main observable diagnostic for explosion asymmetry is found to be the ratio of the width of a probable [ ] blend near 5200 to that of [ ] @xmath116300 , 6364 .
maeda et al . ( 2002 ) find models of spherical explosions to be incapable of yielding velocity ratios greater than @xmath12 3:2 , whereas aspherical models can generate ratios of more than 2:1 .
although a specific degree of asphericity is not given , maeda et al .
( 2002 ) find the nebular - phase line - width ratio of [ ] to [ ] for sn 1998bw to be inconsistent with a spherical explosion . in this paper
, we report on spectropolarimetric observations obtained for the recent peculiar type ic sn 2002ap in m74 ( kinugasa et al .
2002 ; filippenko & chornock 2002 ; mazzali et al .
2002 ; gal - yam , ofek , & shemmer 2002 ) during the photospheric phase , and two nebular - phase optical flux spectra .
we describe the observations in [
sec : observations ] , present our results and analysis in
[ sec : resultsandanalysis ] , and discuss our conclusions in [ sec : conclusions ] .
additional details of the observations and analysis are given in appendix a. as this paper was nearing completion , a submitted paper by kawabata et al .
( 2002 ) describing similar spectropolarimetric data appeared on astro - ph ; we briefly compare our results with that work in
[ sec : conclusions ] .
we obtained spectropolarimetry of sn 2002ap on 2002 february 14 and march 7 ( 16 and 37 days , respectively , after the time of explosion derived by mazzali et al .
2002 ) , with the low - resolution imaging spectrometer ( oke et al .
1995 ) in polarimetry mode ( lrisp ; cohen 1996 ) at the cassegrain focus of the keck - i 10-m telescope .
after spending nearly three months in solar conjunction , sn 2002ap emerged from behind the sun in june , and we obtained optical flux spectra of it on 2002 june 8 and june 17 ( 131 and 140 days , respectively , after the time of explosion ) with the kast double spectrograph ( miller & stone 1993 ) at the cassegrain focus of the shane 3-m telescope at lick observatory . on the second june epoch ,
spectropolarimetry of five distant galactic stars near to the l - o - s of sn 2002ap was also obtained in order to quantify the galactic interstellar polarization ( isp ) component contributing to the observed polarization of sn 2002ap .
( due to the very limited time of visibility , spectropolarimetry of sn 2002ap itself was not obtained on this date . )
the slit position angle was set to within @xmath13 of the parallactic angle ( filippenko 1982 ) for all observations in order to minimize differential light loss , and the total - flux spectra from all epochs were corrected for continuum atmospheric extinction and telluric absorption bands ( e.g. , wade & horne 1988 ; bessell 1999 ) .
a journal of observations is given in table 1 .
the polarimetry data were reduced according to the methods outlined by miller , robinson , & goodrich ( 1988 ) and detailed by leonard et al .
( 2001 ) and leonard & filippenko ( 2001 ) .
additional details of the observations and analysis , including an investigation of the potential impact of second - order light and instrumental polarization on our results , are given in appendix a. the polarization angle ( p.a . )
offset between the half - wave plate and the sky coordinate system was determined for all polarimetry epochs ( at both keck and lick ) by observing the polarized standard star bd @xmath14 and setting its @xmath15-band polarization position angle ( i.e. , @xmath16 , the debiased , flux - weighted average of the polarization angle over the wavelength range 50505950 ; see leonard et al . 2001 ) equal to @xmath17 , the value cataloged by schmidt , elston , & lupie ( 1992 ) . to check the precision of the p.a
. offset at keck , we also observed the polarized standard star hd 127769 on both nights , and found its p.a . to agree to within @xmath18 of the values measured by clemens & tapia ( 1990 ) and hiltner ( 1956 ) . since the values individually reported by the two catalogs disagree by @xmath19 in p.a . and @xmath20 in polarization level for this standard star ,
we do not ascribe much significance to the disagreement found between our measurements and the cataloged values .
it is possible that this star exhibits polarization variability .
the observed null standard star , hd 57702 ( mathewson & ford 1970 ) , was found to be null to within @xmath21 on both nights at keck .
( see appendix a for details of the small amount of instrumental polarization that may have existed in one of our settings during the february 14 epoch . ) at lick , the p.a .
offset was checked by observing hd 204827 ; its p.a .
, measured to be @xmath22 , differs by only @xmath23 from the value cataloged by schmidt et al .
the null standards observed at lick , bd + @xmath24 and hd 212311 , were both found to be null to within @xmath25 .
the results for sn 2002ap are shown in figures [ fig:1 ] and [ fig:2 ] . before estimating and then removing the isp ( [ sec : removalofisp ] ) , we make a few general remarks about the observed polarization .
the nature of the polarization of sn 2002ap changed significantly between our two observations . on day 16 ,
the observed polarization rises from @xmath26 near 4000 to @xmath27 at 10,000 , with the general rise punctuated by some features that may be associated with strong features seen in the total - flux spectrum .
the p.a . also changes by about @xmath28 across the spectrum .
the notable feature in the polarization data from day 37 is an enormous polarization change associated with the p - cygni profile of the near - infrared ( ir ) triplet .
the polarization changes between the two observational epochs , as well as the strong modulations across spectral lines , indicate that some of the observed polarization is intrinsic to sn 2002ap , and is not interstellar in nature . in order to progress with further interpretation , however , we must attempt to estimate and remove the isp from our data . from reddening considerations we do not expect a large isp , since any dust that polarizes the sn light should also redden it , and sn 2002ap is not highly reddened : @xmath29 mag ( schlegel , finkbeiner , & davis 1998 ) and @xmath30 mag ( klose , guenther , & woitas 2002 ) , for a total estimated reddening of @xmath31 mag . since the empirical galactic limit for the maximum observed polarization for a given reddening
is @xmath32 ( serkowski , mathewson , & ford 1975 ; see , however , leonard et al .
2002a for a possible violation of this inequality for dust in an external galaxy ) , this implies an upper limit to the total ( galactic plus host galaxy ) isp of only @xmath33 . one way to estimate
the actual total isp level is to rely on the data obtained for sn 2002ap itself , along with some theoretical assumptions .
figure [ fig:3 ] shows the observed polarization data in the @xmath34-@xmath35 plane for both epochs .
the majority of the data from each epoch appears to scatter about straight , but different , lines in the @xmath34-@xmath35 plane .
such behavior is indicative of an axisymmetric asphericity in which the intrinsic p.a .
is wavelength independent , a trait that has been identified in previous sn polarization studies ( e.g. , howell et al .
2001 ; leonard et al .
if axisymmetry is suspected , then the isp must lie along the best - fitting line to the points in the @xmath34-@xmath35 plane ; since the dominant axis for the two epochs prefer different p.a.s ( @xmath36 on days 16 and 37 , respectively ) , the intersection of the two lines provides an estimate of the isp . from figure [ fig:3 ]
we see that this occurs near @xmath37 , or @xmath38 . since isp varies with wavelength according to a `` serkowski '' curve ( whittet et al .
1992 ) , it has a wavelength dependence ; for our purposes , we take the derived isp location to represent the value at @xmath39 , the average wavelength of maximum isp found from studies of galactic stars . a second method to estimate
the total isp is to assume that some spectral region is intrinsically _ completely _ unpolarized , with all of the _ observed _ polarization therefore coming from isp .
there is both theoretical and empirical evidence that this may be the case in the ir emission profile on day 37 . since directional information is lost as photons are absorbed and reemitted in a line ( hflich et al .
1996 ) , broad , unblended emission features have traditionally been treated as intrinsically unpolarized in sn studies . in their spectropolarimetric study of an early - time , emission - dominated spectrum of sn 1998s , leonard et al .
( 2000 ) demonstrate that the basic paradigm of polarized continuum light being diluted by unpolarized emission lines is quite consistent with that data : the degree of depolarization across the broad - line features of sn 1998s changes in direct proportion to strength of the line at each wavelength across the profile . in the case of the profile of sn 2002ap , however , we see that this simple conceptual scheme fails , since the _ entire _ emission component , and even some of the absorption component , is polarized at roughly the same level ( fig .
[ fig:2 ] ) ; such behavior is also seen in the h@xmath40 profiles of the type ii - p sn 1999em ( leonard et al .
one possible interpretation of this is that not only are the emission - line photons unpolarized , but the underlying continuum photons are as well , having been absorbed and reemitted by the optically thick line .
such a situation is also described by howell et al .
( 2001 ) for an sn ia .
if we accept this premise , and estimate the isp from a 200 wide bin centered on the emission peak at 8578 , we derive @xmath41 .
fitting a serkowski law isp curve ( whittet et al .
1992 ) to this value with an assumed wavelength of maximum polarization of @xmath42 , we derive @xmath43 at @xmath44 , or @xmath45 . this result is quite similar to that derived earlier by the @xmath34-@xmath35 plane method , and is also consistent with the value kawabata et al .
( 2002 ) found using a similar technique ( @xmath46 ) .
the consistency between our two independent total isp estimates affords some confidence that we have determined an accurate value . taking the average of the two results yields an estimate of the total isp of @xmath47 , or @xmath48 at @xmath49 . since the reddening from milky way ( mw ) dust
is thought to dominate the total reddening of sn 2002ap , we would expect galactic isp to make up the bulk of the total isp . to check this assertion
, we obtained spectropolarimetry of five distant galactic stars near to the l - o - s of the object ( the catalog of heiles 2000 records no polarization measurements of any galactic star within @xmath50 of the l - o - s to sn 2002ap , the nominal angular separation over which some consistency of isp can be assumed ) , and list the results in table 2 .
although the values have some scatter , they are generally consistent with each other ( they all fall in the same quadrant of the @xmath34-@xmath35 plane ) and confirm a non - negligible polarization in this area of the sky from galactic dust . calculating the mean and @xmath51 scatter of the individual values , we obtain @xmath52 , which yields @xmath53 at @xmath54 .
this is to be compared with the total isp estimated by the earlier techniques of @xmath55 .
the general agreement of all of these values affirms our contention that galactic isp makes up the bulk of the total isp , and also strengthens our belief that the total isp estimated by the earlier techniques must be nearly correct .
we therefore adopt @xmath48 at @xmath56 for @xmath42 , and proceed to remove this isp from our data . may be somewhat less than the nominal value of @xmath42 . given the inherent difficulty of accurately measuring such low polarization values ( see , e.g. , leonard et al .
2001 ) , however , we choose to use the nominal value of @xmath42 to correct our data . ]
the resulting isp - corrected , intrinsic sn polarization and p.a . are displayed in figure [ fig:4 ] .
the outstanding feature of the intrinsic polarization on day 16 is its strong increase with wavelength , from @xmath57 near 4000 to @xmath58 at the cutoff near 10,000 .
such a precipitous rise in polarization at optical and near - infrared wavelengths is completely unlike the serkowski law dependence of isp , and must be intrinsic to the sn .
a similar redward rise in intrinsic polarization was seen in spectropolarimetry of the type ia sn 1999by , where it was attributed to the decreasing importance of line opacities , and the increased influence of continuum electron scattering at longer wavelengths ( howell et al .
in particular , numerous lines , largely due to iron - group elements , dominate the spectrum at @xmath59 and significantly depolarize the light , whereas the red end of the spectrum suffers less severe line - blending , allowing the intrinsically polarized continuum light to dominate .
we suspect that a similar effect may be operating here , and that the high polarization ( @xmath60 ) measured at the red end of the spectrum in fact represents a lower limit on the true level of continuum polarization undiluted by unpolarized line photons at this epoch . a detail of the region near the strong polarization increase associated with the absorption trough of the ir triplet on day 37
is shown in figure [ fig:5 ] . in the absorption trough
there is an apparent change in observed polarization of @xmath61 , where @xmath62 and @xmath63 are the individual changes in the normalized stokes parameters compared with the nearby continuum value ( note that @xmath64 is an _ isp - independent _ quantity ; see leonard & filippenko 2001 ) .
it is possible to interpret this sharp polarization increase within the basic model first proposed by mccall ( 1984 ) and later extended by jeffery ( 1991 ) , that polarization peaks are naturally associated with absorption minima due to selective blocking of forward - scattered ( and hence less polarized ) light in p - cygni troughs .
with this simple model in mind , leonard & filippenko ( 2001 ) showed that a lower bound could be placed on the true intrinsic continuum polarization level of @xmath65 where @xmath66 is defined by equation [ eqn:1 ] , @xmath67 equals the interpolated value of the continuum flux at the location of the line trough , and @xmath68 is the total flux at the line s flux minimum .
the main difficulty in the application of this to sn 2002ap is interpolating the continuum flux across the profile ; our adopted estimate of the continuum level is indicated in figure [ fig:5]a . with @xmath69 and @xmath70 , we find that @xmath71 for this epoch .
we therefore see evidence for significant intrinsic continuum polarization on both day 16 and day 37 after explosion .
when modeled in terms of the oblate , electron - scattering atmospheres of hflich ( 1991 ) , the inferred continuum polarization demands an asphericity of at least @xmath2 on day 16 and @xmath3 on day 37 ; the actual value naturally depends on the ( unknown ) viewing orientation .
it is also possible that ionization asymmetry or clumping in the ejecta may contribute to the polarization , which would reduce the inferred bulk asymmetry ( see , e.g. , hflich , khokhlov , & wang 2001 ) .
detailed theoretical modeling of both the total flux and polarization data may set constraints on the viewing orientation ( e.g. , hflich , wheeler , & wang 1999 ) , and also serve to check the specific assumptions that underlie our estimates of the intrinsic continuum polarization .
figure [ fig:6 ] shows the total flux spectrum of sn 2002ap obtained after the sn reappeared following solar conjunction ; we have combined the observations taken 131 and 140 days after the explosion in order to improve the signal - to - noise ratio .
the spectrum is characterized by strong emission profiles superimposed on a very weak continuum , implying that the sn has entered the nebular phase ( maeda et al .
2002 ; sollerman et al . 2000 ) .
the greater strength of the lines relative to the continuum level in the spectrum of sn 2002ap compared with other sne ic at similar , although slightly earlier epochs ( fig . 6 ) , is consistent with the rapid early spectral evolution reported by mazzali et al . ( 2002 ) , who found that sn 2002ap evolved at about twice the rate of sn 1997ef .
early - time spectra of sn 2002ap showed very broad spectral features , indicating high velocities in the outer ejecta ( @xmath8 km s@xmath6 ; mazzali et al .
2002 ) . given the extreme breadth of the early - time spectral features , one might expect the nebular - phase broad - line widths ( taken as the fwhm of a gaussian fit to the continuum - subtracted line profile ; see matheson et al .
2001 ) for sn 2002ap to be greater than those measured in other nebular sne ic .
however , for [ ] @xmath116300 , 6364 we measure 8600 km s@xmath6 , consistent with the average value ( @xmath72 km s@xmath6 ) of 9 sne ic reported by matheson et al .
( 2001 ) , and for [ ] @xmath117291 , 7324 , we measure 9800 km s@xmath6 , also consistent with the matheson et al .
( 2001 ) value ( @xmath73 km s@xmath6 ) . at this stage
, there is evidently not significantly more line emission coming from high - velocity gas in this `` hypernova '' than is typical for `` ordinary '' sne ic .
similarly normal expansion velocities were also found at late times for the hypernova sn 1998bw ( patat et al .
2001 ) . in addiion , the identification of spectral features extending out to nearly @xmath8 km s@xmath6 in very early - time spectra of the type ii - p sn 1999gi ( leonard et al .
2002b ) suggests that the presence of high - velocity material may be common among core - collapse sne , but is only easily identified if very early - time spectra are obtained .
figure [ fig:7 ] shows the [ ] @xmath116300 , 6364 and ] @xmath744571 line profiles for day 140 ( shown instead of the day 131 spectrum due to its superior resolution ) .
a striking feature of both profiles is the very sharp peaks to the flux distribution , which may indicate the presence of a narrow - line component , redshifted by about 580 km s@xmath6 with respect to the systemic velocity of the host galaxy m74 . given the complex nature of the profiles , it is not certain whether these narrow features are resolved in our spectrum , which has a resolution of @xmath75 km s@xmath6 at 6300 .
patat et al . (
2001 ) observed redshifted emission peaks in spectra of sn 1998bw at a similar phase .
this emission could come from an excess of gas existing in lower - velocity regions close to the center of the sn nebula .
it is also possible that it is coming from material previously lost by the progenitor star and now existing ( in clumps ? ) in the circumstellar environment of the sn ( e.g. , sn 1999cq ; matheson et al .
. continued monitoring at sufficient spectral resolution is urged in order to trace the development of these narrow features .
given the spectropolarimetric evidence for asphericity during the photospheric phase ( [ sec : intrinsicpolarization ] ) , it is worthwhile to investigate whether the nebular - phase line profiles show any evidence for asymmetry in the line - emitting regions at these later epochs . as discussed in [ sec : introduction ] , the main spectral diagnostic of asymmetry was found by maeda et al .
( 2002 ) to be a large ratio ( @xmath76 ) between the width of the [ ] blend near 5200 and that observed in [ ] @xmath116300 , 6364 ( see fig .
[ fig:6 ] for the identification of these features ) .
one difficulty with this diagnostic , however , is the complex nature of the [ ] blend s profile , which might include contributions from other lines ( e.g. , [ ] @xmath77 ; [ ] @xmath78 ) that can substantially widen the observed profile , especially during the early nebular phase .
it is also difficult to estimate the `` continuum '' underlying this feature , since it evidently drops significantly across the profile . in their study , maeda et al .
( 2002 ) analyzed the feature in a nebular spectrum of sn 1998bw taken 201 days after @xmath79-band maximum , in which they felt confident that no extraneous lines contributed to the profile .
given the rapid spectral development of sn 2002ap , it is possible that this region in our spectrum , from @xmath80 days after @xmath79 maximum , is similarly uncontaminated .
detailed spectral models are needed to confirm this assumption , but they are beyond the scope of this paper . if we assume that our feature is the same as that observed in sn 1998bw at the later epoch , then we derive a velocity ratio of @xmath12 2.8:1 for the features .
this value is similar to that found for sn 1998bw , which would therefore imply an aspherical explosion for sn 2002ap in the same way that it was inferred for sn 1998bw .
however , the relatively early phase of our spectrum , the difficulty inherent in measuring the width of the [ ] feature , and the potential for significant line blending make us view this preliminary result with some degree of skepticism .
there is strong evidence from spectropolarimetry and tentative evidence from nebular - phase line profiles that sn 2002ap is not spherical .
it is useful to compare the spectropolarimetry of sn 2002ap with the small but growing number of core - collapse events for which polarimetry measurements exist .
the lower bounds on the continuum polarization level of sn 2002ap ( @xmath81 and @xmath82 on days 16 and 37 , respectively ) are larger than those typically derived for sne ii - p ( wheeler 2000 ) , but not nearly as great as that inferred for the sn ic 1997x ( wang et al .
might be due to line - feature modulations similar to what is seen in the ir triplet of sn 2002ap during the second epoch , and not to intrinsically polarized continuum light . ] a more direct comparison to a similar event can be made with the spectropolarimetry presented by patat et al .
( 2001 ) for the hypernova sn 1998bw .
patat et al . (
2001 ) present spectropolarimetry from two epochs , @xmath83 and + 10 days after @xmath79-band maximum ( our observations of sn 2002ap occurred on days 7 and 28 relative to the date of @xmath79 maximum derived by gal - yam et al .
2002 ) with coverage from 4000 to 6900 .
although the isp is unknown for this event , pc , which is insufficient to fully sample the dust in the galactic plane . ]
we note that the overall level of the observed polarization ( @xmath84 ) , and the marked increase in the polarization level from blue to red wavelengths ( particularly in the earliest epoch ) , are quite similar to the behavior of sn 2002ap .
in addition to the high inferred continuum polarization , another interesting finding for sn 2002ap is that while the data from both epochs suggest a substantial degree of axisymmetry ( fig .
[ fig:3 ] ) , the dominant axes have quite different polarization angles , changing from @xmath85 on day 16 to @xmath86 on day 37 ( [ sec : removalofisp ] ) .
this behavior is different from that observed in other core - collapse events , most notably those resulting from progenitors with massive envelopes intact at the time of explosion . for sn 1999em , a classic type ii - p event , leonard et al .
( 2001 ) show that a common , wavelength - independent p.a .
exists for all observational epochs , which includes days 12 , 45 , 54 , 164 , and 168 after explosion .
this suggests a more complex morphology for the thinning ejecta of sn 2002ap . as mentioned in
[ sec : introduction ] , a potentially relevant result of recent simulations involving explosive nucleosynthesis in aspherical , jet - induced sn explosions is that intermediate - mass and heavy elements ( e.g. , fe ) as well as freshly synthesized radioactive material such as @xmath87ni are ejected ( at high velocity ) primarily along the poles whereas elements synthesized in the progenitor ( e.g. , he , c , ca , o ) are preferentially located at lower velocities near the equatorial plane in the expanding ejecta ( maeda et al . 2002 ; khokhlov & hflich 2001 ; hflich et al .
although the p.a . in the isp - corrected data from day 37 ( bottom , right panel of fig
. [ fig:4 ] ) is difficult to define in some regions due to the low polarization level , there is evidently a change near @xmath88 . below this wavelength , @xmath89 , a value similar to the one that characterizes the day 16 data . above @xmath90
, @xmath91 tends to dominate ; this is especially easy to see ( and believe , due to the high polarization ) in the ir absorption trough ( see also fig .
[ fig:3 ] ) .
might this change in p.a .
be due to the different spatial distributions of the iron - group elements , which dominate at blue wavelengths , from the elements synthesized by the progenitors , such as ca and o , that are responsible for much of the opacity at red wavelengths ( see fig .
[ fig:2]a ) ? along these same lines , an explanation for the nearly constant p.a . on day 16 at all wavelengths , and its similarity to the p.a . of day 37 at blue wavelengths , could be due to the photosphere at early times being located in the highest - velocity ( i.e. , polar ) material . by day 37 ,
perhaps the sn has cooled sufficiently to expose the inner , slower - moving equatorial regions , and we are witnessing the competition between these two different distributions of material . to be sure , a small change in the isp estimate , or even the wavelength of maximum polarization of the isp , could alter the inferred p.a . of the continuum in the second epoch considerably .
further interpretation of the polarization change near @xmath92 in the second epoch is probably not warranted .
the @xmath93 p.a .
difference between the the day 16 data and the region associated with the ir trough on day 37 , however , remains regardless of the isp choice , and may well indicate a true difference in the distribution of ca relative to iron - group elements .
just prior to completion of this research , a paper describing similar data appeared on astro - ph by kawabata et al .
although the observational epochs are slightly different ( kawabata et al .
present data from 2002 february 9.211.3 and march 8.210.2 ) , the general results are similar : an observed polarization level in the february epochs of @xmath94 that may rise somewhat toward the red ( the increase is not as obvious as it is in our data since their data does not extend beyond 8300 ) and a dramatic polarization increase in the ir triplet s absorption trough followed by depolarization in the emission profile in the march data . one interesting speculation made by kawabata et al .
( 2002 ) is that much of the polarized continuum at early times may result from scattering off of electrons in a jet , or bipolar jets , of material emitted from the sn during the explosion .
supporting this hypothesis is the general similarity demonstrated between the intrinsic polarized flux ( i.e. , @xmath95 ) and the total flux spectrum redshifted by @xmath96 , the presumed speed of the jet .
we also find qualitative agreement between the intrinsic polarized flux and the total flux redshifted by @xmath97 during our february epoch ( fig .
[ fig:8 ] ) , although we have not investigated how significant the correlation is .
we do note , though , that the speculation by kawabata et al .
( 2002 ) is consistent with the scenario proposed above that the polarization p.a . on day 16
is dominated by the distribution of the polar jet material . countering the jet hypothesis , however , is the recent assertion by berger et al .
( 2002 ) that the extreme faintness of sn 2002ap at radio wavelengths is inconsistent with the presence of a jet , regardless of the viewing geometry . in any event , additional spectropolarimetry of young sne ic is certainly warranted in order to further test the jet model .
the total flux spectra obtained in the early - nebular phase confirm that sn 2002ap continues the rapid spectral evolution previously witnessed during the photospheric phase by mazzali et al .
it is characterized by a very weak continuum and broad emission lines , which demonstrate that it is already well along the transition to the fully nebular phase .
the lines have widths similar to those observed in `` normal '' nebular sne ic .
there are , however , unusual narrow lines superimposed on some of the broad - line profiles , including especially those of [ ] @xmath116300 , 6364 and ] @xmath744571 .
it is possible that the ratio of the width of the [ ] blend near 5200 compared with that of [ ] @xmath116300 , 6364 is indicative of asymmetry in the line - emitting region , but additional , later - time spectra are needed to confirm this result .
some of the data presented herein were obtained at the w. m. keck observatory , which is operated as a scientific partnership among the california institute of technology , the university of california , and the national aeronautics and space administration .
the observatory was made possible by the generous financial support of the w. m. keck foundation .
this research has made use of the nasa / ipac extragalactic database ( ned ) , which is operated by the jet propulsion laboratory , california institute of technology , under contract with nasa , as well as the simbad database , operated at cds , strasbourg , france .
this research was supported in part by nasa through the american astronomical society s small research grant program .
our work was also funded by nasa grants go-8648 , go-9114 , and go-9155 from the space telescope science institute , which is operated by aura , inc . , under nasa contract nas 5 - 26555 .
additional funding was provided to a.v.f . by nsf grant ast-9987438 .
we thank a. j. barth and e. c. moran for help with some of the observations , w. li for useful discussions , and an anonymous referee for helpful suggestions that resulted in an improved manuscript .
the assistance of the staffs at lick and keck observatories is greatly appreciated .
the observational approach that we have taken at keck in the past for our program to study sn geometry has been to obtain spectropolarimetry with a 600 groove mm@xmath6 grating blazed at 5000 , which yields a spectral range of about 43506850 in the red arm of lris ( leonard et al .
2000 ; leonard et al . 2001
; leonard & filippenko 2001 ) .
the observing procedure of sn 2002ap on february 14 and march 7 differed in two ways from our traditional approach .
first , in an effort to take advantage of the extended coverage possible with the recently commissioned blue arm of the lris double spectrograph ( lris - b ; an overview of the lris - b instrument is available at http://alamoana.keck.hawaii.edu/inst/lris/lrisb.html ) , on february 14 we used the d680 dichroic filter to split the spectrum at 6800 and send the blue and red light to the respective arms of lris ( hereafter referred to as the `` uv '' and `` ir '' settings , respectively ) . to our knowledge , these are the first spectropolarimetric observations to be published using lris - b .
second , on both nights we observed sn 2002ap using the red side alone ( i.e. , no dichroic ) with a 300 groove mm@xmath6 grating blazed at 5000 , which yielded coverage in the range 39308830 ( hereafter referred to as the `` opt '' setting ) .
the lack of an order - blocking filter in this setting makes second - order light contamination beyond @xmath98 a concern . in this appendix
, we discuss the performance of lris - b for spectropolarimetry and the impact of second - order light on the data acquired with the opt setting . for the uv / ir setup ,
a grism with 300 groove mm@xmath6 blazed at @xmath99 dispersed the light on the blue side , and a grating with 400 groove mm@xmath6 blazed at @xmath100 dispersed the light on the red side ; an og570 order - blocking filter was also inserted on the red side . in principle , such a setup yields coverage down to the atmospheric cutoff at @xmath123200 .
however , we found the calibration of the spectropolarimetry to be problematic at blue wavelengths . at these wavelengths the poor efficiency of the blue ccd chip and grism ( see http://alamoana.keck.hawaii.edu/inst/lris/ccdblue_qe.html and + http://alamoana.keck.hawaii.edu/inst/lris/grism_list.html ) , along with the inefficiency of the `` hn-22 uv / optical polaroid '' polarizing filter , which is used to make the wavelength - dependent position angle correction curve ( see miller et al . 1988 ;
ogle et al .
1999 ) , probably contribute to the difficulty we encountered . when we attempted to calibrate the polarization standard bd @xmath14 ( schmidt et al . 1992 ) , strong , artificial p.a .
rotations were evident below @xmath101 , indicating that the derived p.a .
correction curve was not appropriate at these wavelengths .
the data were also extremely noisy , especially for red objects .
fortunately , a new polarizer for the blue side that should allow calibration of polarization further into the blue has recently been installed and is currently being tested . in the end , we concluded that reliable spectropolarimetry on the blue side was possible for sn 2002ap down to 3800 .
the total flux spectrum was calibratable down to 3400 .
an additional difficulty with the uv setting is the known problem of `` ghost '' images of second - order dispersed red - side light appearing on the blue side .
although the intensity of the `` ghost '' images is reported to be no more than @xmath102 of the red - side intensity , such contamination , if variable between observations , could corrupt spectropolarimetry results in the uv setting .
these uncertainties , as well as some previous problems that we have had calibrating spectropolarimetry data obtained with a dichroic ( see , e.g. , leonard et al .
2001 ) , led us to approach the lris - b spectropolarimetry with some caution .
the opt setting fortunately provides a check on the uv data from this first epoch . in a similar way ,
the ir setting provides a check on possible second - order light contamination effects at red wavelengths in the opt setting . on the red side ,
the `` ir800 ir '' polarizer , which is used to make the wavelength - dependent position angle correction curve , does not produce @xmath103% polarized light at wavelengths beyond @xmath104 .
however , unlike the case on the blue side , consistent results ( i.e. , no unusual p.a . or polarization level changes ) were obtained for all objects and standards out to @xmath105 m , indicating that the derived p.a .
correction curve is accurate .
this builds confidence in the calibration of the spectropolarimetry of sn 2002ap at red wavelengths as recorded by the ir setting . to check for instrumental polarization ,
the null standard hd 57702 ( mathewson & ford 1970 ) was observed on february 14 with all three settings ( fig .
[ fig:9 ] ) .
the polarization measured with both the opt and ir settings are in good agreement and effectively null . also impressive
is the apparently good calibration of the ir polarization through the region affected by the dichroic cutoff .
similarly promising results were also obtained for the polarization standard bd @xmath14 ( schmidt et al .
1992 ) and other science objects observed on this night .
although the signal - to - noise ratio decreases sharply for both uv and ir near 6800 , it may be possible to trust the polarization data in this region with no gap in coverage .
the comparison between the uv and opt settings for hd 57702 , however , reveals that a small ( @xmath106 ) amount of instrumental polarization likely exists in the uv setting .
this is also supported by similar discrepancies between the uv and opt settings in other objects data on this night . for hd 57702 , mathewson & ford ( 1970 ) list @xmath107 . from the opt setting data we measure @xmath108 ( statistical error only ) , while the uv setting yields @xmath109 .
correcting for instrumental polarization at the @xmath110 level is usually of dubious merit , since the intrinsic polarization of null standards is generally of this order .
nonetheless , since the opt setting value is closer to the cataloged value , and we have never found significant instrumental polarization to be evident on the red side of lris , we attempted to correct for this small degree of instrumental polarization in the uv data . to do this , we first derived the offsets in the raw stokes parameters between the uv and opt measurements of hd 57702 , and then used these offsets to correct the other objects uv data ( see leonard et al .
we found that this correction consistently improved the agreement between the two settings , although in some cases the agreement between the two settings was still not perfect . since improvement was seen with the correction , though , we adopted it and applied it to the uv data obtained for sn 2002ap .
the ultimate cause of this small amount of instrumental polarization in our uv setting is unknown .
although spectropolarimetry is routinely obtained at keck using settings nearly identical to our opt setting ( e.g. , brotherton et al . 2000 ; ogle et al .
1999 ) , this was the first time that we personally had obtained data significantly beyond 7000 without the use of an order - blocking filter .
this lead to some concern over the possibility of second - order light contaminating the data beyond @xmath98 .
the degree to which second - order light affects red - side data has been investigated in detail by putney & cohen ( 1996 ) for the 300 groove mm@xmath6 grating in the red camera , and they show that the percentage of blue light appearing as second - order red light in the range 64009000 is at most @xmath111 of the incident blue flux . furthermore , from observations of polarized stars made with and without an order - blocking filter in place , putney & cohen ( 1996 ) conclude that second - order light effects have negligible impact on polarimetry data , except perhaps for objects with highly wavelength - variable polarization ( potentially a concern for sn 2002ap ) , and/or very blue flux .
the redundancy of having the opt data beyond @xmath112 also covered by the ir setting , in which an order - blocking filter was used ( in addition to the dichroic ) , allows us to directly test the conclusions of putney & cohen ( 1996 ) for sn 2002ap in the feb .
figure [ fig:10 ] shows the polarimetry data for sn 2002ap obtained on this night in all three settings .
the agreement at red wavelengths is indeed quite good , and suggests that second - order light effects are not a concern . at blue wavelengths
, there is relatively good agreement between opt and uv , although some discrepancies in the polarization level at the @xmath113 level do exist .
given the concerns with the uv data , the demonstrated lack of second - order light effects at red wavelengths in the opt setting , and the very limited availability of sn 2002ap at the second epoch ( it was approaching solar conjunction and visible for less than 1 hour immediately after sunset ) ,
we decided to observe sn 2002ap using only the opt setting on march 7 .
thus , direct comparison between these data and those for which an order - blocking filter was used is not possible for this epoch .
however , we can estimate the fraction of the observed red flux contributed by second - order blue light through a procedure similar to that described by putney & cohen ( 1996 ) , and compare this with the amount from the first epoch , for which second - order effects were shown to be negligible .
since the second - order light effects derived by putney & cohen ( 1996 ) were not derived using observations made with the polarimetry optics in the light - path , we determined our own values by observing the blue flux standard feige 34 ( stone 1977 ; massey & gronwall 1990 ) in the opt setting , both with and without the gg495 order - blocking filter , as well as with the uv / ir setup .
all observations were taken with the achromatic half - wave plate at a position angle of @xmath114 , which produces two parallel , perpendicularly polarized beams on the ccd chip .
the beams were individually flux calibrated and then combined to yield the total flux spectrum .
following the procedure of putney & cohen ( 1996 ) , we calculate the fraction of blue light appearing as second - order contamination in the red end as where @xmath116 is the flux - calibrated data measured from 64008828 in the opt setting without an order - blocking filter in place , @xmath117 is the flux - calibrated data in the same region observed with an order - blocking filter , and @xmath118 is the flux - calibrated data measured below @xmath119 .
we then smoothed @xmath120 and interpolated over telluric features ( e.g. , the a - band ) to produce the result shown in figure [ fig:11 ] , which is , in fact , quite similar to the result derived by putney & cohen ( 1996 ) .
for both the february and march epochs , and show the results in figure [ fig:12 ] .
( flux - calibrated data were available for sn 2002ap down to @xmath122 on february 14 and @xmath123 on march 7 ; see table 1 . )
it is clear that second - order light does not make an appreciably larger contribution to the march 7 data than it did on february 14 . in both epoch
s data , the contribution is less than 1.3% of the observed flux .
this similarity , especially when coupled with the smooth , non - variable nature of the polarization in the range @xmath123@xmath124 ( fig . [ fig:2 ] ) , implies similarly negligible second - order light effects in the march data , at least at wavelengths beyond 7800 , which includes the important ir region .
additional confidence that second - order light does not compromise the data from the march epoch is gained from the favorable comparison with the spectropolarimetry data that kawabata et al .
( 2002 ) obtained at nearly the same epoch , presumably without any second - order light contamination .
we therefore conclude that second - order light contamination has had a negligible impact on our data for both epochs , and concur with the conclusions of putney & cohen ( 1996 ) that second - order light is not a concern for the spectropolarimetry of objects that do not have either highly wavelength - variable polarization at blue wavelengths or extremely blue color . | the early - time optical spectrum of the type ic supernova ( sn ) 2002ap was characterized by unusually broad features , leading some authors to designate it a hypernova .
" we present optical spectropolarimetry of this object 16 and 37 days after the estimated date of explosion . after correcting for interstellar polarization
, we find evidence for a high level of intrinsic continuum polarization at both epochs : @xmath0 on day 16 and @xmath1 on day 37 .
prominent line polarization is also seen , especially in the trough of the ca ii near - infrared triplet during the second epoch . when interpreted in terms of the oblate , electron - scattering model atmospheres of hflich ( 1991 )
, our results imply an asymmetry of at least @xmath2 ( day 16 ) and @xmath3 ( day 37 ) .
the data suggest a fair degree of axisymmetry , although the polarization angle of the dominant polarization axes are different by about @xmath4 between the two epochs , implying a complex morphology for the thinning ejecta .
in particular , there exists some spectropolarimetric evidence for a different distribution of ca relative to iron - group elements .
we also present flux spectra of sn 2002ap taken 131 and 140 days after the explosion .
the spectra are characterized by a very weak continuum and broad emission lines , indicating that sn 2002ap has entered the nebular phase .
the spectral features are substantially similar to those of `` normal '' sne ic in the nebular phase , and the emission lines are not significantly broader .
however , some of the broad lines are very sharply peaked , and may possess a narrow component ( probably unresolved by our spectra , fwhm @xmath5 km s@xmath6 ) that is redshifted by @xmath7 km s@xmath6 with respect to the systemic velocity of the host galaxy . |
the problem of inferring interactions couplings in complex systems arises from the huge quantity of empirical data that are being made available in many fields of science and from the difficulty of making systematic measurements on interactions . in biology , for example
, empirical data on neurons populations , small molecules , proteins and genetic interactions , have largely outgrown the understanding of the underlying system mechanisms . in all these cases the inverse problem , whose aim is to infer some effective model from the empirical data with just partial a priori knowledge , is of course extremely relevant . statistical physics , with its set of well - understood theoretical models , has the crucial role to provide complex , but clear - cut benchmarks ,
whose direct solution is known and that can therefore be used to develop and test new inference methods . in a nutshell , the equilibrium approach to inverse problems consists inferring some information about a system defined through an energy function @xmath0 starting from a set of sampled equilibrium configurations .
suppose @xmath1 i.i.d sampled configurations @xmath2 generated by a boltzmann distribution of an unknown energy function @xmath3 , @xmath4 are given .
the posterior distribution of @xmath3 ( also called likelihood ) is given by @xmath5 , where @xmath6 represents the average of @xmath3 over the given sample configurations and @xmath7 the prior knowledge about @xmath3 .
the parameter @xmath1 plays the role of an inverse temperature : when @xmath1 is very large , @xmath8 peaks on the maximums ( with respect to @xmath3 ) of the log - likelihood @xmath9 .
the problem of identifying the maximum of @xmath10 can be thought of as an optimization problem , normally very difficult both because the space of @xmath3 is large and because @xmath11 is very difficult to estimate by itself on a candidate solution .
several methodological advances and stimulating preliminary applications in neuroscience have been put forward in the last few years @xcite .
still the field presents several major conceptual and methodological open problems related to both the efficiency and the accuracy of the methods .
one problem which we consider here is how to perform inference when data are not coming from a uniform sampling over the equilibrium configurations of a system but rather they are taken from a subset of all the attractive states .
this case arises for instance when we consider systems with multiple attractors and we want to reconstruct the interactions couplings from measurements coming from a subset of the attractors . in what follows ,
we take as model system the hopfield model over random graphs in its memory phase ( i.e. with multiple equilibrium states ) and show how the interaction couplings can be inferred from data taken from a subset of memories ( states ) .
this will be done by employing the bethe equations ( normally used in the ergodic phase where they are asymptotically exact ) by taking advantage of a certain property of their multiple fixed points in the non - ergodic phase .
the method can be used to infer both couplings and external local fields .
we also show how from the inference method one can derive a simple unsupervised learning protocol which is able to learn patterns in presence of week and highly fluctuating input signals , without ever reaching a spin glass like saturation regime in which all the memories are lost .
the technique that we will discuss is based on the so called cavity method and leads to a distributive algorithmic implementation generically known as message passing scheme .
the paper is organized as follows .
first , in section [ s2 ] we define the problem , and make connections with related works .
section [ s3 ] is concerned with the inference problem in non - ergodic regimes , for which a simple algorithmic approach is presented . in section [ s4 ]
we apply the technique to the finite connectivity hopfield model in the memory phase .
section [ s5 ] shows how the approach can be turned into an unsupervised learning protocol .
conclusions and perspectives are given in section [ s6 ] .
the hopfield model is a simple neural network model with pair - wise interactions which behaves as an attractor associative memory @xcite .
its phase diagram is known exactly when memories are random patterns and the model is defined over either fully connected or sparse graphs @xcite .
reconstructing interactions in the hopfield model from partial data thus represents a natural benchmark problem for tools which pretend to be applied to data coming from multi electrode measurements from large collections of neurons .
the underlying idea is that a statistically consistent interacting model ( like the hopfield model ) inferred from the data could capture some aspects of the system which are not easy to grasp from the raw data @xcite .
here we limit our analysis to artificial data . in the hopfield model
the couplings @xmath12 are given by the covariance matrix of a set of random patterns which represent the memories to be stored in the system .
we will use the model to generate data through sampling and we will aim at inferring the couplings .
the structure of the phase space of the hopfield model at sufficiently low temperature and for a not too large number of patterns is divided into clusters of configurations which are highly correlated with the patterns .
we will proceed by sampling configurations from a subset of clusters and try to infer the interactions .
the simple observation that we want to exploit is the fact that fluctuations within single clusters are heavily influenced by the existence of other clusters and thus contain information about the total system .
we consider a system of @xmath13 binary neurons @xmath14 ( or ising spins ) interacting over a random regular graph of degree @xmath15 ; that is every node has a fixed number @xmath15 of neighbors which are selected randomly .
the connectivity pattern is defined by the elements @xmath16 of the adjacency matrix .
the ( symmetric ) interactions between two neighboring neurons are given by the hebb rule , i.e. @xmath17 , where @xmath18 are the patterns to be memorized and @xmath19 is their number . at finite temperature @xmath20 , we simulate the system by a glauber dynamics @xcite ; starting from an initial configuration , the spins are chosen in a random sequential way and flipped with the following transition probability @xmath21 where @xmath22 is the local field experienced by spin @xmath23 and @xmath24 is an external field .
we use @xmath25 to denote the set of neighbors interacting with @xmath23 .
the process satisfies detailed balance and at equilibrium the probability of steady state configurations @xmath26 is given by the gibbs measure @xmath27}e^{\beta \sum_i \theta_i \sigma_i+\beta \sum_{i < j } j_{ij}\sigma_i \sigma_j},\ ] ] where @xmath28 $ ] is a normalization constant , or partition function . in the memory phase , the system will explore configurations close to a pattern , provided that the initial configuration lies in the basin of attraction of that pattern . in the following we use the wording patterns , basins of attraction or states equivalently . in a given state the average activity ( or magnetization ) and correlations are denoted by @xmath29 and @xmath30 where the averages are taken with the gibbs measure inside that state .
informally , a gibbs state corresponds to a stationary state of the system , and so defined by the average macroscopic quantities in that state .
suppose that starting from some random initial configuration we observe the system for a long time , measuring @xmath1 configurations .
the standard way to infer interactions couplings @xmath31 , and external fields @xmath32 is by maximizing the log - likelihood of @xmath33 , given the experimental data @xcite , namely @xmath34,\end{aligned}\ ] ] where @xmath35 and @xmath36 are the experimental magnetizations and correlations and @xmath37 is the free energy .
one can exploit the concavity of the log - likelihood and use a gradient ascent algorithm to find the unique parameters maximizing the function .
however , this needs an efficient algorithm to compute derivatives of the free energy @xmath38 $ ] , which in general is a difficult task .
a well known technique which can be used for not too big systems is of course the monte carlo method ( see e.g. @xcite ) .
though under certain limiting assumptions , there exist good approximation techniques which are efficient , namely mean field , small - correlation and large - field expansions @xcite .
in this paper we resort to the mean - field cavity method , or belief propagation ( bp ) , to compute the log - likelihood ( see e.g. @xcite ) .
this technique is closely related to the thouless - anderson - palmer ( tap ) approximation in spin glass literature @xcite .
the approximation is exact on tree graphs and asymptotically correct as long as the graph is locally tree - like or the correlations between variables are sufficiently weak . in spin glass jargon ,
the approximation works well in the so called replica symmetric phase . in the bp approach ,
the marginals of variables and their joint probability distribution ( which is assumed to take a factorized form where only pair correlations are kept ) are estimated by solving a set of self - consistency functional equations , by exchanging messages along the edge of the interaction graph ( see ref .
@xcite for comprehensive review ) . a message ( typically called `` cavity belief '' )
@xmath39 is the probability that spin @xmath23 takes state @xmath40 ignoring the interaction with its neighbor @xmath41 , i.e. in a cavity graph .
we call this probability distribution a cavity message . assuming a tree interaction graph we can write an equation for @xmath39 relating it to other cavity messages @xmath42 sent to @xmath23 : @xmath43 as in cavity graphs the neighboring variables are independent of each other .
these are bp equations and can be used even in loopy graphs to estimate the local marginals .
the equations are solved by starting from random initial values for the cavity messages and updating them in some random sequential order till a fixed point is reached . upon convergence
the cavity messages are used to obtain the local marginals or `` beliefs '' : @xmath44 these marginals are enough to compute the magnetizations @xmath45 and correlations @xmath46 and thus can be used for maximizing the log - likelihood by updating the parameters as @xmath47 with @xmath48 and positive . repeating this procedure for sufficient times leads to an estimate of the unknown parameters . assuming that the external fields are absent , the inference error can be written as : @xmath49 a more accurate estimate of the correlations can be obtained by exploiting the fluctuation - response theorem @xmath50 .
this method , called susceptibility propagation @xcite , uses derivatives of cavity messages ( cavity susceptibilities ) .
its time complexity grows as @xmath51 , to be compared with the @xmath52 complexity of bp equations . in this paper
we will work exclusively with the bp estimate which is simple and accurate enough for our studies .
actually , if one is interested only on correlations along the edges of a sparse graph , the bp estimation would be as good as the one obtained by susceptibility propagation .
the reader can find more on the susceptibility propagation in @xcite .
in an ergodic phase a system visits all configuration space . sampling for a long time
is well represented by the measure in ( [ gibbs ] ) . in a non - ergodic phase , as happens for the hopfield model in the memory phase , the steady state of a system is determined by the initial conditions .
starting from a configuration close to pattern @xmath53 , the system spends most of its time ( depending on the size of system ) in that state .
we indicate the probability measure which describes such a situation by @xmath54 , that is the gibbs measure restricted to state @xmath53 . if configurations are sampled from one state , then the expression for the log - likelihood in ( [ logl ] )
should be corrected by replacing @xmath55 with @xmath56 , the free energy of state @xmath53 .
still the log - likelihood is a concave function of its arguments and so there is a unique solution to this problem .
it is well known that the bethe approximation is asymptotically exact in the ergodic phase ( @xcite ) . in this case , the gibbs weight can be approximately expressed in terms of one- and two- point marginals @xmath57 , @xmath58 as follows : @xmath59 the above equation is exact only asymptotically ( on a replica - symmetric system ) ; it can be used for inference in at least two ways : the simplest one is by replacing @xmath57 and @xmath58 in the above expression by their experimental estimation ( given as input of the inference process ) , equating ( [ eq : pbethe ] ) to ( [ gibbs ] ) and solving for @xmath31 and @xmath32 .
this is known as the `` independent pairs '' approximation .
a second one , often more precise but computationally more involved , is to search for a set of @xmath60 and a corresponding @xmath60-fixed point of bp equations , such that the bethe estimation @xmath61 , @xmath62 of the two- and one - point marginals match the experimental input as accurately as possible . in a non - ergodic phase ,
it is known however that bp equations typically do not converge or have multiple fixed points .
this is normally attributed to the fact that the bp hypothesis of decorrelation of cavity marginals fails to be true .
when a bp fixed point is attained , it is believed to approximate marginals inside a single state ( and not the full gibbs probability ) , as the decorrelation hypothesis are satisfied once statistics are restricted to this state @xcite .
the fact that bp solutions correspond to restriction to subsets of the original measure may suggest that there is little hope in exploiting ( [ eq : pbethe ] ) on such systems .
fortunately , this is not the case . for every finite system , and every bp fixed point @xmath53 the following holds ,
@xmath63 } \prod_i \pi^\alpha_i(\sigma_i ) \prod_{i < j}\frac{\pi^\alpha_{ij}(\sigma_i,\sigma_j)}{\pi^\alpha_i(\sigma_i)\pi^\alpha_j(\sigma_j)}. \label{eq : pbethe2}\ ] ] a proof of a more general statement will be given in appendix [ app - bp ] .
as in the ergodic case , ( [ eq : pbethe2 ] ) can be exploited in at least two ways : one is by replacing @xmath64 and @xmath65 by their experimental estimation inside a state and solving for @xmath60 the identity between ( [ eq : pbethe2 ] ) and ( [ gibbs ] ) , exactly like in the independent pairs approximation , as if one just forgets that the samples come from a single ergodic component .
a second one is by inducing bp equations to converge on fixed points corresponding to appropriate ergodic components . in this paper
we will take the latter option .
please notice that the second method , as an algorithm , is more flexible with respect to the first one ; indeed , there is no reason to have a bp fixed point for any set of experimental data , especially when the number of samples is not too large .
it means that matching exactly the data with those of a bp fixed point is not always possible .
therefore , a better strategy would be to find a good bp solution which is close enough to the experimental data . ignoring the information that our samples come from a single ergodic component would result in a large inference error due to the maximization of the wrong likelihood .
as an example , we take a tree graph with ising spins interacting through random couplings @xmath66 , in zero external fields @xmath67 .
choose an arbitrary pattern @xmath68 and fix a fraction @xmath69 of the boundary spins to the values in @xmath68 .
for @xmath70 the system would be in paramagnetic phase for any finite @xmath71 , therefore the average overlap of the internal spins with the pattern would be zero .
on the other hand , for @xmath72 and low temperatures the overlap would be greater than zero , as expected from a localized gibbs state around pattern @xmath68 . in this case the observed magnetizations are nonzero and without any information about the boundary condition we may attribute these magnetizations to external fields which in turn result to a large inference error in the couplings .
equivalently , we could put the boundary spins free but restrict the spin configurations to a subspace , for instance a sphere of radius @xmath73 centered at pattern @xmath68 in the configuration space @xmath74 .
that is , the system follows the following measure : @xmath75 where @xmath76 is an indicator function which selects configurations in the subspace @xmath74 . by the bp approximation
we can compute the magnetizations @xmath77 and the correlations @xmath78 , see appendix [ app - bpd ] for more details
. taking these as experimental data , we may perform the inference by assuming that our data represent the whole configuration space .
this again would result to a large inference error ( for the same reason mentioned before ) whereas taking into account that the system is limited to @xmath74 , we are able to infer the right parameters by maximizing the correct likelihood ; i.e. replacing the total free energy @xmath55 in the log - likelihood with @xmath79 , the free energy associated to subspace @xmath74 . in figure
@xmath80 we display the inference error obtained by ignoring the prior information in the above two cases .
notice that in principle the error would be zero if we knew @xmath74 and the boundary condition .
as it is seen in the figure , the error remains nonzero when the boundary spins are fixed ( @xmath72 ) even if sampling is performed over the whole space .
inference error on a cayley tree when the leaves are free ( @xmath70 ) or fixed ( @xmath72 ) to random configuration @xmath68 .
the data come from subspace @xmath74 ( a sphere of radius @xmath73 centered at @xmath81 ) . in the inference algorithm
we ignore the boundary condition and that samples are restricted to @xmath74 .
the internal nodes have degree @xmath82 and size of the tree is @xmath83.,width=377 ]
the hopfield model can be found in three different phases . for large temperatures
the system is in the paramagnetic phase where , in the absence of external fields , magnetizations @xmath45 and overlaps @xmath84 are zero . if the number of patterns is smaller than the critical value @xmath85 , for small temperatures the system enters the memory phase where the overlap between the patterns and the configurations belonging to states selected by the initial conditions can be nonzero . for @xmath86 , the hopfield model at low temperature
enters in a spin glass phase , where the overlaps are typically zero . in fully connected graphs @xmath87 and in random poissonian graphs @xmath88 where @xmath89 is the average degree @xcite .
take the hopfield model with zero external fields and in the memory phase .
we measure samples from a glauber dynamics which starts from a configuration close to a pattern @xmath90
. the system will stay for a long time in the state @xmath90 and is thus well described by the restricted gibbs measure @xmath91 . in a configuration @xmath26 , the local field seen by neuron @xmath23 is @xmath92 . if @xmath90 corresponds to the retrieved pattern , the first term ( signal ) would have the dominant contribution to @xmath93 . the last term ( noise )
is a contribution of the other patterns to the local field . to exploit this information
, we look for a set of couplings that result to a gibbs state equivalent to the observed state of the system . one way to do
this is by introducing an auxiliary external field pointing to the experimental magnetizations , i.e. @xmath94 , for a positive @xmath95 ; we may set the couplings @xmath12 at the beginning to zero and compute our estimate of the correlations @xmath46 by the bp algorithm .
this can be used to update the couplings by a small amount in the direction that maximizes the likelihood , as in ( [ update - j ] ) ( we do not update the external fields which for simplicity are assumed to be zero ) .
this updating is repeated iteratively , decreasing @xmath95 by a small amount in each step .
the procedure ends when @xmath95 reaches the value zero .
the auxiliary field is introduced only to induce convergence of the equations towards a fixed point giving statistics inside a particular state .
figure [ f2 ] compares the inference error obtained with the above procedure for several values of temperature in the memory phase . for the parameters in the figure , the inferred couplings from one basin were enough to recover the other two patterns . in the figure
we also see how the error decreases by taking larger number of samples from the system .
inference error versus number of samples for different temperatures .
the data are extracted from one pure state of the hopfield model in the memory phase ( @xmath96 ) .
size of the system is @xmath97 , each spin interacts with @xmath98 other randomly selected spins , and number of stored patterns is @xmath99 . in the inference algorithm we use @xmath100.,width=377 ] in general we may have samples from different states of a system .
let us assume that in the hopfield model we stored @xmath19 patterns by the hebb rule but the samples are from @xmath101 basins .
the estimated correlations @xmath102 in any state @xmath103 should be as close as possible to the experimental values @xmath104 .
a natural generalization of the previous algorithm is the following : as before we introduce external fields @xmath105 for each state @xmath106 . at fixed positive @xmath95
we compute the estimated bp correlations for different states .
each of these estimations can be used to update the couplings as in the single state case
. specifically , this amounts to make a single additive update to the couplings by the average vector @xmath107 given by @xmath108 , where @xmath109 and @xmath110 . indeed , the addends of @xmath111 will be typically linearly independent , so @xmath112 will imply @xmath113 for @xmath114 .
we then decrease @xmath95 and do the bp computation and update steps .
again we have to repeat these steps until the external field goes to zero .
figures [ f3 ] and [ f4 ] show how the inference error changes with sampling from different states .
notice that if we had an algorithm that returns exact correlations , an infinite number of samples from one state would be enough to infer the right interactions in the thermodynamic limit .
however , given that we are limited by the number of samples , the learning process is more efficient if this number is taken from different states instead of just one .
evolution of the inference error with update iterations .
the data obtained by sampling from one or several pure states of the hopfield model in the memory phase ( @xmath96 ) .
the total number of samples in each case is @xmath115 .
size of the system is @xmath97 , each spin interacts with @xmath98 other randomly selected spins , and number of stored patterns is @xmath99 . in the inference algorithm we use @xmath100.,width=377 ] inference error and number of states which are stable and highly correlated with the patterns after sampling from @xmath101 pure states of the hopfield model in the memory phase ( @xmath96 ) .
size of the system is @xmath97 , each spin interacts with @xmath116 other randomly selected spins , and number of stored patterns is @xmath117 . in the inference algorithm we use @xmath100 and number of samples
is @xmath118.,width=377 ]
hebbian learning is a stylized way of representing learning processes . among the many oversimplifications that it involves
there is the fact that patterns are assumed to be presented to the networks through very strong biasing signals . on the contrary it is of biological interest to consider the opposite limit where only weak signals are allowed and retrieval takes place with sizable amount of errors . in the spin language we are thus interested in the case in which the system is only slightly biased toward the patterns during the learning phase . in what follows
we show that one can `` invert '' the inference process discussed in the previous sections and define a local learning rule which copes efficiently with this problem . as first step
we consider a learning protocol in which the patterns are presented sequentially and in random order to the system by applying an external field in direction of the pattern , that is a field @xmath119 with @xmath120 .
we assume that initially all couplings are zero .
depending on the strength of the field , the system will be forced to explore configurations at different overlaps with the presented pattern @xmath106 .
a small @xmath95 corresponds to a weak or noisy learning whereas for large @xmath95 the system has to remain very close to the pattern .
what is a small or large @xmath95 , of course depends on the temperature and strength of the couplings .
here we assumed the couplings are initially zero , so @xmath121 defines the boundary between weak and strong fields .
the learning algorithm should indeed force the system to follow a behavior that is suggested by the auxiliary field .
therefore , it seems reasonable if we try to match the correlations in the two cases : in absence and presence of the field .
notice to the similarities and differences with the first part of the study . as before we are to update the couplings according to deviations in the correlations .
but , here the auxiliary field is necessary for the learning ; without that the couplings would not be updated anymore .
moreover , it is obvious that we can not match exactly the correlations in absence and presence of an external field .
we just push the system for a while towards one of the patterns to reach a stationary state in which all the patterns are remembered . for any @xmath95 we can compute the correlations @xmath122 by either sampling from the glauber dynamics or by directly running bp , with external fields @xmath119 . at the same time
we can compute correlations @xmath102 by the bp algorithm in zero external fields and with initial messages corresponding to pattern @xmath106 .
then we try to find couplings which match the correlations in the two cases , namely we update the couplings by a quantity @xmath123 .
the process is repeated for all couplings and for @xmath124 iterations with the same pattern @xmath106 .
next we switch to some other randomly selected pattern @xmath90 and the whole process is repeated for @xmath125 learning steps .
notice that here @xmath95 is fixed from the beginning .
the above learning protocol displays a range of interesting phenomena .
firstly one notices that for @xmath126 ( i.e. for very large external fields ) and @xmath127 ( i.e. for very high temperature or isolated neurons ) the above learning results to the hebb couplings of the hopfield model . in figure [ f5 ]
we compare the histogram of learned couplings for small and large @xmath95 with the hebbian ones . comparing the histogram of hebbian couplings with that of learned couplings for small and large values of the external field .
size of the system is @xmath97 , each spin interacts with @xmath128 other randomly selected spins . here
we are to store @xmath117 random and uncorrelated patterns . in the learning algorithm
we use @xmath96 and @xmath129.,width=377 ] the number of patterns @xmath130 which are highly correlated with stable configurations depends on the strength of external fields .
we consider that pattern @xmath106 is `` learned '' if there is a gibbs state with nonzero overlap @xmath131 that is definitely larger than the other ones @xmath132 .
figures [ f6 ] and [ f7 ] show how these quantities evolve during the learning process and by increasing the magnitude of external filed .
evolution of the average overlap and fraction of successfully learned patterns in the learning algorithm .
size of the system is @xmath97 , each spin interacts with @xmath128 other randomly selected spins . in the learning algorithm we set @xmath133 , @xmath96 , @xmath129 , and number of patterns that are to store is @xmath117.,width=377 ] average number of successfully learned patterns in the learning algorithm for different values of @xmath95 .
size of the system is @xmath97 , each spin interacts with @xmath134 other randomly selected spins .
the learning algorithm works at @xmath135 , and number of patterns that are to store is @xmath136 .
the average is taken over @xmath137 realizations of the patterns.,width=377 ] for small @xmath19 nearly all patterns are learned , whereas , for larger @xmath19 some patterns are missing .
a large number of patters can thus be learned at the price of smaller overlaps and weaker states .
that is , the average overlap in successfully learned patterns decreases continuously by increasing @xmath19 , approaching the paramagnetic limit . in figure [ f8 ]
we compare this behavior with that of hebb couplings .
as the figure shows , there is a main qualitative difference between hebbian learning of the hofield model and the protocol discussed here . in the former case when the number of stored patterns exceeds some critical value the systems enters in a spin glass phase where all memories are lost and the bp algorithm does not converge anymore . on the contrary , in our case many patterns can be stored without ever entering the spin glass phase ( for a wide range of choices of @xmath95 ) .
the bp algorithm always converges , possibly to a wrong fixed point if the corresponding pattern is not stored .
average number of successfully learned patterns in the learning algorithm and hebb rule versus @xmath19 , the number of patterns that are to store .
the inset shows the average overlap .
size of the system is @xmath97 , each spin interacts with @xmath134 other randomly selected spins .
the learning algorithm works at @xmath138 . the average is taken over @xmath137 realizations of the patterns.,width=377 ] population dynamics
is usually used to obtain the asymptotic and average behavior of quantities that obey a set of deterministic or stochastic equations @xcite .
for instance , to obtain the phase diagram of the hopfield model with population dynamics one introduces a population of @xmath139 messages representing the bp cavity messages in a reference state , e.g. pattern @xmath140 @xcite .
then one updates the messages in the population according to the bp equations : at each time step a randomly selected cavity message is replaced with a new one computed by @xmath141 randomly selected ones appearing on the r.h.s . of the bp equations .
in each update , one generates the random couplings @xmath142 by sampling the other @xmath143 random patterns .
after a sufficiently large number of updates , one can compute the average overlap with the condensed pattern @xmath90 to check if the system is in a memory phase .
the stability of condensed state would depend on the stability of the above dynamics with respect to small noises in the cavity messages .
if @xmath139 is large enough , one obtains the phase diagram of hopfield model in the thermodynamic limit averaged over the ensemble of random regular graphs and patterns . we used the above population dynamics to obtain the phase diagram of the hopfield model on random regular graphs ,
see figure [ f9 ] .
the phase diagram of hopfield model on random regular graphs of degree @xmath98 obtained with population dynamics ( @xmath144 ) .
horizontal axes is number of patterns @xmath19 and vertical axes is temperature @xmath20 .
the paramagnetic , memory and spin glass phases are labeled with @xmath145 and @xmath146 , respectively.,width=377 ] in order to study the new learning protocol we need a more sophisticated population dynamics .
the reason is that in contrast to hebb couplings , we do not know in advance the learned couplings . in appendix [ app - pop ]
we explain in more details the population dynamics that we use to analyze the learning process studied in this paper . the algorithm is based on @xmath19 populations of bp messages and one population of couplings .
these populations represent the probability distributions of bp messages in different states and couplings over the interaction graph . for a fixed set of patterns
@xmath147 we update the populations according to the bp equations and the learning rule , to reach a steady state .
figure [ f10 ] displays the histogram of couplings obtained in this way . in the figure we compare two cases of bounded and unbounded couplings . in the first case
the couplings should have a magnitude less than or equal to @xmath148 whereas in the second case they are free to take larger values .
we observe a clear difference between the two cases ; when @xmath95 is small , the couplings are nearly clipped in the bounded case whereas the unbounded couplings go beyond @xmath149 . however , in both cases there is some structure in the range of small couplings . increasing the magnitude of @xmath95 we get more and more structured couplings .
for very large fields they are similar to the hebb couplings . for small fields
the histogram of the couplings is very different from the hebb one , though the sign of the learned and the hebbian couplings is the same .
there are a few comments to mention here ; in the population dynamics we do not have a fixed graph structure and to distinguish @xmath19 patterns from each other we have to fix them at the beginning of the algorithm .
moreover , we have to modify the bp equations to ensure that populations are representing the given patterns , see appendix [ app - pop ] . and
finally the outcome would be an average over the ensemble of random regular graphs , for a fixed set of patterns .
having the stationary population of couplings , one can check the stability of each state by checking the stability of the bp equations at the corresponding fixed point .
the maximum capacity that we obtain in this way for the learned couplings is the same as the hebb one whereas on single instances we could store much larger number of patterns .
the reason why we do not observe this phenomenon in the population dynamics resides in the way that we are checking stability ; the fixed patterns should be stable in the ensemble of random regular graphs .
in other words , checking for stability in the population dynamics is stronger than checking it in a specific graph .
the main result of our analysis consists in showing that the distribution of the couplings arising from the bp learning protocol is definitely different from the hebbian one .
the histogram of learned couplings obtained by the population dynamics in random regular graphs of degree @xmath128 .
number of patterns that are to store is @xmath117 . in the upper panel
we compare the two cases of learning with bounded and unbounded couplings for a small external field . in the lower panel
we compare the hebb rule with the learning algorithm for a large external field . in the algorithm we use @xmath150 , @xmath151 , and @xmath152.,width=377 ]
we studied the finite connectivity inverse hopfield problem at low temperature , where the data are sampled from a non - ergodic regime .
we showed that the information contained in the fluctuations within single pure states can be used to infer the correct interactions .
we also used these findings to design a simple learning protocol which is able to store patterns learned under noisy conditions .
surprisingly enough it was possible to show that by demanding a small though finite overlap with the patterns it is possible to store a large number of patterns without ever reaching a spin glass phase .
the learning process avoids the spin glass phase by decreasing the overlaps , as the number of patterns increases . a separate analysis which is similar to the one presented in ref .
@xcite ( and not reported here ) shows that the equations can be heavily simplified without loosing their main learning capabilities . in this paper we focused on a simple model of neural networks with symmetric couplings .
it would be interesting to study more realistic models like the integrate and fire model of neurons with general asymmetric couplings .
moreover , instead of random unbiased patterns one may consider sparse patterns which are more relevant in the realm of neural networks .
the arguments presented in this paper can also be relevant to problem of inferring a dynamical model for a system by observing its dynamics . in this case , a system is defined solely by its evolution equations and one can not rely on the boltzmann equilibrium distribution . still it is possible to try to infer the model by writing the likelihood for the model parameters given the data and given the underlying dynamical stochastic process .
a mean - field approach has been recently described in @xcite .
we actually checked this approach in our problem and observed qualitatively the same behavior as the static approach .
in fact , which method is best heavily depends on the type of data which are available .
the work of was partially supported by a _
programma neuroscienze _ grant by the compagnia di san paolo and the ec grant 265496 .
a @xmath153 limit version of this result ( except the determination of the value of the constant @xmath154 ) appeared in @xcite .
this result is valid for general ( non - zero ) interactions . for a family of `` potentials '' @xmath155 , where we denote by @xmath156 the subvector of @xmath157 given by @xmath158 .
we will use the shorthand @xmath159 or @xmath160 to mean @xmath161 . _
proposition_. given a factorized probability function @xmath162 and a bp fixed point @xmath163 and plaquette marginals @xmath164 and single marginals @xmath165 for every @xmath160 , then @xmath166 _ proof_. using the fact that @xmath167 and @xmath168 , we obtain @xmath169 , then using the definitions : @xmath170 this proves that a fixed point can be interpreted as a form of reparametrization of the original potentials .
in fact , a sort of converse also holds : _ proposition _ : if @xmath171 satisfies a bethe - type expression @xmath172 with @xmath173 for every @xmath159 .
then there exists a bp fixed point @xmath163 such that @xmath174 .
_ proof _ : choose any configuration @xmath175 .
we will use the following notation : @xmath176 , and @xmath177 . define @xmath178 , normalized appropriately .
afterwards , we can define @xmath179 . by definition of @xmath180 we have @xmath181 . similarly , but using ( [ eq : bethe3 ] ) , and noting by @xmath182 , we have also @xmath183 .
then @xmath184 , and thus @xmath185 .
this also implies that @xmath186 , proving that the first bp equation is satisfied . by definition of @xmath180 , @xmath187 where @xmath188 . moreover using ( [ eq : bethe3 ] )
, we can conclude that also @xmath189 where @xmath190 this implies that @xmath191 .
but we also have that @xmath192 , so @xmath174 as desired .
now @xmath193 by hypothesis , so @xmath194 and this proves that the second bp equation is also satisfied .
consider the ising model on a tree graph of size @xmath13 with couplings @xmath31 and external fields @xmath32 .
suppose that we are given a reference point @xmath68 in the configuration space @xmath195 and the following measure @xmath196 where @xmath74 is a sphere of radius @xmath73 centered at @xmath68 .
by distance of two configurations we mean the hamming distance , i.e. number of spins which are different in the two configurations .
the aim is to compute thermodynamic quantities like average magnetizations and correlations in an efficient way .
we do this by means of the bethe approximation and so bp algorithm .
first we express the global constraint @xmath197 as a set of local constraints by introducing messages @xmath198 that each node sends for its neighbors . for a given configuration @xmath26
, @xmath199 denotes the distance of @xmath26 from @xmath68 in the cavity graph @xmath200 which includes @xmath23 and all nodes connected to @xmath41 through @xmath23 . with these new variables we can write bp equations as @xmath201 where @xmath202 is an indicator function to check the constrains on @xmath203 and @xmath204 . starting from random initial values for the bp messages we update them according to the above equation .
after convergence the local marginals read @xmath205 where in @xmath206 we check if @xmath207 .
these marginals will be used to compute the average magnetizations and correlations .
notice that when the graph is not a tree we need to pass the messages @xmath198 only along the edges of a spanning tree ( or chain ) which is selected and fixed at the beginning of the algorithm .
consider @xmath19 patterns @xmath208 , where @xmath209 and @xmath210 goes from @xmath148 to @xmath211 , which is equivalent to the size of system .
the patterns , learning rate @xmath212 and parameter @xmath95 are fixed at the beginning of the algorithm . to each patten
we assign a population of messages @xmath213 where @xmath214 ( @xmath15 is the node degree ) .
these are to represent the normalized bp messages that we use in the learning algorithm .
besides this we have also a population of couplings @xmath215 .
notice to the maximum we are taking in the last step .
this is to ensure that bp messages in population @xmath106 are related to pattern @xmath223 .
we do these updates for @xmath224 iterations , where in each iteration all members of a population are updated in a random sequential way . | we discuss how inference can be performed when data are sampled from the non - ergodic phase of systems with multiple attractors .
we take as model system the finite connectivity hopfield model in the memory phase and suggest a cavity method approach to reconstruct the couplings when the data are separately sampled from few attractor states .
we also show how the inference results can be converted into a learning protocol for neural networks in which patterns are presented through weak external fields .
the protocol is simple and fully local , and is able to store patterns with a finite overlap with the input patterns without ever reaching a spin glass phase where all memories are lost . |
over the last six years there has been large interest in the ads / cft correspondences providing the gravity duals of non relativistic gauge theories relevant in solid state physics @xcite .
a non - relativistic field theory should have a dual made out of a non - relativistic theory of gravity .
the mathematical structure of such a theory - the newton
cartan ( nc ) space - time - is rather baroque .
it consists of a torsion free non
metric connection @xmath1 , a parallel degenerate metric @xmath2 and a parallel one
form @xmath3 in the kernel of this metric @xcite .
in this paper we shall construct a non - relativistic twistor theory for newton - cartan space - times .
there are several equivalent definitions of twistors for flat minkowski space , but they lead to nonequivalent pictures once a non - relativistic limit is taken .
one could define twistors as spinors for the conformal group @xmath4 and then take a wigner contraction corresponding to a limit where the speed of light becomes infinite @xcite .
this approach has to be abandoned in the search for an analogue of a curved twistor correspondence @xcite as a general space time does not admit any conformal isometries .
thus the linear / group theoretic structures of the twistor space become irrelevant .
we shall instead concentrate on the holomorphic geometry of the twistor space and its one - dimensional complex submanifolds . in the non - relativistic limit the normal bundle of the twistor curves corresponding to space - time points jumps from @xmath5 to @xmath0 , where @xmath6 is the holomorphic line bundle over @xmath7 with the chern class @xmath8 .
this results in the conformal structure becoming degenerate .
we shall show that the newtonian space - times are unstable under the general kodaira deformation of the twistor complex structure .
the newton
cartan connections can nevertheless be reconstructed from merkulov s generalisation of the kodaira map augmented by a choice of a holomorphic line bundle over the twistor space trivial on twistor lines .
the coriolis force may be incorporated by holomorphic vector bundles , which in general are non
trivial on twistor lines .
the resulting geometries agree with non
relativistic limits of anti - self - dual gravitational instantons .
the paper is organised as follows : in the next section we shall review the newton - cartan geometry , and show that in the presence of the time - independent coriolis force , the general newton
cartan connection satisfying the einstein equations is determined by a couple of harmonic functions on @xmath9 . in section [ nr_limit ]
we shall explain how the nc structures arise as limits from general relativity .
we shall give examples of nc limit of the lorentzian taub - nut metric , anti self dual ( asd ) gravitational waves , and gibbons hawking gravitational instantons .
parameter family of anti self
dual ricci flat metrics [ gh_intro ] g_=(1+v)(dx^2+dy^2+dz^2)+(d+ ^3/2 a)^2 depending on a harmonic function @xmath10 and a one form @xmath11 on @xmath9 such that @xmath12 admits a newtonian limit @xmath13 with the unparametrised geodesic equations of @xmath1 given by @xmath14 . in section [ new_spin ]
we shall introduce the spinor calculus adopted to a 3 + 1 splitting , leading to newtonian spinors . in section [ twistor_sect ] we shall review the relativistic twistor theory , and proceed to construct its non - relativistic limit .
this will be done by realising the relativistic twistor space as an affine line bundle over the total space of the bundle @xmath15 , taking a limit of the resulting patching matrix ( section [ sect_jump_lines ] ) and then ( section [ nrte ] ) by directly analysing the incidence relation for a complexified minkowski space @xmath16 with a speed of light @xmath17 .
we shall prove the following [ theo_1 ] let @xmath18 be a one - parameter family of rank - two holomorphic vector bundles over @xmath7 determined by a patching matrix @xmath19 1 .
if @xmath20 is finite , then the holomorphic sections of @xmath21 have normal bundle @xmath22 .
the four dimensional moduli space @xmath16 of these sections carries a flat conformal structure such that two points in @xmath16 are null separated iff the corresponding sections intersect at one point in @xmath18 .
if @xmath23 then the holomorphic sections of @xmath21 have normal bundle @xmath24 .
the four dimensional moduli space @xmath16 of these sections admits a fibration @xmath25 defined by a closed one form @xmath26 , and a contravariant metric @xmath2 of rank @xmath27 which is non
degenerate on @xmath28 .
3 . for any @xmath17 ( finite or not )
there exists an involution @xmath29 which restricts to an antipodal map on each section , and such that the @xmath30invariant sections form a real four - manifold @xmath31 with structures as in @xmath32 and @xmath33 .
this theorem will be established in section [ sect_jump_lines ] .
we shall explicitly construct the sections of @xmath18 which will allow us to identify @xmath20 with the speed of light .
if @xmath23 then @xmath34 and @xmath35 , but for finite @xmath20 there exist holomorphic splitting matrices @xmath36 and @xmath37 such that @xmath38 and @xmath39 . for all non
zero @xmath20 there exists a holomorphic fibration @xmath40 .
this fibration is trivial if @xmath23 , which gives rise to a global twistor function on @xmath41 and consequently to a closed one form @xmath3 on @xmath16 .
the conformal freedom in @xmath3 and @xmath2 are both fixed by a holomorphic canonical one form on the base of the fibration @xmath42 . in section [ sec_kodaira ]
we shall discuss the kodaira deformation theory of the complex structure underlying the non - relativistic twistor space .
we shall exhibit a class of deformations which do not preserve the type of the normal bundle of twistor curves , and lead to gibbons hawking metrics .
this is a consequence of the non - vanishing of the obstruction group @xmath43 , where @xmath44 is the normal bundle to non relativistic twistor lines
. we shall also discuss deformations for which the twistor curves do not change their holomorphic type .
in section [ sec_merkulov ] we shall use merkulov s relative deformation theory to give a construction of newtonian connections on @xmath31 with no coriolis term .
we shall establish [ theo_3 ] there is a 1 - 1 correspondence between line bundles over @xmath41 which are trivial on @xmath30invariant sections of @xmath42 , and newtonian connections on @xmath31 .
it may appear that all theorem [ theo_3 ] does is to reinterpret the penrose transform between @xmath30invariant sections of @xmath45 and solutions to the laplace equation on @xmath9 .
in fact we are saying much more than that .
the newtonian connections of theorem [ theo_3 ] will arise as a family of morphisms [ mer_map_i ] n_(^2n_^ * ) tm_^2(t^*m _ ) , where @xmath46 is the normal bundle to the correspondence space @xmath47 in the product manifold @xmath48 and where @xmath49 is a rational curve in @xmath41 corresponding to a point @xmath50 in the complexified newtonian space
time @xmath16 .
the contribution from the @xmath51 factor in ( [ mer_map_i ] ) corresponds ( by the serre duality ) to a zero - rest - mass field on @xmath16 arising from a gravitational potential @xmath10 in the gibbons hawking family ( [ gh_intro ] ) with @xmath52 . in section [ sparling_section ]
rank two vector bundles which restrict to @xmath24 on twistor curves will lead to connections with non
zero coriolis term . in appendix 1
we shall exhibit and interpret a contour integral formula @xmath53 for solutions to the 2 + 1 schrodinger equation @xmath54 .
finally in appendix 2 we shall show how the spin connection in the nonlinear graviton construction arises as a ward transform of a rank two holomorphic vector bundle over the relativistic deformed twistor space .
we are grateful to christian duval , george sparling and paul tod for helpful discussions .
this work started when md was visiting the institute for fundamental sciences ( imp ) in tehran in april 2010 .
md is grateful to imp for the extended hospitality when volcanic eruption in iceland halted air travel in europe .
the work of jg has been supported by an stfc studentship .
the trajectories of test particles in newtonian physics with a chosen universal time @xmath55 are integral curves @xmath56 of the system of odes [ geodesics_v ] = -v , where @xmath57 is the newtonian potential .
one normally interprets these equations as giving rise to curved paths in a fixed three - dimensional space .
the geometric perspective of cartan @xcite differs in that the particle trajectories are instead regarded as geodesics of some connection in a four - dimensional space time @xmath31 . to introduce
the necessary structures consider a four manifold @xmath31 equipped with a triple @xmath58 , where @xmath2 is a degenerate contravariant metric of signature @xmath59 , @xmath3 is a one form which belongs to a kernel of @xmath2 ( when viewed a map from @xmath60 to @xmath61 ) , and @xmath1 is a torsion free affine connection on @xmath31 with covariant derivative @xmath1 which keeps @xmath2 and @xmath3 parallel . in local coordinates [ compatibility ] h = h^ab_a_b , = _ ad x^a , _ a h^bc=_a_b=0 .
the last two conditions do not specify @xmath1 uniquely .
the most general connection satisfying ( [ compatibility ] ) is parametrised by an otherwise irrelevant choice of time like vector @xmath62 such that @xmath63 and a choice of two
form @xmath64 .
the most general connection such that @xmath65}=0 $ ] is given by the christoffel symbols [ g_connection ] _
ab^c = h^dc(_a h_bd+_bh_ad-_d h_ab)+_(a_b)u^c + _ ( af_b)dh^dc , where @xmath66 , and @xmath67 is a degenerate metric uniquely determined by the conditions @xmath68 connections of the form ( [ g_connection ] ) are called _
galilean_. a galilean connection is called _ newtonian _ if the two - form @xmath64 is closed , or equivalently if the trautman condition @xcite [ trautman ] h^a[br^c]_(de)a=0 holds . here
@xmath69 is the curvature of @xmath1 .
the condition ( [ trautman ] ) is required for consistency with a non - relativistic limit of general relativity ( see section [ nr_limit ] ) .
even this additional requirement allows connections which are more general than one needs to reproduce the newtonian theory ( see section [ corio_ex ] ) . for a given @xmath70
the pair @xmath71 is defined up to gauge transformations @xmath72 where @xmath73 is an arbitrary one form , and @xmath74 thus , if @xmath1 is a newtonian connection ( so that @xmath75 ) , then one can locally set @xmath76 and instead work with a non trivial vector field @xmath62 .
this approach is adapted in @xcite .
the absence of torsion of @xmath1 , and the consistency conditions ( [ compatibility ] ) imply the existence of a function @xmath77 such that @xmath78 .
the three dimensional distribution spanning the kernel of @xmath3 is integrable in the sense of the frobenius theorem , and the manifold @xmath31 admits a fibration over a real line @xmath79 called the time axis , with a coordinate @xmath55 @xmath80 thus the space
time @xmath31 in the newton
cartan theory is a fibre bundle over a universal time axis equipped with a riemannian metric @xmath2 on the fibers .
the einstein field equations are [ field_eq ] r_ab=4g_a_b , where @xmath81 is the ricci tensor of @xmath1 , the function @xmath82 is the mass density and @xmath83 is the gravitational constant . these equations
in particular imply that the metric @xmath2 on three - dimensional spatial fibres is flat .
this allows an introduction of the _ galilean coordinates _
@xmath84 such that ( in general indices @xmath85 run from 0 to 3 and @xmath86 run from 1 to 3 ) , @xmath87 let us spell out some physical consequences of the newton
cartan formalism . in general
, the free falling particles follow the geodesics of @xmath1 which however do not need to be affinely parametrised .
the statement that two events are simultaneous in newtonian physics is invariant and coordinate independent .
this is a consequence of the existence of the one
form @xmath3 and the resulting fibration @xmath88 .
on the other hand one can not say that two events occur at the same point in 3-space but at different times . to be able to compare two non - simultaneous events one needs to make an additional choice of galilean coordinates such that the spatial part of the newton - cartan connection vanishes .
the existence of such coordinates is a consequence of @xmath89 , and the flatness of @xmath2 which itself follows from the field equations ( [ field_eq ] ) .
moreover the distance between two events is only invariantly defined if these events are simultaneous .
take @xmath2 to be a flat metric on the spatial fibres , i. e. @xmath90 , @xmath91 and consider @xmath92 with all other components of @xmath1 equal to zero .
in this case @xmath93 and the geodesics of @xmath1 parametrised by @xmath55 are the integral curves of the newton equations of motion ( [ geodesics_v ] ) .
the coordinate transformation [ gali ] * x
* = r(t)*x*+*a*(t ) , where @xmath94 is an element of @xmath95 which is allowed to depend on @xmath55 , gives rise to the ` coriolis forces ' given by the connection components @xmath96 a combination of an inertial force @xmath97 and a centrifugal force @xmath98 also appear as components of the connection as @xmath99 the form ( [ geodesics_v ] ) of the geodesic equations is preserved only if @xmath100 .
the newtonian connection of the previous subsection can be modified to a galilean connection which includes the general space - time dependent coriolis force @xmath101 . in this case
the closed two - form @xmath64 is @xmath102 and the corresponding non - zero connection components are [ cor_con ] _
ab^c= h^cd_(af_b)d , _
00^i =- e^i , _ 0j^i=^i , where @xmath78 , @xmath103 is a degenerate metric of signature @xmath104 such that @xmath105 which in our case is chosen to be @xmath106 .
the field equations together with the trautman condition imply @xmath107 , for some function @xmath108 , and [ new_ricci ] = 2 |w|^2 , ^2 w=0 . at this stage
we can not assume that @xmath109 .
this can be either put in by hand , or derived from an asymptotic condition that @xmath110 at spatial infinity .
this condition , together with the uniqueness theorem for the laplace equation implies that @xmath111 can be set to zero .
this is because if @xmath112 , then @xmath113 , which can be set to zero by a choice of an orthogonal matrix @xmath114 .
alternatively , the closure of @xmath64 implies that @xmath115 arises from a potential if @xmath111 does not depend on @xmath55 .
if @xmath109 , then @xmath116 so that the connection is determined by two harmonic functions . in section ( [ sparling_section ] ) we shall show how two independent harmonic functions arise from holomorphic vector bundles over a newtonian twistor space @xmath117 which restrict to a non - trivial vector bundle @xmath24 on each twistor line .
in this section we shall show how newton
cartan geometry arises as a degenerate limit of general relativity @xcite .
let @xmath118 be a family of pseudo - riemannian manifolds parametrised by @xmath119 , and such that @xmath120 where @xmath103 is a contravariant tensor on @xmath31 of signature @xmath59 .
the following result has been established in @xcite [ theo_kunz ] let @xmath121 be an @xmath122independent one form such that @xmath123 , normalised by @xmath124 . if the one form @xmath3 is closed , then the levi - civita connection of @xmath125 has a well defined limit @xmath1 as @xmath122 goes to zero .
it is a newton - cartan connection ( [ g_connection ] ) with @xmath75 .
the newton
cartan structure resulting from theorem [ theo_kunz ] is @xmath126 .
the ricci tensor of @xmath125 also has a well defined limit , and choosing a perfect fluid energy - momentum tensor for the relativistic theory leads ( in the limit ) to the field equations ( [ field_eq ] ) .
see @xcite for details of this derivation . another way to proceed @xcite is to define a one - parameter family of structures @xmath127 such that @xmath128 are parallel with respect to @xmath1 and @xmath129 . for @xmath119
this gives pseudo - riemannian geometry with @xmath130 , and for @xmath131 the newton
cartan theory with @xmath132 , and @xmath133 .
we shall now give three examples of newtonian limits corresponding to particular solutions of einstein equations .
the first example is a minor modification from @xcite ( where some factors appear to be wrong ) , and corresponds to a non zero coriolis force and a newton
cartan structure of the type ( [ cor_con ] ) .
the remaining two examples give limits of anti - self - dual solutions and thus correspond to analytic continuation of the newton
cartan theory .
the resulting geodesic equations can be analytically continued to newton s equations of motion despite the fact that asd gravitational instantons in riemannian signature , or asd @xmath134waves in neutral signature do not admit lorentzian analytic continuations .
our motivation for considering the limits of asd solutions will become clear in section [ sec_ward ] when we construct non relativistic twistor spaces from a limiting procedure .
consider a one - parameter family @xcite of lorentzian taub - nut metrics , with the mass parameter @xmath135 , and the nut charge @xmath136 @xmath137 where @xmath138 employing the cartesian coordinates @xmath139 @xmath140 yields the gravitational acceleration @xmath141 and the coriolis force @xmath142 both forces arise from scalar potentials @xmath10 and @xmath143 , but only @xmath108 is a harmonic function .
the two potentials satisfy @xmath144 which implies the ricci - flat condition ( [ new_ricci ] ) .
let @xmath10 and @xmath11 be respectively a harmonic function , and a one form on @xmath9 which satisfy the abelian monopole equation @xmath145 where @xmath146 is the hodge endomorphism of the flat metric on @xmath9 .
parameter family of gibbons hawking metrics @xcite [ gh_earlier ] g=(1+v)(dx^2+dy^2+dz^2)+(d+ ^3/2 a)^2 has anti self dual weyl tensor and is ricci flat .
conversely , any riemannian asd ricci
flat ( and so hyper - khler ) four - manifold admitting a killing vector preserving the hyper - kahler structure is of the form ( [ gh_earlier ] ) .
the newtonian limit of ( [ gh_earlier ] ) is [ ghcon ] h^ij=^ij,_= ^ij , = d , where all other components of @xmath147 vanish .
for example @xmath148 corresponds to the asd taub - nut metric .
it is an asymptotically locally flat ( alf ) gravitational instanton .
the metric ( [ gh_earlier ] ) is everywhere regular , and asymptotically approaches the non
trivial @xmath149 fibration over @xmath150 with the monopole number @xmath151 . in the newtonian limit
the fibration becomes trivial , so that the connection is asymptotically flat , but the genuine singularity arises at @xmath152 . the most general asd ricci - flat metric admitting a parallel killing vector is of the form @xcite [ pp_wave_m ] g = dwdx+dzdy+(w , y)dw^2 . if the coordinates @xmath153 on @xmath31 , and the arbitrary function @xmath154 are taken to be real , then the metric has signature @xmath155 .
set @xmath156 and redefine @xmath157 .
this , for a given @xmath157 , gives a one parameter family of metrics @xmath158 with a newtonian limit [ newtonian_pp ] h_0 ( g^-1)=-_u_u+2_y_z , _ 0 ( g)=dtdt , _
tt^z=- , where now @xmath159 .
this structure can , for certain @xmath157s , be analytically continued to a newtonian theory so that @xmath160 now the limit is @xmath161 this appears to be different than ( [ newtonian_pp ] ) , but is nt , as the coordinate transformation @xmath162 where @xmath163 gives @xmath164 , where the signature of @xmath2 is @xmath104 .
let @xmath165 be a ( pseudo ) riemannian four manifold . locally there exist complex
rank - two vector bundles @xmath166 and @xmath167 over @xmath31 equipped with covariantly constant symplectic structures @xmath168 and @xmath169 parallel w.r.t the levi civita connection and such that [ can_bun_iso ] t m is a canonical bundle isomorphism , and @xmath170 for @xmath171 and @xmath172 .
we use the conventions of penrose and rindler @xcite , where the spinor indices are capital letters , unprimed for sections of @xmath166 and primed for sections of @xmath167 . for example @xmath173
denotes a section of @xmath166 , and @xmath174 a section of @xmath175 .
the symplectic structures @xmath176 and @xmath177 ( such that @xmath178 ) are used to lower and raise the spinor indices according to @xmath179 .
the properties of the complex conjugation on spinors depends on the signature of @xmath180 . * if @xmath180 is lorentzian , then the spinor complex conjugation is a map @xmath181 given by @xmath182 linear @xmath183 transformations on @xmath166 induce lorentz transformations on vectors , and @xmath184 . * if @xmath180 is riemannian , then the complex conjugation preserves the type of spinors , i.e. it maps sections of @xmath166 to sections of @xmath166 and sections of @xmath167 to sections of @xmath167 .
it is given by [ riem_rel ] p^a=(p^0 , p^1)^a= ( , - ) , q_a=(q_0 , q_1)_a=(- , ) . in the riemannian case
the structure group is not simple , and @xmath185 , where the spin groups @xmath186 and @xmath187 act linearly on sections of @xmath166 and @xmath167 respectively .
now assume that @xmath180 has lorentzian signature , and chose a time like unit vector @xmath188 to perform the 3 + 1 split of @xmath180 .
this vector gives a map from @xmath166 to @xmath167 given by @xcite [ 31spinors ] p^a^ap^a , where we have chosen [ unit_vector ] t^aa=(o^ao^a+^a^a ) where @xmath189 is the normalised basis of @xmath167 , i.e. @xmath190 , and @xmath191 is the normalised basis of @xmath166 . under this map
the isomorphism ( [ can_bun_iso ] ) becomes @xmath192 where @xmath193 is a rank
one vector bundle , and @xmath194 is a rank three vector bundle isomorphic to a bundle of self - dual two - forms on @xmath31 .
the isomorphism ( [ 3spinor ] ) gives the orthogonal decomposition : [ decomposition ] v^aav^aat_a^b=v^(ab)+ g(v , t)^ab. the symmetric matrix @xmath195 represents a three - vector in the space @xmath196const and the multiple of @xmath197 is the component in normal direction .
consider the flat metric @xmath198 where [ point_matrix ] x^aa= ( cc ct+z & x+iy + x - iy&ct - z ) ( so that @xmath199 ) .
the map ( [ 31spinors ] ) gives @xmath200 where [ three_matrix ] x^ab= ( cc x - iy & -z + -z&-x - iy ) . in our discussion of the newtonian limit of general relativity in section [ nr_limit ] we needed to single out a time like one form @xmath133 in the kernel of the three
metric @xmath2 .
we shall use the dual vector
field to perform the 3 + 1 decomposition ( [ decomposition ] ) , and construct the isomorphism ( [ 31spinors ] ) .
therefore the two component spinor formalism for newton
cartan structures involves only one type of spinors ( which we choose to be sections of @xmath167 ) .
it is natural to ask whether the newton
cartan connection on @xmath61 is induced from some spin connection on @xmath201 .
the answer to this question appears to be ` no ' which is related to the fact that the galilean boosts do not have a double cover .
a newton cartan connection takes its values in the lie algebra of the galilean group .
however we find that the homogeneous galilean transformations @xmath202 can not be realised spinorially as [ gal_spin ] y^a_b_c_d_b where @xmath203 or @xmath204 , and @xmath205 is trace - less . the condition ( [ gal_spin ] ) forces @xmath206 which then implies @xmath207 .
the choice of the unit vector ( [ unit_vector ] ) breaks the lorentz invariance .
a subgroup of @xmath183 which acts on @xmath166 and @xmath167 , and preserves ( [ unit_vector ] ) is @xmath186 .
its action on spinors only generates rotations , and not lorentz boosts .
the more general transformations of the form @xmath208 do not work either .
it is nevertheless possible to construct a spin connection in the newtonian limit , although to achieve a non zero result the speed of light has to be regarded as a part of the constant metric on the fibres of @xmath60 in the cartan formalism . to see it consider the post
newtonian limit of gr given by [ post_new ] g =-
c^2e^2v / c^2 dt^2+_ijdx^idx^j . employing the spin frame @xmath209 yields a spin connection which vanishes in the limit @xmath210 .
let us instead consider @xmath211 so that ( [ post_new ] ) is given by @xmath212 .
then the cartan equations @xmath213 together with a limit @xmath214 yield the non vanishing connection given by @xmath215
in this section we shall recall basic facts about twistor theory , following @xcite and @xcite .
the theory was put forward by roger penrose @xcite in the late 1960s .
the primary motivation was to unify general relativity and quantum theory in a non local theory based on complex numbers , but a good starting point for our discussion is penrose s contour integral formula for wave equation in minkowski space @xcite [ penrose_form ] ( x , y , z , t)=_^1 f((x+iy)+(ct+z),(ct - z ) + ( x - iy ) , ) d , where @xmath216 is a closed contour and the function @xmath217 is holomorphic on an intersection of two open sets in @xmath7 away from poles inside @xmath218 .
differentiating inside the integral shows that [ wave_eq ] -- -=0 .
the formula ( [ penrose_form ] ) gives real solutions to the wave equation in minkowski space from holomorphic functions of three variables , and the next step is to recognise these variables as coordinates in some complex three dimensional manifold .
the five dimensional space @xmath219 of light rays in the minkowski space is a hyper surface in a three - dimensional complex manifold @xmath220 called the projective twistor space .
let @xmath221 with @xmath222 be homogeneous coordinates of a twistor .
the minkowski space and the twistor space are linked by the incidence relation [ 4d_incidence ] ( cc ^0 + ^1 ) = ( cc ct+z & x+iy + x - iy & ct - z ) ( cc _ 0 + _ 1 ) where @xmath223 are coordinates of a point in minkowski space .
it then follows that if two points in minkowski space are incident with the same twistor , then they are separated by a null geodesic .
the spinor form of the incidence formula above is [ twistor_equation ] ^a = x^aa_a , where the displacement vector @xmath224 is given by ( [ point_matrix ] ) .
let @xmath225 represent homogeneous coordinates of a twistor .
the non projective twistor space @xmath226 is equipped with a hermitian inner product [ inner_null ] ( z , |z)=^a|_a+|^a_a. the signature of @xmath227 is @xmath228 and endomorphisms of @xmath229 preserving @xmath227 as well as the orientation of @xmath230 form a group @xmath4 which is locally isomorphic to the conformal group of the minkowski space .
the projective twistor space is divided into three regions depending on the sign of @xmath231 .
the hypersurface [ pnspace ] pn=\{(^a , _
a ) , ( z , |z)=0 } is preserved by the conformal transformations of the minkowski space . fixing the coordinates
@xmath232 of a space time point @xmath50 in ( [ 4d_incidence ] ) gives a projective line @xmath49 in @xmath233 .
if the coordinates of @xmath50 are real , then @xmath234 is contained in the hypersurface @xmath219 .
conversely , fixing a point in @xmath219 gives a light
ray in the minkowski space .
points in @xmath235 can be interpreted in terms of the complexified minkowski space @xmath236 where they correspond to null two dimensional planes ( @xmath237planes ) with self dual tangent bi - vector .
this , again , is a direct consequence of ( [ 4d_incidence ] ) where now the coordinates @xmath224 are complex ( figure 1 ) . * figure 1 . * _ twistor relativistic incidence relation .
two points on an @xmath237plane in @xmath16 correspond to two twistor lines passing through a point in @xmath238 .
_ there is a @xmath7 worth of @xmath237planes through each point of @xmath16 . for any @xmath239 a corresponding @xmath237plane
is spanned by a two - dimensional twistor distribution _ a=^a_aa,_aa= .
[ alpha_dist ] finally we can give a twistor interpretation of the contour integral formula ( [ penrose_form ] ) .
consider a function @xmath240 which is holomorphic on an intersection of two open sets covering @xmath238 ( one of these sets is defined by @xmath241 and the other by @xmath242 ) .
restrict @xmath217 to a rational curve @xmath243 and integrate @xmath217 along a contour in @xmath234 .
this yields ( [ penrose_form ] ) with @xmath244 .
the function @xmath217 , when regarded as defined on the non projective twistor space , must be homogeneous of degree @xmath245 .
our construction of a non - relativistic twistor space for the flat newtonian space time will proceed in two steps .
we shall first represent the relativistic twistor space introduced in the last section as an affine line bundle over @xmath246 , and show that in the limit @xmath214 the normal bundle of sections of @xmath247 jumps from @xmath22 to @xmath24 .
this will be done is section [ sect_jump_lines ] . in section [ nrte ]
we shall relate the jumping lines approach to a non - relativistic limit of the 3 + 1 split of the twistor incidence relation .
let @xmath248 $ ] be homogeneous coordinates on @xmath7 .
cover @xmath7 with two open sets [ uuopen_sets ] u=\{[]^1 , _ 10 } , = \{[]^1 , _ 00 } and set @xmath249 on @xmath250 .
the birkhoff - grothendieck theorem states that any rank-@xmath251 holomorphic vector bundle over @xmath7 is isomorphic to a direct sum of line bundles @xmath252 for some integers @xmath253 .
moreover the transition matrix @xmath254 of this bundle can be written as [ bh_split ] f= ( ^-m_1 , ^-m_2 , ,
^-m_k ) h^-1 , where @xmath255 and @xmath256 are holomorphic .
let @xmath257 be a one - parameter family of rank - two vector bundles determined by a patching matrix @xmath258 where @xmath20 is a constant .
if @xmath23 then @xmath34 is the patching matrix for @xmath259 with @xmath36 and @xmath37 in ( [ bh_split ] ) both equal to the identity matrix . from now on we shall identify the constant @xmath20 with the speed of light , and refer to @xmath117 as the newtonian twistor space . if @xmath260 then @xmath261 which is of the form ( [ bh_split ] ) with @xmath262 thus @xmath263 if the speed of light is non - zero and finite .
this is the relativistic twistor space , with the holomorphic sections of @xmath264 parametrised by points in the complexified minkowski space @xmath16 .
let @xmath224 given by ( [ point_matrix ] ) be a displacement vector of a point in @xmath16 from the origin , and let @xmath265 be a unit vector ( [ unit_vector ] ) with respect to the flat metric ( [ metric_md ] ) .
set @xmath266 and define inhomogeneous coordinates @xmath267 and @xmath268 in pre - images of @xmath62 and @xmath269 in @xmath18 by [ patch_u ] ( q=2 , t= ) u and @xmath270 on the pre - image of @xmath250 in @xmath18 we have @xmath271 or @xmath272 note that on @xmath41 there exists a global twistor function of weight 0 , given by @xmath273 . restricting the inhomogeneous coordinates to a twistor line @xmath274 gives @xmath275 where @xmath276 .
thus , if @xmath23 , then @xmath277 on the twistor lines , where @xmath55 is the global time coordinate on the complexified newtonian space - time . the algebraic geometry of holomorphic sections of @xmath21 determines the metric structure on @xmath16 : two points in @xmath16 are null separated iff the corresponding sections intersect at one point in @xmath18 .
infinitesimally , a vector in @xmath278 is null if the corresponding section of @xmath279 vanishes at one point .
this condition is equivalent to the existence of the unique solution @xmath280 for a simultaneous system @xmath281 where @xmath282 and @xmath188 are given by ( [ rest_lines ] ) , and the variation @xmath283 is w.r.t the moduli @xmath232 and not @xmath284 . if @xmath20 is finite , then the conformal structure on @xmath16 is determined by @xmath285 , and the conformal factor is fixed by a two - form @xmath286 on the fibres of @xmath18 .
if @xmath23 , then the the conformal structure on minkowski space degenerates in the following way : the equation @xmath287 implies that @xmath288 , and in the newtonian limit the null separation implies simultaneity of events .
we also deduce the existence of a canonical closed one
form @xmath289 on @xmath41 which yields @xmath290 on @xmath16 .
this is the _ clock _ of newton - cartan theory which defines a fibration of @xmath16 over the time axis [ fibration_over_c ] m_m_/()=. the section of @xmath279 will have a single zero if the two roots of the remaining equation @xmath291 coincide . evaluating the discriminant of
the quadratic form leads to a degenerate conformal structure on @xmath16 , which is determined by @xmath292 on three dimensional fibres of ( [ fibration_over_c ] ) .
the conformal factor is fixed by the canonical @xmath246valued one - form @xmath293 on the fibres of @xmath294 by @xmath295 .
the relation between @xmath67 and @xmath103 is provided by a unique choice of a vector field @xmath62 such that @xmath296 , and @xmath297 .
this gives a degenerate contravariant metric on @xmath298 such that @xmath123 .
the discussion of the intrinsic definition of the connection is postponed to section [ sec_merkulov ] .
now we shall move on to discuss the reality conditions and establish point ( 3 ) in theorem [ theo_1 ] .
the euclidean real slice of the complexified minkowski space @xmath16 is characterised by an involution @xmath29 given by [ o2inv ] ( q , , t)=(- , - , -+ ) so that @xmath299 on the projective twistor space .
this involution has no fixed points , because on the non projective twistor space it squares to minus identity , and so for any @xmath300\in pt_c$ ] there is a unique line joining @xmath300 $ ] to @xmath301 $ ] .
these are the _ real twistor curves _ , which correspond to points in the euclidean slice of @xmath16 .
these @xmath30invariant sections are characterised by @xmath302 so that @xmath139 are real .
so that @xmath304 , and @xmath305 is purely imaginary . thus the real newtonian twistor lines in @xmath41 are [ newton_lines ] q=-(x+iy)-2z+^2 ( x - iy ) , = ^-2q , t==iwhere @xmath306 are real coordinates on @xmath31 .
@xmath307 it is worth remarking that in the limit when @xmath210 the real ( i.e. @xmath30invariant ) twistor curves lie on the hypersurface ( [ pnspace ] ) . to see it consider the inner product @xmath308 , where @xmath227 is given by ( [ inner_null ] ) .
then @xmath309 thus the newtonian twistor space is divided into two regions @xmath310 and @xmath311 separated by the five dimensional space @xmath312 of null twistors @xmath313 .
the difference between the lorentzian reality conditions ( twistor curves contained in @xmath314 ) and riemannian reality conditions ( twistor curves invariant under an anti - holomorphic involution ) disappears in the newtonian limit @xmath214 . to sum up , we have exhibited a realisation of the relativistic twistor space @xmath22 as an affine line bundle over the total space of the line bundle @xmath315 . in the limit @xmath23 the holomorphic structure of @xmath18 changes discontinuously : the total space of @xmath18 , and the normal bundle of the twistor curves both jump to @xmath24 .
note that this jumping is a purely holomorphic phenomenon : the bundles @xmath18 are all the same for any @xmath20 finite or not from the topological perspective .
our findings agree with the analysis of @xcite ( as well as with the recent work @xcite ) , where the jump from @xmath22 to @xmath24 was interpreted as a singularity of the conformal structure . combining the @xmath316 splitting ( [ three_matrix ] ) with the incidence relation ( [ twistor_equation ] ) yields [ n_twistor ] x^ab_b-ct ^a=^a where @xmath317 is given by ( [ three_matrix ] ) , and @xmath318 .
this equation blows up in the limit @xmath214 but the twistor function @xmath319 remains finite . if a pair @xmath320 solves the twistor equation ( [ n_twistor ] ) , then
so does @xmath321 for any constant spinor @xmath322 .
the image of the two dimensional twistor distribution ( [ alpha_dist ] ) tangent to @xmath237-planes under ( [ decomposition ] ) is @xmath323 in the newtonian limit this gives an integrable 2-plane distribution ( the mini - twistor distribution ) on the constant time three - dimensional fibres of @xmath324 @xmath325 the non - relativistic twistor space @xmath117 is the three - dimensional space of leaves of @xmath326 in the correspondence space @xmath327 . to specify a leaf
pick a point on the time axis , and choose a two plane ( a point in a mini - twistor space @xmath246 ) which is null with respect to the holomorphic metric @xmath328 on the fibres of ( [ fibration_over_c ] ) .
thus , in the newtonian limit the @xmath237planes become space - like , and they lie on the fibres of ( [ fibration_over_c ] ) ( figure 2 ) . * figure 2 . *
_ an @xmath237plane @xmath329 intersects hyper - planes @xmath330 of constant time in @xmath16 in null lines @xmath157 . in the non
relativistic limit @xmath329 becomes a subspace of a constant - time hyper - plane @xmath331 .
_ a point @xmath332 gives a time coordinate and a @xmath7 worth of 2-planes on the three dimensional spatial fiber .
this is a rational curve @xmath49 in @xmath259 lying in a fibre @xmath333 .
the normal bundle @xmath279 of a rational curve @xmath234 corresponding to @xmath334 is @xmath335 as we have established in section [ sect_jump_lines ] .
this can also be seen from the exact sequence @xmath336 as the last map is , in the spinor notation , given by @xmath337 clearly projecting onto @xmath338 , and the decomposition of @xmath230 into @xmath339 is given by ( [ decomposition ] ) .
this is a complexified picture .
if the reality conditions ( [ riem_rel ] ) and ( [ newton_lines ] ) are imposed , then the intersection of a complex @xmath237-plane with a real @xmath340 slice is a real straight line - a geodesic of the spatial metric - with a tangent vector @xmath341 ( figure 3 ) .
* figure 3 . * _ a real section of the non relativistic incidence relation . oriented straight lines in the fibres of @xmath88
correspond to points in @xmath117 .
two simultaneous events @xmath342 and @xmath343 in newtonian space
time correspond rational curves @xmath344 intersecting at two points in @xmath41 . a time like geodesic between two non simultaneous events @xmath342 and @xmath345 corresponds to a ruled surface in @xmath41 . _ the twistor curves invariant under the involution ( [ o2inv ] ) correspond to points in the flat newtonian space - time @xmath346 .
the points in @xmath41 correspond to oriented straight lines on the fibres of @xmath88 .
thus two real curves @xmath347 and @xmath348 in @xmath117 either do not intersect , or intersect at two points @xmath329 and @xmath349 which correspond to a straight line ( with two possible orientations ) joining @xmath350 and @xmath351 in @xmath31 .
a way to introduce the curvature on @xmath16 in relativistic twistor theory is to deform the complex structure of @xmath18 .
the kodaira theorems @xcite guarantee that , under the additional assumption of the deformation being ` small ' , the deformed twistor space will still admit a four - dimensional moduli space @xmath16 of rational curves with normal bundle @xmath22 .
the nonlinear graviton theorem of penrose @xcite then implies that @xmath16 admits a holomorphic conformal structure with anti - self - dual ( asd ) weyl tensor , and moreover that all asd conformal structures correspond to some deformed twistor spaces .
if the deformation preserves the fibration of @xmath18 over @xmath7 , and the symplectic two - form @xmath352 on the fibres of this fibration , then the resulting conformal structure contains a ricci - flat metric . in this section
we shall show that the kodaira deformation theory , when applied to a newtonian twistor space @xmath117 , in general leads to the jumping lines phenomenon : the deformed twistor curves have normal bundle @xmath22 , and thus @xmath16 carries a non degenerate metric , rather than a newton cartan structure .
we can phrase it by saying that the newtonian space
times are unstable under the general kodaira deformations , but we should note that this instability is a purely holomorphic feature of the underlying twistor space , and not a dynamical process . to consider a sub - class of deformations we shall use our construction ( section [ sect_jump_lines ] ) of the undeformed newtonian twistor space @xmath117 as a total space of a trivial holomorphic line bundle over a total space of @xmath315 @xmath353 where sections of @xmath354 restricted to sections of @xmath315 are twistor curves . this motivates a replacing the undeformed twistor space @xmath117 by @xmath355 where @xmath356 is a non - trivial line bundle becoming trivial on each twistor line .
to construct @xmath356 as a deformation of @xmath357 consider the deformed patching relations [ gh_defo ] = t+f , where @xmath358 represents the cohomology class , and @xmath122 is a deformation parameter .
restrict this cohomology class to a section of @xmath294 and pull back to the correspondence space so that @xmath359 , where @xmath2 and @xmath360 are holomorphic in the open sets @xmath62 and @xmath269 defined by ( [ uuopen_sets ] )
. therefore @xmath361 where @xmath55 does not depend on @xmath362 by the liouville theorem and thus can be used as a coordinate on the moduli space of curves in @xmath356 . the deformed @xmath7s ( in a patch containing @xmath363 )
are @xmath364 where @xmath365 here @xmath366 is a spinor ( a choice of which arises because of the non - uniqueness of the splitting ) .
the contours @xmath218 and @xmath367 are homologous to the equator of the twistor curve @xmath49 in an intersection of two open sets covering @xmath234 , and such that @xmath368 surrounds the point @xmath369 in @xmath234 ( figure 4 ) .
* figure 4 . * _ splitting formula . the contour @xmath370 surrounds a point @xmath371 .
_ the ward transform @xcite of @xmath356 gives [ ward_1 ] _
ah=_ab^b , where @xmath372 is given by [ twistor_monopole ] _ ab= _
.dand the one form @xmath373 together with a function @xmath10 satisfy the monopole equation [ monopole ] dv=*da or @xmath374 therefore @xmath375 is not a twistor distribution , as it does not lie - derive @xmath376 . the deformed twistor distribution (
also called a lax pair ) is @xmath377 where @xmath378 are four vector fields on @xmath16 .
this is a jumping line phenomenon : the deformation changes the normal bundle to @xmath379 .
the vector fields in the lax pair ( [ deformed_twist ] ) @xmath380 give rise , by @xmath381 to a conformally rescaled gibbons - hawking metric .
imposing the reality conditions @xmath305 , where @xmath382 and choosing a conformal factor @xmath10 gives the asd , ricci
flat metric @xmath383 therefore the deformation ( [ gh_defo ] ) does not preserve the holomorphic type of twistor curves .
they jump from the newtonian @xmath24 type to relativistic @xmath22 .
this was to be expected as the general deformations will not preserve the type of the newtonian normal bundle @xmath384 .
this is because @xmath385 .
if we regard a section of @xmath386 as a column vector with first entry taking values in @xmath387 and the second entry taking values in @xmath388 , the endomorphisms of @xmath386 are represented by two by two matrices of the form [ matrix_b ] ( cc & + & ) where @xmath389 .
thus @xmath390 @xmath391 as the second entry in the sum does not vanish .
the argument above shows that general kodaira deformations of @xmath117 do not preserve the type of the normal bundle .
we have seen that a deformation which does not preserve the one
form @xmath289 certainly leads to jumping lines .
we can look at the restricted class of deformations which preserve @xmath289 , and thus do not change the normal bundle .
this requires the coefficient @xmath392 in the matrix ( [ matrix_b ] ) to vanish .
one way to achieve it is to take the twistor space to be a product @xmath393 , where @xmath394 is a two dimensional complex manifold with a three parameter family of curves with normal bundle @xmath246 .
thus @xmath394 is a mini twistor space for a general einstein
weyl structure in three dimensions @xcite , and the corresponding @xmath31 is a product of an einstein weyl three
manifold with a line .
this is not the structure we are seeking , as the metric ( or in fact a conformal structure ) on the three dimensional spatial slices of @xmath31 will in general be curved , whereas we require it to be flat . in the remainder of this section
we shall explore another possibility , and deform the relation between a projective and non - projective twistor space in a way which preserves the one
form @xmath289 .
this is analogous to the construction of @xcite which in turn was motivated by ward s twisted photon construction @xcite .
the canonical bundle of @xmath117 restricted to a twistor line is @xmath395 .
the argument works for @xmath18 with any non zero @xmath20 , as restricting the canonical bundle @xmath396 to a line @xmath234 with a normal bundle @xmath279 gives [ canonical_b ] |_l = t^*l^2(n(l_p))^*=(-4 ) , as @xmath397 .
let us assume that the line bundle @xmath398 is non
trivial on @xmath117 , but trivial when restricted to twistor lines , and deform the twistor space in a way which preserves the fibration over @xmath7 .
therefore @xmath399 where @xmath359 is the cohomology class defining @xmath356 which gives rise to the abelian monopole ( [ ward_1 ] ) .
this cohomology class coincides with that from section [ instability ] and its splitting is given by ( [ splitting_h ] ) .
the deformed twistor distribution [ def_dist ] l_a=_a-^b_ab , = ^a lie derives @xmath400 . to this end
we note that the @xmath401 term can be removed from the lax pair by a mobius transformation of @xmath362 iff @xmath402 is a gradient : if @xmath403 , then setting @xmath404 gives @xmath405 in this case the deformed twistor distribution in frobenius integrable for any function @xmath62 .
the original non linear graviton construction of penrose @xcite yields ricci flat asd metrics from the holomorphic geometry of associated twistor spaces . in a far reaching generalisation of this construction merkulov
@xcite developed a twistor correspondence between complete analytic families of complex submanifolds on a given complex manifold , and a class of torsion free affine connections ( which he called the @xmath406connections ) on the moduli space of these submanifolds .
this has subsequently led to the celebrated solution of the holonomy problem @xcite .
in this section we shall adapt merkulov s construction to the newtonian twistor correspondence , and show that the twistor space @xmath41 introduced earlier in this paper leads to a family of newton cartan connections depending on five arbitrary functions .
two of these functions can be set to zero by fixing the overall conformal factors in the degenerate newtonian metric @xmath2 , and the associated clock one form @xmath3 in the kernel of @xmath2 .
the remaining three functions correspond to the gravitational force . by analysing the non relativistic limit of the gibbons hawking twistor space
we shall show that the gravitational force arises from a harmonic gravitational potential which is given by a choice of a holomorphic line
bundle over @xmath41 trivial on twistor lines .
let @xmath407 be an ideal of holomorphic functions on @xmath16 which vanish to order @xmath251 at @xmath332 .
therefore @xmath408 .
the second order tangent bundle @xmath409}m_{\mathbb{c}}$ ] is defined as a union over all points in @xmath16 of second order tangent spaces @xmath409}_pm_{\mathbb{c}}\equiv ( j_p/{j_p}^3)^*$ ] .
thus an element of @xmath410}_pm_{\mathbb{c}})^*$ ] consists of the first two non zero terms of a taylor expansion of a function vanishing at @xmath50 , and a section of @xmath409}m_{\mathbb{c}}$ ] is a second order linear differential operator @xmath411}=v^a(x)\p_a+v^{ab}(x)\p_a\p_b$ ] .
we also have an exact sequence @xmath412}m_{\mathbb{c}}\rightarrow \mbox{sym}^2(tm_{\mathbb{c}})\rightarrow 0\ ] ] such that @xmath413 , and @xmath414 . a torsion
free affine connection @xmath1 on @xmath415 is then equivalent to a linear map [ m_maps ] : t^[2 ] m_tm _ , where @xmath416})=(v^a+{\gamma^a}_{bc}v^{bc})\p_a\ ] ] for some functions @xmath417 which are to be identified with the christoffel symbols of @xmath1 . in @xcite merkulov
proposed a twistor construction of a class of maps ( [ m_maps ] ) . in some cases (
e.g. the nonlinear graviton construction ) merkulov s construction leads to a unique connection .
we will show that in the case of the non relativistic twistor theory a unique newtonian connection in merkulov s class is provided by a natural choice of a holomorphic line bundle over @xmath41 . the general construction , adapted to three dimensional complex manifolds @xmath18 which fiber holomorphically over @xmath7 , and admit a moduli space @xmath16 of holomorphic curves can be summarised as follows .
let @xmath418 and @xmath419 be a covering of @xmath18 by two open sets which is a deformation of the covering ( [ patch_flat ] ) . thus @xmath420
are holomorphic coordinates in @xmath418 and @xmath421 are holomorphic coordinates on @xmath419 .
the patching relation on the overlap @xmath422 is @xmath423 and @xmath424 , where @xmath425 and @xmath426 .
we shall make use of the kodaira isomorphism , which relates vector fields on @xmath16 to global sections of the normal bundle @xmath46 to the correspondence space @xmath427 in the product manifold @xmath428 .
the associated double fibration @xmath429 yields @xmath430 where @xmath431 is the rational curve in @xmath18 corresponding to a point in @xmath16 . the patching for this normal bundle is given by @xmath432_{\mathcal{f}}=\begin{pmatrix}\p_t\widetilde{t } & \p_q\widetilde{t}\\ \p_{t}\widetilde{q } & \p_q\widetilde{q } \end{pmatrix}_{\mathcal{f}}\ ] ] where the subscript @xmath433 means the pull - back to the correspondence space . to exhibit
the kodaira isomorphism take a vector field given by a differential operator @xmath434 on @xmath433 , and apply it to the twistor functions @xmath435 to obtain a global section of @xmath46 given by @xmath436 the map ( [ m_maps ] ) arises by taking a section of @xmath409}m$ ] and similarly applying the associated differential operator to the twistor functions .
we find @xmath437}(\tilde{w}^a)=f_{\ , b}^{a } v^{[2]}(w^b)+ f_{\ , bc}^{a}v^{ab}\partial_{a } w^{b}\partial_{b } w^{c } \label{eq : obstruction}\ ] ] @xmath438_{\mathcal{f}}.\ ] ] the term involving @xmath439 is an obstruction to ( [ eq : obstruction ] ) representing a global section of @xmath46 ; only if @xmath439 splits as @xmath440 for some 0-cochain @xmath441 of @xmath442 will we arrive at @xmath437}(\tilde{w}^a)+ v^{ab}\tilde{\sigma}_{\ , ef}^{a}\partial_{a}\tilde{w}^{e}\partial_{b}\tilde{w}^{f } = f_{\ , b}^{a } v^{[2]}(w^a)+ v^{ab}f_{\ , b}^{a}\sigma_{\ , ef}^{b}\partial_{a}w^{e}\partial_{b}w^{f}.\ ] ] in solving the splitting problem ( [ eq : splittingproblem ] ) we have constructed a global section of @xmath46 out of a section of @xmath409}m$ ] , and via the kodaira isomorphism we have therefore constructed a vector field out of a section of @xmath409}m$ ] .
this is the map ( [ m_maps ] ) , and we can read off the connection components from @xmath443 one can check that @xmath444 transforms correctly as an affine connection under a change of coordinates .
the splitting problem ( [ eq : splittingproblem ] ) is solvable for any patching in the isomorphism class @xmath46 iff [ coho_m ] h^1(^1 , n_(^2n_^*))=0 .
the procedure is however non unique ( and hence leads to a family of connections ) unless @xmath445 vanishes .
otherwise an element of this cohomology group can be added to @xmath441 . in the relativistic case
@xmath446 which implies that ( [ coho_m ] ) holds , @xmath447 , and therefore the splitting is unique . to illustrate how this procedure works in practice we will consider the simple twistor space ( originally due to sparling ) described by the infinitesimal deformation of the flat relativistic twistor space @xmath18 given by ( [ ap_ham ] ) , where @xmath448 . integrating the deformation equations and introducing the inhomogeneous coordinates on the deformed twistor space by @xmath449 on @xmath418 and @xmath450 on @xmath419 yields [ defor_s ] ^0=w^0,^1=w^1+(w^0)^3()^2 .
the holomorphic splitting of these relations , together with the liouville theorem imply the existence of coordinates @xmath451 on @xmath16 such that the twistor curves pulled back from @xmath418 to @xmath452 are @xmath453 the asd ricci
flat metric resulting from the nonlinear graviton construction is of the form ( [ pp_wave_m ] ) for @xmath454 .
we shall find the connection on @xmath16 directly .
we find @xmath455 and all other @xmath456 .
the splitting problem ( [ eq : splittingproblem ] ) can be solved uniquely to give @xmath457 leading to a connection whose only non - vanishing components are @xmath458 we have recovered the levi civita connection of ( [ pp_wave_m ] ) .
we shall now construct the general class of connections on the moduli space @xmath16 of twistor lines in @xmath459 using ( [ patch_u ] ) and ( [ merpatch ] ) we find @xmath460 the condition @xmath456 implies that the connections will be constructed only out of global sections of @xmath442 . in the relativistic case
this would imply @xmath461 , but for @xmath117 we have @xmath462 , so @xmath463 therefore we expect @xmath444 to depend on five arbitrary functions .
the global sections of @xmath442 which constitute the newtonian 0-cochain @xmath464 are of the form @xmath465 ( with all other components of @xmath466 set to zero ) for five arbitrary functions @xmath467 on @xmath16 .
using ( [ intermedia_eq ] ) and ( [ readoffgamma ] ) , and imposing the reality conditions ( [ newton_lines ] ) yields the non - vanishing connection components @xmath468 where we used the isomorphism ( [ 3spinor ] ) to replace @xmath469 by a spatial vector @xmath470 .
the vector @xmath471 yields the gravitational attraction , whilst @xmath227 and @xmath472 are consistent with @xmath89 and @xmath473 , where the degenerate metric @xmath2 , and clock one form @xmath3 are defined up to some conformal factors .
when we fix these conformal factors as in section [ sect_jump_lines ] , so that @xmath474 and @xmath475 , then @xmath476 .
in particular , we note that this construction can not generate a coriolis force . what remains , then , is a connection whose non
zero @xmath477 components are determined by a global section @xmath478 of @xmath388 . using the serre duality between the elements of @xmath479 and @xmath480 we can fix this connection by [ serre1 ] e_ab= _ g_(-4 ) _
a_b.d , where @xmath481 is an element of @xmath482 restricted to a twistor line ( [ rest_lines ] ) .
the resulting @xmath483 satisfies the zero - rest - mass field equation @xmath484 . in what follows
we shall demonstrate that the expression ( [ serre1 ] ) arises naturally from a limiting procedure applied to the unique splitting of the 0-cochain corresponding to the gibbons hawking metric .
in section [ gh_limit ] we have demonstrated that any newtonian connection with the only non - zero components given by @xmath485 arises as a newtonian limit of the gibbons hawking metric ( [ gh_earlier ] ) .
the twistor space of ( [ gh_earlier ] ) is an affine line bundle over the total space of @xmath246 described by a cohomology class @xmath486 .
this is in fact what we have recovered in our analysis in section [ instability ] leading to the jumping phenomenon and the kodaira instability of the newtonian twistor space .
let us assume that the cohomology class is represented by a function @xmath487 in a patch ( [ patch_u ] ) in @xmath488 , and define [ sec_ward1 ] = q , = t -- f , with @xmath489 giving the patching relation for the undeformed relativistic twistor space @xmath18 ( compare [ patch_flat ] ) .
we then have @xmath490_{\mathcal{f}}\right)\\ 0 & \lambda^{-2 } \end{pmatrix}\,\,\,,\\ f_{\ , ab}^{q}&=&0\,\,\,\,\,\,\mbox{and}\,\,\ , f_{\ , ab}^{t}=\begin{pmatrix}f_{\,tt}^{t } & f_{\,t q}^{t}\\ f_{\ , qt}^{t } & f_{\ , qq}^{t } \end{pmatrix}_{\mathcal f}=\begin{pmatrix}0 & 0\\ 0 & -\frac{1}{c^{3}}\left[\frac{\partial^{2}f}{\partial q^{2}}\right]_{\mathcal{f } } \end{pmatrix}.\end{aligned}\ ] ] for this twistor space we have @xmath491 ( for finite , non zero @xmath20 ) and so one can find the unique solution @xmath492 to the splitting problem ( [ eq : splittingproblem ] ) .
the solution is given by @xmath493 where @xmath218 is a contour enclosing @xmath494 , and @xmath495 is the global twistor function homogeneous of degree @xmath496 on @xmath41 ( compare formula ( [ patch_u ] ) ) restricted to a twistor line .
thus one finds that the only parts of @xmath497 which do not vanish in the newtonian limit are @xmath498 and @xmath499 , constituting a global section of @xmath388 and giving rise to a non - zero @xmath477 via merkulov s procedure above .
this provides a way of fixing the newtonian 0-cochain @xmath500 : we identify it with the @xmath214 limit of the gibbons - hawking 0-cochain [ sigmaequalssigmainfty-1 ] \ { , } _ = _
c ( \ { , } _ c ) .
the only non vanishing components are given by @xmath498 and @xmath499 in ( [ intermedia_eq ] ) , where @xmath483 is a zero - rest - mass field on @xmath16 .
comparing this limit with ( [ serre1 ] ) and replacing @xmath501 by @xmath502 inside the integral yields @xmath503 where @xmath217 is an element of @xmath45 restricted to a twistor line @xmath504 and @xmath10 is the harmonic function which ( up to a constant multiple ) gives the newtonian potential in agreement with ( [ ghcon ] ) .
we conclude that the limiting procedure yields a canonical element of @xmath482 in ( [ serre1 ] ) .
it is given by @xmath505 . to complete the proof
we shall show that @xmath506 gives rise to a line bundle @xmath507 which is trivial on twistor curves .
let @xmath508 be a covering of @xmath117 and let @xmath509 be a local trivialisation of @xmath510 .
the holomorphic patching function @xmath511 is given by @xmath512 .
the multiplicative splitting of @xmath64 reduces to the additive splitting ( [ splitting_h ] ) of @xmath217 , and it gives rise to the abelian monopole ( [ ward_1 ] ) , where @xmath513 satisfy ( [ monopole ] ) . the harmonic function @xmath10 is identified with the newtonian potential from section [ newto_ex ] .
this construction gives all newtonian connections with no coriolis force , as in section [ gh_limit ] we have shown that all such connections arise as limits of metrics from the gibbons hawking class .
@xmath307 consider a one - parameter family of gibbons hawking metrics with linear potential @xcite @xmath514 where @xmath52 .
this family is asd and ricci flat for all @xmath119 . in the limiting case
@xmath515 the metric blows up , the inverse metric degenerates to @xmath516 , but the levi civita connection has the finite limit with the only non - zero components given by @xmath517 the cohomology class defining the line bundle from theorem [ theo_3 ] @xmath518 gives rise to the monopole ( [ twistor_monopole ] ) with @xmath519
theorem [ theo_3 ] gives a satisfactory twistor construction of a newtonian connection with no coriolis term .
for a fixed degenerate metric @xmath2 , the general newton
cartan connections correspond to closed two
forms @xmath64 , which ( given a canonical 3 + 1 splitting induced by the one form @xmath3 ) have well defined electric and magnetic parts ( compare formulae ( [ g_connection ] ) and ( [ cor_con ] ) ) . thus we can identify the connection with a couple of gauge fields on @xmath31 , and construct it from a generalisation of ward transform @xcite adapted to newtonian settings .
the gauge fields needed to reconstruct the gravitational and coriolis parts of the connection consist of a couple of abelian monopoles .
the one forms in the monopoles will give rise to an electric and magnetic fields on @xmath31 , and these fields will in turn give rise to a connection .
this is a legitimate approach , as electromagnetic forces are inertial in the newton - cartan theory : the equations of motion of a charged particle under the influence of the electric field @xmath115 and magnetic field @xmath520 [ ini_em ] = * e*+2 * b * can be reinterpreted as a geodesic motion corresponding to a connection @xmath521 where now @xmath115 is the gravitational acceleration , and @xmath111 is the coriolis force - compare ( [ cor_con ] ) .
the field @xmath111 is a gradient of a harmonic function @xmath108 , and ( assuming that @xmath111 is time
independent ) @xmath109 is also a gradient , but @xmath10 is not necessarily harmonic .
however @xmath522 is harmonic .
let @xmath16 be a moduli space of holomorphic sections of the newtonian twistor space @xmath523 , and let @xmath524 be the five dimensional correspondence space . here the _ twistor line _
@xmath49 is a rational curve in @xmath117 with normal bundle @xmath357 corresponding to @xmath332 .
the correspondence space can be identified with the projective spin bundle @xmath525 .
let @xmath526 denote the corresponding fibration .
thus a pre - image of any twistor line in @xmath452 can be pulled back to @xmath167 , where @xmath527 .
consider a vector bundle @xmath528 such that @xmath529 where @xmath530 .
therefore the restriction of @xmath510 to a twistor line is a non - trivial vector bundle , and the pull back to a pre - image of @xmath234 in @xmath452 of the patching matrix of @xmath510 can not be split in a way which leads to the ward twistor correspondence @xcite .
however pulling back @xmath510 to the total space of @xmath167 , and restricting it to a pre - image @xmath531 of @xmath234 in @xmath167 is a trivial bundle .
the corresponding pulled - back @xmath532 patching matrix @xmath533 , where @xmath282 is given by ( [ rest_lines ] ) , satisfies and @xmath37 can be derived from those in ( [ bh_split ] ) using @xmath534 ] @xmath535 where @xmath536 and @xmath537 are holomorphic in @xmath62 and @xmath538 respectively , where now @xmath62 and @xmath538 are pre - images of the standard covering of @xmath234 in @xmath531 .
note that @xmath36 and @xmath539 are not homogeneous on @xmath531 and so do not descend down to @xmath452 .
let @xmath540 be the a rank - two twistor ( @xmath237-plane ) distribution on @xmath167 , such that the three dimensional space of leaves of the rank - three distribution @xmath541 in @xmath167 is @xmath117 .
therefore @xmath542 , and the matrix components of @xmath543 are homogeneous functions on the fibres of @xmath167 , with the coefficients given by functions on @xmath31 .
let us rewrite this condition as [ sparling_1 ] _
a ( cc v_-m & r_-n + w_-m & s_-n ) = ( cc v_-m & r_-n + w_-m & s_-n ) ( cc _ a & _ a + _ a & _
a ) , h= ( cc v_-m & r_-n + w_-m & s_-n ) . here
@xmath544 and @xmath545 denote a pair of local sections of @xmath510 , where @xmath546 are homogeneous of degree @xmath547 and @xmath548 are homogeneous of degree @xmath549 when regarded as functions on the fibres of @xmath201
. the polynomials @xmath550 will give rise to potentials for _ higher spin fields _ on @xmath31 , i.e. sections of various powers of @xmath201 satisfying some field equations @xcite . to construct these fields , and find the corresponding field equations rewrite ( [ sparling_1 ] ) as @xmath551 therefore , as @xmath552 is homogeneous of degree one
, we conclude that @xmath553 and @xmath554 are homogeneous of degree one , @xmath555 is homogeneous of degree @xmath556 , and finally @xmath557 .
equivalently @xmath558 where @xmath559 the @xmath183 irreducible components give rise to two spin-2 fields , two functions , one spin-@xmath560 field and one spin @xmath561 field .
@xmath562 to find the field equations satisfied by these potentials , we shall impose the integrability conditions on ( [ sparling_2 ] ) arising from @xmath563 .
contracting each equation in ( [ sparling_2 ] ) with @xmath564 and using ( [ sparling_2 ] ) to eliminate the derivatives of @xmath36 gives @xmath565 decomposing this last set of equations into irreducible parts yields the final system ( which is the non relativistic version of the sparling equations @xcite ) @xmath566 where @xmath567 and @xmath568 .
the first two equations give a pair of abelian monopoles @xmath569 and @xmath570 satisfying @xmath571 thus @xmath62 and @xmath108 are two harmonic functions , which give rise to electric and magnetic fields in ( [ cor_con ] ) by @xmath572 if the gauge group is taken to be @xmath183 , then @xmath573 .
let @xmath18 be a twistor space corresponding to a flat space - time with the speed of light @xmath20 .
thus , if @xmath260 , then @xmath18 is the total space of a rank - two vector bundle @xmath574 , and @xmath117 is the @xmath357 newtonian twistor space .
as a special case of the construction presented in the previous section , consider a vector bundle @xmath575 , such that @xmath576 where @xmath279 is the holomorphic normal bundle to @xmath577 .
let @xmath578 be the canonical bundle of @xmath18 , so that @xmath579 : this true both for finite , and infinite @xmath20 ( as discussed in section [ tf_def]).define a rank - two vector bundle @xmath580 if @xmath260 , then @xmath581 is trivial when restricted to twistor lines , and its ward correspondence gives a solution to anti - self - dual yang - mills field on the complexified minkowski space . in the newtonian case
@xmath582 corresponds to @xmath583 in ( [ sparling_1 ] ) .
the resulting potentials are two abelian monopoles , one spin-4 field and one spin-2 field subject to equations ( [ s_equations ] ) .
one can now identify the field strengths of these potentials with various components of the newton - cartan connection .
note that it is consistent to set @xmath584 .
[ sec_schrodinger ] it is known ( see e.g. @xcite ) that the free schrdinger equation in @xmath585 space
time dimensions arises as a null reduction of the wave equation in @xmath586 dimensions .
we shall use this observation together with the twistor contour integral formula ( [ penrose_form ] ) to construct a contour integral formula for a @xmath587dimensional schrdinger equation . consider the wave equation ( [ wave_eq ] ) which we shall write as @xmath588 .
let @xmath589 be null coordinates , so that @xmath590
. then [ dirac_ansatz ] ( x , y , u , v)=e^-(x , y , u ) satisfies @xmath588 iff [ schrodinger ] i=-(+ ) which is the free schrdinger equations in @xmath591 dimensions . now consider the integral formula ( [ penrose_form ] ) , where @xmath592 is a twistor cohomology class restricted to a twistor line ( [ 4d_incidence ] ) .
the ansatz ( [ dirac_ansatz ] ) implies that @xmath593 , where @xmath594 . solving this equation for @xmath217 , and substituting back into ( [ penrose_form ] ) yields the contour integral formula [ schro_cont ] ( x , y , u)=_^1 e^-mi(x - iy ) g(x+iy+u , ) d , where @xmath180 is an element of @xmath595 restricted to a line @xmath596 in @xmath597 .
the complex two fold @xmath394 is the twistor space of the flat holomorphic projective structure on a two dimensional space @xmath598 .
if @xmath599 ( with @xmath600 ) are local coordinates of a point @xmath601 , then the corresponding line @xmath602 is given by @xmath603 , where @xmath604 is a coordinate on the fibres of @xmath605 .
the formula ( [ schro_cont ] ) is also valid if @xmath606 , where the real and imaginary parts of @xmath607 are harmonic functions on @xmath608 depending on a parameter @xmath609 . in this case ( [ schro_cont ] ) is the radon transform of @xmath180 .
let us give an example , where @xmath610 .
let @xmath611 , and let @xmath218 be a circle centred at the origin of the @xmath284plane .
then @xmath612 which is indeed a solution of the @xmath587 schrdinger equation ( [ schrodinger ] ) .
the contour integral formula ( [ schro_cont ] ) is , on dimensional grounds , only tangentially related to the rest of this paper where the non
relativistic theories in @xmath316dimensions are considered ( hence its place in the appendix ) .
it may however be relevant in newton
cartan theories in @xmath587 dimensions @xcite .
[ sec_app ] in the original nonlinear graviton construction @xcite , and its modification involving non zero cosmological constant @xcite the spin connection on @xmath613 can be constructed directly from the ward correspondence applied to a certain rank - two sub - bundle of the tangent bundle of the twistor space .
this construction is mentioned in @xcite , but not implemented explicitly .
below we show how carry the construction over in a way , which at the linearised level , agrees with the spin connection arising for plebanski 2nd heavenly equations @xcite .
let @xmath614 be the holomorphic canonical line bundle of the twistor space which restricts to @xmath395 on twistor lines .
if @xmath238 corresponds to an asd einstein metric ( with or without @xmath406 ) , then there exists a one - form @xmath615 ( which is given by @xmath616 in the vacuum case . )
this one - form defines a rank - two sub - bundle @xmath617 consisting of vectors annihilated by @xmath615 . the bundle @xmath618 restricts to @xmath386 on each twistor line .
the bundle @xmath619 is therefore trivial on twistor lines , and ( by the standard ward transform @xcite ) corresponds to an asd gauge field : a spin connection on @xmath166 .
this agrees with the gauge field arising from the sparling equation in section [ sparling_section ] .
both spin bundles have a twistorial construction as @xmath620 where @xmath49 is a twistor curve corresponding to @xmath332 .
assume that the asd metric on @xmath16 is vacuum , and so the twistor space fibres over @xmath7 .
let @xmath621 . to construction
the connection on @xmath166 consider the inifinitesimal deformation [ ap_ham ] ^a=^a+ and let @xmath622 , so that @xmath623 .
the relation @xmath624 gives @xmath625 to construct the splitting @xmath626 consider @xmath627 so that , to the first order in @xmath122 , @xmath628 the freedom in splitting the rhs is measured by elements of @xmath629 . choosing a constant spinor @xmath366 yields @xmath630 the usual liouville argument gives @xmath631 for some matrix @xmath632 which does not depend on @xmath633 .
equivalently @xmath634 therefore the linearised ward transform gives a connection @xmath635 this is consistent , in the linearised 2nd plebanski gauge , with the expression for the asd weyl spinor as @xcite @xmath636 ( in plebanski s gauge the quadratic term in the connection on @xmath166 contracts to zero , and the connection on @xmath167 vanishes . ) . | we develop a non relativistic twistor theory , in which newton
cartan structures of newtonian gravity correspond to complex three manifolds with a four parameter family of rational curves with normal bundle @xmath0 .
we show that the newton cartan space - times are unstable under the general kodaira deformation of the twistor complex structure . the newton
cartan connections can nevertheless be reconstructed from merkulov s generalisation of the kodaira map augmented by a choice of a holomorphic line bundle over the twistor space trivial on twistor lines .
the coriolis force may be incorporated by holomorphic vector bundles , which in general are non
trivial on twistor lines .
the resulting geometries agree with non relativistic limits of anti - self - dual gravitational instantons . |
einstein s theory of the special relativity establishes that the velocity @xmath0 of the light in vacuum is an invariant under a reference frame change @xcite .
superluminal single objects under lorentz transformations lead to violation of relativistic causality principle , and in turn , to the paradox of effect preceding its cause @xcite .
nevertheless , many experiments with faster - than-@xmath0 light propagation were done and discussed @xcite .
in fact , special relativity theoretical framework stands even without assuming that @xmath0 is the highest possible speed @xcite .
research on superluminality refers mostly to barrier tunneling by radiation pulses @xcite or by single photons @xcite , and to active media crossing @xcite .
barrier tunneling of light pulses is substantially governed by very low transmission coefficients and by an almost linear spectral component time delay @xmath1 .
some authors state that the outgoing light pulse after crossing the barrier is so much weaker than the incoming one ( or photon crossing probability so scarce ) that any possible information carried by the pulse is destroyed , therefore causality principle is not broken down @xcite .
others @xcite maintain that no propagation can occur inside the barrier , hence it is not the case to speak of advancing velocity . in optical pulse propagation experiments within the so - called fast - light media , that is media with anomalous dispersion @xcite ( precisely , with gain - assisted linear anomalous dispersion ) pulse shape is preserved and phase varies almost linearly with frequency in the region of interest .
the graph slope @xmath2 leads to a group velocity @xmath3 which exceeds the speed of light in vacuum and can even become negative @xcite . a superluminal experiment carried out with a microwave pulse crossing a birefringent two - dimensional crystal resulted in a clear superluminal group velocity .
this was measured using the interference of pulses which had traveled along the two crystal axes @xcite set in such a way that pulse polarization of the incident and detected fields relative to the crystal fast axis could be well controlled .
the experimental results on these systems renewed the debate about superluminal propagation and information velocity .
the discussion focuses on the concept that the speed of a light pulse crossing a medium is not precisely defined because a pulse is an ensemble of optical components traveling at different and well defined phase velocities @xmath4 , where @xmath5 is the refractive index of the optical material at a given frequency .
the pulse peak usually travels at the group velocity @xmath6 where @xmath7 is the group index and @xmath8 is the wavepacket central frequency @xcite .
the wave nature of the wavepacket allows superluminal light propagation .
arguments concentrate on the fact that @xmath3 does not coincide with the information velocity @xmath9 and there is a debate about the nature of these velocities @xcite . since the analysis of the problem by sommerfield and brillouin it is discussed that the `` front '' velocity of a square pulse does not exceed @xmath0 , and refs . @xcite
suggest a non
analytic point of the pulse amplitude as transporting information , by observing that this is a generalization of the `` front '' point of the pulse . in all experiments carried out so far ,
the temporal forward shift of the pulse or of the single photon wavepacket is much smaller than the total length involved , and this necessarily poses interpretation problems . in this respect ,
the definition of the information velocity as the propagation speed of a particular point in the profile @xcite leads to information velocities always less or equal to @xmath0 .
experiments show that the characteristic of a light pulse for providing superluminal effects is its nature of being a superposition of monochromatic waves . within this view , the possibility of a superluminal effect with a single photon lies on the fact that a photon is always a superposition of monochromatic fock states @xmath10 ( encompassing a frequency bandwidth due to heisenberg principle and the fact that the photon is generated in a definite space region ) .
we would like to underline that in quantum mechanics any single particle is a superposition of many states ( another example , a moving electron is a superposition of momentum eigenstates @xmath11 ) , even if they are detected in laboratory as a single event ( _ i.e. _ , a single _ click _ ) .
therefore , quantum mechanics allows superluminal propagation of single particles .
we are going to present and discuss an experiment where a single photon , created in a non - linear crystal by the down - conversion of a cw laser light @xcite , is operated through its spectral components in such a way that a clear temporal shift with respect to the
non - acted upon photon is detected via an interferometric measurement .
this experiment pertains to the class of superluminal experiments . the way we carry out the experiment ( acting upon one of the two generated photons only ) does not allow us to claim strictly that we are operating with one single photon wavepacket .
the description of the interference between optical components can be also done by viewing the observed light temporal shift as a result of the interference of very weak light beams with femtosecond coherence length .
anyway , for simplicity and convenience , we will use the view of running photons . in our experiment
the photon velocity can become apparently superluminal as a result of the interference of photon optical components whose phases are acted upon by the optical system described below . because of the interference process , the observed result is not at odds with causality . in order to avoid issues about the kind and the physical meaning of the different definitions of velocity
, we set the experiment in such a way that the shift between the normal and `` superluminal '' photons is notably wider than the width of the corresponding coherence length .
sketch of the experimental apparatus .
laser : a diode laser of 405 nm wavelength , 40 mw power .
bbo : beta - barium - borate nonlinear crystal .
hwp : half wave plate .
g1 , g2 : gratings with 1200 lines / mm .
l1 , l2 : lenses with f = 100 mm .
slm : a liquid crystal spatial light modulator with 640 pixels .
c1 , c2 : calcite crystals .
p : polarizer oriented at 45@xmath12 .
d : optic coupler + multimode optic fiber + single photon detector . ]
in figure [ f : apparato ] we show the experimental device consisting , in sequence , of a cw pump laser , a non - linear bbo crystal that generates photons via parametric down - conversion , a half - wave - plate ( hwp ) for inducing a polarization rotation , a 4f - system with a phase mask ( a slm , _ spatial light modulator _ ) in the middle providing a time delay among optical components of the horizontally polarized photon beam @xcite , and finally an interferometer followed by a single photon detector for measuring the time delay .
a pair of photons , usually called _ signal _ and _ idler _ photons , is generated by a parametric non - collinear down - conversion @xcite of a cw 40 mw 405 nm almost monochromatic laser radiation within a non - linear 3 mm thick bbo crystal .
the state of this photon pair can be written as @xcite @xmath13 where @xmath14 is the frequency shift with respect to the central frequency @xmath8 and @xmath15 indicates the horizontal polarization .
the function @xmath16 , defined in @xcite , gives the probability amplitude .
the signal photon enters the 4f - system , where the required linear delay @xmath17 is generated by the mask .
the 4f - system consists of two gratings with 1200 lines / mm and of two 100 mm focal lenses . by means of this device
the photon spectral components are spatially dispersed in a linear way by the first grating , and then focused onto the liquid crystal mask array of pixels ( our mask is composed by 640 pixels 100 @xmath18 m wide ) capable of setting the relative phases almost at will . finally the optical components are again synthesized by the second grating .
the transmission coefficient of the 4f - system depends on the efficiency of the two gratings , and in our apparatus it is around 50% .
this device is more flexible than the fast - light medium and , more importantly , allows the manipulation of each single optical component . in this context
we observe that our setup allows us to act separately , almost at will , upon the photon optical component phases , contrary to all other experiments in the literature where a dispersion law is imposed by a medium . in our case ,
the entering light is opened up along a plane by means of the fourier components spatial expansion , while in the other experiments the components were forced to propagate along the same original line within the acting medium .
now we analyze the propagation of the signal photon along the experimental apparatus . a half - wave - plate ( hwp ) set in front of the the 4f - system
rotates the photon polarization of a suitable angle @xmath19 ( see below ) , hence the state of the entering photon is changed into a superposition of a horizontal and a vertical polarized states : @xmath20 then the signal photon propagates through the 4f - system . the spectral phase function that we introduce is applied only on the horizontal polarization , while the vertical polarization experiences only the delay due to the transit through the mask pixels , becoming therefore our time reference . considering only the path sections which have different optical thicknesses for h and v polarizations , we obtain the two phase variations : @xmath21 where @xmath22 is the spectral phase function imposed by mask pixels . in our case
we introduce a linear function @xmath23 , with @xmath24 a constant parameter .
the times @xmath25 and @xmath26 are the time delays due to the pixel slab crossing .
we found experimentally that @xmath27 fs .
incidentally , in the setting of the diagnostic system we also had to take into consideration the fact that the two transmission coefficients @xmath28 and @xmath29 of the 4f - system are different ( this is due to the different transmission efficiencies of the gratings for the two polarizations )
. the interferometer placed after the 4f - system to detect the signal photon at the optical system output is made by two calcite crystals , an hwp and a polarizer set at 45@xmath12 .
this device , described in @xcite , interchanges the two polarizations and causes a time delay which can be changed simply by rotating the second crystal .
we also have to take into account a certain dispersion introduced by the crystals because they are relatively long .
however this dispersion , described by the parameter @xmath30 , can be assumed to be nearly equal for the two paths with a very good approximation . the photon state propagation within crystals
is then described by the following spectral phase variation : @xmath31 the last step to be analyzed is the propagation through the polarizer placed at @xmath32 .
this is a crucial element in the photon detection .
in fact , the polarizer mixes up the h and v polarization states with the result that the two states interfere and a pattern of fringes within the coherence length are created . summing up the state evolution along the entire path ,
the signal component @xmath33 at the output read @xmath34 | 45^o , \omega \rangle \ , = \ , a(\omega ) | 45^o , \omega \rangle \hspace{0.3 cm } & \end{aligned}\ ] ] the probability of having one count relative to the signal photon , ignoring the idler one , is given by the trace of the density matrix @xmath35 where : @xmath36 after some mathematics we get @xmath37\end{aligned}\ ] ] where @xmath38 is obtained with a proper rotation of the hwp set in front of the 4f - system , and @xmath39 is the time delay introduced by the interferometer . the interference records of the advanced ( b ) and retarded ( c ) photon - wavepackets , referred to the reference @xmath40 ( a ) .
the top frame shows an expanded view of the interference fringes . ]
this result accounts for our experimental data shown in figure [ f : misure ]
. curve ( a ) is the reference case @xmath40 ( which sets the origin of the time scale ) .
curves ( b ) and ( c ) present a lead and a lag of @xmath41 fs , respectively .
all data show that the interference fringes occur within a coherence length of @xmath42 .
considering the transit time of the photon light from the source up to the end of the apparatus ( that is looking at the optical system as a long black - box ) , we observe that the horizontal and vertical polarized parts take different time intervals . the two parts travel at the same velocity within the vacuum sections and within the lenses ( which are isotropic )
, hence we may say that their velocities are different within the mask slab of @xmath43 m thickness . using the delay between v and h states measured at the end of the apparatus
, we may define the group velocity of the horizontal polarization as @xmath44 where @xmath45 is the group velocity of the vertical polarization , derived from the manufacturer mask characteristics .
the group velocity defined in this way would result greater than @xmath0 when @xmath24 is lower than @xmath46 fs and even negative for @xmath47 fs .
we must observe that this overall view of the light transmission would raise problem with respect to the causality principle @xcite , because of a photon propagation mathematically superluminal , and does not consider the real physics of the phenomenon .
the overall result is readily explained by following the spatially sectioned sub - light - packets crossing the mask pixels .
each one of these sub - packets has the limited spectrum selected by the pixel dimension .
that portion of the spectrum corresponds to a coherence length of 3 ps .
these sub - packets have subluminal velocity in every part of the device , including the mask . according to this view , the recombination of the sub - packets on the second grating leads to either the forward or backward shift of the photon with respect to the non - acted photon state , depending on the setting of the component phases .
this superluminal effect was already observed in refs .
@xcite .
the question of the information velocity in our experiment does not fit either the discussion presented so far or the debate in progress about the matter , that is , within the models of pulse reshaping and consideration of peculiar points of the pulse ( such as front or non - analytic ones ) . in our experiment photon reconstruction may occur within the entire 3 ps coherence length of a sub - wavepacket , which means a shift backwards or forwards of the reconstructed envelope which is very far from the tail of the reference one .
an observation is in order : the delay introduced by the lenses , which is about 30 ps , is larger than the 3 ps of maximum advance allowed , and this does not allow direct measurement of `` superluminality '' downstream the 4f - system .
however , in principle , one could substitute the refractive optics with parabolic mirrors @xcite , this way eliminating the causes of the delay .
the propagation of the spectral sub - light - packets crossing the pixels is certainly in agreement with causality .
in fact , considering kramers - kronig relations for the sub - packets , we can represent the evolution by means of a green function that satisfies the requirements of causality . for different pixels they are independent of one another ,
so the phase of each pixel can be programmed at will .
there is no contradiction , then , in saying that the propagation is causal , although the photon moves far ahead its coherent length . from this analysis
one can infer that the relevant time is not the coherence time of the photon , but the coherence time of the sub - packet that reaches the single pixel .
we have performed an experiment on a superluminal shift wider than the coherence length , hence more noticeable than those observed in all other experiments carried out so far .
the overall result of figure [ f : misure ] indicates clearly that we have induced a large superluminal group velocity on the radiation traveling inside the apparatus .
our experiment can also be described by considering a single photon propagating within the apparatus .
the superluminal group velocity is such that the preserved photon envelope shows up at distances from the vacuum site which are much larger than the photon coherence length , a result that is not possible with the other experimental layouts .
this result was obtained with an optical system capable of managing the single component phases of the radiation independently , at variance with all other previous experiments .
we have shown that our observations are consistent with the principle of causality even if the nominal group velocity is highly superluminal . by considering this experiment extendable to single photons , we observe that the results would have a physical content different in essence with respect the complementary result obtained with sub - picosecond laser pulses @xcite .
in fact , while the detection of the propagation speed of particular points of a light pulse profile ( as for instance the front edge ) can be in principle experimentally measured , it can not be considered in the case of a single photon because a point within its wavepacket profile is meaningless , and represents only the probability amplitude of obtaining a clic ( _ i.e. _ of detecting the photon ) .
the superluminality experiment with single photons could be carried out neatly thanks to the exploitation of the particular technique of the spatial light modulator which allowed to manage the spectral components while substantially maintaining their amplitudes .
we wish to acknowledge fruitful discussions with c. maroli and s. olivares , and the extended help in the experimental work by a. schiavi .
we would like to remark the support by m. giammarchi .
one of the authors thanks v.s .
olkhovsky for stimulating interest in the subject .
the research was partly supported by european contract rii3-ctphi506395care .
einstein a 1905 _ ann .
* 17 * 891 einstein a , lorentz h a , minkowski h and weyl h 1952 _ the principle of relativity , collected papers _ ( dover , new york )
brillouin l 1960 _ wave propagation and group velocity _ ( academic press , new york ) garrison j c ,
mitchell m w , chiao r y and bolda e l 1998 _ phys .
lett . _ a * 245 * 19 longhi s , marano m , laporta p and belmonte m 2001 _ phys .
_ e * 64 * 055602 spielmann ch , szipcs r , stingl a and krausz f 1994 _ phys .
* 73 * 2308 steinberg a m , kwiat p g and chiao r y 1993 _ phys .
lett . _ * 71 * 708 wang l j , kuzmich a and dogariu a 2000 _ nature ( london ) _ * 406 * 277 and references therein von ignatowsky w a 1910 _ verh . deutsch .
_ * 12 * 788 mark j and leblond l 1976 _ am .
j. phys . _
* 44 * 271 lugiato l a and gorini v 1972 _ j. math .
phys . _ * 13 * 665 olkhovsky v s and recami e 1992 _ phys .
rep . _ * 214 * 339 wang l - g , liu
n - h , lin q. and zhu s - y 2003 _ phys .
_ e * 68 * 066606 and references therein solli d r , mccormick c f , ropers c , morehead j j , chiao r y and hickmann j m 2003 _ phys .
lett . _ * 91 * 143906 winful h g 2003 _ phys .
lett . _ * 90 * 023901 brunner n , scarani v , wegmuller m , legr m and gisin n 2004 _ phys .
lett . _ * 93 * 203902 agrawal g p 2001 _ nonlinear optics _ ( academic press , san diego ) stenner m d , gauthier d j and neifeld m a 2003 _ nature ( london ) _ * 425 * 695 hong c k and
mandel l 1985 _ phys .
rev . _ a * 31 * 2409 wefers m m and nelson k a 1994 _ j. opt .
b * 12 * 1343 weiner a m 2000 _ rev .
instrum . _
* 71 * 1929 cialdi s and boscolo i 2004 _ nucl . instrum .
methods _ a * 526 * 239 joobeur a , saleh b e a , larchuk t s and teich m c 1996 _ phys .
rev . _ a * 53 * 4360 cialdi s , castelli f , boscolo i and paris m g a 2008 _ appl .
opt . _ * 47 * 1832 gogo a , snyder w d and beck m 2005 _ phys . rev . _ a * 71 * 052103 peatross j , glasgow s a and ware m 2000 _ phys .
* 84 * 2370 reitze d h , weiner a m , and leaird d e 1992 _ appl .
_ * 61 * 1260 | we present experimental results relative to superluminal propagation based on a single photon traversing an optical system , called 4f - system , which acts singularly on the photon s spectral component phases .
a single photon is created by a cw laser light down conversion process .
the introduction of a linear spectral phase function will lead to the shift of the photon peak far beyond the coherence length of the photon itself ( an apparent superluminal propagation of the photon ) .
superluminal group velocity detection is done by interferometric measurement of the temporal shifted photon with its correlated untouched reference .
the observed superluminal photon propagation complies with causality .
the operation of the optical system allows to enlighten the origin of the apparent superluminal photon velocity .
the experiment foresees a superluminal effect with single photon wavepackets . |
the coupling of complex magnetic field and plasma generates variety of magnetohydrodynamic ( mhd ) waves and oscillations in various solar structures .
these mhd waves and oscillations are one of the important candidates for coronal heating and solar wind acceleration .
the idea of exploiting observed oscillations as a diagnostic tool for determining the physical conditions of the coronal plasma was first suggested by roberts et al .
( 1984 ) . until recently
, the application of this idea has been marked by the lack of high - quality observations of coronal oscillations .
however , this situation has changed dramatically , especially due to space - based observations by the solar and heliospheric observatory ( soho ) , the transition region and coronal explorer ( trace ) , and most recently with the high resolution spectra from the hinode spacecraft .
the fast kink wave is most frequently observed mode as it can be directly detected by periodic spatial displacement of coronal loop axis ( nakariakov et al .
1999 ; aschwanden et al . 1999 ; wang & solanki 2004 ) .
using temporal series image data from cds / soho , recently oshea et al .
( 2007 ) have found the first evidence of fast - kink standing oscillations in the cool transition region loops as well . on the other hand
, the fast mhd sausage wave causes the variation of pressure and magnetic field in a coronal loop and therefore can be observed as intensity oscillations or periodic modulation of coronal radio emission ( nakariakov et al .
recently , srivastava et al . (
2008 ) have reported the signature of the leakage of chromospheric magnetoacoustic oscillations into the corona near the south pole .
these observations provide a basis for the estimation of coronal plasma properties ( nakariakov & ofman 2001 ) .
recently , verwichte et al . (
2004 ) have detected interesting phenomenon of simultaneous existence of fundamental and first harmonics of fast kink oscillations ( see also de moortel & brady 2007 ; van doorsselaere et al .
however , the ratio between the periods of fundamental and first harmonics @xmath3 was significantly shifted from 2 , which later was explained as a result of longitudinal density stratification in the loop ( andries et al .
2005 ; mcewan et al . 2006 ) .
the rate of the shift allows to estimate the density scale height in coronal loops , which can be few times larger compared to its hydrostatic value ( aschwanden et al .
2000 ) . in this paper
, we report the first evidence of fundamental and first harmonics of fast sausage oscillations in cool postflare loop .
we use high spatial ( 1.3 arc sec ) and temporal ( 5 and 10 s ) resolution h - alpha observations from the 15 cm solar tower telescope at aries , nainital , india , to study the oscillations in the relative intensity . using the standard wavelet tool
, we find multiple oscillation periods at different parts of the loop , which are interpreted as the fundamental and first harmonics of fast sausage mode . in section 2
, we describe the observations and data reduction .
we describe the wavelet analysis in section 3 . in section 4 , we present our theoretical model and its link to coronal seismology .
we present results and discussion in the last section .
the observations of post flare loops have been carried out with 15 cm , f/15 coud solar tower telescope at aries , nainital , india , equipped with bernhard hale h - alpha filter ( @xmath4 = 6563 and p.b .
0.5/0.7 ) , and pxl photometric ccd camera .
the filter is a birefringent lyot type filter , tunable @xmath5 1 from central wavelength ( 6563 ) with the step of 0.1 . during our observations ,
we set it at central wavelength 6563 with passband 0.5 . the image size has been enlarged by a factor of two using a barlow lens .
the 512 @xmath6 512 pixels , 12-bit frame transfer ccd camera has a square pixel of 15 _ @xmath7m@xmath8 _ corresponding to a 0.65 arc sec pixel size .
the spatial resolution of our observations is 1.30 arc sec .
the read out noise for the system is 31 e@xmath9 with a gain of 44.1 e@xmath9/adu .
dark current of the camera is 56 e@xmath10 .
the dark current is integrated over the 512 x 512 pixels .
the camera controller of the system has a variable read - out rate from 0.5 to 2 mhz.the variable read out rate of our ccd provides us the facility of fast imaging of the flares at different rates . in these observations ,
we have used a constant read out rate of 2 mhz .
the the ccd chip ( eev 37 ) is cooled up to -25@xmath11 c using liquid cooling system .
200 pixel ( or 130 arc sec @xmath6 130 arc sec ) ( left panel).the aligned soho / eit fe ix / x 171 image of the same loop system is presented in right panel.,title="fig:",width=226 ] 200 pixel ( or 130 arc sec @xmath6 130 arc sec ) ( left panel).the aligned soho / eit fe ix / x 171 image of the same loop system is presented in right panel.,title="fig:",width=222 ] we analyze the postflare loop system between 01:00:52 u.t . and
01:58:28 ut on may 2 , 2001 , at a cadence of 5 s and 10 s with an exposure time of 30 ms .
the post flare loop system was associated with m1.8 class limb flare which occured in noaa active region ar 9433 ( n15 , w88 ) . since , the temporal evolution / changes were very fast during the onset of post flare loop system , we took fast sequence of images with 5 s cadence . as time progresses in the gradual / decay phase , the temporal changes become slow compared to initial phase of post flare loop system .
so , we took the sequence of images with 10 s cadence .
hence , there are two cadences , initially 5 s and later on 10 s. we have started our fast imaging with 5 s cadence on 00:52:11 ut . in the beginning of our observations ,
the flare was in decay phase and was enveloped by complex post flare loop system .
hence , we were unable to select the distinct loop initially .
the post flare loop system became more relaxed and distinctly visible around 01:00:52 ut , then we have selected images of a clearly visible loop for our study after 01:00:52 ut . in our analysis , first 92 frames were with 5 s cadence , while rest of the frames were with 10 s cadence .
the total number of the frames used in this study is 390 .
we find that the post flare loop system shows the similar structure / evolution in soho / eit ( 171 , 195 , 284 , 304 ) and h - alpha observations.the post flare loop system which was visible at higher temperature ( 6.0@xmath610@xmath12 k3.0@xmath610@xmath13 k ) with eit , has also seen in h - alpha at the lower temperature ( @xmath0 10@xmath12 k ) after its cooling . in fig 1 ( top panel ) , the image of the post flare loop system in high resolution h - alpha has been shown , while the same loop system observed with soho / eit fe ix / x 171 has been shown in the bottom panel.both images show the clear shape of the dynamic post flare loop that we have selected . the loop top and foot point of this prominent loop marked by arrows in these two images .
since the spatial resolution of our h - alpha observations is high in comparison to the soho / eit , hence we can not see the features in eit as sharp as in h - alpha image .
we have estimated the length of the selected loop in h - alpha and soho / eit fe ix / x 171 images .
we find the lengths @xmath2 97 mm and @xmath2 100 mm in the h - alpha and euv images respectively .
the estimated loop lengths at two wavelengths are almost equal which clearly shows the presence of the same loop at different temperatures
. the lower atmosphere may not necessarily be an optically thin atmosphere , and many bright structures cross to each other in the field of view .
however , the chromospheric post flare loop system is very complex and dynamic in itself , instead of simpler and static .
the features are temporally changing very fast in the observations .
so , we carefully selected the image frames of the clearly visible loop in order to study the oscillations .
the sky condition was very good during the observations .
several dark and flat field images have also been taken to calibrate the h - alpha images . using iraf and sswidl softwares ,
the image processing has been performed .
the details of the instrument can be obtained from ali et al .
( 2007 ) and joshi et al . (
we use the morlet wavelet tool to produce the power spectrum of oscillations in h - alpha .
the morlet wavelet suffers from the edge effect of time series data .
this effect is significant in regions defined as cone of influence ( coi ) .
the details of wavelet procedure , its noise filtering , coi effects etc . are given in torrence & compo ( 1998 ) .
randomisation technique evaluates the peak power in the global wavelet spectrum , which is just the average peak power over time and similar to a smoothed fourier power spectrum .
this technique compares it to the peak powers evaluated from the @xmath14 equally likely permutations of the time series data , assuming that n values of measured intensities are independent of n measured times if there is no periodic signal .
the proportion of permutations which gives the value greater or equal to the original peak power of the time series , will provide the probability of no periodic component ( p ) . the percentage probability of periodic components presented in the data will be @xmath15 , and 95 % is the lowest acceptable probability for real oscillations .
we set 200 permutations for the reliable estimation of p , and hence the probability of real oscillations .
the details of randomisation technique to obtain the statistically significant real oscillation periods are given by nemec & nemec ( 1985 ) and oshea et al .
we do not remove any upper / longer period intervals and associated powers of our time series data during wavelet analysis .
we choose the running average option of oshea s wavelet tool , and smooth the original signal by window width 60 .
average smoothing process is used to reduce the noise in the original signal in order to get a real periodicity , and it is based on the low pass filtering methods .
the running average process smooth the original signal by the defined scalar width .
this process use the smooth subroutine available with idl tool kit .
the smooth function returns a copy of array smoothed with a boxcar average of the specified width .
the result has the same type and dimensions as array .
the fitted signal is then subtracted from the original signal , and gives the resultant signal for the wavelet analysis .
the maximum allowed period from coi , where edge effect is more effective , is 1259 s. hence , the power reduces substantially beyond this threshold . in our wavelet analysis
, we only consider the power peaks and corresponding real periods below this threshold .
we performed the wavelet analysis at different parts ( near the apex and foot point ) of the selected loop .
we have chosen a location of _ ( x , y ) _
= ( 133th pixel , 117th pixel ) near the apex , where the dominant periodicity is @xmath0 587 s with the probability of @xmath0 99 % 100 % ( fig .
we have again chosen a location of _ ( x , y ) _
= ( 58th pixel , 104th pixel ) near the foot point of the same loop , where the dominant oscillation period is @xmath0 349 s with the probability of @xmath0 99 % 100 % ( fig .
it should be noted that we have chosen a box of width 4 at these locations to extract the light curves with good signal - to - noise ( s / n ) ratios .
boxes with the width 4 have been chosen near the loop - top and foot point with respect to their reference coordinates as mentioned in the figure captions of the wavelet diagrams .
we extract maximum intensity from the chosen box .
our box of width 4 fits the loop dimension best .
however , we have checked the results at other widths , e.g. , 2 , 3 , and 6. the boxes with widths 2 and 3 still remain inside the loop dimension .
however , the box with width 6 crosses the loop , which may not be appropriate especially near the limb where the background emissions and structures may be more effective compared to far off the limb . from the boxes of width 2 and 3
, we get the same global periodicities at maximum power as we have obtained in the case of box width 4. however , the power distribution in the intensity wavelet is slightly different for each case .
this is obvious because we extract lower intensity with smaller box and higher intensity with larger box .
we get a periodicity of @xmath2 493 s and @xmath2 587 s respectively near the loop foot point and loop apex using the box of width 6. the periodicity at apex is similar to the previous findings because the limb effects are not much far off the limb . the difference in the periodicity near loop foot point shows the out side effects near the limb , which came into action due to the larger box size compared to the loop dimension .
however , the similar periodicities ( @xmath2 349 near the foot point and @xmath2 587 near the apex ) with well fitted or smaller boxes ( of widths 2 , 3 , and 4 ) , clearly show that we are inside a well isolated loop structure .
however , we should take care of the dimension of the box during wavelet analysis especially near the limb.thus , the selected loop intensities show the oscillations with significantly different periods near the apex ( @xmath0 587 s ) and footpoint ( @xmath0 349 s ) respectively .
these observation may lead an interesting consequences for loop oscillation phenomena .
the intensity oscillation with the period of @xmath16 587 s likely represents the fundamental harmonic of either fast sausage ( nakariakov et al . 2003 ) or slow magnetoacoustic ( nakariakov et al . 2004 ) waves .
the loop length is estimated as @xmath17 97 mm with corresponding phase speed as @xmath18 330 km / s .
the estimated phase speed is much higher than the sound speed corresponding to cool h - alpha line .
therefore , we suggest that the oscillation is due to a fast sausage mode .
however , the dispersion relation of fast sausage waves in straight coronal loop includes both trapped and leaky modes depending on the loop parameters ( edwin & roberts 1983 ; nakariakov et al . 2003 ; aschwanden et al .
due to low sound speed , we can easily use the cold plasma approximation .
hence , the cut - off wavenumber @xmath19 is ( edwin and roberts 1983 ; roberts et al .
1984 ) @xmath20^{1/2}{j_0\over a},\ ] ] where @xmath21 and @xmath22 are alfvn speeds inside and outside the loop , @xmath23 is the loop radius and @xmath24 is the first zero of the bessel function @xmath25 .
the modes with @xmath26 are trapped in the loop , while the modes with @xmath27 are leaky .
we may assume @xmath28 , then the cut - off wavenumber is ( aschwanden et al .
2004 ) @xmath29^{1/2}{j_0\over a},\ ] ] where @xmath30 is the electron density ratio inside and outside the loop .
the width of selected loop can be roughly estimated as @xmath31 6 mm .
hence , the trapped global sausage mode can be realized only if the density ratio is @xmath32 thus , only very dense loop can support non - leaky global sausage mode in the estimated length and width ( nakariakov et al .
2003 ; aschwanden et al .
the postflare loops usually have a very high density contrast of the order of @xmath33@xmath34 .
the selected loop is cool post flare one , therefore the estimated density ratio falls in the expected range .
thus , the trapped sausage mode may occur in the loop , however , wave leakage can not be ruled out .
the fundamental harmonic of sausage oscillation has pressure antinode at the loop apex ( upper panel in fig .
4 ) . therefore
, it modifies the density leading to the intensity oscillation .
however , it has pressure node at the footpoints , and should not lead to remarkable intensity oscillation .
the observations show strong oscillations at the loop apex with the period of @xmath0 587 s and almost no indication of the oscillation with this period near foot point ( see fig .
2 ) . therefore ,
the oscillation with @xmath0 587 s is probably due to the fundamental mode of sausage oscillations . on the other hand ,
the first harmonic of sausage oscillation has pressure node at the loop apex and antinodes at the mid parts ( lower panel in fig .
4 ) . therefore
, the first harmonic should show the intensity oscillation near the foot point and almost no oscillation at the loop apex ( see fig .
3 ) . indeed
, the oscillation with the period of @xmath0 349 s has high probability near foot point and no significant probability near the apex .
therefore , we interpret the oscillation with the period of @xmath0 349 s as the first harmonic of fast sausage waves in the coronal loop . in the curved geometry , this mode is identical to sausage swaying mode ( daz et al .
2006 ) .
it must be mentioned that 349 s oscillation may be a consequence of nonlinear excitation from the fundamental harmonic , however , this possibility can be ruled out by two reasons : first , the amplitude of oscillation is not very strong and leaving a little space for nonlinear interaction ; secondly , the period of the first harmonic is not exactly twice of the period of fundamental mode , which was expected from the resonance condition . as a consequence of above discussions
, we suggest for the first observational evidence of the fundamental and its first harmonics of fast sausage oscillations in cool postflare loop .
the ratio between the periods of the fundamental and the first harmonics of sausage waves is proportional to @xmath1 1.68 , which is significantly shifted from 2 .
similar phenomena were observed for fast kink oscillations ( verwichte et al .
2004 ; de moortel & brady 2007 ; van doorsselaere et al .
the deviation of @xmath3 from 2 in homogeneous loops is very small due to the wave dispersion ( mcewan et al .
but longitudinal density stratification causes significant shift of @xmath3 from 2 ( andries et al .
2005 ; mcewan et al . 2006 ; van doorsselaere et al .
2007 ) .
using the results of andries et al .
( 2005 ) , we estimate the density stratification along the loop as @xmath35 1.8 , where @xmath36 is the density scale height .
the calculation of mcewan et al .
( 2006 ) gives the approximately same value .
thus , both calculation yields the same density scale height as @xmath2 17 mm for the loop length of 97 mm .
it must be mentioned that seismologically estimated scale height of 17 mm is larger comparing to the hydrostatical scale height at h - alpha temperature .
this can be explained by two reasons : first , it is possible that the medium is not in equilibrium inside postflare loops ; second , some mechanism ( e.g. , the variation of loop cross section with height ) other than the density stratification causes the deviation of @xmath3 from 2 .
recently , verth and erdlyi ( 2008 ) have studied the effect of magnetic stratification on loop transverse kink oscillations and found that the loop divergence may have the significant effect on the period ratio almost similar to the density stratification .
it would be interesting to study the effect of magnetic stratification on fast sausage oscillations .
however , even if the loop divergence effect reduces the estimated scale height to @xmath2 7 - 8 mm , it still remains larger than equilibrium scale height .
therefore , we suggest that the post flare loops are not in equilibrium , which may cause plasma motions along the loop . in the movie of the observations ,
we clearly see the unidirectional mass motion from western foot point to eastern foot point in the selected loop which may indicate its departure from hydrostatic equilibrium . however , it will be interesting to search further the observational evidence in high resolution space data .
the question is opened for further discussion .
it is not entirely clear from wavelet analysis that whether the global oscillation represents trapped or leaky mode .
we see in fig .
2 that the oscillation persists at least for 45 min , i.e. , for @xmath2 4 - 5 wave periods .
however , the relative intensity plot in fig .
2 ( upper panel ) likely shows that the oscillation amplitude decreases after 35 min .
as intensity oscillations are due to the variation in plasma density , hence the decrement may reflect the process of wave damping .
since the oscillations are in sausage mode , the resonant absorption is unlikely to occur
. therefore , the wave leakage in the surroundings , is the most probable candidate for the wave damping .
if the oscillation represents the global leaky mode instead of trapped one , then its phase speed would be close to the external alfvn speed ( pascoe et al .
2007 ) , which enables us to estimate its value as @xmath37 km / s .
however this estimation is done for the straight magnetic cylinder , while noting that the curvature probably enhances the wave leakage ( verwichte et al .
2005 ; daz et al .
2006 ; selwa et al .
2007 ) . for comparison , we may estimate the external alfvn speed from the wave leakage in curved magnetic slab ( daz et al .
2006 ) . using equation ( 25 ) of daz et al .
( 2006 ) , we may write the expression for the external alfvn speed as @xmath38 where @xmath39 is the damping time . from fig .
2 , we estimate the damping time as @xmath40 then using the density ratio of @xmath41 we estimate the value of the external alfvn speed as @xmath42 km / s ( note , that this estimation is done for a curved slab , therefore eq .
( 3 ) , which is obtained for a straight loop , is not valid here ) .
using high spatial ( 1.3 arc sec ) and temporal ( 5 and 10 s ) resolution h - alpha observations , we found intensity oscillations with different periodicity in different parts of cool postflare loop .
the intensity shows the oscillation with the period of @xmath0 587 s at the loop apex and with the period of @xmath0 349 s near loop footpoint .
we interpret the oscillations as signature of the fundamental and first harmonics of fast sausage mode .
it is difficult to say whether the oscillations are due to the trapped or leaky modes .
however , intensity plot likely shows the decrement of the oscillation amplitude , therefore we suggest the presence of leaky nature of these modes .
seismologically estimated alfvn speed outside the loop is @xmath2 300 - 330 km / s . using the period ratio @xmath3
, we also estimate the density scale height in the loop as @xmath2 17 mm .
this value is much higher than the equilibrium scale height at low h - alpha temperature , therefore we suggest that cool postflare loops are not in hydrostatic equilibrium . in the movie of the observations ,
we clearly see the unidirectional mass motion from western foot point to eastern foot point in the selected loop which indicates its departure from hydrostatic equilibrium . in conclusions , we report the first observational evidence of multiple oscillations of fast sausage modes in cool chromospheric postflare loop .
future detailed observational search should be investigated , especially using recent space - based observations .
aks wishes to express his gratitude to prof .
ram sagar ( director , aries ) for valuable suggestions and encouragements , and dr .
e. oshea for providing wavelet tool .
tvz acknowledges the georgian national science foundation for the financial support through its grant gnsf / st06/4 - 098.we wish to express our gratitude to the anonymous referee for his valuable comments which considerably improved our manuscript . | using simultaneous high spatial ( 1.3 arc sec ) and temporal ( 5 and 10 s ) resolution h - alpha observations from the 15 cm solar tower telescope at aries , we study the oscillations in the relative intensity to explore the possibility of sausage oscillations in the chromospheric cool postflare loop .
we use standard wavelet tool , and find the oscillation period of @xmath0 587 s near the loop apex , and @xmath0 349 s near the footpoint .
we suggest that the oscillations represent the fundamental and the first harmonics of fast sausage waves in the cool postflare loop .
based on the period ratio @xmath1 1.68 , we estimate the density scale height in the loop as @xmath2 17 mm .
this value is much higher than the equilibrium scale height corresponding to h - alpha temperature , which probably indicates that the cool postflare loop is not in hydrostatic equilibrium .
seismologically estimated alfvn speed outside the loop is @xmath2 300 - 330 km / s .
the observation of multiple oscillations may play a crucial role in understanding the dynamics of lower solar atmosphere , complementing such oscillations already reported in the upper solar atmosphere ( e.g. , hot flaring loops ) .
[ firstpage ] sun : chromosphere sun : loops mhd waves . |
[ [ section ] ] let @xmath10 be a simple lie algebra of simply - laced type @xmath11 , and let @xmath12 be a maximal nilpotent subalgebra . let @xmath13 be the canonical basis of the quantum enveloping algebra @xmath14 @xcite and @xmath15 the basis dual to @xmath13 .
when @xmath16 tends to 1 , these two bases specialize to bases @xmath17 and @xmath18 of @xmath19 and @xmath20 $ ] , respectively . here
@xmath21 stands for a maximal unipotent subgroup of a complex simple lie group @xmath22 with lie algebra @xmath10 .
let @xmath23 denote an indexing set for the simple roots of @xmath10 . given a finite - dimensional @xmath23-graded vector space @xmath24 with graded dimension @xmath25 , we denote by @xmath26 the corresponding nilpotent variety ,
see @xcite .
this variety can be seen as the variety of modules over the preprojective algebra @xmath27 attached to the dynkin diagram of @xmath10 , with underlying vector space @xmath24 @xcite . for a variety @xmath28
let @xmath29 be the set of irreducible components of @xmath28 .
lusztig has shown that there are natural bijections @xmath30
@xmath31 where @xmath32 ( resp .
@xmath33 ) is the subset of @xmath13 ( resp .
@xmath34 ) consisting of the elements of degree @xmath25 . kashiwara and saito @xcite proved that the crystal basis of @xmath14 can be constructed geometrically in terms of these irreducible components ( this was a conjecture of lusztig ) .
this paper is motivated by several problems about the bases @xmath13 and @xmath15 and their relations with the varieties @xmath26 and the preprojective algebra @xmath27 .
[ [ section-1 ] ] one problem , which was first considered by berenstein and zelevinsky @xcite , is to study the multiplicative structure of the basis @xmath15 .
two elements @xmath35 and @xmath36 of @xmath15 are called _ multiplicative _ if their product belongs to @xmath15 up to a power of @xmath16 .
it was conjectured in @xcite that @xmath35 and @xmath36 are multiplicative if and only if they @xmath16-commute .
we refer to this as the bz - conjecture .
the conjecture was proved for types @xmath37 and @xmath38 @xcite , and it also holds for @xmath39 @xcite .
more recently , marsh and reineke observed a strong relationship between the multiplicative structure of @xmath15 and properties of the irreducible components of the varieties @xmath26 .
they checked @xcite that for @xmath10 of type @xmath40 , if the irreducible components @xmath41 and @xmath42 are the closures of the isomorphism classes of two indecomposable @xmath27-modules @xmath43 and @xmath44 , then @xmath45 and @xmath46 are multiplicative if and only if @xmath47 .
this was verified by a case - by - case calculation , using the fact that for type @xmath40 the preprojective algebra is of finite representation type , that is , it has only a finite number of isomorphism classes of indecomposable modules @xcite .
they also calculated many examples in type @xmath39 and conjectured that this property still holds in this case ( note that @xmath27 is again representation - finite for @xmath39 ) . but a conceptual explanation was still missing .
let @xmath48 denote the subset of @xmath49 consisting of all @xmath27-modules @xmath50 isomorphic to @xmath51 with @xmath52 and @xmath53 .
it follows from a general decomposition theory for irreducible components of varieties of modules developed in @xcite that the condition @xmath54 for some @xmath55 is equivalent to @xmath56 being an irreducible component of @xmath49 . in @xcite counterexamples to the bz - conjecture
were found for all types other than @xmath4 with @xmath9 . in particular in type @xmath6 ,
for a certain @xmath24 of dimension 8 one can find an irreducible component @xmath57 of @xmath26 such that @xmath58 where @xmath59 and @xmath3 are two irreducible components of @xmath60 , see also @xcite .
this seems to be the smallest counterexample to the bz - conjecture in type @xmath61 .
moreover , it also shows that the result of marsh and reineke does not generalize to @xmath6 .
note however that the bz - conjecture was proved for large families of elements of @xmath34 @xcite .
for example , in type @xmath61 it holds for quantum flag minors , and the reformulation in terms of direct sums of irreducible components is also valid @xcite .
so one would like to get a better understanding of the relationship between multiplicativity of elements of @xmath15 and direct sum decompositions of irreducible components of varieties of @xmath27-modules .
[ [ section-2 ] ] another interesting problem concerns the singular supports of the simple perverse sheaves used by lusztig @xcite to define the canonical basis @xmath13 .
let @xmath62 be a dynkin quiver , which is obtained from the dynkin diagram of @xmath10 by choosing an orientation .
let @xmath63 be the affine space of representations of @xmath62 with underlying finite - dimensional @xmath23-graded vector space @xmath24 .
this is a finite union of isomorphism classes ( or orbits ) @xmath64 . in lusztig
s geometric construction , the elements of @xmath65 are given by the perverse extensions @xmath66 of the constant sheaves @xmath67 on the orbits @xmath64 . in @xcite lusztig considered the singular supports @xmath68 of these sheaves and showed that they are unions of irreducible components of @xmath26 ( independent of the chosen orientation of the dynkin diagram of @xmath10 ) .
he conjectured that in fact each @xmath68 is irreducible , equal to the closure @xmath69 of the conormal bundle of @xmath64 .
unexpectedly , kashiwara and saito @xcite produced a counterexample to this conjecture .
they exhibited two orbits @xmath70 for type @xmath6 such that @xmath71 the corresponding vectors @xmath72 and @xmath73 of @xmath13 have principal degree 16 , and apparently this is the smallest counterexample in type @xmath61 .
it turns out that this counterexample is dual to the counterexample above for @xmath15 , in the sense that @xmath74 and @xmath75 , see ( * ? ? ?
* remark 1 ) .
one motivation for this paper was to find an explanation for this coincidence .
[ [ section-3 ] ] what makes these problems difficult is that , although the canonical basis reflects by definition the geometry of the varieties @xmath76 , we want to relate it to the geometry of some other varieties , namely the irreducible components of the nilpotent varieties @xmath26 .
it is natural to think of an intermediate object , that is , a basis reasonably close to the canonical basis , but directly defined in terms of the varieties @xmath26 .
lusztig @xcite has constructed such a basis @xmath77 and called it the _ semicanonical basis_. this is a basis of @xmath19 ( not of the @xmath16-deformation @xmath14 ) which gives rise , like @xmath17 , to a basis in each irreducible highest weight @xmath78-module .
let @xmath79 denote the basis of @xmath20 $ ] dual to @xmath80 .
our first main result is the following : [ thma ] if @xmath81 and @xmath42 are irreducible components such that @xmath82 is an irreducible component of @xmath49 , then @xmath83 .
in other words , the dual semicanonical basis @xmath84 satisfies the multiplicative property which was expected to hold for the dual canonical basis @xmath18 .
an irreducible component @xmath85 is called _ indecomposable _ if @xmath57 contains a dense subset of indecomposable @xmath27-modules .
by @xcite , every irreducible component @xmath57 of @xmath26 has a _
canonical decomposition _
@xmath86 where the @xmath87 are indecomposable irreducible components .
our theorem implies that @xmath88 hence @xmath84 has a natural description as a collection of families of monomials in the elements indexed by indecomposable irreducible components .
such a description of @xmath84 resembles the description of @xmath18 for type @xmath89 obtained by berenstein and zelevinsky .
[ [ section-4 ] ] so a natural question is how close are the bases @xmath84 and @xmath18 ? in type @xmath61 , berenstein and zelevinsky
@xcite proved that all minors of the triangular matrix of coordinate functions on @xmath21 belong to @xmath18 .
we prove that they also belong to @xmath84 . hence using @xcite
, it follows that @xmath90 contains all multiplicative products of flag minors .
however the two bases differ in general .
more precisely we have : [ thmb ] the bases @xmath84 and @xmath18 coincide if and only if @xmath10 is of type @xmath4 with @xmath91 .
for example in type @xmath6 , we deduce from equation ( [ ce1 ] ) and theorem [ thma ] that @xmath92 ( where for simplicity we use the same notation @xmath93 and @xmath94 for the specializations at @xmath95 ) . nevertheless ,
since @xmath84 and @xmath18 have lots of elements in common , we get an explanation why the bz - conjecture ( or rather its reformulation in terms of irreducible components of varieties of @xmath27-modules ) holds for large families of elements of @xmath18 .
of course , by duality , these results also allow to compare the bases @xmath80 and @xmath17 .
in particular , returning to the example of @xcite , we can check that @xmath96 and this is probably the smallest example in type @xmath61 for which the canonical and semicanonical bases differ . one may conjecture that , in general , the elements @xmath97 occurring in the @xmath80-expansion of @xmath98 are indexed by the irreducible components @xmath57 of @xmath68 , so that @xmath68 is irreducible if and only if @xmath99 .
( there is a similar conjecture of lusztig @xcite for the `` semicanonical basis '' of the group algebra of a weyl group obtained from the irreducible components of the steinberg variety . ) assuming this conjecture we get an explanation of the relationship between the counterexamples to the conjectures of berenstein - zelevinsky and lusztig .
[ [ section-5 ] ] in the last part of the paper , we consider the first case which is not well understood , namely type @xmath6 . in this case , the preprojective algebra @xmath27 is representation - infinite , but it is still of tame representation type @xcite .
motivated by our description of @xmath84 in terms of indecomposable irreducible components of varieties of @xmath27-modules , we give a classification of the indecomposable irreducible components for the case @xmath6 .
we also give an explicit criterion to decide when the closure of the direct sum of two such components is again an irreducible component .
these results are deduced from @xcite , in which a general classification of irreducible components of varieties of modules over tubular algebras is developed .
they are naturally formulated in terms of the ringel bilinear form @xmath100 of a convex subalgebra @xmath101 of a galois covering of @xmath27 .
the algebra @xmath101 is a tubular algebra of type @xmath7 and the corresponding 10-dimensional infinite root system @xmath102 is an elliptic root system of type @xmath8 in the classification of saito @xcite , with a 2-dimensional lattice of imaginary roots .
note that the irreducible component @xmath57 of equation ( [ ce1 ] ) corresponds to a generator of this lattice .
( this is an a posteriori justification for calling @xmath103 an imaginary vector in @xcite . )
the ringel form @xmath100 allows to define a distinguished coxeter matrix @xmath104 of order @xmath105 acting on @xmath102 .
we prove the following : [ thmc ] there is a one - to - one correspondence @xmath106 between the set of schur roots of @xmath102 and the set of indecomposable irreducible components of the nilpotent varieties of type @xmath6 which do not contain an indecomposable projective @xmath27-module .
moreover @xmath107 is an irreducible component if and only if the schur roots @xmath108 and @xmath109 satisfy certain conditions which are all expressible in terms of @xmath100 and @xmath104 . we also explain how to translate from the language of roots to the language of multisegments , which form a natural indexing set of canonical and semicanonical bases in type @xmath61 . [ [ section-6 ] ] the paper is organized as follows . in section [ var ]
we recall the general theory of varieties of modules .
we explain a general decomposition theory for irreducible components of such varieties .
this is followed in section [ prepro ] by a short introduction to preprojective algebras .
then we recall the concept of a constructible function in section [ constf ] .
following lusztig @xcite , we review in section [ semi ] the definition of the semicanonical basis of @xmath19 , which is obtained by realizing @xmath19 as an algebra @xmath110 of constructible functions on the nilpotent varieties . in order to study the dual semicanonical basis and its multiplicative properties we also need to describe the natural comultiplication of @xmath19 in terms of @xmath110 .
this was not done in @xcite , so we provide this description in section [ comult ] . in section [ multipli ]
we introduce the dual semicanonical basis @xmath84 of @xmath111 and prove theorem [ thma ] .
note that for this theorem we do not restrict ourselves to types @xmath11 , and only assume that @xmath12 is the positive part of a symmetric kac - moody lie algebra .
we end this section with the proof of the `` only if '' part of theorem [ thmb ] . in section [ emb ]
we embed @xmath111 into the shuffle algebra .
this gives a practical way of computing elements of @xmath84 .
we use this to prove that in type @xmath61 all nonzero minors in the coordinate functions of @xmath21 belong to @xmath84 . in the rest of the paper we focus on the dynkin cases @xmath112 . in section [ galois ]
we consider a galois covering @xmath113 of the algebra @xmath27 , with galois group @xmath114 , and we use it to calculate the auslander - reiten quiver of @xmath27 for @xmath9 .
we also introduce an algebra @xmath101 whose repetitive algebra is isomorphic to @xmath113 . for @xmath9 , @xmath101 has finite representation type , while for @xmath115 it is a tubular algebra of tubular type @xmath7 . in section [ indmult ]
we recall from @xcite that the indecomposable irreducible components of @xmath27 are in one - to - one correspondence with the @xmath114-orbits of schur roots of @xmath113 .
we also describe the map which associates to such a schur root the multisegment indexing the corresponding indecomposable irreducible component .
the component graphs for the representation - finite cases @xmath37 , @xmath38 and @xmath39 are constructed in section [ comp234 ] , and the corresponding graphs of prime elements of @xmath116 are described in section [ prime234 ] . in section [ compare ]
we prove the `` if '' part of theorem [ thmb ] .
all the remaining sections are devoted to the case @xmath6 . in section [ tubweight ]
we relate the category of @xmath113-modules to the category @xmath117 of modules over the tubular algebra @xmath101 and to the category @xmath118 of coherent sheaves on a weighted projective line @xmath119 of type @xmath7 in the sense of geigle and lenzing @xcite . in section [ roots ]
we consider the grothendieck groups @xmath120 .
they are naturally endowed with a ( non - symmetric ) bilinear form @xmath121 ( the ringel form ) and a coxeter matrix .
this gives rise to an elliptic root system of type @xmath8 .
we give an explicit description of its set of positive roots and of the subset @xmath122 of schur roots . in section [ category ] , we show that @xmath122 naturally parametrizes the @xmath114-orbits of schur roots of @xmath113 , hence also the indecomposable irreducible components of @xmath27 .
then section [ extensions ] describes the component graph of @xmath27 for type @xmath6 , thus making precise the statements of theorem [ thmc ] .
section [ proof1 ] consists of the proof of theorem [ thm1 ] .
we conclude by noting the existence of similar results for type @xmath123 and by pointing out some possible connections with the theory of cluster algebras of fomin and zelevinsky ( section [ concl ] ) .
section [ pictures ] contains a collection of pictures and tables to which we refer at various places in the text .
[ [ section-7 ] ] throughout , we use the following conventions . if @xmath124 and @xmath125 are maps , then the composition is denoted by @xmath126 .
similarly , if @xmath127 and @xmath128 are arrows in a quiver , then the composition of @xmath129 and @xmath130 is denoted by @xmath131 .
modules are always assumed to be left modules .
all vector spaces are over the field @xmath132 of complex numbers .
we set @xmath133 , @xmath134 and @xmath135
. we also set @xmath136 and @xmath137 .
[ [ section-8 ] ] a _ quiver _ is a quadruple @xmath138 where @xmath23 and @xmath139 are sets with @xmath23 non - empty , and @xmath140 are maps such that @xmath141 and @xmath142 are finite for all @xmath143 .
we call @xmath23 the set of _ vertices _ and @xmath139 the set of _ arrows _ of @xmath62 . for an arrow @xmath144 one
calls @xmath145 the starting vertex and @xmath146 the end vertex of @xmath129 .
a _ path _ of length @xmath147 in @xmath62 is a sequence @xmath148 of arrows such that @xmath149 for @xmath150 .
set @xmath151 and @xmath152 .
additionally , for each vertex @xmath143 let @xmath153 be a path of length 0 . by @xmath154 we denote the _ path algebra _ of @xmath62 , with basis the set of all paths in @xmath62 and product given by concatenation .
a @xmath155 for @xmath62 is a linear combination @xmath156 where @xmath157 and the @xmath158 are paths of length at least two in @xmath62 with @xmath159 and @xmath160 for all @xmath161 .
thus , we can regard a relation as an element in @xmath154 .
an ideal @xmath162 in @xmath154 is _ admissible _ if it is generated by a set of relations for @xmath62 .
note that this differs from the usual definition of an admissible ideal , where one also assumes that the factor algebra @xmath163 is finite - dimensional .
[ [ section-9 ] ] a map @xmath164 such that @xmath165 is finite is called a _
dimension vector for _ @xmath62 .
we also write @xmath166 instead of @xmath167 , and we often use the notation @xmath168 . by @xmath169
we denote the semigroup of dimension vectors for @xmath62 .
let @xmath170 be the category of finite - dimensional @xmath23-graded vector spaces .
thus , the objects of @xmath170 are of the form @xmath171 where the @xmath172 are finite - dimensional vector spaces , and only finitely many of the @xmath172 are nonzero .
we call @xmath173 the _ dimension vector _ or _ degree _ of @xmath24 .
the morphisms in @xmath170 are just linear maps respecting the grading .
direct sums in @xmath170 are defined in the obvious way .
a _ representation _ of @xmath62 with underlying vector space @xmath174 is an element @xmath175 for a representation @xmath176 and a path @xmath148 in @xmath62 set @xmath177 then @xmath50 _ satisfies a relation _
@xmath178 if @xmath179 . if @xmath102 is a set of relations for @xmath62 , then let @xmath180 be the set of all representations @xmath181 which satisfy all relations in @xmath102 .
this is a closed subvariety of @xmath63 .
let @xmath182 be the algebra @xmath163 , where @xmath162 is the admissible ideal generated by @xmath102 .
note that every element in @xmath183 can be naturally interpreted as an @xmath182-module structure on @xmath24 , so we shall also write @xmath184 this is the affine variety of @xmath182-modules with underlying vector space @xmath24 .
a _ dimension vector for _
@xmath182 is by definition the same as a dimension vector for @xmath62 , that is , an element of @xmath169 . for @xmath185
we call @xmath186 the _ dimension vector _ of @xmath50 .
[ [ section-10 ] ] define @xmath187 this algebraic group acts on @xmath188 as follows . for @xmath189 and @xmath190
define @xmath191 the @xmath192-orbit of an @xmath182-module @xmath185 is denoted by @xmath193 .
two @xmath182-modules @xmath194 are isomorphic if and only if they lie in the same orbit . for a dimension vector @xmath195 for @xmath182 set @xmath196 thus @xmath197 for all @xmath174 .
for this reason , we often do not distinguish between @xmath198 and @xmath188 .
any problems arising from this can be solved via the existence of an isomorphism between these two varieties which respects the group actions and the gradings .
[ [ section-11 ] ] by abuse of notation , we identify @xmath143 with the dimension vector mapping @xmath199 to 1 and @xmath200 to @xmath201 .
if @xmath24 is an @xmath23-graded vector space with @xmath202 , then the variety @xmath188 consists just of a single point and is denoted by @xmath203 $ ] .
the corresponding 1-dimensional @xmath182-module is denoted by @xmath204 .
an element @xmath185 is said to be _ nilpotent _ if there exists an @xmath205 such that for any path @xmath206 of length greater than @xmath21 we have @xmath207 . by @xmath208
we denote the closed subset of nilpotent elements in @xmath188 .
the nilpotent elements are exactly the @xmath182-modules whose composition series contains only factors isomorphic to the simple modules @xmath204 , @xmath143 . [ [ section-12 ] ] an irreducible component @xmath209 is called _ indecomposable _ if it contains a dense subset of indecomposable @xmath182-modules .
let @xmath210 be the set of indecomposable irreducible components of @xmath188 .
let @xmath211 be the set of all irreducible components of varieties of @xmath182-modules , and set @xmath212 [ [ section-13 ] ] let @xmath213 , @xmath214 be irreducible components of @xmath182-modules , and set @xmath215 .
let @xmath216 be the set of modules in @xmath188 , which are isomorphic to @xmath217 with @xmath218 for all @xmath199 .
the closure @xmath219 is irreducible , but in general it is not an irreducible component . according to @xcite any irreducible component @xmath220 has a decomposition @xmath221 for certain indecomposable irreducible components @xmath222 .
moreover the components @xmath223 are uniquely determined up to reordering .
this is called the _ canonical decomposition _ of @xmath57 . for irreducible components @xmath2 and
@xmath3 of @xmath182-modules define @xmath224 this is the dimension of the extension group @xmath225 for all @xmath226 in a certain dense open subset of @xmath227 . for irreducible components
@xmath228 it is known that @xmath229 is an irreducible component if and only if @xmath230 for all @xmath231 , see @xcite .
the _ component graph _ @xmath232 of @xmath182 is defined as follows .
the vertices of @xmath232 are the elements in @xmath233 .
there is an edge between vertices @xmath2 and @xmath3 if and only if @xmath234 is again an irreducible component , or equivalently if @xmath235
[ [ section-14 ] ] assume that @xmath138 is a finite quiver without loops .
( a _ loop _ is an arrow @xmath129 with @xmath236 , and @xmath62 is _ finite _ if @xmath23 is finite .
note that this implies that @xmath139 is finite as well . ) the _ double quiver _
@xmath237 of @xmath62 is obtained from @xmath62 by adding for each arrow @xmath144 an additional arrow @xmath238 .
define @xmath239 , @xmath240 , @xmath241 and @xmath242 . for
any @xmath143 let @xmath243 be the _ gelfand - ponomarev relation _ associated to @xmath199 .
this is a relation for @xmath244 .
the _ preprojective algebra _
@xmath245 corresponding to @xmath62 is defined as @xmath246 where @xmath162 is generated by the relations @xmath247 , @xmath143 . these algebras were introduced and studied by gelfand and ponomarev , compare also @xcite . [ [ section-15 ] ] for a preprojective algebra @xmath248 and some @xmath174 set @xmath249 lusztig proved that @xmath26 has pure dimension @xmath250 , i.e. all irreducible components of @xmath26 have dimension @xmath250 . usually the varieties @xmath26 are called _
if @xmath62 is a dynkin quiver , then one might call them just _ preprojective varieties _ , since the nilpotency condition follows automatically in these cases , as shown by the next proposition . for a preprojective algebra
@xmath245 the following are equivalent : * @xmath245 is finite - dimensional ; * @xmath251 for all @xmath174 ; * @xmath62 is a dynkin quiver .
the equivalence of ( a ) and ( c ) is classical ( see for instance @xcite ) .
the implication ( c ) @xmath252 ( b ) is proved by lusztig @xcite , and the converse by crawley - boevey @xcite .
[ [ section-16 ] ] the following remarkable property of preprojective algebras was proved in @xcite .
[ propcb ] for finite - dimensional modules @xmath28 and @xmath253 over a preprojective algebra @xmath27 we have @xmath254 most preprojective algebras are of wild representation type .
the following proposition lists the exceptional cases .
we refer to @xcite and @xcite for further details .
[ infinite_type ] let @xmath245 be a preprojective algebra .
then the following hold : * @xmath245 is of finite representation type if and only if @xmath62 is of dynkin type @xmath4 with @xmath5 ; * @xmath245 is of tame representation type if and only if @xmath62 is of dynkin type @xmath6 or @xmath123 .
[ [ section-17 ] ] let @xmath28 be an algebraic variety over @xmath132 , endowed with its zariski topology .
a subset @xmath182 of @xmath28 is said to be constructible if it is a finite union of locally closed subsets .
it is easy to see that if @xmath28 is irreducible and if we have a partition @xmath255 into a finite number of constructible subsets , then there exists a unique @xmath256 containing a dense open subset of @xmath28 .
a function @xmath257 is _ constructible _ if it is a finite @xmath132-linear combination of characteristic functions @xmath258 for various constructible subsets @xmath182 .
equivalently , @xmath259 is finite and @xmath260 is a constructible subset of @xmath28 for all @xmath261 .
the set of constructible functions on @xmath28 is denoted by @xmath262 .
this is a @xmath132-vector space .
[ [ rhoz ] ] if @xmath57 is an irreducible component of @xmath28 and @xmath263 , then @xmath264 is a finite partition into constructible subsets , hence there is a unique @xmath261 such that @xmath265 contains a dense open subset of @xmath57 . in other words ,
a constructible function @xmath266 has to be constant on a dense open subset of each irreducible component @xmath57 of @xmath28 .
we denote by @xmath267 the linear form associating to @xmath266 its value on this dense open subset of @xmath57 .
[ [ section-18 ] ] by @xmath268 we denote the euler characteristic of a constructible subset @xmath182 . if @xmath269 , then @xmath270 .
hence @xmath271 can be regarded as a `` measure '' on the set of constructible subsets of @xmath28 .
for @xmath263 it is then natural to define @xmath272 this is a linear form on @xmath262 . more generally , for a constructible subset @xmath182 of @xmath28 we write @xmath273
[ [ section-19 ] ] in this section we assume that @xmath138 is a finite quiver without loops , and as before for @xmath174 let @xmath274 be the corresponding nilpotent variety .
we denote by @xmath275 the subspace of @xmath276 consisting of the constructible functions which are constant on the orbits of @xmath192 .
[ [ section-20 ] ] for @xmath277 such that @xmath278 , lusztig @xcite defines a bilinear map @xmath279 as follows .
let @xmath280 .
define @xmath281 to be the variety of all @xmath23-graded subspaces @xmath282 of @xmath24 such that @xmath283 and @xmath284 .
in other words , @xmath281 is the variety of all @xmath27-submodules of @xmath50 with dimension vector @xmath285 .
for such a @xmath282 let @xmath286 and @xmath287 be the elements induced by @xmath50 , and let @xmath288 and @xmath289 be the elements obtained by transporting @xmath290 and @xmath291 via some isomorphisms @xmath292 and @xmath293 . for @xmath294 and @xmath295
define @xmath296 by @xmath297 following @xcite define @xmath298 [ [ section-21 ] ] let @xmath299 ( resp .
@xmath300 ) be a @xmath301-orbit ( resp .
a @xmath302-orbit ) . for @xmath280 let @xmath303 be the constructible subset of @xmath281 consisting of all @xmath27-submodules @xmath304 of @xmath50 of isomorphism type @xmath305 and such that @xmath306 has isomorphism type @xmath307 .
the above definition yields @xmath308 note that in general the variety @xmath26 has infinitely many orbits .
( indeed , by proposition [ infinite_type ] the algebra @xmath27 has in general infinite representation type , hence , by the validity of the second brauer - thrall conjecture ( see @xcite ) , it has in general an infinite number of non - isomorphic representations of a given dimension ) .
therefore the support of the function @xmath309 may consist of an infinite number of orbits .
[ [ section-22 ] ] let @xmath310 where @xmath24 runs over the set of all isomorphism classes of vector spaces in @xmath170 .
( for example , we can take @xmath311 . ) the operation @xmath312 defines the structure of an @xmath169-graded associative @xmath132-algebra on @xmath313 . for @xmath143
, we recall that @xmath203 $ ] denotes the variety @xmath26 where @xmath202 .
( @xmath203 $ ] is just a single point . )
following @xcite define @xmath110 to be the subalgebra of @xmath314 generated by the functions @xmath315}$ ] , @xmath143 .
we set @xmath316 [ [ section-23 ] ] for two distinct vertices @xmath317 , let @xmath318 denote the negative of the number of arrows @xmath144 such that @xmath319 .
set also @xmath320 .
let @xmath10 be the symmetric kac - moody lie algebra over @xmath132 with cartan matrix @xmath321 .
let @xmath12 be a maximal nilpotent subalgebra of @xmath10 , and let @xmath19 be its enveloping algebra .
we denote by @xmath322 the chevalley generators of @xmath19 .
the defining relations of @xmath19 are @xmath323 where @xmath324 .
the algebra @xmath19 is @xmath169-graded by assigning to @xmath325 the degree @xmath199 .
it is known that the number of irreducible components of lusztig s nilpotent variety @xmath26 is equal to the dimension of the homogeneous component of @xmath19 of degree @xmath25 .
this was proved by lusztig @xcite when @xmath10 is of finite or affine type and by kashiwara and saito @xcite in general .
[ [ section-24 ] ] lusztig has proved that there is an algebra isomorphism @xmath326 given by @xmath327}$ ] .
to do this he constructed for every @xmath23-graded vector space @xmath24 a @xmath132-basis @xmath328 of @xmath329 , naturally labelled by the irreducible components of @xmath26 . using the notation of section [ rhoz ] , it is characterized by @xmath330 in other words ,
the function @xmath97 is the unique element of @xmath329 equal to @xmath331 on a dense open subset of @xmath57 and equal to @xmath201 on a dense open subset of any other irreducible component @xmath2 of @xmath26 , see ( * ? ? ?
* lemma 2.5 ) .
the basis of @xmath19 obtained by transporting via @xmath332 the collection @xmath333 where @xmath24 ranges over the set of all isomorphism classes of vector spaces in @xmath170 , is called the _ semicanonical basis _ of @xmath19 and
is denoted by @xmath80 .
[ ex1 ] let @xmath62 be the quiver with two vertices @xmath331 and @xmath334 and one arrow @xmath127 .
thus @xmath62 is a dynkin quiver of type @xmath37 .
\(a ) let @xmath335 with @xmath336 . then @xmath337 the variety @xmath26 has two irreducible components @xmath338 the group @xmath339 acts on @xmath26 with three orbits @xmath340 .
we have @xmath341 } * \1_{z[2 ] } , \qquad f_{z ' } = \1_{z ' } = \1_{z[2 ] } * \1_{z[1]}.\ ] ] ( b)let @xmath335 with @xmath342 . then @xmath343 and @xmath344 has dimension @xmath345 .
the variety @xmath26 has three irreducible components @xmath346 we have : @xmath347 } * \1_{z[2]}*\1_{z[2]}*\1_{z[1 ] } ) = { \frac{1}{2}}(\1_{z[2 ] } * \1_{z[1]}*\1_{z[1]}*\1_{z[2]}),\ ] ] @xmath348 } * \1_{z[1]}*\1_{z[2]}*\1_{z[2 ] } ) , \quad f_{z '' } = \1_{z '' } = { \frac{1}{4}}(\1_{z[2 ] } * \1_{z[2]}*\1_{z[1]}*\1_{z[1]}).\ ] ] note that @xmath349 and @xmath350 .
[ [ section-25 ] ] next , we consider composition series of modules over preprojective algebras .
let @xmath351 denote the set of pairs @xmath352 where @xmath353 is a sequence of elements of @xmath23 and @xmath354 .
the integer @xmath355 is called the length of @xmath352 .
given @xmath356 such that @xmath357 , we define a _ flag _ in @xmath24 of type @xmath352 as a sequence @xmath358 of graded subspaces of @xmath24 such that @xmath359 for @xmath360 .
thus @xmath361 is equal to @xmath201 or @xmath331 .
so these are complete flags , with possible repetition of some subspaces .
( it will be convenient below to allow such flags with repeated subspaces . )
we denote by @xmath362 the variety of flags of type @xmath352 .
when @xmath363 ( flags without repetition ) , we simply write @xmath364 .
let @xmath365 .
a flag @xmath366 is @xmath50-_stable _ if @xmath367 for all @xmath368 .
we denote by @xmath369 ( resp .
@xmath370 ) the variety of @xmath50-stable flags of type @xmath352 ( resp . of type @xmath371 ) . note that an @xmath50-stable flag is the same as a composition series of @xmath50 regarded as a @xmath27-module .
[ [ section-26 ] ] for @xmath356 with @xmath372 , define @xmath373}^{c_1 } * \cdots * \1_{z[i_m]}^{c_m } \in { { \mathcal m}}_\v.\ ] ] if @xmath374 for all @xmath368 , we simply write @xmath375 instead of @xmath376 . in general ,
@xmath377 where @xmath378 is the subword of @xmath371 consisting of the letters @xmath379 for which @xmath380 . by definition ,
the functions @xmath381 , where @xmath378 runs over all words , span @xmath110 .
the following lemma results immediately from the definition of the product @xmath312 of constructible functions .
[ lem2 ] let @xmath280 .
we have @xmath382 retain the notation of example [ ex1 ] ( b ) .
let @xmath383 be given by the following matrices ( with respect to some fixed bases of @xmath384 and @xmath385 ) @xmath386 let us calculate @xmath387 . to construct a flag @xmath388 we first have to choose a line @xmath389 in the 2-dimensional vector space @xmath390 .
we may take @xmath389 to be \(a ) the 1-dimensional image of @xmath50 , or \(b ) any line except this one . in case ( a )
the module @xmath391 induced by @xmath50 in the quotient @xmath392 is the semisimple module .
so we get @xmath393 in case ( b ) , @xmath394 where @xmath395 and at the next stage @xmath396 must be taken as the kernel of @xmath391 ( no choice ) , and @xmath397 is also completely determined .
thus , in case ( b ) we get @xmath398 so finally , @xmath399 [ [ section-27 ] ] in this section we assume that @xmath10 is a simple finite - dimensional lie algebra .
equivalently , @xmath62 is a dynkin quiver .
then @xmath19 has a pbw - basis @xmath400 associated to this quiver @xmath62 .
the image @xmath401 is easy to describe .
let @xmath402 .
the affine space @xmath63 can be regarded as a subset of @xmath26 by identifying it to the set of @xmath280 with @xmath403 for every @xmath404 .
clearly this is an irreducible component of @xmath26 .
moreover by our assumption it has finitely many @xmath192-orbits .
[ pbw ] let @xmath64 be a @xmath192-orbit in @xmath63 .
there exists a unique @xmath405 whose restriction to @xmath63 is the characteristic function of @xmath64 .
the collection of all @xmath406 where @xmath64 runs through all @xmath192-orbits in @xmath63 is equal to @xmath407 . by @xcite ,
the map from @xmath329 to @xmath408 sending @xmath266 to its restriction to @xmath63 is a vector space isomorphism .
moreover , the space @xmath409 ( where @xmath24 ranges over all isoclasses of vector spaces in @xmath170 ) endowed with the image of the product @xmath312 is the geometric realization of the hall algebra of @xmath62 over @xmath132 due to schofield ( see @xcite ) . in this
setting , the pbw - basis is the basis of @xmath410 given by the characteristic functions of the @xmath192-orbits in @xmath63 . thus the lemma is proved .
the algebra @xmath19 is in fact a bialgebra , the comultiplication being defined by @xmath411 in this section we describe the comultiplication @xmath412 of @xmath110 obtained by transporting this comultiplication via @xmath413 .
[ [ section-28 ] ] for @xmath414 such that @xmath278 , define a linear map @xmath415 as follows .
we have @xmath416 , so given @xmath417 and @xmath418 we can regard @xmath419 as an element of @xmath26 .
here , @xmath419 denotes the direct sum of @xmath290 and @xmath291 as endomorphisms of @xmath420 and @xmath421 , or equivalently as modules over the preprojective algebra @xmath27 . for @xmath422 , @xmath423 and @xmath424
set @xmath425 this is clearly a constructible function on @xmath426 which is constant on @xmath427-orbits .
[ [ section-29 ] ] let @xmath428 with @xmath429 .
let @xmath430 , @xmath417 , @xmath418 and @xmath431 .
[ homo ] we have @xmath432 where the sum is over all pairs @xmath433 of sequences in @xmath434 such that @xmath435 by definition we have @xmath436 to @xmath437 , we associate the pair of flags @xmath438 we regard @xmath439 as a flag in @xmath421 , and @xmath440 as a flag in @xmath420 by identifying @xmath420 with @xmath441 . clearly , we have @xmath442 and @xmath443 for some sequences @xmath444 , @xmath445 in @xmath434 satisfying the conditions ( [ condition ] ) .
let @xmath446 denote the set of pairs @xmath433 satisfying ( [ condition ] ) . for @xmath447 ,
let @xmath448 be the subset of @xmath370 consisting of those @xmath366 for which @xmath449 .
then clearly we have a finite partition @xmath450 where @xmath451 consists of the pairs @xmath433 such that @xmath448 is nonempty .
now for @xmath452 , the map @xmath453 sending @xmath366 to @xmath454 is a vector bundle , see ( * ? ? ? * lemma 4.4 ) .
hence , @xmath455 and @xmath456 on the other hand , @xmath457 and it only remains to prove that @xmath458 .
clearly @xmath451 , so let @xmath447 .
let @xmath442 and @xmath443 .
the assumption @xmath459 implies that @xmath448 is nonempty .
indeed , the flag @xmath366 in @xmath24 whose @xmath368th subspace is the direct sum of the @xmath368th subspaces of @xmath440 and @xmath439 is @xmath50-stable and by construction @xmath460 .
so @xmath452 and the lemma is proved .
[ [ section-30 ] ] by lemma [ homo ] , the map @xmath461 induces a linear map from @xmath329 to @xmath462 , given by @xmath463 where the pairs @xmath433 satisfy ( [ condition ] ) . taking direct sums
, we obtain a linear map @xmath464 the sum being over all isomorphism types @xmath420 and @xmath421 of @xmath23-graded vector spaces such that @xmath278 . taking direct sums over all isomorphism types @xmath24 , we get a linear map @xmath465 since @xmath466 , @xmath412 is coassociative . since @xmath467
, @xmath412 is cocommutative .
lemma [ homo ] shows that @xmath412 is multiplicative on the elements @xmath375 , that is , for @xmath428 @xmath468 where the product in @xmath469 is defined by @xmath470 under the algebra isomorphism @xmath471 , the map @xmath412 gets identified with the standard comultiplication of @xmath19 .
equation ( [ multdelta ] ) shows that @xmath412 is an algebra homomorphism @xmath472 .
moreover the generators @xmath473}=\psi(e_i)$ ] are clearly primitive .
the result follows .
[ [ section-31 ] ] the vector space @xmath110 is @xmath169-graded , with finite - dimensional homogeneous components .
let @xmath111 denote its graded dual .
given an @xmath23-graded space @xmath24 and an irreducible component @xmath57 of @xmath26 , we have defined a linear form @xmath474 on @xmath276 , see section [ rhoz ] .
we shall also denote by @xmath474 the element of @xmath111 obtained by restricting @xmath474 to @xmath329 and then extending by @xmath201 on all @xmath475 with @xmath476 . note that by ( [ caract ] ) , the basis of @xmath111 dual to the semicanonical basis @xmath477 is nothing but @xmath478 . from now on @xmath79 will be called the _ dual semicanonical basis _ of @xmath111 . [ delta ] for @xmath85 there exists an open dense @xmath192-stable subset @xmath479 such that for all @xmath480 and all @xmath481 we have @xmath482 . for a given @xmath266 ,
this follows from section [ rhoz ] .
moreover , there exists such an open set simultaneously for all @xmath266 because @xmath329 is finite - dimensional .
[ [ section-32 ] ] for @xmath365 define the delta - function @xmath483 by @xmath484 we then have @xmath485 the next lemma follows immediately .
[ lemmaopen ] let @xmath486 and suppose that the orbit of @xmath487 is open dense
. then @xmath488 .
let @xmath489 denote the multiplication of @xmath111 dual to the comultiplication @xmath412 of @xmath110 .
[ lemmaplus ] let @xmath490 and @xmath491 .
we have @xmath492 for @xmath493 , one has @xmath494 [ lem3 ] suppose that @xmath495 is an irreducible component of @xmath26
. then there exists @xmath496 of the form @xmath497 with @xmath498 and @xmath499 .
the direct sum @xmath500 is the image of the morphism @xmath501 defined by @xmath502 since @xmath503 ( resp .
@xmath504 ) is open dense in @xmath505 ( resp . in @xmath506 ) , the subset @xmath507 is open dense in @xmath508 .
now , since @xmath509 is a dominant morphism between irreducible varieties , the image under @xmath509 of @xmath507 contains a dense open subset of @xmath57 , hence it has a nonempty intersection with @xmath510 .
since both @xmath511 and @xmath510 are @xmath192-stable we can find @xmath50 in their intersection of the form @xmath497 with @xmath498 and @xmath499 .
[ [ section-33 ] ] we can now give the proof of theorem [ thma ] .
_ proof of theorem [ thma]._choose @xmath512 as in lemma [ lem3 ] .
then lemma [ delta ] and lemma [ lemmaplus ] yield @xmath513 [ decomp ] let @xmath514 be the canonical decomposition of the irreducible component @xmath57 of @xmath26 . the dual semicanonical basis vector @xmath474 factorizes as @xmath515 for @xmath516 this follows from theorem [ thma ] .
assume that @xmath517 .
by @xcite @xmath518 is an irreducible component .
moreover @xmath519 so by theorem [ thma ] we get @xmath520 .
the result follows by induction on @xmath355 .
the factorization given by corollary [ decomp ] will be called the _ canonical factorization _ of @xmath474 .
[ [ section-34 ] ] we shall now deduce from theorem [ thma ] the proof of the `` only if '' part of theorem [ thmb ] . let @xmath10 be of type @xmath521 or @xmath522 . then , the bases @xmath18 and @xmath84 do not coincide .
assume first that @xmath10 is of type @xmath6 or @xmath123 .
then the preprojective algebra @xmath27 is of tame representation type . in this case
, we have @xmath523 for any irreducible component @xmath57 of @xmath26 , see @xcite
. therefore by theorem [ thma ] and @xcite the square of any vector of @xmath84 belongs to @xmath84 . on the other hand ,
it was shown in @xcite that for the cases @xmath6 and @xmath123 there exist elements of @xmath18 whose square is not in @xmath18 .
they are called _ imaginary vectors _ of @xmath18 .
this shows that @xmath18 and @xmath84 are different in these cases .
now if @xmath10 is not of type @xmath4 with @xmath5 , then the dynkin diagram of @xmath10 contains a subdiagram of type @xmath6 or @xmath123 , and the result follows from the cases @xmath6 and @xmath123 . in the next sections we shall prepare some material for the proof of the `` if '' part of theorem [ thmb ] , to be given in section [ compare ]
we describe a natural embedding of @xmath111 into the shuffle algebra .
this is then used to describe a certain family of elements of @xmath84 in type @xmath4 .
[ [ section-35 ] ] let @xmath524 be the free associative algebra over @xmath132 generated by @xmath23 . a monomial in @xmath525
is called a word .
this is nothing else than a sequence @xmath526 in @xmath23 .
let @xmath527 be the surjective algebra homomorphism given by @xmath528}$ ] , and more generally by @xmath529 .
let @xmath530 denote the graded dual of @xmath525 .
we thus get an embedding of vector spaces @xmath531 let @xmath532\}$ ] denote the basis of @xmath530 dual to the basis @xmath533 of words in @xmath525 .
let @xmath534 .
we have @xmath535 = \sum_\ii \si(\pi(\ii))\ , w[\ii ] = \sum_\ii \si(d_\ii)\ , w[\ii].\ ] ] by lemma [ lem2 ] , we obtain in particular @xmath536.\ ] ] [ [ section-36 ] ] denote by @xmath537 the multiplication on @xmath538 obtained by pushing @xmath489 with @xmath539 , that is , for @xmath540 set @xmath541 [ lemshuf ] the product @xmath537 is the restriction to @xmath538 of the classical shuffle product on @xmath530 defined by @xmath542\shuffle w[i_{m+1},\ldots , i_{m+n } ] = \sum_s w[i_{s(1)},\ldots , i_{s(m+n)}],\ ] ] where the sum runs over the permutations @xmath543 such that @xmath544 this follows easily from lemma [ homo ] and the duality of @xmath489 and @xmath412 .
indeed , @xmath545 \\
& = & \sum_{\ii } ( \si\otimes\tau)(\delta(d_\ii))\,w[\ii]\\ & = & \sum_{\ii } \sum_{(\cc',\cc '' ) } \left(\si(d_{\ii,\cc'})\otimes\tau(d_{\ii,\cc''})\right ) \,w[\ii ] , \end{aligned}\ ] ] where the pairs @xmath433 are as in lemma [ homo ] .
now it is clear that the coefficient of @xmath546 $ ] in this last sum is the same as the coefficient of @xmath546 $ ] in the shuffle product @xmath547\right ) \shuffle \left(\sum_{\jj''}\tau(d_{\jj''})\,w[\jj'']\right)\ ] ] and the lemma is proved .
[ [ section-37 ] ] suppose that @xmath10 is of type @xmath4 .
then @xmath548 $ ] , where @xmath21 is the group of unitriangular @xmath549-matrices .
[ [ section-38 ] ] let us construct an explicit isomorphism @xmath550 \ra { { \mathcal m}}^*.\ ] ] let @xmath551 denote the coordinate function assigning to @xmath552 its entry @xmath553
. then @xmath554 = { \mathbb{c}}[t_{ij}\mid 1 \le i < j \le n+1].\ ] ] it is known that in the isomorphism @xmath20\simeq u(\n)^*$ ] , the natural basis of @xmath20 $ ] consisting of monomials in the @xmath551 gets identified to the dual of the pbw - basis of @xmath19 associated to the quiver @xmath555_{\alpha_1 } & \cdots \ar[l]_{\alpha_2 } & n \ar[l]_{\alpha_{n-1}}}\ ] ] ( see for example @xcite ) .
the @xmath192-orbits of @xmath556 are naturally labelled by the multisegments of degree @xmath25 , and if we denote by @xmath557 the dual in @xmath111 of the pbw - basis @xmath558 in @xmath110 , then more precisely the above isomorphism maps the monomial @xmath559 to the element @xmath560 indexed by the multisegment @xmath561+\cdots+[i_r , j_r-1].\ ] ] for @xmath562 , let @xmath563 $ ] denote an indecomposable representation of @xmath564 with socle @xmath204 and top @xmath565 ( up to isomorphism there is exactly one such representation ) .
then the orbit of @xmath563 $ ] is open dense so @xmath566}$ ] belongs to @xmath111 by lemma [ lemmaopen ] . on the other hand , by lemma [ pbw ] @xmath567}(\pbw_{\mathfrak{m } } )
= \begin{cases } 1 & \text{if $ { \mathfrak{m}}= [ i , j]$},\\ 0 & \text{otherwise . } \end{cases}\ ] ] hence @xmath568 } = \delta_{x[i , j]}$ ] and @xmath129 is the algebra homomorphism determined by @xmath569}$ ] .
[ [ section-39 ] ] if we regard the functions @xmath551 as entries of a unitriangular @xmath549 matrix @xmath570 we may consider some special elements of @xmath20 $ ] given by the minors of this matrix .
[ minor ] the images under @xmath129 of all nonzero minors of the matrix @xmath570 belong to @xmath84 .
we shall use the embedding @xmath539 .
first note that since @xmath563 $ ] has a unique composition series , @xmath571.\ ] ] let @xmath572 be the @xmath573-minor taken on the sequence of rows @xmath574 and the sequence of columns @xmath575 . since @xmath570 is a unitriangular matrix with algebraically independent entries @xmath551 above the diagonal , the function @xmath572 is nonzero if and only if @xmath576 for every @xmath577 .
we shall assume that this condition is satisfied .
let @xmath578 then @xmath579 is a skew young diagram .
we identify it with the following subset of @xmath580 @xmath581 each pair @xmath582 is called a cell of @xmath579 .
let @xmath304 be a standard young tableau of shape @xmath579 , that is , a total ordering @xmath583 of the cells of @xmath579 which is increasing both on the rows and on the columns .
we associate to @xmath304 the element @xmath584 = w[a_t - b_t,\ldots , a_1-b_1]\ ] ] of @xmath530 , where @xmath585 . before we continue with the proof of proposition [ minor ] we need the following lemma .
[ yshuf ] @xmath586,\ ] ] where @xmath304 runs over the set of standard young tableaux of shape @xmath579 .
set @xmath587.\ ] ] we shall prove that @xmath588 by induction on the number @xmath147 of cells of @xmath579 .
if @xmath589 , then @xmath590 = s_{{\lambda}\mu}$ ] for some @xmath199 , and the statement is clear .
so suppose @xmath591 . for @xmath592
define @xmath593 by @xmath594)= \begin{cases } w[i_1,\ldots , i_{s-1 } ] & \text{if $ i_1=i$},\\ 0 & \text{otherwise}. \end{cases}\ ] ] it is immediate to check that @xmath595 is a derivation with respect to the shuffle product , i.e. @xmath596 it is also clear that @xmath597 if and only if @xmath598 for every @xmath199 . note that @xmath599 is the minor on rows @xmath600 and columns @xmath601 of the matrix @xmath602 & w[2,1 ] & \ldots & w[n , n-1,\ldots,1 ] \\ 0 & 1 & w[2 ] & \ldots & w[n , n-1,\ldots,2 ] \\ \vdots & \vdots & \vdots & & \vdots \cr 0 & 0 & 0 & \ldots & 1 \end{pmatrix}\ ] ] where in the expansion of the determinant the shuffle product of the entries is used .
it follows that , if @xmath603 and @xmath604 then @xmath605 where @xmath606 is obtained from @xmath601 by replacing @xmath607 by @xmath608 , and otherwise @xmath609 . on the other hand , @xmath610,\ ] ] where @xmath611 ranges over the young tableaux whose shape is a skew young diagram @xmath612 obtained from @xmath579 by removing an outer cell @xmath613 with @xmath614
it is easy to check that there is one such diagram only if @xmath603 and @xmath604 , and that this diagram then corresponds to the pair @xmath615 above .
so , by induction @xmath616 in this case , and @xmath617 , otherwise
. therefore @xmath618 .
this finishes the proof of lemma [ yshuf ] .
we continue with the proof of proposition [ minor ] .
let @xmath561 + \cdots + [ i_k , j_k-1]\ ] ] be the multisegment corresponding to the pair @xmath619 .
( here we leave out @xmath620 $ ] in case @xmath621 . )
following @xcite this parametrizes a laminated @xmath27-module @xmath622 $ ] , that is , a direct sum of indecomposable subquotients of projective @xmath27-modules .
let @xmath24 be the underlying @xmath23-graded vector space of @xmath622 $ ] .
it is known that the @xmath192-orbit of @xmath622 $ ] is open dense in its irreducible component , hence the function @xmath623}$ ] belongs to the dual semicanonical basis .
now it is easy to see that the types @xmath371 of composition series of @xmath622 $ ] are in one - to - one correspondence with the standard young tableaux of shape @xmath579 , and that for each tableau , the corresponding flag variety @xmath624}$ ] is reduced to a point .
therefore , comparing with lemma [ yshuf ] we see that @xmath625})=\pi^*\alpha(\varphi_{\iu\ju}).\ ] ] hence @xmath626}$ ] belongs to the dual semicanonical basis .
this finishes the proof of proposition [ minor ] .
in order to prove the `` if '' part of theorem [ thmb ] we need to study the canonical decomposition of @xmath627 for type @xmath89 . our main tool for this will be the auslander - reiten quiver of @xmath27 , which we will calculate by using a galois covering @xmath113 of @xmath27 .
this covering is in fact important for all @xmath628 , and it will also play an essential rle in our investigation of type @xmath6 in the last sections of the paper .
so we shall work in type @xmath4 for general @xmath628 in the next two sections , and we shall specify which results are only valid for @xmath629 .
we will also exclude the trivial case @xmath630 and assume that @xmath631 .
[ [ section-40 ] ] for @xmath631 , let again @xmath564 be the quiver @xmath632_{\alpha_1 } & \cdots \ar[l]_{\alpha_2 } & n \ar[l]_{\alpha_{n-1}}}\ ] ] of dynkin type @xmath4 .
let @xmath633 be the preprojective algebra corresponding to @xmath564 .
thus @xmath634 where the double quiver @xmath635 of @xmath564 is @xmath636^{\alpha^*_1}&\ar@<1ex>[l]^{\alpha_1 } 2 \ar@<1ex>[r]^{\alpha^*_2}&\ar@<1ex>[l]^{\alpha_2 } \ar@<1ex>[r]^{\alpha^*_{n-1 } } \cdots & n \ar@<1ex>[l]^{\alpha_{n-1 } } , } \ ] ] and the ideal @xmath637 is generated by @xmath638 [ [ covlam ] ] next , let @xmath639 where @xmath640 is the quiver with vertices @xmath641 and arrows @xmath642 and the ideal @xmath643 is generated by @xmath644 for @xmath115 we illustrate these definitions in figure [ figu1 ] .
denote by @xmath645}}_n$ ] the full ( and convex ) subquiver of @xmath640 which has as vertices the set @xmath646 and denote by @xmath647}}_n$ ] the restriction of @xmath648 to @xmath645}}_n$ ] , see [ redatla ] for an example .
@xmath649^{\alpha^*_{13}}&{}\save[]+<0cm,2ex>*{\vdots}\restore & \ar[ld]^{\alpha_{22 } } 3_2 \ar[rd]^{\alpha^*_{32}}&{}\save[]+<0cm,2ex>*{\vdots}\restore & \ar[ld]^{\alpha_{41 } } 5_1\\ & & \ar[ld]^{\alpha_{12 } } 2_2 \ar[rd]^{\alpha^*_{22 } } & & \ar[ld]^{\alpha_{31 } } 4_1 \ar[rd]^{\alpha^*_{41}}\\ & { 1_2}\ar[rd]^{\alpha^*_{12 } } & & \ar[ld]^{\alpha_{21 } } 3_1 \ar[rd]^{\alpha^*_{31 } } & & \ar[ld]^{\alpha_{40 } } 5_0\\ & & \ar[ld]^{\alpha_{11 } } 2_1 \ar[rd]^{\alpha^*_{21 } } & & \ar[ld]^{\alpha_{30 } } 4_0 \ar[rd]^{\alpha^*_{40}}\\ & 1_1 \ar[rd]^{\alpha^*_{11}}&&\ar[ld]^{\alpha_{20 } } 3_0 \ar[rd]^{\alpha^*_{30}}&&\ar[ld]^{\alpha_{4,-1 } } 5_{-1}\\ & & \ar[ld]^{\alpha_{10 } } 2_0 \ar[rd]^{\alpha^*_{20 } } & & \ar[ld]^{\alpha_{3,-1 } } 4_{-1 } \ar[rd]^{\alpha^*_{4,-1}}\\ & 1_0 & { } \save[]+<0cm,-2ex>*{\vdots}\restore & 3_{-1 } & { } \save[]+<0cm,-2ex>*{\vdots}\restore & 5_{-2}\\ & & & \ar[dd]^f&&\\ & & & & & \\ & & & & & \\ { \lambda}_5 : & 1\ar@<1ex>[r]^{\alpha^*_1}&\ar@<1ex>[l]^{\alpha_1 } 2 \ar@<1ex>[r]^{\alpha^*_2}&\ar@<1ex>[l]^{\alpha_2 } 3\ar@<1ex>[r]^{\alpha^*_3 } & \ar@<1ex>[l]^{\alpha_3 } 4 \ar@<1ex>[r]^{\alpha^*_4 } & \ar@<1ex>[l]^{\alpha_4 } 5\\ } \ ] ] [ [ section-41 ] ]
the group @xmath114 acts on @xmath648 by @xmath132-linear automorphisms via @xmath650 this induces an action @xmath651 @xmath652 where @xmath653 denotes the @xmath648-module obtained from @xmath654 by twisting the action with @xmath611 . roughly speaking
, @xmath655 is the same @xmath648-module as @xmath654 , but `` translated one level upwards '' .
if we consider @xmath648 and @xmath656 as locally bounded categories we have a functor @xmath657 defined by @xmath658 this is a galois covering of @xmath656 in the sense of @xcite , with galois group @xmath114 .
it provides us with the push - down functor @xcite @xmath659 which we also denote by @xmath525 .
it is defined as follows .
let @xmath660 be a @xmath648-module with underlying graded vector space @xmath661 .
then @xmath662 has the same underlying vector space with the grading @xmath663 where @xmath664 , and @xmath662 has maps @xmath665 and @xmath666 .
the next lemma follows from @xcite for @xmath9 , and from @xcite for @xmath115 , as noted in @xcite .
[ push ] let @xmath667 .
any finite - dimensional indecomposable @xmath656-module is isomorphic to @xmath662 for some indecomposable @xmath648-module @xmath50 , which is unique up to a translation @xmath668 by the galois group @xmath114 .
[ [ section-42 ] ] for a dimension vector @xmath669 for @xmath648 define @xmath670 where @xmath671 we have a morphism of varieties @xmath672 where the _ push - down morphism _
@xmath673 is defined by @xmath674 .
for @xmath675 let @xmath676 be the @xmath611th _ shift _ of @xmath677 , that is , @xmath678 thus , if @xmath677 is the dimension vector of a @xmath648-module @xmath654 , then @xmath679 is the dimension vector of the shifted module @xmath680 . in this case
we write @xmath681 .
this defines an equivalence relation @xmath682 . to simplify our notation
, we shall not always distinguish between @xmath677 and its equivalence class .
however , it will always be clear from the context which one is meant . to display a dimension vector ( or its equivalence class ) for @xmath648
we just write down the relevant entries , all other entries are assumed to be 0 . [ [ arq ] ] recall that to any finite - dimensional algebra @xmath182 over a field @xmath368 ( and more generally to any locally bounded @xmath368-algebra or locally bounded category , see @xcite ) is associated a translation quiver @xmath683 called the _ auslander - reiten quiver _ of @xmath182 . it contains a lot of information about the category @xmath684 .
in particular if @xmath182 is representation - finite and standard , one can recover @xmath684 from @xmath683 .
the vertices of @xmath683 consist of the isomorphism classes of indecomposable finite - dimensional @xmath182-modules .
if @xmath654 and @xmath21 are two such modules , then there are @xmath685 arrows from @xmath654 to @xmath21 in @xmath683 , where @xmath686 denotes the radical of the category @xmath684 ( compare ( * ? ? ?
* chapter 2 ) ) .
this means that there is an arrow from @xmath654 to @xmath21 if there exists a nonzero irreducible homomorphism from @xmath654 to @xmath21 . here , by abuse of notation
, we do not distinguish between a module and its isomorphism class .
the quiver @xmath683 is endowed with an injective map @xmath687 , _ the translation _ , defined on the subset of vertices corresponding to non - projective modules . if @xmath654 is indecomposable and non - projective then @xmath688 where @xmath689 is the _ auslander - reiten sequence _ ( or almost split sequence ) ending in @xmath654 .
the _ stable auslander - reiten quiver _
@xmath690 of @xmath182 is obtained from @xmath683 by removing all translates @xmath691 and @xmath692 @xmath693 of the projective vertices @xmath694 and the injective vertices @xmath23 as well as the arrows involving these vertices . thus the translation @xmath687 induces a permutation on the vertices of @xmath690 .
we refer the reader to ( * ? ? ?
* chapter vii ) for more details on auslander - reiten theory , or to @xcite and ( * ? ? ?
* chapter 2 ) for @xmath368 algebraically closed as we assume here .
note however that these papers use slightly different conventions .
[ [ section-43 ] ] in section [ ar ] we display the auslander - reiten quivers of @xmath695 , @xmath696 and @xmath697 . to calculate them
one first determines the auslander - reiten quivers of their coverings @xmath698 , @xmath699 and @xmath700 . indeed the algebras @xmath648 are _ directed _ , that is , there is no sequence of indecomposable @xmath648-modules of the form @xmath701 with all homomorphisms being nonzero and non - invertible .
it follows that the auslander - reiten quiver of @xmath648 can be calculated by a combinatorial procedure known as the _ knitting procedure _ , see for example @xcite . by applying the push - down functor @xmath525 to this quiver ,
one then obtains the auslander - reiten quiver of @xmath656 @xcite . in our pictures ,
each indecomposable @xmath656-module @xmath654 is represented by the dimension vector @xmath677 of a @xmath648-module @xmath702 such that @xmath703 .
one has to identify each vertex in the extreme left column with the vertex in the extreme right column represented by the same dimension vector up to a shift by the galois group @xmath114 .
the auslander - reiten quiver of @xmath696 is shaped like a moebius band , and for @xmath695 and @xmath697 one gets a cylinder .
in particular , we see that @xmath695 has 4 isoclasses of indecomposable modules , @xmath696 has 12 , and @xmath697 has 40 .
we should point out that for @xmath704 there are @xmath656-modules which are not obtained from a @xmath648-module via the push - down functor .
[ [ htheo ] ] for an algebra @xmath182 let @xmath705 be the _ stable category _ of finite - dimensional @xmath182-modules ( @xcite )
. by definition , the objects of @xmath705 are the same as the objects of @xmath684 , and the morphism space @xmath706 is defined as @xmath707 modulo the morphisms factoring through projective modules .
the isomorphism classes of indecomposable objects in @xmath705 correspond naturally to the isomorphism classes of non - projective indecomposable @xmath182-modules . the stable category @xmath705 is no longer abelian , but if @xmath182 is a frobenius algebra then @xmath705 has the structure of a triangulated category ( @xcite ) with translation functor @xmath708 , the inverse of heller s loop functor .
moreover , in this situation we may identify the quiver of the stable category @xmath705 ( see for example @xcite ) with the stable auslander - reiten quiver @xmath690 defined in [ arq ] - just remove the projective - injective vertices from @xmath683 .
[ [ htheo2 ] ] to a finite - dimensional algebra @xmath709 one associates its _ repetitive algebra _
@xmath710 ( @xcite , @xcite ) .
this is an infinite - dimensional algebra without unit .
however it is locally bounded , and its indecomposable projective modules coincide with its indecomposable injective modules , that is , @xmath710 is a _
frobenius algebra_. hence for a finite - dimensional algebra @xmath709 , @xmath711 is a triangulated category , see [ htheo ] .
moreover if @xmath709 has finite global dimension , there is an equivalence of triangulated categories between @xmath711 and the derived category @xmath712 of bounded complexes of @xmath709-modules ( happel s theorem @xcite ) . under this equivalence
the functor @xmath708 corresponds to the translation functor @xmath713 $ ] .
[ [ gtgp ] ] if @xmath709 is a finite - dimensional algebra of finite global dimension we have by @xcite @xmath714 where @xmath715 is the subgroup of @xmath716 generated by the classes of projective @xmath710-modules .
thus we may identify @xmath717 with @xmath718 .
this coincides with the grothendieck group of @xmath711 viewed as a triangulated category . for a @xmath710-module @xmath654
we denote by @xmath719 $ ] its class in @xmath720 . in particular ,
if @xmath654 is projective , then @xmath719 = 0 $ ] .
notice that @xmath719 $ ] depends only on the dimension vector @xmath721 and provides an efficient method for calculating @xmath719 $ ] . in this context
we have @xmath722=\phi [ m]\ ] ] where @xmath104 is the coxeter transformation of @xmath709 and @xmath687 is the self - equivalence of @xmath711 induced by the auslander - reiten translation of @xmath723 .
this follows from happel s theorem [ htheo2 ] , the construction of the auslander - reiten translate in @xcite and the definition of the coxeter transformation .
[ [ redatla ] ] in case @xmath724 we have in @xmath725 a functorial isomorphism @xmath726 which is proved along the lines of @xcite . in particular this implies @xmath727 = -\phi^{-1}[m]$ ] . moreover , if we set @xmath728}}_n$ ] it is easy to see that @xmath729 .
for example @xmath730 is the algebra given by the following quiver with relations : @xmath731 2_2 \ar[rd ] \ar@{--}[dd ] & & \ar[ld ] 4_1 \ar[rd ]
\ar@{--}[dd]\\ 1_2 \ar[rd ] \ar@{--}@/^/[dd ] & & \ar[ld ] 3_1 \ar[rd ] \ar@{--}[dd ] & & \ar[ld ] 5_0 \ar@{--}@/_/[dd]\\ & \ar[ld ] 2_1 \ar[rd ] & & 4_0 \ar[rd ] \ar[dl]\\ 1_1 & & 3_0 & & 5_{-1}}}\end{aligned}\ ] ] here the dotted lines indicate zero relations and commutativity relations .
thus by happel s theorem [ htheo2 ] we have @xmath732 .
in particular we can apply [ gtgp ] . note that @xmath733 is hereditary of type @xmath734 ( resp .
@xmath735 ) for @xmath736 ( resp .
@xmath737 ) , and tilted of type @xmath738 in case @xmath739 .
happel s description of the auslander - reiten quiver of the derived category of a ( piecewise ) hereditary algebra of dynkin type ( * ? ? ?
iv.4.3 ) together with yields the shape of the stable auslander - reiten quiver in these cases . in case
@xmath115 the algebra @xmath730 is tubular and we can derive the structure of @xmath740 from the known structure of the derived category of a tubular algebra @xcite .
we will discuss this case in section [ tubweight ] .
[ [ is ] ] let @xmath741 be the set of dimension vectors of the indecomposable @xmath648-modules .
by @xmath742 we denote the set of _ schur roots _ of @xmath648 , _
i.e. _ the set of dimension vectors of the ( indecomposable ) @xmath648-modules @xmath654 with @xmath743 . for @xmath744 let @xmath745 for @xmath667 this is an irreducible component of @xmath746 .
we write @xmath747 if @xmath748 . for an irreducible component @xmath57 of @xmath749
define @xmath750 clearly , @xmath751 is an irreducible subset of @xmath752 , but in general @xmath751 is not an irreducible component .
the following result was shown in @xcite .
[ thmgs ] assume that @xmath667 . for @xmath753 ,
the variety @xmath754 is an indecomposable irreducible component of @xmath752 .
moreover , @xmath755 induces a bijection from @xmath756 to @xmath757 .
[ [ section-44 ] ] let @xmath758 be the set of multisegments supported on @xmath759 , that is , of the form @xmath760\ ] ] where @xmath761 .
the _ degree _ @xmath670 of @xmath762 is given by @xmath763 there is a one - to - one correspondence @xmath764 between the set @xmath765 of multisegments of degree @xmath195 in @xmath758 and the set of @xmath766-orbits in @xmath767 .
let @xmath768 be the multisegment labelling the unique dense orbit of @xmath767 .
it can be computed recursively by @xmath769 + {
\mathfrak{m}}_{\rm max}(d - e)\ ] ] where @xmath770 and @xmath771 if @xmath772 $ ] and @xmath773 , otherwise .
for example , @xmath774+[2,3]+[3,3]+[5,5].\ ] ] [ [ section-45 ] ] let @xmath775 be the projection morphism from @xmath752 to @xmath767 obtained by forgetting the maps labelled by the arrows @xmath776 .
lusztig @xcite proved that the irreducible components of @xmath752 are the closures of the sets @xmath777 , @xmath778 .
define @xmath779 a multisegment @xmath762 is said to be _ indecomposable _ if @xmath780 is indecomposable . by @xmath781
we denote the set of indecomposable multisegments in @xmath758 .
[ [ section-46 ] ] for a dimension vector @xmath669 for @xmath648 set @xmath782 where @xmath783 .
the variety @xmath784 is irreducible and , in fact , isomorphic to @xmath785 .
there is a @xmath648-module @xmath786 such that the orbit @xmath787 is dense in @xmath784 , and @xmath786 is uniquely determined up to isomorphism .
let @xmath788 be the set of modules @xmath789 which have submodules @xmath790 such that @xmath791 and @xmath792 for all @xmath793 .
thus @xmath794 if and only if @xmath795 for all @xmath796 , and @xmath797 if and only if @xmath798 . set @xmath799 for @xmath800 let @xmath801 be the unique multisegment in @xmath758 such that @xmath802 this defines a map @xmath803 if @xmath751 is an irreducible component , say @xmath780 , of @xmath752 , then @xmath804 . for all dimension vectors @xmath677 for @xmath648
the set @xmath805 is an irreducible component of @xmath749 , and we have @xmath806 one can easily see that @xmath807 then ( * ? ? ?
* theorem 1.3 ) yields that @xmath805 is an irreducible component .
the second part of the lemma follows directly from the definitions .
[ [ section-47 ] ] assume that @xmath667 .
define @xmath808 by @xmath809+[2,2]+[2,4]+[3,3]+[4,5 ] & \text{if $ n=5 $ and $ { { \bf d } } = \left ( \begin{smallmatrix } 1&&0&&0\\ & 1&&1&\\ 1&&2&&1\\ & 1&&1 & \end{smallmatrix } \right)$ } , \vspace{0.5cm}\\ [ 1,2]+[2,4]+[3,3]+[4,4]+2\,[5,5 ] & \text{if $ n=5 $ and $ { { \bf d } } = \left ( \begin{smallmatrix } & 1&&1&\\ 1&&2&&1\\ & 1&&1&\\ 0&&0&&1 \end{smallmatrix } \right)$},\\ { \mathfrak{m}}_{\rm max}({{\bf d } } ) & \text{otherwise}. \end{cases}\ ] ] [ thm1 ] let @xmath810 . the map @xmath811 establishes a bijection from @xmath756 to @xmath781 .
moreover the following diagram commutes : @xmath812^{\theta : { { \bf d } } \mapsto z_{{\bf d } } } \ar[dd]^{\psi } & & { \operatorname{ind}}({\operatorname{irr}}({{\widetilde{{\lambda}}}}_n))/\!\!\sim \ar[dd]^{\eta : z_{{\bf d } } \mapsto \overline{f_{{\bf d}}({\operatorname{gl}}(d ) \times z_{{\bf d}})}.}\\ & & \\ { \operatorname{ind}}(\mm(n ) ) \ar[rr]^{\phi : { \mathfrak{m}}\mapsto z_{\mathfrak{m } } } & & { \operatorname{ind}}({\operatorname{irr}}({\lambda}_n ) ) } \ ] ] the proof of theorem [ thm1 ] will be given in section [ proof1 ] .
[ [ section-48 ] ] in the case that @xmath656 is of finite representation type , the results of the previous sections simplify greatly . [ cases234 ] assume that @xmath813 .
then for each @xmath814 there exists ( up to isomorphism ) exactly one indecomposable @xmath648-module @xmath815 with dimension vector @xmath677 .
furthermore , @xmath816 for all @xmath677 , i.e. @xmath817 therefore , @xmath818 and @xmath819 the first two statements follow from the general theory of directed ( simply connected representation - finite ) algebras ( * ? ? ?
. then we apply theorem [ thmgs ] and theorem [ thm1 ] .
[ [ section-49 ] ] in section [ a4multi ] we list the 40 indecomposable multisegments in @xmath820 labelling the 40 indecomposable @xmath697-modules , and we redisplay the auslander - reiten quiver of @xmath697 with vertices these multisegments .
the translation @xmath687 can be read off by going horizontally two units to the left , for example @xmath821 and so on . note
that @xmath822 have no @xmath687-translate because they are the projective vertices .
[ [ results_extensions ] ] we shall use @xmath823 to describe the pairs @xmath824 of indecomposable @xmath656-modules such that @xmath825 . by proposition [ propcb ] ,
@xmath825 if and only if @xmath826 , so this is a symmetric relation .
recall , that for any finite - dimensional algebra @xmath27 the auslander - reiten formula ( * ? ? ?
* proposition 4.5 ) gives us @xmath827 .
now , if @xmath27 is selfinjective , representation - finite and admits a simply connected galois covering @xmath113 with galois group @xmath22 , we get @xmath828 here @xmath28 and @xmath253 are indecomposable @xmath27-modules , and @xmath829 and @xmath830 are indecomposable @xmath113-modules which under push - down give @xmath28 and @xmath253 , respectively . in this situation
it is easy to determine the dimensions of the summands in the last term using additive functions on the stable auslander - reiten quiver @xmath831 for some dynkin quiver @xmath832 , see @xcite .
as we have seen , this is exactly the situation for @xmath833 , and by there will be at most one @xmath834 with @xmath835 .
since @xmath687 induces a self - equivalence of @xmath836 we have to do this calculation only for one representative @xmath28 of each @xmath687-orbit .
the stable auslander - reiten quivers of @xmath695 , @xmath696 and @xmath697 have 1 , 2 and 6 non - trivial @xmath687-orbits , respectively . for @xmath736
there are only two indecomposable non - projective @xmath656-modules , say @xmath28 and @xmath253 , and @xmath837 . in sections [ exta3 ] and [ exta4 ] we display several copies of the stable auslander - reiten quivers of @xmath696 and @xmath697 . in each copy
we pointed a representative @xmath838 of a @xmath687-orbit , and we marked with the sign @xmath839 all indecomposable @xmath656-modules @xmath654 such that @xmath840 for example , there are 21 indecomposable @xmath697-modules @xmath654 such that @xmath841 [ [ section-50 ] ] we note that the previous description shows in particular that for @xmath91 every indecomposable @xmath656-module @xmath28 satisfies @xmath842 this was first observed by marsh and reineke .
it follows that the orbit closures of the indecomposable @xmath656-modules are the indecomposable irreducible components of the varieties of @xmath656-modules .
therefore the results of [ results_extensions ] give a complete description of the component graph @xmath843 for @xmath5 .
in this section we consider the dual canonical basis @xmath18 of @xmath20 $ ] for type @xmath89 .
using the bijection @xmath844 we may label its elements by multisegments @xmath762 or irreducible components @xmath57 .
we will write @xmath845 or @xmath846 depending on the context .
[ [ section-51 ] ] an element @xmath847 is called _ prime _ if it does not have a non - trivial factorization @xmath848 with @xmath849 .
let @xmath850 be the set of prime elements in @xmath18 .
let @xmath851 be the _ graph of prime elements_. the set of vertices of @xmath851 is @xmath852 , and there is an edge between @xmath35 and @xmath36 if and only if the product @xmath853 is in @xmath18 .
these graphs give a complete description of the basis @xmath18 @xcite .
indeed , @xmath18 is the collection of all monomials of the form @xmath854 where the @xmath855 satisfy for @xmath856 @xmath857 note that for all @xmath858 in @xmath18 , the square of @xmath858 also belongs to @xmath18 .
this is a particular case of the bz - conjecture which holds for @xmath89 .
therefore there is a loop at each vertex of @xmath851 .
similarly , the @xmath628 vertices labelled by an irreducible component containing an indecomposable projective module are connected to every other vertex ( because the corresponding elements of @xmath34 belong to the @xmath16-center ) .
let @xmath859 be the graph obtained from @xmath851 by deleting these @xmath628 vertices and all the loops .
clearly @xmath859 contains all the information .
[ [ b^*a4 ] ] the graphs @xmath860 and @xmath861 have been determined by berenstein and zelevinsky @xcite .
they are respectively dual to an associahedron of type @xmath630 and @xmath38 in the terminology of @xcite . in [ grapha4 ]
we display the graph @xmath862 , which has @xmath863 vertices ( corresponding to the multisegments @xmath864 in the list of [ a4multi ] ) and 330 edges .
we have calculated it by computer using the bz - conjecture . as suggested by zelevinsky
, @xmath862 is dual to an associahedron of type @xmath865 .
the maximal complete subgraphs of @xmath859 all have the same cardinality , namely @xmath331 , @xmath866 and @xmath105 for @xmath867 , @xmath866 and @xmath345 , and are called _
clusters_. there are respectively @xmath334 , @xmath868 and @xmath869 clusters .
[ [ section-52 ] ] the following theorem was proved by marsh and reineke for @xmath870 and conjectured for @xmath739 @xcite .
[ primethm ] for @xmath9 the graph @xmath851 is isomorphic to @xmath843 via the map @xmath871 .
this is checked by using the explicit descriptions of both graphs .
for example , from the first quiver of section [ exta4 ] we get that the vertex labelled by @xmath872 in @xmath873 is connected to all other vertices except @xmath874 the same happens in the graph @xmath875 , as can be seen from section [ grapha4 ] .
in this section we prove the `` if '' part of theorem [ thmb ] .
[ [ section-53 ] ] let @xmath10 be of type @xmath39
. for brevity set @xmath876 for @xmath877 .
[ egalite ] for every indecomposable multisegment @xmath762 of @xmath820 we have @xmath878 for 34 multisegments
@xmath879 out of the 40 elements of @xmath880 the vector @xmath881 is a minor , and the result follows from proposition [ minor ] .
the six elements which are not minors are @xmath881 where @xmath882 . denoting by @xmath883 the multisegment duality of zelevinsky @xcite
, we have @xmath884 denote also by @xmath883 the linear involution on @xmath530 given by @xmath885)=w[i_k,\ldots , i_1].\ ] ] it follows from @xcite that @xmath886 moreover it is known @xcite that for any multisegment @xmath762 we have @xmath887 hence it is enough to prove the lemma for @xmath888 .
this can be checked by an explicit computation in @xmath530 .
the calculation of @xmath889 is easy to perform using the algorithm of @xcite . on the other hand , for @xmath890 a point in the dense orbit of the irreducible component @xmath891
, we have @xmath892 and @xmath893 can be computed via ( [ eqdel ] ) in section [ emb ] .
thus , for @xmath894 we obtain the following expression for both @xmath895 and @xmath896 : @xmath897 + 2\,w[2 , 4 , 3 , 1 , 3 , 2 ] + 2\,w[2 , 4 , 3 , 3 , 1 , 2 ] + w[2 , 1 , 4 , 3 , 2 , 3 ] \\ & & + \ w[2 , 1 , 3 , 4 , 3 , 2 ] + 2\ , w[2 , 1 , 4 , 3 , 3 , 2 ] + w[2 , 3 , 4 , 1 , 3 , 2 ] + w[4 , 2 , 1 , 3 , 2 , 3 ] \\ & & + \ w[2 , 3 , 4 , 3 , 1 , 2 ] + w[4 , 3 , 2 , 1 , 3 , 2 ] + w[4 , 3 , 2 , 3 , 1 , 2 ] + 2\,w[4 , 2 , 1 , 3 , 3 , 2 ] \\ & & + \ w[2 , 4 , 3 , 1 , 2 , 3 ] + w[2 , 4 , 1 , 3 , 2 , 3 ] + 2\,w[2 , 4 , 1 , 3 , 3 , 2 ] + w[2 , 3 , 1 , 4 , 3 , 2 ] \\ & & + \ 2\,w[4 , 2 , 3 , 1 , 3 , 2 ] + w[4 , 2 , 3 , 1 , 2 , 3 ] \end{aligned}\ ] ] the calculations for @xmath898 and @xmath899 are similar and we omit them . [ [ section-54 ] ] we can now finish the proof of theorem [ thmb ] . assume that @xmath10 is of type @xmath39 .
let @xmath900 be the canonical decomposition of an irreducible component @xmath901 . by corollary [ decomp ] @xmath902 all components @xmath903 , @xmath904 are of the form @xmath905 for some indecomposable multisegment @xmath906 .
by proposition [ egalite ] we thus have @xmath907 . moreover ,
using theorem [ primethm ] we get @xmath908 hence @xmath474 belongs to @xmath18 .
thus for dynkin types @xmath4 with @xmath5 the dual canonical basis and the dual semicanonical basis coincide .
for the rest of this article we set @xmath909 , @xmath910 , @xmath911 , @xmath647}}={{\delta}^{[a , b]}}_5 $ ] , @xmath912}}_5 $ ] and @xmath913 . for our convenience we define moreover @xmath914}}={{\delta}^{[a , a]}}$ ] , @xmath915}}$ ] , @xmath916}}$ ] and @xmath917}}$ ] .
note that @xmath918 .
[ [ tub - fam ] ] almost all components of the auslander - reiten quiver of @xmath101 are _
tubes_. this plays a crucial rle in our results , so we shall recall the definition of a tube ( see @xcite , @xcite ) .
let @xmath919 be the quiver with vertices @xmath920 and arrows @xmath921 define a map @xmath687 on the set of vertices by @xmath922 .
for a vertex @xmath50 in @xmath919 and @xmath923 let @xmath924_r = \ { \tau^{ri}(x ) \mid i \in { \mathbb{z}}\}.\ ] ] then @xmath925 is the quiver with vertices @xmath926_r$ ] , and having an arrow @xmath926_r \to [ y]_r$ ] if and only if there is an arrow @xmath927 for some @xmath928_r$ ] and some @xmath929_r$ ] .
the vertex @xmath930_r$ ] in @xmath931 is said to have _ quasi - length _ @xmath608 .
the map @xmath687 induces a map on @xmath931 again denoted by @xmath687 , given by @xmath932_r ) = [ \tau(x)]_r\ ] ] and called the _ translation_. in this way , @xmath931 becomes a translation quiver in the sense of @xcite .
one calls @xmath931 a _ tube of rank _ @xmath577 .
a tube of rank 1 is called a _ homogeneous tube_. the _ mouth _ of a tube is the subset of vertices of quasi - length @xmath331 . sometimes we also consider as a tube a translation quiver @xmath933 whose stable part @xmath934 , obtained by deleting the translates of the projective and injective vertices , is a tube . [ [ tubfam ] ] for an algebra @xmath182 , we call a class of indecomposable @xmath182-modules a tube if the vertices of @xmath683 that belong to that class form a tube @xcite . a family of tubes @xmath935 is called a _ tubular @xmath28-family_. in our situation the index set @xmath28 will always be the projective line @xmath936 .
such a tubular family is said to be of _ type _
@xmath937 if for certain points @xmath938 the corresponding tubes have rank @xmath939 respectively , and for all remaining points the tubes are homogeneous . because of these exceptional points , a better index set is provided by the weighted projective line @xmath119 in the sense of geigle - lenzing @xcite , with exceptional points @xmath938 having respective weights @xmath939 . below we will point out some strong relations between the representation theory of @xmath27 and the weighted projective line @xmath119 of weight type @xmath940 ( see [ artlam ] , [ + roots ] ) . [ [ datub ] ] the algebra @xmath941 is a _ tame concealed algebra _ of type @xmath942 .
this means that @xmath941 is obtained from a hereditary algebra of type @xmath942 by tilting with respect to a preprojective tilting module @xcite .
the tame concealed algebras have been classified by happel and vossieck @xcite , and one can check that @xmath941 belongs to one of the frames in this list .
similarly @xmath943 is a tame concealed algebra of type @xmath944 .
its tubular type is @xmath945 ( see @xcite ) .
there are two indecomposable @xmath943-modules @xmath654 and @xmath946 completely determined ( up to isomorphism ) by their respective dimension vectors @xmath947 it can be shown that @xmath654 and @xmath946 lie at the mouth of a tube of the auslander - reiten quiver of @xmath943 , and that this tube has rank @xmath345 .
moreover , the tubular extension @xmath948[m']$ ] ( see @xcite ) is isomorphic to @xmath101 .
it follows that @xmath101 is a _ tubular algebra _ of tubular type @xmath7 ( see @xcite ) . similarly , @xmath101 can be regarded as a tubular coextension of @xmath941 .
note that the algebras @xmath949}}$ ] and @xmath950}}$ ] are isomorphic to @xmath943 and @xmath941 , respectively , so they are tame concealed . similarly the algebras @xmath951}}$ ] and @xmath952}}$ ]
are isomorphic to @xmath101 and @xmath953 , respectively , hence they are tubular algebras .
thus we may speak of preprojective , regular and preinjective @xmath954}}$]-modules or @xmath955}}$]-modules . [ [ destlam ] ] we are going to define some tubular families of @xmath113-modules . following @xcite
we introduce the following classes of modules .
let @xmath956}}$ ] be the class of indecomposable @xmath957}}$]-modules @xmath654 such that the restriction of @xmath654 to @xmath954}}$ ] is regular and nonzero . similarly , let @xmath958}}$ ] be the class of indecomposable @xmath955}}$]-modules @xmath654 such that the restriction of @xmath654 to @xmath954}}$ ] is preinjective and the restriction to @xmath959}}$ ] is preprojective .
clearly , we may also interpret the classes @xmath956}}$ ] and @xmath958}}$ ] as classes of @xmath113-modules , on which the galois group @xmath114 acts by @xmath960}}={{\mathcal{t}}^{[i+2j ] } } , \qquad ^{(j)}{\mathcal{m}^{[i , i+1]}}={\mathcal{m}^{[i+2j , i+2j+1 ] } } , \qquad ( j\in{\mathbb{z}}).\ ] ] the classes @xmath961}}$ ] and @xmath962}}$ ] are tubular families of type @xmath7 . in section [ tnull ] and [ tinfty ] we display the dimension vectors of the non - homogeneous tubes in these families .
the remaining tubes consist only of the homogeneous @xmath954}}$]-modules .
the classes @xmath958}}$ ] are the regular modules of the tubular algebra @xmath955}}$ ] which do not belong to the first or last tubular family
. we may decompose @xmath958}}$ ] into a disjoint collection of tubular families @xmath963}}_{\lambda})_{{\lambda}\in{\mathbb{q}}^+}$ ] of type @xmath7 ( see @xcite and also [ expldec ] for more details ) .
[ [ section-55 ] ] define @xmath964}}\cup{{\mathcal{t}}^{[0]}}\cup{\mathcal{m}^{[0,1]}}\cup{{\mathcal{t}}^{[1]}}\ ] ] and @xmath965 the next proposition shows that @xmath966 is a transversal of the action of the galois group on the set of indecomposable @xmath113-modules . [ prp : transv ] for each indecomposable @xmath113-module @xmath654 there is a unique @xmath967 such that @xmath968 .
moreover for each @xmath969 there is a unique @xmath834 with @xmath970 . by repeating the argument in @xcite
we find that each indecomposable module is contained in @xmath971}}\cup{{\mathcal{t}}^{[i]}})$ ] . by it only remains to show the unicity of @xmath608 .
this follows from the fact that the classes @xmath972}},{{\mathcal{t}}^{[0]}},{\mathcal{m}^{[0,1]}},{{\mathcal{t}}^{[1]}}$ ] are pairwise disjoint .
indeed @xmath973}}$ ] and @xmath974}}$ ] are disjoint as we can see from the dimension vectors of their objects , see [ tnull ] and [ tinfty ] . on the other hand @xmath972}}$ ] and @xmath975}}$ ] are disjoint because a @xmath954}}$]-module can not be preprojective and preinjective at the same time .
the second claim follows from the first .
indeed , we can determine the class @xmath956}}$ ] or @xmath958}}$ ] which an indecomposable module belongs to by means of @xmath976 only ( see @xcite and [ expldec ] below ) .
[ [ artlam ] ] proposition [ prp : transv ] shows in particular the ( known ) fact that @xmath113 is locally support finite .
moreover the additive closure @xmath977 ( _ i.e. _ the full subcategory of @xmath978 consisting of modules which are direct sums of modules of @xmath966 ) is closed under auslander - reiten sequences . thus by @xcite and @xcite we may identify the auslander - reiten quivers of @xmath977 and of @xmath979 .
these considerations together with imply that the image @xmath980 of @xmath966 in @xmath981 is a transversal for the indecomposable objects in @xmath981 under the action of @xmath114 by powers of the translation functor @xmath708 . on the other hand , since @xmath101 is a tubular algebra of tubular type @xmath7 , we know from @xcite that we have equivalences of triangulated categories @xmath982 where @xmath119 is a weighted projective line of weight type @xmath7 . since @xmath118 is a hereditary category we have the useful decomposition @xmath983.\ ] ] this is similar to the decomposition @xmath984\ ] ] that we have just explained .
although a lot of our intuition on @xmath981 comes from the comparison with @xmath118 , we prefer to avoid using this machinery .
[ [ section-56 ] ] write @xmath985 for the grothendieck group of @xmath117
. we have @xmath986 by identifying the class of a @xmath101-module @xmath654 with its dimension vector @xmath987 .
let @xmath988 be the _
ringel bilinear form _ of @xmath101 , given by @xmath989 see @xcite .
the algebra @xmath101 being tubular has global dimension 2 , so @xmath990 for @xmath991 , and this infinite sum is in fact finite .
explicitly for @xmath992 we have @xmath993 where the first sum runs over all indices @xmath994 , the second one over all pairs @xmath995 such that there is an arrow from @xmath994 to @xmath996 , and in the third sum @xmath997 is the number of relations from @xmath994 to @xmath996 , i.e. @xmath998 [ [ section-57 ] ] let @xmath999 be the matrix giving the ringel form : @xmath1000 where @xmath1001 and @xmath1002 are interpreted as column vectors in @xmath1003 and @xmath1004 means transposition .
_ coxeter matrix _ of @xmath101 by @xmath1005 ( see @xcite ) .
it is easy to check that @xmath1006 the data @xmath1007 is called a _ bilinear lattice _ @xcite .
[ [ sus : isomt ] ] let @xmath1008 be an element of @xmath1009 .
the _ support _ @xmath1010 of @xmath654 is defined as the set of vertices @xmath994 of @xmath1011 such that the @xmath994th component of @xmath976 is nonzero .
for @xmath1012 it will be convenient to identify the grothendieck group @xmath1013}})$ ] with the subgroup of @xmath1014 of elements with support in @xmath1015}}$ ] , see [ covlam ] .
we shall denote by @xmath1016}}$ ] the ringel form on @xmath1013}})$ ] , so that @xmath1017}}$ ] .
let @xmath1018 ( resp .
@xmath1019 ) be the dimension vectors of the indecomposable projective @xmath113-modules with support in @xmath1020}}$ ] ( resp . in @xmath1021}}$ ] ) , that is , @xmath1022 @xmath1023 up to shift these are the dimension vectors of all indecomposable projective @xmath113-modules .
note that these are precisely the dimension vectors of the projective @xmath113-modules that belong to @xmath966 .
the following result is verified by a direct calculation .
[ lem : isom ] the map @xmath1024 $ ] ( see [ gtgp ] ) induces isometries @xmath1025}})\lra k_0({\delta } ) , \quad & { \mathbf{x}}\mapsto { \mathbf{x}}-x^{{\phantom{g}}}_{4_{-1}}{{\mathbf{p}}^{{\phantom{g}}}_{2}}-x^{{\phantom{g}}}_{2_0}{{\mathbf{p}}^{{\phantom{g}}}_{4 } } \\
\iota_-\colon k_0({{\delta}^{[1,2]}})\lra k_0({\delta}),\quad & { \mathbf{x}}\mapsto { \mathbf{x}}-x^{{\phantom{g}}}_{1_3}{{\mathbf{p}}^{{\phantom{g}}}_{1}}-x^{{\phantom{g}}}_{3_2}{{\mathbf{p}}^{{\phantom{g}}}_{3}}-x^{{\phantom{g}}}_{5_1}{{\mathbf{p}}^{{\phantom{g}}}_{5}}\end{aligned}\ ] ] of bilinear lattices , that is , @xmath1026}}={\langle \iota_+({\mathbf{x}}),\iota_+({\mathbf{y}})\rangle},\qquad { { \langle { \mathbf{x}},{\mathbf{y}}\rangle}^{[1,2]}}={\langle \iota_-({\mathbf{x}}),\iota_-({\mathbf{y}})\rangle}.\ ] ] [ [ section-58 ] ] let @xmath1027 } } : k_0({{\delta}^{[i , j ] } } ) \lra { \mathbb{z}}$ ] be the quadratic form @xmath1027}}({\mathbf{d } } ) = { { \langle { \mathbf{d}},{\mathbf{d}}\rangle}^{[i , j]}}$ ] . for simplicity
we write @xmath1028}}={q^{[i , i]}}$ ] and @xmath1029}}$ ] .
define @xmath1030}}$ ] to be the positive generator of the one - dimensional radical of @xmath1028}}$ ] ( recall that @xmath954}}$ ] is tame concealed ) .
this means for example that the support of @xmath1031}}$ ] and of @xmath1032}}$ ] is contained in @xmath1033}}$ ] .
explicitly @xmath1034 } } = \left ( \begin{smallmatrix } & 0&&0\\1&&2&&1\\&3&&3\\1&&2&&1 \end{smallmatrix } \right ) , \qquad { { \mathbf{h}}^{[1 ] } } = \left ( \begin{smallmatrix } & 1&&1\\1&&2&&1\\&1&&1\\0&&0&&0 \end{smallmatrix } \right).\ ] ] moreover @xmath1035}}={^{(i)}{{\mathbf{h}}^{[0]}}}$ ] and @xmath1036}}={^{(i)}{{\mathbf{h}}^{[1]}}}$ ] .
since @xmath1037}}$ ] is positive semidefinite of corank 2 it is easy to see that @xmath1038 } } ) = { \mathbb{z}}\,{{\mathbf{h}}^{[i]}}\oplus{\mathbb{z}}\,{{\mathbf{h}}^{[i+1]}}$ ] .
notice that ( lemma [ lem : isom ] ) @xmath1039}})=-{{\mathbf{h}}^{[1 ] } } , \qquad \iota_+({{{\mathbf{h}}^{[0 ] } } } ) = { { \mathbf{h}}^{[0]}}.\ ] ] it will be convenient to set @xmath1040}}$ ] and @xmath1041}}$ ] .
these are the two generators of @xmath1042 .
[ [ section-59 ] ] let @xmath1043 be the set of _ roots _ of @xmath16
. a root @xmath1044 is called _ imaginary _ if @xmath1045 , and _ real _ otherwise .
let @xmath1046 ( resp .
@xmath1047 ) be the set of imaginary ( resp .
real ) roots of @xmath16 .
the form @xmath16 being positive semidefinite , @xmath1046 consists of the nonzero elements of @xmath1042 .
note that , by ( [ cox ] ) , @xmath1001 is a radical vector if and only if @xmath1048 .
indeed , @xmath1049 and the ringel form @xmath121 is nondegenerate . since @xmath101 has tubular type @xmath7 , the form @xmath16 induces a positive definite form of type @xmath522 on @xmath1050 .
( note that the dynkin diagram of type @xmath1051 is a star with three branches of lengths @xmath1052 . )
thus , @xmath102 is an elliptic root system of type @xmath8 in the sense of saito @xcite .
[ [ + roots ] ] by @xcite , the elements of @xmath1053 are precisely the dimension vectors of the indecomposable @xmath101-modules .
note however that there exist elements in @xmath102 whose coordinates are not all nonnegative or all nonpositive
. for example @xmath1054 following @xcite and @xcite we will introduce another notion of positive root in @xmath102 . define the subset of _ positive roots _ of @xmath16 as @xmath1055 and set @xmath1056 . in our setting @xmath1057
is the set of classes in @xmath1058 of the non - projective ( indecomposable ) @xmath113-modules which belong to @xmath966 as we will see in corollary [ cor : eta ] .
then it follows from [ artlam ] that @xmath1059 , see also @xcite .
one could also derive this fact from a careful analysis of @xmath102 .
our definition of @xmath1057 is partially inspired by the interpretation of the bilinear lattice @xmath1060 as the grothendieck group of the category @xmath1061 , see [ artlam ] .
indeed , since @xmath1061 and @xmath117 are derived equivalent @xcite , they have isomorphic grothendieck groups @xcite . the linear forms @xmath1062 and @xmath1063 correspond to the functions `` rank '' and `` degree '' on @xmath1064 , and @xmath1057 corresponds to the set of classes of indecomposable sheaves .
[ [ section-60 ] ] since for @xmath1065 we have @xmath1066 or @xmath1067 , the quotient @xmath1068 is a well - defined element of @xmath1069 .
for @xmath1070 let @xmath1071 be the set of _ positive roots of slope
_ @xmath1072 .
thus @xmath1073 for @xmath1074 and @xmath1075 such that @xmath1076 or ( @xmath1077 and @xmath1078 ) set @xmath1079 we have @xmath1080 this follows immediately from the equalities @xmath1081 set @xmath1082 thus @xmath1083 [ [ expldec ] ] the above decomposition of @xmath1057 is motivated by the following fact . for @xmath1084 let @xmath1085}}_{\lambda}$ ] be the class of indecomposable @xmath955}}$]-modules @xmath654 with @xmath1086}},{\mathbf{dim}(m)}\rangle}^{[i , i+1]}}}{{{\langle { \mathbf{dim}(m)},{{\mathbf{h}}^{[i+1]}}\rangle}^{[i , i+1]}}}={\lambda}.\ ] ] then @xmath1085}}_{\lambda}$ ] is a tubular family of type @xmath7 and @xmath1087}}={\mathcal{m}^{[i , i+1]}}$ ] .
moreover the set of dimension vectors @xmath1088}}_{\lambda}\}$ ] coincides with the set of roots @xmath1089 of @xmath1090}})$ ] such that @xmath1086}},{\mathbf{v}}\rangle}^{[i , i+1]}}}{{{\langle { \mathbf{v}},{{\mathbf{h}}^{[i+1]}}\rangle}^{[i , i+1]}}}={\lambda},\ ] ] see @xcite .
for @xmath1091 we may now define the tubular family @xmath1092 } } & \text { for } { \lambda}=\infty,\\ { { \mathcal{t}}^{[0,1]}}_{\lambda}&\text { for } 0<{\lambda}<\infty,\\ { { \mathcal{t}}^{[0 ] } } & \text { for } { \lambda}=0,\\ { { \mathcal{t}}^{[-1,0]}}_{-1/{\lambda } } & \text { for } { \lambda } < 0 . \end{cases}\ ] ] clearly we have @xmath1093 .
we shall denote by @xmath1094 the tube of the family @xmath1095 indexed by @xmath1096 .
thus , for all points @xmath50 except the three exceptional points @xmath1097 , @xmath1094 is a homogeneous tube , and @xmath1098 have respective ranks @xmath1099 and @xmath334 .
[ lem : tubes ] for @xmath1091 we have @xmath1100\mid m\in{\mathcal{t}}_{\lambda}\text{\ and $ m$ is non - projective}\}$ ] . for @xmath1101 this may be verified directly by comparing the dimension vectors of the objects in @xmath1102 ( resp .
@xmath1103 ) which we describe in [ tnull ] ( resp . [ tinfty ] ) with the classes in @xmath1104 ( resp .
@xmath1105 ) which we can construct explicitly , see theorem [ thm3 ] . for @xmath1106
the result is clear from the case @xmath1107 of ringel s result which we explained above .
for @xmath1108 it follows from the same result for @xmath1109 after applying the isometry @xmath1110 from lemma [ lem : isom ] and .
[ cor : eta ] the restriction @xmath1111 of the map @xmath1112 $ ] , is well - defined and surjective .
moreover , we have @xmath1113 if @xmath1114 the set @xmath1115 contains precisely one element .
we leave the proof to the reader .
[ [ section-61 ] ] since @xmath101 is tubular of tubular type @xmath7 , we have that @xmath1116 ( see @xcite ) . of course
this can also be verified directly by a simple calculation .
define the _
@xmath1117 of a vector @xmath1118 as the minimal @xmath1119 such that @xmath1120 .
hence , if @xmath1001 is an imaginary root @xmath1121 , and if @xmath1001 is a real root @xmath1122 or @xmath105 . the _ quasi - length _
@xmath1123 of @xmath1001 is defined as the greatest common divisor of the entries in the imaginary root @xmath1124 in other words , writing @xmath1125 we have @xmath1126 .
set also @xmath1127 then @xmath1128 if and only if @xmath1129 where @xmath1130 and @xmath1131 .
these definitions are motivated by the fact that for a non - projective ( indecomposable ) @xmath113-module @xmath654 which has quasi - length @xmath1132 in a tube of rank @xmath577 we have @xmath1133)=\begin{cases}\ell & \text { if $ r \nmid \ell$},\\ \ell / r & \text { if $ r\mid \ell$ } , \end{cases}\qquad { \operatorname{rk}}([m])=\begin{cases } r & \text { if $ r\nmid \ell$},\\ 1 & \text { if $ r\mid \ell$}. \end{cases}\ ] ] ( of course our @xmath1134 should not be confused with the rank function on @xmath1064 ) .
[ [ r - lambda - l - i ] ] for @xmath1070 , @xmath1135 and @xmath1136 let @xmath1137 so that @xmath1138 [ lemrli ] * if @xmath1139 then @xmath1140 ; * if @xmath1141 and @xmath199 divides @xmath1132 , then @xmath1142 ; * if @xmath1141 and @xmath199 does not divide @xmath1132 , then @xmath1143 consists of @xmath199 real roots ; * if @xmath1144 , then @xmath1145 where @xmath1146 and @xmath1147 .
let @xmath1139 .
clearly , @xmath1148 , @xmath1149 , and @xmath1150 .
moreover , by equation ( [ cox ] ) we have @xmath1151 .
hence the @xmath104-orbit of @xmath1001 is contained in @xmath1143 .
conversely , if @xmath1143 is nonempty , then it consists of a single @xmath104-orbit .
this follows from and lemma [ lem : tubes ] ( remember that @xmath1095 is a tubular family of type @xmath7 ) .
note that all the indecomposable modules lying on some homogeneous tube of @xmath1095 and having the same quasi - length @xmath1132 have the same class @xmath1152 in @xmath985 , where @xmath1153 and @xmath1154 .
moreover for @xmath1155 the element @xmath1152 is also the class of @xmath199 modules with quasi - length @xmath1132 lying on the tube of rank @xmath199 of @xmath1095 .
it follows that there is no positive root @xmath1001 with @xmath1156 and @xmath1123 divisible by @xmath199 .
[ [ section-62 ] ] let @xmath1157 . for @xmath1158 and @xmath1159 set @xmath1160}(i ) = r^{\lambda}_\ell(i)\ ] ] where @xmath1161 and @xmath1162 . by lemma [ lemrli ] , @xmath1163}(i)$ ]
has cardinality @xmath199 . in section [ rootlist ]
we list all elements of the sets @xmath1163}(i)$ ] , that is , a distinguished set of @xmath1164 real elements , say @xmath1165 , of @xmath1057 .
we constructed this set in the following way .
note first , that it coincides with the set @xmath1166 now consider the algebra @xmath1167 which is obtained by restricting @xmath101 to the full subquiver of @xmath1033}}$ ] which is obtained by removing the vertices @xmath1168 and @xmath1169 . notice that if @xmath1170 is a real root , then @xmath1171 is a real root with support in @xmath1167 .
this is tilted of type @xmath522 , thus the associated quadratic form @xmath1172 has 240 roots @xmath1173 . find them for example as the orbits of the dimension vectors of the @xmath1174 indecomposable projective @xmath1167-modules under the corresponding coxeter transformation @xmath1175 .
since @xmath1167 is tilted of type @xmath522 , the transformation @xmath1175 has order @xmath1176 ( the coxeter number of @xmath522 ) .
next , define integers @xmath1177 by @xmath1178 for @xmath1179 .
finally set @xmath1180 we can now give the following explicit construction of all real roots in @xmath1057 .
[ thm3 ] let @xmath1070 , @xmath1181 and @xmath1136 not divisible by @xmath199 .
write @xmath1182 with @xmath1074 , @xmath1075 and @xmath1147 ( if @xmath1183 set @xmath1077 and @xmath1184 ) .
write @xmath1185 with @xmath1186 such that @xmath1187
. then @xmath1188}(i ) \}.\ ] ] let @xmath1170 be a real root of rank @xmath199 , @xmath1189 an imaginary root and set @xmath1190 . then @xmath1001 is a real root and @xmath1191 , hence @xmath1001 has also rank @xmath199 .
moreover , if @xmath1192 and @xmath1193}(i)$ ] then @xmath1194 , therefore @xmath1001 belongs to @xmath1143 .
thus , since @xmath1143 and @xmath1195}(i)$ ] both have cardinality @xmath199 we see that the first subset is the translate of the second one by @xmath1196 .
thus , the 240 positive real roots listed in section [ rootlist ] yield a complete description of the infinite set of all positive real roots of @xmath102 .
note that the classes of these 240 roots in @xmath1050 form a finite root system of type @xmath522 , and we recover that @xmath102 is an elliptic root system of type @xmath8 . [ [ section-63 ] ] the set @xmath1197 of _ schur roots _ is defined as @xmath1198 it is easy to see that the set of imaginary schur roots consists of the @xmath1199 with @xmath1131 .
the schur roots can also be characterized in terms of rank and quasi - length , namely @xmath1200 this comes from the identity @xmath1201 thus the set of real schur roots is equal to the union of all subsets @xmath1143 for @xmath1070 , @xmath1155 and @xmath1202 . using [ r - lambda - l - i ] , this implies that @xmath1203 contains exactly @xmath1204 roots , one imaginary and the others real .
note that all the 240 roots listed in section [ rootlist ] are schur roots .
the schur roots of @xmath1057 are related to the schur roots of @xmath113 ( see [ is ] ) in the following way .
suppose that @xmath1205 is non - projective and has quasi - length @xmath1132 in a tube of rank @xmath577 . since all tubes in @xmath1206 are standard , the endomorphism ring of @xmath654 is non - trivial if and only if either @xmath1207 , or @xmath1208 and @xmath654 has a non - trivial endomorphism that factors over a projective module in the same tube .
[ cor : triven ] a module @xmath1205 has trivial endomorphism ring if and only if @xmath719=0 $ ] , or @xmath719\in r_s^+$ ] and @xmath976 does not belong to the following list : @xmath1209 with @xmath1210 and @xmath1211 .
this follows from together with our description of @xmath1102 and @xmath1103 .
we shall now explain how @xmath1057 parametrizes ( i ) the indecomposable @xmath27-modules and ( ii ) the set of dimension vectors of indecomposable @xmath113-modules modulo the galois group action . from ( ii ) we shall deduce the main result of this section , namely the parametrization of the indecomposable irreducible components of @xmath27-modules by @xmath122 .
[ [ section-64 ] ] by [ artlam ] , the indecomposable @xmath27-modules are in one - to - one correspondence with the @xmath113-modules of the class @xmath966 .
this class decomposes into tubular families @xmath1095 as shown in [ expldec ] .
the projective modules @xmath1212 and @xmath1213 appear at the mouth of the non - homogeneous tube @xmath1214 , the module @xmath1215 at the mouth of the non - homogeneous tube @xmath1216 , and the modules @xmath1217 and @xmath1218 at the mouth of the non - homogeneous tube @xmath1219 . as a result
we obtain the following parametrization of the indecomposable @xmath27-modules by @xmath1057 .
[ rootsdimvectors ] let @xmath1070 , @xmath1135 and @xmath1136 . then the following hold : * if @xmath199 does not divide @xmath1132 then there exists a one - to - one correspondence between @xmath1220 and the set of indecomposable @xmath27-modules of quasi - length @xmath1132 in the non - homogeneous tube @xmath1221 with @xmath1222 . this correspondence maps @xmath1223 to @xmath1224 with @xmath1225 = { \mathbf{d}}$ ] . * if @xmath1144 then @xmath1226 where @xmath1146 and @xmath1147 . there is an infinite set of indecomposable @xmath27-modules @xmath1224 with @xmath1225 = { \mathbf{h}}_{a , b}$ ] , parametrized by the weighted projective line @xmath119 .
more precisely , for each ordinary point @xmath1096 there is an indecomposable module of quasi - length @xmath1132 in the tube @xmath1094 , and for each exceptional point @xmath1227 , there are @xmath1228 indecomposable modules of quasi - length @xmath1132 on @xmath1221 . * the only indecomposable @xmath27-modules not appearing in the above lists are the five indecomposable projective modules .
[ [ section-65 ] ] recall from corollary [ cor : eta ] that we have a canonical map @xmath1229 . we are going to define a ` right inverse ' @xmath1230 of @xmath1231 .
define @xmath1232 by @xmath1233 [ deltares ] with the above definition of @xmath1234 we have : * @xmath1234 induces a well - defined and injective map @xmath1235 ; * the only elements of @xmath1236 not in the image of @xmath1230 are @xmath1237 with @xmath1210 and @xmath1211 ; * the map @xmath1230 restricts to a map @xmath1238 .
the only elements of @xmath1239 not in the image of @xmath1240 are @xmath1241 .
first we have to show that @xmath1242 is the dimension vector of an indecomposable @xmath113-module .
this is clear for @xmath1243 .
for @xmath1244 we notice that @xmath1245 gives an isometry @xmath1246}})$ ] , so that @xmath1247 consists of dimension vectors of objects in @xmath972}}$ ] .
the remaining two cases are treated directly .
it is easy to calculate the map @xmath1231 ( corollary [ cor : eta ] ) explicitly with .
it follows that @xmath1248 and the rest of ( a ) follows since @xmath1236 is a transversal for the action of @xmath114 on @xmath1249 , see proposition [ prp : transv ] .
now we obtain @xmath1250 from the description of the fibres of @xmath1231 in corollary [ cor : eta ] , and ( c ) follows from corollary [ cor : triven ] .
[ [ section-66 ] ] collecting the results of theorem [ thmgs ] , theorem [ thm1 ] and proposition [ deltares ] we can now state the following parametrization of the indecomposable irreducible components of varieties of @xmath27-modules and of the corresponding multisegments .
let @xmath1251 be the irreducible components containing the five indecomposable projective @xmath27-modules .
let @xmath1252 denote the corresponding multisegments , namely @xmath1253,\qquad ( j=1,\ldots,5).\ ] ] [ paramet ] * the map @xmath1254 is a one - to - one correspondence from the set @xmath1197 of schur roots of the ringel form @xmath121 in @xmath1003 to the set @xmath1255 . *
the map @xmath1256 is a one - to - one correspondence from @xmath1197 to @xmath1257 .
note that the descriptions of @xmath1258 and @xmath811 are completely explicit , so that we get a very concrete parametrization of the factors arising in the canonical factorization of the elements of @xmath84 .
( i)let @xmath1259
. then @xmath1260 and @xmath1261+[1,2]+[1,3]+2\,[2,3]+[3,4]+[3,5]+[4,4]+[5,5].\ ] ] ( ii)let @xmath1262
. then @xmath1263 and @xmath1261+[2,2]+[2,3]+[3,4]+[4,5].\ ] ]
recall that the component graph @xmath1264 has for vertices the indecomposable irreducible components of the varieties of @xmath27-modules , and two vertices @xmath505 and @xmath506 are connected by an edge if and only if @xmath56 is an irreducible component , or equivalently @xmath1265 .
there are edges from the irreducible components @xmath1266 to every other vertex .
the following theorem describes all remaining edges . in agreement with theorem [ paramet ] ( a )
, we shall label the vertices other than @xmath1267 by the elements of @xmath122 . in order to state the theorem
we introduce the following definition .
we call a pair of schur roots @xmath1268 _ critical _ if the following three conditions hold : * @xmath1269 for some @xmath1270 and @xmath1271 , * @xmath1272 , * @xmath1273 .
[ thm4 ] two schur roots @xmath1001 and @xmath1002 are connected by an edge in @xmath1264 if and only if the following two conditions hold : * @xmath1274 and @xmath1275 , * @xmath1268 is not critical , or @xmath1276 where @xmath1277 .
this follows from an adaptation of ( * ? ? ?
* theorem 1.3 , lemma 6.4 ) .
thus , the edges of the component graph of @xmath27 are completely determined by the bilinear form @xmath121 and the coxeter matrix @xmath104 .
moreover there is an edge between @xmath195 and @xmath1278 if and only if there is an edge between @xmath1279 and @xmath1280 .
it was already proved in @xcite that @xmath509 and @xmath755 are well - defined and bijective .
the map @xmath1281 is bijective by definition .
it remains to explicitly construct the map @xmath1282 .
it is enough to prove theorem [ thm1 ] for @xmath909 , since @xmath656 ( @xmath1283 ) are full convex subalgebras of @xmath1284 .
we will use the following result from ( * ? ? ? *
theorem 1 , 2 ) : let @xmath182 be a tame quasi - tilted basic algebra , and let @xmath1001 be a dimension vector of an indecomposable @xmath182-module
. then @xmath1285 has at most two irreducible components , and @xmath1285 is irreducible if and only if @xmath1001 is not of one of the following forms : * @xmath1286 where @xmath1189 and @xmath1287 are connected positive vectors with disjoint support , @xmath1288 , @xmath1289 and @xmath1290 for all entries @xmath1291 of @xmath1287 , * @xmath1292 where @xmath1189 and @xmath1293 are connected positive vectors with @xmath1294 , @xmath1295 and @xmath1296 .
the algebras @xmath1297}}$ ] and @xmath917}}$ ] are both tubular algebras , in particular they are tame quasi - tilted algebras .
all connected positive vectors @xmath1189 for @xmath955}}$ ] such that @xmath1298}}= 0 $ ] are of the form @xmath1299 } } + b{{\mathbf{h}}^{[i+1]}}$ ] where @xmath1300 .
we have @xmath1301}},{{\mathbf{h}}^{[i+1]}}\rangle}^{[i , i+1]}}= 6 $ ] and @xmath1302}},{{\mathbf{h}}^{[i]}}\rangle}^{[i , i+1 ] } } = -6 $ ] .
it follows that the case @xmath1250 above can not occur for @xmath1303}}$ ] , @xmath1304 . for @xmath1305
we are in case @xmath1306 precisely when @xmath1307 where @xmath1308 and @xmath1309 for @xmath1310 we are in case @xmath1306 precisely when @xmath1311 where @xmath1312 and @xmath1313 let @xmath654 be an indecomposable @xmath113-module , and let @xmath976 be its dimension vector .
if @xmath1314 then @xmath1315 , and if @xmath1316 then @xmath1317 .
one easily checks that @xmath1318 is a schur root if and only if @xmath1319 .
set @xmath1320 and @xmath1321 , @xmath1322 .
assume that @xmath1323 is a schur root whose support lies ( up to shift ) in @xmath101 or @xmath953 , and assume @xmath1324 . by the result mentioned above this
implies that @xmath1325 is irreducible .
thus @xmath1326 and @xmath1327 next , assume @xmath1328 .
thus @xmath1325 has exactly two irreducible components .
furthermore , we know that @xmath1329 for some indecomposable @xmath113-module @xmath1330 . for any vertex @xmath994 of @xmath1011 let @xmath1331 be the corresponding simple @xmath113-module , and let @xmath1332 be the irreducible component consisting of the single point corresponding to @xmath1331
. then @xmath1333 for @xmath1334 and all @xmath793 , and @xmath1335 for @xmath1336 and all @xmath793 .
thus ( * ? ? ?
* theorem 1.2 ) implies that @xmath1337 ( resp .
@xmath1338 ) are irreducible components provided @xmath1334 ( resp .
@xmath1336 ) .
exactly one of these irreducible components lies in @xmath1325 , we denote this component by @xmath1339 .
thus @xmath1325 contains exactly one indecomposable irreducible component , namely @xmath1340 , and exactly one decomposable irreducible component , namely @xmath1339 . if @xmath234 is an irreducible component , then @xmath1341 thus , if @xmath1342 , then @xmath1343 .
this yields @xmath1326 and @xmath1327 for @xmath1344 it is not difficult to construct the module @xmath1330 explicitly .
we get @xmath1345 + [ 2,2 ] + [ 2,4 ] + [ 3,3 ] + [ 4,5]\ ] ] and @xmath1346 + [ 2,4 ] + [ 3,3 ] + [ 4,4 ] + 2\,[5,5].\ ] ] the only schur roots in @xmath1347 whose support is ( up to shift ) not contained in @xmath101 or @xmath953 are @xmath1348 ( these are all in @xmath1102 ) , and @xmath1349 ( these are all in @xmath1103 ) . here @xmath1350 are ( up to shift ) the dimension vectors of the indecomposable projective @xmath113-modules as displayed in [ sus : isomt ] . for each @xmath1351 , there exists an indecomposable @xmath113-module @xmath1330 such that @xmath1352 . for @xmath1353
it is easy to construct @xmath1330 explicitly . in these cases
, it follows that @xmath1326 , thus @xmath1327 for any element @xmath1354 and any @xmath113-module @xmath654 , let @xmath1355 be the linear map defined by the @xmath113-module action of @xmath1356 on @xmath654 , i.e. @xmath1357 . if @xmath654 is a submodule of a module @xmath21 , then we have @xmath1358 for all @xmath1356 .
now let @xmath1359 in section [ tnull ] we can see that @xmath1330 contains two indecomposable submodules @xmath1360 and @xmath1361 with dimension vectors @xmath1362 respectively .
we get @xmath1363 since the support of @xmath1364 and @xmath1365 lies in @xmath953 and @xmath101 , respectively , we get @xmath1366 this enables us to compute the ranks of the maps @xmath1367 and @xmath1368 for all paths @xmath206 in @xmath1011 of the form @xmath1369 , @xmath1370 .
then we can use the above rank inequality , and get the rank of @xmath1371 for any path @xmath206 .
it turns out that @xmath1372 the case @xmath1373 is done in a similar way , and we get again @xmath1374
[ [ section-67 ] ] by proposition [ infinite_type ] , the preprojective algebra @xmath1375 is tame if and only if the quiver @xmath62 is of dynkin type @xmath6 or @xmath123 .
using the same methods as in this paper it is possible to obtain a complete analogue of theorem [ thmc ] for type @xmath123 . in this case
, @xmath27 has a galois covering which is isomorphic to the repetitive algebra of a tubular algebra of type @xmath1376 , and the corresponding root system is an elliptic root system of type @xmath1377 .
we plan to give a detailed analysis of this case in a forthcoming publication .
[ [ section-68 ] ] it is shown in @xcite that for all dynkin types , the algebra @xmath20 $ ] has the structure of an ( upper ) cluster algebra , and that it has finite type as a cluster algebra if and only @xmath10 is of lie type @xmath1378 . in that case
one can associate to @xmath20 $ ] a root system @xmath1379 called its _ cluster type _ , which controls the combinatorics of the cluster variables and of the cluster sets .
more precisely , the cluster variables are parametrized by the set @xmath1380 of almost positive roots of @xmath1379 , and the pairs of cluster variables which can occur simultaneously in a cluster set can be explicitly described in terms of @xmath1379 and a piecewise linear coxeter transformation acting on @xmath1380 .
the cartan matrix @xmath182 of @xmath1379 can be obtained by a certain symmetrization procedure from the principal part @xmath1381 of the exchange matrix of @xmath20 $ ] at certain vertices @xmath147 of its exchange graph ( see @xcite ) . for @xmath10 of lie type @xmath1382 ,
the algebra @xmath20 $ ] has cluster type @xmath1383 respectively @xcite . as mentioned in [ redatla ] , these root systems also occur in our setting in the following way . for @xmath1384
, we have @xmath1385 where @xmath1386 is a quiver of type @xmath1383 respectively . [ [ section-69 ] ] at the moment , there is no notion of cluster type for the algebras @xmath20 $ ] which are not of finite type as cluster algebras .
the results of this paper strongly suggest that if such a cluster type exists for @xmath10 of lie type @xmath6 ( _ resp . _
@xmath123 ) , then it should be the elliptic type @xmath8 ( _ resp . _
@xmath1377 ) in the notation of saito , or the tubular type @xmath7 ( _ resp . _
@xmath1376 ) in the language of ringel . remember in particular that @xmath1387 where @xmath119 is a weighted projective line of type @xmath7 .
[ [ section-70 ] ] here is another remark supporting that guess . for @xmath10 of lie type @xmath6
one can find a vertex @xmath147 of the exchange graph of @xmath20 $ ] at which the principal part of the exchange matrix is ( up to simultaneous permutations of rows and columns ) @xmath1388 we suggest to take as symmetric counterpart of @xmath1381 the matrix @xmath1389 note that @xmath182 has two positive entries off the diagonal , so it differs from the matrix obtained from @xmath1381 by the symmetrizing procedure of @xcite .
it turns out that @xmath182 is exactly the cartan matrix of the root system @xmath8 , that is , the matrix of scalar products of a basis of simple roots in the sense of saito @xcite .
it can be visualized with the help of the dynkin type diagram of figure [ figu2 ] in which an ordinary edge between @xmath199 and @xmath608 means that @xmath1390 and the dashed line between @xmath345 and @xmath1391 means that @xmath1392 ( see @xcite ) .
there is a similar fact for type @xmath123 and @xmath1377 .
@xmath1393\ar@{.}[dd ] & & & & { { \begin{smallmatrix}}&1\\1\\&0\\0{\end{smallmatrix } } } \ar@{.}[dd]\\ & { { \begin{smallmatrix}}&0\\0\\&1\\0{\end{smallmatrix } } } \ar[dr ] & & { { \begin{smallmatrix}}&0\\1\\&0\\0{\end{smallmatrix } } } \ar[ur]\\ & & { { \begin{smallmatrix}}&0\\1\\&1\\0{\end{smallmatrix } } } \ar[ur]&&\\ } \ ] ] @xmath1394\ar@{.}[dd ] & & & & & & { { \begin{smallmatrix}}0&&1\\&1\\1&&0\\&0\\0&&0{\end{smallmatrix}}}\ar@{.}[dd]\\ & { { \begin{smallmatrix}}0&&0\\&0\\1&&0\\&1\\0&&0{\end{smallmatrix } } } \ar[dr ] & & { { \begin{smallmatrix}}0&&0\\&0\\0&&1\\&0\\0&&0{\end{smallmatrix } } } \ar[dr ] & & { { \begin{smallmatrix}}0&&0\\&1\\1&&0\\&0\\0&&0{\end{smallmatrix } } } \ar[dr]\ar[ur]\\ { { \begin{smallmatrix}}0&&0\\&0\\0&&0\\&1\\0&&0{\end{smallmatrix } } } \ar[dr]\ar[ur]\ar@{.}[dd ] & & { { \begin{smallmatrix}}0&&0\\&0\\1&&1\\&1\\0&&0{\end{smallmatrix } } } \ar[dr]\ar[r]\ar[ur ] & { { \begin{smallmatrix}}0&&0\\&1\\1&&1\\&1\\0&&0{\end{smallmatrix } } } \ar[r ] & { { \begin{smallmatrix}}0&&0\\&1\\1&&1\\&0\\0&&0{\end{smallmatrix } } } \ar[dr]\ar[ur ] & & { { \begin{smallmatrix}}0&&0\\&1\\0&&0\\&0\\0&&0{\end{smallmatrix}}}\ar@{.}[dd]\\ & { { \begin{smallmatrix}}0&&0\\&0\\0&&1\\&1\\0&&0{\end{smallmatrix } } } \ar[ur ] & & { { \begin{smallmatrix}}0&&0\\&0\\1&&0\\&0\\0&&0{\end{smallmatrix } } } \ar[ur ] & & { { \begin{smallmatrix}}0&&0\\&1\\0&&1\\&0\\0&&0{\end{smallmatrix } } } \ar[dr]\ar[ur]\\ { { \begin{smallmatrix}}0&&0\\&0\\0&&1\\&1\\1&&0{\end{smallmatrix } } } \ar[ur ] & & & & & & { { \begin{smallmatrix}}1&&0\\&1\\0&&1\\&0\\0&&0{\end{smallmatrix}}}\\ } \ ] ] @xmath1395&&&{{\begin{smallmatrix}}0&&0\\&0&&1\\0&&1\\&1&&0\\1&&0{\end{smallmatrix}}}\ar[dr ] & & & & & & { { \begin{smallmatrix}}1&&0\\&1&&0\\0&&1\\&0&&1\\0&&0{\end{smallmatrix}}}\ar[dr ] & & & \ar@{.}[d]\\ { { \begin{smallmatrix}}0&&0\\&0&&0\\0&&0\\&0&&1\\0&&0{\end{smallmatrix}}}\ar[dr]\ar@{.}[dd ] & & { { \begin{smallmatrix}}0&&0\\&0&&0\\0&&1\\&1&&0\\1&&0{\end{smallmatrix}}}\ar[ur]\ar[dr ] & & { { \begin{smallmatrix}}0&&0\\&0&&1\\0&&1\\&1&&0\\0&&0{\end{smallmatrix}}}\ar[dr ] & & { { \begin{smallmatrix}}0&&0\\&0&&0\\1&&0\\&0&&0\\0&&0{\end{smallmatrix}}}\ar[dr ] & & { { \begin{smallmatrix}}0&&0\\&1&&0\\0&&1\\&0&&1\\0&&0{\end{smallmatrix}}}\ar[dr]\ar[ur ] & & { { \begin{smallmatrix}}1&&0\\&1&&0\\0&&1\\&0&&0\\0&&0{\end{smallmatrix}}}\ar[dr ] & & { { \begin{smallmatrix}}0&&0\\&0&&1\\0&&0\\&0&&0\\0&&0{\end{smallmatrix } } } \ar@{.}[dd]\\ & { { \begin{smallmatrix}}0&&0\\&0&&0\\0&&1\\&1&&1\\1&&0{\end{smallmatrix}}}\ar[ur]\ar[dr ] & & { { \begin{smallmatrix}}0&&0\\&0&&0\\0&&1\\&1&&0\\0&&0{\end{smallmatrix}}}\ar[ur]\ar[dr ] & & { { \begin{smallmatrix}}0&&0\\&0&&1\\1&&1\\&1&&0\\0&&0{\end{smallmatrix}}}\ar[ur]\ar[dr ] & & { { \begin{smallmatrix}}0&&0\\&1&&0\\1&&1\\&0&&1\\0&&0{\end{smallmatrix}}}\ar[ur]\ar[dr ] & & { { \begin{smallmatrix}}0&&0\\&1&&0\\0&&1\\&0&&0\\0&&0{\end{smallmatrix}}}\ar[ur]\ar[dr ] & & { { \begin{smallmatrix}}1&&0\\&1&&1\\0&&1\\&0&&0\\0&&0{\end{smallmatrix}}}\ar[ur]\ar[dr]&\\ { { \begin{smallmatrix}}0&&0\\&0&&0\\1&&1\\&2&&1\\1&&1{\end{smallmatrix } } } \ar[dr ] \ar[ur]\ar@{.}[d ] & & { { \begin{smallmatrix}}0&&0\\&0&&0\\0&&1\\&1&&1\\0&&0{\end{smallmatrix } } } \ar[dr ] \ar[ur ] & & { { \begin{smallmatrix}}0&&0\\&0&&0\\1&&1\\&1&&0\\0&&0{\end{smallmatrix } } } \ar[dr ] \ar[ur ] & & { { \begin{smallmatrix}}0&&0\\&1&&1\\1&&2\\&1&&1\\0&&0{\end{smallmatrix } } } \ar[dr ] \ar[ur ] & & { { \begin{smallmatrix}}0&&0\\&1&&0\\1&&1\\&0&&0\\0&&0{\end{smallmatrix } } } \ar[dr ] \ar[ur ] & & { { \begin{smallmatrix}}0&&0\\&1&&1\\0&&1\\&0&&0\\0&&0{\end{smallmatrix } } } \ar[dr ] \ar[ur ] & & { { \begin{smallmatrix}}1&&1\\&2&&1\\1&&1\\&0&&0\\0&&0{\end{smallmatrix}}}\ar@{.}[d]\\ { { \begin{smallmatrix}}0&&0\\&0&&0\\0&&0\\&1&&0\\0&&0{\end{smallmatrix } } } \ar[r]\ar@{.}[d ] & { { \begin{smallmatrix}}0&&0\\&0&&0\\1&&1\\&2&&1\\0&&1{\end{smallmatrix}}}\ar[r]\ar[dr]\ar[ur ] & { { \begin{smallmatrix}}0&&0\\&0&&0\\1&&1\\&1&&1\\0&&1{\end{smallmatrix } } } \ar[r]\ar[ddr ] & { { \begin{smallmatrix}}0&&0\\&0&&0\\1&&1\\&1&&1\\0&&0{\end{smallmatrix}}}\ar[r]\ar[dr]\ar[ur ] & { { \begin{smallmatrix}}0&&0\\&1&&0\\1&&1\\&1&&1\\0&&0{\end{smallmatrix}}}\ar[r ] & { { \begin{smallmatrix}}0&&0\\&1&&0\\1&&2\\&1&&1\\0&&0{\end{smallmatrix}}}\ar[r]\ar[dr]\ar[ur ] & { { \begin{smallmatrix}}0&&0\\&0&&0\\0&&1\\&0&&0\\0&&0{\end{smallmatrix } } } \ar[r ] & { { \begin{smallmatrix}}0&&0\\&1&&1\\1&&2\\&1&&0\\0&&0{\end{smallmatrix}}}\ar[r]\ar[dr]\ar[ur ] & { { \begin{smallmatrix}}0&&0\\&1&&1\\1&&1\\&1&&0\\0&&0{\end{smallmatrix}}}\ar[r]\ar[ddr ] & { { \begin{smallmatrix}}0&&0\\&1&&1\\1&&1\\&0&&0\\0&&0{\end{smallmatrix}}}\ar[r]\ar[dr]\ar[ur ] & { { \begin{smallmatrix}}0&&1\\&1&&1\\1&&1\\&0&&0\\0&&0{\end{smallmatrix } } } \ar[r ] & { { \begin{smallmatrix}}0&&1\\&2&&1\\1&&1\\&0&&0\\0&&0{\end{smallmatrix}}}\ar[r]\ar[dr]\ar[ur ] & { { \begin{smallmatrix}}0&&0\\&1&&0\\0&&0\\&0&&0\\0&&0{\end{smallmatrix}}}\ar@{.}[d]\\ \ar@{.}[d]{{\begin{smallmatrix}}0&&0\\&0&&0\\0&&1\\&1&&1\\0&&1{\end{smallmatrix } } } \ar[ur ] & & { { \begin{smallmatrix}}0&&0\\&0&&0\\1&&0\\&1&&0\\0&&0{\end{smallmatrix } } } \ar[ur ] & & { { \begin{smallmatrix}}0&&0\\&0&&0\\0&&1\\&0&&1\\0&&0{\end{smallmatrix } } } \ar[ur ] & & { { \begin{smallmatrix}}0&&0\\&1&&0\\1&&1\\&1&&0\\0&&0{\end{smallmatrix } } } \ar[ur ] & & { { \begin{smallmatrix}}0&&0\\&0&&1\\0&&1\\&0&&0\\0&&0{\end{smallmatrix}}}\ar[ur ] & & { { \begin{smallmatrix}}0&&0\\&1&&0\\1&&0\\&0&&0\\0&&0{\end{smallmatrix}}}\ar[ur ] & & { { \begin{smallmatrix}}0&&1\\&1&&1\\0&&1\\&0&&0\\0&&0{\end{smallmatrix}}}\ar@{.}[d]\\ & & & { { \begin{smallmatrix}}0&&0\\&1&&0\\1&&1\\&1&&1\\0&&1{\end{smallmatrix } } } \ar[uur ] & & & & & & { { \begin{smallmatrix}}0&&1\\&1&&1\\1&&1\\&1&&0\\0&&0{\end{smallmatrix } } } \ar[uur ] & & & \\ } \ ] ] @xmath1396&{\mathfrak{m}}_{21 } & = [ 1,3]+[4,4]\\ { \mathfrak{m}}_2 & = [ 2,2]&{\mathfrak{m}}_{22 } & = [ 1,2]+[3,3]+[4,4]\\ { \mathfrak{m}}_3 & = [ 3,3]&{\mathfrak{m}}_{23 } & = [ 1,1]+[2,2]+[3,4]\\ { \mathfrak{m}}_4 & = [ 4,4]&{\mathfrak{m}}_{24 } & = [ 1,1]+[2,4]\\ { \mathfrak{m}}_5 & = [ 1,2]&{\mathfrak{m}}_{25 } & = [ 1,2]+[3,4]\\ { \mathfrak{m}}_6 & = [ 1,1]+[2,2]&{\mathfrak{m}}_{26 } & = [ 1,1]+[2,3]+[4,4]\\ { \mathfrak{m}}_7 & = [ 2,3]&{\mathfrak{m}}_{27 } & = [ 1,2]+[2,3]+[4,4]\\ { \mathfrak{m}}_8 & = [ 2,2]+[3,3 ] & { \mathfrak{m}}_{28 } & = [ 1,2]+[2,4]\\ { \mathfrak{m}}_9 & = [ 3,4]&{\mathfrak{m}}_{29 } & = [ 1,3]+[3,4]\\ { \mathfrak{m}}_{10 } & = [ 3,3]+[4,4]&{\mathfrak{m}}_{30 } & = [ 1,1]+[2,3]+[3,4]\\ { \mathfrak{m}}_{11 } & = [ 1,3]&{\mathfrak{m}}_{31 } & = [ 1,2]+[2,3]+[3,3]+[4,4]\\ { \mathfrak{m}}_{12 } & = [ 1,1]+[2,2]+[3,3]&{\mathfrak{m}}_{32 } & = [ 1,2]+[2,4]+[3,3]\\ { \mathfrak{m}}_{13 } & = [ 1,2]+[3,3]&{\mathfrak{m}}_{33 } & = [ 1,3]+[2,2]+[3,4]\\ { \mathfrak{m}}_{14 } & = [ 1,1]+[2,3]&{\mathfrak{m}}_{34 } & = [ 1,1]+[2,2]+[2,3]+[3,4]\\ { \mathfrak{m}}_{15 } & = [ 1,2]+[2,3]&{\mathfrak{m}}_{35 } & = [ 1,1]+[1,3]+[2,2]+[3,4]\\ { \mathfrak{m}}_{16 } & = [ 2,4]&{\mathfrak{m}}_{36 } & = [ 1,2]+[2,4]+[3,3]+[4,4]\\ { \mathfrak{m}}_{17 } & = [ 2,2]+[3,3]+[4,4]&{\mathfrak{m}}_{37 } & = [ 1,4]\\ { \mathfrak{m}}_{18 } & = [ 2,3]+[4,4]&{\mathfrak{m}}_{38 } & = [ 1,3]+[2,4]\\ { \mathfrak{m}}_{19 } & = [ 2,2]+[3,4]&{\mathfrak{m}}_{39 } & = [ 1,2]+[2,3]+[3,4]\\ { \mathfrak{m}}_{20 } & = [ 2,3]+[3,4]&{\mathfrak{m}}_{40 } & = [ 1,1]+[2,2]+[3,3]+[4,4]\\\end{aligned}\ ] ] @xmath1397&&&{{\mathfrak{m}}_{37}}\ar[dr ] & & & & & & { { \mathfrak{m}}_{40}}\ar[dr ] & & & \ar@{.}[d]\\ { { \mathfrak{m}}_4}\ar[dr]\ar@{.}[dd ] & & { { \mathfrak{m}}_{11}}\ar[ur]\ar[dr ] & & { { \mathfrak{m}}_{16}}\ar[dr ] & & { { \mathfrak{m}}_1}\ar[dr ] & & { { \mathfrak{m}}_{17}}\ar[dr]\ar[ur ] & & { { \mathfrak{m}}_{12}}\ar[dr ] & & { { \mathfrak{m}}_4 } \ar@{.}[dd]\\ & { { \mathfrak{m}}_{21}}\ar[ur]\ar[dr ] & & { { \mathfrak{m}}_7}\ar[ur]\ar[dr ] & & { { \mathfrak{m}}_{24}}\ar[ur]\ar[dr ] & & { { \mathfrak{m}}_{22}}\ar[ur]\ar[dr ] & & { { \mathfrak{m}}_8}\ar[ur]\ar[dr ] & & { { \mathfrak{m}}_{23}}\ar[ur]\ar[dr]&\\ { { \mathfrak{m}}_{35 } } \ar[dr ] \ar[ur]\ar@{.}[d ] & & { { \mathfrak{m}}_{18 } } \ar[dr ] \ar[ur ] & & { { \mathfrak{m}}_{14 } } \ar[dr ] \ar[ur ] & & { { \mathfrak{m}}_{36 } } \ar[dr ] \ar[ur ] & & { { \mathfrak{m}}_{13 } } \ar[dr ] \ar[ur ] & & { { \mathfrak{m}}_{19 } } \ar[dr ] \ar[ur ] & & { { \mathfrak{m}}_{35}}\ar@{.}[d]\\ { { \mathfrak{m}}_2 } \ar[r]\ar@{.}[d ] & { { \mathfrak{m}}_{34}}\ar[r]\ar[dr]\ar[ur ] & { { \mathfrak{m}}_{30 } } \ar[r]\ar[ddr ] & { { \mathfrak{m}}_{26}}\ar[r]\ar[dr]\ar[ur ] & { { \mathfrak{m}}_{27}}\ar[r ] & { { \mathfrak{m}}_{31}}\ar[r]\ar[dr]\ar[ur ] & { { \mathfrak{m}}_3 } \ar[r ] & { { \mathfrak{m}}_{32}}\ar[r]\ar[dr]\ar[ur ] & { { \mathfrak{m}}_{28}}\ar[r]\ar[ddr ] & { { \mathfrak{m}}_{25}}\ar[r]\ar[dr]\ar[ur ] & { { \mathfrak{m}}_{29 } } \ar[r ] & { { \mathfrak{m}}_{33}}\ar[r]\ar[dr]\ar[ur ] & { { \mathfrak{m}}_2}\ar@{.}[d]\\ \ar@{.}[d]{{\mathfrak{m}}_{20 } } \ar[ur ] & & { { \mathfrak{m}}_6 } \ar[ur ] & & { { \mathfrak{m}}_{10 } } \ar[ur ] & & { { \mathfrak{m}}_{15 } } \ar[ur ] & & { { \mathfrak{m}}_9}\ar[ur ] & & { { \mathfrak{m}}_5}\ar[ur ] & & { { \mathfrak{m}}_{20}}\ar@{.}[d]\\ & & & { { \mathfrak{m}}_{39 } } \ar[uur ] & & & & & & { { \mathfrak{m}}_{38 } } \ar[uur ] & & & \\ } \ ] ] @xmath1398&\cdot \ar[dr ] & & \bullet \ar[dr ] & & \cdot \ar[dr]&\ar@{.}[d ] & & & \ar@{.}[d]&x_2 \ar[dr ] & & \bullet \ar[dr ] & & \cdot \ar[dr]&\ar@{.}[d]\\ x_1 \ar[ur]\ar[dr]\ar@{.}[d ] & & \bullet \ar[ur]\ar[dr ] & & \bullet \ar[ur]\ar[dr ] & & \cdot\ar@{.}[d ] & & & \cdot \ar[ur]\ar[dr]\ar@{.}[d ] & & \cdot \ar[ur]\ar[dr ] & & \bullet \ar[ur]\ar[dr ] & & \cdot \ar@{.}[d]\\ & \cdot \ar[ur ] & & \bullet \ar[ur ] & & \cdot \ar[ur ] & & & & & \cdot \ar[ur ] & & \cdot \ar[ur ] & & \bullet \ar[ur ] & } \ ] ] @xmath1399\ar@{.}[dd ] & & { \bullet } \ar[dr ] & & { \cdot } \ar[dr ] & & { \cdot } \ar[dr ] & & { \cdot } \ar[dr ] & & { \bullet } \ar[dr ] & & { \cdot}\ar@{.}[dd ] \\ & { \cdot } \ar[dr]\ar[ur ] & & { \bullet } \ar[dr]\ar[ur ] & & { \cdot } \ar[dr]\ar[ur ] & & { \cdot } \ar[dr]\ar[ur ] & & { \bullet } \ar[dr]\ar[ur ] & & { \cdot } \ar[dr]\ar[ur ] & & \\ { \cdot } \ar[dr]\ar[ur]\ar@{.}[d ] & & { \cdot } \ar[dr]\ar[ur ] & & { \bullet } \ar[dr]\ar[ur ] & & { \cdot } \ar[dr]\ar[ur ] & & { \bullet } \ar[dr]\ar[ur ] & & { \cdot } \ar[dr]\ar[ur ] & & { \cdot}\ar@{.}[d ] \\ { \cdot }
\ar[r]\ar@{.}[d ] & { \cdot } \ar[r]\ar[dr]\ar[ur ] & { \cdot } \ar[r ] & { \cdot } \ar[r]\ar[dr]\ar[ur ] & { \cdot } \ar[r ] & { \bullet } \ar[r]\ar[dr]\ar[ur ] & { \bullet } \ar[r ] & { \bullet } \ar[r]\ar[dr]\ar[ur ] & { \cdot } \ar[r ] & { \cdot } \ar[r]\ar[dr]\ar[ur ] & { \cdot } \ar[r ] & { \cdot } \ar[r]\ar[dr]\ar[ur ] & { \cdot}\ar@{.}[d]\\ { \cdot } \ar[ur ] & & { \cdot}\ar[ur ] & & { \cdot } \ar[ur ] & & { \bullet}\ar[ur ] & & { \cdot } \ar[ur ] & & { \cdot}\ar[ur ] & & { \cdot } } \ ] ] @xmath1400\ar@{.}[dd ] & & { \cdot } \ar[dr ] & & { \bullet } \ar[dr ] & & { \cdot } \ar[dr ] & & { \bullet } \ar[dr ] & & { \cdot } \ar[dr ] & & { \cdot}\ar@{.}[dd ] \\ & { \cdot } \ar[dr]\ar[ur ] & & { \bullet } \ar[dr]\ar[ur ] & & { \bullet } \ar[dr]\ar[ur ] & & { \bullet } \ar[dr]\ar[ur ] & & { \bullet } \ar[dr]\ar[ur ] & & { \cdot } \ar[dr]\ar[ur ] & & \\ x_2 \ar[dr]\ar@{.}[d]\ar[ur ] & & { \bullet } \ar[dr]\ar[ur ] & & { \bullet } \ar[dr]\ar[ur ] & & { \bullet } \ar[dr]\ar[ur ] & & { \bullet } \ar[dr]\ar[ur ] & & { \bullet } \ar[dr]\ar[ur ] & & { \cdot}\ar@{.}[d ] \\ { \cdot } \ar[r]\ar@{.}[d ] & { \cdot } \ar[r]\ar[dr]\ar[ur ] & { \cdot } \ar[r ] & { \bullet } \ar[r]\ar[dr]\ar[ur ] & { \bullet } \ar[r ] & { \bullet } \ar[r]\ar[dr]\ar[ur ] & { \bullet } \ar[r ] & { \bullet } \ar[r]\ar[dr]\ar[ur ] & { \bullet } \ar[r ] & { \bullet } \ar[r]\ar[dr]\ar[ur ] & { \cdot } \ar[r ] & { \cdot } \ar[r]\ar[dr]\ar[ur ] & { \cdot}\ar@{.}[d]\\ { \cdot } \ar[ur ] & & { \cdot}\ar[ur ] & & { \bullet } \ar[ur ] & & { \bullet}\ar[ur ] & & { \bullet } \ar[ur ] & & { \cdot}\ar[ur ] & & { \cdot } } \ ] ] @xmath1400\ar@{.}[dd ] & & { \cdot } \ar[dr ] & & { \cdot } \ar[dr ] & & { \bullet } \ar[dr ] & & { \cdot } \ar[dr ] & & { \cdot } \ar[dr ] & & { \cdot}\ar@{.}[dd ] \\ & { \cdot } \ar[dr]\ar[ur ] & & { \cdot } \ar[dr]\ar[ur ] & & { \bullet } \ar[dr]\ar[ur ] & & { \bullet } \ar[dr]\ar[ur ] & & { \cdot } \ar[dr]\ar[ur ] & & { \cdot } \ar[dr]\ar[ur ] & & \\ { \cdot } \ar[dr]\ar[ur]\ar@{.}[d ] & & { \cdot } \ar[dr]\ar[ur ] & & { \bullet } \ar[dr]\ar[ur ] & & { \bullet } \ar[dr]\ar[ur ] & & { \bullet } \ar[dr]\ar[ur ] & & { \cdot } \ar[dr]\ar[ur ] & & { \cdot}\ar@{.}[d ] \\ x_3 \ar[r]\ar@{.}[d ] & { \cdot } \ar[r]\ar[dr]\ar[ur ] & { \bullet } \ar[r ] & { \bullet } \ar[r]\ar[dr]\ar[ur ] & { \cdot } \ar[r ] & { \bullet } \ar[r]\ar[dr]\ar[ur ] & { \bullet } \ar[r ] & { \bullet } \ar[r]\ar[dr]\ar[ur ] & { \cdot } \ar[r ] & { \bullet } \ar[r]\ar[dr]\ar[ur ] & { \bullet } \ar[r ] & { \cdot } \ar[r]\ar[dr]\ar[ur ] & { \cdot}\ar@{.}[d]\\ { \cdot } \ar[ur ] & & { \cdot}\ar[ur ] & & { \bullet } \ar[ur ] & & { \cdot}\ar[ur ] & & { \bullet } \ar[ur ] & & { \cdot}\ar[ur ] & & { \cdot } } \ ] ] @xmath1400\ar@{.}[dd ] & & { \cdot } \ar[dr ] & & { \cdot } \ar[dr ] & & { \bullet } \ar[dr ] & & { \cdot } \ar[dr ] & & { \cdot } \ar[dr ] & & { \cdot}\ar@{.}[dd ] \\ & { \cdot } \ar[dr]\ar[ur ] & & { \cdot } \ar[dr]\ar[ur ] & & { \bullet } \ar[dr]\ar[ur ] & & { \bullet } \ar[dr]\ar[ur ] & & { \cdot } \ar[dr]\ar[ur ] & & { \cdot } \ar[dr]\ar[ur ] & & \\ { \cdot } \ar[dr]\ar[ur]\ar@{.}[d ] & & { \cdot } \ar[dr]\ar[ur ] & & { \bullet } \ar[dr]\ar[ur ] & & { \bullet } \ar[dr]\ar[ur ] & & { \bullet } \ar[dr]\ar[ur ] & & { \cdot } \ar[dr]\ar[ur ] & & { \cdot}\ar@{.}[d ] \\ { \cdot } \ar[r]\ar@{.}[d ] & { \cdot } \ar[r]\ar[dr]\ar[ur ] & { \cdot } \ar[r ] & { \bullet } \ar[r]\ar[dr]\ar[ur ] & { \bullet } \ar[r ] & { \bullet } \ar[r]\ar[dr]\ar[ur ] & { \bullet } \ar[r ] & { \bullet } \ar[r]\ar[dr]\ar[ur ] & { \bullet } \ar[r ] & { \bullet } \ar[r]\ar[dr]\ar[ur ] & { \cdot } \ar[r ] & { \cdot } \ar[r]\ar[dr]\ar[ur ] & { \cdot}\ar@{.}[d]\\ x_4 \ar[ur ] & & { \bullet}\ar[ur ] & & { \cdot } \ar[ur ] & & { \bullet}\ar[ur ] & & { \cdot } \ar[ur ] & & { \bullet}\ar[ur ] & & { \cdot } } \ ] ] @xmath1400\ar@{.}[dd ] & & { \cdot } \ar[dr ] & & { \bullet } \ar[dr ] & & { \cdot } \ar[dr ] & & { \cdot } \ar[dr ] & & { \bullet } \ar[dr ] & & { \cdot}\ar@{.}[dd ] \\ & x_5 \ar[dr]\ar[ur ] & & { \bullet } \ar[dr]\ar[ur ] & & { \bullet } \ar[dr]\ar[ur ] & & { \cdot } \ar[dr]\ar[ur ] & & { \bullet } \ar[dr]\ar[ur ] & & { \bullet } \ar[dr]\ar[ur ] & & \\ { \cdot } \ar[dr]\ar[ur]\ar@{.}[d ] & & { \cdot } \ar[dr]\ar[ur ] & & { \bullet } \ar[dr]\ar[ur ] & & { \bullet } \ar[dr]\ar[ur ] & & { \bullet } \ar[dr]\ar[ur ] & & { \bullet } \ar[dr]\ar[ur ] & & { \cdot}\ar@{.}[d ] \\ { \cdot } \ar[r]\ar@{.}[d ] & { \cdot } \ar[r]\ar[dr]\ar[ur ] & { \cdot } \ar[r ] & { \cdot } \ar[r]\ar[dr]\ar[ur ] & { \cdot } \ar[r ] & { \bullet } \ar[r]\ar[dr]\ar[ur ] & { \bullet } \ar[r ] & { \bullet } \ar[r]\ar[dr]\ar[ur ] & { \bullet } \ar[r ] & { \bullet } \ar[r]\ar[dr]\ar[ur ] & { \cdot } \ar[r ] & { \cdot } \ar[r]\ar[dr]\ar[ur ] & { \cdot}\ar@{.}[d]\\ { \cdot } \ar[ur ] & & { \cdot}\ar[ur ] & & { \cdot } \ar[ur ] & & { \bullet}\ar[ur ] & & { \bullet } \ar[ur ] & & { \cdot}\ar[ur ] & & { \cdot } } \ ] ] @xmath1400\ar@{.}[dd ] & & { \cdot } \ar[dr ] & & { \cdot } \ar[dr ] & & { \bullet } \ar[dr ] & & { \bullet } \ar[dr ] & & { \cdot } \ar[dr ] & & { \cdot}\ar@{.}[dd ] \\ & { \cdot } \ar[dr]\ar[ur ] & & { \cdot } \ar[dr]\ar[ur ] & & { \bullet } \ar[dr]\ar[ur ] & & { \bullet } \ar[dr]\ar[ur ] & & { \bullet } \ar[dr]\ar[ur ] & & { \cdot } \ar[dr]\ar[ur ] & & \\ { \cdot } \ar[dr]\ar[ur]\ar@{.}[d ] & & { \cdot } \ar[dr]\ar[ur ] & & { \bullet } \ar[dr]\ar[ur ] & & { \bullet } \ar[dr]\ar[ur ] & & { \bullet } \ar[dr]\ar[ur ] & & { \bullet } \ar[dr]\ar[ur ] & & { \cdot}\ar@{.}[d ] \\ { \cdot } \ar[r]\ar@{.}[d ] & x_6 \ar[r]\ar[dr]\ar[ur ] & { \cdot } \ar[r ] & { \bullet } \ar[r]\ar[dr]\ar[ur ] & { \bullet } \ar[r ] & { \bullet } \ar[r]\ar[dr]\ar[ur ] & { \bullet } \ar[r ] & { \bullet } \ar[r]\ar[dr]\ar[ur ] & { \bullet } \ar[r ] & { \bullet } \ar[r]\ar[dr]\ar[ur ] & { \bullet } \ar[r ] & { \bullet } \ar[r]\ar[dr]\ar[ur ] & { \cdot}\ar@{.}[d]\\ { \cdot } \ar[ur ] & & { \cdot}\ar[ur ] & & { \bullet } \ar[ur ] & & { \bullet}\ar[ur ] & & { \bullet } \ar[ur ] & & { \bullet}\ar[ur ] & & { \cdot } } \ ] ] @xmath1403*{{\begin{smallmatrix}}1&&2&&1\\&3&&3\\1&&2&&1{\end{smallmatrix}}}\restore{\phantom{{\begin{smallmatrix}}1\\1\\1{\end{smallmatrix}}}\hspace{0.16cm}}\ar[rd]\ar@{.}[u ] & & { } \save[]*{{\begin{smallmatrix}}1&&2&&1\\&3&&3\\1&&2&&1{\end{smallmatrix}}}\restore{\phantom{{\begin{smallmatrix}}1\\1\\1{\end{smallmatrix}}}\hspace{0.16cm}}\ar[rd ] & & { } \save[]*{{\begin{smallmatrix}}1&&2&&1\\&3&&3\\1&&2&&1{\end{smallmatrix}}}\restore{\phantom{{\begin{smallmatrix}}1\\1\\1{\end{smallmatrix}}}\hspace{0.16cm}}\ar[rd ] & & { } \save[]*{{\begin{smallmatrix}}1&&2&&1\\&3&&3\\1&&2&&1{\end{smallmatrix}}}\restore{\phantom{{\begin{smallmatrix}}1\\1\\1{\end{smallmatrix}}}\hspace{0.16cm}}\ar@{.}[u]\\ & { } \save[]*{{\begin{smallmatrix}}1&&2&&1\\&3&&3\\1&&2&&1{\end{smallmatrix}}}\restore{\phantom{{\begin{smallmatrix}}1\\1\\1{\end{smallmatrix}}}\hspace{0.16cm}}\ar[rd ] & & { } \save[]*{{\begin{smallmatrix}}1&&2&&1\\&3&&3\\1&&2&&1{\end{smallmatrix}}}\restore{\phantom{{\begin{smallmatrix}}1\\1\\1{\end{smallmatrix}}}\hspace{0.16cm}}\ar[rd ] & & & & & & { } \save[]*{{\begin{smallmatrix}}1&&1&&1\\&2&&2\\1&&1&&1{\end{smallmatrix}}}\restore{\phantom{{\begin{smallmatrix}}1\\1\\1{\end{smallmatrix}}}\hspace{0.16cm}}\ar[rd]\ar[ru ] & & { } \save[]*{{\begin{smallmatrix}}0&&2&&0\\&2&&2\\1&&1&&1{\end{smallmatrix}}}\restore{\phantom{{\begin{smallmatrix}}1\\1\\1{\end{smallmatrix}}}\hspace{0.16cm}}\ar[rd]\ar[ru ] & & { } \save[]*{{\begin{smallmatrix}}1&&1&&1\\&2&&2\\0&&2&&0{\end{smallmatrix}}}\restore{\phantom{{\begin{smallmatrix}}1\\1\\1{\end{smallmatrix}}}\hspace{0.16cm}}\ar[rd]\ar[ru]&\\ { } \save[]*{{\begin{smallmatrix}}1&&1&&0\\&2&&1\\1&&1&&0{\end{smallmatrix}}}\restore{\phantom{{\begin{smallmatrix}}1\\1\\1{\end{smallmatrix}}}\hspace{0.16cm}}\ar[ru]\ar@{.}[uu ] & & { } \save[]*{{\begin{smallmatrix}}0&&1&&1\\&1&&2\\0&&1&&1{\end{smallmatrix}}}\restore{\phantom{{\begin{smallmatrix}}1\\1\\1{\end{smallmatrix}}}\hspace{0.16cm}}\ar[ru ] & & { } \save[]*{{\begin{smallmatrix}}1&&1&&0\\&2&&1\\1&&1&&0{\end{smallmatrix}}}\restore{\phantom{{\begin{smallmatrix}}1\\1\\1{\end{smallmatrix}}}\hspace{0.16cm}}\ar@{.}[uu ] & & & & { } \save[]*{{\begin{smallmatrix}}1&&0&&1\\&1&&1\\0&&1&&0{\end{smallmatrix}}}\restore{\phantom{{\begin{smallmatrix}}1\\1\\1{\end{smallmatrix}}}\hspace{0.16cm}}\ar[ru]\ar@{.}[uu ] & & { } \save[]*{{\begin{smallmatrix}}0&&1&&0\\&1&&1\\1&&0&&1{\end{smallmatrix}}}\restore{\phantom{{\begin{smallmatrix}}1\\1\\1{\end{smallmatrix}}}\hspace{0.16cm}}\ar[ru ] & & { } \save[]*{{\begin{smallmatrix}}0&&1&&0\\&1&&1\\0&&1&&0{\end{smallmatrix}}}\restore{\phantom{{\begin{smallmatrix}}1\\1\\1{\end{smallmatrix}}}\hspace{0.16cm}}\ar[ru ] & & { } \save[]*{{\begin{smallmatrix}}1&&0&&1\\&1&&1\\0&&1&&0{\end{smallmatrix}}}\restore{\phantom{{\begin{smallmatrix}}1\\1\\1{\end{smallmatrix}}}\hspace{0.16cm}}\ar@{.}[uu ] } \ ] ] @xmath1404*{{\begin{smallmatrix}}&0&&0\\1&&2&&1\\&3&&3\\1&&2&&1\\&0&&0{\end{smallmatrix}}}\restore{\phantom{{\begin{smallmatrix}}1\\1\\1\\1\\1{\end{smallmatrix}}\hspace{0.20cm}}}\ar[rd ] & & { } \save[]*{{\begin{smallmatrix}}&0&&1\\1&&3&&2\\&4&&4\\2&&3&&1\\&1&&0{\end{smallmatrix}}}\restore{\phantom{{\begin{smallmatrix}}1\\1\\1\\1\\1{\end{smallmatrix}}\hspace{0.20cm}}}\ar[rd ] & & { } \save[]*{{\begin{smallmatrix}}&0&&0\\1&&2&&1\\&3&&3\\1&&2&&1\\&0&&0{\end{smallmatrix}}}\restore{\phantom{{\begin{smallmatrix}}1\\1\\1\\1\\1{\end{smallmatrix}}\hspace{0.20cm}}}\ar[rd ] & & { } \save[]*{{\begin{smallmatrix}}&0&&0\\1&&2&&1\\&3&&3\\1&&2&&1\\&0&&0{\end{smallmatrix}}}\restore{\phantom{{\begin{smallmatrix}}1\\1\\1\\1\\1{\end{smallmatrix}}\hspace{0.20cm}}}\ar[rd ] & & { } \save[]*{{\begin{smallmatrix}}&1&&0\\2&&3&&1\\&4&&4\\1&&3&&2\\&0&&1{\end{smallmatrix}}}\restore{\phantom{{\begin{smallmatrix}}1\\1\\1\\1\\1{\end{smallmatrix}}\hspace{0.20cm}}}\ar[rd ] & & { } \save[]*{{\begin{smallmatrix}}&0&&0\\1&&2&&1\\&3&&3\\1&&2&&1\\&0&&0{\end{smallmatrix}}}\restore{\phantom{{\begin{smallmatrix}}1\\1\\1\\1\\1{\end{smallmatrix}}\hspace{0.20cm}}}\ar[rd]&\\ { } \save[]*{{\begin{smallmatrix}}&0&&0\\1&&2&&1\\&3&&2\\1&&2&&1\\&0&&0{\end{smallmatrix}}}\restore{\phantom{{\begin{smallmatrix}}1\\1\\1\\1\\1{\end{smallmatrix}}\hspace{0.20cm}}}\ar[rd]\ar[ru]\ar@{.}[dd]\ar@{.}[uu ] & & { } \save[]*{{\begin{smallmatrix}}&0&&1\\1&&2&&1\\&3&&3\\1&&2&&1\\&0&&0{\end{smallmatrix}}}\restore{\phantom{{\begin{smallmatrix}}1\\1\\1\\1\\1{\end{smallmatrix}}\hspace{0.20cm}}}\ar[rd]\ar[ru ] & & { } \save[]*{{\begin{smallmatrix}}&0&&0\\1&&2&&1\\&3&&3\\1&&2&&1\\&1&&0{\end{smallmatrix}}}\restore{\phantom{{\begin{smallmatrix}}1\\1\\1\\1\\1{\end{smallmatrix}}\hspace{0.20cm}}}\ar[rd]\ar[ru ] & & { } \save[]*{{\begin{smallmatrix}}&0&&0\\1&&2&&1\\&2&&3\\1&&2&&1\\&0&&0{\end{smallmatrix}}}\restore{\phantom{{\begin{smallmatrix}}1\\1\\1\\1\\1{\end{smallmatrix}}\hspace{0.20cm}}}\ar[rd]\ar[ru ] & & { } \save[]*{{\begin{smallmatrix}}&1&&0\\1&&2&&1\\&3&&3\\1&&2&&1\\&0&&0{\end{smallmatrix}}}\restore{\phantom{{\begin{smallmatrix}}1\\1\\1\\1\\1{\end{smallmatrix}}\hspace{0.20cm}}}\ar[rd]\ar[ru ] & & { } \save[]*{{\begin{smallmatrix}}&0&&0\\1&&2&&1\\&3&&3\\1&&2&&1\\&0&&1{\end{smallmatrix}}}\restore{\phantom{{\begin{smallmatrix}}1\\1\\1\\1\\1{\end{smallmatrix}}\hspace{0.20cm}}}\ar[rd]\ar[ru ] & & { } \save[]*{{\begin{smallmatrix}}&0&&0\\1&&2&&1\\&3&&2\\1&&2&&1\\&0&&0{\end{smallmatrix}}}\restore{\phantom{{\begin{smallmatrix}}1\\1\\1\\1\\1{\end{smallmatrix}}\hspace{0.20cm}}}\ar@{.}[dd]\ar@{.}[uu]\\ & { } \save[]*{{\begin{smallmatrix}}&0&&1\\1&&2&&1\\&3&&2\\1&&2&&1\\&0&&0{\end{smallmatrix}}}\restore{\phantom{{\begin{smallmatrix}}1\\1\\1\\1\\1{\end{smallmatrix}}\hspace{0.20cm}}}\ar[rd]\ar[ru ] & & { } \save[]*{{\begin{smallmatrix}}&0&&0\\1&&1&&0\\&2&&2\\0&&1&&1\\&0&&0{\end{smallmatrix}}}\restore{\phantom{{\begin{smallmatrix}}1\\1\\1\\1\\1{\end{smallmatrix}}\hspace{0.20cm}}}\ar[rd]\ar[ru ] & & { } \save[]*{{\begin{smallmatrix}}&0&&0\\1&&2&&1\\&2&&3\\1&&2&&1\\&1&&0{\end{smallmatrix}}}\restore{\phantom{{\begin{smallmatrix}}1\\1\\1\\1\\1{\end{smallmatrix}}\hspace{0.20cm}}}\ar[rd]\ar[ru ] & & { } \save[]*{{\begin{smallmatrix}}&1&&0\\1&&2&&1\\&2&&3\\1&&2&&1\\&0&&0{\end{smallmatrix}}}\restore{\phantom{{\begin{smallmatrix}}1\\1\\1\\1\\1{\end{smallmatrix}}\hspace{0.20cm}}}\ar[rd]\ar[ru ] & & { } \save[]*{{\begin{smallmatrix}}&0&&0\\0&&1&&1\\&2&&2\\1&&1&&0\\&0&&0{\end{smallmatrix}}}\restore{\phantom{{\begin{smallmatrix}}1\\1\\1\\1\\1{\end{smallmatrix}}\hspace{0.20cm}}}\ar[rd]\ar[ru ] & & { } \save[]*{{\begin{smallmatrix}}&0&&0\\1&&2&&1\\&3&&2\\1&&2&&1\\&0&&1{\end{smallmatrix}}}\restore{\phantom{{\begin{smallmatrix}}1\\1\\1\\1\\1{\end{smallmatrix}}\hspace{0.20cm}}}\ar[rd]\ar[ru]&\\ { } \save[]*{{\begin{smallmatrix}}&0&&1\\1&&2&&1\\&3&&2\\1&&2&&1\\&0&&1{\end{smallmatrix}}}\restore{\phantom{{\begin{smallmatrix}}1\\1\\1\\1\\1{\end{smallmatrix}}\hspace{0.20cm}}}\ar[ru]\ar[rd ] & & { } \save[]*{{\begin{smallmatrix}}&0&&0\\1&&1&&0\\&2&&1\\0&&1&&1\\&0&&0{\end{smallmatrix}}}\restore{\phantom{{\begin{smallmatrix}}1\\1\\1\\1\\1{\end{smallmatrix}}\hspace{0.20cm}}}\ar[ru]\ar[rd ] & & { } \save[]*{{\begin{smallmatrix}}&0&&0\\1&&1&&0\\&1&&2\\0&&1&&1\\&0&&0{\end{smallmatrix}}}\restore{\phantom{{\begin{smallmatrix}}1\\1\\1\\1\\1{\end{smallmatrix}}\hspace{0.20cm}}}\ar[ru]\ar[rd ] & & { } \save[]*{{\begin{smallmatrix}}&1&&0\\1&&2&&1\\&2&&3\\1&&2&&1\\&1&&0{\end{smallmatrix}}}\restore{\phantom{{\begin{smallmatrix}}1\\1\\1\\1\\1{\end{smallmatrix}}\hspace{0.20cm}}}\ar[ru]\ar[rd ] & & { } \save[]*{{\begin{smallmatrix}}&0&&0\\0&&1&&1\\&1&&2\\1&&1&&0\\&0&&0{\end{smallmatrix}}}\restore{\phantom{{\begin{smallmatrix}}1\\1\\1\\1\\1{\end{smallmatrix}}\hspace{0.20cm}}}\ar[ru]\ar[rd ] & & { } \save[]*{{\begin{smallmatrix}}&0&&0\\0&&1&&1\\&2&&1\\1&&1&&0\\&0&&0{\end{smallmatrix}}}\restore{\phantom{{\begin{smallmatrix}}1\\1\\1\\1\\1{\end{smallmatrix}}\hspace{0.20cm}}}\ar[ru]\ar[rd ] & & { } \save[]*{{\begin{smallmatrix}}&0&&1\\1&&2&&1\\&3&&2\\1&&2&&1\\&0&&1{\end{smallmatrix}}}\restore{\phantom{{\begin{smallmatrix}}1\\1\\1\\1\\1{\end{smallmatrix}}\hspace{0.20cm}}}\\ & { } \save[]*{{\begin{smallmatrix}}&0&&0\\1&&1&&0\\&2&&1\\0&&1&&1\\&0&&1{\end{smallmatrix}}}\restore{\phantom{{\begin{smallmatrix}}1\\1\\1\\1\\1{\end{smallmatrix}}\hspace{0.20cm}}}\ar[rd]\ar[ru ] & & { } \save[]*{{\begin{smallmatrix}}&0&&0\\1&&1&&0\\&1&&1\\0&&1&&1\\&0&&0{\end{smallmatrix}}}\restore{\phantom{{\begin{smallmatrix}}1\\1\\1\\1\\1{\end{smallmatrix}}\hspace{0.20cm}}}\ar[rd]\ar[ru ] & & { } \save[]*{{\begin{smallmatrix}}&1&&0\\1&&1&&0\\&1&&2\\0&&1&&1\\&0&&0{\end{smallmatrix}}}\restore{\phantom{{\begin{smallmatrix}}1\\1\\1\\1\\1{\end{smallmatrix}}\hspace{0.20cm}}}\ar[rd]\ar[ru ] & & { } \save[]*{{\begin{smallmatrix}}&0&&0\\0&&1&&1\\&1&&2\\1&&1&&0\\&1&&0{\end{smallmatrix}}}\restore{\phantom{{\begin{smallmatrix}}1\\1\\1\\1\\1{\end{smallmatrix}}\hspace{0.20cm}}}\ar[rd]\ar[ru ] & & { } \save[]*{{\begin{smallmatrix}}&0&&0\\0&&1&&1\\&1&&1\\1&&1&&0\\&0&&0{\end{smallmatrix}}}\restore{\phantom{{\begin{smallmatrix}}1\\1\\1\\1\\1{\end{smallmatrix}}\hspace{0.20cm}}}\ar[rd]\ar[ru ] & & { } \save[]*{{\begin{smallmatrix}}&0&&1\\0&&1&&1\\&2&&1\\1&&1&&0\\&0&&0{\end{smallmatrix}}}\restore{\phantom{{\begin{smallmatrix}}1\\1\\1\\1\\1{\end{smallmatrix}}\hspace{0.20cm}}}\ar[rd]\ar[ru]&\\ { } \save[]*{{\begin{smallmatrix}}&0&&0\\0&&0&&0\\&1&&0\\0&&0&&0\\&0&&0{\end{smallmatrix}}}\restore{\phantom{{\begin{smallmatrix}}1\\1\\1\\1\\1{\end{smallmatrix}}\hspace{0.20cm}}}\ar[ru]\ar@{.}[uu ] & & { } \save[]*{{\begin{smallmatrix}}&0&&0\\1&&1&&0\\&1&&1\\0&&1&&1\\&0&&1{\end{smallmatrix}}}\restore{\phantom{{\begin{smallmatrix}}1\\1\\1\\1\\1{\end{smallmatrix}}\hspace{0.20cm}}}\ar[ru]\ar[rd ] & & { } \save[]*{{\begin{smallmatrix}}&1&&0\\1&&1&&0\\&1&&1\\0&&1&&1\\&0&&0{\end{smallmatrix}}}\restore{\phantom{{\begin{smallmatrix}}1\\1\\1\\1\\1{\end{smallmatrix}}\hspace{0.20cm}}}\ar[ru ] & & { } \save[]*{{\begin{smallmatrix}}&0&&0\\0&&0&&0\\&0&&1\\0&&0&&0\\&0&&0{\end{smallmatrix}}}\restore{\phantom{{\begin{smallmatrix}}1\\1\\1\\1\\1{\end{smallmatrix}}\hspace{0.20cm}}}\ar[ru ] & & { } \save[]*{{\begin{smallmatrix}}&0&&0\\0&&1&&1\\&1&&1\\1&&1&&0\\&1&&0{\end{smallmatrix}}}\restore{\phantom{{\begin{smallmatrix}}1\\1\\1\\1\\1{\end{smallmatrix}}\hspace{0.20cm}}}\ar[ru]\ar[rd ] & & { } \save[]*{{\begin{smallmatrix}}&0&&1\\0&&1&&1\\&1&&1\\1&&1&&0\\&0&&0{\end{smallmatrix}}}\restore{\phantom{{\begin{smallmatrix}}1\\1\\1\\1\\1{\end{smallmatrix}}\hspace{0.20cm}}}\ar[ru ] & & { } \save[]*{{\begin{smallmatrix}}&0&&0\\0&&0&&0\\&1&&0\\0&&0&&0\\&0&&0{\end{smallmatrix}}}\restore{\phantom{{\begin{smallmatrix}}1\\1\\1\\1\\1{\end{smallmatrix}}\hspace{0.20cm}}}\ar@{.}[uu]\\ & & & { } \save[]*{{\begin{smallmatrix}}&1&&0\\1&&1&&0\\&1&&1\\0&&1&&1\\&0&&1{\end{smallmatrix}}}\restore{\phantom{{\begin{smallmatrix}}1\\1\\1\\1\\1{\end{smallmatrix}}\hspace{0.20cm}}}\ar[ru ] & & & & & & { } \save[]*{{\begin{smallmatrix}}&0&&1\\0&&1&&1\\&1&&1\\1&&1&&0\\&1&&0{\end{smallmatrix}}}\restore{\phantom{{\begin{smallmatrix}}1\\1\\1\\1\\1{\end{smallmatrix}}\hspace{0.20cm}}}\ar[ru]\\ } \ ] ] @xmath1405*{{\begin{smallmatrix}}&&1\\&2&&2\\2&&3&&2\\&2&&2\\&&1{\end{smallmatrix}}}\restore{\phantom{{\begin{smallmatrix}}1\\1\\1{\end{smallmatrix}}}\hspace{0.16cm}}\ar[dr]\ar@{.}[uu ] & & { } \save[]*{{\begin{smallmatrix}}&&0\\&1&&1\\1&&2&&1\\&1&&1\\&&0{\end{smallmatrix}}}\restore{\phantom{{\begin{smallmatrix}}1\\1\\1{\end{smallmatrix}}}\hspace{0.16cm}}\ar[rd ] & & { } \save[]*{{\begin{smallmatrix}}&&0\\&1&&1\\1&&2&&1\\&1&&1\\&&0{\end{smallmatrix}}}\restore{\phantom{{\begin{smallmatrix}}1\\1\\1{\end{smallmatrix}}}\hspace{0.16cm}}\ar[rd ] & & { } \save[]*{{\begin{smallmatrix}}&&1\\&2&&2\\2&&3&&2\\&2&&2\\&&1{\end{smallmatrix}}}\restore{\phantom{{\begin{smallmatrix}}1\\1\\1{\end{smallmatrix}}}\hspace{0.16cm}}\ar@{.}[uu]&&&\\ & { } \save[]*{{\begin{smallmatrix}}&1&&1\\1&&2&&1\\&1&&1{\end{smallmatrix}}}\restore{\phantom{{\begin{smallmatrix}}1\\1\\1{\end{smallmatrix}}}\hspace{0.16cm}}\ar[rd ] & & { } \save[]*{{\begin{smallmatrix}}&1&&1\\1&&2&&1\\&1&&1{\end{smallmatrix}}}\restore{\phantom{{\begin{smallmatrix}}1\\1\\1{\end{smallmatrix}}}\hspace{0.16cm}}\ar[rd ] & & & & & & { } \save[]*{{\begin{smallmatrix}}&&0\\&1&&1\\1&&2&&1\\&1&&1\\&&1{\end{smallmatrix}}}\restore{\phantom{{\begin{smallmatrix}}1\\1\\1{\end{smallmatrix}}}\hspace{0.16cm}}\ar[rd]\ar[ur ] & & { } \save[]*{{\begin{smallmatrix}}&&0\\&1&&1\\1&&1&&1\\&1&&1\\&&0{\end{smallmatrix}}}\restore{\phantom{{\begin{smallmatrix}}1\\1\\1{\end{smallmatrix}}}\hspace{0.16cm}}\ar[rd]\ar[ur ] & & { } \save[]*{{\begin{smallmatrix}}&&1\\&1&&1\\1&&2&&1\\&1&&1\\&&0{\end{smallmatrix}}}\restore{\phantom{{\begin{smallmatrix}}1\\1\\1{\end{smallmatrix}}}\hspace{0.16cm}}\ar[rd]\ar[ur]&\\ { } \save[]*{{\begin{smallmatrix}}&1&&0\\1&&1&&0\\&1&&0{\end{smallmatrix}}}\restore{\phantom{{\begin{smallmatrix}}1\\1\\1{\end{smallmatrix}}}\hspace{0.16cm}}\ar[ru]\ar@{.}[uu ] & & { } \save[]*{{\begin{smallmatrix}}&0&&1\\0&&1&&1\\&0&&1{\end{smallmatrix}}}\restore{\phantom{{\begin{smallmatrix}}1\\1\\1{\end{smallmatrix}}}\hspace{0.16cm}}\ar[ru ] & & { } \save[]*{{\begin{smallmatrix}}&1&&0\\1&&1&&0\\&1&&0{\end{smallmatrix}}}\restore{\phantom{{\begin{smallmatrix}}1\\1\\1{\end{smallmatrix}}}\hspace{0.16cm}}\ar@{.}[uu ] & & & & { } \save[]*{{\begin{smallmatrix}}&&0\\&0&&0\\0&&1&&0\\&0&&0\\&&0{\end{smallmatrix}}}\restore{\phantom{{\begin{smallmatrix}}1\\1\\1{\end{smallmatrix}}}\hspace{0.16cm}}\ar[ru]\ar@{.}[uu ] & & { } \save[]*{{\begin{smallmatrix}}&&0\\&1&&1\\1&&1&&1\\&1&&1\\&&1{\end{smallmatrix}}}\restore{\phantom{{\begin{smallmatrix}}1\\1\\1{\end{smallmatrix}}}\hspace{0.16cm}}\ar[ru]\ar[rd ] & & { } \save[]*{{\begin{smallmatrix}}&&1\\&1&&1\\1&&1&&1\\&1&&1\\&&0{\end{smallmatrix}}}\restore{\phantom{{\begin{smallmatrix}}1\\1\\1{\end{smallmatrix}}}\hspace{0.16cm}}\ar[ru ] & & { } \save[]*{{\begin{smallmatrix}}&&0\\&0&&0\\0&&1&&0\\&0&&0\\&&0{\end{smallmatrix}}}\restore{\phantom{{\begin{smallmatrix}}1\\1\\1{\end{smallmatrix}}}\hspace{0.16cm}}\ar@{.}[uu]\\ & & & & & & & & & & & { } \save[]*{{\begin{smallmatrix}}&&1\\&1&&1\\1&&1&&1\\&1&&1\\&&1{\end{smallmatrix}}}\restore{\phantom{{\begin{smallmatrix}}1\\1\\1{\end{smallmatrix}}}\hspace{0.16cm}}\ar[ru ] } \ ] ] @xmath1404*{{\begin{smallmatrix}}0&&&&0\\&1&&1\\1&&2&&1\\&1&&1\\0&&&&0{\end{smallmatrix}}}\restore{\phantom{{\begin{smallmatrix}}1\\1\\1\\1\\1{\end{smallmatrix}}\hspace{0.20cm}}}\ar[rd ] & & { } \save[]*{{\begin{smallmatrix}}0&&&&1\\&1&&2\\1&&3&&1\\&2&&1\\1&&&&0{\end{smallmatrix}}}\restore{\phantom{{\begin{smallmatrix}}1\\1\\1\\1\\1{\end{smallmatrix}}\hspace{0.20cm}}}\ar[rd ] & & { } \save[]*{{\begin{smallmatrix}}0&&&&0\\&1&&1\\1&&2&&1\\&1&&1\\0&&&&0{\end{smallmatrix}}}\restore{\phantom{{\begin{smallmatrix}}1\\1\\1\\1\\1{\end{smallmatrix}}\hspace{0.20cm}}}\ar[rd ] & & { } \save[]*{{\begin{smallmatrix}}0&&&&0\\&1&&1\\1&&2&&1\\&1&&1\\0&&&&0{\end{smallmatrix}}}\restore{\phantom{{\begin{smallmatrix}}1\\1\\1\\1\\1{\end{smallmatrix}}\hspace{0.20cm}}}\ar[rd ] & & { } \save[]*{{\begin{smallmatrix}}1&&&&0\\&2&&1\\1&&3&&1\\&1&&2\\0&&&&1{\end{smallmatrix}}}\restore{\phantom{{\begin{smallmatrix}}1\\1\\1\\1\\1{\end{smallmatrix}}\hspace{0.20cm}}}\ar[rd ] & & { } \save[]*{{\begin{smallmatrix}}0&&&&0\\&1&&1\\1&&2&&1\\&1&&1\\0&&&&0{\end{smallmatrix}}}\restore{\phantom{{\begin{smallmatrix}}1\\1\\1\\1\\1{\end{smallmatrix}}\hspace{0.20cm}}}\ar[rd]&\\ { } \save[]*{{\begin{smallmatrix}}0&&&&0\\&1&&1\\1&&2&&0\\&1&&1\\0&&&&0{\end{smallmatrix}}}\restore{\phantom{{\begin{smallmatrix}}1\\1\\1\\1\\1{\end{smallmatrix}}\hspace{0.20cm}}}\ar[rd]\ar[ur]\ar@{.}[uu]\ar@{.}[dd ] & & { } \save[]*{{\begin{smallmatrix}}0&&&&1\\&1&&1\\1&&2&&1\\&1&&1\\0&&&&0{\end{smallmatrix}}}\restore{\phantom{{\begin{smallmatrix}}1\\1\\1\\1\\1{\end{smallmatrix}}\hspace{0.20cm}}}\ar[rd]\ar[ur ] & & { } \save[]*{{\begin{smallmatrix}}0&&&&0\\&1&&1\\1&&2&&1\\&1&&1\\1&&&&0{\end{smallmatrix}}}\restore{\phantom{{\begin{smallmatrix}}1\\1\\1\\1\\1{\end{smallmatrix}}\hspace{0.20cm}}}\ar[rd]\ar[ur ] & & { } \save[]*{{\begin{smallmatrix}}0&&&&0\\&1&&1\\0&&2&&1\\&1&&1\\0&&&&0{\end{smallmatrix}}}\restore{\phantom{{\begin{smallmatrix}}1\\1\\1\\1\\1{\end{smallmatrix}}\hspace{0.20cm}}}\ar[rd]\ar[ur ] & & { } \save[]*{{\begin{smallmatrix}}1&&&&0\\&1&&1\\1&&2&&1\\&1&&1\\0&&&&0{\end{smallmatrix}}}\restore{\phantom{{\begin{smallmatrix}}1\\1\\1\\1\\1{\end{smallmatrix}}\hspace{0.20cm}}}\ar[rd]\ar[ur ] & & { } \save[]*{{\begin{smallmatrix}}0&&&&0\\&1&&1\\1&&2&&1\\&1&&1\\0&&&&1{\end{smallmatrix}}}\restore{\phantom{{\begin{smallmatrix}}1\\1\\1\\1\\1{\end{smallmatrix}}\hspace{0.20cm}}}\ar[rd]\ar[ur ] & & { } \save[]*{{\begin{smallmatrix}}0&&&&0\\&1&&1\\1&&2&&0\\&1&&1\\0&&&&0{\end{smallmatrix}}}\restore{\phantom{{\begin{smallmatrix}}1\\1\\1\\1\\1{\end{smallmatrix}}\hspace{0.20cm}}}\ar@{.}[uu]\ar@{.}[dd]\\ & { } \save[]*{{\begin{smallmatrix}}0&&&&1\\&1&&1\\1&&2&&0\\&1&&1\\0&&&&0{\end{smallmatrix}}}\restore{\phantom{{\begin{smallmatrix}}1\\1\\1\\1\\1{\end{smallmatrix}}\hspace{0.20cm}}}\ar[rd]\ar[ru ] & & { } \save[]*{{\begin{smallmatrix}}0&&&&0\\&1&&0\\1&&1&&1\\&0&&1\\0&&&&0{\end{smallmatrix}}}\restore{\phantom{{\begin{smallmatrix}}1\\1\\1\\1\\1{\end{smallmatrix}}\hspace{0.20cm}}}\ar[rd]\ar[ru ] & & { } \save[]*{{\begin{smallmatrix}}0&&&&0\\&1&&1\\0&&2&&1\\&1&&1\\1&&&&0{\end{smallmatrix}}}\restore{\phantom{{\begin{smallmatrix}}1\\1\\1\\1\\1{\end{smallmatrix}}\hspace{0.20cm}}}\ar[rd]\ar[ru ] & & { } \save[]*{{\begin{smallmatrix}}1&&&&0\\&1&&1\\0&&2&&1\\&1&&1\\0&&&&0{\end{smallmatrix}}}\restore{\phantom{{\begin{smallmatrix}}1\\1\\1\\1\\1{\end{smallmatrix}}\hspace{0.20cm}}}\ar[rd]\ar[ru ] & & { } \save[]*{{\begin{smallmatrix}}0&&&&0\\&0&&1\\1&&1&&1\\&1&&0\\0&&&&0{\end{smallmatrix}}}\restore{\phantom{{\begin{smallmatrix}}1\\1\\1\\1\\1{\end{smallmatrix}}\hspace{0.20cm}}}\ar[rd]\ar[ru ] & & { } \save[]*{{\begin{smallmatrix}}0&&&&0\\&1&&1\\1&&2&&0\\&1&&1\\0&&&&1{\end{smallmatrix}}}\restore{\phantom{{\begin{smallmatrix}}1\\1\\1\\1\\1{\end{smallmatrix}}\hspace{0.20cm}}}\ar[rd]\ar[ru]&\\ { } \save[]*{{\begin{smallmatrix}}0&&&&1\\&1&&1\\1&&2&&0\\&1&&1\\0&&&&1{\end{smallmatrix}}}\restore{\phantom{{\begin{smallmatrix}}1\\1\\1\\1\\1{\end{smallmatrix}}\hspace{0.20cm}}}\ar[rd]\ar[ru ] & & { } \save[]*{{\begin{smallmatrix}}0&&&&0\\&1&&0\\1&&1&&0\\&0&&1\\0&&&&0{\end{smallmatrix}}}\restore{\phantom{{\begin{smallmatrix}}1\\1\\1\\1\\1{\end{smallmatrix}}\hspace{0.20cm}}}\ar[rd]\ar[ru ] & & { } \save[]*{{\begin{smallmatrix}}0&&&&0\\&1&&0\\0&&1&&1\\&0&&1\\0&&&&0{\end{smallmatrix}}}\restore{\phantom{{\begin{smallmatrix}}1\\1\\1\\1\\1{\end{smallmatrix}}\hspace{0.20cm}}}\ar[rd]\ar[ru ] & & { } \save[]*{{\begin{smallmatrix}}1&&&&0\\&1&&1\\0&&2&&1\\&1&&1\\1&&&&0{\end{smallmatrix}}}\restore{\phantom{{\begin{smallmatrix}}1\\1\\1\\1\\1{\end{smallmatrix}}\hspace{0.20cm}}}\ar[rd]\ar[ru ] & & { } \save[]*{{\begin{smallmatrix}}0&&&&0\\&0&&1\\0&&1&&1\\&1&&0\\0&&&&0{\end{smallmatrix}}}\restore{\phantom{{\begin{smallmatrix}}1\\1\\1\\1\\1{\end{smallmatrix}}\hspace{0.20cm}}}\ar[rd]\ar[ru ] & & { } \save[]*{{\begin{smallmatrix}}0&&&&0\\&0&&1\\1&&1&&0\\&1&&0\\0&&&&0{\end{smallmatrix}}}\restore{\phantom{{\begin{smallmatrix}}1\\1\\1\\1\\1{\end{smallmatrix}}\hspace{0.20cm}}}\ar[rd]\ar[ru ] & & { } \save[]*{{\begin{smallmatrix}}0&&&&1\\&1&&1\\1&&2&&0\\&1&&1\\0&&&&1{\end{smallmatrix}}}\restore{\phantom{{\begin{smallmatrix}}1\\1\\1\\1\\1{\end{smallmatrix}}\hspace{0.20cm}}}\\ & { } \save[]*{{\begin{smallmatrix}}0&&&&0\\&1&&0\\1&&1&&0\\&0&&1\\0&&&&1{\end{smallmatrix}}}\restore{\phantom{{\begin{smallmatrix}}1\\1\\1\\1\\1{\end{smallmatrix}}\hspace{0.20cm}}}\ar[rd]\ar[ru ] & & { } \save[]*{{\begin{smallmatrix}}0&&&&0\\&1&&0\\0&&1&&0\\&0&&1\\0&&&&0{\end{smallmatrix}}}\restore{\phantom{{\begin{smallmatrix}}1\\1\\1\\1\\1{\end{smallmatrix}}\hspace{0.20cm}}}\ar[rd]\ar[ru ] & & { } \save[]*{{\begin{smallmatrix}}1&&&&0\\&1&&0\\0&&1&&1\\&0&&1\\0&&&&0{\end{smallmatrix}}}\restore{\phantom{{\begin{smallmatrix}}1\\1\\1\\1\\1{\end{smallmatrix}}\hspace{0.20cm}}}\ar[rd]\ar[ru ] & & { } \save[]*{{\begin{smallmatrix}}0&&&&0\\&0&&1\\0&&1&&1\\&1&&0\\1&&&&0{\end{smallmatrix}}}\restore{\phantom{{\begin{smallmatrix}}1\\1\\1\\1\\1{\end{smallmatrix}}\hspace{0.20cm}}}\ar[rd]\ar[ru ] & & { } \save[]*{{\begin{smallmatrix}}0&&&&0\\&0&&1\\0&&1&&0\\&1&&0\\0&&&&0{\end{smallmatrix}}}\restore{\phantom{{\begin{smallmatrix}}1\\1\\1\\1\\1{\end{smallmatrix}}\hspace{0.20cm}}}\ar[rd]\ar[ru ] & & { } \save[]*{{\begin{smallmatrix}}0&&&&1\\&0&&1\\1&&1&&0\\&1&&0\\0&&&&0{\end{smallmatrix}}}\restore{\phantom{{\begin{smallmatrix}}1\\1\\1\\1\\1{\end{smallmatrix}}\hspace{0.20cm}}}\ar[rd]\ar[ru]&\\ { } \save[]*{{\begin{smallmatrix}}0&&&&0\\&0&&0\\1&&0&&0\\&0&&0\\0&&&&0{\end{smallmatrix}}}\restore{\phantom{{\begin{smallmatrix}}1\\1\\1\\1\\1{\end{smallmatrix}}\hspace{0.20cm}}}\ar[ru]\ar@{.}[uu ] & & { } \save[]*{{\begin{smallmatrix}}0&&&&0\\&1&&0\\0&&1&&0\\&0&&1\\0&&&&1{\end{smallmatrix}}}\restore{\phantom{{\begin{smallmatrix}}1\\1\\1\\1\\1{\end{smallmatrix}}\hspace{0.20cm}}}\ar[ru]\ar[rd ] & & { } \save[]*{{\begin{smallmatrix}}1&&&&0\\&1&&0\\0&&1&&0\\&0&&1\\0&&&&0{\end{smallmatrix}}}\restore{\phantom{{\begin{smallmatrix}}1\\1\\1\\1\\1{\end{smallmatrix}}\hspace{0.20cm}}}\ar[ru ] & & { } \save[]*{{\begin{smallmatrix}}0&&&&0\\&0&&0\\0&&0&&1\\&0&&0\\0&&&&0{\end{smallmatrix}}}\restore{\phantom{{\begin{smallmatrix}}1\\1\\1\\1\\1{\end{smallmatrix}}\hspace{0.20cm}}}\ar[ru ] & & { } \save[]*{{\begin{smallmatrix}}0&&&&0\\&0&&1\\0&&1&&0\\&1&&0\\1&&&&0{\end{smallmatrix}}}\restore{\phantom{{\begin{smallmatrix}}1\\1\\1\\1\\1{\end{smallmatrix}}\hspace{0.20cm}}}\ar[ru]\ar[rd ] & & { } \save[]*{{\begin{smallmatrix}}0&&&&1\\&0&&1\\0&&1&&0\\&1&&0\\0&&&&0{\end{smallmatrix}}}\restore{\phantom{{\begin{smallmatrix}}1\\1\\1\\1\\1{\end{smallmatrix}}\hspace{0.20cm}}}\ar[ru ] & & { } \save[]*{{\begin{smallmatrix}}0&&&&0\\&0&&0\\1&&0&&0\\&0&&0\\0&&&&0{\end{smallmatrix}}}\restore{\phantom{{\begin{smallmatrix}}1\\1\\1\\1\\1{\end{smallmatrix}}\hspace{0.20cm}}}\ar@{.}[uu]\\ & & & { } \save[]*{{\begin{smallmatrix}}1&&&&0\\&1&&0\\0&&1&&0\\&0&&1\\0&&&&1{\end{smallmatrix}}}\restore{\phantom{{\begin{smallmatrix}}1\\1\\1\\1\\1{\end{smallmatrix}}\hspace{0.20cm}}}\ar[ru ] & & & & & & { } \save[]*{{\begin{smallmatrix}}0&&&&1\\&0&&1\\0&&1&&0\\&1&&0\\1&&&&0{\end{smallmatrix}}}\restore{\phantom{{\begin{smallmatrix}}1\\1\\1\\1\\1{\end{smallmatrix}}\hspace{0.20cm}}}\ar[ru]\\ } \ ] ] @xmath1407 , \left [ \begin { smallmatrix } & 0&&0 & \\\noalign{\medskip}0&&1&&1\\\noalign{\medskip}&1&&2 & \\\noalign{\medskip}0&&1&&1\end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 0&&0&\\\noalign{\medskip}1&&1&&0 \\\noalign{\medskip}&2&&1&\\\noalign{\medskip}1&&1&&0 \end { smallmatrix } \right ] ] \ ] ] @xmath1408 , \left [ \begin { smallmatrix } & 1&&0 & \\\noalign{\medskip}1&&2&&1\\\noalign{\medskip}&2&&2 & \\\noalign{\medskip}0&&1&&1\end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 0&&1&\\\noalign{\medskip}1&&2&&1 \\\noalign{\medskip}&2&&2&\\\noalign{\medskip}1&&1&&0 \end { smallmatrix } \right ] ] \ ] ] @xmath1409 , \left [ \begin { smallmatrix } & 0&&1 & \\\noalign{\medskip}0&&1&&1\\\noalign{\medskip}&0&&1 & \\\noalign{\medskip}0&&0&&0\end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 1&&0&\\\noalign{\medskip}1&&1&&0 \\\noalign{\medskip}&1&&0&\\\noalign{\medskip}0&&0&&0 \end { smallmatrix } \right ] ] \ ] ] @xmath1407 , \left [ \begin { smallmatrix } & 0&&0 & \\\noalign{\medskip}0&&1&&0\\\noalign{\medskip}&1&&1 & \\\noalign{\medskip}1&&0&&1\end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 0&&0&\\\noalign{\medskip}0&&1&&0 \\\noalign{\medskip}&1&&1&\\\noalign{\medskip}0&&1&&0 \end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 0&&0 & \\\noalign{\medskip}1&&0&&1\\\noalign{\medskip}&1&&1 & \\\noalign{\medskip}0&&1&&0\end { smallmatrix } \right ] ] \ ] ] @xmath1410 , \left [ \begin { smallmatrix } & 0&&0 & \\\noalign{\medskip}1&&1&&1\\\noalign{\medskip}&2&&2 & \\\noalign{\medskip}0&&2&&0\end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 0&&0&\\\noalign{\medskip}1&&1&&1 \\\noalign{\medskip}&2&&2&\\\noalign{\medskip}1&&1&&1 \end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 0&&0 & \\\noalign{\medskip}0&&2&&0\\\noalign{\medskip}&2&&2 & \\\noalign{\medskip}1&&1&&1\end { smallmatrix } \right ] ] \ ] ] @xmath1411 , \left [ \begin { smallmatrix } & 0&&0 & \\\noalign{\medskip}1&&2&&1\\\noalign{\medskip}&2&&2 & \\\noalign{\medskip}1&&1&&1\end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 1&&1&\\\noalign{\medskip}1&&3&&1 \\\noalign{\medskip}&3&&3&\\\noalign{\medskip}1&&2&&1 \end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 0&&0 & \\\noalign{\medskip}1&&1&&1\\\noalign{\medskip}&2&&2 & \\\noalign{\medskip}0&&1&&0\end { smallmatrix } \right ] ] \ ] ] @xmath1408 , \left [ \begin { smallmatrix } & 0&&0 & \\\noalign{\medskip}0&&1&&0\\\noalign{\medskip}&1&&1 & \\\noalign{\medskip}0&&0&&0\end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 0&&0&\\\noalign{\medskip}1&&1&&1 \\\noalign{\medskip}&1&&1&\\\noalign{\medskip}0&&1&&0 \end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 1&&1 & \\\noalign{\medskip}1&&2&&1\\\noalign{\medskip}&2&&2 & \\\noalign{\medskip}1&&1&&1\end { smallmatrix } \right ] ] \ ] ] @xmath1412 , \left [ \begin { smallmatrix } & 1&&1 & \\\noalign{\medskip}2&&3&&2\\\noalign{\medskip}&3&&3 & \\\noalign{\medskip}1&&2&&1\end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 1&&1&\\\noalign{\medskip}1&&3&&1 \\\noalign{\medskip}&3&&3&\\\noalign{\medskip}1&&1&&1 \end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 0&&0 & \\\noalign{\medskip}1&&2&&1\\\noalign{\medskip}&2&&2 & \\\noalign{\medskip}0&&1&&0\end { smallmatrix } \right ] ] \ ] ] @xmath1413 , \left [ \begin { smallmatrix } & 1&&1 & \\\noalign{\medskip}1&&2&&1\\\noalign{\medskip}&2&&2 & \\\noalign{\medskip}0&&1&&0\end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 0&&0&\\\noalign{\medskip}1&&1&&1 \\\noalign{\medskip}&1&&1&\\\noalign{\medskip}0&&0&&0 \end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 1&&1 & \\\noalign{\medskip}1&&3&&1\\\noalign{\medskip}&2&&2 & \\\noalign{\medskip}1&&1&&1\end { smallmatrix } \right ] ] \ ] ] @xmath1409 , \left [ \begin { smallmatrix } & 0&&0 & \\\noalign{\medskip}0&&1&&0\\\noalign{\medskip}&0&&0 & \\\noalign{\medskip}0&&0&&0\end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 1&&1&\\\noalign{\medskip}1&&1&&1 \\\noalign{\medskip}&1&&1&\\\noalign{\medskip}0&&1&&0 \end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 0&&0 & \\\noalign{\medskip}0&&0&&0\\\noalign{\medskip}&0&&0 & \\\noalign{\medskip}0&&-1&&0\end { smallmatrix } \right ] ] \ ] ] @xmath1414 , \left [ \begin { smallmatrix } & 1&&1 & \\\noalign{\medskip}1&&1&&1\\\noalign{\medskip}&1&&1 & \\\noalign{\medskip}0&&0&&0\end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 0&&0&\\\noalign{\medskip}0&&1&&0 \\\noalign{\medskip}&0&&0&\\\noalign{\medskip}0&&-1&&0 \end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 1&&1 & \\\noalign{\medskip}1&&2&&1\\\noalign{\medskip}&1&&1 & \\\noalign{\medskip}0&&1&&0\end { smallmatrix } \right ] ] \ ] ] @xmath1407 , \left [ \begin { smallmatrix } & -1&&0 & \\\noalign{\medskip}0&&0&&0\\\noalign{\medskip}&0&&0 & \\\noalign{\medskip}0&&0&&0\end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 1&&0&\\\noalign{\medskip}1&&1&&0 \\\noalign{\medskip}&1&&1&\\\noalign{\medskip}0&&1&&1 \end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 0&&0 & \\\noalign{\medskip}0&&0&&0\\\noalign{\medskip}&0&&1 & \\\noalign{\medskip}0&&0&&0\end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 0&&-1&\\\noalign{\medskip}0&&0&&0 \\\noalign{\medskip}&0&&0&\\\noalign{\medskip}0&&0&&0 \end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 0&&1 & \\\noalign{\medskip}0&&1&&1\\\noalign{\medskip}&1&&1 & \\\noalign{\medskip}1&&1&&0\end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 0&&0&\\\noalign{\medskip}0&&0&&0 \\\noalign{\medskip}&1&&0&\\\noalign{\medskip}0&&0&&0 \end { smallmatrix } \right ] ] \ ] ] @xmath1410 , \left [ \begin { smallmatrix } & -1&&0 & \\\noalign{\medskip}0&&0&&0\\\noalign{\medskip}&1&&0 & \\\noalign{\medskip}0&&0&&0\end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 0&&0&\\\noalign{\medskip}1&&1&&0 \\\noalign{\medskip}&1&&1&\\\noalign{\medskip}0&&1&&1 \end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 1&&0 & \\\noalign{\medskip}1&&1&&0\\\noalign{\medskip}&1&&2 & \\\noalign{\medskip}0&&1&&1\end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 0&&-1&\\\noalign{\medskip}0&&0&&0 \\\noalign{\medskip}&0&&1&\\\noalign{\medskip}0&&0&&0 \end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 0&&0 & \\\noalign{\medskip}0&&1&&1\\\noalign{\medskip}&1&&1 & \\\noalign{\medskip}1&&1&&0\end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 0&&1&\\\noalign{\medskip}0&&1&&1 \\\noalign{\medskip}&2&&1&\\\noalign{\medskip}1&&1&&0 \end { smallmatrix } \right ] ] \ ] ] @xmath1415 , \left [ \begin { smallmatrix } & -1&&1 & \\\noalign{\medskip}0&&1&&1\\\noalign{\medskip}&2&&1 & \\\noalign{\medskip}1&&1&&0\end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 0&&0&\\\noalign{\medskip}1&&1&&0 \\\noalign{\medskip}&2&&1&\\\noalign{\medskip}0&&1&&1 \end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 0&&0 & \\\noalign{\medskip}1&&1&&0\\\noalign{\medskip}&1&&2 & \\\noalign{\medskip}0&&1&&1\end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 1&&-1&\\\noalign{\medskip}1&&1&&0 \\\noalign{\medskip}&1&&2&\\\noalign{\medskip}0&&1&&1 \end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 0&&0 & \\\noalign{\medskip}0&&1&&1\\\noalign{\medskip}&1&&2 & \\\noalign{\medskip}1&&1&&0\end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 0&&0&\\\noalign{\medskip}0&&1&&1 \\\noalign{\medskip}&2&&1&\\\noalign{\medskip}1&&1&&0 \end { smallmatrix } \right ] ] \ ] ] @xmath1416 , \left [ \begin { smallmatrix } & 1&&0 & \\\noalign{\medskip}1&&2&&1\\\noalign{\medskip}&2&&3 & \\\noalign{\medskip}1&&2&&1\end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 0&&0&\\\noalign{\medskip}0&&1&&1 \\\noalign{\medskip}&2&&2&\\\noalign{\medskip}1&&1&&0 \end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & -1&&0 & \\\noalign{\medskip}0&&1&&1\\\noalign{\medskip}&2&&1 & \\\noalign{\medskip}1&&1&&0\end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 0&&1&\\\noalign{\medskip}1&&2&&1 \\\noalign{\medskip}&3&&2&\\\noalign{\medskip}1&&2&&1 \end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 0&&0 & \\\noalign{\medskip}1&&1&&0\\\noalign{\medskip}&2&&2 & \\\noalign{\medskip}0&&1&&1\end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 0&&-1&\\\noalign{\medskip}1&&1&&0 \\\noalign{\medskip}&1&&2&\\\noalign{\medskip}0&&1&&1 \end { smallmatrix } \right ] ] \ ] ] @xmath1417 , \left [ \begin { smallmatrix } & 1&&0 & \\\noalign{\medskip}1&&2&&1\\\noalign{\medskip}&3&&3 & \\\noalign{\medskip}1&&2&&1\end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & -1&&0&\\\noalign{\medskip}0&&1 & & 1\\\noalign{\medskip}&2&&2&\\\noalign{\medskip}1&&1&&0 \end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 0&&0 & \\\noalign{\medskip}1&&2&&1\\\noalign{\medskip}&3&&2 & \\\noalign{\medskip}1&&2&&1\end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 0&&1&\\\noalign{\medskip}1&&2&&1 \\\noalign{\medskip}&3&&3&\\\noalign{\medskip}1&&2&&1 \end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 0&&-1 & \\\noalign{\medskip}1&&1&&0\\\noalign{\medskip}&2&&2 & \\\noalign{\medskip}0&&1&&1\end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 0&&0&\\\noalign{\medskip}1&&2&&1 \\\noalign{\medskip}&2&&3&\\\noalign{\medskip}1&&2&&1 \end { smallmatrix } \right ] ] \ ] ] @xmath1418 , \left [ \begin { smallmatrix } & 0&&0 & \\\noalign{\medskip}1&&2&&1\\\noalign{\medskip}&2&&3 & \\\noalign{\medskip}0&&2&&1\end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 1&&0&\\\noalign{\medskip}2&&2&&1 \\\noalign{\medskip}&3&&3&\\\noalign{\medskip}1&&2&&1 \end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 0&&0 & \\\noalign{\medskip}0&&2&&1\\\noalign{\medskip}&2&&3 & \\\noalign{\medskip}1&&1&&1\end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 0&&0&\\\noalign{\medskip}1&&2&&1 \\\noalign{\medskip}&3&&2&\\\noalign{\medskip}1&&2&&0 \end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 0&&1 & \\\noalign{\medskip}1&&2&&2\\\noalign{\medskip}&3&&3 & \\\noalign{\medskip}1&&2&&1\end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 0&&0&\\\noalign{\medskip}1&&2&&0 \\\noalign{\medskip}&3&&2&\\\noalign{\medskip}1&&1&&1 \end { smallmatrix } \right ] ] \ ] ] @xmath1419 , \left [ \begin { smallmatrix } & 0&&0 & \\\noalign{\medskip}1&&1&&1\\\noalign{\medskip}&2&&2 & \\\noalign{\medskip}0&&1&&1\end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 0&&0&\\\noalign{\medskip}1&&2&&0 \\\noalign{\medskip}&2&&2&\\\noalign{\medskip}1&&1&&1 \end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 1&&0 & \\\noalign{\medskip}1&&2&&1\\\noalign{\medskip}&2&&3 & \\\noalign{\medskip}0&&2&&1\end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 0&&0&\\\noalign{\medskip}1&&1&&1 \\\noalign{\medskip}&2&&2&\\\noalign{\medskip}1&&1&&0 \end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 0&&0 & \\\noalign{\medskip}0&&2&&1\\\noalign{\medskip}&2&&2 & \\\noalign{\medskip}1&&1&&1\end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 0&&1&\\\noalign{\medskip}1&&2&&1 \\\noalign{\medskip}&3&&2&\\\noalign{\medskip}1&&2&&0 \end { smallmatrix } \right ] ] \ ] ] @xmath1420 , \left [ \begin { smallmatrix } & 0&&0 & \\\noalign{\medskip}1&&1&&0\\\noalign{\medskip}&2&&1 & \\\noalign{\medskip}0&&1&&0\end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 0&&0&\\\noalign{\medskip}1&&1&&1 \\\noalign{\medskip}&1&&2&\\\noalign{\medskip}0&&1&&1 \end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 1&&0 & \\\noalign{\medskip}1&&2&&0\\\noalign{\medskip}&2&&2 & \\\noalign{\medskip}1&&1&&1\end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 0&&0&\\\noalign{\medskip}0&&1&&1 \\\noalign{\medskip}&1&&2&\\\noalign{\medskip}0&&1&&0 \end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 0&&0 & \\\noalign{\medskip}1&&1&&1\\\noalign{\medskip}&2&&1 & \\\noalign{\medskip}1&&1&&0\end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 0&&1&\\\noalign{\medskip}0&&2&&1 \\\noalign{\medskip}&2&&2&\\\noalign{\medskip}1&&1&&1 \end { smallmatrix } \right ] ] \ ] ] @xmath1421 , \left [ \begin { smallmatrix } & 0&&0 & \\\noalign{\medskip}1&&2&&1\\\noalign{\medskip}&2&&3 & \\\noalign{\medskip}1&&1&&1\end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 1&&0&\\\noalign{\medskip}1&&3&&1 \\\noalign{\medskip}&3&&3&\\\noalign{\medskip}1&&2&&1 \end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 0&&1 & \\\noalign{\medskip}1&&2&&2\\\noalign{\medskip}&3&&3 & \\\noalign{\medskip}1&&2&&0\end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 0&&0&\\\noalign{\medskip}1&&2&&1 \\\noalign{\medskip}&3&&2&\\\noalign{\medskip}1&&1&&1 \end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 0&&1 & \\\noalign{\medskip}1&&3&&1\\\noalign{\medskip}&3&&3 & \\\noalign{\medskip}1&&2&&1\end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 1&&0&\\\noalign{\medskip}2&&2&&1 \\\noalign{\medskip}&3&&3&\\\noalign{\medskip}0&&2&&1 \end { smallmatrix } \right ] ] \ ] ] @xmath1411 , \left [ \begin { smallmatrix } & 0&&0 & \\\noalign{\medskip}0&&1&&0\\\noalign{\medskip}&1&&1 & \\\noalign{\medskip}0&&0&&1\end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 0&&0&\\\noalign{\medskip}1&&1&&0 \\\noalign{\medskip}&1&&1&\\\noalign{\medskip}0&&1&&0 \end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 1&&0 & \\\noalign{\medskip}1&&1&&1\\\noalign{\medskip}&1&&2 & \\\noalign{\medskip}0&&1&&1\end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 0&&0&\\\noalign{\medskip}0&&1&&0 \\\noalign{\medskip}&1&&1&\\\noalign{\medskip}1&&0&&0 \end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 0&&0 & \\\noalign{\medskip}0&&1&&1\\\noalign{\medskip}&1&&1 & \\\noalign{\medskip}0&&1&&0\end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 0&&1&\\\noalign{\medskip}1&&1&&1 \\\noalign{\medskip}&2&&1&\\\noalign{\medskip}1&&1&&0 \end { smallmatrix } \right ] ] \ ] ] @xmath1422 , \left [ \begin { smallmatrix } & 0&&0 & \\\noalign{\medskip}0&&2&&1\\\noalign{\medskip}&2&&2 & \\\noalign{\medskip}1&&1&&0\end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 0&&1&\\\noalign{\medskip}1&&2&&2 \\\noalign{\medskip}&3&&2&\\\noalign{\medskip}1&&2&&0 \end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 0&&1 & \\\noalign{\medskip}1&&2&&1\\\noalign{\medskip}&3&&2 & \\\noalign{\medskip}1&&1&&1\end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 0&&0&\\\noalign{\medskip}1&&2&&0 \\\noalign{\medskip}&2&&2&\\\noalign{\medskip}0&&1&&1 \end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 1&&0 & \\\noalign{\medskip}2&&2&&1\\\noalign{\medskip}&2&&3 & \\\noalign{\medskip}0&&2&&1\end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 1&&0&\\\noalign{\medskip}1&&2&&1 \\\noalign{\medskip}&2&&3&\\\noalign{\medskip}1&&1&&1 \end { smallmatrix } \right ] ] \ ] ] @xmath1423 , \left [ \begin { smallmatrix } & 1&&0 & \\\noalign{\medskip}2&&3&&1\\\noalign{\medskip}&3&&3 & \\\noalign{\medskip}1&&2&&1\end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 1&&1&\\\noalign{\medskip}1&&3&&2 \\\noalign{\medskip}&3&&4&\\\noalign{\medskip}1&&2&&1 \end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 0&&0 & \\\noalign{\medskip}1&&2&&1\\\noalign{\medskip}&3&&2 & \\\noalign{\medskip}1&&1&&0\end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 0&&1&\\\noalign{\medskip}1&&3&&2 \\\noalign{\medskip}&3&&3&\\\noalign{\medskip}1&&2&&1 \end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 1&&1 & \\\noalign{\medskip}2&&3&&1\\\noalign{\medskip}&4&&3 & \\\noalign{\medskip}1&&2&&1\end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 0&&0&\\\noalign{\medskip}1&&2&&1 \\\noalign{\medskip}&2&&3&\\\noalign{\medskip}0&&1&&1 \end { smallmatrix } \right ] ] \ ] ] @xmath1424 , \left [ \begin { smallmatrix } & 0&&0 & \\\noalign{\medskip}1&&1&&1\\\noalign{\medskip}&2&&1 & \\\noalign{\medskip}0&&1&&0\end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 0&&1&\\\noalign{\medskip}1&&2&&1 \\\noalign{\medskip}&2&&2&\\\noalign{\medskip}1&&1&&1 \end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 1&&0 & \\\noalign{\medskip}1&&2&&0\\\noalign{\medskip}&2&&2 & \\\noalign{\medskip}0&&1&&1\end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 0&&0&\\\noalign{\medskip}1&&1&&1 \\\noalign{\medskip}&1&&2&\\\noalign{\medskip}0&&1&&0 \end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 1&&0 & \\\noalign{\medskip}1&&2&&1\\\noalign{\medskip}&2&&2 & \\\noalign{\medskip}1&&1&&1\end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 0&&1&\\\noalign{\medskip}0&&2&&1 \\\noalign{\medskip}&2&&2&\\\noalign{\medskip}1&&1&&0 \end { smallmatrix } \right ] ] \ ] ] @xmath1425 , \left [ \begin { smallmatrix } & 0&&0 & \\\noalign{\medskip}1&&2&&1\\\noalign{\medskip}&2&&2 & \\\noalign{\medskip}1&&1&&0\end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 1&&1&\\\noalign{\medskip}1&&3&&2 \\\noalign{\medskip}&3&&3&\\\noalign{\medskip}1&&2&&1 \end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 0&&1 & \\\noalign{\medskip}1&&2&&1\\\noalign{\medskip}&3&&2 & \\\noalign{\medskip}1&&1&&0\end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 0&&0&\\\noalign{\medskip}1&&2&&1 \\\noalign{\medskip}&2&&2&\\\noalign{\medskip}0&&1&&1 \end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 1&&1 & \\\noalign{\medskip}2&&3&&1\\\noalign{\medskip}&3&&3 & \\\noalign{\medskip}1&&2&&1\end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 1&&0&\\\noalign{\medskip}1&&2&&1 \\\noalign{\medskip}&2&&3&\\\noalign{\medskip}0&&1&&1 \end { smallmatrix } \right ] ] \ ] ] @xmath1426 , \left [ \begin { smallmatrix } & 0&&1 & \\\noalign{\medskip}1&&3&&2\\\noalign{\medskip}&3&&3 & \\\noalign{\medskip}1&&2&&0\end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 1&&1&\\\noalign{\medskip}2&&3&&2 \\\noalign{\medskip}&4&&3&\\\noalign{\medskip}1&&2&&1 \end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 0&&1 & \\\noalign{\medskip}1&&3&&1\\\noalign{\medskip}&3&&3 & \\\noalign{\medskip}1&&1&&1\end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 1&&0&\\\noalign{\medskip}2&&3&&1 \\\noalign{\medskip}&3&&3&\\\noalign{\medskip}0&&2&&1 \end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 1&&1 & \\\noalign{\medskip}2&&3&&2\\\noalign{\medskip}&3&&4 & \\\noalign{\medskip}1&&2&&1\end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 1&&0&\\\noalign{\medskip}1&&3&&1 \\\noalign{\medskip}&3&&3&\\\noalign{\medskip}1&&1&&1 \end { smallmatrix } \right ] ] \ ] ] @xmath1408 , \left [ \begin { smallmatrix } & 0&&0 & \\\noalign{\medskip}1&&0&&0\\\noalign{\medskip}&1&&0 & \\\noalign{\medskip}0&&0&&0\end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 0&&0&\\\noalign{\medskip}0&&1&&0 \\\noalign{\medskip}&0&&1&\\\noalign{\medskip}0&&0&&1 \end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 1&&0 & \\\noalign{\medskip}1&&1&&0\\\noalign{\medskip}&1&&1 & \\\noalign{\medskip}0&&1&&0\end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 0&&0&\\\noalign{\medskip}0&&0&&1 \\\noalign{\medskip}&0&&1&\\\noalign{\medskip}0&&0&&0 \end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 0&&0 & \\\noalign{\medskip}0&&1&&0\\\noalign{\medskip}&1&&0 & \\\noalign{\medskip}1&&0&&0\end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 0&&1&\\\noalign{\medskip}0&&1&&1 \\\noalign{\medskip}&1&&1&\\\noalign{\medskip}0&&1&&0 \end { smallmatrix } \right ] ] \ ] ] @xmath1412 , \left [ \begin { smallmatrix } & 0&&1 & \\\noalign{\medskip}1&&1&&1\\\noalign{\medskip}&2&&1 & \\\noalign{\medskip}0&&1&&0\end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 0&&0&\\\noalign{\medskip}1&&1&&0 \\\noalign{\medskip}&1&&1&\\\noalign{\medskip}0&&0&&1 \end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 1&&0 & \\\noalign{\medskip}1&&2&&0\\\noalign{\medskip}&1&&2 & \\\noalign{\medskip}0&&1&&1\end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 1&&0&\\\noalign{\medskip}1&&1&&1 \\\noalign{\medskip}&1&&2&\\\noalign{\medskip}0&&1&&0 \end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 0&&0 & \\\noalign{\medskip}0&&1&&1\\\noalign{\medskip}&1&&1 & \\\noalign{\medskip}1&&0&&0\end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 0&&1&\\\noalign{\medskip}0&&2&&1 \\\noalign{\medskip}&2&&1&\\\noalign{\medskip}1&&1&&0 \end { smallmatrix } \right ] ] \ ] ] @xmath1427 , \left [ \begin { smallmatrix } & 0&&1 & \\\noalign{\medskip}1&&2&&1\\\noalign{\medskip}&3&&1 & \\\noalign{\medskip}1&&1&&0\end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 0&&1&\\\noalign{\medskip}1&&2&&1 \\\noalign{\medskip}&2&&2&\\\noalign{\medskip}0&&1&&1 \end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 1&&0 & \\\noalign{\medskip}2&&2&&0\\\noalign{\medskip}&2&&2 & \\\noalign{\medskip}0&&1&&1\end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 1&&0&\\\noalign{\medskip}1&&2&&1 \\\noalign{\medskip}&1&&3&\\\noalign{\medskip}0&&1&&1 \end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 1&&0 & \\\noalign{\medskip}1&&2&&1\\\noalign{\medskip}&2&&2 & \\\noalign{\medskip}1&&1&&0\end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 0&&1&\\\noalign{\medskip}0&&2&&2 \\\noalign{\medskip}&2&&2&\\\noalign{\medskip}1&&1&&0 \end { smallmatrix } \right ] ] \ ] ] @xmath1428 , \left [ \begin { smallmatrix } & 1&&0 & \\\noalign{\medskip}1&&3&&1\\\noalign{\medskip}&2&&3 & \\\noalign{\medskip}1&&1&&1\end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 1&&1&\\\noalign{\medskip}1&&3&&2 \\\noalign{\medskip}&3&&3&\\\noalign{\medskip}1&&2&&0 \end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 0&&1 & \\\noalign{\medskip}1&&2&&2\\\noalign{\medskip}&3&&2 & \\\noalign{\medskip}1&&1&&0\end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 0&&1&\\\noalign{\medskip}1&&3&&1 \\\noalign{\medskip}&3&&2&\\\noalign{\medskip}1&&1&&1 \end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 1&&1 & \\\noalign{\medskip}2&&3&&1\\\noalign{\medskip}&3&&3 & \\\noalign{\medskip}0&&2&&1\end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 1&&0&\\\noalign{\medskip}2&&2&&1 \\\noalign{\medskip}&2&&3&\\\noalign{\medskip}0&&1&&1 \end { smallmatrix } \right ] ] \ ] ] @xmath1429 , \left [ \begin { smallmatrix } & 1&&1 & \\\noalign{\medskip}1&&4&&2\\\noalign{\medskip}&3&&4 & \\\noalign{\medskip}1&&2&&1\end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 1&&1&\\\noalign{\medskip}2&&3&&2 \\\noalign{\medskip}&4&&3&\\\noalign{\medskip}1&&2&&0 \end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 0&&1 & \\\noalign{\medskip}1&&3&&2\\\noalign{\medskip}&3&&3 & \\\noalign{\medskip}1&&1&&1\end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 1&&1&\\\noalign{\medskip}2&&4&&1 \\\noalign{\medskip}&4&&3&\\\noalign{\medskip}1&&2&&1 \end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 1&&1 & \\\noalign{\medskip}2&&3&&2\\\noalign{\medskip}&3&&4 & \\\noalign{\medskip}0&&2&&1\end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 1&&0&\\\noalign{\medskip}2&&3&&1 \\\noalign{\medskip}&3&&3&\\\noalign{\medskip}1&&1&&1 \end { smallmatrix } \right ] ] \ ] ] @xmath1430 , \left [ \begin { smallmatrix } & 1&&1 & \\\noalign{\medskip}2&&3&&2\\\noalign{\medskip}&3&&3 & \\\noalign{\medskip}1&&2&&0\end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 1&&1&\\\noalign{\medskip}1&&3&&2 \\\noalign{\medskip}&3&&3&\\\noalign{\medskip}1&&1&&1 \end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 0&&1 & \\\noalign{\medskip}1&&3&&1\\\noalign{\medskip}&3&&2 & \\\noalign{\medskip}1&&1&&0\end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 1&&1&\\\noalign{\medskip}2&&3&&2 \\\noalign{\medskip}&3&&3&\\\noalign{\medskip}0&&2&&1 \end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 1&&1 & \\\noalign{\medskip}2&&3&&1\\\noalign{\medskip}&3&&3 & \\\noalign{\medskip}1&&1&&1\end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 1&&0&\\\noalign{\medskip}1&&3&&1 \\\noalign{\medskip}&2&&3&\\\noalign{\medskip}0&&1&&1 \end { smallmatrix } \right ] ] \ ] ] @xmath1431 , \left [ \begin { smallmatrix } & 1&&1 & \\\noalign{\medskip}1&&3&&1\\\noalign{\medskip}&2&&3 & \\\noalign{\medskip}1&&1&&1\end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 1&&0&\\\noalign{\medskip}1&&2&&1 \\\noalign{\medskip}&2&&2&\\\noalign{\medskip}0&&1&&0 \end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 0&&1 & \\\noalign{\medskip}1&&2&&2\\\noalign{\medskip}&2&&2 & \\\noalign{\medskip}1&&1&&0\end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 1&&1&\\\noalign{\medskip}1&&3&&1 \\\noalign{\medskip}&3&&2&\\\noalign{\medskip}1&&1&&1 \end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 0&&1 & \\\noalign{\medskip}1&&2&&1\\\noalign{\medskip}&2&&2 & \\\noalign{\medskip}0&&1&&0\end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 1&&0&\\\noalign{\medskip}2&&2&&1 \\\noalign{\medskip}&2&&2&\\\noalign{\medskip}0&&1&&1 \end { smallmatrix } \right ] ] \ ] ] @xmath1432 , \left [ \begin { smallmatrix } & 1&&1 & \\\noalign{\medskip}1&&2&&1\\\noalign{\medskip}&2&&2 & \\\noalign{\medskip}0&&1&&1\end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 0&&0&\\\noalign{\medskip}1&&1&&0 \\\noalign{\medskip}&1&&1&\\\noalign{\medskip}0&&0&&0 \end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 1&&0 & \\\noalign{\medskip}1&&2&&1\\\noalign{\medskip}&1&&2 & \\\noalign{\medskip}0&&1&&1\end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 1&&1&\\\noalign{\medskip}1&&2&&1 \\\noalign{\medskip}&2&&2&\\\noalign{\medskip}1&&1&&0 \end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 0&&0 & \\\noalign{\medskip}0&&1&&1\\\noalign{\medskip}&1&&1 & \\\noalign{\medskip}0&&0&&0\end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 0&&1&\\\noalign{\medskip}1&&2&&1 \\\noalign{\medskip}&2&&1&\\\noalign{\medskip}1&&1&&0 \end { smallmatrix } \right ] ] \ ] ] @xmath1433 , \left [ \begin { smallmatrix } & 1&&1 & \\\noalign{\medskip}1&&3&&2\\\noalign{\medskip}&2&&3 & \\\noalign{\medskip}1&&1&&1\end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 1&&1&\\\noalign{\medskip}1&&3&&1 \\\noalign{\medskip}&3&&2&\\\noalign{\medskip}1&&1&&0 \end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 0&&1 & \\\noalign{\medskip}1&&2&&2\\\noalign{\medskip}&2&&2 & \\\noalign{\medskip}0&&1&&0\end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 1&&1&\\\noalign{\medskip}2&&3&&1 \\\noalign{\medskip}&3&&2&\\\noalign{\medskip}1&&1&&1 \end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 1&&1 & \\\noalign{\medskip}1&&3&&1\\\noalign{\medskip}&2&&3 & \\\noalign{\medskip}0&&1&&1\end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 1&&0&\\\noalign{\medskip}2&&2&&1 \\\noalign{\medskip}&2&&2&\\\noalign{\medskip}0&&1&&0 \end { smallmatrix } \right ] ] \ ] ] @xmath1413 , \left [ \begin { smallmatrix } & 1&&0 & \\\noalign{\medskip}1&&1&&0\\\noalign{\medskip}&1&&1 & \\\noalign{\medskip}0&&0&&1\end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 0&&0&\\\noalign{\medskip}0&&1&&0 \\\noalign{\medskip}&0&&1&\\\noalign{\medskip}0&&0&&0 \end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 1&&0 & \\\noalign{\medskip}1&&1&&1\\\noalign{\medskip}&1&&1 & \\\noalign{\medskip}0&&1&&0\end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 0&&1&\\\noalign{\medskip}0&&1&&1 \\\noalign{\medskip}&1&&1&\\\noalign{\medskip}1&&0&&0 \end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 0&&0 & \\\noalign{\medskip}0&&1&&0\\\noalign{\medskip}&1&&0 & \\\noalign{\medskip}0&&0&&0\end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 0&&1&\\\noalign{\medskip}1&&1&&1 \\\noalign{\medskip}&1&&1&\\\noalign{\medskip}0&&1&&0 \end { smallmatrix } \right ] ] \ ] ] @xmath1434 , \left [ \begin { smallmatrix } & 1&&1 & \\\noalign{\medskip}2&&2&&1\\\noalign{\medskip}&2&&2 & \\\noalign{\medskip}0&&1&&1\end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 1&&0&\\\noalign{\medskip}1&&2&&0 \\\noalign{\medskip}&1&&2&\\\noalign{\medskip}0&&0&&1 \end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 1&&0 & \\\noalign{\medskip}1&&2&&1\\\noalign{\medskip}&1&&2 & \\\noalign{\medskip}0&&1&&0\end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 1&&1&\\\noalign{\medskip}1&&2&&2 \\\noalign{\medskip}&2&&2&\\\noalign{\medskip}1&&1&&0 \end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 0&&1 & \\\noalign{\medskip}0&&2&&1\\\noalign{\medskip}&2&&1 & \\\noalign{\medskip}1&&0&&0\end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 0&&1&\\\noalign{\medskip}1&&2&&1 \\\noalign{\medskip}&2&&1&\\\noalign{\medskip}0&&1&&0 \end { smallmatrix } \right ] ] \ ] ] @xmath1435 , \left [ \begin { smallmatrix } & 1&&1 & \\\noalign{\medskip}1&&3&&1\\\noalign{\medskip}&2&&2 & \\\noalign{\medskip}1&&1&&0\end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 1&&1&\\\noalign{\medskip}1&&2&&2 \\\noalign{\medskip}&2&&2&\\\noalign{\medskip}0&&1&&0 \end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 0&&1 & \\\noalign{\medskip}1&&2&&1\\\noalign{\medskip}&2&&1 & \\\noalign{\medskip}1&&0&&0\end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 1&&1&\\\noalign{\medskip}1&&3&&1 \\\noalign{\medskip}&2&&2&\\\noalign{\medskip}0&&1&&1 \end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 1&&1 & \\\noalign{\medskip}2&&2&&1\\\noalign{\medskip}&2&&2 & \\\noalign{\medskip}0&&1&&0\end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 1&&0&\\\noalign{\medskip}1&&2&&1 \\\noalign{\medskip}&1&&2&\\\noalign{\medskip}0&&0&&1 \end { smallmatrix } \right ] ] \ ] ] @xmath1436 , \left [ \begin { smallmatrix } & 0&&1 & \\\noalign{\medskip}0&&1&&1\\\noalign{\medskip}&1&&1 & \\\noalign{\medskip}0&&0&&0\end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 0&&0&\\\noalign{\medskip}1&&1&&0 \\\noalign{\medskip}&1&&0&\\\noalign{\medskip}0&&0&&0 \end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 1&&1 & \\\noalign{\medskip}1&&2&&1\\\noalign{\medskip}&1&&2 & \\\noalign{\medskip}0&&1&&1\end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 1&&0&\\\noalign{\medskip}1&&1&&0 \\\noalign{\medskip}&1&&1&\\\noalign{\medskip}0&&0&&0 \end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 0&&0 & \\\noalign{\medskip}0&&1&&1\\\noalign{\medskip}&0&&1 & \\\noalign{\medskip}0&&0&&0\end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 1&&1&\\\noalign{\medskip}1&&2&&1 \\\noalign{\medskip}&2&&1&\\\noalign{\medskip}1&&1&&0 \end { smallmatrix } \right ] ] \ ] ] @xmath1437 , \left [ \begin { smallmatrix } & 1&&1 & \\\noalign{\medskip}1&&2&&1\\\noalign{\medskip}&2&&1 & \\\noalign{\medskip}0&&1&&0\end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 0&&1&\\\noalign{\medskip}1&&1&&1 \\\noalign{\medskip}&1&&1&\\\noalign{\medskip}0&&0&&0 \end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 1&&0 & \\\noalign{\medskip}1&&2&&0\\\noalign{\medskip}&1&&1 & \\\noalign{\medskip}0&&0&&1\end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 1&&1&\\\noalign{\medskip}1&&2&&1 \\\noalign{\medskip}&1&&2&\\\noalign{\medskip}0&&1&&0 \end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 1&&0 & \\\noalign{\medskip}1&&1&&1\\\noalign{\medskip}&1&&1 & \\\noalign{\medskip}0&&0&&0\end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 0&&1&\\\noalign{\medskip}0&&2&&1 \\\noalign{\medskip}&1&&1&\\\noalign{\medskip}1&&0&&0 \end { smallmatrix } \right ] ] \ ] ] @xmath1438 , \left [ \begin { smallmatrix } & 1&&1 & \\\noalign{\medskip}1&&2&&1\\\noalign{\medskip}&2&&1 & \\\noalign{\medskip}1&&0&&0\end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 0&&1&\\\noalign{\medskip}0&&2&&1 \\\noalign{\medskip}&1&&1&\\\noalign{\medskip}0&&0&&0 \end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 1&&1 & \\\noalign{\medskip}2&&2&&1\\\noalign{\medskip}&2&&1 & \\\noalign{\medskip}0&&1&&0\end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 1&&1&\\\noalign{\medskip}1&&2&&1 \\\noalign{\medskip}&1&&2&\\\noalign{\medskip}0&&0&&1 \end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 1&&0 & \\\noalign{\medskip}1&&2&&0\\\noalign{\medskip}&1&&1 & \\\noalign{\medskip}0&&0&&0\end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 1&&1&\\\noalign{\medskip}1&&2&&2 \\\noalign{\medskip}&1&&2&\\\noalign{\medskip}0&&1&&0 \end { smallmatrix } \right ] ] \ ] ] @xmath1409 , \left [ \begin { smallmatrix } & 0&&0 & \\\noalign{\medskip}1&&0&&0\\\noalign{\medskip}&0&&0 & \\\noalign{\medskip}0&&0&&0\end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 1&&0&\\\noalign{\medskip}0&&1&&0 \\\noalign{\medskip}&0&&1&\\\noalign{\medskip}0&&0&&1 \end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 0&&0 & \\\noalign{\medskip}0&&0&&0\\\noalign{\medskip}&0&&0 & \\\noalign{\medskip}0&&0&&-1\end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 0&&0&\\\noalign{\medskip}0&&0&&1 \\\noalign{\medskip}&0&&0&\\\noalign{\medskip}0&&0&&0 \end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 0&&1 & \\\noalign{\medskip}0&&1&&0\\\noalign{\medskip}&1&&0 & \\\noalign{\medskip}1&&0&&0\end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 0&&0&\\\noalign{\medskip}0&&0&&0 \\\noalign{\medskip}&0&&0&\\\noalign{\medskip}-1&&0&&0 \end { smallmatrix } \right ] ] \ ] ] @xmath1414 , \left [ \begin { smallmatrix } & 0&&0 & \\\noalign{\medskip}1&&0&&0\\\noalign{\medskip}&0&&0 & \\\noalign{\medskip}-1&&0&&0\end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 1&&0&\\\noalign{\medskip}1&&1&&0 \\\noalign{\medskip}&0&&1&\\\noalign{\medskip}0&&0&&1 \end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 1&&0 & \\\noalign{\medskip}0&&1&&0\\\noalign{\medskip}&0&&1 & \\\noalign{\medskip}0&&0&&0\end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 0&&0&\\\noalign{\medskip}0&&0&&1 \\\noalign{\medskip}&0&&0&\\\noalign{\medskip}0&&0&&-1 \end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 0&&1 & \\\noalign{\medskip}0&&1&&1\\\noalign{\medskip}&1&&0 & \\\noalign{\medskip}1&&0&&0\end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 0&&1&\\\noalign{\medskip}0&&1&&0 \\\noalign{\medskip}&1&&0&\\\noalign{\medskip}0&&0&&0 \end { smallmatrix } \right ] ] \ ] ] @xmath1439 , \left [ \begin { smallmatrix } & 0&&1 & \\\noalign{\medskip}1&&1&&0\\\noalign{\medskip}&1&&0 & \\\noalign{\medskip}0&&0&&0\end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 1&&0&\\\noalign{\medskip}1&&1&&0 \\\noalign{\medskip}&0&&1&\\\noalign{\medskip}-1&&0&&1 \end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 1&&0 & \\\noalign{\medskip}1&&1&&0\\\noalign{\medskip}&0&&1 & \\\noalign{\medskip}0&&0&&0\end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 1&&0&\\\noalign{\medskip}0&&1&&1 \\\noalign{\medskip}&0&&1&\\\noalign{\medskip}0&&0&&0 \end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 0&&1 & \\\noalign{\medskip}0&&1&&1\\\noalign{\medskip}&1&&0 & \\\noalign{\medskip}1&&0&&-1\end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 0&&1&\\\noalign{\medskip}0&&1&&1 \\\noalign{\medskip}&1&&0&\\\noalign{\medskip}0&&0&&0 \end { smallmatrix } \right ] ] \ ] ] @xmath1440 , \left [ \begin { smallmatrix } & 1&&1 & \\\noalign{\medskip}0&&2&&1\\\noalign{\medskip}&1&&1 & \\\noalign{\medskip}1&&0&&0\end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 0&&1&\\\noalign{\medskip}0&&1&&1 \\\noalign{\medskip}&1&&0&\\\noalign{\medskip}0&&0&&-1 \end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 0&&1 & \\\noalign{\medskip}1&&1&&1\\\noalign{\medskip}&1&&0 & \\\noalign{\medskip}0&&0&&0\end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 1&&1&\\\noalign{\medskip}1&&2&&0 \\\noalign{\medskip}&1&&1&\\\noalign{\medskip}0&&0&&1 \end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 1&&0 & \\\noalign{\medskip}1&&1&&0\\\noalign{\medskip}&0&&1 & \\\noalign{\medskip}-1&&0&&0\end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 1&&0&\\\noalign{\medskip}1&&1&&1 \\\noalign{\medskip}&0&&1&\\\noalign{\medskip}0&&0&&0 \end { smallmatrix } \right ] ] \ ] ] @xmath1441 , \left [ \begin { smallmatrix } & 1&&1 & \\\noalign{\medskip}0&&2&&1\\\noalign{\medskip}&1&&1 & \\\noalign{\medskip}0&&0&&0\end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 0&&1&\\\noalign{\medskip}1&&1&&1 \\\noalign{\medskip}&1&&0&\\\noalign{\medskip}0&&0&&-1 \end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 1&&1 & \\\noalign{\medskip}1&&2&&1\\\noalign{\medskip}&1&&1 & \\\noalign{\medskip}0&&0&&1\end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 1&&1&\\\noalign{\medskip}1&&2&&0 \\\noalign{\medskip}&1&&1&\\\noalign{\medskip}0&&0&&0 \end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 1&&0 & \\\noalign{\medskip}1&&1&&1\\\noalign{\medskip}&0&&1 & \\\noalign{\medskip}-1&&0&&0\end { smallmatrix } \right ] , \left [ \begin { smallmatrix } & 1&&1&\\\noalign{\medskip}1&&2&&1 \\\noalign{\medskip}&1&&1&\\\noalign{\medskip}1&&0&&0 \end { smallmatrix } \right ] ] \ ] ] , a class of weighted projective curves arising in representation theory of finite - dimensional algebras . in : singularities , representation of algebras , and vector bundles ( lambrecht , 1985 ) , 265297 , lecture notes in math . 1273 , springer - verlag , berlin ( 1987 ) . | we study the multiplicative properties of the dual of lusztig s semicanonical basis .
the elements of this basis are naturally indexed by the irreducible components of lusztig s nilpotent varieties , which can be interpreted as varieties of modules over preprojective algebras .
we prove that the product of two dual semicanonical basis vectors @xmath0 and @xmath1 is again a dual semicanonical basis vector provided the closure of the direct sum of the corresponding two irreducible components @xmath2 and @xmath3 is again an irreducible component .
it follows that the semicanonical basis and the canonical basis coincide if and only if we are in dynkin type @xmath4 with @xmath5 .
finally , we provide a detailed study of the varieties of modules over the preprojective algebra of type @xmath6 .
we show that in this case the multiplicative properties of the dual semicanonical basis are controlled by the ringel form of a certain tubular algebra of type @xmath7 and by the corresponding elliptic root system of type @xmath8 .
rsum .
nous tudions les proprits multiplicatives de la base duale de la base semi - canonique de lusztig .
les lments de cette base sont naturellement paramtrs par les composantes irrductibles des varits nilpotentes de lusztig , qui peuvent tre interprtes comme varits de modules sur les algbres prprojectives .
nous dmontrons que le produit de deux vecteurs @xmath0 et @xmath1 de la base semi - canonique duale est encore un vecteur de la base semi - canonique duale si la somme directe des composantes irrductibles @xmath2 et @xmath3 est encore une composante irrductible .
il en rsulte que les bases canonique et semi - canonique ne concident que pour le type de dynkin @xmath4 avec @xmath9 .
finalement , nous tudions en dtail les varits de modules sur lalgbre prprojective de type @xmath6 .
nous montrons que dans ce cas les proprits multiplicatives de la base semi - canonique duale sont controlles par la forme de ringel dune algbre tubulaire de type @xmath7 et par le systme de racines elliptique de type @xmath8 qui lui est associ . |
ocean circulation impacts weather , climate , marine fish and mammal populations , and contaminant transport , making ocean dynamics of great industrial , military and scientific interest .
a major problem in the field of ocean dynamics involves forecasting , or predicting , a variety of physical quantities , including temperature , salinity and density . by fusing recently measured data with detailed flow models ,
it is possible to achieve improved prediction .
therefore , forecasts occur with increased accuracy @xcite . as a result , ocean surveillance may be improved by incorporating the continuous monitoring of a region of interest .
researchers have used surface drifters and submerged floats to acquire data for many years .
more recently , sensing platforms such as autonomous underwater gliders @xcite have been developed .
the gliders can operate in both littoral ( coastal ) and deep - ocean regimes , and may be used for data acquisition , surveillance and reconnaissance .
one drawback in the use of gliders involves their limited amount of total control actuation due to energy constraints , such as short battery life . for applications such as regional surveillance , energy constraints
may be alleviated by taking advantage of the dynamical flow field structures and their respective body forces .
autonomous underwater gliders are subjected to drift due to hydrodynamic forces .
this drift can be extremely complicated since the velocity fields found in the ocean are aperiodic in time and feature a complex spatial structure . instead of constantly reacting to the drift ( and thereby expending energy )
, one can minimize the glider s energy expenditure by taking advantage of the underlying structure found in geophysical flows .
however , in order to harness the ocean forces to minimize energy expenditure during control actuation , one needs to analyze the structures from the correct dynamical viewpoint .
the potential for dynamical systems tools to shed light on complex , even turbulent , flow fields such as ocean dynamics , has been understood for decades .
work in this area has intensified in the past decade with new focus placed on the improved lagrangian perspective that dynamical systems approaches may provide , especially for complex , aperiodic flows in which the traditional eulerian perspective on the flow is unhelpful or even misleading . during boundary layer separation , for example ,
sheet - like flow structures coincide with fluid particles being ejected from the wall region , a technologically important feature that an eulerian framework fails to capture in unsteady cases , but which has a clear lagrangian signature that can be identified using finite - time lyapunov exponents @xcite . in liquid jet breakup ,
a similar ejection process occurs during primary atomization , but a limited lagrangian perspective is provided by the liquid - gas interface , revealing liquid sheets and ligaments that precede droplet formation @xcite .
the development of these critical fluid sheets and ligaments can be traced to unsteady , finite - amplitude global flow structures @xcite .
dynamical systems tools thus provide a new approach to the characterization and control of these important flows .
geophysical flows offer another example where an eulerian framework is ineffective in the diagnosis of large lagrangian structures and the measurement of transport . for prediction and control of particle dynamics in large surveillance regions of interest ,
lagrangian structures of geophysical flows need to be characterized in both deterministic and stochastic settings .
the field of geophysical fluid dynamics ( gfd ) involves the study of large - scale fluid flows that occur naturally .
gfd flows are , by nature , aperiodic and stochastic .
the data sets describing them are usually finite - time and of low resolution .
established tools of dynamical systems have proven to be less effective in these cases : while providing some insight , they can not provide realistic or detailed flow field data relevant to the trajectories of tracer particles . for this , dynamical systems tools can be applied to fluid flows in an alternative way , by interpreting the eulerian flow field @xmath0 as a dynamical system @xmath1 describing the trajectories of tracer particles .
the phase space and real space are identical here and due to the incompressibility condition @xmath2 , the resulting dynamical system is conservative .
this `` chaotic advection '' approach originated with aref @xcite and demonstrated that even simple laminar two - dimensional ( 2d ) periodic and three - dimensional ( 3d ) steady flows @xcite could lead to complex , chaotic particle trajectories .
figure [ fig : ocean ] illustrates an example of how particles in a periodic flow may exhibit unexpected trajectories . in this figure , a single - layer quasi - geostrophic beta plane is being driven by a bimodal wind - stress with a small amplitude periodic perturbation .
details of this ocean model can be found in appendix [ sec : ocean_model ] . in the last two decades , this approach has led to new tools including the study of transport by coherent structures @xcite , lobe dynamics @xcite , distinguished trajectories @xcite , and global bifurcations @xcite .
even though the transport controlling structures in gfd flows are inherently complicated and unsteady , their understanding is necessary to the design of glider controls . to overcome this obstacle
we combine a set of dynamical systems tools that have proven effective in higher dimensions @xcite and stochastic problems @xcite . in this paper
, we will consider a well - known driven double - gyre flow as an example to illustrate our prediction / control framework @xcite .
this model can be thought of as a simplified version of the double - gyre shown in fig .
[ fig : ocean ] which is a solution to a realistic quasi - geostrophic ocean model .
it should be noted that our methods are general and may be applied to any flow of interest .
the goal of our approach is the production of a complete picture of particle trajectories and tracer lingering times that enables one to design a control strategy that limits tracers from switching between gyres .
( color online ) streamfunction ( color ) and two particle trajectories within a single - layer quasi - geostrophic ocean model ( see appendix [ sec : ocean_model ] ) subject to low - amplitude periodic forcing whose mean gives a steady double - gyre / western boundary current flow solution .
initial points on the trajectory are identified with open circles .
, width=321,height=321 ] the techniques we will use to analyze the dynamical systems are based on both deterministic and stochastic analysis methods , and reveal different structures depending on the system under examination . in the deterministic case , the lagrangian coherent structures reveal much about transport and information related to the basin boundaries .
they also pinpoint regions of local sensitivity in the phase space of interest . since basin boundaries are most sensitive to initial conditions , uncertainties in the data near the boundaries generate obstructions to predictability .
therefore , highly uncertain regions in phase space may be revealed by computing local probability densities of uncertain regions in the deterministic case , but using noisy initial data .
finally , in the time - dependent stochastic case , we can describe which sets contain very long term , albeit finite , trajectories .
the sets are almost invariant due to the stochastic forcing on the system , which causes random switching between the almost invariant sets .
the tools we use here are based on the stochastic frobenius - perron operator theory .
once the full structure of almost invariant sets is identified along with regions of high uncertainty , control strategies may be designed to maintain long time trajectories within a given region with minimal actuation . in fig .
[ fig : ocean ] , one can see that one particle s trajectory remains in the lower gyre for a long period of time , while the other particle escapes from the lower gyre to the upper gyre .
the tools we will develop and outline in this article will enable one to know if and when a control force must be actuated to prevent the particle from escaping .
the layout of the paper is as follows . in sec .
[ sec : themodel ] we present the stochastic double - gyre system and examine the deterministic dynamical features . in sec .
[ sec : ftle ] , we show how to use the finite - time lyapunov exponents to describe transport , and we show in sec . [ sec : unc ] how to quantify uncertainty regions .
we then turn to the stochastic system , and describe in sec .
[ sec : ais ] how to compute almost invariant sets .
section [ sec : control ] contains a discussion of our corral control strategy , and sec .
[ sec : conc ] contains the conclusions and discussion .
a simple model of the wind - driven double - gyre flow is provided by & = -a(f(x , t))(y)-x + _ 1(t),[e : toygyre_a ] + & = a(f(x , t))(y)-y + _ 2(t),[e : toygyre_b ] + & f(x , t)=(t + ) x^2+(1 - 2(t + ) ) x.[e : toygyre_c ] when @xmath3 , the double - gyre flow is time - independent , while for @xmath4 , the gyres undergo a periodic expansion and contraction in the @xmath5-direction . in eqs .
( [ e : toygyre_a])-([e : toygyre_c ] ) , @xmath6 approximately determines the amplitude of the velocity vectors , @xmath7 gives the oscillation frequency , @xmath8 determines the amplitude of the left - right motion of the separatrix between the two gyres , @xmath9 is the phase , @xmath10 determines the dissipation , and @xmath11 describes a stochastic white noise with mean zero and standard deviation @xmath12 , for noise intensity @xmath13 .
this noise is characterized by the following first and second order statistics : @xmath14 , and @xmath15 for @xmath16 . in the rest of the article we shall use the following parameter values : @xmath17 , @xmath18 , @xmath19 , and @xmath20 .
we consider the dynamics restricted to the domain @xmath21 and @xmath22 .
an unforced , undamped autonomous version of the double - gyre model was studied by rom - kedar , et.al .
@xcite , and an undamped system with different forcing was studied by froyland and padberg @xcite . prior to defining the control of certain sets of the stochastic system
, we first describe the important dynamical features of the deterministic part of eqs .
( [ e : toygyre_a])-([e : toygyre_c ] ) .
there are two attracting periodic orbits , which correspond to two fixed points of the poincar map defined by sampling the system at the forcing period .
for each fixed point , there corresponds a left and a right basin of attraction .
local analysis on the poincar section about the fixed points reveals the attractors to be spiral sinks , which generate the global double - gyre . a representative basin map at the phase @xmath23 is shown in fig . [ fig : basin ] .
one can see the complicated basin boundary structure in which the basins of attraction are intermingled , a signature of the existence of a fractal basin boundary .
there are also several unstable ( saddle ) periodic orbits corresponding to fixed points that lie along or close to the domain boundary .
due to the intermingling of the basin boundaries , one can expect that small perturbations , or uncertainties , in initial conditions near the basin boundary will generate large changes in dynamical behavior .
this can occur deterministically , or when noise is added to the system .
therefore , we quantify regions of uncertainty in phase space in both deterministic and stochastic settings in sections iv and v respectively .
( color online ) the basin poincar map for the deterministic part of eqs .
( [ e : toygyre_a])-([e : toygyre_c ] ) at phase @xmath23 .
the locations of the attracting fixed points corresponding to periodic orbits are denoted by the stars .
the coloring represents the convergence rate to the attractors . as shown by the color bar ,
positive values converge to the left basin and negative values converge to the right basin .
the largest magnitude values converge the fastest .
four saddle fixed points are located within the boundaries of the domain , and are denoted by large dots .
the remaining two are located at @xmath24 and @xmath25.,width=321,height=160 ]
one method that can be used to understand transport and which quantifies localized sensitive dependence to initial conditions in a given fluid flow involves the computation of finite - time lyapunov exponents ( ftle ) . in a deterministic setting
, the ftle also gives an explicit measure of phase space uncertainty .
given a dynamical system , one is often interested in determining how particles that are initially infinitesimally close behave as time @xmath26 .
it is well - known that a quantitative measure of this asymptotic behavior is provided by the classical lyapunov exponent @xcite . in a similar manner , a quantitative measure of how much nearby particles
separate after a specific amount of time has elapsed is provided by the ftle . in the early 1990s ,
pierrehumbert @xcite and pierrehumbert and yang @xcite characterized atmospheric structures using ftle fields .
in particular , their work enabled the identification of both mixing regions and transport barriers .
later , in a series of papers published in the early 2000s , haller @xcite introduced the idea of lagrangian coherent structures ( lcs ) in order to provide a more rigorous , quantitative framework for the identification of fluid structures .
haller @xcite proposed that the lcs be defined as a ridge of the ftle field , and this idea was formalized several years later by shadden , lekien and marsden @xcite . when computing the ftle field of a dynamical system , these lcs , or ridges , are seen to be the structures which have a locally maximal ftle value .
although the ftle / lcs theory can be extended to arbitrary dimension @xcite , in this article we consider a 2d velocity field @xmath27 given by the deterministic part of eqs .
( [ e : toygyre_a])-([e : toygyre_c ] ) which is defined over the time interval @xmath28\subset\mathbb{r}$ ] and the following system of equations : & ( t;t_i,_0)=((t;t_i,_0),t),[e : xdot ] + & ( t_i;t_i,_0)=_0,[e : xic ] where @xmath29 , @xmath30 , and @xmath31 .
as previously stated , the trajectories of this dynamical system in the infinite time limits can be quantified with the system s lyapunov exponents .
if one restricts the lyapunov exponent calculation to a finite time interval , the resulting exponents are the ftle . in practice ,
the ftle computation involves consideration of nearby initial conditions and the determination of how the trajectories associated with these initial conditions evolve in time .
therefore , the ftle provides a local measure of sensitivity to initial conditions and measures the growth rates of the linearized dynamics about the trajectories .
since the details of the derivation of the ftle @xcite as well as applications that employ the ftle @xcite have appeared in the literature , we shall only briefly summarize the procedure . the solution of the dynamical system given by eqs .
( [ e : xdot])-([e : xic ] ) from the initial time @xmath32 to the final time @xmath33 can be viewed as the flow map @xmath34 which is defined as follows : @xmath35 we consider an initial point located at @xmath36 at @xmath37 along with a perturbed point located at @xmath38 at @xmath37 . using a taylor series expansion
, one finds that @xmath39 dropping the higher order terms , the magnitude of the linearized perturbations is given as @xmath40 where @xmath41 is the right cauchy - green deformation tensor and is given as follows : @xmath42 with * denoting the adjoint . then the ftle can be defined as @xmath43 where @xmath44 is the maximum eigenvalue of @xmath41 . for a given @xmath45 at an initial time @xmath32 , eq .
( [ e : sigma ] ) gives the maximum finite - time lyapunov exponent for some finite integration time @xmath46 ( forward or backward ) , and provides a measure of the sensitivity of a trajectory to small perturbations . the ftle field given by @xmath47
can be shown to exhibit ridges of local maxima in phase space .
the ridges of the field indicate the location of attracting ( backward time ftle field ) and repelling ( forward time ftle field ) structures . in 2d space ,
the ridge is a curve which locally maximizes the ftle field so that transverse to the ridge one finds the ftle to be a local maximum .
figure [ fig : ftle ] shows a snapshot taken at phase @xmath23 of the forward time ftle field calculated using the deterministic form of eqs .
( [ e : toygyre_a])-([e : toygyre_c ] ) for a finite integration time @xmath48 .
one can see in fig .
[ fig : ftle ] that there are ridges ( in red ) of locally maximal ftle values .
these ridges or lcs effectively separate the phase space into distinct dynamical regions . for the deterministic system
, a particle placed in one of these distinct regions will remain in that region as the system evolves in time .
notice also that the red ridges appear to be a subset of the basin boundary set separating the two basins of attraction shown in fig .
[ fig : basin ] .
( color online ) forward ftle flow field computed using the deterministic form of eqs .
( [ e : toygyre_a])-([e : toygyre_c ] ) and shown at phase @xmath23 .
the integration time is @xmath48 with an integration step size of @xmath49 and a grid resolution of @xmath50 in both @xmath5 and @xmath51 .
, width=321,height=160 ] in the stochastic system , the noise acts continuously on a particle placed in one of these distinct basins , and therefore , it is possible that the particle will cross the lcs and move into the other basin .
even though we find the lcs using the deterministic system , the location of these structures are a valuable tool in understanding how a particle escapes from the basin in which it is initially placed . coupled with the ideas of uncertain sets and almost invariant sets described below
, we will use the lcs information provided by the ftle field to determine appropriate regions of control . in this way , it is possible to increase the loitering times of a particle in a basin while minimizing the amount of control actuation .
in this section , we describe a method that measures the fraction of uncertainty for regions in phase space .
nonlinear systems that possess multiple attractors typically can be extremely sensitive with respect to the choice of initial conditions .
that is , very small changes , or uncertainty , in the initial conditions can lead to different attractors @xcite .
sensitivity here is measured in the asymptotic time limit . when applied to high - dimensional attractors ,
long - time sensitivity is measured with respect to parameter sensitivity @xcite .
in contrast to the asymptotic definition , we measure uncertainty in phase space with respect to perturbations in initial data by computing exponential changes in distances over short time intervals .
( color online ) the uncertainty measure for the deterministic part of eqs .
( [ e : toygyre_a])-([e : toygyre_c ] ) for the phase @xmath23 . a 200 x 100 regular grid of initial conditions is used .
the color represents the fraction of trajectories that diverge from the trajectory of the center of each ball tested .
the computation was performed using balls of radius @xmath52 with twenty points chosen at random within each ball .
the trajectories were computed until time @xmath53 using a threshold parameter of @xmath54.,width=321,height=160 ] the sensitivity of the deterministic system to the initial conditions is quantified by defining the pointwise uncertainty measure . for a point in phase space @xmath55 chosen at random , a ball of radius @xmath8 is constructed about that point , @xmath56 . for @xmath57
randomly chosen points in the ball @xmath56 , we record the number of trajectories , @xmath58 , at time @xmath59 , that diverge beyond a certain distance from the trajectory of @xmath55 . by diverge , we mean that we test if the distance between the trajectories at a given time is greater than a threshold parameter .
the test indicates the fraction of initial conditions in the ball that diverge beyond a certain distance in a given finite time .
if there are multiple attractors present , then as @xmath60 , the uncertain points correspond to those points that go to another basin . for finite @xmath59 , as the radius of the ball is decreased , one obtains more accurate approximations of the basin boundary of the system .
this is due to the fact that the dynamics are most sensitive with respect to the choice of initial conditions when the points straddle the basin boundary separating the basins of attraction .
figure [ fig : uncertain ] shows a sample calculation of the uncertainty measure for the fixed phase @xmath23 . in this calculation
, a 200 x 100 regular grid is used where each grid point coincides with the center of a ball of radius 0.02 . by comparing the ftle field shown in fig .
[ fig : ftle ] with the uncertainty measure shown in fig .
[ fig : uncertain ] , one can see that the location of the ftle ridge is contained within the uncertainty region .
in addition , it should be noted that the uncertainty measure captures the complicated fractal structure of the basin boundary shown in fig .
[ fig : basin ] , as radius @xmath8 approaches zero . for the stochastic double - gyre
, it is expected that the regions defined by the uncertain sets over all phases , @xmath61 will be the most active regions in which to actuate if one wishes to increase the residence time of a particle in one particular basin .
we now refine this notion in the discrete case by defining the actual sets which will be almost invariant in the stochastic case .
one may think of the almost invariant sets as the complement of the most uncertain regions in phase space .
as mentioned above , approximating the almost invariant sets is another method that can quantify where particles loiter in phase space for a stochastic system .
computation of the almost invariant sets will be achieved using the stochastic frobenius - perron ( sfp ) operator .
while the transition probabilities can be approximated for continuous time @xcite , we take advantage of the fact that the double - gyre system is a periodically forced system .
therefore , the dynamics are sampled at a particular phase of the period . for small noise ,
contributions to the probability along trajectories come primarily from the difference between the tangent vector and the vector field .
averaging over a period of the drive allows one to sample the noise periodically . in doing this
, we assume that the noise is not in the tail of the probability distribution so that there are no large , rare events .
we therefore construct a poincar map with a specified noise distribution .
this method of discrete sampling quantitatively identifies the complement of the uncertainty region as two almost invariant sets , which are associated with the stable attractors of the associated deterministic system .
although we present the machinery for a fixed phase , it is possible to extend the analysis for all phases of @xmath9 , allowing one to approximate the density over the full forcing period . to be explicit ,
consider a stochastically perturbed map @xmath62 , where @xmath63 is a random variable having the distribution @xmath64 .
let @xmath65 represent the map on the poincar section which samples the flow at time intervals of length @xmath59 , equivalent to the period of the steady states .
the sfp operator for a normal distribution with mean zero and standard deviation @xmath66 is defined as @xcite @xmath67=\frac{1}{\sqrt{2\pi\sigma^{2}}}\int_{m}e^{-\frac{\vert(x - f(y))\vert^{2}}{2\sigma^{2}}}\rho(y)dy.\label{fpnormal}\ ] ] one approximates the density @xmath68 by the finite sum of basis functions , @xmath69 , where @xmath70 , and @xmath71 is an indicator defined on boxes @xmath72 covering the region @xmath73 .
the sfp operator is approximated by the @xmath74 matrix , @xmath75,\phi_{i})=\int_{m}p_{f}[\phi_{j}(x)]\phi_{i}(x)dx\label{eq : aij matrix}\ ] ] for @xmath76 .
therefore , a transition matrix entry @xmath77 value represents how mass , or measure , flows from cell @xmath78 to cell @xmath79 .
details of the method can be found in ref . .
( color online ) the results of approximating the almost invariant sets for the stochastic system given by eqs .
( [ e : toygyre_a])-([e : toygyre_c ] ) at the phase @xmath23 . a 100 x 100 grid of points is used with a standard deviation of @xmath80 .
figure [ fig : almostinvariant](a ) is the mapping of the second eigenvector of the reversible matrix back to the phase space .
the almost invariant sets are represented by the red and the blue regions at the ends of the eigenvector range .
figure [ fig : almostinvariant](b ) is the contour plot of the surface , which shows the transition region that forms a barrier between the almost invariant sets.,title="fig:",height=188 ] ( color online ) the results of approximating the almost invariant sets for the stochastic system given by eqs .
( [ e : toygyre_a])-([e : toygyre_c ] ) at the phase @xmath23 . a 100 x 100 grid of points is used with a standard deviation of @xmath80 .
figure [ fig : almostinvariant](a ) is the mapping of the second eigenvector of the reversible matrix back to the phase space .
the almost invariant sets are represented by the red and the blue regions at the ends of the eigenvector range .
figure [ fig : almostinvariant](b ) is the contour plot of the surface , which shows the transition region that forms a barrier between the almost invariant sets.,title="fig:",height=188 ] from the transition matrix , a reversible markov chain is constructed , satisfying @xmath81 for all @xmath82 .
such a condition implies @xmath83 is an invariant ( stationary ) distribution of @xmath6 @xcite . since in general @xmath6
is not reversible as it is defined in eq .
( [ eq : aij matrix ] ) , a reversible markov chain is constructed from the transition matrix .
let @xmath84 , where @xmath85 and @xmath83 is an invariant probability density of @xmath6 .
the matrix @xmath86 now possesses detailed balance .
since @xmath86 is reversible , it has the properties that its eigenvalues are real and the eigenvectors are orthogonal .
one can then compute a collection of sets which are almost invariant by examining the eigenvectors of @xmath86 .
a gauge of how much measure is ejected each iterate from the defined almost invariant sets is given by the eigenvalues .
the first few eigenvalues of @xmath86 will be clustered near unity and their associated eigenstates will be a mixture of almost invariant sets formed from the above basis elements . since the first eigenvector of @xmath86 is a vector of ones , the second eigenvector is used to identify two almost invariant sets emanating from basins of attraction . the fact that the eigenvalues cluster near unity means that very little mass is ejected from the basins each iterate , which is why the basins are defined as almost invariant states .
the results of the almost invariant sets computation are shown in fig .
[ fig : almostinvariant ] for the poincar map of the double - gyre at @xmath23 .
the second eigenvector can be visualized by a roughly piecewise constant function , with a transition region connecting the two pieces .
the eigenvalue associated with this eigenvector is 0.9996 .
the higher level represents the left basin ( red ) and the lower level represents the right basin ( blue ) . for this computation ,
a 100 x 100 grid of points is used with a standard deviation of @xmath80 . as expected
, the complement set of the left and right almost invariant basins agrees approximately with the location of the uncertain region from fig .
[ fig : uncertain ] .
( color online ) a graph of the @xmath87coordinate for two simulations of the stochastic system given by eqs .
( [ e : toygyre_a])-([e : toygyre_c ] ) . for continuous noise with standard deviation @xmath88 , one does not observe switching events ( red ) .
however , for the larger standard deviation of @xmath80 , one can see frequent switching between the two basins ( black).,width=321,height=160 ] since we have now computed the location of the lagrangian coherent structures and have approximated the most uncertain regions and almost invariant regions of phase space , we can consider loitering , or residence times , within a defined region .
moreover , we may use actuation to control trajectories in the system by increasing the loitering times within an almost invariant set . to quantify the residence time of a particle in a basin
, we consider the double - gyre system with continuously added gaussian noise . as the standard deviation , @xmath66 , of the noise
is increased , trajectories wander in larger neighborhoods about the stable steady states and frequently switch from one basin to another , as shown in fig .
[ fig : switching ] . these basins
are approximated by the almost invariant sets at a fixed phase , shown in fig .
[ fig : almostinvariant ] .
the control strategy employs a monitoring ball ( that we refer to as a control region ) that covers the almost invariant set with the center defined as the left non - boundary steady state when time @xmath89 .
when a trajectory passes beyond a threshold distance from the center @xmath90 of the control region , a radial force is turned on .
this force moves the trajectory back towards the center of the region .
this control strategy can be modeled as & = -a(f(x , t))(y)-x + _ 1(t)-c(x - x_s)(r(x , y)-r_0),[e : xdot_c ] + & = a(f(x , t))(y)-y + _ 2(t)-c(y - y_s)(r(x , y)-r_0),[e : ydot_c ] + & f(x , t)=(t + ) x^2+(1 - 2(t + ) ) x , + & r(x , y)= , [ e : r_c ] where @xmath91 is the magnitude of the control force , @xmath92 is the radial threshold , and @xmath93 is a heaviside function . by adjusting @xmath94 and @xmath91 ,
it is possible to control the number of actuations and the effectiveness of the control scheme .
we have designed this specific control strategy to take advantage of the underlying flow structure in order to corral the particles into the almost invariant set .
we therefore refer to our scheme as corral control .
the control scheme is tested by observing the changes in the almost invariant sets .
figure [ fig : almostinvariant_control ] shows the results of approximating the almost invariant sets for the stochastic system with control applied to the left basin given by eqs .
( [ e : xdot_c])-([e : r_c ] ) for @xmath95 .
as expected , there is an increase in the size of the left ( red ) almost invariant set and a decrease in the size of the right ( blue ) almost invariant set when compared to fig .
[ fig : almostinvariant ] .
( color online ) the results of approximating the almost invariant sets for the stochastic system with control applied to the left basin given by eqs .
( [ e : xdot_c])-([e : r_c ] ) at the phase @xmath23 . a 100 x 100 grid of points with a standard deviation of @xmath80 is used .
similar to figure [ fig : almostinvariant](a ) , this graph is the mapping of the second eigenvector of the reversible matrix back to the phase space .
the almost invariant sets are represented by the red and the blue regions at the ends of the eigenvector range .
one can see that the fig .
[ fig : almostinvariant_control ] left basin ( red ) is larger than the fig .
[ fig : almostinvariant](a ) left basin , while the fig .
[ fig : almostinvariant_control ] right basin ( blue ) is smaller than the fig .
[ fig : almostinvariant](a ) right basin.,height=302 ] we have analyzed how individual trajectories escape under this control algorithm . by measuring the most frequent path of an escape event from the left basin to the right basin
, one can observe that the trajectories follow the ftle ridge towards a periodic saddle orbit on the @xmath5-axis .
topologically , the ftle ridge loops around the almost invariant sets , corralling points in the outlying region to the other side .
the goal is to reduce the number of escape events by a control algorithm that takes advantage of the topology of the system .
figures [ fig : topology](a)-(c ) illustrate the topology ( as does a movie which can be found as auxiliary online material ) . in these figures , @xmath96 particles were placed in the left basin near the fixed point and their trajectories were tracked .
if a particle crossed the @xmath97 threshold , the particle was considered to have escaped from the left basin into the right basin . in figs .
[ fig : topology](a)-(c ) , the color contours show the probability density of the trajectory path prehistory for all the particles ( there are 4361 of them ) that escaped from the left basin between phases @xmath98 .
the prehistory shows that there are many paths to escape .
however , the high probability density in the region of the saddle ( near the point @xmath99 ) shows that most of the particles have been funneled into the saddle region just prior to escape from the left basin . also shown in the figures
are the position of the particles ( black dots ) before escape , the ftle contour , and a circular region of radius @xmath100 that denotes the control region .
these last three features are shown at phase @xmath101 in fig .
[ fig : topology](a ) , at phase @xmath102 in fig .
[ fig : topology](b ) , and at @xmath103 in fig .
[ fig : topology](c ) .
the control force used in this simulation is @xmath104 . in fig .
[ fig : topology](a ) , one can see that although many particles are outside the control region , the majority of the particles are inside the control region . for those particles that are inside the control region ,
no control is being applied .
in addition , the majority of the particles lie on one side of the ftle ridge , which is bounding the almost invariant set ( from comparison with fig .
[ fig : almostinvariant ] ) .
as time evolves , one can see in fig .
[ fig : topology](b ) that the ftle ridge sweeps up the left side of the figure , and one can see from the prehistory that about half of the particles sweep up the left side , staying on the exterior side of the ftle ridge outside the control region so that the control is actuated .
the other half of the particles remain on the interior side of the ftle ridge inside the control region so that no control is actuated . eventually , as can be seen in fig .
[ fig : topology](c ) , the ftle ridge enters the region of uncertainty ( see fig . [
fig : almostinvariant ] ) , and all the particles cross over the ftle ridge .
all the particles now have left the control region .
however , even though the control is being actuated , it can not overcome the underlying dynamics found in this system and all the particles will proceed towards the saddle and finally escape . to prevent escape and to increase loitering time , one can use the knowledge of the location of the ftle ridge , uncertainty regions and almost invariant sets to design a control region .
for example , by choosing a control region with a smaller radius , one can prevent the particles from approaching the ftle ridges and uncertain areas .
with such a small radius , however , the number of control actuations will be very large .
control actuation may be optimized by using a control region with the largest possible radius so that the region does not intersect the ftle structures .
( color online ) probability density of the trajectory path prehistory for particles that escaped from the left basin based on a threshold of @xmath97 between phases @xmath98 .
these paths form a subset of @xmath96 random particle simulations .
also shown is the position of the particles ( black dots ) before escape , the ftle ridge ( black curve ) , and the control region of radius @xmath100 ( red ) at ( a ) phase @xmath105 , ( b ) phase @xmath102 , and ( c ) phase @xmath106 .
the control force is @xmath104.,width=311 ] ( color online ) probability density of the trajectory path prehistory for particles that escaped from the left basin based on a threshold of @xmath97 between phases @xmath98 .
these paths form a subset of @xmath96 random particle simulations .
also shown is the position of the particles ( black dots ) before escape , the ftle ridge ( black curve ) , and the control region of radius @xmath100 ( red ) at ( a ) phase @xmath105 , ( b ) phase @xmath102 , and ( c ) phase @xmath106 .
the control force is @xmath104.,width=311 ] + ( color online ) probability density of the trajectory path prehistory for particles that escaped from the left basin based on a threshold of @xmath97 between phases @xmath98 .
these paths form a subset of @xmath96 random particle simulations .
also shown is the position of the particles ( black dots ) before escape , the ftle ridge ( black curve ) , and the control region of radius @xmath100 ( red ) at ( a ) phase @xmath105 , ( b ) phase @xmath102 , and ( c ) phase @xmath106 .
the control force is @xmath104.,width=311 ] we vary the control parameter , @xmath91 , to study its effect on the mean escape time .
one expects that an increase in @xmath91 will lead to a corresponding increase in the mean time to escape .
note that since the control force is finite , there is always a finite probability of escape even if control is actuated .
the control algorithm decreases the probability of trajectories visiting the regions that have a high probability of transition , and the mean time to escape reflects the decrease in pdf in the transition regions
. figure [ fig : escape](a ) shows the natural log of the average escape time as a function of the inverse of the noise intensity .
one can see that as the control force is increased , the mean time to escape increases with the slope depending linearly on the control parameter . since small changes in @xmath91 translate into exponential changes in residence times , one recovers a linear relation of the slope of the exponent as a function of @xmath91 which can be seen in fig . [
fig : escape](b ) .
( color online ) ( a ) natural log of the average escape time as a function of the inverse of the noise intensity .
the control radius is @xmath100 and a threshold of @xmath97 is used to identify escape from the left basin .
the results are averaged over 1000 simulations and the slopes of the linear fit to each data set is noted on the right .
( b ) the slopes of the average escape time linear fits from fig .
[ fig : escape](a ) as a function of the control parameter @xmath91.,title="fig:",width=302 ] ( color online ) ( a ) natural log of the average escape time as a function of the inverse of the noise intensity .
the control radius is @xmath100 and a threshold of @xmath97 is used to identify escape from the left basin .
the results are averaged over 1000 simulations and the slopes of the linear fit to each data set is noted on the right .
( b ) the slopes of the average escape time linear fits from fig .
[ fig : escape](a ) as a function of the control parameter @xmath91.,title="fig:",width=245 ] another consideration is that one would like to minimize the number of actuations for the particle , thereby preserving the energy needed to employ the control scheme . as a consequence , this could extend the use of a battery in an autonomous underwater glider .
figure [ fig : actuate ] shows the natural log of the average number of actuations per unit time as a function of the control radius . for @xmath107 , the ftle ridge does not intersect the control region and
the average number of actuations follows the scaling of the average escape time . beyond that
, the number of actuations increases as the control scheme works to move the trajectory back inside the control region .
therefore , the number of actuations per unit time follows the average escape rate for @xmath107 and the minimum number of actuations can be found for the largest control region inside the ftle ridge . natural log of the average number of actuations per unit time as a function of the control radius for parameters @xmath108 and @xmath95 .
the mean number of actuations @xmath109 was calculated over 100 simulations of time series with @xmath110 , or until an escape event occurred .
the error bars show one standard deviation of the data from the mean .
overlaid is a linear fit for the data points @xmath107 , since for this range of @xmath111 the control region does not intersect the ftle ridge .
, width=321 ] to understand the effectiveness of the control scheme and its scaling properties , we approximate the natural probability density function ( pdf ) and the almost invariant set in the following 1d example . consider a stochastic system , where the deterministic part has a stable focus at the origin surrounded by an unstable limit cycle . in the associated deterministic system , the points inside the limit cycle would spiral toward the stable focus , while the points outside would diverge .
the stochastic perturbations allow trajectories to escape across the limit cycle ( and then diverge ) , creating an almost invariant set .
this captures the local behavior in one basin of the double - gyre .
now consider a control region with radius @xmath112 , where @xmath8 determines the distance from the attractor for which the control is actuated .
then the system has the form @xmath113 = \left[\begin{array}{cc } -\lambda & \omega \\ - \omega & -\lambda\end{array}\right ] \left[\begin{array}{c } x \\ y \end{array}\right ] + \lambda\left[\begin{array}{c } x \\ y \end{array}\right](x^{2}+y^{2 } ) - c\theta(r-\epsilon ) \left[\begin{array}{c } x \\ y \end{array}\right ] + \bm{\eta}(t ) , \label{eq : unstable lc controller}\ ] ] where @xmath114 . as before , @xmath91 is the magnitude of the control force and @xmath93 is a heaviside function .
each component of @xmath115^{\mbox{t}}$ ] , is a random variable with intensity @xmath13 . to find the pdf , we switch to polar coordinates using the time - dependent change of variables given by @xmath116 and @xmath117 .
this results in the following transformed stochastic system : @xmath118 here , we find that @xmath119 and the noise vanishes entirely from the @xmath120 equation . assuming the transformed noise term still has intensity @xmath13 , and @xmath121 is a normalization constant , the probability density @xmath122 is given by @xmath123 the pdf is now fully specified by the deterministic flow and noise drift , control strength and actuation region , and can be plotted for various parameters . in fig .
[ fig : pdf ] one can see how an increase in the control strength decreases the probability of the trajectory moving outside the boundary of the control region .
the inset figure shows that the local minimum of the pdf occurs at @xmath124 , which is the location of the unstable limit cycle .
the pdfs branch at the point of control at @xmath125 .
the thick solid curve represents the dynamics with no control .
it follows that a trajectory escaping outside the limit cycle will diverge to infinity , and the pdf increases as one increases @xmath111 .
( color online ) pdf as a function of the control radius @xmath111 for varying control force @xmath91 found using eq .
( [ eq : radial pdf ] ) .
continuous noise is used with @xmath126 , @xmath127 , and @xmath128 .
the normalization constant , @xmath121 , was set in each case so that the area under the curve is one .
the inset shows a close - up of the pdf near the attractor for the case of no control.,width=321 ] in addition , one can perform the one - dimensional almost invariant set analysis for this example .
the transport matrix is computed using a grid of 500 intervals on the domain and gaussian noise with standard deviation of @xmath129 .
the point of control is located at @xmath130 , to the left of the basin boundary at @xmath124 .
figure [ fig : ais_control](a ) , shows the expansion of the left almost invariant set to the right as the control is increased .
this information is contained in the second eigenvector of the transport matrix .
the lower and upper almost constant regions of the function represent location of the sets .
figure [ fig : ais_control](b ) , shows the movement of the transport region to the right as the control is increased .
this information is contained in the third eigenvector of the transport matrix .
the maximum value of the function represents the location of the transition region .
notice that it is the complement of the almost invariant sets .
both the pdf and almost invariant set analysis demonstrate the change that the control algorithm has on the natural dynamics of the stochastic system .
the control algorithm decreases the probability of trajectories visiting regions where deterministic dynamics would cause them to diverge .
this moves the effective basin boundary farther from the attractor , increasing its size . by using sufficient control radius and force , any trajectory that would naturally diverge
can be redirected towards the attractor .
therefore , the almost invariant set can be expanded to the desired size at a cost of the number of control actuations .
( color online ) the results of the almost invariant sets analysis for varying control force @xmath91 in eq .
( [ eq : radial lc controller ] ) with @xmath127 , @xmath131 , and gaussian noise with standard deviation of @xmath132 .
figure [ fig : ais_control](a ) shows the almost invariant set for varying control force @xmath91 .
the second eigenvector of the transport matrix is mapped back to the domain @xmath111 .
notice that as @xmath91 increases , the right basin extends to the right .
figure [ fig : ais_control](b ) shows the transition set for varying control force @xmath91 .
the third eigenvector of the transport matrix is mapped back to the domain @xmath111 .
notice that as @xmath91 increases , the transition region moves to the right . , title="fig:",width=302 ] ( color online ) the results of the almost invariant sets analysis for varying control force @xmath91 in eq .
( [ eq : radial lc controller ] ) with @xmath127 , @xmath131 , and gaussian noise with standard deviation of @xmath132 .
figure [ fig : ais_control](a ) shows the almost invariant set for varying control force @xmath91 .
the second eigenvector of the transport matrix is mapped back to the domain @xmath111 .
notice that as @xmath91 increases , the right basin extends to the right .
figure [ fig : ais_control](b ) shows the transition set for varying control force @xmath91 .
the third eigenvector of the transport matrix is mapped back to the domain @xmath111 .
notice that as @xmath91 increases , the transition region moves to the right .
, title="fig:",width=302 ] additionally , it is possible to quantify the relation between the mean escape time and the potential defined by the pdf . using the well - known kramer s escape rate @xcite
, one can predict the rate at which a particle can escape over a potential barrier under brownian motion . in one dimension ,
the escape time of a particle from a potential defined by @xmath133 is @xmath134}$ ] , where @xmath135 is the location of the attractor and @xmath136 is the location of the escape point . from the radial control
pdf given by eq .
( [ eq : radial pdf ] ) , one can see that the potential function exponent depends linearly on the control amplitude @xmath91 .
all of these one - dimensional analytic results are consistent with the behavior seen for the double - gyre system with control that was presented in the preceding subsections .
using modern dynamical systems theory , we have analyzed the flow of a stochastic double - gyre as a model of a wind - driven ocean .
the existence of multiple almost invariant sets combined with highly uncertain regions in phase space gives rise to stochastic switching between the basins .
the uncertain regions generate high probability transition sets which are the primary cause of switching from one basin to another .
we showed that the high probability of transitions may be approximated by computing uncertain regions with respect to initial conditions and the use of ftle .
the system displays maximum sensitivity near the deterministic basin boundary trajectory .
knowledge of the ftle ridges in a given flow , in conjunction with knowledge of uncertain regions and almost invariant sets , enables one to predict the location of escape trajectories from one basin to another . as in previous work @xcite
, these transition regions may be monitored directly to predict future switching events .
this knowledge , in turn , allows for set - based corral control methods to be used to inhibit the escape event and increase the loitering time within the basin .
almost invariant sets were computed explicitly using the frobenius - perron theory for discrete stochastic systems .
the sets provided approximate target regions in which to control particles which act as sensors in large surveillance regions .
that is , given initial conditions in one of the almost invariant basins , the particles will stay in the sets for long , but finite time .
although computed for a discrete map model , the sets approximate well the continuous noise almost invariant sets .
they also form sets of the most certain points in that they are the complement of the uncertain region , which was computed based on initial condition uncertainty . since the almost invariant sets are indicative of long - time , but finite , dynamics , we used these sets or subsets to define corral control regions in which we wish to maintain the residence times in one basin with minimal control actuation . to this end , we designed a simple discrete controller to extend the residence times in the continuous noise double - gyre .
we find that even small control amplitudes have an exponential increase on the residence times .
this is quite similar to the well - known problem of noise - induced escape from a potential well , even though the basin boundaries may be fractal in the deterministic case .
we saw that if the control radius defined a set which did not intersect the ridges defined by the ftle , the number of actuations per unit time followed an exponential law as a function of noise intensity .
this can be understood completely from the switching rate theory . on the other hand ,
if the control radius intersected the ftle ridge , the uncertainty regions were breached , and the control actuation rate no longer follows a nice scaling law , but rather some complicated function of control radius @xmath111 .
the model presented here is a simplified version of a double - gyre flow that is a solution to a realistic quasi - geostrophic ocean model , and the vehicles controlled are point particles .
however , the techniques here can be extended to full ocean and glider models in the future .
the extensions to vehicles with real mass means that inertial effects will need to be included , as well extensions to 3d gfd models .
however , the machinery presented here is well - suited to the description of sets in higher dimensions , and we expect that the monitoring of large surveillance regions in the ocean by gliders will be enhanced by implementing our corral control method .
additionally , the study and control of coupled systems , with and without delay , is of interest .
we gratefully acknowledge support from the office of naval research . e.f .
is supported by the naval research laboratory ( award no .
n0017310 - 2-c007 ) .
the ocean model results in fig . [ fig : ocean ] were obtained by numerical solution of barotropic quasi - geostrophic flow in a single layer basin @xmath137\times[0,1]$ ] on a @xmath138-plane .
the governing non - dimensionalized equation for the fluid streamfunction @xmath139 is : @xmath140 where @xmath141 is the jacobian operator , @xmath142 and forcing is provided by a wind stress curl , @xmath143 , that is prescribed as follows to form a double - gyre circulation with a weak periodic `` seasonal '' variation : @xmath144 where the amplitude @xmath145 and frequency @xmath146 were used to produce fig .
[ fig : ocean ] .
this system is characterized by the non - dimensional parameters @xcite @xmath147 where @xmath138 is the rotation parameter , @xmath86 is the bottom friction and @xmath148 and @xmath149 are respectively the characteristic length and velocity scales of the basin . the parameters @xmath150 and @xmath8 correspond to the relative length scales of the stommel and inertial layers , respectively . to produce fig . [ fig : ocean ] , we used @xmath151 and @xmath152 .
the above model has been numerically integrated using second - order spatial differences and second - order runge - kutta time stepping for the streamfunction , and a fourth - order runge - kutta algorithm to compute the lagrangian trajectories of tracer particles . in fig .
[ fig : ocean ] , a grid of resolution @xmath153 was used and a dimensionless time step of @xmath154 .
coordinates of tracer particles are independent of the grid while flow velocities at the particle locations are found using bilinear interpolation from the grid values .
an initially static ocean spins - up in a few hundred time steps .
if @xmath155 the spun - up solution is stationary , while non - zero @xmath10 leads to a superimposed oscillatory behavior .
the tracers are held in place until spin - up is complete . c. c. eriksen , t. j. osse , r. d. light , t. wen , t. w. lehman , p. l. sabin , j. w. ballard , and a. m. chiodi , `` seaglider : a long - range autonomous underwater vehicle for oceanographic research , '' ieee j. oceanic eng . *
26 * , 424 ( 2001 ) .
k. ide , d. small , and s. wiggins , `` distinguished hyperbolic trajectories in time - dependent fluid flows : analytical and computational approach for velocity fields defined as data sets , '' nonlinear proc . geoph .
* 9 * , 237 ( 2002 ) .
r. meucci , d. cinotti , e. allaria , l. billings , i. triandaf , d. morgan , and i. b. schwartz , `` global manifold control in a driven laser : sustaining chaos and regular dynamics , '' physica d * 189 * , 70 ( 2004 ) .
s. c. shadden , f. lekien , and j. e. marsden , `` definition and properties of lagrangian coherent structures from finite - time lyapunov exponents in two - dimensional aperiodic flows , '' physica d * 212 * , 271 ( 2005 ) .
m. branicki and s. wiggins , `` finite - time lagrangian transport analysis : stable and unstable manifolds of hyperbolic trajectories and finite - time lyapunov exponents , '' nonlinear proc .
* 17 * , 1 ( 2010 ) .
w. tang , p. w. chan , and g. haller , `` accurate extraction of lagrangian coherent structures over finite domains with application to flight data analysis over hong kong international airport , '' chaos * 20 * , 017502 ( 2010 ) . | we consider the problem of stochastic prediction and control in a time - dependent stochastic environment , such as the ocean , where escape from an almost invariant region occurs due to random fluctuations .
we determine high - probability control - actuation sets by computing regions of uncertainty , almost invariant sets , and lagrangian coherent structures .
the combination of geometric and probabilistic methods allows us to design regions of control that provide an increase in loitering time while minimizing the amount of control actuation .
we show how the loitering time in almost invariant sets scales exponentially with respect to the control actuation , causing an exponential increase in loitering times with only small changes in actuation force .
the result is that the control actuation makes almost invariant sets more invariant .
* prediction and control of the motion of an object in time - dependent and stochastic environments is an important and fundamental problem in nonlinear dynamical systems .
one of the main goals of control is the design of a theory that can take unstable states and render them stable .
for example , small perturbations at the base of an inverted pendulum will stabilize the inverted state .
noise poses a greater problem for deterministically controlled states , in that stochastic effects destabilize the states as well as their neighborhoods .
therefore , control theory of stochastic dynamical systems may be addressed by examining the change in stability of certain sets . * we present a variety of geometric and probabilistic set - based methods that enable one to compute controllable sets . for a particle moving under the influence of deterministic and stochastic forces with no control ,
these sets determine regions which are unstable in the sense that the particle will leave the set after a sufficiently long time .
controls added to particle dynamics to increase the time to escape ( or loitering time ) have a strong dependence on the probabilistic and geometric set characteristics .
the determination of controls and associated sets allow for an increase in the amount of time the particle can loiter in a particular region while minimizing the amount of control actuation .
our theoretical analysis shows how an increase in the strength of the control force leads to a decrease in the probability that an object will escape from the control region .
in fact , we have found that small changes in the control actuation force have an exponential effect on the loitering time of the object .
additionally , we show how the exponential increase in escape times from the controlled sets is related to the problem of noise - induced escape from a potential well . |
the infrared astronomy satellite mission ( iras ) @xcite successfully performed the first all - sky survey at infrared wavelengths , providing many fruitful results . after this pioneering mission over two decades ago , the akari satellite ( previously known as astro - f ) @xcite is performing the next generation all - sky survey primarily with the fis ( far - infrared surveyor ) focal plane instrument in 4 far - ir bands ( 50200 @xmath0 m range ) to improved spatial resolutions and wider wavelength coverage than the its predecessor iras [ see @xcite for detailed specifications and hardware performance of the fis instrument ] . although the telescope apertures of iras ( 60 cm ) and akari ( 68.5 cm ) are similar , the effective beam size of the akari mission ( 0.5@xmath1 @xmath2 0.9@xmath1 ) is much smaller than that of iras ( 2@xmath1 @xmath2 5@xmath1 ) because the sizes of the detector pixels in iras were much larger than the size of the diffracted beam of the telescope . due to this relatively large beam size
, the astronomical observations made in the far - ir bands of the iras mission were presumed to have been affected by the fluctuation noise arising from both background structures and many overlapping extragalactic point sources below the resolution of the beam . as the resolution of the telescope / instrument beam increases ,
the measurements of point source fluxes are less prone to the fluctuation noises .
thus , fluctuation noise needs to be more carefully treated for accurate extraction of sources and flux measurements [ see jeong et al .
( 2005 , 2006 ) for realistic estimations of these noises for the akari mission ] . from observations of sources for those with iras fluxes made during the performance verification ( pv ) phase of the akari
, we investigate how the resolved background near a point source can affect the measurement of fluxes by comparing the measured fluxes .
this paper is organized as follows . in section [ sec : ass_akari ] , we describe the data and the data reduction .
we present the comparison of our results between iras flux and akari flux in section [ sec : comp_flux ] .
the probable detections of new akari sources are explained in section [ sec : new_akari ] .
we summarize our results in section [ sec : summary ] .
the data used in this work are from akari all - sky survey observations performed during the two week - long pv phase from april 24th 2006 to may 7th 2006 .
the observations cover a range in galactic latitude between -50@xmath3 to 50@xmath3 as shown in figure [ fig : scanmap_pv ] .
( 90 mm , 80mm)fig01a.ps ( 90 mm , 80mm)fig01b.ps for the data processing , we have used the latest pipeline developed for the all - sky survey @xcite .
the source extraction was performed only for the akari wide - s band ( 90@xmath0 m ) with the dedicated software for the akari survey . on the purpose of calibrating the absolute flux ,
the observations of asteroids , stars and galaxies with a wide range of fluxes were used , and the uncertainty of flux calibration is expected to be less than 20% for wide - s band @xcite . in order to avoid false detections , we removed any detections near the south atlantic anomaly ( saa ) region , reset or calibration sequences and the edge of a detector array . among the extracted sources , we have selected reliable sources with the following criteria to eliminate false detections : ( 1 ) signal to noise ratio ( s / n ) greater than 15 , ( 2 ) detected in all rows of the detector array , e.g. , 3 rows of detector pixels for the wide - s band , and ( 3 ) confirmed in at least two more orbit ( @xmath4 100 min . per 1 orbit for akari ) ( 4 ) the galactic plane region is excluded as it is observed in a special detector readout mode that needs separate processing [ for the sampling mode of akari observation , see @xcite for detailed information ] . after selecting reliable sources
, we attempted to match them with iras sources within a positional error circle of 100@xmath5 .
the total number of sources in our final selected sample is 150 .
we showed the positional difference between akari and iras sources with a good flux quality in figure [ fig : pos_diff_det ] .
most of positions for iras sources are well consistent with those estimated in akari within a resolution range ( 30@xmath5 ) . the current position information is determined based on the sensors of the onboard attitude and orbit control system ( aocs ) , reprocessed on the ground .
the major source of uncertainty is the alignment between the aocs sensors and the telescope , and its time variation . in the future
, position information will be improved with the pointing reconstruction processing using the data from the focal - plane instruments .
the accuracy of the detected source position will be as good as a few arcsec , which enable us to carry out more detailed analysis of the ` iras ' positions .
figure [ fig : pos_det_el ] shows the distribution of these sources in ecliptic coordinates .
( 90 mm , 90mm)fig02.ps ( 130 mm , 80mm)fig03.ps
the reliably detected sources are used to compare the iras and akari fluxes .
the effective wavelength of the akari wide - s band is 90@xmath0 m while iras has 60@xmath0 m and 100@xmath0 m bands in the far - ir region . in order to make a consistent comparison between the akari and iras fluxes , we computed the iras flux at 90@xmath0 m by fitting the 60 and 100 @xmath0 m data to a black - body curve . since the flux densities of iras point sources listed in the catalogue are non - color corrected ones , we attempted an iterative fitting method which performs successively fitting and color corrections .
we also applied the color correction to the akari flux densities .
figure [ fig : comp_flux ] shows the comparison between the akari and iras fluxes of our detected sources at 90@xmath0 m .
we also plot the error bars for the iras and akari fluxes .
note that a flux error for an akari flux was obtained from the standard deviation of the extracted fluxes for orbit - confirmed detections while the flux error for the iras sources with a bad flux quality was not listed in iras point source catalogue .
the straight line shows the perfect correlation between these two fluxes .
the size of the circle is proportional to the estimated signal - to - noise ratio .
this figure shows that the data lies below the envelope of @xmath6 .
this means that the akari fluxes are in general similar to or smaller than the iras fluxes .
( 130 mm , 130mm)fig04.ps such a trend is more clearly shown in figure [ fig : comp_flux_ratio ] which displays the flux ratio of iras to akari detections .
we checked the quality of the iras flux , by examining the quality flag for the deviating sources .
most of those sources indeed have a moderate or a bad flux quality and are often located near the galactic plane or strong star formation regions ( e.g. , large magellanic cloud ) .
thus , due to a strong background , the flux of iras sources with moderate or bad flux quality flags may not have been accurately estimated .
we also found that the flux is over or underestimated even for iras sources with a high flux quality .
the scattering trend in flux ratio shows a triangular shape ( peaking at 80 jy ) in the upper panel of figure [ fig : comp_flux_ratio ] and a right triangular shape in the lower panel of figure [ fig : comp_flux_ratio ] , respectively . the sensitivity of akari in the low flux range is better than that of iras , which explains that the scatter in the lower panel of figure [ fig : comp_flux_ratio ] is larger for low fluxes than for high fluxes .
( 120 mm , 120mm)fig05a.ps ( 120 mm , 120mm)fig05b.ps the background fluctuations from the surface brightness of extended structure , can easily be mistaken for genuine point sources . since our detections are for bright sources , the contribution of the fluctuation noise will be primarily from sky confusion ( rather than point sources ) .
some authors have estimated the sky confusion from background fluctuations for the akari mission based upon iras and iso data @xcite . in order to check any dependency of the flux uncertainty on the background level , we plot flux ratios as a function of the background emission in figure [ fig : comp_flux_bgr ] .
we assume that the fluctuations from the background are mainly from cirrus emission .
the fluctuations from the zodiacal light is expected to be small at the resolution of the iras mission . for the estimation of the cirrus emission in the akari band , we used the all - sky 100 @xmath0 m dust map generated from the iras and cobe data by schlegel , finkbeiner , and davis ( 1998 ) and the dust model by finkbeiner et al .
( 1999 ) . as shown in the upper panel of figure [ fig : comp_flux_bgr ] , the scatter increases around the 20 mjy / sr level , with an underlying monotonic increase from 20 to 200 mjy / sr .
owing to its higher resolution , akari can resolve the background more effectively .
since the estimated background for iras depends on the relatively larger spatial structure of the background compared to the diffracted beam size [ the source extraction for iras used an annulus of radius 5@xmath7@xmath27@xmath7 in measuring the background @xcite ] , it could not be measured properly on a intensively fluctuating background .
therefore , this trend shows that the local background at least below 5@xmath7 near a source , rather than the global background , is an important factor in the flux estimation .
note that previous estimations of sky confusion @xcite suggested that the fluctuation noise is strongly dependent on the background level with the assumption that the spatial power spectrum from fluctuations of the dust emission has a simple power - law with the power index of @xmath4 - 3 @xcite .
in addition , we have also made a simple estimation of the contamination of the flux by summing the background flux over a detector pixel .
we plot the case for a 10 jy iras source contaminated by a 10 percent background ( see the dotted line in figure [ fig : comp_flux_bgr ] ) .
we found that the flux measurements for some iras sources might be contaminated by the background .
if we plot akari sources matched with iras sources with only high flux quality flags ( see the lower panel of figure [ fig : comp_flux_bgr ] ) , the scatter is reduced , however , visible scatter still remains and slightly increases as the cirrus background becomes larger .
thus , iras flux measurements may have errors even for correctly measured iras sources with a high flux quality . for a more systematic study to confirm this result
, we need to use samples from a wider survey area .
( 120 mm , 120mm)fig06a.ps ( 120 mm , 120mm)fig06b.ps
in this work , we have used only sources matched with iras point source catalogue .
however , we also detected new sources from the akari all - sky survey which are not present in the iras catalogue .
there are many possible candidates for these new akari sources .
first , one iras source may have been resolved into two sources ( see one example shown in figure [ fig : det_new_akari ] ) .
second , some sources may be resolved from the background .
third , the source may be a solar system object , e.g. , asteroid .
fourth , some detections may be false due to the detector characteristics .
on the other hand , there may be false detections from iras .
thus , we expect that the akari all - sky survey can find new extragalactic point sources as well as faint galactic point sources .
( 70 mm , 70mm)fig07a.ps ( 70 mm , 70mm)fig07b.ps ( 70 mm , 70mm)fig07c.ps
by using the akari all - sky survey data taken during the pv phase , we tried to extract reliably detected sources and compared their fluxes with associations in the iras point source catalogue .
the positions of iras sources are mostly well consistent with those estimated from akari .
we found that the original flux measurements of some iras sources appear to have been overestimated or underestimated .
we also confirmed that most of the iras sources with a moderate or a bad quality flux flag in the iras catalogue were quite overestimated . to investigate any possible contamination of the iras source flux from the background in the estimation of the flux
, we checked the background level on each detection of the iras sources .
the difference in the flux between the iras and akari associations was not found to be strongly dependent on the background level .
however , it is suggested that the flux measurement may have been affected by the local background rather than the global background . note that akari can resolve the background and source more effectively than iras due to the higher resolution of the akari observations . in addition
, we found that in some cases , the flux measurements even for correctly measured ( high flux quality ) iras sources may include errors of up to @xmath45 , even though we consider the uncertainty of flux calibration .
however , due to various detector characteristics , there may be false detections as well as detections with inaccurately estimated flux though we have removed false detections and selected reliable detections as described in section [ sec : ass_akari ] .
the akari all - sky survey is expected both to verify the iras sources and find new extragalactic or galactic point sources .
the preliminary catalogue for reliably detected point sources will be released in a forthcoming paper .
the akari project is an infrared mission of the japan space exploration agency ( jaxa ) institute of space and astronautical science ( isas ) , and is carried out with the participation of mainly the following institutes ; nagoya university , the university of tokyo , national astronomical observatory japan , the european space agency ( esa ) , imperial college london , university of sussex , the open university ( uk ) , university of groningen / sron ( the netherlands ) , seoul national university ( korea ) .
the far - infrared detectors were developed under collaboration with the national institute of information and communications technology .
we would like to thank the referee l. viktor t@xmath8th for a careful reading of our paper and many fruitful comments .
jeong acknowledges a japan society for the promotion of science ( jsps ) fellowship to japan .
we thank myungshin i m and bon - chul koo for helpful suggestions . | we present the spatially resolved observations of iras sources from the japanese infrared astronomy satellite akari all - sky survey during the performance verification ( pv ) phase of the mission .
we extracted reliable point sources matched with iras point source catalogue . by comparing iras and akari fluxes
, we found that the flux measurements of some iras sources could have been over or underestimated and affected by the local background rather than the global background .
we also found possible candidates for new akari sources and confirmed that akari observations resolved iras sources into multiple sources .
all - sky survey observations are expected to verify the accuracies of iras flux measurements and to find new extragalactic point sources . |
in recent years , there have been a number of theoretical and experimental studies of the manner in which coulomb blockade effects in a metallic particle or semiconductor quantum dot disappear when the conducting island becomes more electrically connected to its environment .
here we focus on a system in which two symmetric quantum dots are defined by applying negative voltages to gate electrodes that lie on the surface of a semiconductor heterostructure above a two - dimensional electron gas ( 2deg ) .
we assume that the dots are joined by a quantum point contact containing a single orbital channel that is almost perfectly transmitting at the fermi energy , but that the dots are isolated from their respective leads by comparatively large tunnel barriers . in this geometry ,
information about the coulomb blockade energy @xmath0 involved in the transfer of electrons from one dot to the other can be obtained by observing the positions of coulomb blockade peaks in the conductance across the entire system ( from one lead to another ) when that conductance is plotted as a function of voltages on gates coupled to each of the dots .
@xcite previous analyses of the disappearance of the coulomb blockade for two - dot systems @xcite containing such a partially open point contact have characterized the contact by a number @xmath2 of degenerate one - dimensional ( 1d ) channels and by the dimensionless conductance @xmath3 of each channel ( where @xmath10 ) .
@xcite this characterization is complete in the limit where the electronic transmission amplitude through the contact is independent of energy for electron energies that differ from the fermi energy by no more than an amount comparable to @xmath0 .
the energy independence of the transmission amplitude means that this characterization corresponds to a potential barrier which is sufficiently thin that it can be modeled as a delta function .
because the coulomb blockade energy @xmath0 is much smaller than the fermi energy @xmath11 , working in this `` delta - function barrier limit '' can yield good results even though the product of the barrier width and the fermi wave vector @xmath12 is generally much greater than @xmath8 ( with @xmath12 being the value of the fermi wave vector in the 2deg far from the barrier ) .
nevertheless , it seems important to investigate further the consequences of relaxing the assumption of a delta - function barrier limit .
for one thing , it is possible to generate wider barriers using an appropriate gate geometry , and one would like to understand at what point the delta - function - limit calculations break down . secondly , estimating the corrections due to a barrier s finite thickness provides a valuable check on the delta - function - limit results .
another reason for interest is that recent experiments on transmission through quantum point contacts have shown unexpected structure ( e.g. , an apparent conductance plateau near @xmath13 at intermediate temperatures
@xcite ) whose origin is only poorly understood .
such results suggest that transmission through a point contact may have a nontrivial energy dependence , and such a dependence could well have an effect on the breakdown of the coulomb blockade in the coupled - dot geometry . with the aim of explaining such effects in mind , it is a helpful first step to study theoretically the effects of energy dependences in a simpler situation where the barrier potential is known and many - body effects are reduced to a bare minimum . in this vein
we consider here a model in which the two dots are separated by a parabolic barrier of nonzero width and the electron - electron interaction is taken to be constant for any two electrons located anywhere on the same quantum dot .
this model presents challenges for `` bosonization techniques '' that characterize point - contact constrictions as one - dimensional fermionic seas whose low - energy degrees of freedom can be expressed in terms of bosonic density and phase variables .
@xcite in such `` bosonized '' models , the behavior of incompletely opened channels is commonly studied by introducing a zero - width delta - function barrier at a specific point in the one - dimensional sea .
@xcite by way of contrast , this paper seeks to show how a bosonized model of a one - dimensional system can describe the single - particle effects of replacing a delta - function barrier with a barrier that more realistically corresponds to a finite - length constriction i.e . , a barrier of nonzero width and therefore nontrivial single - particle transmission properties .
before proceeding , we should describe more fully what is already understood about systems of two symmetric quantum dots connected by a single orbital channel containing degenerate spin modes .
in such systems , an energy scale @xmath0 characterizes the energy cost of moving electronic charges between different segments of the system .
moreover , the system s coulomb blockade behavior reflects the presence of an energy term proportional to @xmath0 that oscillates between one value held when the total number of electrons on the dots is even and another value held when the total electron number is odd . @xcite previous work , both experimental and theoretical , has shown that as the conductance of the interdot point contact is increased , the energy scale associated with this even - odd alternation is reduced by a factor of @xmath1 , where @xmath14 goes to @xmath8 when the channel is fully open . when the transmission coefficient for electronic motion within this channel is independent of energy ( i.e. , the delta - function barrier limit ) , the factor @xmath1 takes on a universal value determined solely by the number of point contact channels @xmath2 ( the channels being assumed degenerate and uncoupled ) and the dimensionless conductance @xmath3 of each individual channel ( if @xmath15 is the total conductance of the point contact , @xmath16 . ) .
we refer to the value @xmath14 in the factor @xmath1 as the _ fractional peak splitting _ because it has been measured by observing the relative separation of conductance peaks in a series of coulomb blockade experiments .
this paper studies the corrections to @xmath14 , and therefore to @xmath1 , that result when the assumption of a delta - function barrier limit is relaxed in other words , when the interdot barrier is more realistically treated as having a finite height and nonzero width .
a prior study of the effects of a nonzero - width barrier concentrated on the limit of weakly coupled dots ( the limit @xmath17 ) .
@xcite this study revealed that , if the one - dimensional channels between two dots , or between a dot and a lead , contain a tunneling barrier of finite height @xmath18 ( in energy units ) and of nonzero width @xmath19 , the behavior of such systems is responsive to another energy scale , @xmath20 , that characterizes the energy range ( for electrons incident on the barrier ) over which the probability of transmission through the barrier varies substantially .
@xcite for relatively large and shallow dots , such as those that have been constructed in gaas / algaas heterostructures at low temperatures , @xcite the energies @xmath0 and @xmath20 tend to be much smaller than the fermi energy @xmath11 but much larger than both the single - particle level spacing @xmath21 and the thermal energy @xmath22 . as a result
, such systems are characterized by the following hierarchy of energy scales : @xcite @xmath23 because @xmath20 is comparable in size to the intermediate energy scale @xmath0 ( which acts as a kind of `` excitation energy scale '' with respect to independent - particle energies ) , effects from the barrier s finite size can be significant and deserve investigation before they are confidently discarded . our prior study of the weak - coupling limit where @xmath24 found that the finiteness of the barrier leads to an upward correction to the universal @xmath14-versus-@xmath3 curve for a delta - function barrier . in other words , for a given _ small _ value of @xmath3 , the fractional peak splitting @xmath14 is enhanced relative to its value for a delta - function barrier . @xcite this enhancement occurs because , in a channel containing a barrier with a finite energy height ( as opposed to a delta - function barrier ) , electrons can tunnel from one dot to another through largely unreflected states that have single - particle energies greater than the energy that corresponds to the finite barrier s peak .
@xcite in order to study the correction to the universal @xmath14-versus-@xmath3 curve , in our prior work we made several basic assumptions about the system of two symmetric dots connected by a single orbital channel .
first , we assumed that the electrons that enter the point contact between the dots are in the lowest energy eigenstate for motion perpendicular to the channel and can therefore be described by a one - dimensional ( 1d ) schrdinger equation with an effective barrier potential @xmath25 .
we assumed that @xmath25 could be treated as parabolic , with a harmonic oscillator frequency @xmath5 associated with the inverted parabolic well .
( in this case , the transmission energy scale @xmath20 is given by @xmath26 . ) in addition , we assumed that , despite the nonzero width of the barrier , the essential nature of electron - electron interactions was still accurately represented by the standard coulomb blockade model , in which the interaction energy of a quantum dot is analogous to that of a classical capacitor
i.e . , proportional to the charge on the dot squared .
having made these assumptions , we found that , in the limit @xmath17 , the leading behavior of the enhancement to @xmath14 is roughly proportional to @xmath27 .
for the experimental systems with which we are concerned , where @xmath2 is @xmath8 or @xmath28 and where @xmath29 , the size of the enhancement is relatively small .
@xcite this paper extends study of such finite - barrier corrections to the strong - coupling limit where @xmath30 . in doing so
, we repeat the two assumptions described above .
the assumption of a parabolic form for the imposed barrier potential is likely as good as before .
@xcite however , our simplistic treatment of the electron - electron interactions is likely less firm , given that the point contact is now largely open and therefore much more likely to be occupied by electrons interacting in a way not reflected by the standard capacitive coulomb blockade model . nonetheless , because the barrier region is still small in relation to the larger conducting basins of the two dots , and because the effects from a combination of the coulomb blockade model and the interdot barrier s finite size are themselves sufficiently interesting and complex , in this initial stab at the problem of finite barriers we ignore the effects of interactions specific to the point contact , with the understanding that separate efforts to understand those effects should follow .
thus , the basic parameters for our strong - coupling study are the same as those for the weak - coupling limit .
the @xmath2 degenerate spin channels are assumed to contain effectively identical barriers , and the interdot barrier can be modeled as fully parabolic so long as we are concerned only with energies within a restricted range about the barrier peak .
we therefore describe the single interdot orbital channel as containing the potential : @xmath31 see figure 1 .
as we will later see , our ultimate results are independent of the potential s details away from the barrier peak .
key aspects of the barrier potential are the `` depth '' ( @xmath32 ) of its peak relative to the fermi surface and the harmonic oscillator frequency @xmath33 associated with the inverted parabolic well .
@xcite ( note that @xmath34 , which is assumed to be much less than @xmath8 . )
we use the single - particle eigenfunctions that correspond to the above potential to calculate corrections to @xmath14 . to do so , we incorporate these eigenfunctions in a generalization of a bosonization approach that was used earlier in the delta - function barrier limit .
we first find the change in the two - dot system s ground state energy as the value of @xmath35 is increased from zero .
this energy shift @xmath36 is a function of @xmath35 and another dimensionless quantity @xmath37 , which represents a linear combination of the gate voltages applied to the two dots .
@xcite when the total number of electrons on the two dots is even , the fractional peak splitting @xmath14 is given by the following formula : @xmath38 by applying this formula , we find that , consistent with our earlier conjecture , @xcite for both @xmath39 and @xmath40 the leading finite - barrier corrections _ decrease _ the value of @xmath14 observed for a given value of @xmath3 when @xmath3 is close to @xmath8 .
for @xmath41 and @xmath30 , the magnitude of the decrease is proportional to @xmath42 for @xmath6 and to @xmath43 for @xmath7 .
= 0.9 + in sec .
ii , we show how to use the bosonization approach to derive an effective action that captures the coulomb blockade behavior of such a system for an arbitrary number @xmath2 of interdot tunneling channels . in sec .
iii , we use the effective action of sec .
ii to solve for the leading @xmath14-versus-@xmath3 behavior for @xmath6 and @xmath44 . in sec .
iv , we do the same for @xmath7 . in sec .
v , we remark on the significance and limitations of our results .
to calculate the corrections due to the interdot barrier s nonzero width , we start with a fermionic formulation of our system of two symmetric quantum dots connected by a single orbital channel containing an arbitrary number @xmath2 of degenerate spin modes .
since this fermionic problem is effectively one - dimensional , @xcite we can apply to it a variation of the standard technique of bosonization .
@xcite having `` bosonized '' the problem , we integrate out various bosonic degrees of freedom to obtain a low - energy effective action involving only a single set of scalar fields .
the fermionic formulation of the problem of coupled quantum dots is quite straightforward . in a basis of non - interacting single - particle eigenstates ,
the fermionic hamiltonian consists of a kinetic - energy part @xmath45 and a charging - energy part @xmath46 : @xmath47 where @xmath48 is the energy of the single - particle eigenstate ; @xmath49 is the channel index ; @xmath50 is the index for states incident on the barrier from the left ( @xmath51 ) and the right ( @xmath52 ) , respectively ; and @xmath53 is the magnitude of the wave vector associated with motion along the one - dimensional channel ( @xmath12 is the fermi wave vector and @xmath54 is assumed to be smaller than @xmath12 ) .
the operator @xmath55 measures the difference between the number of electrons to the right of the barrier and the number of electrons to the left of the barrier .
if one lets @xmath56 be the occupation number for the left dot and @xmath57 be the occupation number for the right dot , @xmath58 . in order to determine the value of @xmath14 via eq .
[ eq : fgeneral ] , we consider the case where the total number of electrons on the two dots is even , so that @xmath55 is an operator with integer eigenvalues .
( when the total number of electrons is odd , @xmath55 has half - integer eigenvalues . )
the operator @xmath55 can be written in terms of fermionic position operators @xmath59 and @xmath60 as follows : @xmath61 \ , \psi_{\sigma}^{\dagger}(x ) \psi_{\sigma}(x ) \
, , \label{eq : nhat2}\ ] ] where @xmath60 consists of left - incident and right - incident parts , @xmath62 and where @xmath63 is a heaviside step function centered on the dividing line between the `` right '' and `` left '' sides of the barrier at @xmath64 .
if one were bosonizing fermionic operators associated with a basis of plane - wave states , one could bosonize the above hamiltonian by using the standard formula : @xmath65 } \ , \ , , \label{eq : bosplane1}\ ] ] where @xmath66 and @xmath67 are scalar boson fields and @xmath68 is a quantity analogous to a lattice spacing that is used to impose an ultraviolet cutoff at energies approximating @xmath69 on the bosonic degrees of freedom @xcite ( @xmath68 is ultimately meant to be taken to zero a limit that can be reasonably taken when @xmath70 and the approach to modeling the double - dot system can be supposed to work down to length scales @xmath71 , where @xmath72 is much greater than the relevant excitation scale @xmath0 ) .
the annihilation operators associated with the plane - wave states would then be given by the equation : @xmath73 } \ , \ , .
\label{eq : bosplane2}\ ] ] in terms of this plane - wave bosonization scheme , the operator @xmath55 could be simply expressed in terms of the boson fields @xcite : @xmath74 unfortunately , our problem is not so simple . the basis of single - particle states that appears in eq .
[ eq : ferm1 ] is not a plane - wave basis .
the eigenstates are those of a one - dimensional system containing a barrier of finite height and nonzero width , and they therefore include incident , reflected , and transmitted parts . happily , the kinetic - energy part of the hamiltonian @xmath45 does not concern itself with such complications .
so long as the single - particle level spacing can be treated as effectively zero , we can proceed with regard to @xmath45 just as if the single - particle states were plane waves . accordingly , if we focus on bosonizing @xmath45 , we can bosonize the annihilation operators in eq .
[ eq : ferm1 ] in the same way as in eq .
[ eq : bosplane2 ] . under this bosonization scheme , the kinetic energy @xmath45 has the following standard form : @xmath75 ^ 2 + \left [ \partial_{x } \theta_{\sigma } ( x ) \right]^2 \right\}. \label{eq : boskin}\ ] ] our next task is to find a proper bosonized form for the position operators @xmath76 that appear in eq .
[ eq : nhat2 ] . recalling that we already have bosonized forms for the operators @xmath77
, we express the @xmath76 in bosonized form by using the fact that the position operators consist of linear combinations of the operators @xmath77 . in particular , in the continuum limit , @xmath78 where the functions @xmath79 are the single - particle eigenfunctions of the non - interacting one - dimensional system .
@xcite for the potential described by eq .
[ eq : vfunc ] , we have derived these @xmath79 in prior work , using the fact that there are exact solutions ( parabolic cylinder functions ) to the schrdinger equation for a single particle in a parabolic potential .
@xcite using the @xmath79 , we find that , to leading order in the perturbation due to the barrier , @xmath55 equals @xmath80 , where @xmath81 } } { k_2 - k_1 + i \eta } \ , a_{k_1 \sigma 1}^{\dagger } a_{k_2 \sigma , -1 } \ +
\ \text{h.c . }
\right\ } \ , , \label{eq : nparts}\end{aligned}\ ] ] where `` c.c . '' and `` h.c . ''
stand , respectively , for the complex conjugate and hermitian conjugate of the preceding term , and the following identities hold : @xmath82 with @xmath83 being a constant independent of @xmath84 and @xmath85 being the value of @xmath84 at the fermi energy .
@xcite in the above equations , the dimensionless variables @xmath86 measure the energy distance from the finite barrier s peak .
@xcite the scale for these dimensionless energy measures is the harmonic oscillator energy @xmath87 , and in terms of these dimensionless measures @xmath88 the functions @xmath89 and @xmath90 are associated with the single - particle transmission properties of the barrier .
each additive term in @xmath89 is a product of transmission and reflection amplitudes , reflecting the fact that the leading contributions to @xmath91 come from `` overlaps '' between the transmitted part of an original right - mover and the reflected part of an original left - mover and vice versa ) .
meanwhile , @xmath90 represents a phase associated with the scattering of a single - particle wavefunction .
because of the exponential growth ( or damping ) of @xmath89 , we should not be surprised to find that @xmath89 plays a starring role in the nature of our leading - order result whereas , to first approximation , contributions from @xmath90 will prove to be negligible . before we proceed further , we should make three remarks about eq .
[ eq : nparts ] .
first , we should note that derivation of the above equation for @xmath91 ignores contributions to @xmath92 from integrating over some interior portion of the barrier region in which approximating the single - particle wavefunctions by plane waves , or at least by wkb approximations to plane waves , is no longer valid . hence because the exact eigenfunctions in the barrier region ( parabolic cylinder functions ) do not `` explode '' in amplitude as one goes further toward the center of the barrier @xcite as in our derivation of analogous identities in the limit of weakly coupled dots , @xcite we expect additive corrections of rough order @xmath19 to the eq .
[ eq : nparts ] cofactors @xmath93 . consequently ,
to avoid substantial corrections to eq .
[ eq : nparts ] , we must be able to restrict our attention to wave - vectors @xmath94 ( measured from the appropriate fermi wave vector in one dimension ) such that @xmath95 because the exponentially variant function @xmath89 strongly favors values of @xmath96 that correspond to @xmath97 near @xmath98 , the @xmath96 of most interest for calculating corrections to the universal behavior are those for which @xmath99 .
the above restriction on @xmath96 can thus be seen to mean that we need @xmath100 .
for systems ( such as those this paper seeks to model ) in which @xmath101 , the restriction then reduces ( conservatively speaking ) to @xmath102 or , equivalently , @xmath103 . given that the relevant experiments with quantum dot systems @xcite have generally not resolved @xmath3 to increments of @xmath104 , this constraint is not a serious one . as a second point regarding eq .
[ eq : nparts ] , it should be noted that , for @xmath102 , this intermediate result ( and therefore those that follow from it ) is insensitive to details of the barrier away from the barrier peak ( e.g. , the slope discontinuities that occur in our barrier potential at @xmath105 ) so long as @xmath19 is large compared to the inverse fermi wave vector @xmath106 ( with the fermi wavelength itself being of the order of @xmath107 , the wavelength associated with the peak barrier energy , @xmath108 ) .
the condition @xmath109 is typically true of the kinds of quantum dots that have motivated these investigations .
@xcite a third remark about eq .
[ eq : nparts ] is that in deriving @xmath91 , we have ignored a correction involving terms of the form @xmath110 - 1)/(k_2 - k_1 \pm i \eta)$ ] .
we ignored similar `` phase - based '' terms in ref .
the basic reason that we ignore these terms here derives from the fact that , because the characteristic excitation energy @xmath0 is less than @xmath87 , the variation of the numerator of these terms over any region of particular interest should be very small . as a result , in the regions that we generally find to be most important ( those regions near the simple poles where @xmath111 and the magnitudes of other contributing terms tend toward infinity ) , these phase - based terms go to zero
thus , we have good reason to expect that the contribution from these phase - based terms is small compared to the contribution from the terms we keep . indeed ,
if we follow the calculation of the lowest - order contribution from these terms further , we would find that , by reasoning parallel to that used in sec .
iii to perform various wave vector integrals , these terms produce a contour integral with no contribution from residues of the simple poles at @xmath111 and only negligible , higher - order contributions from integration around other singularities and branch cuts . returning to eq .
[ eq : nparts ] and the process of determining the fractional peak splitting , we observe that if the operators @xmath77 were annihilation operators for particles in plane - wave states ( e.g. , if there were no interdot barrier ) , @xmath55 would simply equal @xmath112 . in other words
, @xmath112 bears the same relation to the bosonic fields associated with the barrier - state operators @xmath77 that @xmath55 bears to the bosonic fields associated with the similarly indexed plane - wave operators of eq .
[ eq : bosplane2 ] . from eq .
[ eq : nplane ] , it therefore follows that @xmath113 we do not have a similarly simple formula for @xmath91 . substituting for the operators @xmath77
, we find that @xmath114 } e^{i ( k_1 x_1 + k_2 x_2 ) } } { k_2 - k_1 + i \eta } e^{-i \sqrt{\pi } [ \theta_{\sigma } ( x_1 ) + \phi_{\sigma } ( x_1 ) ] } \right .
\nonumber \\ & & \hspace{0.2 in } \left .
\times \ , e^{-i \sqrt{\pi } [ \theta_{\sigma } ( x_2 ) - \phi_{\sigma } ( x_2 ) ] } \ \ \ + \ \ \ \text{h.c . } \right\ } \label{eq : deltn}\end{aligned}\ ] ] having now obtained bosonized expressions for the separate terms @xmath92 and
@xmath91 , we express the hamiltonian as a whole in bosonized form : @xmath115 @xmath45 is given by eq .
[ eq : boskin ] , and the remaining terms are as follows : @xmath116^{2 } \ , \nonumber \\
h_{c}^{(1 ) } & = & u_{\rho } \ , \ , \delta \hat{n } \left [ \frac{1}{\sqrt{\pi } } \sum_{\sigma } \theta_{\sigma } ( 0 )
- \rho/2 \right ] \ , + \
, \text{h.c . } \ , \nonumber \\
h_{c}^{(2 ) } & = & u_{\rho } \ , ( \delta \hat{n})^2 \ , .
\label{eq : bosham1}\end{aligned}\ ] ] the @xmath117 term can be dropped from the leading - order calculation because , to second - order in @xmath91 ( which due to @xmath118 is itself roughly proportional to @xmath119 at the fermi surface ) , @xmath117 contributes nothing to the @xmath37-dependence of the ground - state energy .
recall from eq .
[ eq : fgeneral ] that it is the @xmath37-dependence of the ground - state energy that provides the basis for calculation of the fractional peak - splitting @xmath14 . to simplify calculation of this @xmath37-dependence
, we shift the @xmath120-fields by a term linear in @xmath37 : @xmath121 .
this transformation leaves us with a hamiltonian in which the @xmath37-dependence appears only in the perturbative factors of @xmath91 .
the transformed ( and truncated ) hamiltonian has the following form : @xmath122 ^ 2 + \left [ \partial_{x } \theta_{\sigma } ( x ) \right]^2 \right\ } \
, \nonumber \\
h_{c}^{(0 ) } & = & \frac{u_{\rho}}{\pi } \left [ \sum_{\sigma } \theta_{\sigma } ( 0 ) \right]^{2 } \ , \nonumber \\
h_{c}^{(1 ) } & = & \frac{u_{\rho}}{\sqrt{\pi } } \sum_{\sigma } \left [ \delta \hat{n } \ , \theta_{\sigma}(0 ) \ , + \ , \theta_{\sigma}(0 ) \ , \delta \hat{n } \right ] \ , , \label{eq : bosham2}\end{aligned}\ ] ] where @xmath91 is now given by the identity : @xmath123 } e^{i ( k_1 x_1 + k_2 x_2 ) } } { k_2 - k_1 + i \eta } \ , e^{-i \pi \rho / n_{\text{ch } } } \ , \right . \nonumber \\ & & \hspace{0.2 in } \times \ , e^{-i \sqrt{\pi } [ \theta_{\sigma } ( x_1 ) + \phi_{\sigma } ( x_1 ) ] } \ , \nonumber \\ & & \hspace{0.2 in } \left .
\times \ , e^{-i \sqrt{\pi } [ \theta_{\sigma } ( x_2 ) - \phi_{\sigma } ( x_2 ) ] } \ \ \ + \ \ \ \text{h.c . } \right\ } \label{eq : nhat2b}\end{aligned}\ ] ] with eq .
[ eq : nhat2b ] , we have reached the end of the road with regard to the hamiltonian approach . to progress further , we shift to an action - based , path - integral formulation . for ease of notation , we will henceforth assume that we have chosen units in which @xmath124 and drop @xmath125 from subsequent intermediate calculations ( although we resuscitate it in stating our final results ) .
having made this choice of units , we proceed by integrating out various degrees of freedom : first , the @xmath126-field degrees of freedom , and second , the @xmath127-field degrees of freedom away from @xmath64 .
the result is an effective action in terms of fields @xmath128 that are equivalent to the @xmath127-field degrees of freedom at @xmath64 .
this effective action , dependent only on the @xmath64 degrees of freedom for the original @xmath127-fields , is analogous , though significantly not equivalent , to that previously obtained for a system containing a delta - function barrier ( see refs . 1 and 2 ) .
we do no more than outline the process of integrating out the @xmath126-fields and @xmath127-fields because the methodology for doing so is fairly straightforward . with regard to integrating out the @xmath126-fields ,
a key point is that , in the path - integral approach , a lagrangian term linear in @xmath129 appears in the action ( just as a @xmath130-term appears in going from the hamiltonian for a single particle to the corresponding lagrangian ) .
having accounted for this term , we eliminate the @xmath126-fields by performing a standard gaussian integration .
the result is an effective action entirely in terms of @xmath127-fields .
having obtained this effective action , we introduce the field @xmath131 and the auxiliary fields @xmath132 .
we use the @xmath133-fields to enforce the identity @xmath134 by adding to the action the following term : @xmath135 \ , .
\label{eq : deltfact}\ ] ] having added @xmath136 to the action , we can substitute @xmath131 for @xmath137 in @xmath138 ( recall eq .
[ eq : bosham2 ] ) .
our next steps are to integrate out the @xmath139-fields and then the @xmath132-fields .
once again , only straightforward albeit tedious
gaussian integrations are required . to leading order in the perturbative factor @xmath91
, the result is an effective action dependent only on the scalar fields @xmath128 and their fourier transforms @xmath140 : @xmath141 } e^{-i ( k_1 x_1 + k_2 x_2 ) } } { k_2 - k_1 - i \eta } \ , \ , \right .
\nonumber \\ & & \hspace{0.5 in } \times \ , \theta_{\sigma_2}^{(0 ) } ( \tau ) \ , \ , e^{-i \int \ !
d\omega \ , h_2(x_1,x_2,\omega ) \ , e^{-i \omega \tau } \
, \tilde{\theta}_{\sigma_1}^{(0)}(\omega ) } \nonumber \\ & & \left . \hspace{0.4 in } + \ \text{m.t . } \right\ } \ , , \label{eq : bosact}\end{aligned}\ ] ] where the `` mirror term '' ( m.t . ) can be obtained by complex conjugating both the factors that precede @xmath142 and the exponential prefactor @xmath143 that precedes the integral over @xmath144 , and where the functions @xmath145 , @xmath146 , @xmath147 , and @xmath148 are given by the following formulae : @xmath149^{1/2 } \nonumber \\ & & \hspace{0.0 in } \times \ , \left [ \frac { ( |x_1| + \alpha)(|x_2| + \alpha ) } { \alpha ( |x_1| + |x_2| + \alpha ) } \right]^ { \text{sgn } ( x_1 ) \text{sgn } ( x_2 ) /2 } \
\nonumber \\ h_2(x_1 , x_2 , \omega ) & = & h_1(x_1,x_2,\omega ) - h_0(x_1,x_2,\omega ) \ ,
\nonumber \\ h_1(x_1 , x_2 , \omega ) & = & \text{sgn}(\omega ) \ , \frac{\text{sgn } ( x_2 ) \left [ 1 - e^{-|\omega x_2|/v_f } \right ] } { 2 \sqrt{\pi } }
\ \nonumber \\ & & \hspace{0.3 in } - \ \ { x_1 \leftrightarrow x_2 \ } \nonumber \\ h_0(x_1 , x_2 , \omega ) & = & \frac{e^{-|\omega x_1|/v_f } \ ,
+ \ , e^{-|\omega x_2|/v_f } } { 2 \sqrt{\pi } } \ , \label{eq : afunc}\end{aligned}\ ] ] where `` @xmath150 '' is the same as the preceding term , except with the roles of @xmath151 and @xmath152 reversed . the effective action of eq .
[ eq : bosact ] is the end result for this section .
as promised , this action depends only on a single set of scalar fields , which are equivalent to the original @xmath127-fields at @xmath64 .
sections iii and iv use this action to solve for the fractional peak splitting @xmath14 in systems with one and two interdot channels , respectively . in both sections ,
the general approach is to treat @xmath153 as the unperturbed action and to solve perturbatively in @xmath154 .
here we consider the single - channel case ( @xmath6 ) , in which electrons of only one spin have enough energy to penetrate into the point contact . starting with the unperturbed action @xmath153 , it is not hard to find that @xmath155 we use this identity to calculate the first - order correction to the ground state energy : @xmath156 .
in particular , we find the following : @xmath157^{2 } \ , \frac{e^{- \alpha |\omega|/v_{f}}}{|\omega| \ , + \ , u_{\rho}/\pi } } \ , \nonumber \\ & & \hspace{-0.3 in } \times \ , \int \
! d\omega \ , h_0(x_1,x_2,\omega ) \ , \frac{e^{- \alpha |\omega|/v_{f}}}{|\omega| + u_{\rho}/\pi } \nonumber \\ & & \hspace{-0.3 in } \times \ , \left [ i \ , e^{i \pi \rho } \ , \frac{e^{-i [ d(\epsilon_2 ) - d(\epsilon_1 ) ] } e^{-i ( k_1 x_1 + k_2 x_2 ) } } { k_2 - k_1 - i \eta } \ + \ \text{c.c . } \right ] , \label{eq : dmess}\end{aligned}\ ] ] where the function @xmath158 is defined below in eq . [ eq : c1 ] . before we apply standard techniques of complex analysis to reduce the integral over @xmath159 to the contributions from the simple poles at @xmath160 , we need to show that we can neglect the effects of the singularities in the expression for @xmath154 that result from the factors @xmath89 and @xmath161 .
it is not too hard to discount the singular behavior at @xmath162 that results from the functional form of @xmath161 ( see eq . [ eq : parts ] ) .
we can rewrite @xmath163 as follows : @xmath164 where @xmath165 is a sort of variant of the heaviside step function that , in the appropriate limits , is @xmath8 for @xmath84 real and negative and is @xmath98 for @xmath84 real and positive . to allow for complex values of @xmath84 , we can define @xmath165 by the formula @xmath166 , where the path @xmath167 in the complex plane starts at @xmath168 ( with @xmath169 being a large and positive real number , ultimately taken to @xmath170 ) and then proceeds to the value @xmath84 by the shortest path that avoids the singularities at @xmath171 and their accompanying branch cuts . in the limit @xmath172 , this definition yields the desired behavior for real values of @xmath84 . on the other hand
, it does leave @xmath165 with two branch points ( at @xmath173 ) , and to these we must add the branch point of @xmath174 at @xmath175 .
fortunately , it is not too hard to see that integrating around these singularities and associated branch cuts leads to contributions to the final result that are higher - order in the expansion parameter @xmath9 than the contributions from the simple poles at @xmath160 .
rough evaluation of the higher - order contributions indicates that they are indeed negligible in the regime where @xmath176 , but are probably not negligible for values of @xmath177 approaching @xmath178 .
having discounted higher - order contributions from singularities other than the simple poles at @xmath160 , we can perform the @xmath159 integral with ease .
the integral over @xmath179 can then be done after the resulting factor @xmath180 is expanded in powers of @xmath181 ( and after it is noted that @xmath182 } = 1 $ ] ) , and the result is the following : @xmath183 where , once again , @xmath184 is the `` dimensionless fermi energy '' relative to the barrier peak , @xmath185 is the harmonic oscillator energy of the parabolic barrier ( recall the discussion of sec .
i ) , and @xmath186 \ , \nonumber \\ & & \hspace{-0.2 in } \times \ , e^{i \epsilon_f \omega_p ( x_1 + x_2)/v_f } \ , f_1(x_1 , x_2 , u_{\rho } , \omega_p ) \ .
\label{eq : j1}\end{aligned}\ ] ] as before , @xmath63 is the heaviside step function and @xmath187 is the fermi velocity .
furthermore , the function @xmath188 has the formula : @xmath189 ^ 2 \frac{e^{-\alpha |\omega|/v_f } } { |\omega| + u_{\rho}/\pi } } \nonumber \\ & & \hspace{-0.2 in } \times \ , \int \
! d\omega \ , h_0(x_1,x_2,\omega ) \ , \frac{e^{-\alpha |\omega|/v_f } } { |\omega| + u_{\rho}/\pi } \ , , \label{eq : f1}\end{aligned}\ ] ] where @xmath190 \nonumber \\ & & \hspace{-0.6 in } - \ , 2 \left ( e^{u_{\rho } |x_1|/ \pi v_f } - 1 \right ) \left ( 1 - \text{sgn}(x_1 ) \ , \text{sgn}(x_2 ) \right ) \nonumber \\ & & \hspace{-0.2 in } \times \int_{1}^{\infty } \ ! dz \ , \ , \frac{e^{- ( \alpha + |x_1| ) ( u_{\rho}/\pi v_f)z}}{z } \nonumber \\ & & \hspace{-0.6 in } + \ , \left ( e^{2 u_{\rho } |x_1|/ \pi v_f } - 1 \right ) \nonumber \\ & & \hspace{-0.2 in } \times \int_{1}^{\infty } \ ! dz \ , \ , \frac{e^{- ( \alpha + 2 |x_1| ) ( u_{\rho}/\pi v_f)z}}{z } \nonumber \\ & & \hspace{-0.6 in } - \ , \text{sgn}(x_1 ) \ , \text{sgn}(x_2 ) \left ( e^{u_{\rho } ( |x_1| + |x_2|)/ \pi v_f } - 1 \right ) \nonumber \\ & & \hspace{-0.2 in } \times \int_{1}^{\infty } \ ! \ ! dz \ , \ , \frac{e^{- ( \alpha + |x_1| + |x_2| ) ( u_{\rho}/\pi v_f)z}}{z } \ .
\label{eq : c1}\end{aligned}\ ] ] although the equation for @xmath36 may not be entirely transparent , it is not hard to confirm , by taking the limit @xmath191 , that the result for @xmath14 that follows from eqs .
[ eq : fgeneral ] and [ eq : en1ch ] agrees with that previously derived for a delta - function barrier .
@xcite the key is to recognize that the relation @xmath192 means that that @xmath193 as @xmath194 , where @xmath195 is the euler - mascheroni constant ( @xmath196 ) .
the behavior for @xmath197 can be found through numerical integration .
the results for various values of @xmath177 are displayed in figure 2 .
as predicted in our study of the weak - coupling limit ( @xmath17 ) , @xcite for a given value of @xmath3 in the vicinity of @xmath8 , the fractional peak splitting @xmath14 is reduced relative to that for a delta - function barrier .
this strong - coupling depression of the @xmath14-versus-@xmath3 curve becomes greater for larger values of @xmath177 .
= 0.9 + for @xmath198 , however , the downward correction is quite small .
since earlier approximations of the barrier as a delta function were designed to predict the behavior of experimental systems in which @xmath198 , our calculation confirms that those prior results were substantially correct .
figure 2 gives a sense of what the leading @xmath9-dependent corrections look like for small and large values of @xmath9 , but , because of the approximations we have made , the results are expected to be reliably accurate only for @xmath176 .
one might wonder what analytic form the leading @xmath177-dependent corrections take . in the weak - coupling limit ( @xmath17 )
, the finiteness of the barrier was shown to result in leading - order corrections proportional to @xmath199 for @xmath41 .
@xcite corrections of this form arose from the fact that transitions into virtual states with energies near or above that of the barrier peak allow for relatively free interdot movement ( with transmission probabilities on the order of @xmath8 ) . by reasoning analogous ( but in a sense opposite ) to that used in studying the weakly coupled system
, we expect that , to leading order , the finite - barrier corrections in the strong - coupling limit , where @xmath30 , are dominated by the enhanced backscattering of states near or below the barrier peak , where reflection probabilities are on the order of @xmath8 , rather than on the order of @xmath119 as is true near the fermi surface . to estimate these corrections from strongly backscattered states , it is best to return to the form of @xmath36 that appeared in eq .
[ eq : dmess ] .
examination of the @xmath200-dependent factors in eq .
[ eq : dmess ] s integrand reveals that the primary contributions for the integral come when the variables @xmath201 have values of rough magnitude @xmath202 . in this range of @xmath201-values , the @xmath94-independent factors of the integrand (
i.e. , all the factors but @xmath203 and those enclosed in the final set of large brackets ) vary relatively slowly , only changing substantially when one or both of the @xmath201 values changes by an addend of order @xmath202 .
consequently , if the wave vectors @xmath94 have magnitudes greater than @xmath204 , the @xmath94-independent factors can , to leading order , be treated as constants for @xmath205 , and the @xmath201-dependence of the integrands in eq .
[ eq : dmess ] can be assumed to be dominated by the factor @xmath206 .
the question then is whether the @xmath96 values that contribute most substantially to the @xmath9-dependence are large enough ( in magnitude ) for this assumption to be justified .
as we have reasoned above , the leading contributions to @xmath9-dependence are expected to come from @xmath94 such that energy of the corresponding eigenstate is near or below the barrier peak .
in other words , the most significant wave vectors are expected to be ones such that @xmath207 .
consequently , we expect these @xmath94 to be negative and to have magnitudes approximating @xmath208 , where ( given eq .
[ eq : oneminusg ] ) @xmath209 as noted in sec .
ii , the anticipated importance of negative wave vectors of approximate magnitude @xmath210 ( which yield values for @xmath97 equal or near to @xmath98 ) means that we must have @xmath103 for our general approach to be valid .
our desire to gain a more compact analytic approximation to the leading @xmath9-dependent behavior now leads us to impose another requirement : @xmath211 . from eq .
[ eq : kzero ] , this requirement means that we need @xmath212 . combining this upper bound on @xmath35 with our previous lower bound , we find that our general approach and the new assumption that we propose to make ( that the @xmath206-factors dominate the @xmath213-dependence of the integrands ) are both valid only if @xmath214 . the new assumption
therefore allows a useful analytic approximation ( i.e. , an approximation with a significant range of @xmath3 values in which it is valid ) when @xmath9 is not much greater than @xmath28 . for @xmath9 values slightly over @xmath178
, on the other hand , the assumption s utility `` breaks down '' entirely , as the room between the upper and lower limits on @xmath35 vanishes . nonetheless , because we have already restricted our attention to @xmath41 , this `` breakdown '' is of no concern ; for the values of @xmath9 of real interest , our new assumption for the purpose of a compact analytic approximation imposes no more restraint than our initial assumption that @xmath30 . having determined that we can assume @xmath206-dominance
, we can now proceed with estimating the corrections due to wave vectors in the vicinity of @xmath210 .
approximation of the integrations based on the above assumptions leads to the conclusion that , to first approximation , the correction to @xmath36 is proportional to @xmath215 .
thus , the leading finite - barrier correction to the fractional peak splitting @xmath14 may be roughly written as @xmath216 , where @xmath217 is a small positive number .
the form of this correction is parallel to that found in the weak - coupling limit ; in both cases , the leading @xmath9-dependent terms are proportional to @xmath218 . the results displayed in figure 2 bear out our approximation to the analytic behavior of the leading strong - coupling corrections .
the graphed corrections for @xmath219 agree with the predicted analytic behavior , producing errors of less than about 10% for the behavior as a function of @xmath35 and of less than about 20% for the behavior as a function of @xmath9 .
the graphed values suggest that the value of @xmath220 is slightly less than @xmath221 .
the downward shift of the @xmath14-versus-@xmath3 curve for one - channel systems ( @xmath6 ) is paralleled by a similar downward shift of the @xmath14-versus-@xmath3 curve for two - channel systems ( @xmath7 ) .
however , the route to the two - channel result is less straightforward than the already somewhat tortuous route taken for the single - channel problem , largely due to the fact that we now have to deal with two @xmath222-fields in eq .
[ eq : bosact ] , rather than just one .
the first step in dealing with this set of fields is to transform them to a more manageable pair the `` charge '' and `` spin '' fields @xmath223 and @xmath224 , respectively which are linear combinations of the @xmath225 and @xmath226 fields : @xmath227 and @xmath228 .
the bosonic action @xmath229 then consists of the sum of separate charge and spin contributions to the unperturbed action @xmath230 , and a perturbative term @xmath231 , which depends on both the charge and the spin fields . in particular , @xmath232 and @xmath233 , where @xmath234 } \ , e^{-i ( k_1 x_1 + k_2 x_2 ) } } { k_2 - k_1 - i \eta } \right . \nonumber \\ & & \hspace{0.1 in } \times \ , \theta_{c } ( \tau ) \ , \ , e^{-(i/2 ) \int \ ! d\omega \ , h_2(x_1,x_2,\omega ) \ , e^{-i \omega \tau } \ ,
\tilde{\theta}_{c}(\omega ) } \nonumber \\ & & \hspace{-0.05 in } \left . + \ \ \text{m.t . } \right\ } \nonumber
\\ & & \hspace{-0.2 in } \times \ , \left\ { e^{-(i/2 ) \int \ ! d\omega \ , h_2(x_1,x_2,\omega ) \ , e^{-i \omega \tau } \ , \tilde{\theta}_{s}(\omega ) } \ \ + \ \ \text{m.t . }
\right\ } \ , , \label{eq : bosact2}\end{aligned}\ ] ] where the `` mirror terms '' ( m.t . ) can be obtained in the same way as for eq .
[ eq : bosact ] . after integrating out the high - energy charge fields ,
one is left with an effective spin action @xmath235 , where @xmath236 is unchanged from eq .
[ eq : bosact2 ] and where @xmath237^{-1/4 } \nonumber \\ & & \hspace{-0.1 in } \times \ , \left [ \frac{(|x_2| + \alpha ) \ , ( |x_1| + \alpha ) } { \alpha \ , ( |x_1| + |x_2| + \alpha ) } \right]^{-\text{sgn}(x_1 ) \text{sgn}(x_2)/4 } \nonumber \\ & & \hspace{-0.1 in } \times \ , e^{- ( \pi/4 ) \int \ ! d\omega \ , \left [ h_0(x_1,x_2,\omega ) \right]^{2 } \frac{e^{- \alpha |\omega|/v_{f}}}{|\omega| \ , + \ , 2 u_{\rho}/\pi } } \nonumber \\ & & \hspace{-0.1 in } \times \ , \int \ !
d\omega \ , h_0(x_1,x_2,\omega ) \ , \frac{e^{- \alpha |\omega|/v_{f}}}{|\omega| + 2 u_{\rho}/\pi } \nonumber \\ & & \hspace{-0.1 in } \times \ , \left [ i \ , e^{i \pi \rho/2 } \ , \frac{e^{-i [ d(\epsilon_2 ) - d(\epsilon_1 ) ] } \ , e^{-i ( k_1 x_1 + k_2 x_2 ) } } { k_2 - k_1 - i \eta } \ + \ \text{c.c . } \right ] \nonumber \\ & & \hspace{-0.1 in } \times \ , \left [ e^{-(i/2 ) \int \ !
d\omega \ , h_2(x_1,x_2,\omega ) \ , e^{-i \omega \tau } \ ,
\tilde{\theta}_{s}(\omega ) } \ \ + \ \ \text{m.t . }
\right ] \ , , \label{spinact1}\end{aligned}\ ] ] where , as was done in the case of the @xmath6 problem , only the lowest - order perturbative term has been retained in the effective action . to exploit the even - odd combinations that occur in the integrals over @xmath144 , we rewrite the action in terms of the following `` even '' and `` odd '' fields : @xmath238 \ , .
\label{eq : eofields}\end{aligned}\ ] ] the nonperturbative term @xmath236 then takes the form @xmath239 \ , .
\label{eq : szoe}\end{aligned}\ ] ] after integrating out the @xmath240-fields , we have a new effective action in terms of the @xmath241-fields .
the `` even '' fields are not to be left alone , however . in anticipation of `` refermionization , '' we insert into the effective action a set of dummy `` odd '' fields free fields that are entirely decoupled from the `` even '' fields .
the nonperturbative , kinetic energy terms of these new @xmath242-fields and @xmath243-fields add to give a total kinetic energy of the same form as that in eq .
[ eq : szoe ] .
this total is in turn rewritten in a form equivalent to that of @xmath236 in eq .
[ eq : bosact2 ] .
our newest , and final , effective action then consists of the sum of this @xmath236-term and the leading perturbative term @xmath244 , which was derived from integrating out the original `` odd '' fields : @xmath245^{2 } \frac{e^{- \alpha |\omega|/v_{f}}}{|\omega| \
, + \ , 2 u_{\rho}/\pi } } \nonumber \\ & & \hspace{0.0 in } \times \ , \int \ !
d\omega \ , h_0(x_1,x_2,\omega ) \ , \frac{e^{- \alpha |\omega|/v_{f}}}{|\omega| + 2 u_{\rho}/\pi } \nonumber \\ & & \hspace{0.0 in } \times \left [ i \ , e^{i \pi \rho/2 } \ , \frac{e^{-i [ d(\epsilon_2 ) - d(\epsilon_1 ) ] } \ , e^{-i ( k_1 x_1 + k_2 x_2 ) } } { k_2 - k_1 - i \eta } \ + \ \text{c.c . } \right ] \nonumber \\ & & \hspace{0.0 in } \times \ , \left [ e^ { ( i/2 ) \int \ ! d\omega \ , h_0(x_1,x_2,\omega ) \ , e^{-i \omega \tau } \ ,
\tilde{\theta}_{s } ( \omega ) } \ + \ \text{m.t . } \right ] \ , .
\label{eq : spinactfin}\end{aligned}\ ] ] as in prior work , @xcite we `` refermionize '' the bosonic action @xmath246 by identifying @xmath247 as a fermionic annihilation operator at the origin of a semi - infinite system , where @xmath248 where @xmath249 this refermionization formula may look peculiar because of its use of a normalization factor and an ultraviolet cutoff ( embedded in @xmath250 ) that depend on position variables , rather than mere constants ( contrast the standard bosonic representation of fermionic position operators in eq .
[ eq : bosplane1 ] ) .
these position - dependent factors work , in tandem with the scalar prefactor @xmath251 for the exponentiated integral , to ensure the correct anticommutation relations for @xmath247 , and their use is expected to be unobjectionable so long as for all @xmath201 of real interest the resulting ultraviolet cutoff is high enough to capture the behavior with which we are concerned . as in the single - channel problem ,
the @xmath201 of real interest satisfy @xmath252 ( as can be seen by doing the integrals over @xmath94 and obtaining a result with factors analogous to those of @xmath253 of eq .
[ eq : f1 ] ) .
the ultraviolet cutoffs in eq .
[ eq : referm2 ] therefore characteristically correspond to energies approximating @xmath254which is much greater than @xmath255 , the characteristic energy for the unperturbed spin degrees of freedom , and ( as we can confirm once we rediscover the leading order behavior produced by @xmath244 ) , much greater than the energy of states characteristically brought into play by the perturbation @xmath244 .
indeed , we have already incorporated an assumption that ultraviolet cutoffs approximating @xmath254 are valid because , in obtaining our effective action , we have only kept the leading order terms from integrating out the charge degrees of freedom
an approximation only expected to be good if the spin - field states of concern have excitation energies significantly less than @xmath0 .
@xcite returning to eq .
[ eq : spinactfin ] , we use the identity @xmath256 = \int \ ! \ !
dk \ , ( f_{k}^{\dagger } + f_{k})$ ] to find that our refermionization scheme produces a fermionic hamiltonian @xmath257 that consists of the following parts : @xmath258 where @xmath259^{2 } \frac{e^{- \alpha |\omega|/v_{f}}}{|\omega| \ , + \ , 2 u_{\rho}/\pi } } \nonumber \\ & & \hspace{-1.0 in } \times \ , \int \
d\omega \ ,
h_0(x_1,x_2,\omega ) \ ,
\frac{e^{- \alpha |\omega|/v_{f}}}{|\omega| + 2 u_{\rho}/\pi } \nonumber \\ & & \hspace{-1.0 in } \times \ , \left [ i \ , e^{i \pi \rho/2 } \ , \frac{e^{-i [ d(\epsilon_2 ) - d(\epsilon_1 ) ] } e^{-i ( k_1 x_1 + k_2 x_2 ) } } { k_2 - k_1 - i \eta } \ + \ \text{c.c . } \right ] \ , .
\label{eq : zmess}\end{aligned}\ ] ] from prior work on the effects of a delta - function barrier , we know how to solve for the fractional peak splitting that such a hamiltonian produces .
@xcite the only difference from the delta - function barrier hamiltonian that we encountered before is that a complicated prefactor @xmath260 replaces the simpler delta - function barrier prefactor @xmath261 , where @xmath195 is the euler - mascheroni constant ( @xmath196 ) .
@xcite thus , to find the leading behavior of the fractional peak splitting as a function of @xmath35 and @xmath9 , we can simply substitute @xmath260 for @xmath262 in the results previously obtained for a delta - function barrier .
as for one - channel systems , we ultimately resort to numerical integration to solve for the fractional peak splitting when the barrier has a nonzero width .
needless to say , it is gratifying that such numerical calculation confirms that , at least through five significant digits , the prefactor @xmath260 converges to the delta - function quantity @xmath262 in the limit @xmath263 .
consequently , in the limit of a narrow barrier , we recover the same result as for a delta - function barrier @xcite a good confirmation both of the robustness of our earlier results and of our success in wending through the complications created by the initial assumption of a nonzero - width barrier .
the approach to performing the quadruple integral of eq .
[ eq : zmess ] is very similar to that used to perform the analogous quadruple integral for @xmath6 and therefore will not be described in depth .
we first integrate over @xmath179 and @xmath159 by techniques of complex analysis similar to those used for @xmath6 , once again focusing on the simple poles at @xmath264 , with the understanding that this imposes a constraint that @xmath265 . for various values of @xmath9
, we then do the remaining integrals over @xmath151 and @xmath152 numerically .
the results are shown in figure 3 . as for one - channel systems
, we see that as @xmath177 increases , the strong - coupling end of the @xmath14-versus-@xmath3 curve shifts downward from the zero - width result . once again
, the corrections for the experimentally realized values of @xmath266 are small , a fact which confirms that the previous assumption of a delta - function barrier @xcite was substantially justified , at least so long as interaction effects peculiar to the barrier region can be ignored .
= 0.9 + as for the single - channel system , one might inquire about the nature of the leading analytic behavior of these corrections .
the answer is essentially parallel , both in reasoning and substance , to that found at the end of sec .
the constraints on @xmath35 are the same ( i.e. , @xmath267 ) , and the leading finite - barrier corrections to @xmath260 are expected to be roughly proportional to @xmath268 when @xmath41 . in other words , @xmath260 can be expanded as follows : @xmath269 where @xmath270 is a small positive number . because , from prior work , the leading corrections to the fractional peak splitting are proportional to @xmath271 at @xmath272 , the leading finite - barrier correction to the fractional peak splitting is roughly proportional to @xmath43 .
the corrections shown in figure 3 follow this predicted behavior quite well : for @xmath273 , the analytic prediction captures the calculated corrections with error margins of about 10% or less for the @xmath35-dependence and of about 20% or less for the @xmath9-dependence .
the numerical results suggest that @xmath270 is slightly less than @xmath274 .
this paper shows that bosonization techniques for studying the behavior of one - dimensional systems need not be abandoned when finite - length barriers ( or constrictions ) are introduced .
the finite - length effects of those barriers can be captured to leading order by a standard perturbative approach , albeit one that requires complicated calculations and a fair degree of care .
thus , bosonization techniques can still be useful when nontrivial behavior in the barrier region is at issue , and the approach presented in this paper may help investigators to distinguish between single - particle and many - particle effects from a barrier s finite length . with regard to the more particular problem of the coulomb blockade behavior of coupled quantum dots ,
our results can be summarized as follows .
this paper shows that at least for one - channel or two - channel systems , the fractional peak splitting @xmath14 of two strongly coupled dots decreases , for a given value of the interdot conductance , as the ratio @xmath9 is increased . for one - channel systems ( @xmath6 )
, the downward correction behaves as @xmath268 in the limit where @xmath41 and @xmath30 .
thus , for @xmath6 , the fractional peak splitting has the following leading - order functional form : @xmath275 where @xmath276 is somewhat less than 0.05 and , as derived in prior work , @xcite @xmath277 ( about @xmath278 ) . in two - channel systems ( @xmath7 ) , and in the same limits , the downward correction due to the finite barrier width behaves as @xmath43 , resulting in leading - order behavior of the form @xmath279 where @xmath280 is somewhat less than @xmath281 and , as derived in prior work , @xmath282 ( or about @xmath283 ) .
( it should be noted that prior results indicate that , in the @xmath7 equation , the next additive term independent of @xmath9 will equal @xmath284 , where @xmath285 .
@xcite because this term linear in @xmath35 comes from higher - order terms in the effective action than those we have considered here ( see , e.g. , eq .
[ eq : spinactfin ] ) , we have omitted it in our calculations and in the graphs of figure 3 in order to have a truer comparison between the @xmath9-dependent terms and those they directly correct .
if need be , the linear term is easy enough to combine with the result expressed in eq .
[ eq : f2approx ] . )
the above results combine with an earlier study of the weak - coupling regime ( @xmath24 ) @xcite to give a more complete understanding of the @xmath14-versus-@xmath3 curve when one leaves the delta - function barrier limit @xmath263 .
in particular , we come to the nontrivial conclusion that , for the experimentally realized values of @xmath266 , @xcite the corrections to the results derived from modeling the barrier as a delta function are not fundamentally substantial ; thus , to this extent at least , earlier theoretical work using a delta - function potential was correct .
in addition to confirming the essential nature of the @xmath14-versus-@xmath3 curve for @xmath41 , this paper has helped us gain a better picture of what happens to the @xmath14-versus-@xmath3 curve for more general values of @xmath9 .
the weak - coupling results suggested that , as @xmath9 is increased , the @xmath14-versus-@xmath3 curve shifts upward for small values of @xmath3 ( i.e. , for @xmath24 ) and becomes flatter for intermediate values of @xmath3 .
the strong - coupling results suggest ( as conjectured ) that , as the same ratio is increased , the @xmath14-versus-@xmath3 curve shifts downward for large values of @xmath3 ( i.e. , for @xmath30 ) and becomes flatter for intermediate values of @xmath3 .
together , the two sets of results suggest that , in the limit of an extremely wide , `` adiabatic '' barrier ( @xmath286 ) , the @xmath14-versus-@xmath3 curve will be essentially flat for values of @xmath3 in an intermediate region between @xmath98 and @xmath8 .
for such an adiabatic barrier , the @xmath14-versus-@xmath3 curve will sit at some essentially constant value of @xmath14 for most of this interval and will turn sharply toward the limiting values of @xmath287 and @xmath288 at the edges . for the moment ,
the intermediate value at which the `` adiabatic '' @xmath14-versus-@xmath3 curve sits remains a mystery .
the weak - coupling results displayed an antisymmetry around @xmath289 that , if true at higher orders , might aid in solving for the fractional peak splitting when @xmath3 takes on intermediate values .
@xcite unfortunately , the strong - coupling results do not exhibit such a simple symmetry . as
figures 1 and 2 indicate , unlike the weak - coupling results , the strong - coupling results do not reveal a common point of intersection for the leading - order curves that correspond to different values of @xmath9 .
the apparent lack of a common pivot will presumably make solution of the intermediate-@xmath3 problem more difficult .
this work was supported by nsf grants no .
dmr99 - 81283 and no .
dmr98 - 09363 , and darpa grant no . 1034315 .
the authors thank r. westervelt , c.h .
crouch , and c. livermore for helpful conversations .
f. r. waugh , m. j. berry , d. j. mar , r. m. westervelt , k. l. campman , and a. c. gossard , phys .
lett . * 75 * , 705 ( 1995 ) ; f. r. waugh , m. j. berry , c. h. crouch , c. livermore , d. j. mar , r. m. westervelt , k. l. campman , and a. c. gossard , phys .
b * 53 * , 1413 ( 1996 ) ; f. r. waugh , ph.d .
thesis , harvard university , 1994 .
liu , phys .
b * 56 * , 6732 ( 1997 ) .
unfortunately , we have not been able to follow the reasoning of liu s paper , which reports results that contrast sharply with our results here and with those derived in refs .
7 - 10 .
it should be noted that the two - dot system that we have described has a close mathematical relation to other systems , including one consisting of a single dot coupled to one lead through a nearly open point contact .
7 and 8 above .
information about the energy of this single dot can be obtained by measuring the capacitance between the dot and a nearby overlying gate .
5 above .
k. j. thomas , j. t. nicholls , m. pepper , w. r. tribe , m. y. simmons , and d. a. ritchie , phys . rev .
b * 61 * , r13,365 ( 2000 ) ; k. j. thomas , j. t. nicholls , n. j. appleyard , m. y. simmons , m. pepper , d. r. mace , w. r. tribe , and d. a. ritchie , phys . rev .
b * 58 * , 4846 ( 1998 ) .
v. j. emery , `` theory of the one - dimensional electron gas , '' in _ highly conducting one - dimensional solids _ , edited by j. t. devreese , r. p. evrard , and v. e. van doren ( plenum , new york , 1979 ) , pp .
247 - 303 .
h. j. schulz , in _ strongly correlated electronic materials : the los alamos symposium , 1993 _ , edited by k. bedell et al .
( addison - wesley , reading , massachusetts , 1994 ) , pp .
in _ mesoscopic quantum physics , les houches , session lxi , 1994 _ , edited by e. akkermans , g. montambaux , j .- l .
pichard , and j. zinn - justin ( elsevier , amsterdam , 1995 ) , pp . 533ff
. because we are only concerned with the movement of electrons through the channel between the dots ( rather than motion within the dots themselves ) , we use free boundary conditions , and the eigenfunctions that result consist of incident , transmitted , and reflected parts . | a pair of quantum dots , coupled to each other through a point contact , can exhibit coulomb blockade effects that reflect the presence of an oscillatory term in the dots total energy whose value depends on whether the total number of electrons on the dots is even or odd . the effective energy associated with this even - odd alternation
is reduced , relative to the bare coulomb blockade energy @xmath0 for uncoupled dots , by a factor @xmath1 that decreases as the interdot coupling is increased .
when the transmission coefficient for interdot electronic motion is independent of energy and is the same for all channels within the point contact ( which are assumed uncoupled ) , the factor @xmath1 takes on a universal value determined solely by the number of channels @xmath2 and the dimensionless conductance @xmath3 of each individual channel .
when an individual channel is fully opened ( the limit @xmath4 ) , the factor @xmath1 goes to zero .
when the interdot transmission coefficient varies over energy scales of the size of the bare coulomb blockade energy @xmath0 , there are corrections to this universal behavior . here
we consider a model in which the point contact is described by a single orbital channel containing a parabolic barrier potential , with @xmath5 being the harmonic oscillator frequency associated with the inverted parabolic well .
we calculate the leading correction to the factor @xmath1 for @xmath6 ( spin - split ) and @xmath7 ( spin - degenerate ) point contacts , in the limit where @xmath3 is very close to @xmath8 and the ratio @xmath9 is not much greater than @xmath8 . calculating via a generalization of the bosonization technique previously applied in the case of a zero - thickness barrier , we find that for a given value of @xmath3 , the value of @xmath1 is increased relative to its value for a zero - thickness barrier , but the absolute value of the increase is small in the region where our calculations apply .
= 10000 2 |
coexistence of magnetism with the mixed state of type - ii superconductors has attracted great interests @xcite .
nearly antiferromagnetic spin fluctuation close to the quantum critical point has been a key to understand the magnetic correlation of high-@xmath1 superconductors .
such an underlying magnetic correlation may affect the local electronic structure at vortex cores .
tl@xmath2ba@xmath2cuo@xmath3 ( tl2201 ) is an overdoped high-@xmath1 superconductor .
the optimal @xmath1 is about 85 k for a reduced oxygen content .
the early studies by tl nmr with 4.26 t @xcite and cu nmr with 7 t @xcite for tl2201 reported an obvious redfiled pattern due to the vortex lattice .
although the magnetic ordering in the mixed state is not expected for tl2201 , because it is a typical overdoped electronic system away from antiferromagnetic instability @xcite , the recent tl nmr study with 2.1 t shows coexistence of local antiferromagnetic ordering at the vortex cores @xcite .
the specific characteristics of tl2201 are cation disorder on the tl site @xcite and magnetic field dependence of the elementary excitations @xcite .
disorder makes the vortex phase rich @xcite .
the effect of these characteristics on the antiferromagnetic vortex core is poorly understood . in this paper , we studied the vortex state of an overdoped high-@xmath7 superconductor tl2201 ( @xmath485 k ) with magnetic field of 2 t along the @xmath5 axis using a @xmath0tl nmr spin - echo technique .
we observed a redfield pattern of the @xmath0tl nmr spectrum at 5 k due to the vortex lattice state without static antiferromagnetic order at the vortex cores .
the powder sample in the present study is the same one as in the previous study @xcite , which was mixed with stycast 1266 epoxy and magnetically aligned along the @xmath5 axis .
a phase - coherent - type pulsed spectrometer was utilized to perform the @xmath0tl nmr ( spin @xmath8=1/2 ) experiments .
all the measurements were done with magnetic field cooling .
the nmr frequency spectra with quadrature detection were obtained by integration of the nuclear spin - echoes with frequency @xmath9 .
the width of the first exciting @xmath10/2-pulse @xmath11 was 4 @xmath12s ( the excited frequency region @xmath1363 khz ) . because of no fine structure but moderately broad nmr spectra ,
high spectral resolution @xcite was not required in the present measurements .
the nuclear spin - lattice relaxation curves @xmath14 ( recovery curves ) were measured by an inversion recovery technique , as functions of a time @xmath15 after an inversion pulse , where the nuclear spin - echo amplitude @xmath16 , @xmath17 $ ] and @xmath15 were recorded .
in contrast to the previous study @xcite , the comb pulse train was not needed .
the measurement of @xmath18 was necessary to determine the function form of @xmath19 .
figure 1 shows @xmath0tl nmr spectra with magnetic field of @xmath20=2 t along the @xmath5 axis at @xmath6=85 and 5 k. the bare resonance line @xmath21 given by @xmath22 ( @xmath23=24.567 mhz / t ) is denoted by the dash line . above @xmath7 ,
the full width at half maximum of the spectrum is about 50 khz , being nearly the same as the reported one @xcite .
the linewidth can be considered to result from a staggered spin susceptibility ( the second moment ) and from a knight shift distribution ( the first moment ) .
the enhancement of the static staggered spin susceptibility along the @xmath5 axis is estimated from the @xmath24cu nuclear spin - spin relaxation rate 1/@xmath25 @xcite .
the @xmath0tl knight shift is understood by the sum of the spin shift @xmath26 due to the cu - to - tl supertransferred hyperfine coupling and the temperature - independent @xmath27 due to residual spin density polarization @xcite .
below @xmath7 , the spectrum peak is largely shifted into the negative shift side ( low frequency region ) , while the tail is still in the positive shift side ( high frequency region ) . since for a singlet @xmath28-wave superconductor , local density of states of electrons
is suppressed away from vortex cores but recovers at the cores @xcite , then we expect that a redfield pattern of the spin shift @xmath29 due to the electron density of states affects the nmr spectrum .
however , the observed negative shift indicates that a redfield pattern of a superconducting diamagnetic shift @xmath30 is predominant .
the @xmath6=5 k nmr spectrum resembles the field distribution due to a slightly distorted vortex lattice @xcite but not a gaussian due to amorphous vortex or vortex glass .
we could not deduce from the nmr spectrum which the vortex lattice is , triangular or square .
the peak frequency can be assigned to the saddle point ( s ) in the field distribution of vortices , the highest frequency edge to the vortex core ( v ) , and the lowest frequency edge to the minimum field position ( c ) mostly away from the vortex cores @xcite .
site - selective nmr is possible for our sample . in fig .
1 , the knight shift at the vortex core v at 5 k is nearly the same as or larger than the normal shift at 85 k. an external magnetic field is squeezed at the vortex core .
we notice that the squeezed magnetic flux density at the core yields a higher magnetic field than an applied external field , so that the shift at the core - site v can be larger than the normal shift . using the additional positive vortex core shift @xmath31(v ) due to the enhanced field and the negative shift @xmath32(c)@xmath330 due to the diamagnetic field
, we obtain @xmath34[1+^{sc}k_{c , field}(\mathrm{v } ) ] \label{e.core}\ ] ] with @xmath35(v)@xmath36 @xmath26(@xmath7 ) @xcite , while @xmath37 .
\label{e.mini}\ ] ] these are illustrated in fig .
a part of the nmr signals at @xmath3849.18 mhz is due to the combined effect of magnetic flux squeezing and of the finite positive spin shift at the vortex core . to our knowledge
, no one has ever pointed out the importance of this effect .
no signal above 49.4 mhz at 5 k , that is , no evidence for the local antiferromagnetic ordering effect at the vortex cores was found for our sample .
the difference in imperfection or in the crystal symmetry @xcite may affect the local order .
figure 3(a ) shows frequency distribution of the recovery curves at 5 k. the recovery curve is close to an exponential function at the peak frequency ( s ) but nonexponential at lower and higher frequency sides .
these results remind us of the in - plane impurity - substitution effect @xcite .
the nuclear moments close to an impurity show nonexponential relaxation , because of random impurity distribution .
figure 3(b ) shows temperature dependence of the recovery curve at @xmath9=49.18 - 49.20 mhz .
one can see nonexponential curves below 30 k and a nearly exponential one at 40 k. the solid curves in fig .
3 are the least - squares fitting results by @xmath39 , \label{e.stre}\ ] ] where @xmath40 , @xmath41 and @xmath42 are the fitting parameters . equation ( [ e.stre ] ) is derived from an impurity - induced nmr relaxation theory @xcite .
@xmath41 is a nuclear spin - lattice relaxation time due to quasi - particle scattering or homogeneous relaxation process , whereas @xmath42 is an induced relaxation time due to random distribution of impurity relaxation centers . since there are no literal localized moments , the finite @xmath42 may result from an extended staggered spin fluctuation around the vortex core @xcite , friedel oscillation around the core @xcite , or thermal vortex motion @xcite . for breakdown of the site - selective nmr ,
the recovery curve of the nuclear magnetization is nonexponential in the mixed state @xcite . in the case
that the magnetic penetration depth @xmath43 is too long , a sharp nmr line with hierarchical @xmath41 components is observed and then the site - selective nmr does not work .
the second moment of the nmr spectrum in the vortex lattice , a guide to broadening , is given by @xmath44 where @xmath45 is the field distribution around the peak @xcite .
for deeply underdoped and heavily overdoped superconductors , the long @xmath43 s are estimated @xcite . for our sample ,
the observed redfield pattern indicates a short @xmath43 .
thus , we suppose one - to - one correspondence of the site to the frequency shift .
let us briefly discuss the reason why the relaxation curve is nonexponential in spite of site selection in nmr .
figure 4(a ) illustrates a perfect square lattice of vortices on a cuo@xmath2 plane for convenience .
the recovery curve at any nuclear site in fig .
4(a ) must be an exponential function , even if local moments or local staggered moments are induced at the vortex cores .
for example , the vortex configurations around all the b sites in fig
. 4(a ) are the same , so that all the recovery curves of the b - site nuclei would become the same .
however , we observed the nonexponential recovery curves at the selected sites .
thus , the vortex lattice is not perfect .
one may call it bragg glass @xcite .
figure 4(b ) illustrates an imperfect vortex lattice with weak disorder . the vortex configuration around the b site in fig .
4(b ) is not unique , so that the assemble average on the vortex distribution leads to nonexponential relaxation .
figure 5(a ) shows @xmath0tl nmr spectra with magnetic field cooling .
the narrow nmr lines between @xmath1=85 k and @xmath6=65 k indicate motional narrowing effect , that is , vortex melting .
the nmr spectra between @xmath6=60 and 10 k are composed of two peaks with a negative and a positive knight shifts .
these peaks indicate coexistence of the vortex solid with liquid , that is , partial melting . at @xmath6=5
k , the redfield pattern indicates the vortex lattice .
the similar , but narrow , coexistence region was first reported for yba@xmath2cu@xmath46o@xmath47 @xcite .
the detailed measurement of vortex phase is in progress for our tl2201 sample
. figures 5(b ) and 5(c ) show 1/@xmath42 and 1/@xmath41 as functions of temperature at c , s , and l on the solid lines in fig .
5(a ) . in fig .
5(b ) , the peak behavior of 1/@xmath42 at l is observed at @xmath48=20 k. since the signal l comes from the vortex liquid in the coexistence region , the peak behavior indicates slowing down of vortex motion toward freezing .
this is not due to the local antiferromagnetic ordering around vortex cores .
@xmath48 is close to the irreversibility line @xcite . the temperature dependence of 1/@xmath41 at c and s in fig .
5(c ) can be understood by the spatial dependence of local density of states of electrons @xcite .
more than ten years have passed since our sample was synthesized .
the vortex phase is sensitive to crystalline imperfection .
the aging effect , e.g. oxygen redistribution , might realize the solid - liquid coexistence .
the effect , however , does not cause @xmath1 distribution in our sample .
we exclude a possibility that the signal l is a lower-@xmath1 part of the sample .
the magnetic susceptibility does not exhibit @xmath1 distribution .
the cu nuclear quadrupole resonance spectrum at @xmath6=4.2 k , which is sensitive to the oxygen distribution , is still similar to the previous report @xcite .
the temperature dependence of 1/@xmath42 of the signal l is not conventional behavior of superconducting onset .
to conclude , we observed a vortex lattice with weak disorder for our sample tl2201 ( @xmath485 k ) with 2 t at 5 k , coexistence of vortex solid with liquid ( @xmath6=10 - 60 k ) , and vortex liquid ( @xmath6=65 - 85 k ) , but not the local antiferromagnetic ordering at the vortex cores .
we thank a. goto for fruitful discussion , and m. kato and t. waki for experimental supports .
this was supported by a grant - in - aid on priority area `` novel quantum phenomena in transition metal oxides , '' from ministry of education , science , sports and culture ( 12046241 ) .
e. demler , w. hanke , and s. c. zhang , cond - mat/0405038 , and references therein .
m. mehring , f. hentsch , hj .
mattausch , and a. simon , solid state commun . * 75 * , 753 ( 1990 ) .
a. goto , y. itoh , h. yasuoka , a. hayashi , and y. ueda , institute for solid state physics ( issp ) report no .
* 2846 * ( 1994 ) .
s. kambe , h. yasuoka , a. hayashi , and y. ueda , phys .
b * 47 * , 2825 ( 1993 ) .
k. kakuyanagi , k.kumagai , y. matsuda , and m. hasegawa , phys .
. lett . * 90 * , 197003 ( 2003 ) .
y. shimakawa , physica c * 204 * , 247 ( 1993 ) .
s. kambe , h. yasuoka , a. hayashi , and y. ueda , phys .
lett . * 73 * , 197 ( 1994 ) .
oleg m. vyaselev , and igor f. schegolev , czechoslovak j. of phys .
* 46 * , 1137 ( 1996 ) suppl . s2 .
s. kambe , h. yasuoka , a. hayashi , and y. ueda , phys .
b * 48 * , 6593 ( 1993 ) .
k. shibata , t. nishizaki , t. sasaki , and n. kobayashi , phys .
b * 66 * , 214518 ( 2002 ) . w. g. clark , m. e. hanson , f. lefloch , and p. segransan , rev .
* 66(3 ) * , 2453 ( 1995 ) .
y. itoh , h. yasuoka , a. hayashi , and y. ueda , j. phys .
b * 63 * , 22 ( 1994 ) . m. takigawa , m. ichioka , and and k. machida , j. phys
. soc . jpn . *
69 * , 3943 ( 2000 ) .
t. m. riseman , j. h. brewer , k. h. chow , w. n. hardy , r. f. kiefl , s. r. kreitzman , r. liang , w. a. macfarlane , p. mendels , g. d. morris , j. rammer , j. w. schneider , c. niedermayer , and s. l. lee , phys .
b * 52 * , 10569 ( 1995 )
. e. h. brandt , phys .
b * 37 * , 2349 ( 1988 )
. a. g. redfield , phys .
rev . * 162 * , 367 ( 1967 ) .
y. itoh , t. machi , c. kasai , s. adachi , n. watanabe , n. koshizuka , and m. murakami , phys .
b * 67 * , 064516 ( 2003 ) .
m. r. machenry , b. g. silbernagel , and j. h. wernick , phys .
* 27 * , 426 ( 1971 ) ; phys . rev . * 5 * , 2958 ( 1972 ) .
h. tsuchiura , m. ogata , y. tanaka , and s. kashiwaya , phys . rev . * 68 * , 012509 ( 2003 ) .
m. machida , and t. koyama , phys .
lett . * 90 * , 077003 ( 2003 ) .
g. b. furman , e. m. kunoff , and s. d. goren , physica c * 266 * , 203 ( 1996 ) , and references therein .
p. pincus , a. g. gossard , v. jaccarino , and j. h. wernick , phys . lett . * 13 * , 21 ( 1964 ) .
niedermayer , c. bernhard , u. binninger , h. gluckler , j. l. tallon , e. j. ansaldo , and j. i. budnick , phys .
lett . * 71 * ( 1993 ) 1764 .
t. giamarchi , and p. le doussal , phys .
b * 52 * , 1242 ( 1995 ) .
a. p. reyes , x. p. tang , h. n. bachman , w. p. halperin , j. a. martindale , and p. c. hammel , phys .
b * 55 * , r14737 ( 1997 ) .
h. n. bachman , a. p. reyes , v. f. mitrovic , w. p. halperin , a. kleinhammes , p. kuhns , and w. g. moulton , phys .
lett . * 80 * , 1726 ( 1998 ) . c. bergemann ,
a. w. tyler , a. p. mackenzie , j. r. cooper , s. r. julian , and d. e. farrell , phys .
b * 57 * , 14387 ( 1998 )
. s. kambe , y. yoshinari , h. yasuoka , a. hayashi , and y. ueda , physica c * 185 - 189 * , 1181 ( 1991 ) . | we report a @xmath0tl nmr study of vortex state for an aligned polycrystalline sample of an overdoped high-@xmath1 superconductor tl@xmath2ba@xmath2cuo@xmath3 ( @xmath485 k ) with magnetic field 2 t along the @xmath5 axis .
we observed an imperfect vortex lattice , so - called bragg glass at @xmath6=5 k , coexistence of vortex solid with liquid between 10 and 60 k , and vortex melting between 65 and 85 k. no evidence for local antiferromagnetic ordering at vortex cores was found for our sample . |
the _ spitzer _ galactic legacy infrared mid - plane survey extraordinaire ( glimpse : * ? ? ?
* ) has discovered a new class of interesting objects that show extended bright emission at the irac band 2 ( 4.5@xmath0 m ) , called extended green objects ( egos , * ? ? ? * ) .
egos are often discussed in terms of candidates of massive young stellar objects ( ysos ) with outflows .
the greenish color is attributed either to h@xmath1 ( @xmath600 , s(9 , 10 , 11 ) ) lines and co ( @xmath610 ) band heads excited by shocks , or scattered light in the outflow cavity ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
@xcite make detailed analysis of the origin of the excess emission based on photometric data of egos and conclude that extra components other than h@xmath1 emission are needed to explain the excess emission at 4.5@xmath0 m and that continuum emission of the scattered light in the outflow cavity is likely the primary origin of the excess , which is supported by @xcite .
supernova remnants ( snrs ) also show excess emission at 4.5@xmath0 m due to h@xmath1 and co molecular bands @xcite .
while h@xmath1 emission is thought to dominate in the line emission , the contribution from co emission becomes comparable to that from h@xmath1 for the h@xmath1 density of a range @xmath7@xmath8 @xcite .
co ( @xmath5 ) emission is not detected in the snr ic443 by ground - based observations @xcite , but it is detected toward the orion molecular cloud ( omc * ? ? ? * ; * ? ? ?
* ; * ? ? ?
* ) , suggesting that co emission could make a contribution to the excess emission .
it is indispensable to make spectroscopic observations in order to study the origin of the excess at 4.5@xmath0 m in egos unambiguously .
however , no direct spectroscopy has so far been made for egos except for the 2@xmath0 m region @xcite due to the lack of sensitive spectrometer in the 4@xmath0 m region . in _ this paper _
, we report the result of spectroscopy in 2.05.3@xmath0 m of one of the egos , g318.05 + 0.09 , made with the infrared camera ( irc ) onboard _ akari _ , which offered high - sensitivity spectroscopy in 2.05.3@xmath0 m even in its warm mission phase @xcite .
the _ akari _ /irc observation of g318.05 + 0.09 was carried out in the director time on 2010 february 18 , which was one of the last observations of the _ akari _ mission ( observation i.d . :
5201570.1 ) .
the observation was made with the slit spectroscopy mode at the ns slit of @xmath9 width with the prism and grism dispersers ( aot ircz4 with c;ns ) .
the prism spectroscopy provides spectra of 2.05.3@xmath0 m with the spectral resolution of about 0.14@xmath0 m at 4@xmath0 m , while spectra of 2.55.0@xmath0 m with the resolution of about 0.03@xmath0 m are obtained with the grism . in the present observation , four exposure frames with the prism
were first taken and then five exposures with grism were performed , which sandwiched a 3.2@xmath0m - band imaging observation in - between that was used for the position determination ( see * ? ? ?
* ; * ? ? ?
* for more details of the irc spectrsocpy ) .
it provided long slit spectroscopy of about @xmath10 long along the slit direction .
the slit location is shown in figure [ fig1 ] superimposed on the _ spitzer_/irac artificial three - color image together with the positions of the known sources in the catalogs ( see table [ tab1 ] ) .
the slit of the present observation runs from an infrared bright region at the western edge , which is most likely to be iras 144985856 , although the catalog position is slightly different from the bright source ( cross # 3 in fig . [ fig1 ] ) .
the slit covers a greenish blob at around its center .
the two distinct sources are clearly seen in the long - slit spectrum image .
we call the bright source @xmath11 and the greenish blob @xmath12 in the following .
spectra are extracted from @xmath13 ( 6 pixels in the slit direction ) areas indicated by the yellow and white boxes both for the grism and prism data .
the former corresponds to @xmath11 and the latter to @xmath12 .
the slit area of @xmath11 does not collect the entire emission from the bright source .
ego g318.05 + 0.09 is defined by a diamond - like shape area of @xmath14 , whose center is located at slightly north of the bright source as marked by cross # 2 .
it extends to the east and west directions symmetrically to enclose the bright source and cover the greenish regions , including @xmath12 ( fig . 3.14 of * ? ? ?
the present _
spectroscopy covers only the eastern part of g318.05 + 0.09 .
the center position of ego g318.05 + 0.09 is located in the region for which the spectrum of @xmath11 is extracted .
the methanol maser source mmb318.050 + 0.087 @xcite is also located within the box of @xmath11 ( cross # 1 ) .
table [ tab1 ] summarizes the central positions of the regions where the spectra are extracted together with the known sources in this region . the data reduction was performed with the latest toolkit for the phase 3 ( warm mission phase ) of the irc spectroscopy , irc_spec_toolkit_p3_20150331 . since the observation was made in the very last period of the warm mission phase of _ akari _ , the effect of hot pixels became severe and the noise level was higher than typical values in the phase 3 spectroscopy @xcite .
to remove residual dark current and possible contribution from the surrounding background / foreground emission , we subtract the emission enclosed by the white rectangle in the eastern part of the slit labeled by bkg in figure [ fig1 ] .
there is a shift in the slit position on the detector array between the grism and prism spectroscopy due to the optical alignment .
the amount of the shift is estimated by matching the bright region and the spectra of the same region are extracted within an accuracy of a half pixel ( @xmath15 ) along the slit direction for the grism and prism modes .
details of the data reduction are summarized in @xcite . due to the relatively large number of hot pixels , a few residual hot pixels
remain after the standard reduction process .
they are masked manually .
the masked pixels are at 4.457@xmath0 m and 4.5154@xmath0 m of the grism spectrum of @xmath11 and at 3.96073.9801@xmath0 m in the grism spectrum of @xmath12 and they do not affect the following discussion . it should also be noted that the absolute calibration may have a large uncertainty ( @xmath16% ) due to the elevated temperature of the detector array at the observing time ( @xmath17k ) because the sensitivity is not well calibrated at warm temperatures , while no change in the spectral response due to the change in the temperature is recognized @xcite .
we have not applied correction for the contamination of the second order light for the grism spectra in the phase 2 ( cold mission phase ) recently reported by @xcite because there is no calibration of the contamination estimate available for the phase 3 data at present .
instead we discard the data in the spectral region where the contamination is expected ( @xmath18@xmath0 m ) . the improved wavelength calibration provided by @xcite
is applied to the present grism spectra .
the correction for wavelengths longer than 4@xmath0 m is not significant ( @xmath19@xmath0 m ) compared to the instrumental spectral resolution .
clllc @xmath11 & 14 53 42.2 & -59 08 48.8 & present work + @xmath12 & 14 53 44.0 & -59 08 54.3 & present work + mmb318.050 + 0.087 & 14 53 42.67 & -50 08 52.4 & methanol maser & 1 + g318.05 + 0.09 & 14 53 42.6 & -59 08 49 & ego ( @xmath14 ) & 2 + iras 144985856 & 14 53 41.6 & -59 08 54 & uc & 3 + source b & 14 53 39.5 & -59 08 52 & x - ray source & 4 + g318.2 + 0.1 & 14 54 50 & -59 46 & snr & outside of fig .
figure [ fig2 ] shows the extracted spectra of @xmath11 and @xmath12 . there is apparent discrepancy between the grism and prism spectra of @xmath11 in 45@xmath0 m .
we attribute the difference to the saturation effect of the prism spectrum since @xmath11 is very bright for the prism spectroscopy in phase 3 .
the grism spectrum does not have a saturation problem at this brightness level and thus we regard it more reliable .
the grism spectrum shows absorption features at around 3.0 , 4.25 , and 4.67@xmath0 m clearly , whose traces are also seen in the prism spectrum .
the former two can be attributed to h@xmath1o and co@xmath1 ice absorption , and the latter to co gas or ice absorption . to estimate the absorption features quantitatively ,
we fit the continuum by a spline curve with anchor points at 2.6 , 2.65 , 3.8 , 4.1 , 4.4 , and 4.9@xmath0 m ( thin brown dotted line in fig .
[ fig2 ] ) . for the h@xmath1o ice absorption profile ,
the laboratory data taken at 10k are adopted @xcite .
we find that an extra `` red wing '' component is needed to fit the 3@xmath0 m absorption feature in addition to the laboratory data profile .
the red wing is seen in many objects , but its origin is not yet fully understood ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) . in the present fit
, it is approximated by a gaussian profile with the central wavelength of 3.35@xmath0 m and the fwhm of 0.31@xmath0 m .
figure [ fig3]a shows details of the fit of the 3@xmath0 m absorption feature in units of the optical depth .
we assume a gaussian function for the co@xmath1 ice feature at 4.25@xmath0 m because it is narrow compared to the present spectral resolution and use the band strengths of co@xmath1 in @xcite to estimate the column density .
the observed profile is well fitted by a gaussian with the central wavelength of 4.25@xmath0 m and the fwhm of 0.036@xmath0 m .
the fwhm is adopted from the study of _
ns spectroscopy of galactic star - forming regions @xcite .
the central wavelength of the absorption peak of co@xmath1 ice is slightly shorter than the typical value of 4.26@xmath0 m obtained in the same study , but it is still within the uncertainty due to the large noise level and the spectral resolution of the present observation . the absorption feature at around 4.67@xmath0 m is broader than the co ice feature @xcite .
we fit it with both co gas and ice components . for co ice
, we adopt the central wavelength of 4.67@xmath0 m and the fwhm of 0.031@xmath0 m obtained from the ns spectroscopy by @xcite and assume the band strength reported in @xcite .
the observed feature can be fitted by absorption of co gas at 20k and co ice as shown in fig .
the contribution from co gas dominates in the absorption . with the present spectral resolution ,
the p and r branches of co gas absorption are not resolved .
the red side is relatively well fitted , while the fit is not good at the blue side , where there might be an additional component at around 4.62@xmath0 m . because of the large noise , however , it is difficult to make further discussion on the additional component .
co gas with 1530k gives similar fits and the gas temperature can not be well constrained . as fig .
[ fig3 ] suggests , the contribution from co ice also has a large uncertainty .
therefore , the present estimate of the column densities of co gas and ice should be taken with caution .
the derived column densities are @xmath20 , @xmath21 , @xmath22 , and ( @xmath23@xmath24 for h@xmath1o ice co@xmath1 ice , co ice , and co gas , respectively .
source @xmath12 shows a peculiar spectrum .
it does not show emission features at 3.33.5@xmath0 m , which are generally seen in star - forming regions and are attributed to polycyclic aromatic hydrocarbons @xcite , nor ice absorption features seen in ysos as in @xmath11 .
it shows two broad peaks at around 4.5 and 4.7@xmath0 m and no significant emission is seen below 4@xmath0 m .
both the prism and grim spectra show similar characteristics and thus we conclude that these are real features despite the noisy spectra . a small difference between the grism and prism spectra
is seen at wavelengths longer than 4.5@xmath0 m .
it can be attributed partly to the large noise in the grism spectrum and/or to a possible difference in the spectrum extracted region due to the shift of the slit positions on the detector array ( see [ sec : obs ] ) .
the @xmath12 spectrum can be accounted for by blue - shifted optically thin emission of the fundamental ro - vibrational transitions ( @xmath5 ) of co molecules with red continuum .
the two peaks are attributed to the p and r branches of co gas emission . since the grism spectrum is very noisy and
does not have data longer than 4.9@xmath0 m , we use the prism spectrum to estimate the temperature and the velocity of the co gas .
we simply assume that the continuum increases linearly with the wavelength and the co emission is approximated by the local thermodynamic equilibrium ( lte ) condition . from the fit , the temperature and the velocity of the co gas
are estimated as @xmath25k and @xmath26kms@xmath4 , respectively .
the grism spectrum requires more steeply increasing continuum , but the best fit parameters of the co gas ( 2000k and @xmath27kms@xmath4 ) are within the uncertainties estimated from the prism spectrum .
the model spectra with the best fit parameters are shown in fig .
[ fig2]b both for the prism ( thick black line ) and grism spectra ( thin black line ) together with the linear continuum ( thick and thin dotted brown lines ) .
the column density of the co gas is estimated from the prism spectrum with the best fit parameters as @xmath28@xmath24 , which does not include the systematic error due to the calibration uncertainty at the time of the observation ( [ sec : obs ] ) .
the shift in the wavelength corresponding to the estimated blue - shift velocity is @xmath29@xmath0 m , which is about a half of the resolution of the prism spectrum , but more than twice of that of the grism spectrum at 4.5@xmath0 m .
the latest wavelength calibration of the grism spectroscopy has been applied ( [ sec : obs ] , * ? ? ?
* ) and the accuracy is estimated to be better than 0.006@xmath0 m , which is much smaller than the shift .
the wavelength calibration of the prism spectroscopy is less accurate and estimated to be better than 1 pixel or 0.04@xmath0 m at 45@xmath0 m @xcite .
the comparison of the prism and grism spectroscopy confirms that both wavelengths match with each other within this accuracy @xcite .
the shift is constrained by the profile shape of the co emission in the fit and non - lte effects could affect the shift estimate . to investigate the reliability of the suggested large blue - shift velocity , we simply calculate the peak positions of the observed spectra and compare them with those of the lte model spectra of different temperatures .
the peak positions are estimated by a gaussian fit in the ranges 4.34.6@xmath0 m and 4.654.9@xmath0 m for the r and p branches , respectively .
we also vary the slope of the underlying continuum for 27mjysr@xmath4@xmath0m@xmath4 , which is determined by the range of the slopes obtained in the co profile fit .
the effect of the continuum slope is found to be smaller ( @xmath30 and @xmath31@xmath0 m for the r and p branches , respectively ) than the uncertainties in the gaussian fits both for the prism and grism spectra .
the peaks of the model co emission are estimated in the same manner .
figure [ fig4 ] plots the difference between the p and r branch peak wavelengths against the estimated peak wavelength of the r branch both for the prism and grism spectra together with those of the model co emission for the temperature range 5004000k with the various blue - shift velocities .
the peak wavelength is sensitive both to the temperature and the blue - shift velocity , but the difference in the branch peaks is sensitive mostly to the temperature .
the latter is used to separate the temperature effect from the velocity effect .
the peak wavelength of the r branch of the prism spectrum agrees with that of the grism spectrum within the calibration uncertainty of the prism spectroscopy .
the prism and grism spectra show compatible values for the difference in the r and p branch peak wavelengths within the uncertainties in the fit .
both data points suggest that the observed spectra can be accounted for by co emission with blue - shift velocities of @xmath3km@xmath4 and that it is difficult to explain the observed peak wavelengths by simple lte models with velocities larger than @xmath32kms@xmath4 . the suggested temperatures of the co emission range from 1200 to 3700k , which are in agreement with the co emission profile fit .
while these analyses suggest a large blue - shift velocity consistently , the present spectra do not have a high signal - to - noise ratio and good spectral resolution .
spectra with higher spectral resolution and better quality are needed to investigate the non - lte effects and confirm the large blue - shift velocity .
source @xmath11 is located close to the position of iras 14498@xmath25856 .
the iras source is classified as an ultra compact ( uc ) region based on the @xmath33 colors and methanol maser emission at 6668 mhz is detected toward this object @xcite .
the position of the maser is accurately measured by @xcite , which matches well with the position of @xmath11 ( fig .
[ fig1 ] and table [ tab1 ] ) .
the methanol maser at 6668 mhz is known as a good tracer for the early stages of high - mass star formation ( * ? ? ?
* and references therein ) .
the center position of the egog318.05 + 0.059 is close to the maser position and it is suggested to be a likely massive yso outflow candidate in the ego catalog @xcite .
the _ akari_/irc spectrum of @xmath11 shows the absorption features of h@xmath1o and co@xmath1 ices clearly .
these ice absorption features are also characteristics of embedded ysos ( e.g. , * ? ? ?
* ; * ? ? ?
the column densities of co@xmath1 and h@xmath1o ices are known to show a correlation in ysos and the ratio depends on the nature of the object @xcite .
the abundance ratio of co@xmath1 to h@xmath1o ices in @xmath11 is @xmath34% , being within the range of the ratios found for weakly processed ices in massive ysos in our galaxy .
the co ice absorption is generally not abundant .
while the relative abundance of co ice to co@xmath1 ice varies with the line of sight @xcite , the present relative abundance ( @xmath35% relative to h@xmath1o ice ) is also within the range found for weakly processed ices in massive ysos ( 38% , * ? ? ?
all of these pieces of evidence suggest that @xmath11 is most likely an embedded massive yso .
@xcite estimate a kinematic distance to the maser source as 4.3 or 11.1kpc from its velocity ( @xmath36kms@xmath4 ) .
an x - ray source has also been detected by _ bepposax _ in this region ( source b in * ? ? ?
source @xmath11 is located within the error circle of the position of the x - ray source ( # 4 in fig .
[ fig1 ] ) .
@xcite suggest that if the x - ray source is the hard x - ray counterpart of the uc , the shorter distance is favored since the x - ray luminosity becomes too high as an uc at 11.1kpc .
the area from which the spectrum of @xmath11 is extracted includes greenish emission as fig .
[ fig1 ] suggests and thus the spectrum may have a contribution from co emission .
the prism spectrum of @xmath11 suggests a peak at around 4.5@xmath0 m . it may mimic the r branch of co emission seen in the spectrum of @xmath12 .
however , the prism spectrum does not show a p - branch peak at around 4.7@xmath0 m clearly as seen in @xmath12 .
therefore , we conclude that the peak at around 4.5@xmath0 m is a result of a combination of the saturation effect and the cold co gas absorption and that there is no clear evidence for warm co emission in the spectrum of @xmath11 .
source @xmath12 shows a quite distinct spectrum .
it can be interpreted in terms of highly blue - shifted co emission with red continuum . in the following ,
we discuss two possibilities for its nature : ( 1 ) outflow associated with the yso and ( 2 ) supernova ejecta associated with a snr . excess emission at 4.5@xmath0 m in egos
has often been discussed in terms of outflows associated with massive ysos .
since @xmath11 is most likely a massive yso , @xmath12 could be related to outflow associated with @xmath11 .
observations with the short - wavelength spectrometer ( sws ) on board the infrared space observatory ( _ iso _ ) detect the fundamental ro - vibrational ( @xmath5 ) emission of co molecules toward the omc-1 outflow region together with emission of recombination lines and h@xmath1 lines @xcite .
the sws observations of omc-1 resolve each transition of the co emission and thus the physical conditions of the co gas and excitation mechanism are investigated in detail @xcite .
the observed spectra show stronger emission of the p branch than that of the r branch .
@xcite conclude that the co emission consists of warm ( @xmath37k ) and hot ( @xmath38k ) components and the p - r - asymmetry can be accounted for by a large co column density ( @xmath39@xmath24 ) .
detection of the co ro - vibrational emission toward omc-1 supports the possible association of @xmath12 with outflow from @xmath11 .
however , the characteristics of omc-1 co emission are different from the spectra of @xmath12 .
there seem to be no recombination lines and h@xmath1 lines in the grism spectrum of @xmath12 , which are seen in omc-1 .
there is excess emission at around the position of 0 - 0 s(11 ) ( 4.1810@xmath0 m ) in the grism spectrum of @xmath12 , but the absence of similar excess emission at 0 - 0 s(9 ) ( 4.6847@xmath0 m ) , 0 - 0 s(10 ) ( 4.4096@xmath0 m ) , and 0 - 0 s(13 ) ( 3.8464@xmath0 m ) suggests that the excess at @xmath40@xmath0 m is not h@xmath1 0 - 0 s(11 ) emission . to discuss the implication of non - detection of h@xmath1 line emission in @xmath12
, we make a simple comparison with the spectrum of omc-1 . the total co @xmath5 emission flux at peak 1 of omc-1 is estimated as @xmath41wm@xmath42 and that of the hot component is @xmath43wm@xmath42 @xcite .
the fluxes of the four strongest lines of h@xmath1 at peak 1 in 3.55@xmath0 m , 0 - 0 s(9 ) , 0 - 0 s(10 ) , 0 - 0 s(11 ) , and 0 - 0 s(13 ) , are estimated as 3.2 , 0.83 , 1.4 , and @xmath44wm@xmath42 , respectively @xcite . the total co @xmath5 emission intensity from @xmath12
is estimated as @xmath45wm@xmath42sr@xmath4 .
if the same ratios of the h@xmath1 line intensities to the total co emission as in omc-1 are assumed , the expected h@xmath1 line intensities in @xmath12 are 15 , 3.8 , 6.5 , and @xmath46 wm@xmath42sr@xmath4 for 0 - 0 s(9 ) , 0 - 0 s(10 ) , 0 - 0 s(11 ) , and 0 - 0 s(13 ) , respectively .
if the h@xmath1 emission is associated only with the hot co component in omc-1 , the expected intensities are 34 , 8.5 , 14 , and @xmath47 wm@xmath42sr@xmath4 for 0 - 0 s(9 ) , 0 - 0 s(10 ) , 0 - 0 s(11 ) , and 0 - 0 s(13 ) , respectively .
three-@xmath48 upper limits of the line intensity in the @xmath12 grism spectrum are estimated as 1.8 , 1.7 , 4.5 , and @xmath49wm@xmath42sr@xmath4 , respectively , for 0 - 0 s(9 ) , 0 - 0 s(10 ) , 0 - 0 s(11 ) , and 0 - 0 s(13 ) . the upper limit on 0 - 0 s(11 )
is relatively large compared to the others due to the excess emission described above .
all of these upper limits are smaller than the expected intensities estimated from the spectrum of peak 1 of omc-1 .
thus , the h@xmath1 line emission relative to the co emission in @xmath12 are weak compared to that in omc-1 .
@xcite show that a high h@xmath1 density ( @xmath50@xmath8 ) is required to account for the observed flux of the hot component of co emission in omc-1 .
they also find that the derived co column density of the hot component ( @xmath51@xmath24 ) is much larger than that expected from the column density of hot h@xmath1 component ( @xmath52@xmath24 , * ? ? ?
* ) , suggesting partial dissociation of h@xmath1 in the region .
@xcite also show that co emission becomes comparable to h@xmath1 emission in irac band 2 only at high gas densities ( @xmath53@xmath8 ) .
non - detection of h@xmath1 line emission in @xmath12 may thus suggest even more h@xmath1-poor conditions than in omc-1 .
a detailed study of the physical conditions of co and h@xmath1 in @xmath12 is definitely needed by observations with higher spectral resolution to make quantitative discussion . it should be noted that the suggested large blue - shift velocity , which also has to be confirmed by further observations , seems to be difficult to be explained if the co emission of @xmath12 originates from outflow associated with @xmath11 .
spectra of @xmath12 show strong similarity to those detected toward ejecta knots in the snr cas a @xcite .
the suggested blue - shift velocity of @xmath12 is in the range of those found in ejecta of cas a ( @xmath54kms@xmath4 ) and is attributed to the ejecta motion .
the cas a spectra do not show recombination lines and h@xmath1 emission lines either , which can be attributed to hydrogen - poor abundance of the ejecta .
ego g318.05 + 0.09 is located within the snr g318.2 + 0.1 , which is recognized by a large ( @xmath55 ) radio shell @xcite . however , there is no direct evidence that @xmath12 is associated with sn ejecta .
no associated infrared emission has been found for this snr region by the study of the glimpse survey @xcite .
the present ego is a very tiny portion of the entire snr g318.2 + 0.1 and this is very different from cas a , in which extended infrared emission associated with snr ejecta features is clearly detected . toward the region of the snr g318.2
+ 0.1 , only _ beppossax _
observations are available in the x - ray @xcite , which have a large error circle ( fig , [ fig1 ] ) and do not provide clear evidence for the association of @xmath12 with sn ejecta . while the characteristics of the spectra of @xmath12 suggest emission from sn ejecta , there is no other supporting evidence for the identification as ejecta at present . if @xmath12 found in the present observation is truly sn ejecta , this will be the second snr that shows co emission at 4@xmath0 m .
sne are thought to be the major dust suppliers in the interstellar space as well as in the early universe @xcite .
however , mid - infrared observations of sne so far suggest that the mass of the warm dust associated with sne falls below @xmath56@xmath57 , significantly lower than theoretical predictions ( * ? ? ?
* references therein ) .
recent far - infrared observations detect a significant amount ( @xmath580.1@xmath57 ) of cold dust associated with a sn or a snr @xcite .
detection of sne and snrs in the far - infrared is , however , limited to a small number of objects and the dust production yield in sne still remains under debate ( e.g. , * ? ? ?
recent observations also reveal the presence of several molecules in sne or snrs , such as sio @xcite , co @xcite , and arh@xmath59 @xcite .
the presence of co gas is of particular interest since co traps the major dust - constituting element carbon and may reduce the carbon dust formation yield significantly @xcite .
co molecules have been detected in about 10 sne within 340 years after explosion .
cas a is a snr of 340 years old and it is the only snr in which both high - j rotational and ro - vibrational lines of warm co gas have been detected @xcite .
processing and evolution of dust and gas in sn ejecta after the explosion , in particular the effects of the reverse shock , have been becoming an important issue for the understanding of the net dust formation yield in sne @xcite .
@xcite show that dust and molecules in ejecta clumps are destroyed by the reverse shock , whereas co gas can reform in the post - reverse shock gas , but dust clusters do not .
processing of dust and gas in the post - reverse shock clumps depends sensitively on the environmental conditions @xcite . while no information is available of the snr with which @xmath12 is associated , if it is confirmed to be sn ejecta , further studies of @xmath60 will provide important information on the study of the physical and chemical conditions of the ejecta .
the presently available data for ego g318.05 + 0.09 are not sufficient to draw a clear conclusion on the nature of @xmath12 .
spectroscopy with high spectral resolution and good quality are definitely needed to study the physical conditions of the co gas and to confirm the large blue - shift velocity .
future observations with _
jwst_/nirspec will certainly give us crucial information for the identification .
x - ray observations with high spatial resolution and good energy resolution of _ chandra _ will also help identify the nature of @xmath12 .
we carried out near - infrared spectroscopy of the ego g318.05 + 0.09 with _ akari_/irc .
we have detected two distinct infrared sources in the region .
one , @xmath11 , is associated with the uc iras 14498@xmath25856 and identified as an embedded massive yso based on the presence of the absorption features due to h@xmath1o and co@xmath1 ice species together with the associated methanol maser emission .
the other greenish - color source , @xmath12 , shows a quite peculiar spectrum , which has double peaks at around 4.5 and 4.7@xmath0 m .
we attribute the emission to highly blue - shifted ( @xmath3kms@xmath4 ) fundamental ro - vibrational transitions ( @xmath5 ) of co molecules of temperatures of 12003700k with red continuum .
we discuss two possibilities of the nature of @xmath12 : outflow associated with @xmath11 and supernova ejecta .
detection of the co fundamental ro - vibrational emission toward omc-1 supports the association of @xmath12 with outflow from the massive yso @xmath11 , although the characteristics of the co emission are different . from simple comparison with the _
iso_/sws spectra of omc-1 , non - detection of h@xmath1 in @xmath12 suggests a more h@xmath1-poor environment in @xmath12 than in omc-1 .
it also seems to be difficult to reconcile the large blue - shift velocity with the outflow origin .
the strong similarity of the @xmath12 spectra to those toward the ejecta knots in the cas a snr suggests that @xmath12 may be associated with sn ejecta .
ego g318.05 + 0.09 is located within the snr g318.2 + 0.1 , however , there is no supporting evidence for infrared emission associated with this snr .
there is no other evidence to relate @xmath12 with sn ejecta .
observations with high spectral resolution and better quality are definitely needed to draw a clear conclusion on the nature of @xmath12 .
the present observation suggests that the greenish color of egos has various origins and that one ego could contain more than one objects of different types .
it demonstrates significance of infrared spectroscopy for the study of the nature of individual objects .
there might be similar co emission associated in egos .
the authors thank the referee w. reach for very useful comments , which greatly improved the paper .
this work is based on observations with _
akari _ , a jaxa project with the participation of esa .
the authors thank all the members of the _ akari _ project for their continuous support .
they would also like to thank fumihiko usui for his help in the data reduction , bon - chul koo for the information on the snr g318.2 + 0.1 , and franois boulanger for useful discussion .
this research has made use 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 .
the laboratory data of the h@xmath1o ice are obtained from the leiden atomic and molecular database .
tm received financial support from a grant - in - aid for jsps fellows .
aa was supported by the university of tokyo research internship program ( utrip ) of 2014 , which initiated this work .
this work is supported by grants - in - aid for scientific research from the japan society for the promotion of science ( nos .
23103004 and 16h00934 ) .
onaka , t. , lorente , r. , ita , y. , ohyama , y. , tanab , t. , & pearson , c. 2009 , akari irc data user manual for post - helium ( phase 3 ) mission , ver1.1 , http://www.ir.isas.jaxa.jp/akari/observation/support/irc/ idum / irc_idum_p3_1.1.pdf | we present the results of near - infrared ( 2.55.4@xmath0 m ) long - slit spectroscopy of the extended green object ( ego ) g318.05 + 0.09 with _
two distinct sources are found in the slit .
the brighter source has strong red continuum emission with h@xmath1o ice , co@xmath1 ice , and co gas and ice absorption features at 3.0 , 4.25@xmath0 m , 4.67@xmath0 m , respectively , while the other greenish object shows peculiar emission that has double peaks at around 4.5 and 4.7@xmath0 m .
the former source is located close to the ultra compact region iras 14498@xmath25856 and is identified as an embedded massive young stellar object ( yso ) .
the spectrum of the latter source can be interpreted by blue - shifted ( @xmath3kms@xmath4 ) optically thin emission of the fundamental ro - vibrational transitions ( @xmath5 ) of co molecules with temperatures of 120003700k without noticeable h@xmath1 and emission .
we discuss the nature of this source in terms of outflow associated with the young stellar object and supernova ejecta associated with a supernova remnant . |
when some form of disorder is introduced in a quantum many - body system , e.g. , random coupling constants , impurities , or dilution of the degrees of freedom , the interplay between quantum fluctuations and disorder effects can lead to unconventional properties and special quantum phases .
@xcite such phenomena have been the subject of enormous interest in both theoretical and experimental physics , e.g. , anderson localization of non - interacting electrons,@xcite anomalous metal - insulator transition in interacting two - dimensional electron systems,@xcite bose - einstein condensation in disordered boson systems,@xcite which can be experimentally realized , e.g. , in @xmath10he films on substrates with disorder @xcite and in quantum magnets with random couplings.@xcite disorder also likely plays an important role in many strongly - correlated systems with superconducting phases.@xcite going beyond ground - state properties , the low - energy excitations can also change completely in the presence of disorder , @xcite and many - body localization of higher excited states @xcite is now also attracting considerable attention . in this paper
we study the effects of disorder on an equilibrium quantum phase transition @xcite between a magnetic and quantum paramagnetic state in a two - dimensional ( 2d ) quantum spin model with tunable couplings .
we also discuss related `` quantum glass '' physics .
the standard 2d heisenberg model , which serves as the foundation of our understanding of the physics of the insulating state of the high - tc cuprates and many other quantum antiferromagnets,@xcite has been the subject of numerous theoretical and computational studies of disorder effects.@xcite experimental realizations also have been investigated extensively.@xcite we here study ground - state and finite - temperature properties of a class of disordered 2d heisenberg models with the goal of establishing some key bench - mark results for quantum criticality and associated glass phases .
this work has a broader context of disorder in boson systems ( to which spin models can be mapped ) , where randomness in the local potentials and couplings can destroy superfluid long - rang order and bring about different types of glass phases @xcite which are distinct from the corresponding quantum - disordered phases in clean systems .
two classes of glass phases have been identified the bose glass , which is compressible at temperature @xmath11 , and the incompressible mott glass ( mg).@xcite the latter is in general believed to exist only in systems with particle - hole symmetry,@xcite though recent computational work has found behaviors typical for mgs also in the absence of this symmetry @xcite and some theoretical arguments also support its possible existence more generally.@xcite based on a strong - disordered renormalization group method applied to an effective model representing the one - dimensional ( 1d ) bose - hubbard model at large integer filling fraction and local particle - hole symmetry , altman _ et al . _
found that the transition into the mg state is in the kosterliz - thouless universality class with a dynamical exponent @xmath12 .
@xcite iyer _
_ extended the calculations to two dimensions and presented evidence for the existence of the mg phase also in this case.@xcite this result agrees with quantum monte carlo ( qmc ) simulations of the 2d bose - hubbard model with random hopping ( which maintains local particle - hole symmetry ) , where a non - trivial dynamic exponent @xmath13 was found .
@xcite this special incompressible glass phase was also recently claimed to be experimentally realized in a doped quantum magnet ( br - dtn ) .
@xcite although there are many computational works identifying intriguing quantum - critical phenomena and ground state phases in various disordered @xmath0 heisenberg quantum spin models,@xcite so far the mg phase has not been the focus on such studies , except for a diluted spin-1 model investigated in ref . .
based on general arguments for the existence of griffiths phases adjacent to critical points in disordered systems,@xcite one should also expect mg phases in spin - isotropic @xmath0 spin system ( which have , in boson language , particle - hole symmetry ) with critical points when randomness is introduced in the couplings in some way .
we here present a detailed study of a class of disordered 2d dimerized @xmath0 quantum spin models , confirming the existence of an mg phase and studying its properties both at @xmath7 and @xmath14 .
in addition to uncovering mg physics , another key aspect of this work is to study the influence of disorder on the quantum - critical scaling behavior . in a pioneering study of the effects of disorder at phase transitions
, harris derived a criterion for the critical exponents based on a classical ising model.@xcite according to the harris criterion , the critical behavior ( universality class ) will not be influenced by disorder if the correlation length exponent @xmath15 satisfies @xmath16 , where @xmath17 is the dimensionality of the system .
if the inequality does not hold , the universality class should change upon the introduction of disorder , so that @xmath16 applies ( i.e. , the disorder is a relevant perturbation ) .
the harris criterion was later rederived under more general conditions,@xcite applying also to a wide range of quantum systems . for a quantum system
it is believed that the dimensionality @xmath17 to use in the inequality should be the spatial dimensionality @xmath17 ( if the disorder is introduced symmetrically in all the spatial dimensions ) , i.e. , not the space - time dimensionality @xmath18 representing the effective statistical - mechanical system obtained in a path - integral representation at @xmath7.@xcite the effective system has columnar disorder with constant couplings along the time dimension .
many studies of classical systems with disorder have been in agreement with the harris criterion . in a work of relevance to 2d heisenberg quantum spins , a @xmath19-dimensional heisenberg model with columnar dilution
was studied and it was found that @xmath20,@xcite while in the clean system @xmath21 of the 3d o(3 ) universality class . however in a class of 2d @xmath0 dimerized quantum spin models with random arrangements of weaker ( intra - dimer ) and stronger ( inter - dimer ) interactions,@xcite no change in the exponents with respect to to the clean 3d o(3 ) exponents were found , and the harris criterion is then violated .
possible violations of the harris criterion in a disordered boson system were discussed in refs . .
the general validity of the harris criterion has also been questioned theoretically,@xcite and as of now there does not appear to be any rigorous way to establish _ a priori _ ( without elaborate explicit calculations of critical exponents ) whether a particular kind of disorder is relevant or not .
our studies of criticality and mg physics reported in this paper were carried out within a dimerized @xmath0 heisenberg model starting from the well - studied columnar dimer pattern.@xcite in a clean system with inter- and intra - dimer couplings @xmath22 and @xmath23 , the nel order of the ground state existing for or @xmath24 vanishes when @xmath25 and the excitation spectrum becomes gapped .
universality of this transition in the o(3 ) class has been confirmed to numerical precision rivaling that in studies of classical models.@xcite we here introduce randomness in the intra - dimer couplings , using a total of three different antiferromagnetic nearest - neighbor interactions , @xmath2 , @xmath3 and @xmath4 , with @xmath1 and setting the inter - dimer coupling @xmath26 .
the stronger , intra - dimer couplings again form columns , with random placements of @xmath23 and @xmath27 bonds within these columns . fig .
[ modelf ] shows an illustration of the system . using the stochastic series expansion ( sse ) qmc method,@xcite we find an mg phase with magnetic susceptibility vanishing as @xmath28 , following the form @xmath8 , with @xmath29 depending on the parameters ( and @xmath30 in the standard gapped phase as in the clean system ) . using various quantities for finite - size scaling
we find no detectable differences between the clean and the random system .
the correlation length exponent @xmath31 , and the harris criterion is therefore violated .
this rest of the paper is organized as follows . in sec .
[ sec : model ] we define the random columnar spin model in detail and discuss the physical observables studied and how we approach the @xmath32 limit in the qmc simulations . in sec .
[ sec : critical ] we present finite - size scaling studies to extract the critical point and the critical exponents .
the susceptibility in the mg phase is discussed in sec .
[ sec : mg ] , followed by a brief summary and discussion in sec .
[ sec : disc ] .
the hamiltonian of the disordered antiferromagnetic columnar - dimerized heisenberg model can be simply written as @xmath33 where @xmath34 is an @xmath0 spin operator residing on an @xmath35 square lattice with even @xmath36 and periodic boundary conditions .
the index pairs @xmath37 denote nearest - neighbor sites with corresponding coupling strengths @xmath38 . as illustrated in fig .
[ modelf ] , the bonds where @xmath39 form columns separated by one bond and the assignment to @xmath23 or @xmath27 is done at random with probability @xmath40 for each choice .
all other couplings are assigned the inter - dimer value @xmath26 . in order to more conveniently describe the intra - dimer couplings in terms of an overall strength and a fluctuation , we parametrize them according to @xmath41 so that the average is @xmath42 and the difference @xmath43 .
normally we chose @xmath44 and the parameter @xmath45 serves as a measure of the strength of the disorder . in the clean limit @xmath46 , when @xmath47 our model reduces to the isotropic heisenberg model , which is nel - ordered at @xmath7 ( but not at @xmath14 by the mermin - wagner theorem ) , while for @xmath48 the ground state is a product of dimer singlets and there is no magnetic order .
several qmc studies have been devoted to the quantum phase transition between these two limiting cases,@xcite which takes place at @xmath49 according to the most precise calculation.@xcite in this paper we focus on a fixed rather strong disorder strength , @xmath50 , where at the critical point the strong to weak dimer ratio is @xmath51 .
we use the sse qmc method to compute several physical quantities averaged over @xmath52 realizations of the random bonds in each case .
we here define the physical quantities studied in this work .
the spin stiffness @xmath53 characterizes the response to twisting the magnetic order , similar to an elastic constant .
it is defined as @xmath54 where @xmath55 is the free energy of the system and @xmath56 is the angle of a twist imposed between two column of spins ( in the @xmath57 or @xmath58 direction of the lattice ) .
the stiffness is evaluated in sse simulations using winding number fluctuations.@xcite another important quantity characterizing the magnetic state is the static spin structure factor @xmath59 , i.e. , the fourier transform of the spin - spin correlation function at wave - vector @xmath60 @xmath61 where @xmath62 is the correlation function @xmath63 which is a function only of the separation between the two spins after disorder - averaging .
further , we study the magnetic susceptibility in real space , i.e. , the linear response of a spin @xmath64 to a field coupled to spin @xmath65 ; @xmath66 where @xmath67 . again , after disorder averaging this susceptibility depends only on the separation @xmath68 of the two spins and we again carry out the fourier transform to wave - vector @xmath69 .
both the structure factor and the susceptibility have simple sse estimators , in the case of the susceptibility based on computing the integration over @xmath70 exactly.@xcite we here present results at the ordering wave - vector @xmath71 , and , in the case of the susceptibility also at @xmath72 where the response to a uniform field reduces to @xmath73 because of the hamiltonian conserving the total magnetization . in order to use finite - temperature qmc simulations to investigate the ground state
, it is necessary to make sure that the @xmath7 convergence is achieved or , when studying a continuous quantum phase transition , to scale the inverse temperature @xmath74 with the system size using the proper dynamic exponent ; @xmath75 . in this case
we do not _ a priori _ know the value of @xmath76 and it is therefore better to converge to the true ground state of the system .
[ converg1 ] and [ converg2 ] show examples of the convergence with increasing @xmath77 of the quantities we discussed above for systems with @xmath78 . here
the error bars primarily reflect sample - to - sample fluctuations affecting the disorder averages .
the same disorder realizations were used for all the temperatures , which implies that the statistical errors at different temperatures are strongly correlated ( again because the errors are dominated by sample - to - sample fluctuations , not the statistical errors of the individual simulations ) , which is clearly visible in the data in figs .
[ converg1 ] and [ converg2 ] .
the ground state of the model always being a singlet , the uniform susceptibility vanishes as @xmath32 , and we will only analyze its @xmath14 behavior . for the other quantities
we find that @xmath79 is sufficient ( for the system sizes considered here ) to achieve reasonable convergence to the ground state , in the sense that remaining temperature effects are much smaller than the statistical errors in the disorder averaging .
this scaling of @xmath77 is also of course suitable for studying @xmath12 criticality ( which is in the end what we will find for the dynamic exponent ) , regardless of any remaining finite - temperature effects .
as eluded to above , even for relatively short sse simulations of individual disorder realizations , the final statistical averages are dominated by the sample - to - sample fluctuations , not the fluctuations of the individual sse averages .
it is therefore desirable to perform short simulations of a large number of samples . in order to avoid potential problems of poorly equilibrated simulations
it is then useful to use the _ @xmath77-doubling procedure_,@xcite where an almost equilibrated configuration to start from in a simulation at inverse temperature @xmath77 is obtained from a prior simulation at @xmath80 by combining two copies of the last sampled configuration into a single configuration with twice the extent in the imaginary time direction . for discussion of this
approach and typical equilibration times we refer to ref . .
we will also discuss a set of calculations aiming at reaching the thermodynamic limit for @xmath14 . in this case , as @xmath81 is lowered , increasingly large system sizes have to be used in order to achieve infinite - size convergence .
this convergence can not easily be achieved for divergent quantities ( at criticality and in the nel phase ) at low @xmath81 , because @xmath36 has to exceed the exponentially divergent @xcite correlation length .
however , for non - divergent quantities such as the uniform susceptibility convergence can be achieved at relatively low temperatures , as we will discuss later .
in this section we use results of large - scale sse studies with the @xmath77-doubling scheme to study critical properties of the ground state of the system at disorder strength @xmath50 .
we use several finite - size scaling approaches to locate the critical coupling ratio @xmath82 where the magnetic long - range order vanishes and extract the dynamical exponent @xmath76 along with with the equilibrium critical exponents @xmath15 and @xmath83 .
we also study the magnetic susceptibility as a function of temperature in the thermodynamic limit .
a common way to locate quantum phase transitions by finite - size scaling is to examine a quantity which is expected to be independent of the system size at the critical point .
a useful quantity for analyzing the loss of magnetic order in a spin systems is the spin stiffness , eq .
( [ spinstiffness ] ) , which close to criticality should obey the scaling form @xmath84 , \label{fsscform}\ ] ] where @xmath15 is the correlation length exponent ( see ref . for a review of the above form and other scaling behaviors of spin systems discussed below ) .
this form holds for the ground state , @xmath85 , and also for low but non - zero temperatures if @xmath77 is scaled proportionally to @xmath86 .
the properly size - scaled stiffness @xmath87 should then be independent of @xmath36 when @xmath88 , provided that the system sizes are sufficiently large for corrections to scaling to be negligible .
if @xmath76 is known , one can then graph @xmath89 versus @xmath82 for several system sizes and examine crossing points of curves for different @xmath36 . in the presence of scaling corrections ( which can not be avoided in practice )
these crossing points will only tend toward the critical point when @xmath90 . for our system at hand here
, we do not _ a priori _ know the value of @xmath76 , but it should be noted that even if the wrong value is used there will still be curve crossings and these will still tend toward the correct critical point when @xmath91 .
however , the points will drift in the vertical direction instead of converging to a stationary point .
for now we use @xmath12 scaling with sse data obtained at inverse temperature @xmath79 , which , as we discussed in sec . [ sec : conv ] , produces results essentially converged to the ground state for the system sizes and @xmath82 values we are considering . with practically converged ground state results we are effectively in the @xmath85 regime and can also test scaling using any value of @xmath76 ( i.e. , without scaling @xmath75 in different simulations for each @xmath76 tested ) . in fig .
[ cross ] we use @xmath12 and plot disorder - averaged results for @xmath92 versus @xmath82 for different @xmath36 and @xmath93 $ ] .
the crossing point is well defined even for rather small systems , with only minor drifts that we will examine in detail below .
the uniform susceptibility @xmath94 is also a very useful quantity to consider in finite - size scaling . at the critical point
we expect @xcite @xmath95 for a 2d system . here
it should again be noted that @xmath96 vanishes for finite @xmath36 when @xmath85 , which implies that in this case ground state results are useless .
instead , the above equation applies for results obtained with @xmath75 . as demonstrated in fig .
[ converg2 ] , for @xmath79 the susceptibility is still clearly non - vanishing and we can attempt @xmath12 scaling . in fig . [ critical ] , panel ( a ) , we graph @xmath97 in the region @xmath98 $ ] for different @xmath36 , and for comparison also replot the stiffness data of fig .
[ cross ] for @xmath82 within this smaller window .
some drifts of the crossing points can be seen in both quantities , more so in @xmath97 than in @xmath99 . to analyze the drifts quantitatively
, we extract crossing points for pairs of systems of size , @xmath100 and @xmath36 . to this end
we fit polynomials of suitable order to the data points and use them to extract the crossing point .
the procedure was repeated @xmath101 times with added gaussian noise ( with standard deviation equal to the sse error bars ) in order to obtain a the statistical errors for the crossing points .
the results for the so obtained finite - size estimates ( crossing points ) @xmath102 are graphed versus @xmath103 in fig .
[ criticalpoint ] .
one in general expects the size corrections to be of the form @xmath104 where @xmath105 is the sum of @xmath106 and a correction exponent .
since the correction in practice is an effective correction containing also higher order contributions we let @xmath105 be different for the two quantities analyzed but the infinite - size critical point @xmath107 is constrained to be the same in both cases .
the fitted functions are also shown in fig . [ criticalpoint ] . estimating the statistical fluctuations of the fits using gaussian noise as above , we finally obtain the critical point value @xmath108 for our model with @xmath50 .
next we analyze the vertical drifts of the crossing points . as shown in fig .
[ criticaly ] the behavior is consistent with a correction linear in @xmath103 , and , thus , there is a well defined crossing point for @xmath91 .
this suggests that in fact the dynamic exponent really is @xmath12 as we assumed in the preceding analysis . to test the value of @xmath76 further
, we carry out a scaling study of the uniform susceptibility in a completely different regime , with simulation data at elevated temperatures and taking the limit @xmath91 . in the thermodynamic limit , in the neighborhood of the critical point
, the susceptibility should follow the form @xcite @xmath109 for @xmath81 above a cross - over temperature when @xmath110 ( and with @xmath111 for @xmath112 and @xmath113 for @xmath112 ) .
exactly at the critical point @xmath114 and the scaling behavior extends to @xmath7 .
the constant @xmath115 is non - universal , depending on a model - specific velocity . in principle
this scaling form allows also for an independent determination of the critical point , in addition to extracting @xmath76 , by locating a @xmath82-value and exponent @xmath76 for which @xmath116 holds .
here we just use the @xmath117 value obtained below and test the scaling with @xmath12 .
first , we check the convergence to the thermodynamic limit .
[ sizeeffect ] shows data sets for different system sizes versus temperature .
the size dependence for fixed @xmath81 is non - monotonic at low @xmath81 . for sufficiently large @xmath36
we expect exponential convergence . as shown in fig .
[ sizeeffect ] , for the largest system studied , @xmath118 , we have almost achieved convergence in the temperature range @xmath119 .
the remaining size effects should be at most the size of a few error bars , which is small on the scale of the overall variations with temperature .
we therefore do not expect any significant distortions by using these data instead of fully converged data .
we plot the @xmath118 data both on log - log and lin - lin scales in fig .
[ zexponent ] .
the behavior shows a remarkable consistency with @xmath12 critical scaling .
a linear fit to the log - log data gives @xmath120 .
moreover , on the lin - lin plot it can be seen that the intercept @xmath121 vanishes ( within the error bars ) , thus also independently confirming the location of the critical point determined above .
our conclusion is , thus , that @xmath76 remains unchanged at unity when disorder is introduced in the system .
this is in contrast to dimer - diluted systems , where different dynamic exponents @xmath122 have been found .
@xcite an unchanged dynamic exponent was also found in dimerized models where the coupling strengths are kept fixed but the configuration of dimers is random.@xcite these results are intriguing as one would normally expect disorder introduced only in the space dimension to destroy the emergent lorenz invariance ( space - time symmetry implying @xmath12 ) present in the clean system . in order to study the disorder effects on the phase transition more completely , we also extract the standard static critical exponents .
we analyze the behaviors of the staggered structure factor @xmath123 and the staggered susceptibility @xmath124 , which are expected to have the following leading size dependencies at the critical point : @xmath125 in order to improve the fits when small systems are included , it is useful to also add corrections to scaling .
we then use the following forms : @xmath126 where of course the subleading exponents @xmath127 and @xmath128 should be smaller than the leading exponents .
the critical susceptibility analyzed only with the leading form ( [ xscale1 ] ) is shown on a log - log plot in fig .
[ linear ] .
an essentially linear behavior is seen , with the slope @xmath129 , i.e. , @xmath130 . without disorder ,
the system is in the standard o(3 ) universality class , where @xmath131.@xcite while the fit visually looks reasonably good , the value of @xmath132 per degree of freedom is close to @xmath133 , indicating an only marginally satisfactory fit .
the quality of the fit can be improved if some of the smaller system sizes are excluded , but then the statistical accuracy of @xmath83 deteriorates .
when we instead include the subleading correction , as shown in fig .
[ scalex ] , the fit improves significantly , with @xmath134 .
interestingly , the exponent then also changes to @xmath135 , which is completely consistent with the above cited best estimate of this exponent in the o(3 ) university class . for the subleading exponent
we obtain @xmath136 , further confirming that the correction is small relative to the leading term .
the above estimate of @xmath83 does not take into account the uncertainty in the critical point . by repeating the fit including the scaling correction at one error bar away from the mean value @xmath137 , i.e. , at @xmath138 and @xmath139 , the leading exponent @xmath140 changes substantially , as also seen in fig .
[ scalex ] . based on the susceptibility
we can then only conclude that @xmath83 is roughly within the range @xmath141 $ ] .
moving now to the staggered structure factor , when plotting it on a log - log scale the slope should be @xmath142 according to eq .
( [ sscale1 ] ) . using the o(3 ) value of @xmath83 and @xmath12
we should have @xmath143 .
we present our sse data along with the best - fit line on a log - log scale at @xmath144 in fig .
[ scalesf ] .
the slope is @xmath145 , a bit smaller than expected , and @xmath146 . in this case
the subleading correction does not improve the fit statistically , but this of course does not exclude that there are some effects of corrections . repeating the fit at @xmath117 plus and minus one error bar again introduces a fluctuation in the slope , and based on all these fits ( with the assumption that @xmath12 ) we have @xmath147 , in excellent agreement with the best @xmath83 values calculated for the o(3 ) class .
note that the primary reason why the statistical accuracy of @xmath83 is better when extracted from fits to @xmath123 than @xmath124 is that no correction to scaling is required in the former case .
the second important static exponent is the correlation length exponent @xmath15 , which governs the divergence of the correlation length and is accessible in finite - size scaling through the argument of scaling functions such as eq .
[ fsscform ] .
this exponent is at the heart of the harris criterion , according to which a unique transition point is possible in the presence of disorder only if the exponent satisfies @xmath148 , where in quantum systems it is believed that @xmath17 should be the spatial dimensionality ( with there being no disorder in the time direction in the path integral representation ) .
thus , the introduction of disorder in a clean system with @xmath149 should lead to a new universality class in which the harris inequality is satisfied .
we have shown above that the exponents @xmath76 and @xmath83 in the @xmath150 model are not changed by the disorder and we now examine @xmath15 .
based on the scaling function ( [ fsscform ] ) for the spin stiffness and assuming @xmath12 , plotting @xmath151 versus @xmath152 , curves for different @xmath36 should collapse onto a common scaling function for large @xmath36 . based on the already extracted exponents
@xmath76 and @xmath83 it seems plausible that the universality class remains the 3d o(3 ) class , and we therefore test data collapse with the corresponding value of @xmath15 , for which the best estimate to our knowledge is from ref .
, @xmath153 .
the result of this procedure is a statistically good collapse of the data , as illustrated in the upper panel of fig .
[ nu ] . to quantify the goodness of the collapse
, we fit a polynomial to all the data points , and this has a satisfactory value of @xmath134 .
if the harris inequality hods , the smallest possible value of the exponent is @xmath154 .
the lower panel of fig .
[ nu ] shows the result of an attempted data collapse with this value .
while some of the data do fall close to a common curve , there are also segments of points that deviate very clearly , and this is quantified with a statistically unsatisfactory goodness of a fit of a polynomial , which has @xmath155 .
thus , while our data do not allow a very precise estimate of @xmath15 , we can clearly see that the harris criterion is violated .
in the previous section we studied the point at which the antiferromagnetic long - range order vanishes in a system with random coupling constants . in disordered systems one in general expects a griffiths phase following this kind of critical point . in the griffiths , or glassy ( here quantum glassy )
phase there are arbitrarily large clusters of the ordered phase inside a background of the disordered phase , leading to singular behaviors .
the critical point extracted in the previous section is that separating the quantum glass and the nel state .
the boundary between the glass state and the eventual quantum paramagnetic state is not easy to determine , because the griffiths singularities are due to large clusters , which are very rare close to the phase boundary .
we here study the finite - temperature behavior of the uniform magnetic susceptibility inside the glass phase and show that it has a particular behavior that in principle should be useful for also detecting the glass paramagnetic phase boundary
. in the quantum paramagnetic phase the gapped magnons in combination with the 2d density of triplets leads to a pure exponential form of the susceptibility , @xmath156 where @xmath157 is the singlet - triplet gap . as we discussed in sec .
[ sec : intro ] , the glass phase of a spin - isotropic heisenberg system is expected to be of the mott type , i.e. , gapless but with the susceptibility still vanishing as @xmath28 . to our knowledge ,
the only quantitative study of the form of the compressibility in a 2d spin system is in ref . , where in a diluted @xmath158 model the stretched exponential form @xmath159 was found , with @xmath160 .
no theoretical explanation of this form was provided . in ref .
, in a study of the compressibility of a disordered boson system , the same general form was found but with @xmath29 varying ( @xmath161 ) , and it was argued that this form follows naturally from a scenario where there is a distribution of cluster sizes of the non - gapped phase inside the gapped phase , with the probability of a site depending to a cluster of a given size @xmath162 decaying exponentially with @xmath162 ( as in percolation theory below the percolation threshold ) , and if the finite - size gaps of those clusters depend as a power - law on their size . for the random dimerized heisenberg system we here also find the stretched exponential form with varying @xmath29 , and
it seems very plausible that this form here is also due to the same conditions as mentioned above ( with finite - size gapped nel - phase clusters in a background of the dominant quantum - paramagnetic phase ) . to study the @xmath81 dependent susceptibility we again first investigated finite size effects and in results presented here go down only to temperatures where we can avoid them .
when the length of the system is larger than @xmath78 finite - size effects can be excluded down to @xmath163 .
the temperature dependence of @xmath164 is shown in fig .
[ glassphasetrans ] for this system size and different @xmath82 values . when @xmath165 , deep inside the quantum paramagnetic phase , the standard exponential form ( [ suscrelation ] ) applies , while in the nel phase the temperature dependence is very weak at low @xmath81 . between these well - understood regimes
we find a clearly decreasing susceptibility as @xmath28 , but not with the pure exponential form ( [ parasusc ] ) . as shown in detail for three @xmath82-values in fig .
[ glassphase ] , we instead see the stretched exponential form ( [ suscrelation ] ) with the exponent @xmath29 depending on @xmath82 .
this form indicates that the system is gapless but there is no magnetic order ( in which case the susceptibility must be non - zero for @xmath32 by hydrodynamic arguments @xcite ) .
the stretched exponential should then be a characteristic of the mg state .
it is not easy to locate the phase boundary between the mg to quantum paramagnet , but in principle a detailed study of the exponent @xmath29 versus @xmath82 should allow for this . systems .
when @xmath166 ( in the nel phase ) , the curves flatten out and converge to a non - zero constant ( flat line ) .
for @xmath167 the @xmath81-linear behavior applies , as shown with the fitted curve .
when @xmath168 a standard exponential decay @xmath169 is found ( fitted curve shown ) , indicating a gapped quantum paramagnet , while for smaller @xmath170 there are significant deviations from the exponential form ( the lines are here drawn to go through the data for the higher temperatures ) , suggesting the system is a mott glass.,width=321 ]
we have introduced a dimerized 2d antiferromagnetic heisenberg model with three coupling constants @xmath171 , where the @xmath23 and @xmath27 couplings reside randomly on bonds forming columns on the simple square lattice .
we have studied the ground state phases and quantum phase transitions of the model using unbiased sse qmc simulations on lattices with several thousand spins . using finite - size and finite - temperature scaling of several different quantities
, we have shown that , in spite of the presence of relatively strong disorder ( the ratio of the two intra - dimer couplings being @xmath172 at the critical point ) , the universality class of the system remains in the same standard o(3 ) universality class as clean dimerized heisenberg models . the harris criterion for relevance of disorder
is therefore violated .
violation of the harris criterion was previously found in a dimer system with no randomness in the values of intra- and inter - dimer couplings , but with the dimers arranged randomly.@xcite evidence is this mounting for very generic non - applicability of the harris criterion in a broad class of dimerized heisenberg models . to our knowledge ,
the only case where the harris criterion was shown to be valid in dimerized quantum spin models was for a bilayer where some fraction of dimers were completely removed in a random fashion.@xcite naively one might not expect these different cases of disorder to lead to different universality classes ( since the symmetries are the same ) and the reasons why they are different certainly deserve further examination . given the loopholes @xcite of the harris original @xcite and generalized @xcite analysis of critical points in the presence of disorder , it is now becoming clear that the issue of relevance or irrelevance is much more complex than previously believed .
we also found a mott glass phase by tuning the disorder and interaction strength beyond the critical point where the nel order vanishes . here
the system is incompressible , as in the quantum paramagnet reached eventually , but there is no magnetic order and the susceptibility follows a stretched exponential form , @xmath8 with @xmath9 , instead of the normal exponential form dictated by the gap in the quantum paramagnetic phase ( i.e. , @xmath173 and @xmath115 equals the singlet - triplet gap ) .
the natural interpretation of the stretched exponential is that , within the griffiths scenario of domains of nel - phase clusters inside the quantum paramagnet , the clusters have finite size gaps decaying as a power - law with the size of the clusters ( as has been argued also in the context of a disordered boson model.@xcite . in a previous study of a disordered @xmath158 system
the same kind of stretched exponential form was found,@xcite but in that case @xmath160 was always observed . in our case
the exponent is clearly varying , as was found also in the glass phase of disordered bosons.@xcite we would like to thank wenan guo for stimulating discussions .
nsm and dxy acknowledge support from national basic research program of china ( 2012cb821400 ) , nsfc-11074310 , nsfc-11275279 , specialized research fund for the doctoral program of higher education ( 20110171110026 ) , and ncet-11 - 0547 .
the work of aws was supported by the nsf under grants no .
dmr-1104708 and phy-1211284 , and he also gratefully acknowledges support from sun yat - sen university for visits during which part of this research was completed .
r. yu , l. yin , n. s. sullivan , j. s. xia , c. huan , a. paduan - filho , n. f. oliveira jr , s. haas , a. steppke , c. f. miclea , f. weickert , r. movshovich , e .-
mun , b. l. scott , v. s. zapf , t. roscilde , and a. kitaev , nature * 489 * , 379 ( 2012 ) . | we use quantum monte carlo simulations to study a disordered @xmath0 heisenberg quantum spin model with three different nearest - neighbor interactions , @xmath1 , on the square lattice .
we consider the regime in which @xmath2 represents weak bonds , and @xmath3 and @xmath4 correspond to two kinds of stronger bonds ( dimers ) which are randomly distributed on columns forming coupled 2-leg ladders . when increasing the average intra - dimer coupling @xmath5 , the system undergoes a n@xmath6el to quantum glass transition of the ground state and later a second transition into a quantum paramagnet .
the quantum glass phase is of the gapless mott glass type ( i.e. , in boson language it is incompressible at temperature @xmath7 ) , and we find that the temperature dependence of the uniform magnetic susceptibility follows the stretched exponential form @xmath8 , with @xmath9 . at the nel glass transition we observe the standard o(3 ) critical exponents , which implies that the harris criterion for the relevance of the disorder is violated in this system . |
the measurement of the solar and atmospheric neutrino oscillation paramaters has provided compelling evidence for physics beyond the standard model ( sm ) of elementary particles .
massive active neutrinos can be naturally accounted for in see - saw type extensions of the sm , where new fermion and/or scalar representations are introduced in the theory with suitable yukawa couplings to the sm lepton doublets @xcite .
the mass of the new physical states is in general unrelated to the electroweak ( ew ) symmetry breaking scale and , therefore , can assume arbitrary large values up to the planck scale . on the purely phenomenological side ,
it is interesting to study see - saw scenarios in which new physics is manifest at the tev scale and can be probed in collider searches , lhc included . in this physical context , the phenomenology of type i see - saw extensions has been studied in detail in @xcite , in a model independent way .
the new particle states in such scenarios consist of at least two heavy sm - singlet fermions , which are conventionally denoted as right - handed ( rh ) neutrinos , @xmath0 ( @xmath1 ) , and give rise , when ew symmetry is broken , to the following mass terms in the lagrangian @xmath2 where @xmath3 ( @xmath4 ) , @xmath5 is the @xmath6 majorana mass matrix of the rh neutrinos and @xmath7 provides the @xmath8 neutrino dirac mass term .
the majorana mass @xmath9 for the active left - handed neutrinos is given by the well known see - saw relation : @xmath10 .
after the diagonalization of the full mass matrix given in ( [ typei ] ) , the charged current ( cc ) and neutral current ( nc ) weak interactions involving the heavy majorana mass eigenstates @xmath11 ( @xmath12 ) can be expressed as @xcite : @xmath13 with @xmath14 at leading order in the see - saw expansion and @xmath15 . the couplings @xmath16 can in principle be sizable , typically @xmath17 for @xmath18 gev .
then , in order to reproduce small neutrino masses via the see - saw mechanism , a `` large '' contribution to @xmath9 from @xmath19 is _ exactly _ cancelled by a negative contribution from a second rh neutrino , say @xmath20 , provided : @xmath21 where @xmath22 is the mass of the rh neutrino @xmath23 . barring accidental cancellations ,
relation ( [ rel0 ] ) is naturally fulfilled in models where an approximately conserved lepton charged exists . in such scenarios _ @xmath19 and @xmath20 form a pseudo - dirac pair and the neutrino oscillation parameters fix the flavour structure of their weak cc and nc couplings to gauge bosons and charged leptons , up to an overall constant @xmath24 _ ( see @xcite for a details ) : @xmath25 where @xmath26 , @xmath27 and @xmath28 are the light active neutrino masses in the case of a normal / inverted hierarchical ( nh / ih ) mass spectrum .
the mass splitting of the two rh neutrinos is highly constrained from the experimental upper limits set in neutrinoless double beta ( - ) decay experiments .
indeed , in this case the effective majorana mass @xmath29 , which controls the -decay rate , receives an additional contribution from the exchange of the heavy majorana neutrinos @xmath30 , which may be sizable / dominant for `` large '' couplings @xmath31 . for @xmath32 ,
given a nucleus @xmath33 , one has ( see @xcite for details ) : @xmath34 where for @xmath35 gev : @xmath36 , @xmath37 gev and @xmath38 .
using eq .
( [ rel0 ] ) , the @xmath30 contribution to the effective majorana mass is simply @xmath39 with @xmath40 . in the case of _ sizable _ couplings of rh neutrinos to the charged leptons , @xmath41 @xmath42 , this contribution can be even as large as @xmath43 ev for @xmath44 and @xmath45 gev @xcite . and
@xmath20 form a pseudo - dirac pair .
notice that this conclusion is valid even in the case in which there is no conserved lepton charge in the limit of zero splitting at tree level between the masses of the pair @xcite . ] an effective majorana mass of this order of magnitude may take place in both types of neutrino mass spectrum and can be accessible in ongoing experiments looking for -decay ( @xmath41 the gerda experiment @xcite , which can probe values of @xmath29 up to @xmath46 ev ) .
[ cols="^,^ " , ] in the scenario under discussion , lepton flavour radiative decays allow to put further constraints on the size of the mixing between light and heavy majorana neutrinos .
the strongest bounds are obtained from the current upper limit on @xmath47 branching ratio @xcite : @xmath48_{\mu j}^ { * } \,\left[\left(1+\eta\right)u \right]_{e j } g\left(\frac{m_{j}^{2}}{m_{w}^{2}}\right ) + \sum\limits_{k=1}^{2 } \left ( rv\right)_{\mu k}^ { * } \left ( rv\right)_{e k } g\left(\frac{m_{k}^{2}}{m_{w}^{2}}\right)\nonumber\\ & & \;\;\;\;\cong\;2\left [ ( rv)_{\mu 1}^{*}\ , ( rv)_{e1}\right ] \left [ g(m_1 ^ 2/m_w^2 ) - g(0)\right]\ , , \label{t2}\end{aligned}\ ] ] where @xmath49 .
the last relation arises from ( [ rel0 ] ) and taking into account that @xmath50 , because of @xmath29 upper limit .
therefore , taking @xmath51 at 90% c.l . from meg experiment @xcite , the following constraint for @xmath52 ( @xmath53 tev )
is derived @xcite @xmath54 this can be recast as an upper bound on the neutrino yukawa parameter @xmath24 ( see figure [ fig2 ] ) @xcite : @xmath55
vs @xmath29 for @xmath56 gev and @xmath57.[fig1],width=434,height=283 ] since the flavour structure of the neutrino yukawa couplings is fixed in the present scenarios , correlations among different low energy leptonic observables may be a relevant signature of tev scale type i see - saw mechanism .
indeed , in the simple extension of the standard model considered , with the addition of two heavy rh neutrinos @xmath19 and @xmath20 at the tev scale , which behave as a pseudo - dirac particle , _ a sizable ( dominant ) contribution of @xmath19 and @xmath20 to the @xmath58-decay rate would imply a `` large '' enhancement of the muon radiative decay rate_. in fact , if @xmath59 , where @xmath60 is given in eq .
( [ mee3 ] ) , it is easy to show that @xcite @xmath61 where @xmath62 ( @xmath63 ) for the nh ( ih ) light neutrino mass spectrum . the analytic relation in eq .
( [ bmeg ] ) is confirmed by the results of the numerical computation reported in figure [ fig1 ] , where it is shown the correlation between the @xmath47 branching ratio and the effective majorana mass in the case of sizable couplings between the rh ( pseudo - dirac pair ) neutrinos and charged leptons . in general , a lower bound on @xmath64 within the meg experiment sensitivity reach is set for both light neutrino mass hierarchies ( normal and inverted ) if a positive signal is detected by gerda , @xmath65 for @xmath66 ev . in conclusion , the observation of @xmath58-decay in the next generation of experiments , under preparation at present , and of the @xmath67 decay in the meg experiment ,
could be the first indirect evidence for the tev scale type i see - saw mechanism of neutrino mass generation .
9 p. minkowski , phys .
b * 67 * ( 1977 ) 421 ; m. gell - mann , p. ramond and r. slansky , _ proceedings of the supergravity stony brook workshop _ , new york 1979 ,
eds . p. van nieuwenhuizen and d. freedman ; t. yanagida , _ proceedinds of the workshop on unified theories and baryon number in theuniverse _ , tsukuba , japan 1979 , ed.s a. sawada and a. sugamoto ; r. n. mohapatra and g. senjanovic , phys .
* 44 * ( 1980 ) 912 ; w. konetschny , w. kummer , phys .
* b70 * ( 1977 ) 433 ; t. p. cheng , l. -f .
li , phys . rev . *
d22 * , 2860 ( 1980 ) ; g. lazarides , q. shafi , c. wetterich , nucl .
* b181 * , 287 ( 1981 ) ; j. schechter , j. w. f. valle , phys .
rev . * d22 * , 2227 ( 1980 ) ; r. n. mohapatra , g. senjanovic , phys .
* d23 * , 165 ( 1981 ) .
a. ibarra , e. molinaro and s. t. petcov , jhep * 1009 * ( 2010 ) 108 [ arxiv:1007.2378 [ hep - ph ] ] .
a. ibarra , e. molinaro and s. t. petcov , phys .
d * 84 * ( 2011 ) 013005 [ arxiv:1103.6217 [ hep - ph ] ] .
a. a. smolnikov [ gerda collaboration ] , arxiv:0812.4194 [ nucl - ex ] .
j. adam _ et al . _
[ meg collaboration ] , [ arxiv:1107.5547 [ hep - ex ] ] . | we consider the low energy constraints that can be applied to type i see - saw extensions of the standard model in which the right - handed neutrinos are taken at the electroweak scale . in the reported scenarios , the flavour structure of the charged current and neutral current weak interactions of the standard model leptons with the heavy right - handed neutrinos
is essentially determined by the neutrino oscillation parameters . in this case
, correlations among different measurable phenomena in the lepton sector may provide compelling indirect evidence of low energy see - saw mechanism of neutrino mass generation . |
form factors of the electromagnetic processes @xmath7 ( @xmath8 ) provide information on the strength of the electromagnetic amplitude in @xmath9 decays .
such information is indispensable to separate the strong interaction amplitude for other @xmath9 decay modes , and even more importantly , to determine the relative phase between the electromagnetic and strong amplitudes @xcite .
this phase has been found to be orthogonal in virtually all @xmath10 decays , such as @xmath11 @xcite , @xmath12 @xcite , @xmath13 @xcite , @xmath14 @xcite , and @xmath15 @xcite modes .
it is suggested that this orthogonality also exists in @xmath16 decays and is universal @xcite .
the born - order cross section for the electromagnetic process @xmath17 @xmath18 in continuum production can be expressed as @xcite @xmath19 , @xmath20 is the momentum of either the vector or pseudoscalar meson in the vp decay , and the form factor @xmath21 is an @xmath22 dependent variable @xcite . by measuring the cross section of @xmath23 in continuum production and correcting for the effect of initial state radiation , the form factor @xmath24 can be determined .
in contrast , for @xmath23 at the @xmath16 peak , the total cross section includes contributions from both resonance production and decay and the continuum process , although the interference between them can be neglected @xcite . by separating the one - photon annihilation and @xmath16 resonance contributions , both the form factor at resonance and the branching fractions of @xmath25 can be obtained .
the branching fractions of @xmath26 and @xmath2 provide useful information on the quark content of @xmath27 and @xmath28 mesons @xcite . in this paper , we report measurements for the cross sections and form factors of @xmath29 , @xmath1 , and @xmath2 at center of mass energies 3.650 , 3.686 , and 3.773 gev , and the branching fractions of @xmath30 @xmath29 , @xmath1 , and @xmath2 at 3.686 gev .
the beijing spectrometer ( besii ) is a conventional cylindrical magnetic detector that is described in detail in ref .
@xcite . a 12-layer vertex chamber ( vc ) surrounding the beryllium beam pipe
provides input to the event trigger , as well as coordinate information . a forty - layer main drift chamber ( mdc ) located just outside
the vc yields precise measurements of charged particle trajectories with a solid angle coverage of @xmath31 of @xmath32 ; it also provides ionization energy loss ( @xmath33 ) measurements which are used for particle identification . momentum resolution of @xmath34 ( p in gev/@xmath35 ) and @xmath33 resolution for hadron tracks of @xmath36 are obtained .
an array of 48 scintillation counters surrounding the mdc measures the time of flight ( tof ) of charged particles with a resolution of about 200 ps for hadrons . outside the tof counters , a 12 radiation length , lead - gas barrel shower counter ( bsc ) ,
operating in limited streamer mode , measures the energies of electrons and photons over @xmath37 of the total solid angle with an energy resolution of @xmath38 ( @xmath39 in gev ) . a solenoidal magnet outside the bsc provides a 0.4 t magnetic field in the central tracking region of the detector .
three double - layer muon counters instrument the magnet flux return and serve to identify muons with momentum greater than 500 mev/@xmath35 .
they cover @xmath40 of the total solid angle . in this analysis , a geant3 based monte carlo package ( simbes ) with detailed consideration of the detector performance ( such as dead electronic channels )
the consistency between data and monte carlo ( mc ) has been carefully checked in many high purity physics channels , and the agreement is reasonable .
the generators * ee2vp * @xcite and * howl - vp * @xcite , together with simbes , are used to determine the detection efficiencies for the one photon annihilation and resonance decay processes .
the data samples used for this analysis consist of @xmath41 @xmath42 events @xcite , @xmath43 pb@xmath44 of continuum data at @xmath45 @xcite , and 17.3 ( 1@xmath463% ) pb@xmath44 at the @xmath47 @xcite .
the channels studied are @xmath29 , @xmath1 , and @xmath2 , where @xmath48 decays to @xmath49 , @xmath50 to @xmath51 , @xmath27 to @xmath52 , and @xmath28 to @xmath53 .
a neutral cluster is considered to be a photon candidate if it is located within the bsc fiducial region , the energy deposited in the bsc is greater than 50 mev , the first hit appears in the first 6 radiation lengths , and the angle between the cluster development direction in the bsc and the photon emission direction from the beam interaction point ( ip ) is less than @xmath54 . for the @xmath1 channel ,
tighter requirements are applied : the energy deposited in the bsc must be greater than 100 mev , and the angle in the @xmath55 plane ( perpendicular to beam direction ) between the neutral cluster and the nearest charged track must be greater than @xmath56 .
each charged track is required to be well fit by a three - dimensional helix , to originate from the ip region , @xmath57 cm and @xmath58 cm , and to have a polar angle @xmath59 .
here @xmath60 , @xmath61 , and @xmath62 are the x , y , and z coordinates of the point of the closest approach of the track to the beam axis .
the tof and @xmath33 measurements for each charged track are used to calculate @xmath63 values and the corresponding confidence levels @xmath64 for the hypotheses that a track is a pion , kaon , or proton , where @xmath65 ( @xmath66 ) is the particle type . for events involving the final state @xmath29 and @xmath2 ,
at least half of charged pion candidates in each event are required to have @xmath67 larger than 0.01 , while for events involving the final state @xmath1 , both charged pion candidates must satisfy this requirement . for this channel ,
the events are required to have two good charged tracks with net charge zero and four or five photon candidates . a four constraint ( 4c )
kinematic fit under the hypothesis @xmath68 is performed .
if the number of selected photons is larger than four , the fit is repeated using all possible combinations of photons , and the one with the smallest @xmath69 is chosen .
the confidence level of the 4c fit is required to be larger than 0.01 .
in addition , we require that @xmath70 @xcite for the assignment @xmath71 must be smaller than that for @xmath72 in order to suppress possible @xmath73 background . among the four selected photons ,
there are three possible combinations to compose two @xmath74s : ( @xmath75 , @xmath76 ) , ( @xmath77 , @xmath78 ) , and ( @xmath79 , @xmath80 ) .
events where one and only one combination have both @xmath74 candidates satisfying @xmath81 are kept for further analysis . using the above selection ,
@xmath82 candidate events are obtained .
[ fresult1 ] shows the @xmath83 invariant mass distributions , where @xmath84 is the lower energy @xmath74 from the two @xmath74 candidates .
clear @xmath48 signals are seen in all three data samples at @xmath853.650 , 3.686 , and 3.773 gev .
the @xmath83 invariant mass distribution is fit with a shape for the signal determined by mc simulation plus a polynomial background , and @xmath86 , @xmath87 , and @xmath88 events are observed in the data samples at @xmath853.650 , 3.686 , and 3.773 gev , with @xmath89 , @xmath90 , and @xmath91 statistical significance @xcite , respectively .
the @xmath83 invariant mass distributions for @xmath92 candidate events at @xmath85 ( a ) 3.650 , ( b ) 3.686 , and ( c ) 3.773 gev . @xmath84 and @xmath93 are the low and high energy pions .
the curves show the fit described in the text .
, scaledwidth=45.0% ] for this channel , the events are required to have two good charged tracks with net charge zero and two photon candidates .
a 4c kinematic fit to the hypothesis @xmath94 is performed , and its confidence level is required to be larger than 0.01 and larger than that for the assignment @xmath95 to suppress possible @xmath95 background . the energy of each photon is required to be less than 1.7 gev to reject backgrounds from @xmath96 .
an additional requirement @xmath97 is applied to further suppress background from non-@xmath50 decay .
[ fresult2 ] shows the @xmath52 invariant mass distributions for @xmath98 candidate events .
clear @xmath27 signals are seen in all three data samples at @xmath853.650 , 3.686 , and 3.773 gev .
the @xmath99 invariant mass spectrum is fitted with a shape for the @xmath27 signal determined by mc simulation plus a polynomial background , and @xmath100 , @xmath101 , and @xmath102 events are observed in the data samples at @xmath853.650 , 3.686 , and 3.773 gev , respectively .
the statistical significances are @xmath103 , @xmath104 , and @xmath105 , respectively .
the @xmath52 invariant mass distributions for @xmath106 candidate events at @xmath85 ( a ) 3.650 , ( b ) 3.686 , and ( c ) 3.773 gev .
the curves show the fit described in the text.,scaledwidth=45.0% ] here , events with four good charged tracks with net charge zero and two or three photon candidates are selected .
a 4c kinematic fit is performed for the hypothesis @xmath107 .
if the number of selected photons is larger than 2 , the fit is repeated using all combinations of photons , and the one with the smallest @xmath69 is chosen .
the confidence level of the 4c fit is required to be larger than 0.01 and larger than that for @xmath108 .
the two photons are required to come from @xmath27 decay ( @xmath109 ) .
background from @xmath110 is rejected with the requirement that the mass recoiling from any @xmath51 pair satisfies @xmath111 . in order to suppress background from non-@xmath50 decay ,
an additional requirement @xmath112 is applied , where @xmath113 runs over all possible @xmath51 pairs .
[ fresult3 ] shows the @xmath53 invariant mass distributions for @xmath114 candidate events .
@xmath115 events are obtained in the data sample at @xmath116 by fitting the @xmath53 invariant mass spectrum with an @xmath28 shape obtained by mc simulation plus a polynomial for background .
the statistical significance for the @xmath28 signal is 3.1@xmath117 .
only 1 event is observed in each of the two data samples at @xmath853.650 and 3.773 gev , and the respective background is zero and 0.64 events as estimated from sidebands ; the corresponding upper limit for the observed @xmath2 event is calculated to be 4.4 and 3.9 events , respectively , with the scheme of j. conrad @xcite .
the @xmath118 invariant mass distributions for @xmath119 candidate events ( solid histogram ) and @xmath120 monte carlo simulation ( dotted histogram ) with arbitrary normalization for @xmath85 ( a ) 3.650 , ( b ) 3.686 , and ( c ) 3.773 gev .
the curve in ( b ) shows the fit described in the text.,scaledwidth=45.0% ]
at the @xmath16 peak , the cross section is from resonance decay , one - photon annihilation , and their interference , all electromagnetic processes .
the observed events are separated for the @xmath16 decay and one - photon annihilation processes using the scheme proposed by p. wang @xcite , considering the detection - efficiency difference between the resonance decay and one - photon annihilation , and neglecting the tiny interference ( see table [ tffwpi ] ) , that is , @xmath121 where @xmath122 is the integrated luminosity , @xmath117 the cross section , and @xmath123 the detection efficiency .
table [ tffwpi ] lists detailed information on the separation of the signal events observed in the data sample at @xmath124 . for the data sample at @xmath45 ,
the fraction of cross section from the @xmath16 resonance is negligible ( @xmath125 ) compared to that of the continuum , as estimated using the scheme proposed by p. wang @xcite .
therefore we attribute all observed events to one - photon annihilation . for the data sample at @xmath126 , the resonance part of the cross section for @xmath127 , at born order ,
is @xmath128 where @xmath129 is the partial width to the final state @xmath130 ( @xmath131 ) and is related to @xmath132 and the corresponding form factor is @xcite : latexmath:[\[\gamma_{f } = \frac{\gamma_{ee } q^3_{f}}{m_{\psi(3770 ) } } and @xmath132 of @xmath47 are @xmath134 and @xmath135 , respectively @xcite , the fraction of the cross section from @xmath47 resonance production is negligible ( @xmath136 ) compared to that from one - photon annihilation .
the @xmath16 tail at @xmath126 contributes no more than 2.0% for the observed events in these final states .
therefore , the observed events are taken to be totally from one - photon annihilation . [ cols="<,^,^,^,^,^,^,^ " , ] + energy dependence of the @xmath137 form factor .
curve ( a ) is calculated with eq .
( [ formula2 ] ) , while curve ( b ) is calculated with eq .
( [ formula1 ] ) . ,
scaledwidth=45.0% ] fig .
[ ffwpi ] shows the measured results of @xmath138 from our measurements , cmd-2 @xcite , and dm2 @xcite , and the calculated value of @xmath138 at @xmath139 @xcite : @xmath140 where @xmath141 @xcite , and the other quantities are taken from the pdg @xcite .
curve ( a ) is predicted by j .-
m . grard and g. lpez castro @xcite as : @xmath142 with the decay constants of @xmath48 and @xmath143 of @xmath144 and @xmath145 .
curve ( b ) is predicted by victor chernyak @xcite : @xmath146 where @xmath147 and @xmath148 are the masses of @xmath149 and @xmath150 , respectively . from fig .
[ ffwpi ] , our results agree with the description of eq .
( [ formula2 ] ) .
in conclusion , we determine branching fractions for @xmath151 , @xmath1 , and @xmath2 and the form factors @xmath152 , @xmath153 , and @xmath154 at @xmath853.650 , 3.686 , and 3.773 gev for the first time .
the branching fractions of @xmath7 and @xmath1 in our measurement are consistent with those of cleoc @xcite .
the bes collaboration thanks the staff of bepc for their hard efforts .
this work is supported in part by the national natural science foundation of china under contracts nos .
19991480 , 10225524 , 10225525 , the chinese academy of sciences under contract no .
kj 95t-03 , the 100 talents program of cas under contract nos .
u-11 , u-24 , u-25 , and the knowledge innovation project of cas under contract nos .
u-602 , u-34 ( ihep ) ; by the national natural science foundation of china under contract no .
10175060 ( ustc ) , and no .
10225522 ( tsinghua university ) ; and by the department of energy under contract no .
de - fg03 - 94er40833 ( u hawaii ) .
99 p. wang , c. z. yuan and x. h. mo , & 69&057502(2004 ) .
m. suzuki , & 63&054021(2001 ) .
dm2 collab . , j. jousset , & 41&1389(1990 ) .
mark iii collab . ,
d. coffman , & 38&2695(1988 ) .
m. suzuki , & 60&051501(1999 ) .
l. k@xmath155pke and n. wermes , phys .
rep . * 174 * , 67 ( 1989 ) .
fenice collab . , r. baldini , & 444&111(1998 ) .
p. wang , c. z. yuan and x. h. mo , & 574&41(2004 ) .
p. wang , x. h. mo and c. z. yuan , & 557&192(2003 ) .
victor chernyak , hep - ph/9906387 .
grard and g. lpez castro , & 425&365(1998 ) .
bes collab .
, j. z. bai , nucl .
. meth . * a 458 * , 627 ( 2001 ) .
l. l. ma , hep&np * 28 * , 1021 ( 2004 ) . in the generator for @xmath16
decays into vp mesons , the angular distribution is described by @xmath156 @xmath157},$ ] where @xmath158 is the angle between the vector meson and the positron direction . for two - body decays , such as @xmath159 , @xmath160 and @xmath161
are the polar and azimuthal angles of the momentum of the @xmath162 with respect to the helicity direction of the @xmath50 . for @xmath163 , @xmath160 and @xmath161
represent the angles of the normal to the decay plane with respect to the helicity direction of the @xmath48 .
x. h. mo , hep&np * 28 * , 455 ( 2004 ) .
s. p. chi , x. h. mo and y. s. zhu , measurement of the integrated luminosity at @xmath164 , 3686 gev for the bes detector ( to be published at hep&np 28 ( 2004 ) ) .
bes collab .
, j. z. bai , measurement of cross sections for @xmath165 and @xmath166 production in @xmath167 annihilation at @xmath126 ( submitted to ) .
the combined @xmath168 , @xmath169 , is defined as the sum of the @xmath168 values of the kinematic fit ( @xmath170 ) and those from each of all particle identification assignments : @xmath171 the significance @xmath172 is calculated by @xmath173^{1/2}$ ] , where @xmath174 and @xmath175 is the likelihood function value in the fit with and without signal , respectively .
j. conrad , & 67&012002(2003 ) .
we use the modified likelihood ratio ordering including the systematic uncertainties of signal ( gaussian parametrization ) and background ( flat parametrization ) in confidence interval construction .
particle data group , s. eidelman , & 592&1(2004 ) .
bes collab .
, j. z. bai , & 70&012005(2004 ) . t. sjostrand , comput .
phys . commun . *
82 * , 74 ( 1994 ) .
j. c. chen , & 62&034003(2000 ) .
g. bonneau and f. martin , & 27&381(1971 ) .
cmd-2 collab .
, r. r. akhmetshin _ et al .
_ , & 466&392(1999 ) .
dm2 collab .
, d. bisello _ et al .
( proc . suppl . )
* b 21 * , 111 ( 1991 ) .
cleo collab .
, n. e. adam , hep - ex/0407028 . | cross sections and form factors for @xmath0 , @xmath1 , and @xmath2 at center of mass energies of 3.650 , 3.686 , and 3.773 gev are measured using data samples collected with the besii detector at the bepc .
also , the branching fractions of @xmath3 , @xmath1 , and @xmath2 are determined to be @xmath4 , @xmath5 , and @xmath6 , respectively . |
a common characteristic of massive data sets whose major purpose of study is to discover the association between response and predictors is that the number of predictors is larger than the number of independent individuals .
although linear regression or its generalizations are useful tools to detect associations , some computational and theoretical issues are still remained unsolved for massive data .
many statistical approaches have been developed during the past two decades in many aspects . in this work ,
we concentrate on the screening problem for binary response regressions . considering all predictors in one linear model
is not practical without placing restrictions on the parameter space . instead , one may design screening statistics to rule out unimportant predictors before building final models .
a screening statistic is defined as a surrogate measure of the underlying association between the response and a predictor .
fan and lv ( 2008 ) and fan and song ( 2010 ) propose the concept of sure screening " : a screening statistic possesses the sure screening property if that the statistic is relatively small if the true association is negligible or 0 .
their sure independence screenings were designed toward this end .
fan and lv ( 2008 ) proposes the sure independence screening ( sis ) for linear regression .
they choose the predictors with large absolute covariances between response and predictors as important predictors and then build final regression models based on these important predictors . as we will show later , the covariance is proportional to the slope of the simple linear regression so hereafter , we use slope instead of covariance if there is no confusion .
the computation of sis is extraordinarily fast because it only involves centering and inner products but no matrix inversion .
note that when the term linear regression " is applied , we generally presume that the response is continuous or more restrictively , the response follows normal distribution . either way
, least - square estimation can be applied to estimate the regression coefficients .
one of our major interest is the consequence of applying least - square estimation to binary response data . as we will show later , under some conditions , it is useful for screening but not for estimation and prediction .
for binary response regressions , it is popular to choose logit , probit or complementary log - log link function ( mccullagh and nelder , 1989 ) to formulate the likelihood .
many statistical softwares perform estimation and testing tasks well . however , there are two major issues on using these models .
first , the choice of link function is essential to estimation science different link functions yield different regression coefficient estimates .
li and duan ( 1989 ) proves that , the maximum likelihood estimate ( mle ) is consistent to the true regression coefficient up to an unknown constant when the link function is misspecified .
so , when the true link function is unknown , the regression coefficient estimates is always questionable .
second , the mle of regression coefficient is sometimes unidentifiable , unique or finite mle does not exist ( albert and anderson , 1984 ) .
these two reasons urge us to find a computational efficient procedure for screening rather than merely using traditional logistic or probit regressions .
the rest of this article is arranged as follows . in section 2
, we review some useful results of linear model as well as the sure independence screening in linear model ( fan and lv , 2008 ) and in generalized linear regression ( fan and song , 2010 ) .
moreover , we show that , for binary response regressions , the screening statistics of both sis and the newly proposed least - square screening ( less ) converge in probability to its linear model counterpart up to a constant when the predictors follow multivariate normal and the link function belongs to a class of scale mixture of normals .
simulation and data analysis are shown in section 4 followed by our concluding remarks in section 5 .
we begin with matching parameters of the true model and parameters of working models .
consider the linear model @xmath0 where @xmath1 , @xmath2 follows @xmath3 , and @xmath4 and @xmath5 s are independent .
assume that not all of @xmath6 s are 0 .
denote @xmath7 and @xmath8 as the @xmath9 element of @xmath10 .
let ( [ eq : truemodel ] ) be the true model and call the predictors with non - zero ( zero ) regression coefficients as active ( inactive ) predictors .
as taught in the first course of linear regression , the least - square estimator of the slop of the working model @xmath11 converges in probability to @xmath12 and hence @xmath13 where @xmath14 , @xmath15 , and @xmath16 is a diagonal matrix with diagonal elements @xmath17 . for a more general case ,
let the working model be @xmath18 where @xmath19 .
define @xmath20 and @xmath21 and partition the regression coefficient vector as @xmath22 where @xmath23 and @xmath24 are regression coefficients corresponding to @xmath25 and @xmath26 , respectively .
further , define the partition of the variance covariance matrix as @xmath27\ ] ] with respect to @xmath25 and @xmath26 , too
. then we have @xmath28 which implies that the least - square estimate of the working model ( [ eq : working1 ] ) converges in probability to @xmath29 this suggests that @xmath30 if either @xmath31 or @xmath32 which is actually the partial orthogonality condition defined in huang , horowitz and ma ( 2008 ) .
in other words , to successfully estimate regression coefficients ( @xmath33 ) without contamination ( @xmath34 ) , a subset of predictors , say @xmath25 , should be chosen so that @xmath25 and @xmath26 are uncorrelated or none of predictors in @xmath26 is active . note that , the multiple regression with @xmath35 predictors is a misspecified model for the true model ( [ eq : truemodel ] ) as long as these @xmath35 predictors do not include all active predictors .
a well - known result in linear model literature is that if the working model is misspecified , the least - square estimator of the regression coefficient is biased .
the asymptotic bias can be quantified explicitly by ( [ eq : main ] ) . however , to our best knowledge , there is no such expression under binary response regressions .
the score equation is applied to link the true model parameters and the maximum likelihood estimates of the parameter under a specific misspecified model .
suppose the true model is @xmath36 and @xmath37 , where @xmath38 is the so called logit link .
assume that the true link is known and a simple working model @xmath39 is specified .
consequently , the score function converges in probability to @xmath40 and thus , the maximum likelihood estimator of @xmath41 converges to @xmath42 such that @xmath43= e\left [ x_1 \frac{1}{1+\exp\{-\gamma_0 - \sum_{j=1}^p x_j\gamma_j\ } } \right ] - e\left[x_1 \frac{1}{1+\exp\{-\beta_0^{ml } - \beta_1^{ml } x_1\ } } \right]\ ] ] so the relationship between the true model and working model can be quantified by these two expectations .
the calculation of these expectations are not trivial and their numerical evaluations had been studies by crouch and spiegelman ( 1990 ) and monahan and stefanski ( 1992 ) .
one of the major contribution is providing a closed - form expression of these expectations .
it is worth to emphasize that @xmath44 in general and we wish to express @xmath45 in terms of true parameter values @xmath46 s like ( [ eq : main ] ) .
hereafter , denote @xmath47 and @xmath48 as the probability density function ( _ p.d.f . _ ) and cumulative distribution function ( _ c.d.f . _ ) of the standard normal distribution , respectively , and denote @xmath49 as the density function of the normal distribution with mean @xmath50 and variance @xmath51 .
now , we derive the relationship between the regression coefficient of the true model and of the working model for binary response regression models .
let the true model be @xmath52 and the working model be @xmath53 where a function with the subscript @xmath54 means that the function is unknown but one is posited to it , and the function with subscript @xmath55 means that the function is the underlying function .
moreover , we require that @xmath56 as well as @xmath57 is a valid _ c.d.f . _ with the form of scale normal mixture @xmath58 and @xmath59 is a valid density function for either continuous or discrete @xmath60 .
this implies that @xmath61 is a symmetric _ p.d.f . _
such an @xmath62 can be the _
_ of gaussian , logistic , double exponential , student-@xmath63 ( andrews and mallows , 1974 ) , exponential power family ( box and tiao , 1973 ; west , 1987 ) and others .
hereafter , @xmath64 s should satisfy 1 ) @xmath65 , 2 ) @xmath59 is a valid density function , and 3 ) @xmath66 for every @xmath67 and @xmath68 . a sufficient and necessary condition for the existence of @xmath59 is provided by andrews and mallows ( 1974 ) .
following is one of our major conclusion and it is the result of lemma [ lm : ab00 ] and lemma [ lm : probit ] .
theirs proofs are deferred to appendix a. theorem [ th : main ] implies that the least square estimator @xmath69 converges to a value proportional to the desire value ( [ eq : main ] ) and the proportion is expressed in a form of integration .
[ th : main ] under the true model ( [ eq : btrue ] ) and the working model ( [ eq : bworking ] ) , the least - square estimator @xmath70 converges in probability to @xmath71 where @xmath72 .
[ lm : ab00 ] arnold and beaver ( 2000 ) prove that 1 .
linearly skewed normal @xmath73 is defined according to the following equation @xmath74 2 .
if @xmath73 then @xmath75^{-1}\ ] ] [ lm : probit ] under the true model ( [ eq : btrue ] ) with probit link ( @xmath76 ) , @xmath77 under the working model ( [ eq : bworking ] ) , we show that the maximum likelihood estimator of @xmath78 converges in probability to @xmath79 in theorem [ th : mle ] where the proof is rooted from the score equation @xmath80 \right\ } = e\left\{{\bf z}_1\left[h_t({\bf z}^t{{\boldsymbol}\gamma})-h_w({\bf z}^t_1{{\boldsymbol}\beta}^{ml}_1)\right]\right\}$ ] which implies that the theorem holds according to theorem [ th : main ] and we omit .
note that theorem [ th : mle ] also says that when @xmath81 , the link function is misspecified , the regression coefficient estimator of the working model converges to a value which is proportional to the true regression coefficient .
this conclusion is also made in li and duan ( 1989 ) but they do not provide an explicit form of the proportion .
[ th : mle ] under the true model ( [ eq : btrue ] ) and the working model ( [ eq : bworking ] ) , the maximum likelihood estimator @xmath82 converges in probability to @xmath83 where @xmath84 .
the integrations involved in theorem [ th : main ] and theorem [ th : mle ] are not easy to evaluate because @xmath85 and @xmath59 may not have simple forms .
for example , stefanski ( 1990 ) shows that when @xmath86 is the density function of the logistic distribution , @xmath87 and @xmath88 .
a friendly computation form can be @xmath89 which can be numerically evaluated by many computer softwares easily when all parameters and @xmath90 are specified . finally , we emphasize that the inverse function of the complementary log - log link is the _ c.d.f .
_ of extreme value distribution , and since the corresponding _ p.d.f .
_ is not symmetric around 0 , theorem 1 does not apply to this case .
in this section , we formally define the screening statistics including the sis and the less . for the @xmath91th individual , @xmath92 ,
let the true model be @xmath93 and the working model for the @xmath94th screening statistic be @xmath95 denote the likelihood of the working model as @xmath96 .
similarly , define the least - square objective function as @xmath97 . for the sis ( fan and song , 2010 )
, the screening statistic is @xmath98 .
we further distinguish it into two methods : sis with probit link ( sisp ) , @xmath99 , and sis with logit link ( sisl ) , @xmath100 where @xmath101 .
moreover , the screening statistic of less is @xmath102 .
in addition , because @xmath103 plays no role in screening so we omit but readers should know that above maximizations are taken with respect to @xmath103 and @xmath104 simultaneously . by theorems [ th : main ] and [ th : mle ] , we conclude that @xmath105 , @xmath106 , and @xmath107 where @xmath108 is the contamination due to other active predictors .
if @xmath109 then @xmath110 implies @xmath111 .
a general condition for @xmath109 is that both @xmath112 and @xmath10 are sparse ( fan and lv , 2008 ) .
finally , the sis recommends choosing @xmath113 largest @xmath114 s to build final models where @xmath115 and @xmath116 is the sample size .
we shall follow this rule for sisp and less procedures . following ( true ) model
is designed to demonstrate the consequence of model misspecification .
suppose that predictors @xmath117 follows normal distribution with zero mean and the true covariance structure is either autoregression ( ar1 ) , @xmath118 , or compound symmetry ( cs ) , @xmath119 where @xmath120 if @xmath121 is true and @xmath122 otherwise .
let the true model be @xmath123 and @xmath124 where @xmath125 . by theorem [ th : main ] ,
@xmath126 s can be evaluated according to the true model .
unfortunately , there is no closed - form expression for @xmath127 s and therefore , we estimate @xmath128 s by averaging 100 maximum likelihood estimates where each of them was calculated based on 200 independent samples drawn from the true model . the results are shown in figure [ fig : ass ] .
this figure expresses the intuition that the sure screening property does not hold for all kind of data sets . from figure
[ fig : ass ] , when the predictors have the ar1 correlation structure ( the left panel ) , both less and sis seems to have high chance to detect all active predictors . on the other hand ,
when predictors have the cs correlation structure ( the right panel ) , the predictor @xmath129 is barely chosen by all mentioned screening method .
a simple scheme was applied to demonstrate theorem 1 .
let the true model be @xmath130 and @xmath131 where the link function is either the probit link or the logit link .
moreover , we generate normal predictors @xmath132 with zero mean and aforementioned covariance structures ar1 and cs . for each dataset , 200 _ i.i.d .
_ samples were generated and 100 datasets were simulated .
table [ tb : bias ] shows that if the link function and the true regression coefficients are known then the least - square regression coefficient estimates with adjustment ( @xmath133 s ) seem to be consistent .
.biases of adjusted least square estimation ( @xmath125 ) [ cols="<,<,>,>,>,>,>,>,>",options="header " , ] + bold - faced numbers are gene ids that discovered by all of three screening methods .
in this work , we investigate the screening statistics , sisl , sisp , and less , in population level .
the quality of screening depends on the sparseness of the true regression coefficient and the sparseness of the covariance structure of predictors .
this conclusion can be drawn from ( [ eq : main ] ) and numerically proved by our simulations .
also , none of these methods satisfies the sure screening property defined by fan and lv ( 2008 ) without placing some restrictions on these two sparseness .
fortunately , in many experiments , it is nature to make these sparseness assumptions .
the leukemia dataset demonstrated in section 4.2 is a good example .
first , it is reasonable to assume that only a few genes affects the subtypes of leukemia and second , indeed , the covariance among gene expressions is low on average .
moreover , dudoit _ et al . _
( 2002 ) analyze this dataset via many classification methods . among these approaches ,
the diagonal linear discriminate analysis , assuming that all gene expressions are uncorrelated , beats others .
this may imply that the sparseness of the covariance structure assumption and the gaussian assumption on gene expressions fit the observed data well .
although both the sis s and the less are defined in a very restrictive manner , we suggest two rules to apply them . first , if the covariance structure is rich then aforementioned screening approaches would fail . under this circumstance , using other delicate statistical approaches is recommended .
second , if predictors do not follow normal distribution then the sample size should be large for an acceptable screening result .
when there is no significant violation of above conditions and the dataset is huge , we advocate using less because it is very computational efficient . for the leukemia dataset , sisl and sisp cost 23 and 31 seconds , respectively , whereas less cost less than 1 second .
00 albert , a. and anderson , j.a .
( 1984 ) on the existence of maximum likelihood estimates in logistic regression models " , _ biometrika _ , * 71 * , 1 10 .
andrews , d.f . and mallows , c.l .
( 1974 ) scale mixtures of normal distributions " , _ journal of the royal statistical society , series b _ , * 36 * , 99 102 .
arnold , b.c . and beaver , r.j .
( 2000 ) hidden truncation models " , _ sankhy@xmath134 : the indian journal of statistics _ , * 62 * , 23 35 .
biswas , a. and hwang , j .- s .
( 2002 ) a new bivariate binomial distribution " , _ statistics & probability letters _ , * 60 * , 231 240 .
box , g.e.p . and tiao , g.c .
( 1973 ) _ bayesian inference in statistical analysis _ , addison - wesley .
crouch , e.a . and spiegelman , d. ( 1990 ) the evaluation of integrals of the form @xmath135 : applications to logistic - normal models " , _ journal of the american statistical associations _ , * 85 * , 464 467 .
dudoit , s. , fridlyand , j. and speed t.p .
( 2002 ) comparison of discrimination methods for the classification of tumors using gene expression data " , _ journal of the american statistical association _
, * 97 * , 77 87 .
fan , j. and lv , j. ( 2008 ) sure independence screening for ultrahigh dimensional feature space " , _ journal of the royal statistical society , series b _ , * 70 * , 849 911 .
fan , j. and song , r. ( 2010 ) sure independence screening in generalized linear models with np - dimensionality " , _ the annals of statistics _ , * 38 * , 3567 3604 .
huang , j. , horowitz , j. and ma , s. ( 2008 )
asymptotic properties of bridge estimators in sparse high - dimentional regression models " , _ annals of statistics _ ,
* 36 * , 587 613 .
li , k .- c . and duan , h. ( 1989 ) regression analysis under link violation " , _ the annals of statistics _ , * 17 * , 1009 1052 .
mccullagh , p. and nelder , j.a .
( 1989 ) _ generalized linear models _
, second edt . , chapman & hall / crc monahan , j. and stefanski , l.a .
( 1992 ) normal scale mixture approximations to @xmath136 and computation of the logistic - normal integral " , in _ handbook of the logistic distribution _ , pp 529 540 , n. balakrishnan , editor .
marcel dekker , new york .
stefanski , l.a .
( 1990 ) a normal scale mixture representation of the logistic distribution " , _ statistics & probability letters _ , * 11 * , 69 70 . west , m. ( 1987 ) on scale mixtures of normal distributions " ,
_ biometrika _ , * 74 * , 664 668 .
proof of theorem [ th : main ] note that the least square estimator converges in probability to the inverse of the covariance matrix among predictors multiplied by the covariance between the response and the predictors .
the former is trivial and thus we show the latter .
we begin with the conditional expectation @xmath137 where @xmath138 , @xmath139 and @xmath140 . according to the proof of lemma 2 , @xmath141 \phi({\bf z}_1 ; { { \boldsymbol}\mu } , { { \boldsymbol}\omega } ) d { \bf z}_1\\ & = \int_{{\mathbb{r}}^+ } \left [ \int_{{\mathbb{r } } } { \bf z}_1 \phi\left(\frac{c_1 + { { \boldsymbol}\gamma}_1^t{\bf z}_1}{\sigma}\right ) \phi({\bf z}_1 ; { { \boldsymbol}\mu } , { { \boldsymbol}\omega } ) d { \bf z}_1 \right ] q(\sigma)d\sigma \end{split}\ ] ] and therefore @xmath142 where @xmath143 , @xmath144 , @xmath145 , and @xmath146 .
after change of variable , we have @xmath147 last , since @xmath148 exists , @xmath149 exists .
proof of lemma [ lm : probit ] define @xmath150 .
then @xmath151 where @xmath152 and @xmath140 .
let @xmath153 .
the conditional expectation @xmath154 can be rewritten as @xmath155 further , let @xmath156 and @xmath157 . by lemma [ lm : ab00 ] , above integration becomes @xmath158 or equivalently @xmath159 where @xmath160 .
since @xmath161 and @xmath162 , @xmath163\\ & + { { \boldsymbol}\omega}{{\boldsymbol}\gamma_1 } \int_{{\mathbb{r } } } \phi\left ( w ; - \gamma_0 , 1 + { { \boldsymbol}\gamma}_1^t{{\boldsymbol}\omega}{{\boldsymbol}\gamma}_1\right ) \phi(w ; 0 , { \bf b}^t{{\boldsymbol}\sigma}_{22}{\bf b } ) dw . \end{split}\ ] ] again , by lemma [ lm : ab00 ] , @xmath164 and therefore , after some algebra , @xmath165 .
let @xmath166 follow bivariate binomial distribution with marginal distribution @xmath167 and @xmath168 .
following biswas and hwang ( 2002 ) , the joint distribution can be @xmath169 where @xmath170 @xmath171/(1+\alpha)$ ] , @xmath172 , and @xmath173 is a carefully chosen constant .
note that @xmath174 is not necessarily a probability mass function for arbitrary @xmath175 .
a sufficient condition to make @xmath176 a probability mass function is that @xmath177 moreover , the correlation coefficient between @xmath178 and @xmath179 is @xmath180 given by biswas and hwang ( 2002 ) .
consequently , an algorithm to simulate correlated @xmath181 is as follows : 1 .
sample @xmath182 from @xmath183 .
sample @xmath184 from @xmath185 .
set @xmath186 .
2 . for simulating @xmath187 , sample @xmath188 from @xmath183 and @xmath189 from @xmath190 .
3 . given @xmath191 , if the sufficient condition ( [ eq : suf ] ) is not satisfied , let @xmath192 .
sample @xmath187 from the conditional probability ( [ eq : cond ] ) with @xmath191 .
5 . let @xmath193 .
goto 2 . while @xmath194 and stop while @xmath195 .
note that , setting @xmath196 makes two binomial random variables independent .
so step 3 .
enforces two consecutive variables to be independent with probability roughly equal to 0.1 according to simulation .
moreover , the resulting correlation ranges from 0.2 to 0.8 with median 0.4 by simulation . | screening before model building is a reasonable strategy to reduce the dimension of regression problems .
sure independence screening is an efficient approach to this purpose .
it applies the slope estimate of a simple linear regression as a surrogate measure of the association between the response and the predictor so that the final model can be built by those predictors with steep slopes .
however , if the response is truly affected by a nontrivial linear combination of some predictors then the simple linear regression model is a misspecified model . in this work
, we investigate the performance of the sure independence screening in the view of model misspecification for binary response regressions .
both maximum likelihood screening and least square screening are studied with the assumption that predictors follow multivariate normal distribution and the true and the working link function belong to a class of scale mixtures of normal distributions .
link function , logistic model , probit model , sure independence screening |
all radio telescopes can be classified as either filled or unfilled - aperture antennas .
the simplest filled - aperture antennas are single - dish telescopes .
the desired astronomical object to be observed and the particular scientific goals determine which type of radio telescope to use . in general ,
single - dishes are considered as tools for low spatial resolution observations , while interferometers are used for high resolution observations . while compact objects are more suited for interferometric observations , extended objects
are commonly observed with single - dishes as interferometers can not faithfully recover information on the largest spatial scales .
however , in many scientific cases it is essential to obtain high spatial resolution observations of large objects , and to accurately represent emission present over a wide range of spatial scales .
a simple recipe you may follow in such cases is : * observe ( mosaic ) your object with an interferometer , * observe your object with a single - dish , * cross - calibrate the two data sets , and then * combine the single - dish and interferometer data .
this combination of single - dish and interferometer data , when observing extended objects , is referred to as the short - spacings correction .
this ` simple ' recipe may be considered as an artistic touch to the interferometric images as it makes them look much nicer but still preserves their high spatial resolution .
this results from : ( a ) inclusion of more resolution elements , those seen by a single - dish ; and ( b ) reconstruction of image artifacts . at the same time , these images contain information about the total power , and can be used to measure accurate flux densities , column densities , masses , etc . from a pure historical perspective
, the short - spacings correction bridges the gap between the two classes of radio telescopes , essentially obtaining the best of ` both worlds ' , that is the high spatial resolution information provided by interferometers , and the low spatial resolution , including the total power , information provided by single - dishes .
this article will explain what the short - spacing problem is , how it is manifested , and how we can , both theoretically and practically , solve this problem .
section 2 depicts very briefly the fundamentals of interferometry , defines the spatial frequency domain , and draws an analogy between a single dish and an interferometer .
the effects of missing short spacings are demonstrated in section 3 , as well as prospects for solving the problem .
section 4 considers the cross - calibration of interferometer and single dish data which is a precursor to any combination method .
methods for data combination are discussed in section 5 and section 6 , and compared in section 7 .
a very brief review of the basics of interferometry is necessary right at the beginning of this article , in order to define and explain some terms that will be used further on .
however , we do not want to go deeply into interferometry , as there is a vast literature available on this topic , starting with `` interferometry and synthesis in radio astronomy '' ( thompson , moran , & swenson 1986 ) and `` synthesis imaging in radio astronomy '' ( taylor , carilli , & perley 1999 ) .
the fundamental idea behind interferometry is that a fourier transform relation exists between the sky radio brightness distribution @xmath0 and the response of a radio interferometer . if the distance between two antennas ( the baseline ) is @xmath1 , then the so - called visibility function , @xmath2 , is given by : @xmath3 d\omega \;.\ ] ] here , @xmath4 is an antenna reception pattern , or * primary beam * , and @xmath5 is the vector difference between a given celestial position and the central position of the field of view .
the * aperture synthesis technique * is a method of solving equation [ e : int - basic ] for @xmath6 by measuring @xmath7 at suitable values of @xmath1 . to simplify equation [ e : int - basic ] , a more convenient , right - hand rectilinear ,
coordinate system is introduced in fugure [ f : interferometer ] .
coordinates of vector @xmath1 in this system are @xmath8 , where the direction to the source center @xmath9 defines the @xmath10 direction , and @xmath11 and @xmath12 are baseline projections onto the plane perpendicular to the @xmath9 direction , towards the east and the north , respectively .
a synthesized image in the @xmath13 plane represents a projection of the celestial sphere onto a tangential plane at the source center . in certain conditions , that is in the case of an earth tracking , east - west interferometer array ,
with the @xmath10-axis lying in the direction of the celestial pole , further simplifications of equation [ e : int - basic ] are possible : @xmath14 \frac{dl dm } { \sqrt{1-l^{2}-m^{2 } } } \;.\ ] ] therefore , the visibility function @xmath15 can be expressed as the fourier transform of a modified brightness distribution @xmath16 .
coordinates @xmath11 and @xmath12 ( @xmath17 ) are measured in units of wavelength and the @xmath18 plane is called * the spatial frequency domain*. these are effectively projections of a terrestrial baseline onto a plane perpendicular to the source direction .
the @xmath13 plane is referred to as * the image domain*. to obtain @xmath19 , from equation [ e : v ] , an inverse fourier transform of @xmath15 is required , meaning that a complete sampling of the spatial frequency domain is essential . in practice however , a bit more than a simple inversion is needed as only limited sampling of the @xmath18 plane is available
. for a given configuration of antennas any interferometer array has a limited range of baselines , lying between a minimum , @xmath20 , and maximum , @xmath21 , baselines . as an example , 5 antennas of the australia telescope compact array ( atca ) form 10 baselines , with @xmath22 m and @xmath23 m for a particularly compact configuration .
the final resolution ( @xmath24 ) is inversely related to the maximum baseline by @xmath25 . in the case of an earth tracking interferometer , as the earth rotates the baseline projections on the @xmath18 plane trace a series of ellipses .
the parameters of each ellipse depend upon the declination of the source , the length and orientation of the baseline , and the latitude of the center of the baseline ( thompson et al . 1986 ) .
the ellipses are concentric for a linear array ( e.g. atca ) . for a 2-dimensional array ( e.g. the very large array , vla ) , the ellipses are not concentric and so can intersect .
as each baseline traces a different ellipse , the ensemble of ellipses indicates the spatial frequencies that can be measured by the array ( see thompson et al .
at each sampling interval , the correlator measures the visibility function for each baseline , thus resulting in a number of samples being measured over elliptical tracks in the @xmath18 plane .
hence , the resultant interferometer @xmath18 coverage will always be , more or less , incomplete , having a hole in the center of the @xmath18 plane whose diameter corresponds to the minimum baseline , gaps between measured elliptical tracks , and gaps between each adjacent samples on each elliptical track .
the ensemble of ellipses ( loci ) is known as the * transfer * or * sampling function * , @xmath26 .
an example of the sampling function obtained with the vla is shown in fugure [ f : tracks12 ] .
hence , if @xmath15 is a true ( ideal ) visibility function , the measured ( observed ) visibilities ( @xmath27 ) can be expressed as : @xmath28 @xmath29 is usually representable by a set of @xmath30-functions , between the lowest and the highest spatial frequency sampled by the interferometer ( corresponding to the shortest and the longest baselines , respectively ) .
the fourier transform of equation [ e : int1 ] gives the observed sky brightness distribution @xmath31 ( so called * ` dirty ' image * ) : @xmath32 where @xmath33 is * the synthesized or ` dirty ' beam * , which is the point source response of the interferometer .
as usually , asterisks ( @xmath34 ) are used to denote convolution .
when imaging , incomplete @xmath18 coverage leads to severe artifacts , such as negative ` bowls ' around emission regions and negative and positive sidelobes ( cornwell , braun , & briggs 1999 ) .
we return to this in section [ s : effects ] the determination of @xmath0 from @xmath31 in the deconvolution process , requires beforehand interpolation and extrapolation of @xmath27 for missing data due to the discontinuous nature of @xmath29 ( cornwell & braun 1989 ) .
this process works well when a compact configuration of antennas is used and when the source is small enough , with angular size @xmath35 ( bajaja & van albada 1979 ) . for imaging larger objects , with angular size @xmath36 ,
a significant improvement in filling in the @xmath18 coverage can be achieved by using the * ` mosaicing technique ' * , where observations of many pointing centers are obtained and ` pasted ' together ( see holdaway 1999 ) .
mosaicing effectively reduces the shortest projected baseline to @xmath37 , where @xmath38 is the diameter of an individual antenna .
nevertheless , the center of the @xmath18 plane still suffers if significant large scale structure is present .
this lack of information for very low spatial frequencies ( around the center of the @xmath18 coverage ) in an interferometric observation is usually referred to as the * ` short - spacings problem'*. extended objects with angular size @xmath36 can be observed with a single - dish .
let us now think of a single - dish in a slightly unusual way , imagining filled apertures consisting of a large number of small panels packed closely together
. then all these panels can act as interferometer elements with their signals being combined together at the focus , making a so called phased or adding array ( see contribution by d. emerson in this volume ) .
the distance between each two panels corresponds to a baseline , as shown in fugure [ f : dish ] .
the baseline distribution then monotonically decrease from zero at the center up to the maximum baseline , determined by the single - dish diameter @xmath39 .
this is also shown in fugure [ f : dish ] .
one observation with a single - dish provides a total flux density measurement , corresponding to the zero spacing , @xmath40 .
however , if a single - dish scans across an extended celestial object , it measures not only a single spatial frequency , but a whole range of continuous spatial frequencies all the way up to a maximum of @xmath41 ( ekers & rots 1979 ) .
hence , a single - dish behaves as an interferometer with an almost infinite number of antennas , and therefore has a continuous range of baselines , from zero up to @xmath41 .
the nice thing about this representation is that we can now use the same mathematical notation to describe both single - dishes and interferometers .
the observed sky brightness distribution @xmath42 in the case of single - dish observations is then given by : @xmath43 with @xmath44 being the * single - dish beam * pattern .
the fourier transform of equation [ e : sd1 ] gives the observed single - dish ` visibilities ' , @xmath45 : @xmath46 where @xmath47 is the fourier transform of the single - dish beam pattern which , unlike @xmath29 , is a continuous function between zero and the highest spatial frequency sampled by the single - dish .
determination of @xmath0 from @xmath42 requires deconvolution , but no interpolation of @xmath45 is needed since this is a continuous function .
as shown in equation [ e : int ] , the sky brightness distribution can be reconstructed , in the case of interferometric observations , by deconvolving the ` dirty ' image with the synthesized beam . as an example , fugure [ f : effect2 ] shows the result of the hi ` mosaic ' observations of the small magellanic cloud ( smc ) with the atca .
more information about these observations and data reduction is available in stanimirovic et al .
the two adjacent panels on the right side show right ascension ( ra ) cuts through the image .
negative bowls ( shown in white on the image ) are seen around emission peaks ( shown in black ) , as well as in ra cuts .
these are typical interferometric artifacts resulting from an incomplete @xmath18 coverage . a simple graphical explanation of why this happens , borrowed from braun & walterbos ( 1985 ) , is shown in fugure [ f : effect1 ] for the case of a point source .
the solid vertical line in fugure [ f : effect1 ] distinguishes the spatial frequency ( a ) from the image domain ( b ) .
the distribution of measured spatial frequencies , or what we have already defined as a transfer ( or sampling ) function , @xmath29 , is given on the left side , while its fourier transform , that is the synthesized beam , @xmath33 , is shown in the right .
an exclusion of the central values from the spatial frequency domain , is equivalent to a subtraction of a broad pedestal in the image domain , resulting in the presence of a deep negative ` bowl ' around the observed object , as seen in fugure [ f : effect2 ] .
this demonstrates simply how severe the effects of missing short - spacings can be .
the larger the object is relative to the reciprocal of the shortest measured baseline one tries to image , the more prominent the short - spacing problem becomes .
there are two main questions concerning the short - spacings problem : 1 . how to provide ( observe ) missing short spacings to interferometric data ; 2 . how to combine short - spacing data with those from an interferometer
. in answering the first question ,
all solutions can be grouped into two array schemes : * homogeneous * , having all antennas of the same size , and * heterogeneous * , based on observations obtained with different - sized antennas .
there are many possibilities concerning the heterogeneous arrays , such as using smaller arrays and even a hierarchy of smaller arrays .
the simplest option , though , is a single - dish telescope with a diameter ( @xmath41 ) larger than the interferometer s minimum baseline .
we briefly touch here on some pros and cons for both array schemes .
one of the difficulties in providing short - spacings with a single - dish is that it is hard to provide a large single - dish which would have the sensitivity equivalent to that of an interferometer ( holdaway 1999 ) .
also , single - dish observations are complex ( they require a lot of separate pointing centers to cover a large object ) and very sensitive to systematic errors . using theoretical analysis , numerical simulations and observational tests , cornwell , holdaway , & uson ( 1993 )
show that a homogeneous array in which the short - spacings are obtained from single antennas of the array , allows high quality imaging .
they find that a key advantage over the large single - dish scheme is pure simplicity , which is an important factor for the complex interferometric systems . as both interferometric and total power data
are obtained with the same array elements , no cross - calibration is required in this case .
note that in this case total - power and interferometric observations have to be synchronized which is not a simple task because of different observing techniques involved ( e.g. the single - dish observations require frequency or position switching modes ! ) .
this turns out to be an especially difficult task for the continuum observations .
however , to _ fully _ fill in the central gap in an interferometer @xmath18 coverage and _ preserve sensitivity _ at the same time , the heterogeneous array scheme appears more advantageous .
this has been recently recognized in the planning of the future atacama large millimeter / submillimeter array ( alma ) .
imaging simulations have shown that antenna pointing errors of only a few percent of the primary beam width produce large errors in the visibilities in the central @xmath18 plane , causing a large degradation of image quality ( see morita 2001 ) . to compensate for this problem , an additional , smaller array of 6 8 m dishes ,
the so called alma compact array ( aca ) , has been proposed to provide short baselines . in answering the second question ,
methods for the combination of interferometer and single - dish data can be grouped into two classes : data combination in the spatial frequency domain ( bajaja & van albada 1979 ; vogel et al .
1984 ; roger et al . 1984 ; wilner & welch 1994 ; zhou , evans , & wang 1996 ) , and data combination in the image domain ( ye & turtle 1991 ; stewart et al . 1993 ; schwarz & wakker 1991 ; holdaway 1999 ) .
each approach can be realized through a number of different methods .
both approaches are very common and are becoming a standard data processing step . as the most common scheme of a heterogeneous array involves use of a large single - dish telescope
, we proceed to consider this particular case further .
before adding short - spacing data , it is necessary to be sure that both the interferometer and single - dish data sets have identical flux density scales . as calibration is never perfect , the calibration differences between the two sets of observations can be significant in some cases ( e.g. observations spread over a long period of time , different data quality , use of different flux density scales for calibration , quality of calibrators , etc . ) .
this results in a small but appreciable difference in the measured flux densities .
we define the calibration scaling factor , @xmath48 , as the ratio of the flux densities of an unresolved source in the single - dish and interferometer maps : @xmath49 in the case of perfect calibration , @xmath50 .
however , @xmath51 otherwise , and needs to be determined very accurately .
unfortunately , it is hard to find suitable compact sources to directly determine @xmath48 .
hence , the best way to estimate @xmath48 is to compare the surface brightness of the observed object in the overlap region of the @xmath18 plane , see fugure [ f : uv - scheme ] .
this region should correspond to angular sizes to which both telescopes are sensitive . for
a source of brightness , @xmath0 , both the interferometer and single dish should measure within this region the same , @xmath52 , and calibration errors will appear as : @xmath53 for an extended source @xmath54 and @xmath55 are often , for convenience , expressed in units of jy beam@xmath56 not jy sr@xmath56 , and so will be different numbers because of the different beams considered ( with beam areas @xmath57 and @xmath58 , respectively ) . for this purpose ,
an estimate of the resolution difference between the two data sets ( @xmath59 ) is also needed . to determine @xmath48 the following steps
are required : 1 .
scale the single - dish data by @xmath60 to account for the difference in brightness caused _ only _ by different resolutions , 2 .
fourier transform the interferometer and scaled single - dish images , 3 .
deconvolve the single - dish data ( by dividing them by the fourier transform of the single - dish beam ) , and 4 .
compare ` visibilities ' in the overlapping region of spatial frequencies .
several important issues should be considered here : * when fourier transforming in step 2 watch for edge - effects ! to avoid nasty edge - effects in some cases apodizing of both interferometer and single - dish images may be required in order to make the image intensities smoothly decrease to zero near the edges .
* step 3 requires a very good knowledge of the single - dish beam ! to make things even harder , the fwhm of the single - dish beam and the calibration scaling factor are highly coupled ( sault & killeen 1998 ) . therefore , an error in the single - dish beam model has the same effect in the overlapping region as an error in the flux density scale . if the single - dish beam is poorly known , @xmath48 will be a quadratic function of distance in the fourier plane , in the first approximation ( see stanimirovic 1999 ) . * a sufficient overlap in spatial frequency
is required for step 4 . assuming a gaussian - tapered illumination pattern for a single - dish , and considering a cut - off level of 0.2 for reliable data
, we can estimate the minimum diameter , @xmath38 , of a single - dish necessary to provide all spacings shorter than @xmath20 for a given interferometer : @xmath61 in order to have a reasonable overlap of spatial frequencies so that @xmath48 can be derived , a slightly larger single - dish is required with @xmath62 .
for example , for the atca shortest baseline of 31 m , the single - dish providing short spacings should have diameter of @xmath63 m. therefore , the 64 m parkes telescope can do a really great job . also , while the 100-m green bank telescope will be able to provide short - spacing data for the vla c and d arrays ( with @xmath64 m ) , only arecibo could do so for the vla b array ( with @xmath65 m ) .
as shown by bajaja & van albada ( 1979 ) , the true missing short - spacing visibilities can be provided from the function @xmath15 in equation [ e : sd2 ] , if the single - dish is large enough to cover the whole central gap in the interferometer @xmath18 coverage .
the deconvolution of the single - dish data gives the true single - dish visibilities , where @xmath66 , by : @xmath67 function @xmath15 can be then substituted in equation [ e : int1 ] , after rescaling by @xmath48 , everywhere in the @xmath18 plane where equation [ e : sd3 ] holds .
this would provide the resultant @xmath18 coverage having the inner visibilities from the single - dish data only ( rescaled to match the interferometer flux - density scale due to the calibration differences ) , and the outer visibilities from just the interferometer data .
this is effectively feathering or padding the interferometer visibilities with the single - dish data .
depending on the type of input images used , and/or the type of weighting applied within the region of overlapping spatial frequencies , there are several applications of this technique ( roger et al .
1984 ; vogel et al . 1984 ; wilner & welch 1994 ; zhou et al . 1996 ) .
however , in all cases , for a good data combination the single - dish data set should fulfill the following two conditions . * a sufficiently fine sampling of the single - dish data at the nyquist rate ( 2.4 pixels across the beamwidth )
is required to avoid aliasing during deconvolution ( vogel et al .
the single - dish data must also have the same coordinate system as the interferometer data .
therefore , it is sometimes necessary to re - grid single - dish images . *
visibilities derived from the single - dish data should have a signal - to - noise ratio comparable to those of the interferometer in the overlapping region in order not to degrade the combined map ( vogel et al . 1984 ) .
this linear data combination in the spatial frequency domain is very widely used and is implemented in several packages for radio data reduction , e.g. task imerg in aips , image tool in aips++ and task immerge in miriad . as an example we will discuss immerge here .
immerge takes as input a clean ( deconvolved ) high - resolution image , and a non - deconvolved , low - resolution image .
these images are fourier transformed ( labeled as @xmath68 and @xmath69 ) and combined in the fourier domain applying tapering functions , @xmath70 and @xmath71 , such that their sum is equal to the gaussian function having a fwhm of the interferometer , @xmath24 : @xmath72 function @xmath71 is a gaussian with the fwhm of the single dish .
the low - resolution visibilities are multiplied by @xmath48 to account for the calibration differences .
the final resolution is that of the interferometer image .
the tapering functions @xmath73 and @xmath74 are shown in fugure [ f : immerge - beam ] , together with the tapering function of the merged data set , for the case of the atca interferometer and the parkes single - dish telescope .
immerge can also estimate the calibration scaling factor , @xmath48 , by comparing single - dish and interferometer data in the region of overlapping spatial fequencies specified by the user .
as an example of how immerge works in practice , fugure [ f : fourier_comb ] shows a sequence of images at various stages of data processing : before merging , after fourier transforming , and the final version .
there are two distinct methods for data combination in the image domain .
the first , the ` linear combination ' method ( stanimirovic et al . 1999 ) , merges data sets in a simple linear fashion before final deconvolution , while the second ( sault & killeen 1998 ) , the non - linear method , combines all data during the deconvolution process . the theoretical basis for merging before deconvolution is the linear property of the fourier transform : a fourier transform of a sum of two functions is equal to the sum of the fourier transforms of the individual functions .
therefore , instead of adding two maps in the fourier domain and fourier transforming the combined map to the image domain , one can produce the same effect ( fill in missing short - spacings in an interferometer @xmath18 coverage ) by adding maps in the image domain .
this method was first applied by ye & turtle ( 1991 ) , and stewart et al .
( 1993 ) . as we have seen from equations [ e : int ] and [ e : sd1 ] ,
the interferometer and single - dish data obey the convolution relationship .
the dirty images and beams can be combined to form a composite dirty image ( @xmath75 ) and a composite beam ( @xmath76 ) with the following weighting : @xmath77 where @xmath78 is an estimate of the resolution difference between the two data sets .
the convolution relationship , @xmath79 , still exists between the composite dirty image , @xmath80 , and the true sky brightness distribution , @xmath0 .
deconvolving the composite dirty image with the composite beam hence solves for @xmath0 .
there is no single existing program ( task ) that employs this method , but a linear combination of maps can be easily obtained in any package for radio data reduction , followed by a favorite choice of a deconvolution algorithm . as an example , fugure [ f : image_comb ] shows the ` linear combination ' method applied on the atca mosaic and parkes telescope hi observations of the smc at several stages of data processing . note that @xmath33 in the case of mosaic observations represents a whole cube of beams , one for each pointing in the mosaic .
the combined dirty image was deconvolved using miriad s maximum entropy algorithm ( sault , staveley - smith , & brouw 1996 ) .
the model was finally restored with a 98-arcsec gaussian function . besides the missing information in the center of the @xmath18 plane , an interferometer @xmath18 coverage suffers from spatial - frequency gaps . since the missing information can be introduced in an infinite number of ways , the convolution equation ( @xmath81 ) has a non - unique solution .
hence , the deconvolution has the task of selecting the ` best ' image from all those possible ( cornwell 1988 ) .
since deconvolution has to estimate missing information , a non - linear algorithm must be employed ( cornwell 1988 ; sault et al .
cornwell ( 1988 ) and sault et al .
( 1996 ) showed that the deconvolution algorithms implementing the so called ` joint ' deconvolution scheme , whereby observations of many pointing centers are combined prior to deconvolution , produce superior results in the case of mosaicing , since more information is fed to the deconvolver .
we expect that the same argument might apply for the addition of the single - dish data , resulting in the merging before and during deconvolution being more successful than the merging of clean images in the spatial frequency domain .
the maximum entropy method ( mem ) is one of the non - linear deconvolution algorithms .
it selects the deconvolution solution so it fits the data and , at the same time , has a maximum ` entropy ' .
cornwell ( 1988 ) explains this entropy as something which when maximised produces a positive image with a compressed range in pixel values .
the compression criterion forces the final solution ( image ) to be smooth , while the positivity criterion forces interpolation of unmeasured fourier components .
one of the commonly used definitions of entropy is : @xmath82 where @xmath83 is the brightness of @xmath84th pixel of the mem image and @xmath85 is the brightness of @xmath84th pixel of a ` default ' image incorporated to allow _ a priori _ knowledge to be used ( @xmath86 is the base of the natural logarithms ) . the requirement that the final image fits the data is usually incorporated in a constraint such that the fit @xmath87 of the predicted visibilities to those observed ( cornwell 1988 ) is close to the expected value : @xmath88 with @xmath89 being the number of independent pixels in the map and @xmath90 being the noise variance of the interferometer data .
the single - dish data can be incorporated during the maximum - entropy deconvolution process in two ways . 1 .
* the ` default ' image * + the easiest way is to use the single - dish data as a ` default ' image in equation [ e : mem ] since , in the absence of any other information or constraints , this forces the deconvolved image to resemble the single - dish image in the spatial frequency domain where the interferometer data contribute no information .
since this method puts more weight on the interferometer data wherever it exists , the size of the overlapping region plays a very important role ( holdaway 1999 ) .
as large an overlap of spatial frequencies as possible is required to provide good quality interferometer and single - dish data within this region , in order to retain the same sensitivity over the image .
* the ` joint ' deconvolution * + the second way maximizes the entropy while being subject to the constraints of fitting both data sets simultaneously : @xmath91 the ` joint ' deconvolution method provides also an alternative way , completely performed in the image domain , for determining the calibration scaling factor . maximizing the entropy , while fitting both data sets , a ` joint ' deconvolution algorithm can iteratively solve for @xmath0 and @xmath48 simultaneously . as a schematic example
, fugure [ f : mem ] shows the non - linear approach for data combination performed by miriad s program mosmem .
the qualitative comparison of all four methods for the short - spacings correction addressed in sections 5 and 6 , is shown in fugure [ f:4methods ] for the 169 velocity channel of the smc data .
all four images have the same grey - scale range ( @xmath92 to 107 k ) and are remarkably similar .
they all have the same resolution and show the same small and large scale features with only slightly different flux - density scales .
no signs of interferometric artifacts are visible on any of the images .
this demonstrates that all four methods for the short - spacings correction give satisfactory results in the first approximation , when an _ a priori _ determined calibration scaling factor is used .
a similar conclusion was reached in wong & blitz ( 2000 ) where results of data combination using immerge and the ` linear combination ' method ( section [ s : linear ] ) were compared for the case of bima and kitt peak 12-m telescope co observations . .1truein
ccccc method & total flux & min & max & noise + & ( jy ) & ( jy beam@xmath56 ) & ( jy beam@xmath56 ) & ( mjy beam@xmath56 ) + a & 5600 & @xmath93 & 1.97 & 30 + b & 6500 & @xmath94 & 2.16 & 32 + c & 6300 & @xmath94 & 2.00 & 28 + d & 5900 & @xmath95 & 2.00 & 29 + the quantification of the quality of an image depends on the scientific questions we want to address ( cornwell et al .
1993 ) and is therefore case specific .
something that any short - spacings correction must fulfill , though , is that the resolution of the final image should be the same as for the interferometer data alone , while the integrated flux density of the final image should be the same as measured from the single - dish data alone .
table [ t : comparison ] shows measurements of the total flux density , minimum / maximum values and noise level in the four resultant images .
all four images have very comparable noise levels , minimum values and maximum values .
the last two images also have very comparable ( within 3% ) total flux density , relative to the parkes value alone , while the first two have lower ( by 8% ) and higher ( by 7% ) flux densities , respectively .
the differences come , most likely , from the different weighting of the single - dish data employed by the different methods . while immerge slightly over - weights very short interferometric spatial frequencies , the ` linear combination ' method slightly over - weights single - dish data in the region of overlap .
this results in the total flux density being slightly lower in the first case , and slightly higher in the second case .
a few ( general ) remarks on all four methods : * the ` feathering ' or fourier method ( a ) is the fastest and the least computer intensive way to add short - spacings .
it is also the most robust way relative to the other three methods which all require a non - linear deconvolution at the end . *
the great advantage of the ` linear combination ' method is that it does not require either fourier transformation of the single - dish data , which can suffer severely from edge effects , nor deconvolution of the single - dish data which is especially uncertain and leads to amplification of errors . *
the ` default ' image method shows surprisingly reliable results when a significantly large single - dish is used . * adding during deconvolution when fitting both data sets simultaneously provides , theoretically , the best way to do the short - spacing correction .
however , this method depends heavily on a good estimate of the interferometer and single - dish noise variances .
in this article the need for , and methods of combining interferometer and single - dish data have been explained and demonstrated .
this combination is an important step when mapping extended objects and it is becoming a standard observing and data processing technique .
after a brief introduction to interferometry in section 2 the short - spacings problem and general approaches for its solution were discussed in section 3 . to _ fully and accurately _ fill in the missing short - spacings to interferometric data , a heterogeneous array scheme seems to be preferable .
a sufficiently large single - dish , with diameter at least 1.5 greater larger than the shortest interferometer baseline , provides the simplest option . in order to cross - calibrate single - dish and interferometer data sets , a significant overlap of spatial frequencies
is required .
four different combination methods , two linear and two non - linear , for the short - spacings correction have been discussed and the results of applying these methods to the case of hi observations of the smc presented .
linear methods are data combination in the spatial frequency domain and the ` linear combination ' method , while data combination during deconvolution provides two non - linear methods .
all four techniques yield satisfactory and comparable results .
many thanks to darrel emerson , chris salter and john dickey for reading the article and providing valuable suggestions for improvement .
i am also grateful to matthew wyndham for his help with figures , as well as assisting with last minute crises in organizing the meeting .
cornwell , t. j. , holdaway , m. a. , uson , j. m. 1993 , a&a , 271 , 697 cornwell , t. , & braun , r. 1989 , in asp conf .
vol . 6 , synthesis imaging in radio astronomy , ed .
r. perley , f. schwab , & a. bridle ( san francisco : asp ) , 167 | while , in general , interferometers provide high spatial resolution for imaging small - scale structure ( corresponding to high spatial frequencies in the fourier plane ) , single - dishes can be used to image the largest spatial scales ( corresponding to the lowest spatial frequencies ) , including the total power ( corresponding to zero spatial frequency ) . for many astrophysical studies ,
it is essential to bring ` both worlds ' together by combining information over a wide range of spatial frequencies .
this article demonstrates the effects of missing short - spacings , and discusses two main issues : ( a ) how to provide missing short - spacings to interferometric data , and ( b ) how to combine short - spacing single - dish data with those from an interferometer . |
recent works demonstrated that the quantum procedures can be examined more clearly using a topological approach rather than the classical presentations which utilize the matrices of complex numbers in an orderly manner ( @xcite , @xcite , @xcite ) . in this new approach pioneered by the works of abramsky and coecke one
replaces the matrices with geometrical primitives .
vicary has recently provided the topological analysis of deutsch - jozsa , hidden subgroup and single - shot grover algorithms by overcoming the difficulties that come from the use of an oracle and group representation theory @xcite .
his work not only enable new and simple proofs of correctness but also produce new generalizations of these algorithms . in @xcite ,
the author focuses on the case where a single iteration of the grover algorithm is sufficient to find a marked element in a set .
this algorithm can work correctly if exactly @xmath0 of the elements are marked . in this paper
, we consider in a similar fashion the more general case where one element is marked in a set @xmath1 of size @xmath2 .
we perform @xmath3 iteration using topological formalism .
this paper is organized as follows : in section 2 , we explain conventional grover algorithm .
section 3 is devoted to the topological semantics that we use in the paper .
finally , we give the presentation of the grover algorithm using topological semantics and prove its correctness in section 4 .
grover search algorithm is introduced to find a marked element in an array @xmath4 of size @xmath5 ( see @xcite ) .
the classical algorithms do this job in @xmath6 queries in average while it takes @xmath3 queries for grover search algorithm .
the marked element @xmath7 is defined using an indicator function @xmath8 where @xmath9 is the group of integers under addition modulo @xmath10 . here
@xmath11 is equal to @xmath12 for @xmath13 and to @xmath14 for all other elements of @xmath1 .
we then define the unitary operator @xmath15 as follows : @xmath16 here @xmath17 is an element of @xmath1 and @xmath18 is an element of @xmath9 .
the symbol @xmath19 is the addition in the group @xmath9 .
we can now introduce the quantum circuit for the grover algorithm as follows : where @xmath20 is @xmath21-fold hadamard gate and the unitary matrix @xmath22 is the inversion about the mean . @xmath23 and @xmath22 combined together separates the desired state after @xmath3 consecutive applications .
in other words , after @xmath3th step the probability of measuring any state except @xmath7 will be close to zero while probability of measuring @xmath7 will be greater than @xmath24 .
hence , the final measurement in the algorithm will give us the desired element @xmath7 .
in this section we present topological semantics that we need to build the grover algorithm .
these geometric primitives correspond to properties of finite groups and finite sets .
the content of this section can be found in the appendix of @xcite .
one can also refer to @xcite and @xcite for the mathematical foundation of the notation . here
we start with the identity map on a finite hilbert spaces .
this is represented by a vertical wire .
the following diagram represent a linear map @xmath25 the identity on the @xmath12-dimensional hilbert space is represented as the empty diagram : one can change the relative heights of the boxes and move the components around .
let @xmath26 $ ] be the complex vector space whose basis elements are the elements of a finite set @xmath1 .
for the sake of brevity , we denote this vector space by @xmath1 .
let @xmath27 be a basis element of @xmath1 .
we define the maps @xmath28 and @xmath29 as follows : @xmath30 @xmath31 the maps @xmath32 and @xmath33 gives @xmath1 a commutative algebra structure and they are represented as follows : their adjoints @xmath34 and @xmath35 are constructed in the following way : a function @xmath36 between finite sets extends to a linear map by @xmath37 .
the following diagram corresponds to this function : we have also morphisms with following two conditions these are called _ comonoid homomorphism _ conditions .
a chosen element @xmath38 corresponds to a function @xmath39 , which we denote graphically as : the deletion operator @xmath35 reduce the elements to unit scalars : @xmath17 and its adjoint @xmath40 are related as follows : let @xmath41 be a group . linearizing the multiplication function @xmath42 and unit element @xmath43 , we obtain the following diagrams for finite group @xmath41 , the representations are defined by the maps @xmath44 where @xmath45 is the algebra of @xmath46 matrices .
the representations have the following properties : we finalize this section with the following equation which is the result of basic theorems in representation theory ( the reader can refer to @xcite for detailed explanation ) : here the sum is over equivalence classes of irreducible representations of @xmath41 and @xmath47 is adjoint of an irreducible representation .
the topological structure of the ( multi - step ) grover algorithm as follows : the projective measurement is performed in the base system @xmath48 . @xmath49 corresponding the inversion about the mean ( i.e. @xmath22 ) can be decomposed : next we use the last identity of the section 3 to convert the algorithm to the following form : we now neglegt second part of this system as it is a product state . using the decomposition of @xmath49 we rewrite the system : the last equality is result of the identities stated in the last section .
it is clear that for the representation @xmath50 we have @xmath51 as a factor in our final measurement . for an unmarked element @xmath52
, we obtain the factor @xmath53 in the measurement as a result of the following topological identity : hence for such an unmarked element we will get the following result : @xmath54\ ] ] here the first summand in the bracket is the result of the figures which have one @xmath55 box while the second summand is from the figures without any @xmath55 boxes .
this result can be simplified to the following expression : @xmath56 hence @xmath57 is the probability of measuring an unmarked element .
we note that @xmath57 is less than @xmath24 and approaches to zero as @xmath21 gets larger .
the find the probability for the marked element we now calculate the total probability for unmarked elements .
since there are @xmath58 unmarked elements , we consider the value @xmath59 .
it is straightforward to check this value is less than @xmath24 and actually goes to zero as @xmath21 gets larger .
this implies that the probability of measuring the marked element will be at least @xmath24 after @xmath3 iterations .
this proves the the correctness of the algorithm .
one can also see the reason why the algorithm is repeated @xmath3 times by examining the expression @xmath60 where the power of the terms are the result of the the number of the iteration .
in this paper , we present multi - shot grover search algorithm using a topological semantics originated from categorical quantum mechanics program .
this new method provides a new and simple proof of the grover algorithm and opens the ways to new generalizations .
99 s. abramsky and b. coecke , _ a categorical semantics of quantum protocols _ , proceedings of the 19th annual ieee symposium on logic in computer science , pages 415 - 425 , 2004 .
ieee computer science press . | it has been shown in recent years that quantum information has a topological nature ( @xcite , @xcite , @xcite ) . in @xcite , vicary undergoes the study of quantum algorithms using this new topological approach .
the advantage of this new formalism is that it allows new proofs of correctness of the algorithms such as deutsch - jozsa , hidden subgroup and single - shot grover algorithms .
it also provide more clear insight to the generalizations of these algorithms . in this paper ,
we consider ( multi - step ) grover algorithm from this new perspective .
h *p * l o _ sl h
p l o i _ [ section ] [ th]corollary [ th]lemma [ th]proposition [ th]claim [ th]definition [ th]example [ th]lemma |
solving numerically some nonlinear partial differential equations , for example by using finite elements or finite volumes , often amounts to the resolution of some nonlinear system of equations of the form : [ eq : syst ] ( ) = _ ^n , where @xmath0 is the number of degrees of freedom and can be large .
one of the most popular method for solving the systems of the form is the celebrated newton - raphson method .
if this iterative procedure converges , then its limit is necessarily a solution to .
however , making the newton method converge is sometimes difficult and might require a great expertise .
nonlinear preconditioning technics have been recently developed in improve the performance of the newton s method , see for instance @xcite .
complex multiphase or unsaturated porous media flows are often modeled thanks to degenerate parabolic problems .
we refer to @xcite for an extensive discussion about models of porous media flows . for such degenerate problems , making newton s method converge is often very difficult .
this led to the development of several strategies to optimize the convergence properties , like for instance the so - called continuation - newton method @xcite , or trust region based solvers @xcite .
an alternative approach consist in solving thanks to a robust fixed point procedure with linear convergence speed rather that with the quadratic newton s method ( see for instance @xcite ) .
comparisons between the fixed point and the newton s strategies are presented for instance in @xcite ( see also @xcite ) .
combinations of both technics ( perform few fixed points iterations before running newton s algorithm ) was for instance performed in @xcite .
our strategy consists in reformulating the problem before applying newton s method .
the reformulation consists in changing the primary variable in order to improve the behavior of newton s method .
we apply this strategy to the so called _ richards equation _
@xcite modeling the unsaturated flow of water within a porous medium .
extension to more complex models of porous media flows will be the purpose of the forthcoming contribution @xcite .
denote by @xmath1 some open subset of @xmath2 ( @xmath3 ) representing the porous medium ( in the sequel , @xmath1 will be supposed to be polyhedral for meshing purpose ) , by @xmath4 a finite time horizon , and by @xmath5 the corresponding space - time cylinder .
we are interested in finding a saturation profile @xmath6 $ ] and a water pressure @xmath7 such that [ eq : richards ] _
t s - ( ( s)(p - ) ) = 0 , where the mobility function @xmath8 \to \r_+$ ] is a nondecreasing @xmath9 function that satisfies @xmath10 and @xmath11 , and where @xmath12 stands for the gravity vector . in order to ease the reading ,
we have set the porosity equal to @xmath13 in and neglected the residual saturation .
the pressure and the water content are supposed to be linked by some monotone relation [ eq : capi ] s = s(p ) , q where @xmath14 is a non - decreasing function from @xmath15 to @xmath16 $ ] . in what follows ,
we assume that @xmath17 for all @xmath18 , that corresponds to assuming that the porous medium is water wet , and that @xmath19 , implying in particular that @xmath20 . as a consequence of the lipschitz continuity of @xmath21 and of the integrability of @xmath14 on @xmath22 , one has [ eq : kirchbound ] ( s ) l^1(_- ) .
we denote by @xmath23 with the convention that @xmath24 if @xmath25 typical behaviors of @xmath21 and @xmath14 are depicted in figure [ fig : func ] . in the case where the domain @xmath1 and the finite time @xmath26 are large
, a classical hyperbolic scaling consisting in replacing @xmath27 by @xmath28 leads to the problem @xmath29 where the function @xmath30 is deduced from @xmath14 by @xmath31 letting @xmath32 tend to @xmath33 leads the maximal monotone capillary pressure graph @xmath34 & \text{if}\;p = 0 , \\ 1 & \text{if}\ ; p>0 . \end{cases}\ ] ] the richards equation then degenerates into a hyperbolic - elliptic problem .
our purpose can be extended to the degenerate case even though the graph @xmath35 lack regularity thanks to the so - called semi - smooth newton method ( cf .
moreover , because of the hyperbolic degeneracy , additional entropy criterions _ la _
carrillo @xcite are required in order to characterize the relevant solution . in order to simplify our purpose as much as possible , we focus on the case @xmath36 .
the problem is complemented by the initial condition [ eq : init ] s_|_t=0 = s_0 l^(;[0,1 ] ) , and by dirichlet boundary conditions on the pressure : [ eq : dirichlet ]
p_|_(0,t ) = p_d .
the regularity requirements on the boundary condition will be specified later on . at least from a mathematical point of view
, it is natural to solve by choosing @xmath37 or the kirchhoff transform @xmath38 ( to be defined later on at ) as a primary unknown then to deduce @xmath39 .
however , this approach lacks efficiency when one aims to solve the richards equation numerically , especially for dry media , i.e , when the saturation @xmath40 is close to @xmath33 . in this latter situation
, it turns out that the newton methods encounters difficulties to converge for solving the nonlinear system obtained thanks to standard implicit numerical methods ( say @xmath41-finite elements @xcite , mixed finite elements @xcite , or finite volumes @xcite ) .
a better choice as a primary unknown in the dry regions is the saturation @xmath40 , @xmath37 or @xmath38 being computed thanks to the inverse function of @xmath14 or @xmath42 respectively .
but choosing the saturation @xmath40 as the primary variable yields difficulties in the saturated regions , i.e. , where @xmath43 .
hence , a classical approach for solving numerically the richards equation consists in applying the so - called _ variable switch _ , that consists in changing the primary variable following the physical configuration ( see , e.g. , @xcite ) .
the main feature of our contribution consists in parametrizing the graph @xmath14 in order to stabilize the newton algorithm without implementing the possibly complex variable switch procedure .
this procedure is inspired from the one proposed by j. carrillo @xcite ( see also @xcite for numerical issues ) to deal with hyperbolic scalar conservation laws with discontinuous flux w.r.t .
the unknown .
let us introduce two continuously differentiable nondecreasing functions @xmath44\quad \text{and}\quad p : ( \tau_\star , \infty ) \to ( -\infty,\infty),\ ] ] where @xmath45 may be equal to @xmath46 , such that @xmath47 and @xmath48 this enforces in particular that @xmath49 and @xmath50 . in the case where @xmath51 , the functions @xmath40 and @xmath37 are then continuously extended into constants on @xmath52 .
it is assumed that the parametrization function @xmath40 satisfies [ eq : hyp - energy - bounded ] .
the kirchhoff transform @xmath53 is defined by [ eq : u ] u ( ) = _ 0^(s(a ) ) p(a ) da , _ .
it follows from the integrability property that [ eq : u_star ] u_:= _ _ u ( ) . for technical reasons ,
the function @xmath38 is artificially extended into a continuous onto function from @xmath15 to @xmath15 by setting [ eq : u - extended ] u ( ) = -_+ u _ , 0 _ , @xmath54 s()+u ( ) > 0 @xmath55 since @xmath56 for all @xmath57 .
since @xmath14 is an absolutely continuous function , this implies in particular that @xmath58 a.e . in @xmath59 .
such a parametrization of the graph @xmath14 always exists but is not unique . for instance , on can choose the parametrizations defined by @xmath60 or @xmath61 .
as it will appear in the analysis carried out in the core of the paper , a convenient parametrization should satisfy : there exist @xmath62 and @xmath63 such that [ eq : param - nondeg ] _ ( s ( ) , u ( ) ) ^ , .
thus , we assume that holds for the analysis .
this ensures in particular that the functions @xmath40 and @xmath38 are lipschitz continuous : [ eq : su - lip ] s_^ , u_^. for technical reasons that will appear in the analysis , we also assume that there exists @xmath64 such that [ eq : u - coercive ] c(u()+1 ) , 0 .
it is also assumed that [ eq : liminf - p ] p ( ) > 0 .
this assumption is very naturally satisfied for any non - degenerate parametrization in the sense of of a reasonable function @xmath14 , but unphysical counterexamples can be designed , enforcing us to set as an assumption .
the function @xmath40 from @xmath65 $ ] to @xmath16 $ ] is nondecreasing and onto .
therefore , one can define the function @xmath66 \to [ \tau_\star , 0]$ ] by [ eq : s-1 ] s^-1(a ) = \ { x 0 | s(x ) = a } , a .
this allows to define an initial data @xmath67 as @xmath68 such that @xmath69 . choosing @xmath70 as the primary variable leads to the following doubly
degenerate parabolic equation @xmath71 this equation turns to [ eq : richards - tau ] _
t s ( ) + ( ( s ( ) ) - u ( ) ) = 0 q , at least if @xmath72 ( this will be ensured , cf . theorem [ thm : weak ] ) .
it is relevant to impose the boundary condition [ eq : dirichlet - tau ] _
|_(0,t ) = p^-1(p_d)=:_d _ . as a counterpart of .
it is finally assumed that @xmath73 can be extended to the whole @xmath74 in a way such that [ dirichlet - tau - reg ] _
d c^1(q ) , _ .
the regularity required on @xmath73 is not optimal and can be relaxed .
however , the treatment of the boundary condition is not central in our purpose , hence we stick to [ def : weak ] a measurable function @xmath75 is said to be a weak solution to the problem , , if @xmath76 , if @xmath77 , and if , for all @xmath78 , one has @xmath79 the following statement summarizes known results about the weak solutions . [ thm : weak ] there exists a unique weak solution @xmath80 to the problem , , in the sense of definition [ def : weak ] .
moreover , @xmath72 a.e . in @xmath81 , @xmath82;l^p(\o))$ ] for all @xmath83 , and , given two solutions @xmath84 corresponding to two initial data @xmath85 and @xmath86 , we have [ eq : compl1 ] _ (s((,t ) ) - s((,t)))^_(s_0 ( ) - s_0())^ , t .
existence of weak solutions have been proved by by alt and luckhaus in their seminal paper @xcite .
we refer to @xcite ( see also @xcite ) for extended details on the uniqueness proof and on the comparison principle .
the time continuity of the saturation can be proved as in @xcite . in [
sec : scheme ] , we present an implicit monotone finite volume scheme @xcite designed for approximating the entropy solution @xmath70 of and . first , we describe in [ ssec : mesh ] how the domains @xmath1 ( and then @xmath81 ) has to be meshed .
in particular , the mesh has to fulfill the so - called _ orthogonality condition _ so that the diffusion fluxes can be discretized using a simple two - point flux approximation @xcite .
the finite volume scheme is described in [ ssec : scheme ] .
this scheme yields a nonlinear system of equations [ eq : syst - n ] _
n(^n ) = , n \{1 , ,
n } to be solved at each time step .
the existence and the uniqueness of the solution @xmath87 to the nonlinear system is proved at
[ ssec : anal - scheme ] .
once we know that the scheme admits one unique solution @xmath87 , we discuss its effective computation thanks to newton s method in [ sec : newton ] .
it is in particular proved in [ ssec : und ] that the jacobian matrix is uniformly non - degenerate , so that we can use newton - kantorovich theorem to claim the convergence of newton s method .
the quantification of the error linked to the inexact resolution of the nonlinear system is performed in [ ssec : inexact ] .
finally , some numerical results are presented in [ sec : num ] in order to illustrate the efficiency of our approach .
in this work , we only consider cylindrical discretizations of @xmath81 that consist in discretizing space and time separately .
the approximation of the diffusive fluxes we propose relies on the so - called _ two - point flux approximation_. this approximation is consistent if the problem is isotropic and if the mesh satisfies the so - called orthogonality condition ( see e.g. @xcite ) .
[ def : mesh ] an _ admissible mesh _ @xmath88 of @xmath1 is given by a set @xmath89 of disjointed open bounded convex subsets of @xmath1 called _ control volumes _ , a family @xmath90 of subsets of @xmath91 called _ edges _ contained in hyperplanes of @xmath2 with strictly positive measure , and a family of points @xmath92 ( the so - called _ cell centers _ ) .
it is assumed that the mesh integrates the whole @xmath1 , i.e. , @xmath93 .
the boundary of the control volumes are made of edges , i.e. , for all @xmath94 , there exists a subset @xmath95 of @xmath90 such that @xmath96 .
furthermore , @xmath97 . for any @xmath98 with @xmath99 ,
either the @xmath100-dimensional lebesgue measure of @xmath101 is @xmath33 , or @xmath102 for some @xmath103 . in the latter case ,
we write @xmath104 .
we denote by @xmath105 the set of the internal edges , and by @xmath106 of the boundary edges . finally , the family of points @xmath92 is such that @xmath107 ( for all @xmath94 ) and , if @xmath104 , it is assumed that the straight line @xmath108 is orthogonal to @xmath109 .
for all @xmath110 , there exists one unique cell @xmath111 such that @xmath112 .
then we denote by @xmath113 the projection of @xmath114 over the hyperplane containing @xmath109 , and we assume that @xmath113 belongs to @xmath109 . in what follows , we denote by @xmath115 the @xmath116-dimensional lebesgue measure of the control volume @xmath94 , and by @xmath117 the @xmath100-lebesgue measure of the edge @xmath118 . for all @xmath119 ,
we denote by @xmath120 . since @xmath104 is supposed to be orthogonal to @xmath121 , then @xmath122 .
we define the _ transmissibilities _
@xmath123 by @xmath124 the space of the degrees of freedom ( including those prescribed by the boundary condition ) is @xmath125 while the interior degrees of freedom ( for which a nonlinear system has to be solved ) are described by the space @xmath126 [ def : time - disc ] a _ _
t__ime discretization of @xmath127 is given by an integer value @xmath128 and a sequence of real values @xmath129 .
for all @xmath130 the time step is defined by @xmath131 .
[ def : space - time - disc ] a space - time discretization @xmath132 of @xmath81 is a family @xmath133 where @xmath134 is an admissible mesh of @xmath1 in the sense of definition [ def : mesh ] and @xmath135 is a discretization of @xmath127 in the sense of definition [ def : time - disc ] .
the spaces of the degrees of freedom are defined by @xmath136 and @xmath137 let @xmath138 , then we denote by @xmath139 for @xmath140 . following the approach proposed in @xcite
, we introduce reconstruction operators .
first , we define the linear operator @xmath141 by @xmath142 it is extended into the time - and - space reconstruction linear operator @xmath143 by setting @xmath144 , \qquad \forall \bv = { ( v_k^n , v_\sig^n)}_{k\in\tt , \sig \in \ee_{\rm ext}}^{1\le n \le n } \in \left(\x_\tt\right)^n.\ ] ] the study to be performed also requires the introduction of a so - called discrete gradient
. we will remain sloppy about the construction of the discrete gradient .
we only highlight the properties we will use in the sequel .
[ lem : reconstruct ] let @xmath89 be an admissible discretization of @xmath1 in the sense of definition [ def : mesh ] . there exists a linear operator @xmath145 such that for all @xmath146 and @xmath147 in @xmath148 , one has @xmath149 moreover , let @xmath150 be a sequence of admissible discretizations of @xmath1 in the sense of definition [ def : mesh ] such that @xmath151 tends to @xmath33 while @xmath152 remains bounded as @xmath153 tends to @xmath154 , and let @xmath155 be a family such that @xmath156 for all @xmath157 and @xmath158 then there exists @xmath159 such that , up to an unlabeled subsequence , one has [ eq : compact ] _ _ m _ m v l^2( ) , and [ eq : consist ] _
_ m _ m v l^2()^d .
additionally , if @xmath160 is discretized into @xmath161 by setting @xmath162 then @xmath163 converges strongly in @xmath164 towards @xmath165 as @xmath153 tends to @xmath166 .
note that the discrete gradient reconstruction used in @xcite ( which is the usual one for two - point flux approximations of diffusion operators ) does not meet the requirements of lemma [ lem : reconstruct ] since is not fulfilled .
however , a reconstruction as prescribed by lemma [ lem : reconstruct ] can be obtained as a particular case of the so - called sushi scheme on super - admissible meshes " ( cf .
* lemma 2.1 ) ) . finally , the reconstruction operators @xmath167 and @xmath145 are extended to the space - times framework into @xmath168 and @xmath169 defined for all @xmath170 by [ eq : piddgraddd ] _ ( , t ) = _ ^n , _ ( , t ) = _ ^n , t ( t^n-1,t^n ] .
the initial data @xmath85 is discretized into @xmath171 by setting [ eq : sk0 ] s_k^0 = 1m_k_k s_0 ( ) , k. notice that @xmath172 since @xmath173 .
we define @xmath174 as [ eq : btau0 ] ^0 = s^-1(^0 ) , so that @xmath175 the boundary condition @xmath176 is discretized by setting for all @xmath177 [ eq : tauddisc ] _ d,^n=
_ d ( _ , t^n ) , _ ext , _ d , k^n= _ d(_k , t^n ) , k .
then we denote by @xmath178 , and by @xmath179 .
it follows from formula and from the regularity of @xmath73 that @xmath180 let @xmath181 .
assume that the state @xmath182 is known .
the implicit finite volume scheme is obtained by writing the local conservation of the volume of each fluid on the control volumes , i.e. , [ eq : scheme ] m_k + _ _ k f_k,^n = 0 , k , where @xmath183 denotes the outward w.r.t . @xmath111 flux across the edge @xmath109 at time step @xmath184 .
denote by @xmath185 the outward w.r.t .
@xmath111 normal to @xmath109 , and by @xmath186 for all @xmath112 and all @xmath187 .
denote by @xmath188 then the fluxes @xmath183 across @xmath189 is defined by [ eq : fksig ] f_k,^n = m _ ( ( s(_k^n ) ) g_k,^+ - ( s(_k,^n ) ) g_k,^- ) + a _ ( u(_k^n ) - u(_k,^n ) ) .
note in particular that the scheme is locally conservative , i.e. , @xmath190 combining , the scheme can be rewritten in a condensed form as [ eq : scheme2 ] _
k^n-1 , ( _ l^n)_lk , ( _ d,^n)__k , ext)=0 , k , where @xmath191 is nondecreasing w.r.t . its first argument and nonincreasing w.r.t .
the others thanks to the monotonicity of the functions @xmath192 and @xmath38 .
it is worth noticing that [ eq : div - nulle ] _ _ k m_g_k , = 0 , k .
therefore , the convective flux balance can be reformulated , yielding @xmath193 in this section , we analyze the system obtained for a fixed admissible discretization @xmath132 of @xmath81 . in what follows ,
we denote by @xmath194 the following lemma is a discrete counterpart of the @xmath195-contraction principle on the exact solution .
[ lem : contract ] let @xmath196 and @xmath197 be two elements of @xmath198 , and let @xmath199 and @xmath200 in @xmath198 be two corresponding solutions , then [ eq : contract - disc ] _ | _
s(^n ) - _ s(^n ) | _ | _ s(^n-1 ) - _ s(^n-1 ) | .
it follows from the monotonicity of @xmath191 that @xmath201 since @xmath202 is either equal to @xmath203 or to @xmath204 , one has [ eq : comp - top ] _
k(_k^n_k^n , _
k^n-1_k^n-1 , ( _ l^n _ l^n)_lk , ( _ d,)__k , ext ) 0 .
similar calculations lead to [ eq : comp - bot ] _
k(_k^n_k^n , _
k^n-1_k^n-1 , ( _ l^n _ l^n)_lk , ( _ d,)__k , ext ) 0 . summing with and over @xmath94 yields .
[ lem : unique ] given @xmath205 , then there exists at most one solution @xmath206 to the scheme . as a direct consequence of lemma [ lem : contract ] , @xmath207 for all @xmath94 . subtracting the system yielding @xmath208 to the one corresponding to @xmath199 leads to @xmath209 where we have set @xmath210 .
it follows from classical arguments that @xmath211 .
bearing the nondegeneracy condition in mind , we get that @xmath212 .
the following lemma shows that the solution @xmath87 to the scheme is always greater than @xmath213 .
therefore , the extension we chose for the function @xmath38 does not affect the result .
[ lem : pos ] let @xmath214 be the solution to , then @xmath215 for all @xmath216
. there is nothing to prove if @xmath217 , hence let us assume that @xmath218 is finite .
let @xmath111 be a cell such that @xmath219 for all @xmath112 , and assume that @xmath220 . since @xmath221 and @xmath222 , one gets that @xmath223 the extension of @xmath38 ensures that @xmath224 .
the left - hand side of the above relation is therefore negative , hence a contradiction with the assumption @xmath220 .
let @xmath177 , then define @xmath225 and @xmath226 by @xmath227 it is easy to verify ( see e.g. @xcite ) that [ eq : borne - eee ] 0 ^n ( ) m_(1-s_l^1(_+ ) + s_l^1(_- ) ) , _ .
moreover , it follows from the @xmath228 regularity of @xmath73 and from the fact that @xmath229 that [ eq : diffeee ] |^n ( ) - ^n-1()| ^n m__t _ d_. we define the lipschitz continuous function @xmath230 by [ eq : xi ] ( ) = _ 0^ a = _ 0^ a , , then it follows from cauchy - schwartz inequality that [ eq : xi2 ] ( a - b)(u(a ) - u(b ) ) ( ( a ) - ( b))^2 , ( a , b ) ^2 .
moreover , the lipschitz continuity of @xmath231 implies that @xmath232 hence it follows from assumption that [ eq : xi - coercive ] c ( ( ) + 1 ) , _ + , then , in particular , one has [ eq : xi - coercive2 ] _ ( ) = + .
[ lem : wbv ] let @xmath87 be a solution to the scheme . then there exists @xmath233 depending only on @xmath1 and @xmath73 such that the following estimate holds : [ eq : energy ] ^n(^n ) + ^n ( c_1 + 12_|_()|^2 ) ^n-1(^n-1 ) .
we multiply the equation by @xmath234 and sum over @xmath187 .
using , this provides [ eq : t123 ] t_1 + ^n ( t_2 + t_3 ) = 0 , where @xmath235 it follows from the convexity of @xmath236 ( see e.g. ( * ? ? ?
* proposition 3.7 ) ) that @xmath237 then thanks to , one gets that [ eq : t1 ] t_1
^n(^n ) - ^n-1(^n-1 ) - ^n m__t _ d_. the term @xmath238 can be estimated following the path of @xcite .
denote by @xmath239 the function defined by @xmath240 since @xmath241 , one has @xmath242 for all @xmath243 , and thus @xmath244 moreover , since @xmath245 , one has @xmath246 therefore , for all @xmath247 , one has @xmath248 thanks to and to assumption , one has @xmath249 hence @xmath250 is bounded .
simple calculations show that for all @xmath251 , one has @xmath252 rewriting [ eq : t21 ] t_2
= t_21 + t_22 with @xmath253 we get that @xmath254 using the boundedness of @xmath250 , we get that [ eq : t21 ] t_21 - m_ || _ . on the other hand , a classical reorganization of the term @xmath255 provides @xmath256 therefore , it follows from the regularity of @xmath73 and from the boundedness of @xmath21 that [ eq : t22 ] t_22 - || _ ( d m__d_- m_ _ d _ ) . finally ,
it results from , from the relation @xmath257 , and from that @xmath258 therefore , it follows from that [ eq : t3 ] t_3 12 _ |_(^n)|^2 - 2 .
putting in ends the proof of lemma [ lem : wbv ] .
[ prop : existence ] let @xmath259 , there exists a unique solution @xmath260 to the scheme . moreover , it satisfies @xmath215 for all @xmath261 .
the uniqueness of the solution was proven at lemma [ lem : unique ] while the fact that @xmath262 was the purpose of lemma [ lem : pos ] .
therefore , it only remains to show the existence of a solution .
it follows from estimate that @xmath263 since @xmath231 is coercive , there exists @xmath264 depending on the data ( among which @xmath196 , @xmath73 , the mesh @xmath89 and the time step @xmath265 ) such that @xmath266 this estimate allows us to make use of a topological degree argument ( see @xcite ) to prove the existence of one solution to the scheme .
the convergence of the scheme can be proved following the path proposed in @xcite .
enhanced convergence properties can be obtained thanks to the recent contribution @xcite .
but this is not the goal of this paper .
we are mainly interested in the practical computation of the approximate solution at fixed discretization parameters .
in particular , we focus on the behavior the newton s method .
the numerical scheme amounts for all @xmath140 to the nonlinear system [ eq : syst - n2 ] _
n(^n ) = ( f_k^n(^n))_k= , _
n ^2(^#;^ # ) , where @xmath267 assume that the jacobian matrix @xmath268 of @xmath269 at @xmath87 is not singular ( this will be shown for nondegenerate parametrizations , cf .
proposition [ prop : jac ] ) .
approximating the solutions to the system with the newton method consists in the construction of a sequence @xmath270 defined by [ eq : newton ] ^n,0 = ^n-1 , + ^n , k+1 = ^n , k - ^-1 _ n(^n , k ) .
if the method converges , the solution @xmath87 is then defined as [ eq : btau - n ] ^n = _ k ^n , k .
moreover , the convergence speed is asymptotically quadratic if @xmath271 , i.e. , [ eq : quadra ] ^n , k - ^n c ^n , k-1 - ^n^2 , k k _ where , denoting by @xmath272 a neighborhood of @xmath87 in @xmath273 , the quantity @xmath264 ( as well as @xmath274 ) depends on [ eq : cond - quadra - loc ] _ ( ^n ) ^-1 _ ( ^n ) d^2_n ( ) .
since @xmath87 is unknown , a sufficient condition to ensure that holds for some @xmath64 is the following uniform non - degeneracy condition [ eq : undc ] _
^ # ^-1 < .
we cite here a simplified version of the so - called newton - kantorovich theorem ( see , e.g. , @xcite ) .
we refer to @xcite for a quantitative version of the theorem , and to @xcite to a non - smooth version .
[ thm : kanto ] assume that there exists two positive quantities @xmath275 and @xmath276 such that [ eq : kanto-1 ] _ _ _ _ n ( ) c_2 _ _ ^-1 c_3 , then there exists @xmath277 such that @xmath278 our strategy in the sequel is to prove that a non - degenerate parametrization ( in the sense of ) yields estimates in the subordinate matrix @xmath13-norm .
[ rem : non - smooth ] the assumption @xmath279 can be relaxed .
more precisely , newton method can be extended to the case where @xmath280 is merely semi - smooth @xcite following the way proposed by qi and sun in @xcite .
quadratic convergence is preserved in this nonsmooth case provided @xmath269 is semi - smooth of order 1 , cf .
* theorem 3.2 ) .
our study can be extended to this more general case , but , for the sake of simplicity , we have chosen to reduce our presentation to the classical smooth framework .
because of the elliptic degeneracy of the problem when @xmath282 , we can not apply the the results of @xcite to get estimates on the jacobian matrix @xmath283 . we need to introduce some technical material to circumvent this difficulty .
let us first define that notion of @xmath284-transmissive path ( see also @xcite ) let @xmath285 , let @xmath286 and let @xmath287
a row - wise @xmath284-transmissive path @xmath288 of length @xmath289 from @xmath290 to @xmath291 associated to the matrix @xmath292 consists in a list @xmath293 with a. @xmath294 , @xmath295 , and @xmath296 if @xmath297 ; b. for all @xmath298 , one has @xmath299 .
the list @xmath293 is a column - wise @xmath284-transmissive path associated to the matrix @xmath292 if it is a row - wise @xmath284-transmissive path associated to @xmath300 .
[ def : ddm ] let @xmath301 be such that @xmath302 . 1 .
a matrix @xmath303 is said to be a row - wise @xmath304-m - matrix if a. for all @xmath305 , one has @xmath306 , @xmath307 for all @xmath308 and @xmath309 ; b. the set @xmath310 is not empty , and for all @xmath311 , there exists a @xmath284-transmissive path @xmath288 with @xmath312 .
a matrix @xmath303 is said to be a column - wise @xmath304-m - matrix if @xmath300 is a row - wise @xmath304-m - matrix .
it is well known that @xmath281-matrices are invertible matrices .
the goal of lemma [ lem : row ] and of corollary [ coro : column ] is to get uniform estimates on the inverse of a uniform m - matrix .
let us stress that the estimates we obtain are far from being optimal in the applications we have in mind , namely the finite volume discretization of richards equation .
[ lem : row ] let @xmath313 be a row - wise @xmath304-m - matrix , then there exists @xmath264 depending only on @xmath314 and @xmath128 such that @xmath315 for the ease of reading , we denote @xmath316 instead of @xmath317 , and @xmath318 instead of @xmath319 .
the property @xmath320 is equivalent to [ eq : pf1 ] 1 ^ -1=_=1 = .
let @xmath321 with @xmath322 .
we assume , without loss of generality that there exists @xmath323 such that @xmath324 . in the sequel ,
we denote by @xmath325 the maximum length of a transmissive path , i.e. , [ eq : ll ] : = _ j _ ( ) ^c _ j_()length ( ( i , j ) ) n-1 .
assume first that @xmath326 , then @xmath327 , hence @xmath328 .
assume now that @xmath329 , and let @xmath330 be a transmissive path with @xmath294 , @xmath331 and @xmath332 .
denote by @xmath333 let us show by induction that [ eq : hr ] z_k_p 1- c_p p \{0 ,
, l}. since @xmath334 , the relation holds for @xmath335 .
now suppose that holds for some @xmath336 .
one knows from that @xmath337 hence , using , that @xmath338 , and that @xmath339 for all @xmath340 , one gets that @xmath341 since @xmath342 and @xmath343 , this leads to @xmath344 so that the proof of is complete . as a consequence
, one gets that @xmath345 , and thus that @xmath346 using that @xmath331 , we obtain that @xmath347 therefore , @xmath348 since the length @xmath289 of the path @xmath349 is bounded by @xmath325 defined by .
[ coro : column ] let @xmath313 be a column - wise @xmath304-m - matrix , then there exists @xmath264 depending only on @xmath314 and @xmath128 such that @xmath350 [ prop : jac ] assume that @xmath40 and @xmath38 satisfy he non - degeneracy condition , then there exist @xmath275 depending only on @xmath351 and @xmath89 , and @xmath276 depending only on @xmath352 and @xmath89 such that @xmath353^{-1 } \|_1 \le c_3 , \qquad
\forall \btau \in \r^{\#\tt}.\ ] ] let us first make the jacobian matrix @xmath354 explicit : @xmath355 proving that @xmath356 is easy since all the coordinates of @xmath283 are uniformly bounded w.r.t .
@xmath357 thanks to the upper bound on @xmath358 and @xmath359 .
let us now prove that @xmath360^{-1 } \|_1 \le c_3 $ ] .
thanks to corollary [ coro : column ] , it suffices to check that @xmath283 is a column - wise @xmath361-m - matrix where @xmath284 and @xmath362 depend on the prescribed quantities .
first , it is easy to check that [ eq:(i ) ] j_lk^n ( ) 0 l k , _
l j_lk^n ( ) 0 , 0 _ 1 j_kk^n ( ) , where @xmath363 therefore , condition ( i ) in definition [ def : ddm ] is fulfilled .
it follows from the non - degeneracy condition that for all @xmath364 and all @xmath187 , either @xmath365 or @xmath366 .
let @xmath111 be such that @xmath365 , then @xmath367 whence @xmath368 . on the other hand , if @xmath366 and @xmath369 , then @xmath370 as a consequence , if @xmath371 , one has necessarily that @xmath366 and @xmath372
. but in this case , @xmath373 the matrix @xmath374 defined by @xmath375 is irreducible and admits @xmath284-transmissive paths from @xmath111 to @xmath289 for all @xmath376 .
this ensures that @xmath377 admits a @xmath284-transmissive path from any cell @xmath378 to any cell @xmath379 .
therefore , condition ( ii ) of definition [ def : ddm ] is fulfilled .
@xmath380 is then a row - wise @xmath304-m - matrix , thus @xmath377 is a column - wise @xmath304-m - matrix .
the following corollary is a straightforward compilation of newton - kantorovich theorem [ thm : kanto ] with proposition [ prop : jac ] .
[ coro : local - conv ] there exists @xmath381 such that the newton method converges as soon as @xmath382 .
the radius @xmath383 appearing in corollary [ coro : local - conv ] can be estimated thanks to @xcite as soon as the second order derivatives @xmath384 and @xmath385 of @xmath40 and @xmath38 are uniformly bounded .
it appears in a fairly natural way that @xmath383 is a non - decreasing function of @xmath265 . then choosing @xmath265 small enough
, the variation between the time steps @xmath386 and @xmath184 is small and @xmath387 therefore , the convergence of the newton method is ensured if one uses an adaptive time - step algorithm ( as for instance in @xcite ) .
assume that @xmath388 is exactly known , then the exact solution @xmath87 is obtained as the limit of @xmath389 . computing the exact value of @xmath87 is impossible and a convenient criterion must be adopted in order to stop the iterative procedure .
this yields errors that accumulate along time .
the goal of this section is to quantify the error induced by the inexact resolution of the nonlinear system . in this section
, we assume that @xmath388 is exact , and we want to quantify the error corresponding to one single iteration .
in what follows , we consider the following residual based stopping criterion : [ eq : stopping ] _
n(^n , k)_1 = _ k |f_k(_k^n , k)| ^n for some prescribed tolerance @xmath36 .
we denote by @xmath390 when is fulfilled and the loop is stopped .
then the non - degeneracy of @xmath391^{-1}$ ] provides directly the following error estimate : @xmath392 more than in the variable @xmath70 , whose physical sense is unclear , we are interested in evaluating the error on the reconstructed saturation profile .
thanks to the lipschitz continuity of @xmath40 recall the non - degeneracy assumption , we have [ eq : err - tau ] _ s(^n _ ) - _ s(^n)_l^1( ) c_4 ^n , where @xmath393 . in the previous paragraph , it was assumed that @xmath388 was exactly known . in practice , one may consider that we know @xmath394 exactly , but merely the approximation of @xmath395 obtained after stopping the newton iterative procedure after a finite number of iterations .
denote by @xmath396 the iterated exact solutions to the scheme , and by @xmath397 the iterated inexact solutions obtained _ via _ the newton method with the stopping criterion .
define @xmath398 by [ eq : ffneps ] _
n , ( ) = _ n ( ) + s(^n-1)- s(_^n-1 ) . in particular , one has @xmath399 for all @xmath357 and the results of [ ssec : und ] still hold for @xmath400 instead of @xmath269 .
the inexact newton method then writes 1 .
initialization : define @xmath401 by .
2 . from @xmath386 to @xmath184 : 1 . set @xmath402 2 .
iterate newton s algorithm until @xmath403 then , as claimed by the following statement , the @xmath404 error on the reconstructed saturation growth at most linearly with time . in particular , no exponential amplification of the error occurs in this context .
[ prop : err - inexact ] let @xmath405 be the exact solution to the scheme , and let @xmath406 be the approximate solution computed by the inexact newton method , then [ eq : err - inexact ] _ s(^n _ ) - _ s(^n)_l^1( ) c_4 t^n , n \{0 , , n } , where @xmath407 is introduced was .
we perform the proof by induction .
estimate clearly holds for @xmath408 since the initialization is exact .
now assume that it holds for @xmath409 .
denote by @xmath410 the exact solution of the system @xmath411 .
then it follows from the discussion carried out in [ sssec : onestep ] that @xmath412 on the other hand , applying lemma [ lem : contract ] , one gets that @xmath413 one concludes thanks to the triangle inequality .
in the previous section , we have shown that as long as the parametrization @xmath414 satisfies the condition , the newton s method applied to the problem exhibits local convergence . moreover , due to proposition [ prop : err - inexact ] , the error resulting from an inexact newton s method can be efficiently controlled .
remark that the constant @xmath407 in depends on the nonlinearities @xmath14 and @xmath415 only through the quantities @xmath416 and @xmath417 .
therefore , the estimate is robust with respect to the hydrodynamic properties of the soil .
we illustrate this fact by numerical experiments presented below . as an example of mobility / capillary pressure relations we consider a popular brooks - corey model @xcite for which we compare the efficiency of newton s method resulting from the parametrization @xmath61 and the one satisfying with @xmath418 .
let @xmath419 and @xmath420 , in brooks - corey model the saturation and the mobility functions are given by @xmath421 and @xmath422 providing in terms of kirchhoff transform @xmath423 with @xmath424 and @xmath425 .
remark that the parametrization based on @xmath61 and @xmath426 , referred as @xmath427formulation , do not satisfy since the derivative of @xmath428 is singular at @xmath429 . as an alternative we consider the parametrization @xmath414 defined by the equation @xmath430 and the condition @xmath431 , to which we refer as @xmath432formulation .
we obtain the following explicit formulas @xmath433 and @xmath434 with @xmath435 .
we consider a bidimensional porous domain @xmath436 , which is initially very dry with @xmath437 in @xmath1 .
the water is injected at the pressure @xmath438 through the portion of an upper boundary @xmath439 .
the gravity vector is given by @xmath440 and a zero flux boundary condition is prescribed on @xmath441 .
the computations are performed on two quasi - uniform space discretizations of @xmath1 composed of 396 and 1521 vorono cells referred as mesh 1 and mesh 2 .
the final time is set to @xmath442 and the time step is equal to @xmath443 .
figure [ fig_case_1_snap ] shows , for @xmath444 and @xmath445 , the distribution of saturation and generalized pressure @xmath38 at different times .
the results are visualized on triangular delaunay mesh dual to mesh 1 . in order to challenge the robustness of both formulations , we set @xmath445 and we let the parameter @xmath446 take values in the set @xmath447 . for each value of @xmath446 we compute , using @xmath432formulation and tolerance @xmath448 , the reference solution denoted by @xmath449
. then , for both formulations and for the values of @xmath450 , we perform the calculations measuring the total number of newton s iteration and the deviation , in the discrete @xmath451 norm , of the `` observable '' variables @xmath38 and @xmath40 from the reference solution . for a given value of @xmath446 and of the tolerance @xmath32 ,
we denote by @xmath452 and @xmath453 the approximate solution of obtained using the @xmath427formulation and @xmath432formulation respectively .
the error produced by inexact newton s method is measured by the quantities @xmath454 and @xmath455 for @xmath427formulation , and @xmath456 and @xmath457 for @xmath432formulation .
figure [ fig_case_1 ] exhibits , for meshes 1 and 2 , the behavior of the relative error @xmath458 and @xmath459 , @xmath460 as the function of an average number of iterations per time step required by newton s method in order to converge up to the given tolerance @xmath32 .
we observe that in order to achieve the same precision @xmath427formulation require a much larger number of iterations then @xmath432formulation .
moreover the number of newton s iterations , for @xmath427formulation , increases with @xmath446 , whereas @xmath70-formulation remains robust with respect to this parameter .
the contrast in the efficiency of two formulations is amplified as the mesh is refined .
the goal of this test case is to give a numerical evidence that the inexact newton s method applied to the @xmath38-formulation produces large errors due to troubles in the conservation of mass .
to do so , we prescribe a zero - flux condition on the whole boundary of @xmath436 , whereas the initial saturation satisfies @xmath461 with @xmath462 .
we set @xmath444 and @xmath445 .
the effects of gravity neglected so that the flow is only driven by diffusion .
figure [ fig_case_2_snap ] exhibits , for different times , the saturation field associated with the reference solution , which is computed using @xmath432formulation and the tolerance @xmath463 . since the flow is very slow the large time steps are needed , we set @xmath464 and @xmath465 .
computations are performed using mesh 1 .
let @xmath466 , we define the relative mass conservation error by @xmath467 and @xmath468 figure [ fig_case_2_errs ] exhibits the @xmath469 relative saturation error @xmath470 , @xmath471 , and the relative mass conservation error @xmath472 and @xmath473 as the functions of @xmath32 .
as one can see the @xmath469 error produced by @xmath427formulation is dominated by the mass conservation error .
remark that even for the rather small values @xmath474 or @xmath475 , the error produced by @xmath427formulation is still significant ( see figures [ fig_case_2_cut ] ) . in contrast
@xmath432formulation leads to mush smaller errors and , for any value of @xmath32 , conservatives the mass up to a precision of order @xmath476 .
the very high accuracy for mass conservation observed with the @xmath432formulation can be explained as follows . with the values of the parameters
@xmath477 we have chosen , one has @xmath478 for @xmath479 $ ] and hence for all @xmath480 in view of initial and boundary conditions .
in addition , in view of we have @xmath481 therefore , at each step of inexact newton s method , the flux contribution globally offset , hence we have @xmath482 since @xmath483 is linear , one has @xmath484 , which implies that the mass is exactly conserved ( assuming that linear algebraic computations are exact ) at each iteration of newton s method .
k. brenner , m. groza , l. jeannin , r. masson , and j. pellerin .
immiscible two - phase darcy flow model accounting for vanishing and discontinuous capillary pressures : application to the flow in fractured porous media . ,
2016 . to appear . c. cancs and c. guichard
. numerical analysis of a robust free energy - diminishing finite volume scheme for degenerate parabolic equations with gradient structure .
preprint hal : hal-01119735 , accepted for publication in found .
2016 .
j. droniou , r. eymard , t. gallout , and r. herbin .
gradient schemes : a generic framework for the discretisation of linear , nonlinear and nonlocal elliptic and parabolic equations .
, 23(13):23952432 , 2013 .
r. eymard , t. gallout , and r. herbin .
discretization of heterogeneous and anisotropic diffusion problems on general nonconforming meshes sushi : a scheme using stabilization and hybrid interfaces . , 30(4):10091043 , 2010 .
f. a radu , i. s. pop , and p. knabner .
newtontype methods for the mixed finite element discretization of some degenerate parabolic equations . in
_ numerical mathematics and advanced applications _ , pages 11921200 .
springer , 2006 .
konstantin brenner + laboratoire jean - alexandre dieudonn , universit de nice sophia antipolis , + team coffee inria sophia antipolis mditerrane , + 06108 nice cedex 02 , france . | the nonlinear systems obtained by discretizing degenerate parabolic equations may be hard to solve , especially with newton s method . in this paper , we apply to richards equation a strategy that consists in defining a new primary unknown for the continuous equation in order to stabilize newton s method by parametrizing the graph linking the pressure and the saturation .
the resulting form of richards equation is then discretized thanks to a monotone finite volume scheme .
we prove the well - posedness of the numerical scheme .
then we show under appropriate non - degeneracy conditions on the parametrization that newton s method converges locally and quadratically .
finally , we provide numerical evidences of the efficiency of our approach .
richards equation , finite volumes , newton s method , parametrization * ams subjects classification .
* 65m22 , 65m08 , 76s05 |
about 50% of the known main - sequence - stars belong to the class of binary or multiple stellar systems . due to inherent difficulties in monitoring the radial velocities of multi - star systems ,
these systems have been mostly ignored by researchers performing searches for extra - solar planets . however , in last years , due to improving precision and advances of detection methods , binaries have become increasingly interesting targets of planet searches , and several research programs looking for planets in binary systems are in progress .
the @xmath0 centauri binary is of special interest .
being the closest stellar system to the sun , as well as composed of two stars very similar to the sun , it has been always a major target for extrasolar planet surveys , although it was not until recently that a putative planet has been found orbiting around @xmath0 centauri b with a period of 3.236 days ( @xcite ; for an opposing view see @xcite ) .
furthermore , studies of the planet formation suggest that more planets around the star b may exist , particularly inside the _ habitable zone _ ( hz ) @xcite .
although these planets are yet to be discovered in binary systems , recent works have shown that terrestrial planets can form and have a stable orbit around a star of a binary in a close s - type system , depending on the binary orbital and physical parameters @xcite .
the high mass of the disturbing stellar companion and the frequent high eccentricity / inclination of its relative orbit make the perturbation theories based on classical lagrangian expansions inapplicable in this case .
a possible approximation is the expansion of the disturbing function using the legendre polynomials .
however , although this development avoids the constraints due to the high eccentricities and inclinations , its convergence is very slow , allowing its application only in the case of hierarchical systems @xcite .
an alternative to overcome the problem is the extension of a perturbation theory up to second order in the disturber mass .
this was done in , where the developed model was applied to the @xmath2 cephei binary system .
although the results obtained have shown an improvement over the first - order model , the authors have emphasized that the model was not workable due to its high complexity ; thus they preferred the use of empiric correction terms , specific for the @xmath2 cephei configuration . in this work ,
we perform a study of the stability of the planetary motion in binary systems combining the analytical and numerical approaches . for this
, the stability concept is separated into the short - term stability and the long - term stability .
the short - term stability is based on the hill criterion , which does not require any kind of expansion and averaging of the disturbing function .
in such a way , it does not introduce the constrains on neither the mass of the disturbing body nor the eccentricity / inclination of its orbit .
the hill criterion is a powerful approach , which has been used for over a century @xcite ; an in - depth review of its applications can be found in @xcite .
analytical approaches based on the hill stability concept have been developed in @xcite , @xcite , among others , also as some empirical relations between stable orbits and the physical parameters of the binary . in our paper
, we develop a variation of the hill approach , referring to it as maximum apocentric distance ( or macd criterion ) , and apply it to the @xmath0 centauri stars a and b. to simplify the applications of the method , we present the explicit formula for the maximal apocentric distances as a function of the physical and orbital parameters of the stellar binary . the study of the long - term stability of a binary - planet system ( of the order of the lifetime of the stars ) is done for the s - type ( satellite type ) planetary motion , when the planet revolves around one of the stellar components @xcite .
the planet motion is modeled semi - analytically in the frame of the secular three - body problem @xcite and applied to the @xmath0 centauri hypothetical earth - like planets placed in the hzs of the stars a and b. the topology of the phase space is investigated , without any restriction on the magnitude of the eccentricities and inclinations of the planets .
the boundaries of the secular stability of the system , as a function of the masses and the orbital parameters of the planets , are then obtained . notwithstanding
the averaging being done at first order in masses , the relatively large separation between the stars allows a good agreement of the analytical results with those obtained numerically .
the direct numerical investigations are done through the dynamical mapping of the regions of habitability of both stars in the @xmath0 centauri system .
the physical and orbital parameters of the @xmath0 centaury binary used in this paper were taken from .
they are the masses of the components a and b , @xmath3 and @xmath4 , respectively , and the eccentricity @xmath5 and the semimajor axis @xmath6 @xmath7 ( or @xmath8au , if the parallax of @xmath9 @xmath10 is adopted ) of the relative orbit .
this paper is structured as in the following . in the next section , we study the short - term stability of the system , introducing a stability criterion and applying it to both the planar and 3d cases for planets on s - type orbits . in section [ sec2 ]
we study the long - term stability , applying the semi - analytical 3d model and comparing the results to those obtained through dynamical mapping . in section [ sec3 ]
, we investigate whether the planets remain inside the hz around the central star , regardless of large amplitudes of short - term oscillations provoked by the stellar companion .
finally , the summary is given in section [ sum ] .
in this section , we introduce the stability criterion for s - type planetary orbits in binary star systems based on the concept of the hill stability ( * ? ? ? * ; * ? ? ? * among others ) and apply it to both the planar and 3d motions of the fictitious planets in the @xmath0 centaury system .
our approach is based on the classical solution of the restricted three - body problem and , in particular , on the zero velocity surface method .
zero velocity surfaces are levels of the @xmath11function defined by the position of the planet with respect to the two stars @xcite .
the position vectors are defined with respect to the center of mass of the system in the non - inertial rotating reference frame , in which the two stars always lie along the @xmath12axis .
let us consider a system composed by two stars , 1 and 2 , and a planet @xmath13 .
the @xmath11 function , also known as jacobi integral or relative energy , can be written as @xmath14 , \label{super1}\ ] ] where @xmath15 and @xmath16 ( @xmath17 ) are the stellar masses and the position vectors and @xmath18 is the position vector of the planet ; @xmath19 is the total mass of the system and @xmath20 .
figure [ zerovelab ] shows the positions of the stars @xmath21 and @xmath22 separated by the distance @xmath23 . in the rotating frame ,
the function @xmath24 allows five equilibrium solutions ( three co - linear and two triangular lagrangian points ) , all located in the plane of the motion of the stars , which will be chosen as a reference plane . for our purpose , the co - linear solutions , and , particularly , @xmath25 solution , are of special interest .
these solutions are located on the @xmath12axis and their co - ordinates can be obtained searching for singular points on the surface ( [ super1 ] ) , where the vectors @xmath26 are just replaced by scalar @xmath12component of the corresponding body .
it should be noted that the expression ( [ super1 ] ) was developed in the uniformly rotating reference frame , when two stars move on circular orbits around the baricenter of the system and their mutual distance @xmath23 is unchanged . to apply it to the case when the stars have elliptic orbits ,
it is reasonable to adopt a minimal value for @xmath23 , when the stars are at the pericenter of their relative orbit . in this configuration ,
the perturbations on the s - like planetary orbit are maximal and possible instabilities may appear .
figure [ zerovelab ] shows the projection of one zero - velocity surface on the reference plane ( red curve ) ; the surface was calculated using the orbital elements of the @xmath0 centauri binary and the minimal distance between the stars .
this level holds the lagrangian solution @xmath25 , which appears as a singular saddle - like point on the @xmath12axis .
the same level intersects the @xmath12axis at two other points ; the positions of these points , @xmath27 , can be calculated from the condition solution ( red curve ) , obtained for the stars at the pericenter of their relative orbit .
the circumferences of stability , of radii @xmath28 around @xmath21 , and @xmath29 around @xmath22 , are shown by blue and green curves , respectively . ]
@xmath30 we assume that these intersections define a sphere of stability around each star . by definition , the sphere of stability is centered on the corresponding star and its radius is given by @xmath31 , where @xmath27 is the abscissa of the closest to the star intersection , while the abscissa of the star is @xmath32 .
the projections of the spheres of stability on the reference plane are shown in figure [ zerovelab ] , as a blue circumference centered on the star 1 and a green circumference centered on the star 2 .
it is worth noting that equation ( [ super3 ] ) , which is just a polynomial function of @xmath27 , can be easily solved numerically . for the sake of simplicity , we present here approximate expressions that allow us to obtain the radii .
the expressions are obtained applying the method of least squares . without loss of generality , we assume that the star 1 is a central star , while the star 2 is a perturber . in the case when the stellar mass ratio @xmath33 belongs to the interval @xmath34 , the radius of the sphere of stability is given as @xmath35 where @xmath36 and @xmath37 are the semimajor axis and the eccentricity of the stellar relative orbit , respectively . for the stellar mass ratio @xmath33 from the interval @xmath38 , the radius of the sphere is given as @xmath39
both approximations have errors smaller than @xmath40 of the exact solutions , for @xmath41 .
the error drops to less than @xmath42 , considering the interval of eccentricities @xmath43 .
now we introduce our criterion of stability .
it states that the stable motion of the planet orbiting one of the stars must satisfy , at any instant , the following condition @xmath44 where @xmath45 and @xmath46 are the planetary osculating semimajor axis and eccentricity , with respect to the corresponding central star , and @xmath47 is the radius of the sphere of stability of this star .
the interpretation of the expression ( [ spseq1 ] ) is simple : since the orbital evolution of the planet is confined to the inside of the circumference of radius @xmath47 , even at the most distant position from the central star ( at apocenter ) , the stability of this motion is guaranteed by the deficit of the relative energy ( [ super1 ] ) needed to cross the zero - velocity level holding the @xmath25 solution .
it is worth noting that , due to the restricted three - body approximation , the size of the domains of stability is defined by solely the physical and orbital parameters of the stars .
hereafter we will refer to our criterion as _ maximum apocentric distance _ or macd criterion .
it assures that planetary orbits that exceed the critical distance with respect to the central star ( equivalent to the radius of the sphere of stability of this star ) will be unstable .
however , it is worth stressing that this condition does not guarantee the long - term ( of order of the lifetime of the stars ) stability of the planet motion .
the secular dynamics of the system and its stability will be discussed in section [ sec2 ] .
centauri a , in logarithmic scale .
the top axis is the planetary semimajor axis in unit of the semimajor axis of the binary relative orbit .
the frequency unit is @xmath48 . ] in this section , we apply the macd
criterion to analyze the stability of planetary s - type orbits in the @xmath0 centauri ab binary .
the first step is to determine the radii of the spheres of stability of the stars a and b. in the case when the star a is a central star and the star b is a perturber , the stellar mass ratio is @xmath49 and the expression ( [ approximation1 ] ) must be used . in the case when the central star is b and the perturbing star is a , @xmath50 and the expression ( [ approximation2 ] ) is applied . as a result , for given values of the semimajor axis and eccentricity of the relative stellar orbit , @xmath51au and @xmath52
, we obtain the critical distances for the stars a and b as @xmath53au and @xmath54au , respectively .
the spheres of stability of the stars a and b are shown in figure [ zerovelab ] , where @xmath55 and @xmath56 . in the following step ,
we numerically integrate the system , in order to determine the stability of the planetary orbit defined by the condition ( [ spseq1 ] ) . in this stage ,
the choice of the suitable time - length is important : it should be neither too large to raise the computational costs , nor too small to lose information on the proper oscillations in the planetary motion .
these oscillations are usually separated in the short - period terms , associated with the keplerian motions of the planet and the perturbing star , and the long - period secular terms , associated with the precession of the planetary orbit .
eventually , depending on the ratio of the mean motions of the planet and the perturbing star , there are also mean - motion resonance terms whose periods are intermediate lying between the short and the long periods ; the resonant conditions will not be studied in this paper . although the macd - criterion is applied to assess the short - term stability of the planet motion , the long - period secular terms are still important .
this is due to the fact that , during the precession , the pericenter line of the planetary orbit becomes , at some instant , aligned with the pericenter line of the orbit of the binary .
if , at this instant , the perturbing star is at the pericenter of its relative orbit , the short - period perturbations become more intense and instabilities may appear .
thus , the choice of the integration timespan must account for the proper secular period of the planetary motion . to determine the secular period
, we apply the _ dynamical power spectrum _
method @xcite .
this method is based on the fourier decomposition of oscillations of the orbital elements and provides an information on the proper frequencies of the system under study and their evolution with changing parameters . in our study , we have chosen the semimajor axis of the planet as a parameter , which varied in the range between 0.5au and 3.5au .
the proper frequencies of the s - type planet orbiting @xmath0 centauri a obtained for each value of planetary semimajor axis over @xmath57 years , are presented in figure [ dinamespec ] .
we can clearly observe the presence of the three distinct frequency bands associated to three degrees - of - freedom of the problem . the first band , located in the high - frequency interval @xmath58 , is composed of the orbital frequency of the planet , @xmath59 , and its linear combinations with other proper frequencies .
according to the third kepler s law , this frequency decreases with decreasing @xmath45 .
the second band is confined to the very close neighborhood of @xmath60 and corresponds to the orbital frequency of the binary relative orbit , @xmath61 .
the third band is located in the low - frequency interval @xmath62 and corresponds to the secular frequency , @xmath63 .
this frequency tends to increase with increasing @xmath45 , due to increasing perturbations on the planetary orbit from the stellar companion .
the overlap of the bands occurs at @xmath64au , introducing the long - term instabilities in the planetary motion .
the dense vertical lines visible already at @xmath65au in figure [ dinamespec ] , are associated with the mean - motion resonances in this region .
finally , for @xmath66au , mostly all orbits are disrupted during @xmath57 years . from figure [ dinamespec ]
, we obtain that the secular period of the planet in the @xmath0 centauri system varies approximately from @xmath67years up to @xmath68years .
assuming the integration timespan within this range , we can now numerically integrate the planetary path , in order to verify the stability of the system defined by the condition ( [ spseq1 ] ) .
the quality of the prediction obtained with the macd - criterion can be verified by constructing dynamical maps around each star of the @xmath0 centauri binary . for this
, we use a grid of @xmath69 initial conditions on the ( @xmath70 ) representative plane and a terrestrial planet of mass @xmath71 . without loss of generality ,
we fix the initial values of the mean longitudes at @xmath72 and of the longitudes of pericenter at @xmath73 and @xmath74 .
each initial condition is numerically integrated using two different timespans : one is equal to @xmath67 years ( @xmath75 orbital periods of the binary ) and the second is equal to @xmath68 years ( @xmath76 orbital periods of the binary ) , which cover , respectively , @xmath77 and @xmath78 secular periods at @xmath79au . the solutions obtained
are then analyzed applying the macd - criterion , in order to calculate the maximal apocentric distances that the planet reaches over the assumed timespan .
the values obtained are presented using the grey scale code in figure [ zerovel - all ] , where the domains of the strongly unstable motion characterized by collisions and escape of the planet are dashed . on the planes we also plot , by the red curves , the levels whose values correspond to the radii of the spheres of stability , obtained using equations [ approximation1 ] and [ approximation2 ] as @xmath80au , for the star a , and @xmath81au , for the star b. as expected , the apocentric distances increase with the increasing planetary semimajor axis and decay with the increasing eccentricity . in the domains of the regular motion , the levels are smooth functions of the orbital elements of the planet .
this behaviour is changed when the initial conditions approach the domains of the unstable motion , whose boundaries are defined , according to our macd - criterion , by the critical distances shown by the red curves in figure [ zerovel - all ] . in this case , the critical levels show erratic behaviour , characteristic of chaotic orbits .
it is worth noting the good agreement between the positions of the critical levels ( red curves ) and the boundaries of the domains of instabilities provided by purely numerical simulations .
this confirms the quality of our criterion predictions .
moreover , the comparison of the topology of the dynamical maps obtained over 1 secular period of the planets ( top row in figure [ zerovel - all ] ) and those obtained over 10 secular periods ( bottom row ) , shows no significant differences .
this means that the time - length of 1 secular period is a good choice for the application of the macd - criterion .
a similar criterion was proposed in @xcite : it states that , for planetary motion around one of the stars to be stable , the initial value of the semimajor axis of the planet must obey a relationship @xmath82 , where @xmath83 is a function of the physical and orbital characteristics of the stars . by means of a series of numerical integrations ,
the authors have found an empirical expression for @xmath83 as @xmath84 where @xmath85 and @xmath36 and @xmath37 are the semimajor axis and the eccentricity of the stellar relative orbit , respectively .
the approximation is valid in the range @xmath86 , for @xmath87 , with the error smaller than 11% ( as claimed by the authors ) . in their approach
, the authors assumed that the planet orbits were initially circular , which means that the results obtained using equation ( [ holwie ] ) are valid only along the @xmath88axis in figure [ zerovel - all ] .
for purpose of comparison with our criterion predictions , we re - write the holman & wiegert ( 1999 ) criterion as @xmath89 and apply it to the @xmath0 centauri binary .
the result is shown by the blue curves in figure [ zerovel - all ] .
the correlation with the macd
criterion is notable , but it is clear that the @xcite criterion overestimates the instabilities in the planet motion , in this way , excluding from the study a significative portion of still stable configurations .
it is worth noting that the holman & wiegert ( 1999 ) criterion ( [ holwie ] ) , also as other empirical criteria ( e.g. , , @xcite , @xcite , among others ) , introduces a concept of _ grey area _ in the parameter space of the problem , in which the state of a system changes from stability to instability ( * ? ? ?
* chapter 11 ) .
the grey area is defined as the confidence interval which measures the reliability of the estimation obtained from the fitting procedure . applying the formula ( [ holwie ] ) to the @xmath0 centauri system and using the traditional law of error propagation @xcite
, we obtain the confidence intervals for the stars a and b as @xmath90au and @xmath91au , respectively .
the borders of the corresponding grey areas on the ( @xmath46,@xmath45)plans are shown by the cyan - colored curves in figure [ zerovel - all ] . however , it must be emphasized here that the grey area idea does not apply to the macd criterion ( [ spseq1 ] ) , which is founded on the basic concept of the hill stability and can not be interpreted in statistical terms .
due to short time intervals of numerical integrations , the method can still originate effects analogous to ` grey areas ' ( seen as red - colored zones in figure [ zerovel - all ] ) , but these zones rapidly decrease with increasing integration timespans . also , it is worth noting here that the purposes of both methods are different .
the purpose of the @xcite criterion is a prompt estimation of the stability of one initially circular configuration of a putative planet in the binary , given the distance of the planet to the central star . notwithstanding the facility in implementation ( it does not require numerical integrations ) , this method still suffers on some limitations .
one of these is due to assumption that the planet starts on zero - eccentricity orbit . due to very strong gravitational perturbations produced by the stellar companion on the planetary motion
, this assumption can scarcely be correct .
another limitation is related to the detection of the planets in binary systems and consequent determination of their orbits through the best - fitting procedures .
it seems unlikely that the sets of the osculating orbital elements of the planets provided will be compatible with the circular orbits needed in applying the @xcite criterion .
our macd criterion can overcome this limitation , since it is based on the implementation of numerical integrations which use the sets of the orbital parameters of the planets as input .
the macd criterion can be easily extended to the inclined to the reference plane orbits . indeed ,
since the maximum distance of the planet from the central star in space is still its apocentric distance , we can apply the same condition of stability @xmath92 , where @xmath47 is a radius of the sphere of stability given by equations ( [ approximation1 ] ) , for the star a , and ( [ approximation2 ] ) , for the star b. to test the stability of the inclined planetary orbits , we construct the dynamical maps , each with a @xmath93 grid of initial conditions , on the ( @xmath45 , @xmath94 ) representative planes , with the planetary eccentricity fixed at @xmath95 , again with the planetary mass fixed at @xmath71 .
our previous study @xcite pointed out that the initial angular conditions of the secular angle @xmath96 and the argument of pericenter @xmath97 can be fixed at either @xmath98 or @xmath74 , without loss of generality .
the maps obtained are shown in figure [ zerovel-3d ] .
we can see that the boundaries of the stable motion are well defined by the macd
criterion ( red levels ) , when the apocentric distances approach the critical values @xmath99au and @xmath100au . the maximum values of the planetary semimajor axis , for which the motions still remain stable , are @xmath101au and @xmath102au , for the stars a and b , respectively . for smaller values of the semimajor axis , there are zones of regular motion for any values of the planetary inclination .
the stable motion is robust at inclinations below @xmath103 . above this inclination
, we can detect sporadic chaotic solutions ( red dots ) deeply inside the region of the stable motion .
this is an indicator that the analysis of the short - term stability is not sufficient in this case and must be extended to a longer length - times .
this will be done in the next section , where we will show that the instabilities at @xmath104 are originated by the lidov kozai secular resonance .
the method described in this section provides a global view of the short - term stability of the system .
it is conceptually simple , fast and easily implementable .
it allows us to detect promptly the domains of regular motion , avoiding , in this way , long - term integrations and analysis of unstable orbits .
however , due to the analysis of orbital behaviour of the planets over short timespans , this approach provides no information on the long - term ( of order of the lifetime of the stars ) stability of the system .
the specific study on the stability of the planetary motion in the binary over the age of the system will be done in the next section .
under a long - term " stability we mean the stability of the system over the lifetime of the central star .
one way to assess the long - term stability of the planet motion is direct n - body numerical simulations , which , depending on the stellar types and ages , must be done over several billions of years .
this approach suffers from several limitations and the most serious from these is extremely high computing costs .
an alternative is the dynamical mapping of the domain under study .
this approach is also founded on purely numerical integrations , but the time intervals of the simulations are much shorter in this case , and , therefore , we are allowed to explore a large set of initial conditions . analyzing the topological structure of dynamical maps , we can detect the mechanisms which could induce long - term instabilities in the planetary motion , and determine domains of their action .
the topology analysis is based on the analytical modeling of the general three - body problem briefly described in the following ( for more details , see @xcite ) . in the jacobian reference , the hamiltonian which describes the motion of the planet ( @xmath105 ) around one of the stars ( @xmath106 ) , with the second star ( @xmath107 ) as the disturber ,
is written as @xcite @xmath108 where : @xmath109 @xmath110 and @xmath111 is the osculating semimajor axis of @xmath112-th orbit , @xmath113 is the position vector of @xmath107 relative to the center of mass of @xmath106 and @xmath105 and @xmath114 is the distance between @xmath15 and @xmath107 ; @xmath115 is the gravitational constant .
hereafter , the indices @xmath116 stand for the central star , planet and perturbing star orbits , respectively .
associated with the keplerian part of the hamiltonian , a set of mass - weighted delaunay s elliptic variables is introduced as @xmath117 where @xmath118 and @xmath119 are the eccentricities and inclinations , respectively ; @xmath120 , @xmath121 , @xmath122 and @xmath123 .
we assume that the planet and the perturbing star are not involved in any mean - motion resonance of low order .
thus we can perform numerically the averaging of the hamiltonian ( [ eq : eq0 ] ) with respect to the mean anomalies of the planet and the disturber @xcite .
the secular hamiltonian is then defined by @xmath124 where the keplerian part is constant and therefore needs not be considered . in this section
, we construct the dynamical maps of coplanar motions , using a grid of @xmath125 osculating initial conditions on the ( @xmath126)-plane and fixing the initial angular variables at @xmath127 and @xmath128 .
each initial condition was integrated over @xmath68 years ( @xmath129 orbital periods of the binary or @xmath130 secular periods , see figure [ dinamespec ] ) , with the planetary mass fixed at @xmath131 .
the solution obtained was fourier analyzed , in order to calculate the spectral number corresponding to the used initial condition .
the spectral number defines the degree of chaos in the planetary motion : its small values correspond to regular motions , while its larger values correspond to unstable motions @xcite .
the results are shown in figure [ logn - planar ] .
the grey color scale is used to plot the spectral number : light regions correspond to small numbers and regular motions , while the increasingly dark tones indicate large numbers and unstable orbits .
the hatched patterns correspond to the domains where the macd criterion of stability is not satisfied over the chosen integration timespan ( see sect . [ sec1 - 1 ] ) .
the blue curves show the families of the stable stationary solutions of the secular problem given by the hamiltonian model ( [ eq : eq3 ] ) , frequently referred to as mode i of motion @xcite ; these solutions are also known as forced eccentricities of the secular problem . in the planar case ,
the secular problem is reduced to one degree of freedom and the secular solutions are obtained by plotting the energy and angular momentum levels on the representative plane ( for a detailed description , see @xcite ) . the secular planetary motion projected on the ( @xmath126)-plane is an oscillation around the corresponding equilibrium belonging to the blue curve .
the closer the initial condition is to the mode i solution , the smaller is the amplitude of the oscillation of its eccentricity .
this fact explains the very stable , nearly harmonic , dynamics of the planet in the vicinity of the blue curve ( light grey region ) , even at larger values of @xmath45 .
however , it is worth stressing that the initial conditions used in construction of the dynamical maps are osculating orbital elements ; thus , the comparison of the results of numerical integrations to those provided by the secular model ( [ eq : eq3 ] ) should be done carefully .
moreover , the mass of the disturbing body is comparable to the mass of the central star ; this fact reduces the domain of the application of the first - order in masses secular model ( [ eq : eq3 ] ) .
we plot in figure [ logn - planar ] , by the red curves , the families of the secular stationary solutions obtained through the purely numerical filtering procedure , as described in .
the deviation of the analytically obtained family from that obtained numerically ( with no restriction on the perturbing mass ) clearly occurs at larger semimajor axes , when the perturbations due to the stellar companion become stronger . since , in this region , the planetary motion is stable only in the very narrow vicinity of the mode i , the correct determination of the family is essential in this case .
the two families match each other at small semimajor axes of the planet , up to @xmath132au ( or @xmath133 ) , for small eccentricities .
the dynamical maps in figure [ logn - planar ] show that the planetary dynamics in this case is very regular ( white color ) , even at very - high eccentricities .
it is worth noting that the hzs around the stars are located close to these regions ( see section [ sec3 ] ) . when the semimajor axis of the planetary orbit increases , the perturbations due to the stellar companion become stronger and the planetary motion increasingly non - harmonic ( except in the vicinity of the mode i of motion ) .
the dark tones on the dynamical maps in figure [ logn - planar ] confirm this fact .
several high order mean - motion resonances , from 13/1 to up 20/1 , are dense in these regions ; they appear as thin horizontal strips on the dynamical maps
. the phenomenon can be also observed in the power dynamical spectrum in figure [ dinamespec ] , where the resonant initial conditions appear as vertically scattered dots , at @xmath134au .
since these resonances can destabilize the planetary motion , the long - living planets are not expected to be found in this region .
finally , at relative distances above 0.14 ( in units of the semimajor of the binary orbit ) , the planetary motion is strongly chaotic and the planets are ejected from the binary ( hatched regions ) .
the boundary of these domains match the boundaries of short - term instabilities showed in figure [ zerovel - all ] .
it is worth to emphasize the features of the planet motion near the mode i stationary solutions .
being robust and stable , even at large distances from the central star , this is a favorite location of several known planetary extra - solar systems ( e.g. , @xmath135 andromedae c - d planet pair ) .
@xcite shows that , during migration , the planets are guided toward the stationary configurations , independently on the specific migration mechanism .
moreover , shows that the planet formation is preferential on the orbital configurations corresponding to the mode i stationary solutions .
we construct the dynamical maps on the plane ( @xmath46,@xmath94 ) of the initial osculating values of the planetary eccentricity and the inclination , with a terrestrial planet of mass @xmath136 .
the spectral analysis method was applied on the @xmath137grid , with the spacings @xmath138 and @xmath139 .
the initial semimajor axis of the planet was fixed at @xmath140au ; for this value , the planet is distant from the perturbing star and the first - order hamiltonian model ( [ eq : eq3 ] ) can be used to investigate the topology of the dynamical maps .
the maps are shown in figure [ fig : fig4 ] .
the positive and negative values on the @xmath141axis stand for the initial values of the angle @xmath142 of @xmath98 and @xmath74 , respectively ( see @xcite , for more details ) .
the power dynamical spectrum in figure [ dinamespec ] allows us to estimate the secular period of the planet at @xmath143au as @xmath144 years .
this means that an appropriate timespan for simulations of the secular dynamics must be of order of at least @xmath57 years .
the chosen timespan of integration for our dynamical map was 500,000 years ( @xmath145 orbital periods of the binary ) , for each initial condition on the grid ; the orbits disrupted during this time are shown by the hatched pattern in figure [ fig : fig4 ] .
the surviving orbits are shown by the grey scale code , from light tones , for nearly periodic orbits , to darker tones , for increasingly unstable orbits .
the two color curves represent the periodic solutions obtained through the secular model ( [ eq : eq3 ] ) .
the blue curves are location the of the stable periodic solutions , while the red curves are locations of the unstable solution .
the long - living motions are expected to be found near the blue curves and far from the red curves .
indeed , the analysis of the structure of the dynamical maps of the representative planes reveals that , for low inclinations @xmath146 , the maps are dominated by a light grey background , which means that the low - inclination secular dynamics is regular . in contrast , the vicinity of the red curves is dominated by the darker tones , indicating unstable motions .
the instability is related to the existence of separatrices between different regimes of motion .
their domains , at high inclinations , show complex dynamical behaviour , with the presence of several regimes of resonant motion ( the detailed description of these regimes of secular motion can be found in @xcite ) .
the dominating behaviour is the @xmath46@xmath94 coupling , or lidov kozai resonance , which destabilizes the nearly circular orbits at @xmath147 .
for higher values of eccentricity , the planetary dynamics is characterized by the coupled large variation of the eccentricity and inclination and the libration of the angle @xmath148 around @xmath149 .
the domains of the lidov - kozai resonance are located on the lower half - planes in figure [ fig : fig4 ] ; the behaviour of the system is very regular in these regions .
the @xmath150 secular resonance also exists in the high - inclination regions ; its domains are thin white strips along the red curves . in this case , the secular angle @xmath150 librates either around @xmath151 or @xmath74 .
it is worth emphasizing that we have detected a large portion of the chaotic orbits ( black tones on the graphs in figure [ fig : fig4 ] ) , but only a small quantity of these orbits are actually ejected from the system ( dashed regions ) .
this is because the chaotic processes are slow in the secular system and the timespan of 500,000 years is not sufficient for detecting the planetary escapes .
however , the escapes are imminent and , as a consequence , it will be impossible to detect the planets inside these regions of high instabilities .
regardless of the robust long - term stability of the planetary motion in the large domains around the stars of @xmath0 centauri , the planetary evolution inside the hzs , should be analyzed , in order to verify whether the planet motion is confined inside the hz .
the concept of the hz itself , even when introduced for single stars ( e.g. , @xcite ) , is very complex . in binary stars ,
the definition of the hz is further complicated due to several additional aspects .
one is the combined radiation of both stellar components on the terrestrial planet studied recently in several papers ( e.g. , @xcite , @xcite ) .
another aspect concerns the large oscillations of the orbital elements of the planet , that could affect the conditions of habitability of the planet or even drive the planet far from the hz .
this behaviour was studied in @xcite , where the authors considered the averaged effects produced by the secondary star on the variation of the eccentricity of the planet , yet in the case of coplanar and initially circular planetary orbits . in this work
we explore the conditions for which the planets remain inside the hz , in the case of eccentric and inclined orbits . for this , we use the different definitions of the hzs around the @xmath0 centauri stars given in the papers @xcite , @xcite and @xcite(see table [ tablehz ] ) .
it should be emphasized that these definitions are not equally weighted .
in fact , the kh13 definition can be considered more realistic because it takes into account the presence of an atmosphere on a terrestrial or super - earth planet . however , the detailed discussion on the hz calculations is out of the scope of this paper . for each definition from table [ tablehz ] ,
we calculate the maximum and minimum distances of the planet from the central star , for different initial eccentricities and inclinations of the planet orbit and the initial semimajor axis fixed at @xmath152au , around @xmath0 centauri a , and @xmath153au , around @xmath0 centauri b ; these values roughly correspond to the averaged positions of all hzs defined in table [ tablehz ] . .
the inner and outer boundaries of the hzs ( in au ) and the hzs width in the @xmath0 centauri ab binary , calculated in forgan ( 2012 ) , eggl et al.(2013 ) and kaltenegger & haghighipour ( 2013 ) ( kh13 ) ( conservative and optimistic definitions ) . [ cols="<,^,^,^,^",options="header " , ] [ tablehz ] the numerical test is done using numerical integrations of planetary orbits over 10,000 years .
the values of the maximal and minimal distances obtained are compared to the width of the hzs , in order to infer whether the planet remain within the hz boundaries over 10,000 years ( @xmath76 orbital periods of the binary ) .
the difference between the maximal and minimal distances of the planet from the central star is plotted on the ( @xmath46 , @xmath94)planes in figure [ graf2-mapsp ] , using the grey scale code : the difference is smaller in the light regions and larger in the dark regions .
the regions in white tone correspond to the minimal possible deviations of the planet from the center of the hz .
the simple comparison with the graphs in figure [ fig : fig4 ] allows us conclude that these regions are associated with the mode i secular solutions whose location is given by blue curves in figure [ fig : fig4 ] .
an interesting feature of this analysis is that initially circular orbits exhibit significant deviations from the center of the hz , that , for high inclinations , could result in the planets spending some time outside the hzs . as discussed in @xcite ,
the planet remaining inside the hz during its evolution is not the only requirement for the planet to host life , and it is neither an excluding factor . even if the planet spends some time of its orbit outside the hz , as long as the mean radiation received is still within the hz limits , the planet could be considered habitable , although in a more restrict way .
nevertheless , it is expected that orbits with smaller oscillations are able more likely to sustain some kind of life .
it is also interesting to compare the orbital evolution of a terrestrial s - type planet in @xmath0 centauri ab with the evolution of the terrestrial planet in a planetary system , with a jupiter - sized planet .
the comparison is done using the configurations defined by the orbital parameters of @xmath0 centauri ab , except , in the case of the planetary system , the binary companion is replaced by a jupiter - mass planet . the results are shown in figure [ orbits ] , where the variation of the orbital distance of the planet revolving around @xmath0 centauri b is plotted by the red curve , in the case when the perturber is the star @xmath0 centauri a ( @xmath154 ) , and by the black curve , when the perturber is a hypothetical jupiter - mass planet ( @xmath155 ) . in both simulations ,
the same orbital and physical parameters of the planet and perturbing body were used , except the mass of the perturber .
the initial semimajor axis and the eccentricity of the planet were fixed at @xmath153au ( inside the hz of the star ) and @xmath156 , respectively ; note that this value of the eccentricity corresponds to the actual eccentricity of earth .
the initial orbital elements of the perturber were the same of the relative orbit of the @xmath0 centauri binary .
centauri b when the perturber is @xmath0 centauri a ( red ) and a jupiter - mass planet ( blue ) .
] we can observe in figure [ orbits ] that the amplitudes of the oscillation of the orbital distance of the planet are of the same order of magnitude , despite that the perturbing masses differ in order 3 .
this result is consistent with the theoretical predictions of the restricted three - body models ( e.g. , ) which state that the forced eccentricity of the test particle is independent on the masses of the large bodies ( at least , in first order in masses ) .
therefore , the orbital distance oscillation of the planet in the binary system is similar to that of a terrestrial planet perturbed by a giant planet .
the main difference is the timescale of the oscillation : it is of order of @xmath157 years , in the binary system , and of order of @xmath158 years , in the planetary system .
however , this difference seems to be not important in the context of the continuously habitable zone .
in addition , we have obtained that , in the case of the real earth - jupiter system , the forced eccentricity of the earth s orbit is equal to @xmath159 ; in the case of a putative terrestrial planet orbiting @xmath0 centauri b , the forced eccentricity of the planetary orbit is equal to @xmath160 . calculating the variation of the orbital distances in both cases , we have found that its amplitude is @xmath161au , for the earth s orbit , and @xmath162au , for the planet located near the mode i stationary solution of the hz of @xmath0 centauri b , that is , both variations are of same order of amplitude .
this result brings positive perspectives of finding habitable planets in binary star systems .
in this work , we present a methodology to analyze the planetary dynamics of s - type orbits in binary star systems . we develop a short - term stability criterion ( macd criterion ) , which is based on the hill concept of stability . since this method does not require any kind of expansion and averaging of the disturbing function , there are no constrains on neither the mass of the disturbing body nor the eccentricity / inclination of its orbit .
we apply our approach to the @xmath0 centauri binary system with a hypothetical earth - like planet , orbiting one of the stars ( s - type orbit ) , and show that the purposed criterion provides a global view of the stable and unstable domains in the space of the orbital elements .
it is worth emphasizing that our approach is simple , fast and easily implementable .
its main purpose is to detect promptly the domains of regular motion , avoiding , in this way , long - term integrations and analysis of unstable orbits .
it also allows us to quickly assess the stability of the best - fit solutions in the determination of the planetary orbits hosted by binary stars .
the long - term stability of the planetary orbits in the hzs of the @xmath0 centauri binary system is studied using the secular hamiltonian approach .
we show that the behaviour of the planets is well described by the first - order in masses model .
the secular planetary motion is regular , even at high eccentricity domains of the hzs .
there are no significant mean - motion resonances that could destabilize the secular dynamics .
the stability of the planets is also detected through the 3d analysis , at low and moderate inclinations of the planetary orbits ( @xmath163 ) . for higher inclinations ,
the dominating regime of motion is the lidov kozai resonance , which is separated from the purely secular regime of motion through the layers of strongly chaotic motion . even though there are domains of very stable motion in the lidov
kozai resonance , the large excursions of the planetary eccentricities and inclinations seem to be unfavorable for the evolution of life inside the hzs . beyond the hz , where high - order mean - motion resonances take place , the first - order model fails .
thus , a second - order in masses model , as well as the modeling of the resonant behaviour , would be required to further study of this region . analyzing the dynamics of the planets evolving inside the hzs of the @xmath0 centaury stars
, we obtain that the planetary orbits will remain within the boundaries of the hzs , if their eccentricities and inclinations are constrained to values close to the stationary solutions of the secular problem , the mode i solutions . in this favorable scenario ,
the amplitude of the orbital distance variations are comparable to the amplitudes of the earth s orbit , with the only difference being the timescales of the variations .
the real effects of this on the planetary climate and geological conditions are still poorly known and it is expected that further interdisciplinary studies could bring to light considerable new information on this issue .
this work was supported by the so paulo state science foundation , fapesp ( grant 2010/01209 - 2 ) , and the brazilian national research council , cnpq .
this work has made use of the facilities of the computation center of the university of so paulo ( lcca - usp ) and of the laboratory of astroinformatics ( iag / usp , nat / unicsul ) , whose purchase was made possible by the brazilian agency fapesp ( grant 2009/54006 - 4 ) and the inct - a .
the authors are grateful to dr .
haghighipour for numerous helpful suggestions / corrections on this paper .
everhart , e. 1985 , dynamics of comets : their origin and evolution , proceedings of iau colloq .
83 , held in rome , italy , june 11 - 15 , 1984 . edited by andrea carusi and giovanni b. valsecchi .
dordrecht : reidel , astrophysics and space science library .
volume 115 , 1985 , p.185 , 185 | in this paper , we investigate whether hypothetical earth - like planets have high probability of remaining on stable orbits inside the habitable zones around the stars a and b of @xmath0 centauri , for lengths of time compatible with the evolution of life .
we introduce a stability criterion based on the solution of the restricted three - body problem and apply it to the @xmath0 centauri system . in this way
, we determine the regions of the short - term stability of the satellite - type ( s - type ) planetary orbits , in both planar and three - dimensional cases .
we also study the long - term stability of hypothetical planets through the dynamical mapping of the habitable zones of the stars .
the topology of the maps is analyzed using the semi - analytical secular hamiltonian model and possible processes responsible for long - lasting instabilities are identified .
we verify that the planetary motion inside the habitable zones is regular , regardless of high eccentricities , for inclinations smaller than @xmath1 .
we show that the variation of the orbital distance of the planet located in the habitable zones of the binary is comparable to that of earth , if the planet is close to the mode i stationary solution .
this result brings positive expectations for finding habitable planets in binary stars .
celestial mechanics ; stars : binaries : general ; planets and satellites : dynamical evolution and stability ; planets and satellites : individual : @xmath0 centauri ; |
the successful detection @xcite of bose - einstein condensates ( bec ) in dilute weakly - interacting bosonic atoms employing magnetic trap at ultra - low temperature stimulated intense theoretical studies on different aspects of the condensate @xcite .
the experimental magnetic trap could be either spherically symmetric or axially symmetric .
the properties of an ideal condensate at zero temperature are usually described by the time - dependent , nonlinear , mean - field gross - pitaevskii ( gp ) equation @xcite which incorporates appropriately the trap potential as well as the interaction between the atoms forming the condensate .
one of the problems of interest in bose - einstein condensation is the study of the oscillation in the condensate induced by a sudden change in the interaction acting on it .
there have been both experimental @xcite and theoretical @xcite studies of the motion of the condensate under a sudden or continuous variation of the trapping frequencies .
there has also been similar studies induced by a sudden variation of the atomic scattering length . in the latter class
there have been experimental studies of explosive emission of atoms from a condensate when the scattering length is suddenly turned negative ( attractive ) from positive ( repulsive ) , thus leading to collapse and explosion of the condensate @xcite .
this could also lead to the observation of a soliton train @xcite .
there have been theoretical attempts to explain the dynamics of the condensate for collapse and explosion @xcite as well as for soliton train @xcite using the gp equation .
resonance is an interesting feature of any oscillation under the action of an external periodic force manifesting in a very large amplitude .
usually , a resonance appears when the frequency of the external periodic force becomes equal to that of the natural oscillation of the system @xcite .
such a resonance appears in many areas of physics .
there have been studies of resonance in different problems of linear dynamics and the appearance of resonance is well understood in these problems both analytically and numerically .
the investigation of resonance in nonlinear dynamics is far more interesting and nontrivial .
consequently , resonances in nonlinear problems are not so well understood .
this makes the study of resonance in the dynamics generated by the nonlinear gp equation of general interest .
apart from mere theoretical interest , such resonance in the oscillation of the bec governed by the gp equation can also be studied experimentally so that one can compare theoretical prediction with experiment .
the possibility of generating a resonance in the oscillation of a bec subject to a varying trapping potential has been investigated theoretically @xcite .
the variation of trapping potential in the gp equation corresponds to a linear external perturbation .
here we investigate the resonance dynamics of the bec subject to a periodic variation of the scattering length , which should be considered as a nonlinear periodic force acting in the gp equation .
such a variation of the scattering length is possible near a feshbach resonance by varying an external magnetic field @xcite . for experimentalists
a periodic variation of the external magnetic field is easier to implement and the magnetic field is nonlinearly related to the scattering length .
however , a simple periodic variation of the scattering length is a cleaner theoretical possibility and we shall apply such a variation in this study .
the appearance of a resonance in the oscillation of a bec due to a periodic variation of the scattering length has been postulated recently in a spherically symmetric trap @xcite . here
we present a complete study of the resonance dynamics in this case .
in addition , we extend this investigation to the more realistic and complicated case of an axially symmetric trap as well as to vortex states in such a trap . in the simplest case of a damped classical oscillator under a periodic external force with the natural frequency of oscillation , the resonance appears due a constant phase difference between the oscillation and the external force @xcite .
this leads to a gradual increase in kinetic energy over each cycle of oscillation during an interval of time and hence in the amplitude of oscillation in this interval .
eventually , the rate of loss of energy due to damping is exactly compensated for by the increase in energy from the external force and the oscillator oscillates with a large constant amplitude at large times at resonance .
the constant phase difference between the external force and the oscillation is possible only at resonance when the two frequencies match and is necessary for resonance .
the situation is much more complicated in the case of a trapped bec subject to a nonlinear external force due to a periodic variation in the scattering length . when subject to a sudden perturbation , the natural frequency of oscillation of a trapped spherical bec governed by the gp equation with nearly zero nonlinearity is @xmath1 where @xmath2 is the frequency of the existing trap @xcite .
this result is exact in the linear case @xmath3 .
however , for the nonlinear problem @xmath4 , this natural frequency may change . in this paper we shall only be considering small nonlinearity where the natural frequencies can be taken approximately as @xmath1 . in physical analogy with the classical oscillator
, a resonance is expected in the bec oscillation when the scattering length oscillates with frequency @xmath1 or its multiples . in our study
we find that this is indeed the case .
however , a much richer dynamics emerges in this case compared to the case of linear systems .
we find that due to nonlinearity , the phase difference @xmath5 between the bec oscillation and the external nonlinear force varies with time .
consequently , during a certain interval of time the bec may steadily gain energy from the external force in all cycles and the amplitude of oscillation may grow leading to a resonance . in the case of bec
there is no damping and for a constant phase difference @xmath5 the amplitude is expected to grow beyond limit . however , as a result of the variation of the phase @xmath5 with time due the nonlinearity , after some time the system starts to lose energy due to the action of the external force .
eventually , the oscillation of the the system can virtually stop .
by that time the modified phase difference favors the increase of energy and the amplitude starts to grow again .
consequently , the resonance of the system has a peculiar nature ; the amplitude of oscillation passes through pronounced maxima and minima . in a single classical damped oscillator
this is not possible under the action of a periodic force @xcite .
however , this is possible in case of a coupled system with exchange of energy between the two oscillators .
we also investigate the resonant oscillation of an axially symmetric bec due to a periodic variation of the scattering length .
when the frequency of oscillation of the scattering length equals the natural frequency of the radial or axial oscillation of the bec , we find resonance in the radial or axial oscillation , respectively .
in section 2 we describe briefly the time - dependent gp equation with spherical and axial traps .
the essential details of the split - step crank - nicholson method used for numerical solution is also described in this section . in section 3
we report the numerical results of the present investigation of resonance for the spherically symmetric case . in section 4
we present the same for the axially symmetric case for condensates with zero angular momentum as well as for vortex states with nonzero angular momentum .
finally , in section 5 we present a discussion and summary of our study .
at zero temperature , the time - dependent bose - einstein condensate wave function @xmath6 at position @xmath7 and time @xmath8 may be described by the following mean - field nonlinear gp equation @xcite @xmath9\psi({\bf r};\tau)=0 .
\end{aligned}\ ] ] here @xmath10 is the mass and @xmath11 the number of atoms in the condensate , @xmath12 the strength of interatomic interaction , with @xmath0 the atomic scattering length .
the normalization condition of the wave function is @xmath13 in this case the trap potential is given by @xmath14 , where @xmath2 is the angular frequency and @xmath15 the radial distance .
the wave function can be written as @xmath16 .
after a transformation of variables to dimensionless quantities defined by @xmath17 , @xmath18 @xmath19 and @xmath20 , the gp equation in this case becomes @xmath21 \varphi ( x , t)=0 , \end{aligned}\ ] ] where @xmath22 .
the normalization condition for the wave function is @xmath23 the trap potential is given by @xmath24 where @xmath2 is the angular frequency in the radial direction @xmath25 and @xmath26 that in the axial direction @xmath27 .
we are using the cylindrical coordinate system @xmath28 with @xmath29 the azimuthal angle . in this case
one can have quantized vortex states with rotational motion around the @xmath27 axis @xcite .
in such a vortex the atoms flow with tangential velocity @xmath30 such that each atom has quantized angular momentum @xmath31 along @xmath27 axis . the wave function
can then be written as : @xmath32 with @xmath33 . using the above angular distribution of the wave function in ( [ a ] ) , in terms of dimensionless variables @xmath17 , @xmath34 , @xmath35 @xmath36 , and @xmath37 we get @xmath38\varphi ( { x , y};t ) = 0.\end{aligned}\ ] ] the normalization condition of the wave function is @xmath39 we solve the gp equation numerically employing a split - step time - iteration method using the crank - nicholson discretization scheme described elsewhere @xcite .
we discretize the gp equation spanning @xmath40 from 0 to 25 and @xmath41 from @xmath42 to 30 in the axially symmetric case . in the spherically symmetric case
the range of @xmath40 integration was taken from 0 to 30 .
this was enough for having a negligible value of the wave function at the boundaries . to calculate the solution for a specific nonlinearity @xmath43 the time iteration
is started with the known harmonic oscillator solution of the gp equation for nonlinearity @xmath3 .
these initial normalized solutions are @xmath44 @xmath45 ^{1/4}{x^{1+|l| } } e^{-(x^2+\lambda y^2)/4}\ ] ] for ( [ c ] ) and ( [ d ] ) , respectively .
the nonlinearity is then slowly changed in steps of 0.0001 during each time iteration until the desired value of @xmath43 is attained . as a result in the course of time iteration the initial @xmath3 solution slowly converges towards the final solution with the desired @xmath43 .
we study the formation of resonance in the oscillation of the condensate as the scattering length is varied periodically . in the scaled gp equations ( [ c ] ) and ( [ d ] ) the scattering length appears only through the nonlinearity parameter @xmath46 . in particular we consider the following variation of the nonlinearity parameter @xmath47 where @xmath48 and @xmath49 are two constants and @xmath50 is the frequency of variation of the nonlinear parameter . in the gp equation ( [ c ] ) ,
the scaled harmonic oscillator trap frequency is unity ; the radial and axial trap frequencies of ( [ d ] ) are 1 and @xmath51 , respectively . for zero and
small values of nonlinearity the natural frequency of oscillation of ( [ c ] ) is twice the scaled trap frequency , which is 2 .
the natural frequencies of oscillation of ( [ d ] ) in radial and transverse directions are twice the scaled trap frequencies , which are 2 and @xmath52 , respectively
. these natural frequencies of oscillation have been verified experimentally @xcite , as well as in theoretical calculations @xcite .
the perturbation ( [ e ] ) when applied to the gp equation behaves like a nonlinear external force acting in ( [ c ] ) and ( [ d ] ) .
when the frequency @xmath50 of the perturbation coincides with the natural frequencies of oscillation of the gp equation , resonant behaviors are expected .
the classic example of resonance appears when the parameter @xmath48 in ( [ e ] ) is zero . in that case
resonances are expected in the spherically symmetric case for @xmath53 . in the axially symmetric case
they are expected for @xmath53 and @xmath54 in the radial and axial oscillations , respectively . for @xmath55
, resonances also appear provided that the constant @xmath49 is not very small .
we study these possibilities in the next sections .
first we consider the case with @xmath56 in ( [ e ] ) . in this case
the maximum permitted value of @xmath49 is 0.575 . for @xmath57 , the minimum of nonlinearity
@xmath43 could be less than @xmath58 .
it is known that no stable solution of the spherically symmetric gp equation could be obtained for @xmath59 @xcite .
the domain of values @xmath59 corresponds to attractive interaction leading to collapse of the condensate . in this case
the natural frequency of oscillation of the condensate is 2 and resonances are expected for values of @xmath50 equaling multiples of 2 .
we shall mostly concentrate here to the case @xmath60 . in figures 1 ( a ) , ( b ) and
( c ) we plot the root mean square ( rms ) radii @xmath61 vs. time for @xmath62 and @xmath63 and for @xmath60 .
we find that as @xmath49 increases the rms radii can grow to a large value signaling a resonance .
one interesting feature of these resonances is that the oscillation of the rms radius can increase and subsequently decrease and such a cycle can repeat many times
. a larger number of such growth and decay cycles can be accommodated in a given interval of time as @xmath49 increases .
vs. time @xmath64 for @xmath56 and ( a ) @xmath65 , @xmath60 , ( b ) @xmath66 , @xmath60 , ( c ) @xmath67 , @xmath60 , ( d ) @xmath67 , @xmath68 , ( e ) @xmath67 , @xmath69 .
, title="fig : " ] vs. time @xmath64 for @xmath56 and ( a ) @xmath65 , @xmath60 , ( b ) @xmath66 , @xmath60 , ( c ) @xmath67 , @xmath60 , ( d ) @xmath67 , @xmath68 , ( e ) @xmath67 , @xmath69 .
, title="fig : " ] vs. time @xmath64 for @xmath56 and ( a ) @xmath65 , @xmath60 , ( b ) @xmath66 , @xmath60 , ( c ) @xmath67 , @xmath60 , ( d ) @xmath67 , @xmath68 , ( e ) @xmath67 , @xmath69 . ,
title="fig : " ] vs. time @xmath64 for @xmath56 and ( a ) @xmath65 , @xmath60 , ( b ) @xmath66 , @xmath60 , ( c ) @xmath67 , @xmath60 , ( d ) @xmath67 , @xmath68 , ( e ) @xmath67 , @xmath69 .
, title="fig : " ] vs. time @xmath64 for @xmath56 and ( a ) @xmath65 , @xmath60 , ( b ) @xmath66 , @xmath60 , ( c ) @xmath67 , @xmath60 , ( d ) @xmath67 , @xmath68 , ( e ) @xmath67 , @xmath69 .
, title="fig : " ] vs. time @xmath64 for @xmath70 and ( a ) @xmath71 ( b ) @xmath72 , ( c ) @xmath73 , ( d ) @xmath74 , and ( e ) @xmath75 , title="fig : " ] vs. time @xmath64 for @xmath70 and ( a ) @xmath71 ( b ) @xmath72 , ( c ) @xmath73 , ( d ) @xmath74 , and ( e ) @xmath75 , title="fig : " ] vs. time @xmath64 for @xmath70 and ( a ) @xmath71 ( b ) @xmath72 , ( c ) @xmath73 , ( d ) @xmath74 , and ( e ) @xmath75 , title="fig : " ] vs. time @xmath64 for @xmath70 and ( a ) @xmath71 ( b ) @xmath72 , ( c ) @xmath73 , ( d ) @xmath74 , and ( e ) @xmath75 , title="fig : " ] vs. time @xmath64 for @xmath70 and ( a ) @xmath71 ( b ) @xmath72 , ( c ) @xmath73 , ( d ) @xmath74 , and ( e ) @xmath75 , title="fig : " ] in figure 1 ( d ) we plot the results for @xmath67 and @xmath68 .
we find that again one can have large amplitudes of oscillation .
similar resonances appear for @xmath76 etc . however , in this work we shall only study the resonance of lowest order for @xmath60 .
finally , in figure 1 ( e ) we present a off - resonance result for @xmath67 and @xmath77 .
compared to figure 1 ( c ) with same values of @xmath48 and @xmath49 , we find oscillation with highly reduced amplitude in figure 1 ( e ) due to a variation of @xmath50 to an off - resonance value . if @xmath50 is taken further off - resonance the oscillations disappear quickly .
in a problem of resonance of an uncoupled oscillator driven by a periodic external force the oscillator oscillates in phase with the external force .
hence it gains energy during every cycle of oscillation and the amplitude of oscillation keeps on increasing .
so a pertinent question to ask is how can the amplitude of oscillation decrease in the present problem . in this nonlinear problem
the phase between the oscillation at resonance and the external force varies with time .
hence during a certain interval of time the two are in phase and the amplitude of oscillation grows . after the amplitude attains a certain value , the external force and the oscillation become out of phase and the system loses energy in each cycle .
eventually , the amplitude of oscillation reduces to a small value and this growth and decay cycle may repeat for a long time .
due to the nonlinearity in the equation the system moves in and out of resonance when a constant frequency drive is applied .
this leads to oscillations of varying time - dependent amplitude in the rms widths .
next we study the resonance in the situation with @xmath55 and @xmath78 . for a nonzero @xmath48
, pronounced resonances appear as @xmath49 is increased from 0 . in figures 2
( a ) , ( b ) , and ( c ) we plot @xmath61 vs. @xmath64 for @xmath79 , and @xmath80 and 0.5 , respectively .
the amplitude of oscillation increases as @xmath49 is increased .
we again see the growth and decay cycles for small values of @xmath49 .
however , such periodic growth and decay cycles are lost with the increase in @xmath49 as one can find in figure 2 ( c ) for @xmath67 . in figures 2 ( d ) and ( e )
we show the results for @xmath81 and @xmath82 and 2.4 , respectively . with the increase of @xmath49 , the amplitude of oscillation
has increased from figure 2 ( d ) to 2 ( e ) .
however , the periodic growth and decay cycles have been lost in this transition .
there are two time scales in each of the figures .
the first one is the driving frequency and the second one is related to the nonlinear term in the gp equation . for the linear problem
only the first time scale appears and the second frequency is absent . at resonance
the second frequency should increase with the effective nonlinearity of the system although a simple relation between the two can not be obtained . for @xmath56 the effective nonlinearity increases with increasing @xmath49 and at resonance the second frequency
is expected to increase as @xmath49 increases . however , for a positive @xmath48 the effective nonlinearity is large when @xmath49 is large and at resonance second frequency is expected to decrease as @xmath49 increases .
these interesting features of the growth and decay cycles emerges from figures 1 and 2 . at resonance for @xmath56
more growth and decay cycles appear for large values of @xmath49 .
however , quite expectedly for @xmath55 more growth and decay cycles appear for small values of @xmath49 .
next we continue our discussion of resonance to axially symmetric traps .
we find that when the external frequency @xmath50 coincides with the natural frequency of radial ( @xmath83 ) or axial ( @xmath84 ) oscillation , there is resonance in radial or axial oscillation , respectively .
the general nature of this oscillation is similar to that in the spherically symmetric case .
we studied the resonance dynamics for different sets of parameters .
however , in the following we present results only for @xmath56 and @xmath67 for angular momenta @xmath85 and 1 .
we consider both pancake ( @xmath86 ) and cigar - type ( @xmath87 ) deformation of the condensate in our study .
we recall that @xmath88 corresponds to the spherical case considered in the last section . and @xmath89 , respectively , vs. time @xmath64 for @xmath90 @xmath66 , @xmath91 and for ( a ) @xmath92 , @xmath60 , ( b ) @xmath93 , @xmath60 , ( c ) @xmath92 , @xmath94 , ( d ) @xmath93 , @xmath95 .
, title="fig : " ] and @xmath89 , respectively , vs. time @xmath64 for @xmath90 @xmath66 , @xmath91 and for ( a ) @xmath92 , @xmath60 , ( b ) @xmath93 , @xmath60 , ( c ) @xmath92 , @xmath94 , ( d ) @xmath93 , @xmath95 . ,
title="fig : " ] and @xmath89 , respectively , vs. time @xmath64 for @xmath90 @xmath66 , @xmath91 and for ( a ) @xmath92 , @xmath60 , ( b ) @xmath93 , @xmath60 , ( c ) @xmath92 , @xmath94 , ( d ) @xmath93 , @xmath95 .
, title="fig : " ] and @xmath89 , respectively , vs. time @xmath64 for @xmath90 @xmath66 , @xmath91 and for ( a ) @xmath92 , @xmath60 , ( b ) @xmath93 , @xmath60 , ( c ) @xmath92 , @xmath94 , ( d ) @xmath93 , @xmath95 .
, title="fig : " ] here , for angular momentum @xmath91 , we consider the following two values of the axial parameter @xmath96 and @xmath97 as some of the early experiments @xcite were performed with these anisotropies . also , these values of @xmath51 correspond to substantial contraction and elongation along the axial direction and hence to a substantial deviation from the spherical symmetry .
we present results for @xmath98 and @xmath52 in each of these cases . in figures 3 ( a ) ,
( b ) , ( c ) , and ( d ) we plot the rms radial and axial sizes @xmath99 and @xmath89 for @xmath92 and @xmath97 and for @xmath53 and @xmath52 . in figures 3 ( a ) and ( b ) resonance in radial oscillation is achieved for @xmath53 .
this results in the large and rapidly varying rms radii @xmath99 at resonance and a small and slowly varying rms size @xmath100 off the resonance . in figures 3 ( a ) and ( b ) there is one complete growth and decay cycle for @xmath101 .
then the resonance is manifested by a different type of oscillation in @xmath101 .
similar behavior was also found in the spherically symmetric case , e.g. , in figure 2 ( e ) . in figures 3 ( c ) and ( d )
resonance in axial oscillation is achieved for @xmath54 . in these cases
the axial size @xmath100 executes resonance oscillation , whereas the rms radius @xmath102 increases in size representing a swelling . in figure 3 ( c )
the axial trapping frequency @xmath103 is large , which makes the axial oscillation less likely than in the case presented in figure 3 ( d ) , where the small axial trapping frequency @xmath104 makes the axial oscillation more favorable .
consequently , the amplitude at the resonance in axial oscillation in figure 3 ( d ) is about four times more than that in figure 3 ( c ) . from figures 3 ( a ) and ( b ) we find that at the resonance of radial oscillation the axial rms size @xmath105 remains quite constant .
we find in figures 3 ( c ) and ( d ) that at the resonance of axial oscillation the radial rms radius @xmath106 varies more with time .
one interesting feature has appeared in figures 3 ( c ) and ( d ) , where for @xmath93 the radial rms radii @xmath61 vary with time corresponding to a slow radial swelling accompanied by small oscillation and not an oscillation with large amplitude .
we verified , as in the spherically symmetric case , that a smaller value of the constant @xmath49 leads to many growth and decay cycles of the amplitude at resonance in the axially symmetric case ( however , not shown explicitly here ) . here for angular momentum @xmath107 we consider the effect of a periodic variation of the scattering length via ( [ e ] ) on the quantized vortex states .
such states are of great importance to the study of the bec , as they are intrinsically related to the existence of superfluidity @xcite .
the experimental detection of vortex states in bec @xcite makes the resonant oscillation of the vortex states under a periodic variation of scattering length experimentally observable . and
@xmath108 , respectively , vs. time @xmath64 for @xmath90 @xmath66 , @xmath107 and for ( a ) @xmath92 , @xmath60 , ( b ) @xmath93 , @xmath60 , ( c ) @xmath92 , @xmath94 , ( d ) @xmath93 , @xmath95 . ,
title="fig : " ] and @xmath108 , respectively , vs. time @xmath64 for @xmath90 @xmath66 , @xmath107 and for ( a ) @xmath92 , @xmath60 , ( b ) @xmath93 , @xmath60 , ( c ) @xmath92 , @xmath94 , ( d ) @xmath93 , @xmath95 .
, title="fig : " ] and @xmath108 , respectively , vs. time @xmath64 for @xmath90 @xmath66 , @xmath107 and for ( a ) @xmath92 , @xmath60 , ( b ) @xmath93 , @xmath60 , ( c ) @xmath92 , @xmath94 , ( d ) @xmath93 , @xmath95 . ,
title="fig : " ] and @xmath108 , respectively , vs. time @xmath64 for @xmath90 @xmath66 , @xmath107 and for ( a ) @xmath92 , @xmath60 , ( b ) @xmath93 , @xmath60 , ( c ) @xmath92 , @xmath94 , ( d ) @xmath93 , @xmath95 .
, title="fig : " ] to illustrate the resonance we consider the lowest angular excitation of the bec rotating round the axial direction with each atom possessing angular momentum @xmath109 corresponding to @xmath107 .
again we consider the anisotropy parameters @xmath110 and @xmath97 and the external frequencies @xmath53 and @xmath52 . in figures 4 ( a ) , ( b ) , ( c ) , and ( d ) we plot the rms radial and axial sizes @xmath106 and @xmath108 in these cases .
these figures should be compared with the respective zero angular momentum cases shown in figures 3 ( a ) , ( b ) , ( c ) , and ( d ) . in the case of vortex states very orderly ( periodic )
growth and decay cycles of the resonating amplitudes appear for @xmath56 and @xmath66 in ( [ e ] ) .
the periodic growth and decay cycles of the resonating amplitudes are replaced by irregular ones as in figures 3 ( a ) , ( b ) , ( c ) , and ( d ) as the value of the constant @xmath49 is increased ( not shown explicitly in this paper ) .
the off - the - resonance amplitudes @xmath111 @xmath112 in figures 4 ( a ) and ( b ) and @xmath113 in figures 4 ( c ) and ( d ) @xmath111 remain constant without variation in all cases .
however , the period of growth and decay cycles of the resonating amplitude varies from one case to another . in all four cases
we have plotted the amplitudes up to 600 dimensionless time units .
this accommodates about 2.3 growth and decay cycles of the resonating amplitude in figure 4 ( a ) , 1.5 cycles in figure 4 ( b ) , 5 cycles in figure 4 ( c ) , and 0.5 cycles in figure 4 ( d ) .
comparing the situation of figure 4 ( c ) with that in figure 4 ( d ) we see that there is a reduction in the axial trapping frequency as we pass from figure 4 ( c ) to figure 4 ( d ) .
this reduction in the axial trapping frequency should favor the axial oscillation at resonance in figure 4 ( d ) .
consequently , we find that the amplitude of axial oscillation at resonance is much larger in figure 4 ( d ) compared to that in figure 4 ( c ) .
similar effect was also observed in figures 3 ( c ) and ( d ) in the @xmath91 case .
we have studied the occurrence of resonant oscillation in a bec due to a periodic variation in the atomic scattering length @xmath0 .
a variation in the atomic scattering length near a feshbach resonance can be and has been achieved experimentally @xcite by varying an external magnetic field .
this has been executed experimentally in producing a sudden jump in the scattering length from positive ( repulsive ) to negative ( attractive ) which has led to collapse and explosion in the condensate @xcite and several novel phenomena @xcite . here
we demonstrate that interesting resonating oscillation can be generated in the condensate if the scattering length is varied periodically with a given frequency .
this oscillation may be verified experimentally and novel and interesting phenomena may emerge as a consequence of such verification .
mathematically , the periodically oscillating scattering length in the gp equation can be considered as an external nonlinear periodic driving force in the linear schrdinger equation , which is an interesting problem of quantum mechanics and deserves special attention .
here we have considered a full numerical approach to its solution and have revealed interesting results .
one could also employ an analytic perturbative procedure for its solution when the coefficient of the external force is small .
this would be a problem of future interest .
we found that resonances with novel feature can appear in the oscillation of the bec from a periodic variation of the scattering length when the frequency of the oscillation of scattering length @xmath50 coincides with a multiple of the natural frequency of oscillation of the harmonically trapped bec .
for a spherically symmetric condensate , the natural frequency of oscillation of the bec is twice the trapping frequency ( @xmath114 ) .
hence resonances in the bec oscillation appear for @xmath115 etc . for an axially symmetric bec , the natural frequency of oscillation of the bec in radial and axial directions are twice the trapping frequencies in these directions , which are @xmath114 and @xmath116 , respectively .
consequently , resonances in the bec oscillation in the radial and axial directions appear for @xmath115 and @xmath117 , respectively .
the interesting feature of these resonances is that even in the absense of any damping ( dissipative force ) , the amplitude of oscillation at resonance can grow and reduce with time .
such growth and decay cycle of the amplitude may repeat several times .
the period of such growth and decay cycles is @xmath118 units of dimensionless time , which under usual experimental condition with @xmath119 hz / s corresponds to about 0.2 s @xmath111 a period that can be precisely measured and compared with theoretical prediction to test the mean - field theory .
we hope that the present investigation may stimulate further theoretical and experimental studies on this topic .
the work is supported in part by the conselho nacional de desenvolvimento cientfico e tecnolgico and fundao de amparo pesquisa do estado de so paulo of brazil . | using the explicit numerical solution of the axially - symmetric gross - pitaevskii equation we study the oscillation of the bose - einstein condensate induced by a periodic variation in the atomic scattering length @xmath0 . when the frequency of oscillation of @xmath0 is an even multiple of the radial or axial trap frequency , respectively , the radial or axial oscillation of the condensate exhibits resonance with novel feature . in this nonlinear problem without damping , at resonance in the steady state the amplitude of oscillation passes through maximum and minimum .
such growth and decay cycle of the amplitude may keep on repeating .
similar behavior is also observed in a rotating bose - einstein condensate . |
the renormalization group ( rg ) method provides a way to attack , and in many cases solve , the infrared problem of models in field theory , statistical mechanics and many body theory .
infrared problems are ubiquitous and physically natural , since they are inherent to systems with gapless excitations , such as critical systems of statistical mechanics , massless field theories , and systems with spontaneously broken continuous symmetries . in systems of correlated fermions modelling metals , excitations around the fermi surface
are gapless .
the attempt to treat interaction effects by naive perturbation theory leads to a severe infrared problem in the form of divergences due to small denominators .
the rg has been used for weak , short range two body interactions to cure this problem in perturbation theory and to classify the interaction terms by their power counting degree @xcite .
the one body ( or self energy ) terms , quadratic in the fermion fields , are the relevant ones , followed by the marginally relevant parts of the two body interaction , which are quartic in the fermion fields .
all higher terms are power counting irrelevant at weak coupling .
energy terms change the fermi surface , which is the location of the singularity of the fermion propagator . because the singularity has to be kept track of precisely
, neglecting these terms leads to problems even in formal perturbation theory ( provided it is done to higher than second order ) . for this reason , the self
energy terms are the most relevant ones .
the marginally relevant parts of the two body terms determine the quasiparticle interactions and give information about ordering tendencies and collective phenomena .
the hierarchy of equations couples the evolution of the different terms . by the above arguments , a possible way to proceed is to assume that the fermi surface given by the free propagator is already that of the interacting system .
one can then take the dispersion function of a certain form , hence the interacting fermi surface of a certain shape , assuming that this can be realized in an interacting system , and treat the remaining selfenergy effects by keeping them in the form of two legged vertices .
this also involves the assumption that there is no large correction to the quasiparticle weight .
both assumptions can , with considerable technical effort , be verified for the situation of smooth , positively curved fermi surfaces @xcite .
technically , this is done in @xcite using counterterms to fix the fermi surface .
the counterterm function determines the shift from the fermi surface of the free to that of the interacting system .
the above realizability question takes the form of an inversion problem , which is solved under certain conditions in @xcite .
although the importance of the self energy terms is known and clearly acknowledged in the literature , a way to proceed in applied studies has been to focus on the study of the two body interactions while neglecting self energy terms in the flow .
the main motivation for this is simplicity . controlling the relevant selfenergy terms
carefully is technically complicated already in the case of a smooth fermi surface .
the counterterm method of @xcite provides a conceptually clear way of dealing with the problem , but it has been used only in a few applied studies so far . in the two dimensional hubbard model ,
the case with van hove singularities is of interest for high-@xmath0 superconductivity @xcite . for this case , a renormalized expansion using counterterms
is developed , and some regularity properties of the self energy are shown , in @xcite .
however , the question of realizability of singular fermi surface shapes has not been fully answered yet , although it is very important for deciding how generic van hove singularities really are in two dimensional interacting systems @xcite .
the effects of fermi surface dynamics near van hove points can be rather nontrivial , since the fermi surface is most easily moved in the vicinity of a zero of the gradient of the dispersion relation .
this point was first clearly realized in @xcite , and it was shown that quantum fluctuations of the fermi surface can arise and lead to non fermi liquid behaviour @xcite .
in this situation , it is useful to have a method in which the adjustment of the fermi surface is done in the flow itself , because the concept of a non fluctuating fermi surface itself may make sense only down to a certain scale , below which the fluctuations emerge and may become dominant .
the method discussed in this paper allows to use dynamically adjusting scale decompositions and to study fermi surface flows . as a method
, it applies more generally and can be used flexibly to adapt and adjust propagators in bosonic and fermionic ( or mixed ) systems , both for continuous and discrete flows .
variants of the method are used in mathematical studies of fermi surface flows and fermi liquid theory @xcite .
also , the method has already been used to calculate the fermi surface flow for the hubbard model ( see the appendix of @xcite ) .
various renormalization group schemes have already been used to study correlated fermion systems : polchinski s original scheme @xcite , the wick ordered scheme @xcite , the 1pi scheme @xcite .
flow equations for the 2pi functions were introduced @xcite and studied for quasi-1d systems @xcite . in this context
one may well ask why one would want to investigate another scheme . in the following ,
i give some further motivations for this .
as is well known , all the above mentioned schemes are equivalent on the level of untruncated hierarchies , because all of them contain the full information about all green functions , as long as they are kept as infinite hierarchies .
but they start to differ once the approximations that are necessary in nontrivial applications , are made , most importantly truncations of infinite hierarchies , gradient expansions of the generating functionals , or simplifying assumptions about the dependence of the green functions or vertices on momenta and frequencies .
it is then useful to understand which features of a given scheme make it useful for a particular situation .
all of the above mentioned schemes include , in their untruncated form , all self energy effects in the flow , albeit in a rather different way . in their original form ,
the polchinski and wick ordered schemes keep self
energy effects as vertices , while the 1pi scheme includes them in the most natural way , namely by having full propagators on the lines . on the other hand ,
the wick ordered scheme can be transformed so as to have full propagators on the lines ( see section 4.7.2 of @xcite ) , and a similar transformation also applies to the polchinski scheme .
however , the standard momentum space setup of the rg with a cutoff in momentum space near to the fermi surface does not include an adaptive scale decomposition , and thus does not allow for the above mentioned flow of the fermi surface .
this is the case in all schemes , so a correct procedure would require using counterterms even in the 1pi scheme ( in most applied studies , the fermi surface shift was neglected , so that this problem did not appear ) .
alternatively , one can try not to use a cutoff on the fermion momentum @xcite , but power counting has been done rigorously in more than one dimension only in the momentum space scheme @xcite . in two or higher dimensional applications , it is a serious practical problem to take the momentum and frequency dependence of the flowing vertex functions into account .
the full propagator @xmath1 of the 1pi scheme has support on all scales above the cutoff scale .
this makes it necessary to extend the loop integrals over all of momentum space , which in turn requires having an accurate form of @xmath1 , _ and _ of the higher vertex functions , over a wide range of momenta and frequencies .
this has not been achieved so far .
the situation in the other schemes is better in this respect : in the polchinski equation , all internal lines carry a derivative of the cutoff function .
therefore , at scale @xmath2 , the propagators get evaluated only at values of the momentum @xmath3 with @xmath4 .
the integration procedure for polchinski s equation makes the equation nonlocal in the scale parameter in the standard way @xcite , but a significant advantage is that whenever a momentum region @xmath5 further away from the fermi surface contributes to a loop integral in the flow equation , it does so only via a propagator on scale near to @xmath6 , which was already calculated in an earlier stage of the flow , and which does not need to be adjusted in every further integration step . in the wick ordered scheme , all propagators are supported at or below the cutoff , so only small neighbourhoods of the fermi surface occur in the loop integrals at low scales .
thus the propagators and vertices are needed only close to the fermi surface and at small frequencies . in studies and talks by the g " ottingen group @xcite ,
it was pointed out that in simple models , the 1pi scheme has a better strong coupling behaviour than the other schemes .
namely , the 1pi flow equation is accurate in these toy models at much larger coupling values than one might believe at first , far beyond the weak coupling regime , and it remains accurate in a region of intermediate to strong couplings where all other schemes fail .
the dynamically adjusting scheme is examined for this toy model in section 4 .
it turns out that propagator adjustment , namely putting selfenergy effects into the denominator instead of keeping them as vertices , is a necessary , but not sufficient , ingredient in stabilizing the equations at strong coupling .
the polchinski scheme still fails at strong coupling , while the dynamically adjusted wick ordered scheme is accurate up to arbitrarily large couplings .
the analysis also clarifies the reasons for the success of the 1pi scheme . in section 4
, i also give a more general discussion about strong initial couplings , and show that for reasons that are simple , but specific to the rg method , the `` weak coupling '' rg flow is a well defined starting point at strong coupling also in `` real '' models ( such as the two dimensional hubbard model ) , contrary to what one may believe at first , and to what is stated in many places . whether this can be used efficiently in computations
depends on certain relative rates of growth ; this is under investigation and will be discussed at the end .
the partition function of a general bosonic or fermionic theory with source fields @xmath7 ( the generating function for the green functions ) is @xmath8 here the fields @xmath9 are indexed by elements @xmath10 of a finite set @xmath11 .
@xmath11 is taken finite to make the generating function mathematically well defined . in many applications , @xmath12 , where @xmath13 is a finite space - time lattice and
@xmath14 is a finite set of internal indices , but this assumption will not be needed for the general theory . if @xmath11 comes from such a lattice regularization of a continuum theory , the limit of lattice spacing to zero and volume to infinity can be taken later in the rg equations ( and after a volume factor has been divided out in the field independent term ) . readers who trust that all this can be done properly ( or who do nt care ) can also think of a continuous space @xmath15 right away . for bosons ,
the @xmath9 are real valued ; for fermions , each @xmath9 is one of the generators of a grassmann algebra .
spinors and vector fields are included in this setting by an appropriate choice of @xmath14 , and complex bosonic fields as well , by regarding them as two component real fields ( with the corresponding index again part of @xmath14 ) .
the commutation relations are @xmath16 with @xmath17 for bosons and @xmath18 for fermions .
the integration measure is @xmath19 , and @xmath20 with @xmath21 ; ( when considering a lattice approximation , @xmath22 is the volume of the unit cell of the lattice ) .
the action of the operator @xmath23 is @xmath24 ; the matrix elements @xmath25 are in general complex .
an operator @xmath26 is called
_ @xmath27symmetric _ if @xmath28 ( i.e. @xmath29 ) . in the bosonic case , re @xmath26 is also required to have nonnegative eigenvalues , so that for all @xmath30 , @xmath31 ( in the case of complex fields viewed as two component real fields , this corresponds to the usual condition that the hermitian part of the covariance is positive definite ) .
assume that the @xmath23 in @xmath32 is @xmath27symmetric .
if @xmath23 is invertible , both @xmath23 and the _ propagator _ @xmath33 are @xmath27symmetric .
the normalization factor @xmath34 is chosen such that @xmath35 for @xmath36 . for bosons ,
i also assume that @xmath37 is bounded below and that @xmath38 grows faster than linearly in all @xmath39 for large @xmath39 . under these hypotheses ,
the integral in @xmath32 is a convergent finite dimensional integral in the case of bosons . in the case of fermions
, it is a linear functional on a finite dimensional space .
the theory can also contain both bosonic and fermionic fields .
every formal functional integral can be regularized so that it takes this form .
the logarithm of @xmath40 generates the connected green functions of the theory .
assume that @xmath23 is invertible and let @xmath33 .
a shift in the integration variable gives ( using the symmetry of the quadratic part under @xmath41 ) @xmath42 with @xmath43 the normalized gaussian measure with covariance @xmath26 . in this form ,
one can take a limit where some of the eigenvalues of @xmath26 go to zero ; in the bosonic case this corresponds to restricting the integral to the subspace @xmath44 .
typically , this situation arises by introducing cutoff functions that vanish strictly in parts of momentum space ; see the following examples .
although the equations derived in this paper work for the general case set up above , it is useful to keep two prototypical examples in mind . the first example is scalar field theory on a finite lattice @xmath45 where @xmath46 , @xmath47 is the discrete laplacian with ( say ) dirichlet boundary conditions , and @xmath48 , with @xmath49 a polynomial that is bounded below . in that case
, @xmath50 with fourier transform @xmath51 for @xmath52 .
the momentum @xmath53 is removed by the dirichlet condition ; of course , other boundary conditions can be used , too , provided the laplacian is restricted to the complement of its kernel . for details , see , e.g. , chapter 2 of @xcite .
the second example is the regularized partition function for spin @xmath54 fermions on a finite sublattice @xmath55 of @xmath56 , @xmath57 . here
@xmath58 , @xmath59 , with the first index being a charge index and the second the spin index . for @xmath60 , with @xmath61 and @xmath62 , with @xmath63 , @xmath64 where @xmath65 and the fourier transform of @xmath66 is of the form @xmath67 , where the matsubara frequency @xmath68 takes values in odd multiples of @xmath69 times the temperature @xmath70 and @xmath71 is a real valued function that has a nontrivial zero level set , the fermi surface @xmath72 .
@xmath73 is an even element of the grassmann algebra with @xmath74 . in the simplest case
, @xmath75 is a quartic polynomial in the fields , e.g. @xmath76 , but it can be much more general .
the number @xmath77 is an auxiliary quantity coming from an application of the lie
trotter product formula . to recover a representation of the partition function of the many fermion model , as given by a trace over fock space , one must take @xmath78 first . for details , see chapter 4 of @xcite .
the renormalization group comes into play when the inverse of @xmath23 becomes unbounded , or even fails to exist , in the limit @xmath79 , or when the inverse remains bounded , but its norm becomes nonuniform in an essential physical parameter . in the above examples , the fourier transform diagonalizes @xmath23 and one can read off that the eigenvalues go to zero in the limit of interest .
in example 1 above , taking @xmath80 to be a cube of sidelength @xmath81 , @xmath82 with @xmath83 ; clearly , @xmath84 goes to zero for infinitely many @xmath85 in that limit , so that @xmath86 becomes singular .
similarly , in example 2 , one gets zero eigenvalues when the temperature @xmath87 , and when @xmath88 . at small positive @xmath89 ,
@xmath90 is of order @xmath91 . in both cases , @xmath92 , and thus @xmath26 itself , fails to be square integrable in the limit @xmath87 , so that integrals over a square of a propagator ( and higher powers of it ) are ill defined .
this is not just mathematical pedantry but physically relevant if in the case of the many fermion system , @xmath92 were square integrable at @xmath93 , there would be no superconductivity , because the superconducting gap is driven by the @xmath94 from @xmath95 .
similarly , the singularity of @xmath96 in the bosonic case leads to anomalous decay exponents of correlation functions .
hence in the following , a guideline for getting a useful rg equation in these systems is that _ squares of unregularized propagators are to be avoided_. a standard approach to obtain a renormalization group equation @xcite in the above examples is as follows .
introduce a scale parameter @xmath97 , an @xmath98dependent energy , say @xmath99 , which goes to zero in the limit @xmath100 .
choose a fixed decreasing @xmath101 cutoff function @xmath102 with @xmath103 for @xmath104 and @xmath105 for @xmath106 , and set @xmath107 , so that @xmath108 is a smooth partition of unity .
decompose the propagator into an infrared part @xmath109 and a part supported at higher scales , @xmath110 , via @xmath111 and define the effective action @xmath112 the partition function is then expressed in terms of the effective action as @xmath113 the effective action @xmath114 is a function of a fields which , when expanded in the fields , has a quadratic term . in infrared problems ,
this term is relevant in the rg sense .
that is , the order of the singularity may get changed , producing the anomalous exponents , in the scalar field theory example , or , in the many fermion system , the changes to the quadratic term also shift the location of the singularity of the propagator , the fermi surface , in fourier space .
the above scale decomposition does not take this into account , because it zooms in on the fixed fermi surface @xmath115 .
therefore , with such a fixed choice of scale decomposition , counterterms are needed to avoid divergences @xcite .
it is possible to justify the use of counterterms by an inversion theorem @xcite , but it seems desirable to have a simple method for moving the quadratic terms from @xmath114 into the propagator , in a way that the shift of the singularity gets included in the flow automatically . the flow introduced in this paper allows to put self energy terms into the propagator and hence to define a convenient adaptive scale decomposition both in continous ( and in discrete ) renormalization group flows .
it is such that an analogue of @xmath116 continues to hold , but with a propagator that changes dynamically in the flow , and @xmath117 replaced by an interaction from which all , or part of , the quadratic part is removed .
finding a scheme that implements an adjusting scale decomposition in a differential equation is not completely straightforward . on graphical grounds ,
one expects the full propagator to be @xmath118 , where @xmath119 is the dyson self energy .
thus the first idea may be to make the ansatz @xmath120 with @xmath121 , @xmath122 to be determined .
@xmath123 is well - defined because @xmath124 is nonzero only on a subspace where @xmath125 can be inverted , and the cutoff function now adapts dynamically because it depends on @xmath126 .
the propagator of the unintegrated fields must be defined as @xmath127 to ensure that @xmath128 .
but this has a singularity at the noninteracting fermi surface for any @xmath98 , as well as one developing on the new one , so ( apart from being a very unnatural object ) it does not vanish for @xmath100 , so that not all degrees of freedom get integrated out
. then this flow would not give useful information about the model , even if it converged for @xmath100 .
trying instead to keep the form of @xmath116 , but putting @xmath129 leads to disaster : the derivative @xmath130 contains a term @xmath131 , which has a square of the inverse , _ but no infrared cutoff _ because @xmath103 for @xmath104 . in all cases where the self energy shifts the fermi surface , @xmath132 on the singularity set , so that effectively , a square of the propagator appears .
because this propagator is not square integrable , this scheme leads to divergent integrals already in the rg differential equation at fixed @xmath98 .
the method developed in the following sections avoids these problems .
i now return to the general field theoretic situation of section [ secsetup ] and consider the generating function in the form @xmath133 , i.e.@xmath134 instead of @xmath135 i make the slightly more general choice @xmath136 where @xmath137 can be chosen as an operator that cuts off very large energies , chosen such that @xmath138 is @xmath139symmetric , and for bosons , nonnegative ( but @xmath140 is also allowed ) . in applications , such an ultraviolet cutoff allows to start the flow at a scale where degrees of freedom with very high energies have already been integrated over .
the flow is set up by posing restrictions on the general , @xmath98dependent partition function @xmath141 here @xmath142 , @xmath1 , @xmath143 , @xmath144 , _ and _ @xmath145 to be determined as functions of @xmath98 . for fixed @xmath98
, @xmath146 is a complex number , @xmath144 is an operator , @xmath1 and @xmath145 are @xmath27symmetric operators , and @xmath147 is a function of the fields @xmath7 , which , for bosons , has to be such that the integral exists ( see below ) .
i impose the following general conditions .
1 . @xmath148 , @xmath149 , and @xmath150 .
2 . for all @xmath151 and all @xmath7 , @xmath152 .
3 . @xmath153 as @xmath154 . these conditions have the following simple interpretation . 1 .
@xmath155 is the generating function of the model of interest .
2 . the generating function @xmath156 , and hence all correlation functions , are invariant under the rg .
3 . in the end , all degrees of freedom are integrated out . the field
independent term @xmath157 is ( up to a factor @xmath158 ) equal to the free energy .
the term involving @xmath159 naturally arises once one starts adding terms to the denominator of the propagator .
this is most easily seen in the case when one requires that @xmath147 should contain no quadratic term in @xmath7 .
then , if the limit @xmath100 exists and @xmath160 in that limit ( as required by condition 3 ) , @xmath161 because @xmath162 has no quadratic term for any @xmath98 , the same holds for @xmath163 .
therefore the term involving @xmath164 is the only quadratic term and hence @xmath164 is the full propagator .
the limit exists e.g. if the temperature is above all critical temperatures @xcite or if the fermi surface is asymmetric under reflections @xcite .
thus , the full propagator @xmath1 is built up gradually in the rg flow . for finite @xmath98 , @xmath165 has the usual rg interpretation that @xmath166 is the propagator of particles with interaction @xmath162 , but now @xmath162 can be chosen not to contain any self energy terms .
thus , even if the limit @xmath154 can not be taken because the flow is accurate only up to some @xmath167 ( see , e.g. @xcite ) , the effective propagator and interaction can still be used as an input for the theory of the excitations with energy below @xmath168 .
the above conditions do not determine @xmath146 and the functions @xmath169 , @xmath147 , @xmath170 , and @xmath171 uniquely .
they are just the minimal conditions one wants to impose .
the freedom of choosing the various functions can be used to impose the constraint that the quadratic part of @xmath162 should vanish , or that a certain part of it be removed from @xmath162 and put into @xmath171 instead . the flow @xmath116 satisfies h1h3 , but it restricts to the propagator @xmath172 with the fixed scale function @xmath173 of @xmath174 , which does not adjust to a shifting location of the singularity .
this condition is relaxed when one drops @xmath116 in favour of @xmath165 . in particular
, one can choose the functions such that the propagator @xmath171 incorporates all , or part of , the selfenergy corrections .
in applications , this means that one can directly study the flow of the fermi surface , the flow of superconducting gaps and the like , without having to use counterterms .
i seek only solutions to the conditions h1h3for which @xmath142 , @xmath175 and @xmath37 are differentiable in @xmath98 .
then h1 and h2 are equivalent to the initial
value problem @xmath176 from which the rg differential equation will be obtained .
for a @xmath27symmetric operator @xmath177 , the laplacian in field space is defined as @xmath178 where @xmath179 .
[ th1 ] let @xmath180 be given , with @xmath181 @xmath139symmetric for all @xmath98 and bounded for all @xmath98 .
assume that there is a solution @xmath182 to the initial
value problem @xmath183 such that @xmath184 is differentiable and the low energy propagator @xmath185 and the fluctuation propagator @xmath186 exist and are @xmath27symmetric , and that @xmath171 is nonnegative for bosons .
assume that @xmath142 and @xmath37 satisfy the initial
value problem @xmath187 ( thus in particular @xmath188 is differentiable in @xmath98 and at least twice differentiable in @xmath189 ) and the condition that @xmath190 for all @xmath98 . set @xmath191 and let @xmath1 be the solution to @xmath192 then the functions @xmath193 and @xmath37 solve the initial value problem @xmath194 , hence they satisfy h1 and h2 .
this theorem is proven in the next subsection .
* remarks .
* 1 . @xmath145 must be @xmath27symmetric , and for bosons it must be nonnegative . a way to achieve
this is to restrict to @xmath27symmetric @xmath195 that commutes with its adjoint @xmath196 , and to take @xmath197 as a function of @xmath198 , so that the two commute , e.g. @xmath199 .
the conditions on @xmath177 are satisfied in the above examples and in most other interesting applications .
2 . the dependence of the cutoff function @xmath197 on @xmath195 allows for an @xmath98dependent , adaptive scale decomposition .
this makes showing the existence of a solution to @xmath200 nontrivial .
this existence problem will be discussed in more detail in section [ sec:2.9 ] .
the fluctuation propagator @xmath201 given in @xmath202 is now indeed proportional to the derivative of the scale function , thus has an infrared cutoff .
therefore the problem mentioned after @xmath203 is absent from this flow .
the equation for the full propagator @xmath1 can be rewritten as @xmath204 thus in general , @xmath205 , also in the limit as @xmath100 .
this must be so for general reasons , as will be explained in section [ diagrammar ] .
the strategy to satisfy h3 as well is to choose a solution for which the support of @xmath197 becomes empty as @xmath100 , and for which the singularity of @xmath206 as @xmath154 is such that @xmath207 still vanishes in the limit .
theorem [ th1 ] gives a family of solutions to the conditions h1 and h2 parametrized by @xmath208 .
moreover , it will become clear in the proof that even given a fixed @xmath208 , there are many other possibilities to choose the equation for @xmath37 , and as explained , there is a further freedom of choosing @xmath209 as a function of @xmath177 when seeking a solution @xmath210 to @xmath200 .
thus , there seems to be an enormous freedom in solving h1h2 . however , not every choice has good analytical properties , and most solutions diverge already at a small value of @xmath98 . to have a bounded solution on a maximal interval for @xmath98 requires a careful choice of @xmath181 .
for instance , in the fermi surface problem in @xmath211 , @xmath208 must take the fermi surface shift into account fully ; whether it also needs to contain corrections to the fermi velocity and the coefficient of @xmath68 depends on finer details , such as the shape of the fermi surface . if this is done properly , the limit @xmath100 of the flow exists at temperatures above the mean field transition temperature @xcite , and h3 is satisfied . 7 .
compared to the polchinski equation obtained from @xmath116 by differentiating with respect to @xmath98 , which reads @xmath212 ( with @xmath213 as in @xmath174 ) , there are two changes .
first , there are terms of order 0 and 2 in the fields @xmath7 on the right hand side of @xmath214 , which can be chosen to cancel the field independent and quadratic part arising from the first term on the right hand side , so that @xmath215 starts with terms of order @xmath216 in @xmath189 .
second , the fluctuation propagator @xmath201 now includes selfenergy corrections from the integration up to scale @xmath98 ( if @xmath181 is chosen to absorb only selected parts of the quadratic part generated by the integration , e.g. only cancelling the fermi surface shift , then the propagator @xmath195 , and hence @xmath201 , contain only those ) . +
thus @xmath214 has a simple and natural interpretation : the propagator is allowed to change dynamically , and the selfenergy ( or part of it ) can be put into the denominator .
although the rhs of @xmath217 has the same structure as the first term on the rhs of @xmath214 , the change of the quadratic part , and the ensuing replacement of @xmath218 by @xmath201 leads to a major change in the behaviour of the rg equations .
some aspects will be discussed in section [ toysec ] . to make the proof more readable
, i first give an algebraic derivation which applies both to bosons and fermions , and in a subsequent section provide the mathematical justifications of all steps .
define the convolution with the gaussian measure as @xmath219 the partition function is @xmath220 if @xmath221 is differentiable in @xmath98 and @xmath189 , @xmath222 with @xmath223 the equation @xmath224 implies @xmath225 with @xmath226 using @xmath227 gives @xmath228 acting with @xmath229 on @xmath230 and using @xmath231 ( recall @xmath232 for bosons and @xmath233 for fermions ) i get @xmath234 with @xmath235 to simplify the equations , i remove the dilation term @xmath236 , i.e.choose @xmath144 so that @xmath237 this implies @xmath238 , so that @xmath239 this equation makes it natural to regard @xmath181 as an input , and to seek a solution for @xmath145,@xmath37 , and @xmath142 .
once this is done , the remaining functions are obtained by integrating the equations @xmath240 and @xmath241 .
let @xmath145 be of the form @xmath242 with @xmath195 and @xmath197 the solution to @xmath200 .
differentiation with respect to @xmath98 implies that @xmath201 is indeed given by @xmath202 . with the choice @xmath242
, the amputation operator @xmath144 can now also be determined : @xmath243 , with initial condition @xmath244 has the unique solution @xmath245 stated in @xmath246 .
thus theorem [ th1 ] is proven algebraically for bosons and fermions .
adapting the derivation to the case of a theory involving bosons and fermions is straightforward , provided that the interaction @xmath247 is an even element of the grassmann algebra . in the above derivation
, it was assumed 1 . that @xmath37 and @xmath156 are indeed differentiable in @xmath98 and in @xmath7 2 . that @xmath248 holds , and 3 .
that the commutator equations hold , when applied to @xmath249 .
item 1 follows from the existence of the solution to the initial value problem for @xmath142 and @xmath37 , as stated in the hypotheses of theorem [ th1 ] ( and already explained there ) . concerning item 2 ,
note that for any polynomial @xmath250 in the fields , @xmath251 was proven in @xcite both for bosons and fermions , but @xmath249 is in general not a polynomial .
similarly , the commutator equations @xmath252 and @xmath253 hold when applied to polynomials in @xmath7 , but again , the above proof involves an application to @xmath249 . consider first the case of fermions . in the present setup , the grassmann algebra is finite dimensional , so every function of the grassmann variables is a polynomial .
thus @xmath254 and @xmath255 are polynomials in @xmath30 and @xmath256 is a polynomial , hence infinitely differentiable in @xmath7 .
thus @xmath248 holds , and @xmath252 and @xmath253 hold as well
. moreover @xmath257 is differentiable in @xmath98 by hypothesis of theorem [ th1 ] , and @xmath177 is differentiable in @xmath98 because it solves the differential equation @xmath200 , so @xmath145 is differentiable in @xmath98 as well .
by @xmath248 , the convolution in @xmath165 is a polynomial in @xmath171 and in @xmath258 .
thus by expanding @xmath37 in homogeneous parts of degree @xmath259 in @xmath189 ( see also section [ diagrammar ] ) , one easily proves by induction on @xmath260 that @xmath37 , and hence also @xmath156 , is differentiable in @xmath98 . for bosons ,
@xmath261 is not a polynomial , but , under our assumptions , it has a fast decay for @xmath262 .
this decay and some analyticity could be used to justify first truncating @xmath249 to a polynomial and then taking a limit . a different , easier way
to adapt the argument is to avoid @xmath248 , i.e. go back to @xmath263 and use integration by parts arguments of the type @xmath264 to bring the equation into the form @xmath265 where @xmath266 is the difference between the left and right hand sides of @xmath267 . keeping @xmath268 in the derivation and obtaining the general case of nonnegative @xmath145 by a limit ,
an equation of this form implies that @xmath269 .
this yields the same differential equation as the algebraic method of the previous section and thus completes the derivation for bosons . in conclusion ,
all steps in the derivation of the rg equations in the previous subsection are justified mathematically . in the case of fermions ,
the smoothness of @xmath188 in @xmath98 and @xmath189 follows from easy general arguments . for bosons ,
it is part of showing that the solution to the equations for @xmath37 and @xmath142 exists .
for brevity i call the rg differential equation for @xmath270 and @xmath271 derived above the dynamical rg or dynrg equation . expansion of @xmath37 in powers of the fields gives a hierarchy of equations for the @xmath260point functions . in this section ,
i assume that @xmath37 is even and choose @xmath208 and @xmath142 such that @xmath37 does not contain any quadratic and constant part , which gives the hierarchy with the simplest graphical structure .
the expansion in the fields @xmath272 with @xmath273 homogeneous of degree @xmath259 in @xmath189 , leads to the hierarchy ( shown graphically for @xmath208 , @xmath274 and @xmath275 in figure [ fig1 ] ) @xmath276 by @xmath200 , @xmath277 reads @xmath278 so it integrates to @xmath279 for fermions , @xmath280 and @xmath281 , where pf @xmath282 denotes the pfaffian of the antisymmetric matrix @xmath177 . for covariances of the form @xmath283 , pf @xmath284 , so that @xmath285 for bosons , @xmath232 and @xmath286
. thus for real bosons , @xmath287 for complex bosons with a hermitian covariance there is a factor @xmath288 which squares this to @xmath289 .
thus the relation of the operator @xmath177 to the free energy density is simple : if @xmath290 for @xmath100 , the free energy is ( modulo a conventional factor of the inverse temperature ) simply given by @xmath291 this also clarifies why the equation @xmath292 for @xmath1 can not simply be @xmath293 : the exact relation between the full propagator @xmath1 and the free energy involves the luttinger
ward functional .
it is a nice feature of the dynrg flow that the relation between @xmath177 and the free energy @xmath142 is so simple .
clearly , the diagrammatic structure of the hierarchy ( see figure [ fig1 ] ) differs from the polchinski hierarchy obtained from @xmath217 by the absence of one particle functions ( or two point functions ) which get absorbed into @xmath177 via the equation @xmath294 for @xmath208 and @xmath200 .
the standard iterative solution of this hierarchy produces a graphical expansion .
it should be noted that from a purely graphical perspective , the dynamical propagator adjustment does not make any of the functions one irreducible .
this is obvious for the six point and higher functions , but also holds for the graphs contributing to the four point function @xmath274 and to @xmath208 : taking the tree diagram of the six point function and joining two legs of one of the vertices to a tadpole produces a one reducible contribution to @xmath274 , taking two legs of the other vertex and joining them to a tadpole gives a one reducible contribution to @xmath208 .
the point here is that these two point insertions ( the _ single - scale insertions _ defined in @xcite ) are harmless because they do not generate singularities . in the fermi surface example , the size of such an insertion at scale @xmath98 is by power counting ( see @xcite ) of order @xmath295 , while the size of the propagators on the adjoining lines is of order @xmath296 .
thus the combination of two propagators and such a two point insertion is of order @xmath297 , which is the same size as a single propagator .
however , if one inserted a two point graph that is integrated from @xmath298 to @xmath98 , its value would be @xmath299
. then the combination of this insertion with two propagators is of size @xmath300 , hence @xmath301 larger than it should be .
it is contributions of this type that lead to a problem with unrenormalized expansions . here ,
this problem does not arise because renormalization is implemented by a changing propagator . in summary , for the analytic questions , namely which terms are large and which are small , the reducible graphs contributing to @xmath274 and @xmath208 are inessential . expanding in wick ordered polynomials provides essential simplifications in the mathematical analysis of the rg equations @xcite ; moreover the wick ordered scheme has been useful for calculations in two dimensional systems @xcite .
a distinctive feature of the wick ordered equation is that all propagators are supported on scales below @xmath302 , so that only low
energy degrees of freedom enter the integration . in the many
fermion example , this means that all momentum integrals are restricted to a neighbourhood of the fermi surface .
therefore projections to the fermi surface , which are needed to solve the equations numerically , are better under control than in other schemes .
an further advantage is that structures of overlapping loops @xcite become explicit in the equations and make the study of improvements to power counting much easier .
these improvements are needed for showing fermi liquid behaviour in higher dimensions and controlling two particle interactions . a natural ( but not the only ) way to fix the wick ordering covariance is to require that tadpoles be removed . in the case of the fixed scale decomposition
@xmath174 , this is easy to do : the condition to cancel tadpoles is to have @xmath303 , because then the linear term in the rgde drops out @xcite . in this case
, the solution to that equation is simply @xmath304 , with the @xmath109 of @xmath174 . in the adaptive scheme @xmath214
, the corresponding equation is @xmath305 which is _ not _ solved by @xmath306 .
the reason for this is simply that a wick ordering that removes all tadpoles requires information about the propagator at later scales @xmath307 , which requires solving self consistency equations . in theoretical studies ,
this self consistency can be solved by fixed point arguments .
alternatively , one can choose the wick ordering covariance such that it leads to the above improvements without cancelling tadpoles exactly .
the wick ordered equation is obtained from @xmath214 @xmath308 by setting @xmath309 then @xmath310 and the resulting equation is @xmath311 .
\nonumber\end{aligned}\ ] ] here the usual combinatorial device @xcite of introducing two copies @xmath312 and @xmath313 of the field has been used to rewrite the quadratic term , and @xmath314 denotes evaluation at @xmath315 . the choice of @xmath316 as the solution to @xmath317 makes the linear term in the equation drop out , as in @xcite .
there are some further conditions on @xmath316 that were automatically satisfied in the case without propagator adjustment in @xcite .
the support of @xmath316 needs to go to zero as @xmath100 , and @xmath316 needs to satisfy certain power counting bounds to guarantee that wick ordering does not introduce singularities .
for this reason , rewriting the solution in terms of an initial condition as @xmath318 is not useful .
instead , requiring @xmath319 for @xmath100 , the solution is conveniently written as @xmath320 however , in contrast to the situation without adjustment of propagators ( where @xmath304 ) , this is really a self consistency equation because it requires knowledge about the propagator at later rg times . for general purposes , such as order by order arguments and power counting , it suffices to know general properties of @xmath316 , and to solve for @xmath316 inductively while solving the rg equations . in practical applications
, one needs to use approximate solutions @xmath321 instead of @xmath316 , for which @xmath322 is less singular than @xmath316 itself .
the expansion @xmath323 with @xmath324 homogeneous of order @xmath259 in @xmath7 gives the following hierarchy .
for @xmath325 , @xmath326 for @xmath327 @xmath328 and for @xmath329 @xmath330 \nonumber\\ & + & \sum_{\ell \ge 3 } \sum_{\mu=2}^{\ell -1 } \frac{2^{-\ell -1}}{\ell ! } \
; { { { \lbrack\!\lbrack}}}(\delta_r^{(1,2)})^\ell \ ; \delta_f^{(1,2 ) } \ ; { { \cal w}}^{(\mu ) } ( s,{\psi}_1)\ ; { { \cal w}}^{(1+\ell-\mu ) } ( s,{\psi}_2 ) { { { \rbrack\!\rbrack}}}\end{aligned}\ ] ] this hierarchy is shown graphically in figure [ fig2 ] .
the number of loops of the corresponding graphs is @xmath331 .
as is explicit in the above equation , the contributions quadratic in @xmath332 start at two loop for @xmath333 and at three loop for @xmath334 .
the choice @xmath335 removes the term that is linear in @xmath332 .
a graphical analysis of the iteration implies that with this choice , there are no one reducible contributions to @xmath336 and to @xmath208 .
moreover , if the one loop term is dropped from the eq . for the four - point function , then all graphs contributing to @xmath336 and @xmath208 are 2pi .
if the wick ordering covariance @xmath316 is chosen as the solution to @xmath337 that vanishes for @xmath100 , as above , and if the hierarchy is truncated by setting @xmath338 for all @xmath339 , the resulting equation for @xmath336 becomes @xmath340 ^ 2 \right ) { { \cal w}}^{(2 ) } ( s,{\psi}_1)\ ; { { \cal w}}^{(2 ) } ( s,{\psi}_2 ) { { { \rbrack\!\rbrack}}}\end{aligned}\ ] ] the structure of this equation is analogous to that obtained by a similar truncation from the 1pi hierarchy _
after the 1pi hierarchy has been rearranged in katanin s @xcite way _ : eq .
( 25 ) ) in @xcite is to quadratic order in the interaction of the same form as @xmath341 .
it is also analogous to equations ( 4.176 ) ( 4.178 ) in @xcite because there a product of propagators @xmath342 appears in the flow equation as well ; however , there , @xmath109 does not contain self
energy corrections ) . note that in contrast to the 1pi case , the structure of the scale derivative of a square of the laplacian ( which is essential for the solution of the selfconsistency ) appears in the wick ordered dynrg in a natural way ; no rearrangement of the hierarchy is necessary .
it is thus clear that , when restricted to a particular channel , the equation is easily solved by a geometric series in analogy to the procedure in section 4.5.4 . of @xcite . in the present framework , restricting to a channel means that ( possibly after a change of variables ) , the equation for the coefficient function @xmath343 of @xmath336 has the structure @xmath344 which has the structure @xmath345 with the product defined in the way evident from @xmath346 , i.e. regarding pairs @xmath347 as matrix indices . because @xmath341 contains a derivative of the square of the laplacian , @xmath348 is indeed explicitly given as the derivative of a matrix @xmath250 .
the solution of @xmath346 is @xmath349\ ; w(0)\big)^{-1 } .\ ] ] that integrating the rg equation leads to such a resummation was , together with the self energy equation , crucial for accessing the symmetry broken regime with a fermionic flow in @xcite .
the above structure suggests that this can also be done using the dynamic wick ordered hierarchy developed here . in this example ,
one resums in the particle particle channel , so @xmath350 is simply @xmath351 where @xmath352 denotes the non charge invariant propagator with @xmath195 given by the matrix in ( 37 ) , ( 38 ) of @xcite , multiplied by the infrared cutoff function .
however , there is a crucial difference to the 1pi scheme , which i now discuss at the example of the flow of superconductor done in @xcite . by definition , the wick ordering propagators @xmath352 are supported at small scales below @xmath353 , so @xmath354 .
thus @xmath355 , and @xmath356 gives for @xmath100 @xmath357^{-1 } .\ ] ] that is , the solution of the channel rg equation @xmath356 does not lead to the final , but the _ initial _ value of the propagator in the solution of the four point function .
hence , in the solution of the bcs model given by this truncation , it is the initial gap that appears in the solution for the four point function .
this seeming contradiction ( when viewed from the perspective of @xcite , where the initial gap is put in as a symmetry breaking parameter that is subsequently sent to zero ) is resolved when one remembers that in the wick ordering scheme one also has to wick order the initial interaction . the requirement that the initial interaction is wick ordered _ but has no quadratic part in the fields _ leads to the hartree
fock self
consistency equation . in the reduced bcs model ,
this equation yields the exact solution for the superconducting gap .
that is , in the wick ordered dynrg , the initial wick ordering of the bare interaction already gives the exact solution of the model , so the gap is constant and equal to the initial one .
the flow of @xmath336 then gives the exact vertex .
the two loop contributions to @xmath208 , as well as the higher contributions to @xmath336 , all vanish in the thermodynamic limit , as explained in the appendix of @xcite . in this section ,
i discuss a discrete analogue of the flow differential equation , and show that the differential flow with a dynamically changing propagator can be obtained as a limit of this discrete equation .
the successive changes of the propagator in the discrete scheme take the form of a continued fraction expansion for operators .
this approach gives a specific way to check the existence of the solution to the equation @xmath200 , which is fundamental for the flow developed here .
a full mathematical proof of existence is rather nontrivial , but the following gives a clear prescription how to proceed and what to check .
discrete flows are interesting by themselves since they are used in the mathematical studies in this field .
consider again the effective action @xmath358 for a system of fermions with propagator @xmath145 and interaction @xmath37 , defined by @xmath359 @xmath358 is of the form @xmath360 where @xmath361 is independent of @xmath189 , @xmath362 defines a quadratic form that gives the quadratic term of @xmath1 in @xmath189 , and @xmath363 contains all the higher powers of the fields . by successively splitting the covariance in two pieces , integrating over fluctuations , and shifting quadratic parts into the measure , i will now recursively construct a sequence of propagators and interactions @xmath364 with the property that the quadratic part of @xmath365 vanishes .
assume that @xmath75 has no constant and quadratic parts .
split @xmath366 , then @xmath367 by @xmath368 , @xmath369 splits into a constant term @xmath370 , a quadratic term @xmath371 , and a term @xmath372 .
the gaussian measure is @xmath373 , with the normalization factor @xmath374 for the gaussian measure being a power of determinant or pfaffian of @xmath375 , so the quadratic term given by @xmath376 can be shifted into the measure , and get @xmath377 with the propagator @xmath378 which now includes the corrections to the selfenergy given by @xmath376 .
note that @xmath376 is not the proper selfenergy @xmath379 but instead the connected amputated two point function .
the relation to the selfenergy @xmath379 coming from the integration over @xmath380 is @xmath381 if one takes the limit of a strict support condition for @xmath382 , @xmath383 outside the support of @xmath382 , i.e. at low energy scales . in writing @xmath384 ,
i have assumed that the rhs is invertible .
this condition will be further discussed below .
an obvious variant of this formula holds if one decides to put only part of @xmath376 ( e.g. the part that shifts the fermi surface ) into the propagator . in the convolution integral on the right hand side of @xmath385
, @xmath386 appears instead of @xmath189 , so that @xmath387 with @xmath388
. completing the square in the gaussian measure and using the identity @xmath389 , gives @xmath390 iteration of this identity by splitting @xmath391 and proceeding as above gives @xmath392 with recursively determined propagators @xmath393 , interactions @xmath365 and amputation operators @xmath394 consequently , @xmath395 with @xmath396 if the splitting in @xmath397 and @xmath398 is chosen such that @xmath399 as @xmath78 , and if @xmath400 , @xmath365 , @xmath401 , and @xmath402 converge to limits @xmath142 , @xmath37 , @xmath144 , and @xmath1 in that limit , @xmath403 if at each iteration step the full @xmath404 was put into the measure to define @xmath405 , @xmath1 is the full propagator of the model , and @xmath37 generates the connected @xmath260point functions of the model with @xmath406 , with amputation given by @xmath144 .
the interpretation at a finite @xmath77 is similar , except that there still remains an effective interaction for the degrees of freedom that have not yet been integrated over . in the following ,
i discuss two specific prescriptions for the splitting , which cover most of the applications , and derive the corresponding formulas for @xmath401 , @xmath393 , and @xmath402 .
assume that the original covariance @xmath138 is given in terms of an invertible @xmath23 as @xmath407 , and that @xmath23 commutes with its adjoint @xmath408 .
the operator @xmath409 is hermitian and positive and therefore has a spectrum that is a subset of @xmath410 . for @xmath411 ,
let @xmath412 be a @xmath101partition of unity on @xmath413 , where @xmath414 is decreasing and @xmath415 is increasing and both are strictly positive functions .
we also require that @xmath416 the idea being that the @xmath414 provide lower and lower cutoffs as @xmath77 increases .
a possible choice is @xmath417 and @xmath418 , but the details of this choice do not matter for the moment .
the operator @xmath419 , defined via the spectral representation of @xmath420 , is a positive operator that cuts off the parts of the spectrum outside the support of @xmath414 .
it is not a projection because we chose a continuous partition of unity instead of a step function . with this choice of @xmath421 as pointwise
strictly positive , @xmath422 is even invertible . given @xmath23 and the interaction @xmath75 , the sequence is constructed as follows .
set @xmath423 then @xmath424 .
calculate @xmath425 and @xmath426 , and define @xmath427 if @xmath428 is invertible , set @xmath429 . then @xmath430 .
in general , for @xmath411 , and given @xmath365 and an invertible @xmath431 , set @xmath432 and @xmath433 . determine @xmath434 and @xmath435 , and set @xmath436 if @xmath437 is invertible , the iteration can be continued ; otherwise , the iteration stops at @xmath77 .
the thus defined @xmath393 and @xmath397 satisfy @xmath438 so that @xmath439 the condition that @xmath431 is invertible has to be checked in every step because it is , in general , nontrivial , and its failure can mean that there is some instability .
we discuss this in more detail in the application to the fermi surface deformation below .
there is another , less obvious , property of this scheme : in the fluctuation propagator @xmath440 the difference of cutoff functions does not necessarily give a positive operator .
in a bosonic theory , this would already make the fluctuation integral ill defined . for fermions ,
lack of positivity is not a problem for the definition of the fluctuation integral , but numerically , positivity is important . roughly speaking
, the difference of cutoff functions becomes negative if the change of the propagator is so strong that one starts integrating backwards .
one can modify the cutoff function by replacing @xmath177 by @xmath441 , where @xmath156 is some projection , whenever it appears as the argument of a cutoff function .
for instance , in the fermion models , one can keep the cutoff function independent of the matsubara frequency variables .
the following alternative scheme is more straightforward but does not admit a continuum limit if one wants to keep a continuous partition of unity .
given a single partition of unity @xmath442 , with @xmath421 decreasing , and a strictly decreasing sequence of positive numbers @xmath443 , let @xmath444 and @xmath445 .
set @xmath446 then @xmath447 , and both have positive cutoff operators .
again , calculate @xmath425 and @xmath426 , and set @xmath448 . if @xmath428 is invertible , set @xmath449 .
we now proceed in exactly the same way , by splitting @xmath450 and so on .
this builds up a product of cutoff functions , which remains positive because each factor is positive .
more precisely , for @xmath411 , and given @xmath451 , let @xmath452 and set @xmath453 then both @xmath397 and @xmath454 have a positive cutoff factor . after calculation of @xmath404 , set @xmath455 and , if @xmath437 is invertible ( which is the case if @xmath404 is small enough ) , @xmath456 .
again , this sequence satisfies @xmath457 . both of the above schemes can be described as follows .
we have a sequence of invertible operators @xmath431 and positive operators @xmath414 , such that @xmath458 , @xmath459 , and @xmath460 , @xmath461 , and @xmath462 , with @xmath404 given by the fluctuation integral with propagator @xmath433 and interaction @xmath365 .
iteration gives @xmath463 this equation has an easy interpretation : the @xmath358 on the rhs contains a fluctuation integral over fields with propagator @xmath393 and interaction @xmath365 , and the external field is amputated with @xmath431 instead of @xmath23 , as it should be , since @xmath393 is @xmath464 times a cutoff function .
the constant and quadratic terms are the contribution to the free energy density and the full propagator .
the generating functional is thus given by @xmath465 where @xmath466 is the field
independent term and @xmath467 is the contribution to the full propagator up to step @xmath77 in the rg iteration .
i now use the `` difference cutoff '' scheme of section [ diffrgsect ] to obtain a continuum limit of the rg sequence .
let @xmath468 be an initial energy scale , let @xmath469 and choose @xmath470 where @xmath471 is a fixed partition of unity on @xmath413 with strictly positive functions @xmath421 and @xmath472 . since there is a one to
one relation between @xmath77 and @xmath473 and we want to take a continuum limit , we now label the sequence by @xmath473 instead of @xmath77 , and denote @xmath474 , @xmath475 , etc . since in the limit @xmath476 nothing is integrated over
, @xmath477 is of order @xmath478 : @xmath479 because @xmath480 , the fluctuation propagator at @xmath481 is , by @xmath482 , @xmath483 thus , up to terms that vanish as @xmath476 , @xmath484 is given by @xmath485 note that the @xmath486derivative acts only on the cutoff function , but not on the inverse in front .
the differential equation for @xmath487 and the other functions is obtained by doing the fluctuation integral to @xmath488 . using @xmath248 , recalling that @xmath489 , and dropping terms of order @xmath490 , i get @xmath491 thus @xmath492 hence the increment
is simply given by the right hand side of polchinski s equation .
the main difference here is , however , that @xmath493 has no quadratic part .
writing @xmath494 , extracting the quadratic part of @xmath358 , and taking the limit @xmath476 , gives the differential flow equation as follows .
the change of the inverse propagator is given by @xmath495 where @xmath496 is determined by @xmath497 the interaction part obeys @xmath498 these are the equations obtained directly in the continuum in section [ rgdesec ] ( with the change of variables @xmath499 ) .
the discrete scheme also provides insights about the choice of wick ordering in the limit of a differential equation , and it is of independent interest . the point where things start to differ from the non wick ordered scheme is the definition of @xmath500 . instead of @xmath384 , @xmath500
is now defined such that the integral in @xmath501 is rewritten in terms of a wick ordered interaction .
more precisely , first set @xmath502 in @xmath501 and require that @xmath503 where @xmath504 denotes wick ordering of @xmath505 , as in @xcite , and @xmath506 i.e. @xmath507 does not have constant or quadratic parts in @xmath189 .
this last condition leads to a self consistency equation for @xmath500 whose solution differs from @xmath384 . by definition of the gaussian measure
, this means that @xmath508 @xmath509 is obtained by using @xmath510 to rewrite @xmath511 since @xmath512 projects out all terms constant or quadratic in the fields , this gives @xmath513 and @xmath514 where for a function of the fields @xmath515 , @xmath516is the operator in the quadratic form given by @xmath517 , i.e. @xmath518 .
equation @xmath519 is the self consistency equation for @xmath500 .
it takes the form of a generalized hartree fock
type equation because the action of @xmath520 is to create self contractions of the vertex given by @xmath521 with a @xmath500 propagator , and this feeds into @xmath522 in the way that is standard in self consistency equations of mean field type .
now that @xmath509 and @xmath500 are given , one obtains the rewriting of the integral in @xmath501 for @xmath523 by standard shifting of gaussians and redefinitions of the amputation term that acts on the external field @xmath524 .
in this section , i briefly discuss the fermi surface flow in @xmath525 in the scheme without wick ordering . a discrete fermi surface flow with wick ordering is defined and controlled mathematically for @xmath526 in @xcite .
this is one of the examples that motivated the method .
although different methods , such as the @xmath89flow of @xcite have been used to calculate rg flows , and other regularizations , such as a cutoff on the frequencies only , are possible , the momentum space flow remains the only one where precise mathematical estimates for power counting of all green functions , not just the first few truncations , have been given @xcite .
let the covariance and interaction be those of the standard many
fermion system with short range interaction @xmath75 , as described in the introduction , with @xmath78 .
( for more details , see @xcite , chapter 4 ) . in fourier space ,
the covariance is a multiplication operator , namely the inverse of @xmath527 times the unit matrix in spin space . here
@xmath71 is the dispersion relation for free particles , given by the hopping amplitudes of the model . by standard ( formal ) diagrammatics , the full propagator is the inverse of @xmath528 , with @xmath529 the fermion self energy .
one natural choice for would then be to take @xmath177 as the full propagator , with @xmath529 understood to depend on @xmath98 ( but without a cutoff function ) . in the following ,
i shall discuss another choice , where only the frequency zero part of the selfenergy is put into the denominator .
this then implies that @xmath49 still has a quadratic part , but one which vanishes at zero frequency . in cases where the field renormalization is a marginally relevant parameter
, one can also put in the linear part of @xmath530 in @xmath531 .
thus the freedom in chosing @xmath208 and @xmath177 allows to adapt to the situation at hand , while keeping the denominator as simple as possible .
so , let @xmath532 in terms of the scale dependent self
energy @xmath533 , @xmath534 , so that the difference @xmath535 vanishes at @xmath536 .
let @xmath537 be a decreasing cutoff function with @xmath538 for @xmath539 and @xmath540 for @xmath106
. set @xmath541 ( another choice would be to use a cutoff on the frequencies as well : @xmath542 . with this choice , @xmath162 still has a quadratic part , which is of the form @xmath543 with @xmath544 ( here @xmath545 , and @xmath546 @xmath547 with @xmath548 the set of fermionic matsubara frequencies ) .
thus with this choice of moving only frequency independent parts of the selfenergy into the denominator , there remain some quadratic terms in the action .
but by construction , these terms vanish on the fermi surface ( see @xmath549 ) , and hence do not cause divergences .
the choice @xmath550 leads to an adaptive scale decomposition because @xmath209 is supported in the neighbourhood @xmath551 of the set @xmath552 which one may regard as a flowing fermi surface . in the standard , nonadaptive scale decomposition ,
the cutoff function is taken as @xmath553 , whose support shrinks to the free fermi surface as @xmath100 .
consequently , the nonadaptive scale decomposition can be used only if a counterterm is included @xcite or if the fermi surface shift is ignored , as is sometimes the case in approximate studies . in the present case , the terms in @xmath554
are @xmath555 equation @xmath200 can now be rewritten as @xmath556 with @xmath557 the fluctuation propagator @xmath558 contains a term proportional to @xmath559 , because of the adaptive nature of the scale decomposition given by this choice of @xmath209 . to discuss the mathematical properties of the resulting equations , it is useful to introduce the `` tadpole '' operator . for a function
@xmath560 it is the linear operator @xmath561 acting on functions of @xmath562 . with @xmath563 and @xmath564 @xmath565
becomes @xmath566 the operator @xmath567 depends nonlinearly on @xmath568 , but it is linear in @xmath559 . thus already the differential equation for the change of @xmath568 involves an operator inversion , due to the additional term in the fluctuation propagator in @xmath569 , hence coming from the adapting scale decomposition .
the solution strategby involves the following steps . 1 .
show that @xmath570 has a bounded inverse on some open set of @xmath49 and @xmath568 .
this is essentially a restriction on the size of @xmath571 it will fail if @xmath571 gets so large that its backreaction on @xmath568 becomes of order 1 .
a sufficent condition is @xmath572 in a suitable norm .
this allows to rewrite @xmath573 as @xmath574 because @xmath567 depends on @xmath568 , it is necessary to find a set of @xmath568 on which @xmath575 uniformly .
the existence of this set already requires conditions on @xmath71 ; the set is then essentially a ball around @xmath71 .
the general theory of ordinary differential equations then implies existence and uniqueness of the solution for all @xmath98 in some bounded interval , provided that @xmath576 is lipschitz . proving that this condition is satisfied is again nontrivial , in fact it is closely related to questions of possible symmetry breaking .
3 . to get global existence (
i.e. for all @xmath98 , one needs to prove that the flow actually never leaves the ball around @xmath71 .
4 . in steps 13 ,
@xmath49 was still regarded as fixed ; in the true system of equations , @xmath571 depends on @xmath98 and the rhs of the equation for @xmath577 depends on @xmath568 as well .
thus one needs to show that @xmath571 stays bounded .
there is a similar set of steps for the discrete fermi surface flow .
steps 13 are already nontrivial , but the hard part is including 4 , because it requires dealing with the hierarchy as a whole . for the differential equation ,
an all order proof that steps 14 work can be given along the lines of @xcite .
a nonperturbative proof , i.e. without truncation on the hierarchy , remains an open problem , mainly because to this day there are no good nonperturbative bounds for the continous rg equation @xcite , which would be needed in step 4 .
up to now , this step requires an integration over short scale intervals and corresponding estimates on discrete rg transformations @xcite .
such a proof is given in the wick ordered approach , for two dimensional fermi systems , in @xcite .
in particular , the boundedness of @xmath571 necessary in step 4 can be shown in the temperature range of the fermi liquid criterion of @xcite .
these mathematical points also correspond to physically natural problems .
the question of the existence of the inverse of @xmath578 is , in physical terms , basic to the consistency of the method itself : the rg flow is parametrized by energy scales on the ( changing ) kinetic part of the action , i.e. the part quadratic in the fields . the nonquadratic interaction part is not taken into account when labelling the scales .
however , it influences the change @xmath559 of the quadratic part , and the equations become inconsistent when this change becomes larger than the maximal kinetic energy scale itself . note that it is this condition that is important in the flow ; the typical interaction energy , measured by the coupling constant , can be much larger than the scale @xmath302 , and this situation indeed occurs at low scales in all nontrivial models because @xmath579 for @xmath100 .
it is a nontrivial fact of fermionic field theory that the method can be controlled by convergent expansions even in that situation @xcite .
the invertibility of @xmath195 at all scales @xmath98 is related to this ; the @xmath98dependent fermi surface must never leave the region where @xmath580 , otherwise an inconsistency manifests itself as divergences in the equation .
this condition can be satisfied because power counting implies that the selfenergy at scale @xmath98 is of order @xmath302 times the coupling constant ( in fact , improved power counting @xcite gives a better @xmath98behaviour of the bound , which is needed to control derivatvies of the selfenergy in the flow ) .
the counterterms used in @xcite can be obtained from the adaptive flow : if @xmath581 exists , the difference @xmath582 can be taken as the counterterm @xmath142 of @xcite .
this can be used to avoid solving the inversion equation @xcite .
however , the question whether there is a one to one relation between the free and the interacting dispersion function and fermi surface , which is answered by the inversion theorem , is not solved this way .
the need to invert @xmath583 for a tadpole operator @xmath567 also arises in the 1pi scheme when one reexpresses the single scale propagator @xmath144 in terms of the full propagator @xmath584 in the equation for the self energy @xmath585 @xcite . with the tadpole operator defined in a natural way , the equation for the selfenergy in the 1pi hierarchy becomes @xmath586 where @xmath587 is the 1-irreducible two
particle vertex and @xmath1 is the full green function .
this equation can be rewritten as @xmath588 more generally , the occurrence of such operators can be expected in any scheme where the propagator changes , be it with an adaptive scale decomposition , as done here , or without ( as in the usual 1pi hierarchy ) .
the flow discussed here can be adapted to a flow where the density , not the chemical potential @xmath589 , is kept fixed .
this variant was used in the appendix of @xcite to calculate the flow of the fermi surface at fixed density in the hubbard model .
in this section , i study the flow equations for a very simple example , namely the integral over one complex variable , corresponding to the partition function of a zero dimensional complex bosonic field , @xmath590 with @xmath591 , re @xmath592 ( for simplicity , i take @xmath593 real in the calculations below . ) , and @xmath594 a complex source field . for @xmath595 , @xmath596 .
thus it is simply a two dimensional integral .
this analysis is motivated by @xcite , where it was observed that the 1pi rg equations give an accurate solution for a function similar to @xmath597 even for large @xmath598 .
this is unexpected at first sight because the diagrammatic approach to these equations may suggest that they are useful only for small @xmath598 .
moreover , it is interesting because only the combination of rg and irreducibility seemed to induce the good behaviour at large couplings , namely ( 1 ) in contrast to the 1pi scheme , the other schemes studied in @xcite , namely polchinski s original hierarchy and the wick ordered hierarchy of @xcite , were accurate only at small @xmath598 , as one would expect from small coupling schemes and ( 2 ) perturbation theory for the 1pi vertices @xmath599 fails as badly as in the other schemes when the coupling constant is not small .
thus it is not merely irreducibility , but a different feature of the 1pi rg hierarchy , that makes for the difference
. a closer look at the equations reveals a first crucial feature : the 1pi equation contains the selfenergy @xmath600 in the denominator .
when @xmath593 gets large , @xmath600 becomes of order @xmath601 after a very short flow time already , and ( because the signs conspire well in this toy model ) the factor @xmath593 in the numerator is balanced by a factor @xmath593 in the denominator , which prevents the right hand side of the flow equation from getting very large .
this suggests that a dynamic adjustment of the propagator , such as discussed in this paper , improves the behaviour of the other hierarchies as well .
this is the case , but the above explanation is not the full story , as we shall see in the following .
although we have written @xmath40 as a function of @xmath594 , it is understood that it also depends on @xmath602 . indeed , by the @xmath603 symmetry of the measure it depends only on @xmath604 .
a gaussian transformation gives the representation @xmath605 from which analyticity in @xmath606 is obvious .
write @xmath607 rescaling the integration variable in @xmath608 , one gets @xmath609 where @xmath610 is analytic in @xmath611 for @xmath612 .
explicitly , @xmath613 this convergent expansion gives the behaviour at large @xmath593 as @xmath614 and @xmath615 with @xmath616 thus for large @xmath593 , the @xmath617 moments of the integral , which play the role of the @xmath259point functions in this toy model , become small .
in the following i discuss to what extent this can be reproduced by the rg with propagator adjustment .
the usual shift gives @xmath618 with the notations @xmath619 and @xmath620 .
as in the general case , write @xmath621 and impose @xmath622 , @xmath623 , @xmath624 , @xmath625 , @xmath626 , as well as @xmath627 .
the rg equations are derived using straightforward differentiation , in analogy to the general case .
expand @xmath628 and for @xmath325 @xmath629 the initial interaction @xmath630 has no constant and quadratic part .
we now impose that the function @xmath631 has the same properties . by requiring @xmath632 and @xmath633 .
this gives equations for @xmath562 and @xmath3 : @xmath634 finally , we fix the flow by setting @xmath635 where @xmath636 is to be determined and @xmath637 a given function of @xmath98 that decreases from @xmath638 to @xmath639 ( here we take @xmath640 as the final value of the flow parameter . in this simple example
, we do not need to make @xmath209 depend on @xmath641 ) .
we require @xmath641 and @xmath209 to be nonnegative to have the same for @xmath642 , so that the gaussian measure is really a measure .
now set @xmath643 then @xmath644 .
note that @xmath645 is negative because @xmath257 is decreasing ; the sign of @xmath645 is , however , unimportant because it never appears in a measure , only in a laplacian . with our initial condition
, @xmath646 starts out positive ; @xmath646 decreases and @xmath641 increases in the flow .
the system of equations reads @xmath647 the summation in @xmath648 now excludes @xmath649 and @xmath650 because @xmath651 .
the equation for @xmath641 can be rewritten as @xmath652 the `` amputation '' factor @xmath653 is now determined by the analogue of condition @xmath654 : putting @xmath655 and inserting @xmath656 and @xmath657 gives @xmath658 , thus , by the initial condition , @xmath659 . moreover
, a glance at @xmath660 and @xmath661 reveals that @xmath662 .
thus , once @xmath663 is determined , @xmath664 in accordance with the general results of section [ diagrammar ] . there are still infinitely many equations for the @xmath665 .
we now study the accuracy of various truncations as a function of @xmath666 .
suppose we truncate by setting @xmath667 .
then @xmath668 is a constant , and @xmath669 integrates to @xmath670 similarly , the equations for @xmath3 and @xmath671 , which in this truncation read @xmath672 can easily be solved .
because @xmath673 and @xmath674 , the final values @xmath675 , @xmath676 and @xmath677 are @xmath678 , \quad g_1 = \frac{{a_0}}{a_1 ^ 2 } = \frac{{a_0}}{{a_0}^2 + 4 { { \lambda}_0 } } .\ ] ] the thus obtained approximation to @xmath597 is @xmath679 by construction , this approximation is , for small @xmath593 , exact to first order in @xmath593 .
but even at very large @xmath593 @xmath680 that is , the behaviour as a function of @xmath593 comes out correctly as @xmath681 , only the prefactor is too small .
note that @xmath682 vanishes for large @xmath593 , so the absolute error @xmath683 for large @xmath593 .
the difference of the true second moment to the one calculated in this truncation is much bigger : @xmath684 but the qualitative behaviour , namely the vanishing at large @xmath593 , is still reproduced correctly .
the fourth order moment has the right scaling as a function of @xmath593 , but the coefficient is too large .
the usual polchinski hierarchy would not have given even a qualitatively correct behaviour of these coefficients . clearly , the adjustment of the propagator is crucial for large @xmath593 because the end result @xmath685 for the quadratic term is very different from the starting value @xmath686 at large @xmath593 .
however , the adjustment of the propagator alone is not sufficient for obtaining good asymptotics at large @xmath593 because truncations at higher @xmath260 improve the result only for small @xmath593 , but not at large @xmath593 . to see this ,
let us now truncate by setting @xmath687 , so that @xmath688 becomes @xmath689 this corresponds to the usual truncation of the polchinski equation , where the six point function is replaced by the tree contribution ( see figure [ fig3 ] ) . used to obtain the one loop equation for the four point function . in the general case ,
@xmath690 is converted into an integral equation which is then substituted back into @xmath691 , leading to a flow equation for @xmath646 that is nonlocal in the scale parameter @xmath98 . in the toy model , @xmath691 and @xmath690
can be combined to @xmath692 because @xmath693 at small @xmath694 by @xmath688 , @xmath695 with this , @xmath691 becomes @xmath696 and @xmath697 and @xmath669 now become a closed system .
now there is a serious problem : @xmath646 decreases in the flow because @xmath698 is negative , but this decrease never stops : even if @xmath699 , the @xmath593 term in @xmath697 keeps @xmath700 negative , and the flow then goes to negative @xmath646 and becomes unstable .
a numerical analysis of higher truncations shows a partial stabilization , but there is still an instability already at moderately large @xmath593 .
the reason for this behaviour will be discussed in the next subsections .
the above findings for the toy model can be understood easily in more general terms .
the lowest truncation of the hierarchy is well behaved in the toy model because the correction to @xmath701 has a good sign , i.e. @xmath702 remains positive for all @xmath98 because the @xmath703dependent terms get added to it ( in addition , it is reasonably accurate ) .
the higher truncations become unstable at large couplings because the hierarchy is effectively nonlocal in the flow parameter : if one truncates by putting the eight point function to zero , the right hand side of the flow equation for the six point function @xmath704 becomes a tree diagram of four point functions , so that the equation can be integrated , and the equation for the four point function @xmath646 becomes ( assuming @xmath705 ) @xmath706 the graphical representation of this procedure is shown in figure [ fig3 ] . because the entire flow history gets inserted on the right hand side , it does not help if @xmath646 decreases in the flow because there is always the contribution from small @xmath707 where @xmath708 is still close to @xmath593 , hence large .
this is the generalization of the right hand side of @xmath697 , in which the @xmath593 term kept driving the flow to negative @xmath646 even at @xmath699 .
thus the nonlocality in the flow parameter makes it impossible to see in this truncation that the effective coupling constant decreases , _ but remains positive _ , even at large initial coupling .
in contrast , the 1pi equation is local in the flow parameter , and the just described `` infrared asymptotic freedom '' of the toy model at large @xmath593 is seen easily .
thus in conclusion , to use the truncated equations at strong coupling it is necessary that the scheme is stable in the sense that asymptotic freedom is not violated in these truncations even at large couplings . in the light of the above discussion
, one may ask how an adaptive wick ordered scheme would fare in this example , because the wick ordered equation also has the property that it is local in the flow parameter . as shown below
, the wick ordered flow turns out to be well behaved .
in fact , already the initial wick ordering provides a reasonable approximation to the true result at large @xmath593 .
as already discussed in previous sections , wick ordering has to be done self
consistently if one wants to keep the interaction free of quadratic terms .
the initial wick ordering provides , when done self consistently , the solution of the mean field theory ( hartree fock equations ) as a starting point for the flow . in the toy model , self consistent wick ordering means finding @xmath709 such that @xmath710 because @xmath711 and @xmath712 , the analogue of @xmath713 is @xmath714 comparison of the quadratic terms in @xmath715 gives the self consistency condition @xmath716 the integral becomes @xmath717 and the standard shift operations then give @xmath718 in contrast to mean field equations in @xmath57 dimensions , no integrations appear in the self
consistency relation @xmath719 .
it is simply a quadratic equation with solution @xmath720 for small @xmath593 , @xmath721 gives an @xmath722 correction to @xmath701 , and @xmath723 gives the lowest order correction to @xmath724 . for large @xmath593 , @xmath725
originally , the strength of the coupling is given by the `` dimensionless '' ratio @xmath726 .
after the initial wick ordering , this has decreased to @xmath727 thus the `` strong coupling problem '' in this integral is removed already by the initial wick ordering .
motivated by this , one can evaluate the remaining convolution integral in @xmath728 to first order in @xmath593 , to get the approximation @xmath729 at large @xmath593 , inserting @xmath730 gives @xmath731 comparison with @xmath732 and @xmath733 shows that this approximation for @xmath734 differs from the exact value by 8 % , the one for the prefactor of @xmath735 in @xmath685 differs from the exact value by 12 % , and the approximation for the prefactor of @xmath736 in @xmath737 differs from the true value by less than 50 % .
note that these are the prefactors ; the asymptotic behaviour as a function of @xmath593 comes out correctly .
thus a simple initial wick ordering , together with first order perturbation theory , gives a reasonable approximation for @xmath738 even at arbitrarily large @xmath593 .
let us now consider the simplest truncation of the wick ordered flow with this initial condition .
for the toy example , the choice @xmath739 is possible , so we use this to fix the wick ordering covariance .
it has the advantage that @xmath740 is determined without any self
consistency argument , but on the other hand , some tadpole terms remain in the equation . setting @xmath741 with @xmath742 and expanding @xmath743 ,
the rg equations for the wick ordered functions read @xmath744 the initial condition is given by the result of initial wick ordering as ( fixing @xmath745 in the definition of @xmath156 and recalling @xmath721 ) @xmath746 and the function @xmath747 is given by @xmath748 the numerator is the value in this toy model of the two loop sunset diagram which is the only one contributing to @xmath747 in the truncation @xmath749 .
the denominator in the definition of @xmath747 comes from a tadpole term , which is there because @xmath750 .
this is a special case of the term @xmath751 discussed in the previous section .
again , the sign is nice in the toy example for @xmath752 , the denominator is always at least @xmath753 , so this inversion does not introduce singularities . only the equations @xmath754 and @xmath755 are coupled ; once their solution has been obtained , the other two simply follow by integration .
transform to the `` dimensionless '' functions @xmath756 then @xmath757 \ ] ] with @xmath758 and @xmath759 .
given the solution of @xmath760 , @xmath761 is obtained from the solution of @xmath762 equation @xmath760 can be integrated explicitly but it is more instructive simply to note that @xmath763 is decreasing and @xmath764 for all @xmath765 because @xmath766 and the right hand side of @xmath760 is negative and vanishes for @xmath767 ( the last property failed for the polchinski type equation ) .
thus @xmath768 appears only with @xmath769 $ ] . with @xmath770
this implies @xmath771 , hence @xmath772 thus the solution for @xmath763 is bounded above and below by that of the prototypical equation for `` asymptotic freedom '' .
solving this equation and setting @xmath640 gives @xmath773 .
thus there is no instability in the wick ordered scheme . to obtain the coefficient of @xmath736 in @xmath737 , the correct starting values are @xmath770 and @xmath774 . with this , the truncated equations give the approximation @xmath775 which differs from the exact value given in @xmath733 by about 20 % .
the solution for @xmath671 gives @xmath776 .
comparison with @xmath733 shows a deviation of the coefficient of @xmath735 of about @xmath777 . in summary ,
the bad behaviour of the polchinski type scheme at large couplings is absent from the wick ordered adaptive flow .
the standard truncation @xmath778 is stable and reproduces the coefficients of the expansion of the exact result in inverse powers of @xmath593 well , i.e. with errors of order 1 10 % .
of course , the equation can also be used for all ( not only large ) values of @xmath593 , where it has to be evaluated numerically .
the stability properties are independent of @xmath593 .
the numerical values are in good agreement with the exact result .
a similarly good strong coupling behaviour of the wick ordered functions may also be expected in the more general examples discussed in @xcite . in this section
i briefly discuss the question of well definedness of the rg flow in fermionic systems at large couplings .
it has been stated in many places that the rg hierarchy is a diagrammatic method that can not be taken as a starting point at strong coupling .
the true situation is , however , not really so bad : the very strategy of the rg of integrating out degrees of freedom in small steps makes the rg flow well defined also at large values of the initial coupling function .
this is not a special feature of the differential equations but holds also for discrete rg flows , provided the stepsize is chosen small enough . to see this ,
consider a fermionic theory , as defined in the introduction . by integrating the rgde over short intervals , or using a discrete transformation
, one can show that the generating function @xmath779 has the representation @xmath780 where in @xmath781 , the argument @xmath37 appears @xmath782 times and @xmath783 } { \prod\limits}_{t \in t } \delta_c^{(t ) } { \prod\limits}_{q=1}^p { { \cal v}}_q ( \psi_q ) { { { \rbrack\!\rbrack}}}\ ] ] where @xmath784 is the set of all trees on @xmath782 vertices , and @xmath156 is a probability measure ( i.e. positive and normalized , @xmath785 ) , and @xmath786 is a positive definite @xmath787 matrix with diagonal entries equal to @xmath753 .
both @xmath156 and @xmath788 can be written down explicitly , but only the above mentioned properties matter ( see @xcite , where the above formula is derived and explained in more detail ) .
@xmath789 can be thought of as a resummation of the perturbation expansion in terms of trees . because @xmath790 , the sum converges if the contribution of every tree @xmath89 can be bounded by @xmath791 . for fermions
, this can be shown using determinant bounds ( see , e.g. @xcite ) . the crucial point for the present discussion
is that for a short range interaction @xmath37 sufficient conditions for convergence are that the determinant constant @xmath792 of @xmath26 ( see @xcite ) and the decay constant @xmath793 are so small that @xmath794 . in a rg flow with stepsize @xmath478 , the covariance @xmath26 is split into many small parts @xmath795 and in every rg step , the @xmath26 in the above formula is replaced by @xmath795 . because @xmath796 and @xmath797 , the above convergence condition can be satisfied for any @xmath703 by making the rg stepsize @xmath478 small enough .
thus , for small enough @xmath478 the first steps in the rg integration are given by convergent perturbation theory in @xmath703 , and therefore the rg flow can be started at large coupling . the really interesting analytical and physical question is whether a large , repulsive local initial interaction indeed decreases in the course of the flow ( i.e. it gets screened ) , and whether the rate of this decrease is so fast that the flow can be controlled in the regime where the local coupling is not yet small , as is the case in the toy model .
one may indeed expect that a strong on site repulsion gets screened in the course of an rg flow , because even a hard core condition gets softened , thus effectively weaker when one considers larger blocks on the lattice .
namely , if there is a strong on site repulsion , double occupancy of single sites is exponentially suppressed , but hopping into a block of @xmath798 lattice sites , @xmath799 , is still possible because of local fluctuations in the number density , even if the overall density is fixed .
a standard example for a stochastic system with a hopping term and a hard core repulsion is the asymmetric exclusion process .
it was shown in @xcite that averaging in a suitable way effectively allows to _ remove _ the hard core constraint in this model and still get precise asymptotics of the decay of correlations . because the truncations of the correlation hierarchy in this example have the property that they provide successive upper and lower bounds for the exact solution , one can obtain their asymptotic properties in some detail @xcite .
for fermion systems , such monotonicity properties are in general not known , nevertheless averaging , such as provided by the rg , seems a very promising strategy to see screening effects .
in fact , a low order evaluation of the flow at very large scales in the two dimensional hubbard model shows the screening , as well as the generation of the usual antiferromagnetic interaction term @xcite .
the above argument is about the issue of the behaviour of an _ initial _ strong short
range interaction .
the `` flow to strong coupling '' at low scales that has been observed in studies of initially weakly coupled systems is an independent phenomenon ( as discussed in @xcite , such flows may be kept under control for a while by phase space arguments , but eventually , the growth of the coupling function wins ) .
the growth of certain parts of the coupling function in these flows is really due to the emergence of order parameter fluctuations that become more and more long range , hence lead to fermionic interaction vertices that develop singularities in momentum space at a scale where symmetries get broken . in situations
where , e.g. the singular part of the fermionic four point function can be parametrized by an exchange of a boson that becomes massless at a certain scale , approximations that set the momentum of this boson to zero give a coupling constant of order @xmath800 which `` runs to strong coupling '' and diverges at a certain scale . whether the boson can really become massless depends on the situation , in particular the dimensionality of the system
moreover , in systems with long range interactions , the anderson
higgs mechanism may interfere .
however , unless this is the case , the goldstone theorem implies such singularities whenever continuous symmetries are broken .
because such singularities occur only at points , their effect is very different from that of a strong local interaction , which is equally large everywhere in momentum space .
the rg equations developed here allow for a dynamical adjustment of propagators in a convenient and flexible way .
in particular , they allow for countinuously adjusted scale decompositions , which are useful in fermi surface studies , when one wants to avoid the use of counterterms , or fix the density ( instead of the chemical potential ) in rg flows .
a numerical study of the adaptive flow equation discussed here , with fixed density , was done for the two dimensional hubbard model in the appendix of @xcite .
a mathematical study for the case of regular fermi surfaces is in @xcite .
taking the fermi surface deformation into account is important also for the case of singular fermi surfaces , where the question of pinning of the fermi surface at van hove singularities can be addressed using such flows .
this situation is studied in @xcite using counterterms .
the adaptive method is expected to lead to useful results also here . in flows with fermion and boson fields ,
a generalization of the strategy for the adaptive flow becomes obvious : quartic terms in the fermions contribute to the boson propagators , so the dynamical adjustment of the boson propagator allows to take out also certain terms from the fermionic four point function .
a related method has already been used in the 1pi scheme in @xcite ; a straightforward generalization of the method developed here will allow to do it in any scheme .
finally , the role of the propagator adjustment for the strong coupling behaviour of flows was investigated for a very simple toy model .
it was shown that the wick ordered scheme with dynamic propagator adjustment correctly reproduces the asymptotic behaviour for arbitrarily large couplings , and already its simplest truncation gives the coefficients of the leading term in the asymptotic series with an error of order 1 10 % ( depending on the order of the moment ) .
the analysis made clear that the adjustment of propagators is indeed crucial for the success of a scheme at strong coupling , but that in addition , a stability property of the four point flow is necessary .
namely , the rg equations must be such that asymptotic freedom holds for the solution also if the flow is started at a large value of the coupling .
the polchinski scheme does not have the second property , even with an adapted propagator , but the 1pi and wick ordered schemes do .
needless to say , the toy example is very special because there , the signs work out such that a large initial coupling @xmath593 decreases to a very small final value , and the quadratic term @xmath701 increases to a large final value . because only the combination @xmath801 matters , this effectively leads to a very rapid transition to a small coupling situation in the flow , provided the above mentioned stability holds .
this explains the success of the 1pi and wick ordered scheme . in the latter , already the initial wick ordering removes the strong coupling problem right away , if it is done self - consistently
. it would be too much to hope for such benign signs in more general ( and more interesting ) models , but , as discussed in some detail in section [ scgen ] , the flow equations are well defined at strong coupling ( contrary to what has often been stated in the literature ) and there are reasons to be optimistic about screening of initially strong local repulsions in rg flows .
* remarks and acknowledgements . * my research on this topic started in 1998 , but this paper has been held back for a long time for various reasons .
i would like to thank carsten honerkamp and walter metzner for discussions , and the kitp , santa barbara , especially the program _
realistic theories of correlated electron systems _ and the erwin schr " odinger institut , vienna , for hospitality and financial support in various stages of the work . during the last year
, it was also supported by dfg grant sa 1362/11 and nserc of canada .
w. pedra , m. salmhofer , fermi systems in two dimensions and fermi surface flows proceedings of the 14th international congress of mathematical physics , lisbon , 2003 ; on the mathematical theory of fermi liquids at positive temperatures i iii , to appear r. hedden , v. meden , th .
pruschke , and k. schnhammer , j. phys . : condens .
matter * 16 * , 5279 ( 2004 ) , see also v. meden s contribution to the summer school 2004 in brasilia at + http://www.theorie.physik.uni-goettingen.de/ meden / funrg / talk.pdf m. salmhofer , _ perturbative renormalizability of @xmath802 by renormalization group differential equations _ , proceedings of the 2002 hesselberg workshop _ theory of renormalization and regularization _ , + http://wwwthep.physik.uni-mainz.de/@xmath803scheck/hessbg02.html | a class of continuous renormalization group flows with a dynamical adjustment of the propagator is introduced and studied theoretically for fermionic and bosonic quantum field theories . the adjustment allows to include self
energy effects nontrivially in the denominator of the propagator and to adapt the scale decomposition to a moving singularity , and hence to define flows of fermi surfaces in a natural way .
these flows require no counterterms , but the counterterms used in earlier treatments can be constructed using them .
the influence of propagator adjustment on the strong coupling behaviour of flows is examined for a simple example , and some conclusions about the strong coupling behaviour of renormalization group flows are drawn .
[ section ] [ satz]definition [ satz]lemma [ satz]corollary [ satz]remark |
traversable lorentzian wormholes were first considered by morris and thorne @xcite ; because of the interesting unusual properties of these topologically non trivial geometries , in the years after this leading work they have received considerable attention @xcite . a major objection against their actual existence , however , is the necessity of `` exotic '' matter ( not fulfilling the energy conditions ) supporting the wormholes ; within the framework of general relativity this difficulty can not be avoided in the case of compact wormhole geometries .
therefore , wormholes of the thin - shell class @xcite , that is , those which are mathematically constructed by cutting and pasting two geometries , present a particular interest because in such configurations the exotic matter can be restricted to a layer placed at the throat ( see refs . @xcite and references therein ) . in the last three decades , cosmic strings
@xcite have been the object of a thorough study , because of their possible relevance in the process of structure formation in the early universe @xcite , and also because , in principle , they could be detected by gravitational lensing effects @xcite . besides ,
open or closed fundamental strings are the center of present day programs in the search for a unified theory of the fundamental interactions .
then the interest in the gravitational effects of both fundamental and cosmic strings and , in general , in axially symmetric solutions of the equations of gravitation , has been recently renewed ( for instance , see ref .
consequently , cylindrically symmetric wormholes have been studied in the last few years ; see refs .
@xcite . in particular
, a remarkable aspect was noted in ref .
@xcite : cylindrically symmetric wormholes admit two definitions of the throat , and one of them that the geodesics _ restricted to a plane normal to the symmetry axis _ open up can be compatible with the energy conditions even within the framework of general relativity . it has been shown that such cylindrical wormholes satisfying the weak energy condition can not have a flat or local cosmic string asymptotic behavior at both sides of the throat .
wormholes associated with an azimuthal magnetic field and supported by non exotic matter were found ; they are neither symmetric with respect to the throat nor flat or local string like at infinity ( see ref . @xcite and references therein ) .
the possibility of cylindrical thin - shell wormholes supported by a layer of positive energy density placed at the throat was proved in ref .
@xcite within the framework of einstein s gravity .
besides , the linearized stability of generic thin - shell wormholes was studied in ref .
@xcite .
cylindrical spacetime solutions with non vanishing cosmological constant @xmath0 were found in refs .
@xcite ; they are known as the linet
tian metric or the levi - civita metric with a cosmological constant , and are associated to the exterior of a cosmic string .
several interesting features of these geometries were analyzed in refs .
@xcite . a case of particular interest is the one corresponding to a negative cosmological constant , which is asymptotically anti - de sitter .
in recent years , anti - de sitter spacetimes have been of interest within the context of supergravity and string theories ; in particular , they play an important role in m - theory .
the ads / cft correspondence conjecture provides a link between gravity and conformal field theory .
the linet
tian metric with @xmath1 has the form @xcite : @xmath2 , \label{ds2}\ ] ] where @xmath3 @xmath4 @xmath5 , @xmath6 , @xmath7 , and @xmath8 .
the constant @xmath9 is related to the angle defect , and the parameter @xmath10 is related , but not equal , to the linear mass density of the source .
when @xmath11 , we have that @xmath12 and one recovers the levi - civita spacetime @xcite . for the parameter @xmath10 , the physically relevant range @xcite is @xmath13 .
the solution ( [ ds2 ] ) is singular in the axis @xmath14 , except for the cases @xmath15 and @xmath16 .
when @xmath15 the geometry corresponds to the so - called non uniform anti - de sitter universe , which is one of the two known vacuum solutions of the einstein equations with a negative cosmological constant @xcite . if @xmath16 , it has been shown @xcite that with a suitable coordinate change the spacetime corresponds to a black string .
cylindrical thin - shell wormholes were constructed from the linet
tian metric with a positive cosmological constant @xcite , for which the stability was recently studied . in the present article
, we consider the mathematical construction and characterization of cylindrically symmetric thin - shell wormholes with a negative cosmological constant ; in particular , a detailed analysis of their stability under perturbations preserving the symmetry is performed .
two different approaches for the stability analysis are adopted : in the first one , we assume that the equations of state for the static configuration are preserved along the evolution of the perturbation , and in the second one , we assume a linearized equation of state relating each pressure with the surface energy density . we compare the results with those corresponding to the thin - shell wormholes obtained from the levi - civita metric . we adopt units such that @xmath17 .
the mathematical construction of generic thin - shell wormholes with cylindrical symmetry ( see ref .
@xcite ) starts from a metric which , in coordinates @xmath18 , has the form @xmath19 here @xmath20 , @xmath21 , @xmath22 , and @xmath23 are positive functions of the radial coordinate only .
two copies @xmath24 of the region with @xmath25 are taken , and then joined at the hypersurface @xmath26 .
the resulting new manifold @xmath27 is geodesically complete .
if a flare - out condition is satisfied , that is , if the geometry opens up at the radius @xmath28 , the construction determines a cylindrically symmetric thin - shell wormhole with two infinite regions connected by a throat at @xmath29 .
we apply the standard darmois - israel formalism @xcite . at the throat
we adopt coordinates @xmath30 , with @xmath31 the proper time on the shell .
the radius of the throat is , in general , a function of @xmath31 , that is , @xmath32 ; in this case , the shell is defined by @xmath33 . the energy density and pressures of matter on the shell are related with the extrinsic curvature @xmath34 at the two sides of it .
the relation is given by the lanczos equations @xmath35+[k]g_{ij}=8\pi s_{ij } , \label{e7}\ ] ] where @xmath36\equiv { k_{ij}}^+ - { k_{ij}}^-$ ] , @xmath37=g^{ij}[k_{ij}]$ ] is the trace of the extrinsic curvature tensor and @xmath38 is the surface stress - energy tensor .
if we define the unit normals @xmath39 ( so that @xmath40 ) to the surface @xmath41 as @xmath42 then the extrinsic curvature at each side of the shell is given by @xmath43 working in the orthonormal basis @xmath44 , @xmath45 , @xmath46 , for which @xmath47 , we obtain @xmath48\dot{a}^2}{2a(a)\sqrt{b(a ) } \sqrt{1+b(a)\dot{a}^2 } } , \label{e4}\ ] ] @xmath49 and @xmath50 where the dot and the prime stand for @xmath51 and @xmath52 respectively .
therefore the surface energy density @xmath53 and the pressures @xmath54 and @xmath55 for the matter on the shell are given by @xmath56 , \label{e8}\ ] ] @xmath57 \dot{a}^2 + \frac{a'(a)}{a(a ) } + \frac{d'(a)}{d(a ) } \right\ } , \label{e9}\ ] ] @xmath58 \dot{a}^2 + \frac{a'(a)}{a(a ) } + \frac{c'(a)}{c(a ) } \right\ } .
\label{e10}\ ] ] the definition of the wormhole throat usually applied to compact configurations states that it is a _ minimal area surface_. in the cylindrical case , the area function @xmath59 should then increase at both sides of the throat , and this implies that @xmath60 .
we can call it the _ areal _ flare - out condition . if @xmath61 , from eq .
( [ e8 ] ) it follows that the surface energy density is negative , so the matter at the throat is _
exotic_. however , bronnikov and lemos @xcite recently introduced a new throat definition which only requires that the circular radius function @xmath62 has a minimum at the throat ; thus , we can call this one the _ radial _ flare - out condition .
this definition implies @xmath63 , imposing no condition on the sign of @xmath64 ; in principle , depending on the metric considered , this could make possible a _ positive _ energy density . in the case of the wormholes constructed from the spacetime given by eq .
( [ ds2 ] ) , by using the equations above , we find the energy density and the pressures @xmath65}{12 \pi \alpha ( k ) \sinh \left(\sqrt{3|\lambda | } a\right)},\ ] ] @xmath66 ( 1+\dot{a}^2)}{3 \alpha ( k ) \sinh \left(\sqrt{3|\lambda | } a\right)}\right\},\ ] ] @xmath67 ( 1+\dot{a}^2)}{3 \alpha ( k ) \sinh \left(\sqrt{3|\lambda | } a\right)}\right\}.\ ] ] note that there is no dependence with the parameter @xmath9 of the metric .
it is not difficult to see that the energy density at the shell is always negative for all the values of the parameter @xmath10 ( in the range @xmath68 adopted in this work ) and for any throat radius @xmath28 .
this result is valid even if the radial flare - out condition is used .
so the weak energy condition ( which requires @xmath69 and @xmath70 ) can not be fulfilled in our construction , and the matter at the throat is exotic . for comparison , let us analyze the case of wormholes constructed from the levi - civita metric ( @xmath71 ) , for which we obtain that the energy density and the pressures are given by @xmath72 @xmath73,\ ] ] @xmath74.\ ] ] the energy density at the shell is always negative for all the values of @xmath10 and @xmath28 , even if the radial flare - out condition is adopted .
then , the weak energy condition is not satisfied , and the matter at the throat is also exotic .
the static values of the energy density and the pressures are obtained by putting @xmath75 and @xmath76 in eqs .
( [ e8 ] ) , ( [ e9 ] ) , and ( [ e10 ] ) : @xmath77 , \label{e12}\ ] ] @xmath78 , \label{e13}\ ] ] @xmath79 , \label{e14}\ ] ] where @xmath80 is the constant throat radius .
then , the static wormhole associated to the background geometry given by eq .
( [ ds2 ] ) is supported by a shell at the throat with the surface energy density @xmath81}{12 \pi \alpha ( k ) \sinh \left(\sqrt{3|\lambda | } a_0\right)},\ ] ] and the pressures @xmath82 } { 12 \pi \alpha ( k ) \sinh
\left(\sqrt{3|\lambda | } a_0\right)},\ ] ] @xmath83 } { 12 \pi \alpha ( k ) \sinh
\left(\sqrt{3|\lambda | } a_0\right)}.\ ] ] in the @xmath71 case , we find that @xmath84 @xmath85 @xmath86 in what follows , we adopt two possible formalisms for the stability analysis of the static configurations , previously presented in a general form in refs . @xcite and @xcite .
we consider small perturbations , so we assume that the geometry outside the throat remains static . a possible approach to the study of the mechanical stability relies on the fact that we are interested in small perturbations starting from a static solution , so that the evolution of the shell matter can be considered as a succession of static states @xcite .
we summarize here our general analysis for radial perturbations introduced in ref .
@xcite and generalized in ref .
@xcite , in which we fix the form of the equations of state . by using eq .
( [ e12 ] ) , we see that eqs .
( [ e13 ] ) and ( [ e14 ] ) can be recast in the form @xmath87}{a(a_0)[c(a_0)d'(a_0)+c'(a_0)d(a_0 ) ] } \sigma ( a_0 ) , \label{e15}\ ] ] @xmath88}{a(a_0)[c(a_0)d'(a_0)+c'(a_0)d(a_0 ) ] } \sigma ( a_0 ) .
\label{e16}\ ] ] then , the functions @xmath89 , @xmath90 , and @xmath91 determine the equations of state @xmath92 and @xmath93 of the exotic matter on the shell .
hence , we can assume that the equations of state for the dynamic case have the same form as in the static one , i.e. that they do not depend on the derivatives of @xmath94 ; then @xmath92 and @xmath93 are given by eqs .
( [ e15 ] ) and ( [ e16 ] ) with @xmath28 instead of @xmath80 .
therefore , as we have shown in our previous works , even in the most general case of a wormhole connecting two geometries of the form ( [ e1 ] ) a simple second order differential equation for @xmath95 is obtained : @xmath96 now , in our case for both @xmath1 and @xmath97 , the metric has @xmath98 , so that @xmath99 , @xmath100 and then @xmath101 so , when given in terms of the proper time on the shell , the radius of the wormhole throat can only undergo a monotonous evolution with constant speed .
this result is a particular case of the general behavior found in our previous articles .
let us review the formalism introduced in ref .
@xcite , which was previously developed for the spherically symmetric case in refs .
@xcite . from eqs . ( [ e8 ] ) , ( [ e9 ] ) , and ( [ e10 ] )
, we can obtain the conservation equation @xmath102 where @xmath103 and @xmath104 \left [ \frac{c'(a)}{c(a)}+\frac{d'(a)}{d(a)}\right]^{-1 } \right\}. % \label{flu}\ ] ] in the left hand side of eq .
( [ ceq1 ] ) , the first term represents the internal energy change of the throat , and the second and the third terms represent the work done by the internal forces of the throat ; while the right hand side corresponds to a flux . by using that @xmath105 , this conservation equation can be rewritten to give @xmath106'}{\sqrt{c(a ) } } + \mathcal{a } p_z \frac{\left[\sqrt{d(a)}\right]'}{\sqrt{d(a ) } } = \mathcal{a } \mathcal{f}(a ) .
\label{ceq2}\ ] ] the last equation can be formally integrated to obtain @xmath107 if the pressures are known as functions of the energy density . from eq .
( [ e8 ] ) , we can write the equation of motion of the throat in the form @xmath108 with the potential @xmath109^{-2 } \sigma ^2 .
\label{pot}\ ] ] a taylor expansion to second order of the potential @xmath110 around the static solution gives @xmath111 it is easy to see that @xmath112 . taking the derivative of the potential and using the conservation equation ( [ ceq2 ] ) , we have that @xmath113 . using eq .
( [ ceq2 ] ) again and introducing the parameters @xmath114 and @xmath115 , the second derivative of the potential evaluated at @xmath80 reads @xmath116 \omega ( a_0)}{b(a_0)\left[c'(a_0)d(a_0)+c(a_0)d'(a_0)\right ] } , \label{potd2}\ ] ] where @xmath117 \left [ -\frac{a'(a_0)}{a(a_0)}+\frac{a''(a_0)}{a'(a_0)}-\frac{b'(a_0)}{2 b(a_0)}\right ] \nonumber \\ & & -\frac{b'(a_0)}{b(a_0)}-\frac{c'(a_0)}{c(a_0)}+\frac{c''(a_0)}{c'(a_0)}-\frac{d'(a_0)}{d(a_0)}+ \frac{d''(a_0)}{d'(a_0 ) } , \nonumber\end{aligned}\ ] ] and @xmath118 + \frac{d'(a_0)}{d(a_0 ) } \left [ -\frac{b'(a_0)}{2 b(a_0)}-\frac{d'(a_0)}{d(a_0)}+\frac{d''(a_0)}{d'(a_0)}\right].\ ] ] a static solution is stable under radial perturbations if @xmath119 .
so the stability analysis reduces to the study of the sign of the second derivative of the potential .
: linearized stability regions ( in gray ) in the plane @xmath120 , for @xmath121 and several values of @xmath122.,width=453 ] : linearized stability regions ( in gray ) in the plane @xmath120 , for @xmath123 and several values of @xmath122.,width=453 ] ): linearized stability regions ( in gray ) in the plane @xmath120 , for @xmath121 ( left ) and for @xmath123 ( right ) .
the results do not depend on @xmath80 ( see text).,width=453 ] for the static wormhole constructed from the metric given by eq .
( [ ds2 ] ) , we obtain that @xmath124 } { \alpha ( k ) \left [ 2 \alpha ( k ) \cosh \left ( a_0 \sqrt{3|\lambda | } \right)+\zeta ( k)\right ] \sinh^2\left ( a_0 \sqrt{3|\lambda | } \right ) } \nonumber \\ & & \times \left [ ( \eta _ 1+\eta _ 2 - 1 ) \alpha ( k ) \cosh \left ( a_0 \sqrt{3|\lambda | } \right)+ \eta _ 1 \epsilon ( k ) + \eta _ 2 \gamma(k)+\zeta ( k)\right ] .\end{aligned}\ ] ] the functions in the denominator @xmath125 and @xmath126 are always positive , and the function @xmath127 is also positive for all @xmath10 within the range @xmath128 , so using that @xmath129 we can simplify the condition for @xmath119 in the form @xmath130 \left [ ( \eta _ 1+\eta _ 2 - 1 ) \alpha ( k ) \cosh \left ( a_0 \sqrt{3|\lambda | } \right)+ \eta
_ 1 \epsilon ( k ) + \eta _ 2 \gamma(k)+\zeta ( k)\right ] < 0.\ ] ] we see that this inequality is rather complicated to be interpreted analytically .
so we show the results graphically in figs .
[ f1 ] and [ f2 ] , in the plane @xmath120 , for some relevant values of the other parameters . for given values of @xmath0 , @xmath10 , and @xmath80 , in each plot the stable region ( in gray ) is limited by the straight line @xmath131 \eta _ 1 + \left [ \alpha ( k ) \cosh \left ( a_0 \sqrt{3|\lambda | } \right)+ \gamma(k ) \right ] \eta _ 2 = \alpha ( k ) \cosh \left ( a_0 \sqrt{3|\lambda | } \right)-\zeta ( k),\ ] ] which has a positive slope for small values of @xmath122 ; this slope becomes negative as @xmath122 increases . if the stable region lies above or below this line depends also on the sign of the expression @xmath132 , which is positive for small @xmath10 and becomes negative as @xmath10 grows ( the value where it changes sign is determined by @xmath133 ) .
the plots corresponding to small values of @xmath10 , exemplified by fig .
[ f1 ] for @xmath121 , show that the stable region is above the straight line for small values of @xmath122 and is below this line as @xmath122 grows .
when @xmath10 is large , as displayed in fig .
[ f2 ] for @xmath123 , the stable region is above the limiting line for small @xmath122 , below this line when @xmath122 grows , and then above again . from the figures we see that , for appropriate values of @xmath10 and @xmath122 , stable configurations can be found with both @xmath134 and @xmath135 belonging to the interval @xmath136 $ ] , which is an useful feature if one wants to interpret these parameters as the squared velocity of sound in the corresponding directions . but for the exotic matter at the throat , this interpretation is not mandatory @xcite . in the case of wormholes constructed from the levi - civita geometry
, we find that the second derivative of the potential takes the form @xmath137 where @xmath138 , as stated previously .
the denominator is always positive , so the stability condition @xmath119 reduces to @xmath139 which is independent of the throat radius @xmath80 .
we show the results in the plane @xmath120 for different values of the other parameters in fig .
we see that , for a given value of @xmath10 ( two values : @xmath140 and @xmath141 are shown in the plots ) , there is a stable region above the line @xmath142 it is easy to verify that the slope of this line is positive and a decreasing function of @xmath10 .
configurations with both @xmath134 and @xmath135 within the interval @xmath136 $ ] are possible for all values of @xmath10 in the range adopted .
we have constructed cylindrical thin - shell wormholes with anti - de sitter asymptotics , starting from the linet
tian metric with @xmath143 .
we have characterized the matter at the throat , finding that it can not satisfy the energy conditions for any of the two usual flare - out definitions ; i.e. it is always exotic .
we have also studied the case with @xmath97 , corresponding to wormholes constructed from the levi - civita spacetime , which are also always supported by exotic matter at the throat .
we have performed two different stability analyses under perturbations preserving the symmetry .
if we fix the form of the equations of state by assuming that those corresponding to a dynamic throat maintain the form of the static case , the configurations are unstable , as it was previously shown in the general case @xcite . instead , in the linearized stability analysis , stable configurations are possible for suitable values of the parameters .
when @xmath71 the stability does not depend on @xmath80 , while if @xmath129 it depends on the product @xmath122 . for @xmath71 and @xmath129 , stable configurations with both @xmath144 and @xmath145
are possible , so these parameters can be interpreted as the squared velocity of sound in the corresponding directions . in the @xmath1 case ,
this desirable feature is present for any value of @xmath122 when @xmath10 is small and only for low values of @xmath122 if @xmath10 is large .
this work has been supported by universidad de buenos aires and conicet .
n. sen , ann . phys .
( leipzig ) * 378 * , 365 ( 1924 ) ; k. lanczos , ann .
( leipzig ) * 379 * , 518 ( 1924 ) ; g. darmois , mmorial des sciences mathmatiques , fascicule xxv ( gauthier - villars , paris , 1927 ) , chapter 5 . | in this article we investigate cylindrical thin - shell wormholes which are asymptotically anti - de sitter .
we analyze their stability under perturbations preserving the symmetry by using two different methods .
we compare the results with those corresponding to the wormholes constructed from the levi - civita spacetime .
we find that the configurations always require the presence of exotic matter at the throat , and , in the case of the linearized stability analysis , they can be stable for suitable values of the parameters . + pacs number(s ) : 04.20.-q , 04.20.gz , 04.20.jb + keywords : lorentzian wormholes ; exotic matter |
the galactic chemical evolution is the study of the transformation of gas into stars and the resulting evolution of the chemical composition of a galaxy .
the so called `` simple model '' remains an useful guide for understanding the chemical evolution of galaxies since the pioneering works of schmidt ( 1963 ) , searle & sargent ( 1972 ) , tinsley ( 1974 ) , pagel & patchett ( 1975 ) . in order to derive analytical solutions for the chemical evolution of galaxies
one needs to make several hypotheses : the initial mass function ( imf ) should be considered constant , the lifetime of stars should be neglected ( instantaneous recycling approximation , ira ) , and the complete mixing of chemical elements with the surrounding interstellar medium ( ism ) .
these assumptions , in fact , allow us to have analytical expressions for the metallicity evolution of the galaxies in time and in terms of the gas fraction
. one can find analytical solutions even in presence of gas flows ( infall , outflow , galactic fountains ) as shown by tinsley ( 1980 ) , clayton ( 1988 ) , lacey & fall ( 1985 ) , matteucci & chiosi ( 1983 ) , edmunds ( 1990 ) , recchi et al .
( 2008 ) , spitoni et al .
( 2010 ) , peeples & shankar ( 2011 ) , lilly et al .
( 2013 ) , pipino et al .
( 2014 ) , peng & maiolino ( 2014a ) , recchi & kroupa ( 2014 ) , kudritzki et al .
( 2015 ) .
however , in all mentioned works the solutions are obtained only under specific assumptions regarding the infall / outflow rates .
however , we note that analytical solutions are not able to give a complete description of the chemistry of a galaxy , since they fail in following the evolution of elements created on long time - scales , such as iron and nitrogen .
a satisfactory description of the iron evolution requires detailed numerical models relaxing ira and including the chemical enrichment from type ia sne ( e.g. matteucci & greggio 1986 ) allow one to follow in detail the evolution of single elements .
matteucci & greggio ( 1986 ) showed in detail the effect of the time - delay model , already suggested by tinsley ( 1980 ) and greggio & renzini ( 1983 ) , in particular the effect of a delayed fe production by type ia sne on abundance ratios involving @xmath0-elements ( o , mg , si ) .
analytical solution can be , on the other hand , adopted when studying the evolution of oxygen , created on short timescales and tracing the evolution of the global metallicity , z , of which oxygen is the main component . in this work we start by using the formalism described by matteucci ( 2001 ) , recchi et al .
( 2008 ) , and spitoni et al .
( 2010 ) and we show , for the first time , the analytical solution of the evolution of the metallicity of a galaxy in presence of `` environment '' effects coupled with galactic fountains and primordial infall of gas . in this context for `` environment '' effect we mean the situation where a galaxy suffers , during its evolution , infall of enriched gas from another evolving galactic system , and this gas represents an enriched infall variable in time .
the dynamics of interacting systems have been the subject of many papers concerning numerical simulations ( toomre & toomre 1972 ; barnes & hernquist 1992 ; berentzen et al .
2003 ) and spectrophotometric models ( larson & tinsley 1978 ; kennicutt 1990 ; temporin et al .
2003a , 2003b ) .
large amounts of ism can be removed from the main disks of spiral galaxies by different processes : tides due to the gravitational force of a companion , ram pressure stripping during a near head - on collision between gas - rich galaxies , ram pressure stripping by intracluster gas , and galactic winds driven by supernovae .
smith & struck ( 2001 ) using co signatures observed 11 extragalactic tails and bridges in nine interacting galaxy systems .
recently , smith et al .
( 2010 ) using galaxy evolution explorer ( galex ) ultraviolet telescope studied star formation morphology and stellar populations and in 42 nearby optically selected pre - merger interacting galaxy pairs .
tails and bridges structures are often more prominent relative to the disks in uv images compared to optical maps .
this effect is likely due to enhanced star formation in the tidal features compared to the disks rather than to reduced extinction .
we also refer the reader to the review of boselli & gavazzi ( 2006 ) where a comprehensive description of the environment effects on late - type galaxies in nearby clusters is presented . among them
we recall the tidal interactions among galaxy pairs which act on gas , dust and stars , as well as on dark matter and is depending on the gravitational bounding of the various components ; tidal interaction between galaxies and the cluster potential well , and finally the so - called `` galaxy harassment '' : the evolution of cluster galaxies is governed by the combined effect of multiple high speed galaxy - galaxy close ( @xmath1 50 kpc ) encounters with the interaction with the potential well of the cluster as a whole . as stated by davies et al .
( 2010 ) , galaxy - galaxy interactions are particularly important and striking when the speed of the interaction is well matched to the velocities of the stars and gas .
therefore , small galaxy groups can potentially provide the environment for dramatic gravitational disturbances . beyond the local group , the closest example of this is the environment around m81 , the m81 group .
in fact , in m81 group it is evident that various galaxies are connected by flows .
extended filamentary structures external to the disc of m81 , are clearly seen in emission in all of the herschel bands .
these complex interactions can not be described by the simple tools offered , for example , by recchi et al .
( 2008 ) or similar papers in literature , and with this paper we provide a new set of analytical solutions to take into account this kind of interactions in the framework of analytical chemical evolution models .
moreover , an enriched infall of gas can be originated from the galaxy itself .
in fact , in the galactic fountain models ( shapiro & field 1976 , houck & bregman 1990 ) , hot gas is ejected out of the galactic disk by supernova ( sn ) explosions , and part of this gas falls back in the form of condensed neutral clouds which move at intermediate and high radial velocities .
for example , in the galactic fountain model the ejected gas from sn events falls back ballistically ( bregman 1980 ) .
these models are able to explain the vertical motion of the cold and warm gas components observed in several spiral galaxies ( e.g. fraternali et al . 2004 ; boosma et al .
2005 ) . following the analytical implementation of the galactic fountain presented by recchi et al .
( 2008 ) , we will consider such an effect in our new analytical solutions . in the context of cosmological semi - analytical models of galaxy formation and evolution each galaxy
is treated as one unresolved object , using integrated properties to describe the mass of stars , cold gas , hot gas and the black hole . since each component of the galaxy
is represented by one number , the dynamics within the galaxy is not resolved , and one needs to assume with laws for star formation , cooling and feedback that are valid on average for the entire galaxy . in this context , our new analytical solutions shall be extremely useful , because they give simple recipes concerning the time evolution of the global metallicity of a galaxy in different situations including inflow and outflow .
the paper is organized as follows : in sect .
2 we present the main assumptions of the `` analytical '' chemical evolution models , in sect .
3 we describe our assumptions and the system of equation we need to solve , and in sect 4 new analytical solutions are presented . finally ,
our conclusions are summarized in section 5 . in appendix
a we draw the complete expressions of some new analytical solutions we presented in this paper , and in appendix b a list of variables and parameters used throughout the paper is reported .
the main assumptions of the simple model ( tinsley 1980 ) are : 1 .
the imf is constant in time . 2 . the gas is well mixed at any time ( _ instantaneous mixing approximation _ )
stars @xmath2 1 m@xmath3 die instantaneously ; stars smaller than 1 m@xmath3 live forever ( _ instantaneous recycling approximation _ or ira ) .
these simplifying assumptions allow us to calculate analytically the chemical evolution of the galaxies once we have defined the fundamental quantities , such as the returned fraction : @xmath4 ( where @xmath5 is the imf and @xmath6 is the mass of the remnant ) and the yield per stellar generation : @xmath7 ( where @xmath8 is the fraction of newly produced and ejected metals by a star of mass @xmath9 ) .
recently , recchi & kroupa ( 2015 ) applied the integrated galactic initial mass function ( igimf ) to the simple model solution , and @xmath10 and @xmath11 are not constant , but are functions of time , through the time dependence of the star formation rate ( sfr ) and the metallicity of the system .
the well known solution of the so called `` closed - box '' where the system is one - zone and there are no inflows nor outflows with constant mass ( gas plus stars ) is : @xmath12 where @xmath13 is the gas fraction @xmath14 , with @xmath15 .
it is also assumed that the @xmath16 and the initial metallicity of the system is zero .
analytical solutions of simple models of chemical evolution including infall or outflow are known since at least 30 years ( pagel & patchett 1975 , hartwick 1976 , clayton 1988 , twarog 1980 , edmunds 1990 ) . here , we follow the approach and the terminology of matteucci ( 2001 ) , where it was assumed for simplicity _ linear _ flows ( gas flows are proportional to the sfr ) .
therefore , the outflow rate @xmath17 is defined as : @xmath18 where @xmath19 is the sfr , and the infall rate @xmath20 is given by : @xmath21 here @xmath22 and @xmath23 are two proportionality constants @xmath24 .
the first assumption is justified by the fact that , the larger the sfr is , the more intense are the energetic events associated with it ( in particular supernova explosions and stellar winds ) and therefore the larger is the chance of having a large - scale outflow ( see e.g. silk 2003 ) .
a proportionality between @xmath25 and @xmath26 has been discussed by recchi et al .
they tested different prescriptions for the infall of gas showing that their results do not change substantially if a generic exponential infall is assumed .
this demonstrates also that the major source of error in the solutions of the simple models is the ira assumption rather than the assumption of linear flows . in lilly
et al ( 2013 ) and pipino et al . ( 2014 ) , they considered the simple chemical evolution model in the cosmological context where the gas accreted is proportional to the dark matter growth .
in particular , in pipino et al . ( 2014 ) the infall parameter @xmath27 is time dependent and is defined as the ratio between the accretion rate from cosmological simulations and the sfr . in this paper
we do not investigate `` cosmological '' aspects but we study the effects of galaxy - galaxy interactions on the time evolution of galactic metallicity .
the evolution of the metallicity of a system as a function of @xmath13 in the case with only outflows , i.e. @xmath28 and @xmath29 is the following : @xmath30 , \label{zout}\ ] ] with the assumption that at @xmath31 , @xmath32(0)=0 , @xmath33 . in the opposite case ( @xmath34 and @xmath35 ) , assuming for the infalling gas a primordial composition ( e.g @xmath36 ) , we obtain this solution ( matteucci 2001 ) : @xmath37^{\lambda \over { \lambda - 1 } } \biggr\rbrace . \label{zinfall}\ ] ] the general solution presented by matteucci ( 2001 ) for a system described by the simple model in the presence of infall of gas with a general metallicity @xmath38 and outflow is : @xmath39^{\lambda \over { \lambda - \lambda - 1 } } \biggr\rbrace .
\label{eq : sol}\ ] ]
using the formalism introduced by matteucci ( 2001 ) , and adopted later by recchi et al .
( 2008 ) and spitoni et al . ( 2010 ) , for the first time we intend to study the effects of the environment on the chemical evolution of a galaxy by means of analytical solutions .
the environment effects are mimicked by an infall of gas with a time dependent metallicity originated by a nearby galaxy . in our model
we analyze the evolution of the oxygen in a generic galaxy , and assume for the oxygen yield standard values of @xmath40 = 0.01 and the returned fraction @xmath41 . as done in spitoni et al .
( 2010 ) , those values were obtained by adopting the salpeter ( 1955 ) imf , and the stellar yields of woosley & weaver ( 1995 ) for oxygen at solar metallicity .
we consider the chemical evolution of the galaxy1 with initial mass @xmath42 .
we study here the effects of the interaction with another galaxy on the abundance of oxygen @xmath43 .
we assume an environment dependent infall , originated by feedback episodes and galaxy - galaxy interactions .
we assume that the infall is proportional to the gas outflows from nearby galaxies with wind parameter @xmath22 .
first we consider the case with the `` enriched infall '' proportional to the sfr of the galaxy2 : @xmath44 where @xmath45 is the sfr of the galaxy2 , and @xmath46 is the fraction of outflowing gas from galaxy2 which reaches galaxy1 .
we assume in this work that @xmath47 .
in fact , the outflows are generally bipolar and for a lobe approaching galaxy 1 , the second lobe moves away from it .
the value @xmath46=0.5 is an upper limit for the fraction of outflowing gas from galaxy2 to galaxy1 .
only if the outflow is well collimated along a narrow solid angle and only if galaxy1 happens to intercept this galactic wind , @xmath46 can be equal to 0.5 .
for example , from galex uv images of the starbust galaxy m81 given by hoopes et al .
( 2005 ) , it can be seen that the opening angle of the northern side wind is around 55 degrees .
the @xmath48 metallicity of the infall is time and depends on the outflows of the galaxy2 .
the system of equations we have to solve for our galaxy1 is the following : @xmath49 + \psi_2 ( t ) \epsilon \lambda z_{o,2}(t)\big ) , } \label{system1}\ ] ] where @xmath50 is the sfr of the galaxy1 . as mentioned above , @xmath48 is the metallicity of the galaxy2 which is suffering only gas outflows .
we therefore need to solve another system of equations for the evolution of @xmath48 in galaxy2 .
recalling matteucci ( 2001 ) the system to be solved for galaxy2 with only outflow of gas is : @xmath51 . }
\label{system2}\ ] ] we recall that the outflow rate for the galaxy2 is @xmath52 . with equations ( [ system1 ] ) and ( [ system2 ] ) we study an isolated system formed by two galaxies when @xmath53 .
in fact , in this case we have that @xmath54 . on the other hand , imposing @xmath55
, we assume that part of the chemical enriched gas escaping from galaxy2 , ends up in the in intergalactic medium ( igm ) .
the majority of detailed chemical evolution models of galactic systems assumes that the galaxies formed by accretion of primordial gas from the igm .
therefore , we examine here also the more realistic case where both galaxy1 and galaxy2 suffer an inflow of primordial gas .
we assume that the infall rate is proportional of the sfr of the galactic system as discussed in section 2.1 .
hence , for the galaxy1 the primordial infall rate is @xmath56 , whereas is @xmath57 for the galaxy2 .
the systems of equation we need to solve are the following ones : * galaxy1 : @xmath58 + \psi_2 ( t ) \epsilon \lambda z_{o,2}(t)\big ) . } \label{sprimordial1}\ ] ] * galaxy2 : + @xmath59 . } \label{sprimordial2}\ ] ]
in this section we present our new analytical solutions .
first , we discuss the case where the galaxy1 suffers an enriched infall from an evolving companion galaxy ( galaxy2 ) . in the following
we generalize this solution in the case of different sfes for the galaxy1 and galaxy2 .
we also present the case where a primordial infall is taken into account .
the last result is related to `` environment '' effects coupled with galactic fountains . from the system of equations ( [ system2 ] ) we recover @xmath48 , and following matteucci ( 2001 )
we obtain : @xmath60 . \label{z2}\ ] ] assuming a linear schmidt ( 1959 ) law ( @xmath61 ) ( pipino et al .
2014 , recchi et al , 2008 ) , in the systems ( [ system1 ] ) and ( [ system2 ] ) , we have the following expressions of the time evolution of the gas for the galaxy1 ( @xmath62 ) and the galaxy2 ( @xmath63 ) : @xmath64\big ) , \label{mg1_sfe_fix}\ ] ] @xmath65 recalling that the gas fraction is defined as @xmath66 , we can write for the galaxy2 : in terms of the gas mass fraction @xmath13 for the closed - box model ( blue solid line ) and for a model with only gas outflow with @xmath67 ( magenta dotted line ) .
different horizontal solid green lines refer to different galactic time ( 1 , 2 , 4 , 8 , and 13 gyr ) . ]
@xmath68 from the system ( [ system2 ] ) with the initial condition that @xmath69 , we have that : @xmath70 finally , inserting eq .
( [ mtot2 ] ) in eq .
( [ mumeno ] ) , we recover the following time dependent expression for @xmath71 : @xmath72 hence , the expression of @xmath48 as a function of the galactic time becomes : @xmath73=y_o(1 - r)st .
\label{z2}\ ] ] it is worth noting that the evolution in time of @xmath48 does not depend on @xmath22 in the specific case of the schmidt ( 1959 ) sfr with @xmath74=1 , and is the same as in the closed - box one .
in fact , for a closed - box model with the linear schmidt ( 1959 ) law we have this expression for the gas mass : @xmath75 .
following eq .
( [ eq : simple ] ) we derive the expression of the metallicity @xmath76 as a function of time for the closed - box ( @xmath77 ) .
therefore , we have that : @xmath78 in fig .
[ wt ] we draw the evolution of the metallicity @xmath79 as function of the gas fraction @xmath13 for both closed - box model and the one with linear outflows in the case of the linear schmidt ( 1959 ) law , @xmath80 , with wind parameter @xmath22=2 and @xmath81=1 gyr@xmath82 .
horizontal lines indicate different galactic times during the galactic histories .
= 0.05 and @xmath83=0 .
for all the cases we assume , @xmath22=0.4 , @xmath46=0.5 , s=1 gyr@xmath82 .
the magenta short dashed line represents the case with @xmath84=2 , the green dotted line for @xmath84=1 , the solid blue line represents the case with @xmath84=0.5 .
the long dashed grey line is the closed - box solution with the same initial metallicity . _
right panel : _ : same of left panel but with an initial metallicity fixed at @xmath85=0.1 . ] ) from galaxy2 to galaxy1 on the time evolution of the metallicity ( oxygen ) for the galaxy1 .
we assume @xmath85=0.1 , @xmath83=0 , @xmath22=0.4 , s=1 gyr@xmath82 , and @xmath86=1 .
the blue solid line represents the model with @xmath46=0.5 , the case with @xmath46=0.4 is reported with the dotted green line , the short dashed magenta line is the case with @xmath46=0.3 .
the models with @xmath46=0.2 and @xmath46=0.1 are in brown dashed dotted line and violet long dashed dotted line , respectively .
the long dashed grey line is the closed - box solution with the same initial metallicity @xmath85=0.1 . ]
even if the closed - box model and the model with only outflow have the same metallicity in time , that same metallicity is reached for different values of the gas fraction @xmath13 .
the metallicity @xmath48 expressed in eq .
( [ z2 ] ) can be inserted in the system ( [ system1 ] ) , and recalling that @xmath87=@xmath88 , the third equation of the system ( [ system1 ] ) can be rewritten as : @xmath89 @xmath90\big ) ) .
\label{ets}\ ] ] hence , we assume that the sfr follows a schmidt ( 1959 ) law : @xmath91 and @xmath92 , and the differential equation we need to solve when we include eq .
( [ z2 ] ) in eq . ( [ ets ] ) is : @xmath93 @xmath94 \big ) .
\label{nonho}\ ] ] the final expression for @xmath95 in a case where the galaxy1 is affected by the interaction with another galactic system ( galaxy2 ) , and with @xmath96 is the following one : @xmath97 we see that the closed - box solution is still recovered when the initial metallicity @xmath85 for the galaxy1 is considered equal to zero .
the reason for this behaviour is that the outflow from galaxy2 has the same metallicity of the whole galaxy , i.e. it is expressed by eq .
( 20 ) . since the two expressions ( eqs . 20 and 21 ) are identical , and since metallicity is not an additive quantitiy , the expression @xmath95 in eq .
( 24 ) is expected . in fig .
[ window ] we show the time evolution of the metallicity ( oxygen ) where we assume an initial metallicity different from zero in the galaxy1 and as function of different @xmath84 ratios : 2 , 1 and 0.5 . we considering two different initial metallicities @xmath85=0.05 ( left panel ) and @xmath85=0.1 ( right panel ) .
we are aware that these values are extremely large , but here we only like to show the trends of considering different pre - enriched values . we see that at early times the dilution effect is prominent because the infalling gas has a much lower metallicity of that galaxy1 , and as expected the lower the ratio @xmath84 is , more the dilution effect is important . in fig .
[ epsilon ] the effects of the fraction of outflowing gas @xmath46 on the time evolution of the galaxy1 oxygen abundance obtained by eq .
( [ znew ] ) are shown .
assuming @xmath95=0.1 , @xmath84=1 , @xmath22=0.4 , and @xmath81= 1 gyr@xmath82 , we present different models varying @xmath46 : @xmath98 . as expected , the dilution effect is larger assuming higher @xmath46 values .
we discuss now the case where the initial metallicity for galaxy1 and galaxy2 are @xmath85=0 , @xmath99 0 , respectively . in this case
the differential equation we need to solve is : @xmath93 @xmath100 \big ) .
\label{nonho}\ ] ] the new solution is : @xmath101 in fig .
[ window_z2 ] we show the evolution of the oxygen abundance of galaxy1 using the new analytical solution presented in eq .
( [ znew_z2 ] ) with @xmath22=0.4 , @xmath46=0.5 , and @xmath85=0 , considering two different initial metallicities @xmath83=0.05 and @xmath83=0.1 . in this case
we use such high values for the pre - enrichment in order to better visualize the general trends and the effects of different @xmath84 ratios . as expected , at variance with fig .
[ window ] , the smaller is the @xmath84 ratios , the less efficient the chemical evolution is for galaxy1 .
overall , the general solution for the evolution of the oxygen abundance for the galaxy1 with both initial metallicities @xmath85 , @xmath83 different from zero , is given by the following expression : @xmath102 @xmath103 the general solution reported in eq .
( [ znewgen ] ) describes the more realistic scenario where galaxies exhibit different chemical enrichment rates and therefore they posses different metallicity when the interaction starts .
it seems plausible that galaxy2 presents a faster chemical enrichment in systems as the m81 group , where large and metal rich galaxies ( m81 , m82 , and ngc3077 ) interact and eject gas . in other environments , it can be more reasonable that only dwarf galaxies would show galactic winds , and therefore in this situation galaxy2 is less chemically enriched than galaxy1 . in conclusion , both cases @xmath83@xmath104@xmath85 , and @xmath83@xmath105@xmath85 have physical meaning . recently , in recchi et al .
( 2015 ) it was shown that the values of initial metallicities depend on the level of the interaction between galaxies . the evolution of the stellar mass content in the galaxy1 @xmath106 ,
can be simply inferred by the relation @xmath107 .
the time evolution of the total mass of galaxy1 @xmath108 , is given by the first equation of system ( [ system1 ] ) , and with the initial condition @xmath109 we have that : @xmath110 = 0 and @xmath83=0.05 .
for all the cases we assume , @xmath22=0.4 , @xmath46=0.5 , s=1 gyr@xmath82 .
the magenta short dashed line represents the case with @xmath84=2 , the green dotted line for @xmath84=1 , the solid blue line represents the case with @xmath84=0.5 .
the long dashed grey line is the closed - box solution with the same initial metallicity . _
right panel : _ : same of left panel but with the initial metallicity fo galaxy2 fixed at @xmath48=0.1 . ] as a function of the stellar mass @xmath111 for the closed - box model for a galaxy with total mass equal to 10@xmath112 @xmath113 is drawn with the dotted green line . with the solid blue line
we show the results for the galaxy1 with a initial mass of 10@xmath112 @xmath113 , and when we take into account the enriched infall from a companion galaxy with the same initial mass and the wind parameter equal to @xmath22=2 , and @xmath46=0.5 .
the initial oxygen abundances are @xmath85=@xmath83=0 . ] in fig .
[ ev_mstar ] we compare the time evolution of log(@xmath114/@xmath115)+12 as function of stellar mass for the closed - box model with a model where environment effects are taken into account .
the log(@xmath114/@xmath115)+12 quantity is recovered from @xmath95 using the following expression : log@xmath116 + 12 , where 0.75 is the assumed fraction of hydrogen .
the assumed initial mass in each model is @xmath117 m@xmath118 . in the last case , at a fixed stellar mass value
the system shows a smaller oxygen abundance compared to the closed - box evolution .
10@xmath119 , 5@xmath12010@xmath121 , 10@xmath112 , 5@xmath12010@xmath112 , 10@xmath122 @xmath113 , respectively .
we compare them with model results where we consider the interactions with a companion galaxy with the same initial masses .
for all the models it is assumed that the initial oxygen abundances are @xmath85=@xmath83=0 . with the magenta hexagon models with the wind parameter fixed at the value of @xmath123 are presented .
the cases with @xmath124 , @xmath67 , and @xmath125 are represented with blue pentagons , red squares , and brown triangles respectively . ] in fig .
[ mz ] we study the effects of the environment on the mass - metallicity ( mz ) relation by comparing with the observed relation of kewley & ellison ( 2008 ) for sloan digit sky survey ( sdss ) star - forming galaxies . to estimate the amount of gas that resides in each star - forming galaxy , and the gas fraction @xmath13 as a function of the galactic stellar mass , we use the method described in spitoni et al .
( 2010 ) .
we determine the cold gas mass of each galaxy on the basis of its sfr , using the following inverted kennicutt ( 1998 ) relation , which links the gas surface density to the sfr per unit area : @xmath126 where the gas density @xmath127 is expressed in m@xmath128pc@xmath129 , and the sfr surface density @xmath130 in m@xmath131yr@xmath132kpc@xmath129 .
the gas mass @xmath133 ( in @xmath113 ) is given by : @xmath134 where @xmath135 is the scaling radius calculated as in mo et al .
( 1998 ) . at this point
we have a relation between @xmath13 and the stellar mass for each considered galaxy .
first , we follow the evolution of the closed box models with zero initial metallicity for different initial masses @xmath136=@xmath137=@xmath138=8.8@xmath12010@xmath119 , 5@xmath12010@xmath121 , 10@xmath112 , 5@xmath12010@xmath112 , and 10@xmath122 m@xmath118 , respectively . for each model
we compute the time evolution of @xmath139 , @xmath140 , and @xmath141 .
thus , we consider the time @xmath142 where the stellar mass @xmath143 ) and the corresponding @xmath13 belong to the fit reported in fig . 2 of spitoni et al .
( 2010 ) . at this point
the metallicity at the time @xmath142 is computed using eq .
( 21 ) . in fig .
[ mz ] we show the closed - box model results for systems with initial masses @xmath144 = 8.8@xmath12010@xmath119 , 5@xmath12010@xmath121 , 10@xmath112 , 5@xmath12010@xmath112 , and 10@xmath122 m@xmath118 , respectively .
values of spitoni et al .
( 2010 ) for galaxies with total masses as in fig .
we compare them with model results where we consider the interactions with a companion galaxy with the same initial masses .
for all the models it is assumed that the initial oxygen abundances are @xmath85=@xmath83=0 .
models with different wind parameters @xmath22 are indicated with symbols as fig .
[ mz ] . ]
the closed box model is not able to reproduce the observed mz relation , and in spitoni et al .
( 2010 ) we concluded that a galactic wind rate increasing with decreasing galactic mass or a variable imf are both viable solutions for reproducing the mz relation .
we consider also galactic systems with the same initial masses of the closed - box models , but taking into account environment effects : an enriched infall from a galaxy with the same mass and with the same sfe with zero initial metallicities for galaxy1 and galaxy2 : @xmath85=@xmath83=0 . assuming that the initial oxygen abundances are for galaxy 1 @xmath85=0 , and for galaxy2 @xmath83= @xmath145 , respectively . ] in fig .
[ mz ] we show the results when we consider four different inflow parameters : @xmath22=0.1 , 0.5 , 2 , 6 .
we see that at a fixed stellar mass the metallicity of the galactic system drops down in presence of inflow .
we are in agreement with the work of torrey et al .
( 2012 ) , where the effects of dynamical interaction of companion galaxies were studied . in spitoni
et al . ( 2010 ) we have already proved that a combination of primordial infall and outflows with variable wind parameters as functions of the stellar mass is required to reproduce it . here
, we show the effects of this interaction compared to the closed - box case . in torrey
et al . ( 2012 ) galaxy - galaxy interactions do not change the slope of the mz , but the mz is just shifted down in parallel at smaller metallicities . in our case , if we consider a constant @xmath22 ( in fig . [ mz ] connecting the same type and colour points ) for all the galaxy masses we have a flattening of the mz slope compared to the closed - box .
however , from spitoni et al . ( 2010 ) we know that more massive galactic systems are associated to smaller outflow episodes and therefore smaller wind parameters . hence , if we consider a variable @xmath22 ( decreasing towards higher mass galaxies ) we preserve the slope of the mz relation .
our results are also in agreement with the observed depression of 0.05 -0.10 dex found by ellison et al .
( 2008 ) for the mz relation in interacting galaxies , if we consider a variable wind parameter @xmath22 with a maximum value smaller than @xmath22=6 . in the case of enriched infall from the evolution of galaxy2 as a function of @xmath146 .
we fix @xmath46=0.5 , @xmath22=2 , and consider different ratio @xmath147 values : 1 ( green dotted line ) , 0.1 ( magenta short dashed line ) , 0.01 ( grey long dashed line ) .
we compare our results with closed model case ( solid blue line ) .
the initial oxygen abundances for the galaxy1 and galaxy2 are @xmath85=@xmath83=0 . ] in fig .
[ mz_yo ] we show the mz relation obtained by the closed - box models with the same prescriptions adopted in fig .
[ mz ] but with the variable yields @xmath40 of spitoni et al .
( 2010 ) which is able to perfectly reproduce the mz relation .
again , the fact of considering `` environment '' effects leads to a drop in the metallicity of the galaxy1 at a fixed stellar mass compared to the closed - box models .
the fit with the observations deviates for @xmath148 , and this might be a hint that the kind of interactions treated in this paper can not be very general , but it surely applies for specific systems ( such as the m81 group ) . finally , in fig .
[ mz_z2 ] we show the effects of different initial metallicities @xmath83 of galaxy2 on the mz relation . assuming the same model parameters of fig .
[ mz ] but with @xmath149 , we see that at a fixed stellar mass , the decrease of the metallicity is less prominent compared fig .
[ mz ] especially for small stellar masses .
edmunds ( 1990 ) studied the effects of gas flows on the chemical evolution of galaxies showing that it does exist a forbidden area in the plane log(@xmath150 ) versus metallicity above the simple model solution , for models with inflow of unenriched gas or outflows . here , we want to test if this result is still valid in the case of a time dependent enriched infall from a companion galaxy with initial metallicities for galaxy1 and galaxy2 set to zero : @xmath151=0 .
the expression of @xmath152 as a function of time is : @xmath153\big)}{1+\frac{\lambda \epsilon}{1+\lambda}\frac{m_{g2}(0)}{m_{g1}(0 ) } \big(1- e^{-\lambda(1 - r)st}\big)}. \label{mu1}\ ] ] we need an expression for @xmath152 as a function of the metallicity @xmath79 .
we have seen that the solution @xmath95 in the case of zero initial metallicity is @xmath154 .
therefore , we can rewrite eq .
( [ mu1 ] ) as : @xmath155\big)}{1+\frac{\lambda \epsilon}{1+\lambda}\frac{m_{g2}(0)}{m_{g1}(0 ) } \big(1- e^{-\lambda z_1/y_o}\big)}.\ ] ] in fig .
[ logmu ] we show the metallicity @xmath156 in case of enriched infall from the evolution of galaxy2 as the function of @xmath146 .
we fix @xmath46=0.5 , @xmath22=2 , and consider different ratio @xmath84 values:1 , 0.1 , 0.01 .
we see that all the studied cases are in the forbidden area defined by edmunds ( 1990 ) , and they evolve similarly to the closed box model with small differences .
this result , at variance with edmunds ( 1990 ) is due to the fact that here is the first time in which these particular cases are studied and the theorems t(2 ) and t(3 ) of edmunds ( 1990 ) can not be applied because of the presence of enriched infall . in recchi et al .
( 2008 ) , new analytical solutions in the framework of differential winds were presented , namely galactic winds in which the metals are ejected out of the parent galaxy more efficiently than the other elements .
the existence of differential winds has been first introduced in the context of chemical evolution of galaxies by pilyugin ( 1993 ) and marconi et al .
( 1994 ) . a realistic case of variable infall metallicity is represented by the situation in which the metallicity of the infalling gas is set to be always equal to the one of the galactic wind .
this condition implies that the very same gas that has been driven out of the galaxy by energetic events can subsequently fall back to the galaxy , due to the gravitational potential well .
is drawn with the yellow arrow .
the outflowing gas from galaxy1 with parameter @xmath157 and the fraction which fall back ( @xmath158 ) are in gray arrows .
our model is in the framework of differential wind theory when metals are more easily channeled out . ] in presence of galactic fountains when mass ratios @xmath159 and sfe @xmath81=1 gyr@xmath82 , using the new analytical solution presented in eq .
( [ z_fountain ] ) with @xmath46=0.5 , and @xmath22=0.4 , @xmath160 and @xmath161 .
the initial oxygen abundances for the galaxy1 and galaxy2 are @xmath85=@xmath83=0 . the model with @xmath0= 4.5
is drawn with the blue long solid line , the model with @xmath0= 3 is represented by dotted green line solid line , the case with @xmath0= 2 is drawn with with the short dashed magenta line . with the cyan long dashed line the model with @xmath0= 1.5 is represented . ] when we consider different sfes @xmath162 for the galaxy2 following the new analytical solution presented in eq .
( [ sol_sfe_dif ] ) .
we consider the case with @xmath84=1 , @xmath46=0.5 , and @xmath22=0.1 .
the sfe for the galaxy1 is fixed at the value of @xmath163=1 gyr@xmath82 . the model with @xmath162=1 gyr@xmath82
is drawn with the blue solid line , the model with @xmath162=2 gyr@xmath82 is represented by the magenta short dashed line , with the long dashed grey line the model with @xmath162=0.5 gyr@xmath82 is represented .
finally , the model with @xmath162=0.1 gyr@xmath82 is drawn with the dotted green line .
the initial oxygen abundances for the galaxy1 and galaxy2 are @xmath85=@xmath83=0 . ] when we consider different mass ratios @xmath84 , whereas sfes @xmath163 and @xmath162 have been fixed at the values of 1 and 0.1 gyr@xmath82 , respectively .
we use the new analytical solution presented in eq .
( [ sol_sfe_dif ] ) with @xmath46=0.5 , and @xmath22=0.1 . the model with @xmath84= 10@xmath164
is drawn with the grey long solid line , the model with @xmath84= 10@xmath165 is represented by the blue solid line , the case with @xmath84= 1 is drawn with with the dotted green line . with the magenta short dashed line the model with @xmath84= 10@xmath82
is represented .
the initial oxygen abundances for the galaxy1 and galaxy2 are @xmath85=@xmath83=0 . ]
this kind of duty cycle is called galactic fountain ( shapiro & field 1976 , bregman 1980 ) .
spitoni et al .
( 2008 , 2009 ) showed the effect of galactic fountains on a detailed chemical evolution model ( where the instantaneous recycling and mixing approximation were relaxed ) of the milky way .
they discussed the delay in the chemical enrichment due to the fact that the gas takes a finite time to orbit around the galaxy an fall back into the disk . in recchi
et al . ( 2008 ) an analytical solution was presented in the case for galactic fountains . here
, we want to generalize the results of recchi et al .
( 2008 ) , and find a new analytical solution for the metallicity ( oxygen ) in the galaxy1 when the galactic fountain effect is included together with the the enriched infall from the galaxy2 . to consider galactic fountains we have to take into account an outflow episode and a new infall one caused by sn explosion events in the galaxy , with parameters @xmath157 and @xmath158 .
the metallicities of the outflows an fountains are : @xmath166 @xmath167 we are studying galactic fountain effects in the framework of differential winds as done in recchi et al .
( 2008 ) where it was introduced the parameter @xmath168 , in order to take into account for the fact that metals are more easily channeled out from the parent galaxy compared to the unprocessed gas . in fig .
[ fontfig ] we present a sketch of the gas flow patterns for the galaxy1 when galactic fountain are considered .
it is shown the gas inflow from the galaxy 2 with the associated wind parameter @xmath22 , the outflow from galaxy1 ( @xmath157 ) , and the fraction which comes back ( @xmath158 ) .
melioli et al .
( 2015 ) using three - dimensional hydrodynamical simulations investigated the impact of sn feedback in gas rich dwarf galaxies and the formation of galactic fountain and outflows .
they found a similar circulation flow of the one presented in fig . [
fontfig ] : galactic fountains is generally established and the metal - rich sn ejecta is instead more significant fraction ( 25 - 80 percent ) is vented in the intergalactic medium , even . it must be underlined that the treatment of the galactic fountains is actually analogous to what is done in recchi et al .
( 2008 ) with @xmath157 replacing @xmath22 and @xmath158 replacing @xmath23 .
the difference is in the role of galaxy2 .
the system of equations we have to solve for the galaxy1 is the following : @xmath169 + \psi_2 ( t ) \epsilon \lambda z_{o,2}(t)\big ) . }
\label{system1_fount}\ ] ] with @xmath170 .
the differential equation of the metallicity in terms of oxygen for the system is : @xmath93 @xmath171\big ) , \label{}\ ] ] with @xmath172 .
it can be easily proved that the evolution in time of the mass of gas of the galaxy1 is given by : @xmath173 @xmath174 in eq .
( [ z_fountain ] ) of the appendix a we show the complete expression of the analytical solution in presence of galactic fountains . in fig .
[ plotfont ] the effects of different values of the parameter @xmath0 are tested .
the higher @xmath0 is , the less the galaxy gets enriched .
this is because we assumed that @xmath175 and consequently the metals ( in this case oxygen ) which escape , are larger than the ones which rain back into the galaxy .
finally , we test the case where the interacting galaxies have different sfes .
it is well known that different galactic systems can be characterized by different sfes ( matteucci 2001 ) , and generally , higher sfes can be associated to more massive systems . here , we show the new analytical solution for the galaxy1 with a sfe @xmath163 receiving enriched outflow of gas from another galaxy associated with a sfe @xmath162 .
therefore , with the new sfrs @xmath176 and @xmath177 the eq .
( [ mg1_sfe_fix ] ) becomes : @xmath178 @xmath179 with @xmath180 .
it is trivial to see that when @xmath181 eq .
( [ mg1_sfe ] ) is identical to eq .
( [ mg1_sfe_fix ] ) .
the differential equation we have to solve to have the metallicity evolution of the galaxy1 in presence of different sfes is : @xmath182 @xmath183 \big ) .
\label{}\ ] ] finally , the new analytical solution in this case is the following : @xmath184 @xmath185a(t ) - \frac{m_{g1}(0 ) } { m_{g2}(0 ) } \big[b(t)\big ] } { ( -1 + e^{\delta t } ) \epsilon \lambda s_2 + \frac{m_{g1}(0 ) } { m_{g2}(0 ) } \big(s_1 - ( 1 + \lambda ) s_2 \big ) } , \label{sol_sfe_dif}\ ] ] where @xmath20 and @xmath186 are the following time dependent terms : @xmath187 @xmath188 @xmath189 we note that the result depends on the ratio between the initial masses of galaxy1 and galaxy2 .
moreover , as expected , when we impose @xmath181 in eq .
( [ sol_sfe_dif ] ) , the resulting @xmath95 is identical to the one obtained with eq .
( [ znew ] ) . in fig .
[ 1sfev ] we test the effect of different sfes for the galaxy1 and galaxy2 when we consider the same initial gas mass ratio ( @xmath84= 1 ) .
we tested different values for @xmath162 : 2 , 1 , 0.5 , 0.1 gyr@xmath82 .
for the galaxy1 we assume @xmath163=1 gyr@xmath82 in all models .
we obtain , as expected , that smaller values of @xmath162 lead to a smaller chemical enrichment of the system . in the case of @xmath162= 0.1 gyr@xmath82
the metallicity of the system even decreases at later galactic times due to the strong dilution effect of the infalling gas .
in fact , we can consider it like pristine gas because of , as expected , the small sfe value of galaxy2 compared to galaxy1 . in fig .
[ 2sfev ] , fixing @xmath163 and @xmath162 at the constant values of 1 and 0.1 gyr@xmath82 , respectively , we show the effect of different initial gas mass ratios for the galaxy1 and galaxy2 .
we consider four cases : @xmath84= 10@xmath164 , 10@xmath165 , 1 , and 10@xmath82 . because of the choice of an extremely low sfe for galaxy2 we expect that the most important deviation from the closed - box solution is obtained with larger @xmath190 values , hence smaller @xmath84 ratios . in fig .
[ 2sfev ] we confirm it , and the model with @xmath84= 10@xmath164 , is almost identical to the closed - box evolution model .
when we take into account also a primordial infall with wind parameter @xmath23 .
we consider the case when @xmath84=1 , @xmath46=0.5 , @xmath22=0.3 , and @xmath81=1 gyr@xmath82 .
the initial oxygen abundances for the galaxy1 and galaxy2 are @xmath85=@xmath83=0 . the model with @xmath23=0.5
is drawn with the long dashed grey line , the model with @xmath23=0.7 is represented by the magenta short dashed line , the case with @xmath23=0.9 is drawn with the dotted green line . with the solid blue line the model with @xmath23=1.5 is represented . ] when we consider different sfes @xmath162 for the galaxy2 coupled with a primordial infall of gas following the new analytical solution presented in eq .
( [ sfeprim ] ) of appendix a. we consider the case when @xmath84=1 , and @xmath46=0.5 , and the sfe fro the galaxy1 is fixed at the value of @xmath163=1 gyr@xmath82 .
all the models assume @xmath22=0.3 and @xmath23=0.7 .
the initial oxygen abundances for the galaxy1 and galaxy2 are @xmath85=@xmath83=0 . the model with @xmath162=2 gyr@xmath82
is represented by the magenta dotted line , the model with @xmath162=1 gyr@xmath82 is drawn with the blue solid line , the model with @xmath162=0.2 gyr@xmath82 is drawn with the green short dashed line .
finally , the model with @xmath162=0.1 gyr@xmath82 is drawn with the grey long dashed line . ]
we present here the results in the presence of a primordial infall of gas for both galaxy1 and galaxy2 .
first , we need to compute the time evolution for the metallicity @xmath48 when a primordial infall of gas is considered . to this aim
we need the expression of the mass fraction @xmath191 as a function of time . from the system ( [ sprimordial2 ] ) , with the condition that the total initial mass is equal to the gas content @xmath69 we have that : @xmath192 the mass fraction @xmath193 can be written as : @xmath194 we recall that the general solution as function of the gas fraction for a system with primordial infall and outflow is given by eq .
( [ eq : sol ] ) with @xmath36 . finally , if we insert eq .
( [ muprim ] ) in eq .
( [ eq : sol ] ) , we obtain that @xmath195 the new expressions for the time evolution of @xmath62 and @xmath63 are , respectively : @xmath196 \biggr\rbrace , \label{}\ ] ] @xmath197 the differential equation we should to solve in presence of a primordial infall , is the following one : @xmath198 @xmath199 - \epsilon z_{o,1}(t )
\lambda \biggr\rbrace .
\label{z_prim1}\ ] ] the solution of eq .
( [ z_prim1 ] ) for the time evolution of the oxygen abundance for the galaxy1 is then given by : @xmath200 @xmath201 y_o - b2(t)+c2(t ) \big),\ ] ] with @xmath202 @xmath203 and @xmath204.\ ] ] it is easy to show that it is possible to recover the solution of eq .
( [ znew ] ) presented in section 4.1 for @xmath205 . in fig .
[ 2infall_constant ] we show the time evolution of the metallicity @xmath156 when @xmath206=1 , @xmath46=0.5 , and the sfe for the galaxy1 and galaxy2 are fixed at the value @xmath81=1 gyr@xmath82 .
assuming for all the models @xmath22=0.3 , we test different values for the the infall parameter @xmath23 which is associated with the primordial infall : 0.5 , 0.7 , 0.9 , 1.5 .
we note that the effect of dilution of a primordial gas infall overwhelms the enriched infall from the companion galaxy , and the time evolution of the metallicity for the galaxy1 deviates substantially from the closed - box solution ( we have proved in previous sections that the evolution in time of a system with only enriched gas from a companion galaxy follows the closed - box solution ) even with a primordial infall parameter comparable to the one of enriched infall .
we also consider the case with different sfes for the galaxy1 and galaxy2 in presence of primordial infall .
we do not enter into details about the procedures and the relations used to recover the new analytical solution @xmath95 , and we only show the solution in eq.([sfeprim ] ) of appendix a. in fig .
[ infall_prim2 ] we show the effects of different values for sfes of the galaxy2 on the chemical evolution of galaxy1 in presence of primordial infall .
we consider the case when the mass ratio is @xmath84=1 , and @xmath46=0.5 , and the sfe for the galaxy1 is fixed at the value of @xmath207 gyr@xmath82 .
all the models assume @xmath22=0.3 and @xmath23=0.7 .
we consider different models with different @xmath162 values : 2,1 , 0.2 , 0.1 gyr@xmath82 . in fig .
[ infall_prim2 ] it is clearly shown that all the models with different sfes show more or less the same time evolution for the @xmath156 . comparing the models with @xmath162=0.2
gyr@xmath82 and @xmath162=0.1 gyr@xmath82 we note that at late times , as expected , the chemical evolution of galaxy1 , which suffers the enriched infall from galaxy2 with @xmath162=0.2 , shows a higher metallicity .
this behavior is inverted at early times .
this is due to the way in which we consider the primordial infall .
in fact as shown in eq .
( 5 ) , the primordial infall is assumed to be proportional to the sfr of the galaxy , therefore at early times the galaxy2 with @xmath162=0.2 gyr@xmath82 suffers a larger dilution effect than the same system with @xmath162=0.1 gyr@xmath82 . we conclude that in presence of primordial infall the chemically enriched gas coming from galaxy2 has not a big effect on the chemical evolution of the galaxy1 .
in this work we consider only primordial infall from the igm , whereas in peng & maiolino ( 2014b ) it was studied how global environment properties ( overdensity of galaxies ) can modify the metallicity of the igm .
the observed strong correlation between over - density and metallicity for star - forming satellites suggests that the gas infall is getting progressively more metal - enriched in dense regions .
in this paper we presented new analytical solutions for the evolution of the metallicity ( oxygen ) of a galaxy in presence of `` environment '' effects coupled with galactic fountains , and primordial infall of gas .
the main results of our study are the following ones : * if we consider a linear schmidt ( 1959 ) law for the star formation rate we have the same time evolution of the metallicity for both the closed - box model and the one with only outflow .
therefore , in the last case the result does not depend on the wind parameter @xmath22 .
this result is holding when the outflow is not differential .
it is worth to note that even if at the same time those two systems show the same content of metals they show different gas fractions as shown in fig .
the new analytical solution for the evolution of the galaxy1 where we consider the enriched inflow of gas from the galaxy2 with initial metallicities @xmath85 ( galaxy1 ) and @xmath83 ( galaxy2 ) is : + @xmath208 + @xmath103 + where : @xmath22 is the wind parameter , @xmath46 is the fraction of the outflowing gas from galaxy2 to galaxy1 , @xmath81 is the star formation efficiency , and @xmath209 is the ratio between the initial gas masses of the two galaxies . * when we consider the time evolution of a galaxy including an enriched infall due to the interactions with a nearby galaxy assuming no pre - enriched gas for both galaxies , the chemical enrichemnt is less efficient than in the closed - box solution at a fixed stellar mass .
moreover , the mass - metallicity relation for galaxies which suffer a gas infall from an evolving galaxy , is shifted down at lower z values if compared to the closed - box model results for isolated galaxies . *
we show a new analytical solution where we consider galactic fountain effects in the framework of differential winds coupled with the interaction with a nearby galaxy .
* we presented the new solution in case of different sfes .
if the infall gas originated by a nearby galaxy has a smaller sfe than the galaxy suffering the infall , the recipient galaxy will show a less efficient chemical enrichment . in this case
we showed that the smaller is the ratio between the initial gas masses of the recipient and donor galaxy , the less efficient the chemical evolution is . *
a new analytical solution when a primordial infall is coupled with interactions with a nearby galaxy is presented .
the effect of dilution of a primordial gas infall overwhelms the enriched infall from the chemical evolution of a companion galaxy even with a primordial infall parameter comparable to the one of enriched infall .
we have also shown how different sfes for the companion galaxy does not affect the chemical evolution of galaxy1 in presence of primordial infall .
the author thanks f. vincenzo , and v. grieco for several fruitful discussions , and f. matteucci for many useful suggestions and for reading the manuscript .
the author also thanks an anonymous referee whose comments improved noticeably the paper .
the work was supported by prin miur 2010 - 2011 , project `` the chemical and dynamical evolution of the milky way and local group galaxies '' , prot .
2010ly5n2 t .
99 barnes , j. e. , hernquist , l. e. 1992 , ara&a , 30 , 705 berentzen , i. , athanassoula , e. , heller , c. h. , fricke , k. j. 2003 , mnras , 341 , 343 boomsma , r. , oosterloo , t. , fraternali , f. , van der hulst , t. , sancisi , r. 2005 , in extra - planar gas conference , asp conf .
ser . , ed .
r. braun , 331 , 247 boselli , a. , & gavazzi , g. 2006 , pasp , 118 , 517 bregman , j. n. 1980 , apj , 365 , 544 clayton , d. d. 1988 , mnras , 234 , 1 davies , j. i. , wilson , c. d. , auld , r. , et al . , 2010 , mnras , 409 , 102 edmunds , m. g. , 1990 , mnras , 246 , 678 fraternali , f. , oosterloo , t. , sancisi , r. 2004 , a&a , 424 , 485 greggio , l. , renzini , a. 1983 , a&a , 118 , 217 hartwick , f. d. a. , 1976 , apj , 209 , 418 hoopes , c.g . , heckman , t.m . ,
strickland , d.k .
, et al . 2005 , apjl , 619 , l99 houck , j. c. , bregman , j. n. , 1990 , apj , 352 , 506 kennicutt , r. c. 1990 , in iau colloquium 124 , nasa marshall space flight center , paired and interacting galaxies , ed .
j. w. sulentic , w. c. keel , & c. m. telesco , 269 kewley , l. j. , ellison , s. l. 2008 , apj , 681 , 1183 kudritzki , r.p .
, ho , i. , schruba , a. , et al . , 2015 , arxiv:1503.01503 lacey , c. g. , fall , m. 1985 , apj , 290 , 154 larson , r. b. , tinsley , b. 1978 , apj , 219 , 46 lilly s. j. , carollo c. m. , pipino a. , renzini a. , peng y. , 2013 , apj , 772 , 19 marconi , g. , matteucci , f. , tosi , m. 1994 , mnras , 270 , 35 matteucci , f. 2001 , the chemical evolution of the galaxy , assl , kluwer academic publisher matteucci , f. , chiosi , c. , 1983 , a&a , 123 , 121 matteucci , f. , greggio , l. 1986 , a&a , 154 , 279 melioli , c. , brighenti , f. , dercole , a. , 2015 , mnras , 446 , 299 mo , h. j. , mao , s. , white , s. d. m. , 1998 , mnras , 295 , 319 pipino , a. , lilly , s. j. , carollo , c. m.,2014 , mnras , 441 , 1444 pagel , b. e. j. , patchett , b. e. 1975 , mnras , 172 , 13 peeples m. s. , shankar f. , 2011 , mnras , 417 , 2962 peng , y. , maiolino , r. 2014a , mnras , 443 , 3643 peng , y. , maiolino , r. 2014b , mnras , 438 , 262 pilyugin , l. s. 1993 , a&a , 277 , 42 recchi , s. , kroupa , p , mnras , 2015 , 446 , 4168 recchi , s. , kroupa , p , ploeckinger , s. , 2015 , arxiv:1504.02473 recchi , s. , spitoni , e. , matteucci , f. , lanfranchi , g. a. , 2008 , a&a , 489 , 555 salpeter , e. e. 1955 , apj , 121 , 161 schmidt , m. 1963 , apj , 137 , 758 schmidt , m. 1959 , apj , 129 , 243 searle , l. , sargent , w. l. w. 1972 , apj , 173 , 25 shapiro , p. r. , field , g. b. 1976 , apj 205 , 762 silk , j. , 2003 , mnras , 343 , 249 smith , b. j. , giroux , m. l. , struck , c. , hancock , m. , hurlock , s. 2010 , aj , 139 , 1212 smith , b. j. , struck , c. , 2001 , aj , 121 , 710 spitoni e. , calura f. , matteucci f. , recchi s. , 2010 , a&a , 514 , a73 spitoni , e. , recchi , s. , matteucci , f. 2008 , a&a , 484 , 743 spitoni , e. , matteucci , f.,recchi , s. , cescutti , g. , pipino , a. , 2009 , a&a , 504 , 87 temporin , s. , weinberger , r. , galaz , g. , kerber , f. 2003a , apj , 584 , 239 temporin , s. , weinberger , r. , galaz , g. , kerber , f. 2003b , apj , 587 , 660 tinsley , b. m. 1974 , apj , 192 , 629 tinsley , b. m. 1980 , fund . cosmic phys . , 5 , 287 toomre , a. , toomre , j. 1972 , apj , 178 , 623 torrey p. , cox t. j. , kewley l. , hernquist l. , 2012 , apj , 746 , 108 twarog , b. a. , 1980 , apj , 242 , 242 woosley , s.e . ,
weaver , t.a .
1995 , apjs , 101 , 181
* we report here the analytical solution for the time evolution of the oxygen abundance @xmath156 when galactic fountain and the enriched infall of gas from a companion galaxy are taken into account .
this solution is obtained when the initial metallicity of the galaxy1 is assumed equal to zero .
+ @xmath210 + @xmath211 + with + @xmath212 + @xmath213 + @xmath214 + @xmath215 + @xmath216 * the following solution is related to the case with a primordial infall coupled with an enriched infall from a companion galaxy , considering different sfes : + @xmath217 + with : @xmath218 + @xmath219 + @xmath220 + @xmath221 + @xmath222 s_1 - [ 1 + ( e^{\lambda _ * t } - e^{q t } ) ( -1 + \lambda - \lambda ) + \lambda - e^{q t } ( 1 + \lambda ) ] s_2 \big ) yo,\ ] ] + @xmath223\big).\ ] ]
@xmath224 : oxygen abundance of galaxy @xmath225 ; + @xmath226 : mass of gas in galaxy @xmath225 ; + @xmath227 : star formation rate fo galaxy @xmath225 ; + @xmath228 : star formation efficiency of galaxy @xmath225 ; + @xmath229 : gas fraction of galaxy @xmath225 ; + @xmath40 : oxygen yield ; + @xmath11 : returned fraction ; + @xmath22 : outflow parameter ; + @xmath23 : infall parameter ; + @xmath46 : fraction of the outflowing gas from galaxy2 to galaxy1 ; + @xmath157 : outflow parameter connected to galactic fountains ; + @xmath158 : infall parameter connected to galactic fountains ; + @xmath0 : differential wind parameter . | in the last years , more and more interest has been devoted to analytical solutions , including inflow and outflow , to study the metallicity enrichment in galaxies . in this framework
, we assume a star formation rate which follows a linear schmidt law , and we present new analytical solutions for the evolution of the metallicity ( z ) in galaxies . in particular , we take into account environmental effects including primordial and enriched gas infall , outflow , different star formation efficiencies and galactic fountains .
the enriched infall is included to take into account galaxy - galaxy interactions .
our main results can be summarized as : i ) when a linear schmidt law of star formation is assumed , the resulting time evolution of the metallicity z is the same either for a closed - box model or for an outflow model .
ii ) the mass - metallicity relation for galaxies which suffer a chemically enriched infall , originating from another evolved galaxy with no pre - enriched gas , is shifted down in parallel at lower z values , if compared the closed box model .
iii ) when a galaxy suffers at the same time a primordial infall and a chemically enriched one , the primordial infall always dominates the chemical evolution .
iv ) we present new solutions for the metallicity evolution in a galaxy which suffers galactic fountains and an enriched infall from another galaxy at the same time .
the analytical solutions presented here can be very important to study the metallicity ( oxygen ) , which is measured in high - redshift objects .
these solutions can be very useful : a ) in the context of cosmological semi - analytical models for galaxy formation and evolution , and b ) for the study of compact groups of galaxies .
-.6 in [ firstpage ] galaxies : abundances - galaxies : evolution - galaxies : ism |
scaling relations provide powerful insights on the formation and evolution of early - type galaxies . besides linking internal structure and dynamics through the fundamental plane @xcite , they allow to link these properties with those of stellar populations .
the most studied scaling relation between population and dynamics is the mg@xmath3 - @xmath1 relation @xcite , which shows that the mg@xmath3 line index increases with galaxy mass .
this scaling relation may be used to study whether bulge populations are younger than those of elliptical galaxies , with the basic assumption that elliptical galaxies in clusters are old . however , stellar population studies in the wavelength region of the mg@xmath3 index @xcite indicate that mg@xmath3 is affected by age and metallicity ; that mg is overabundant w.r.t .
fe in bright galaxies ; and , therefore , that the mg@xmath3@xmath1 relation for ellipticals is not necessarily a relation for old galaxies @xcite .
the near - ir caii triplet at 8500 is a bright atomic feature that can easily be studied with modern detectors .
it has been used in the past to analyse stellar populations ( e.g. @xcite ) .
recently , a series of papers have been published describing a new stellar library that is used to make a thorough calibration of the caii triplet and other features in the same spectral region in terms of stellar parameters @xcite .
a followup paper ( @xcite , hereafter vaz03 ) presents the behaviour of these features for single stellar populations as predicted by recent evolutionary synthesis models .
* hereafter cen02 ) , have used this calibration to study the near - ir caii triplet in a sample of elliptical galaxies .
they find that cat@xmath0 , the ca triplet index corrected for the presence of paschen lines , decreases with @xmath1 , and that , in the most luminous galaxies , cat@xmath0 is much lower than predicted by models .
the decrease of cat@xmath0 with @xmath1 may be a result of a dwarf - enhanced imf or a low ca / fe abundance ratio for giant ellipticals . in a similar study , ( * ? ? ?
* hereafter sag02 ) find a similar underabundance of cat w.r.t .
models in luminous ellipticals , but do not see any dependence of cat@xmath0 with @xmath1 . here
we analyse a sample of galactic bulges to compare the trends of ca with @xmath1 in bulges to those found in ellipticals .
different abundance ratios might be expected if bulges and ellipticals have different formation histories .
the study offers an opportunity to see whether cat@xmath0 @xmath1 relation may be established in general for the spheroidal components of galaxies .
nineteen early- to intermediate - type galaxies ( s0 to sbc ) , from the sample defined by @xcite , were observed using the isis double spectrograph at the 4.2 m william herschel telescope of the observatorio del roque de los muchachos at la palma in 1997 july 11 - 13 .
minor - axis spectra were obtained in the 8360 - 9170 range , at 1.74 spectral resolution ( fwhm of the arc lines ) .
the slit width was 1.2 to match the seeing at the time of the observations .
a detailed summary of the observations and data reduction can be found in @xcite .
we measured the recently defined line - strength indices cat@xmath0 , cat , pat @xcite and mgi , on the flux calibrated spectra .
the main advantages of these new definitions over the previous ones ( e.g. @xcite ) are that they have been designed to ease problems in the definition of the continuum bands , the contamination of the paschen lines , and the sensitivity to the s / n ratio and velocity dispersion ( see @xcite for more details ) . in order to explore the dependence of the relation of line - strength and velocity dispersion on aperture
, we measured the 4 line - strength indices in three different apertures ( r@xmath5 , r@xmath5/2 , 1.2 x 4 arcsec@xmath6 ) .
the signal - to - noise ratios for the different apertures range from 67 to 256 , 55 to 232 and 75 to 244 per angstrom , respectively .
@xmath7-band effective radii were taken from table 1 of @xcite and converted to the minor axis using high - spatial resolution @xmath8-band ellipticity profiles from @xcite . for a number of our bulges it was not possible to accurately measure the index at aperture sizes @xmath4 r@xmath5/2 due to its size and seeing conditions .
the difference at smaller apertures , however , should be minor since the near - ir caii triplet depends very little with the size of the aperture .
for brevity , only the r@xmath5 , and 1.2 x 4 apertures are shown in fig .
[ fig : indexsigma ] .
the measured values for the r@xmath5/2 aperture have been included in the electronic version of table [ tab : data ] .
kinematic parameters were derived using fourfit developed by @xcite .
a gaussian fit to the losvd was performed to obtain mean radial velocities and velocity dispersions .
we have made use of a single - age , single - metallicity synthetic model ( [ fe / h]=0.0 , t=6.31 gyr ) from as a kinematic template .
this procedure has proven to be very effective in reducing template mismatch and produces as good results as the best stellar templates in this wavelength domain @xcite . tabulated values for the cat@xmath0 index and velocity dispersions along the minor axis of the galaxies
can be found in the same paper .
we broadened all the spectra to 300 km s@xmath9 , slightly higher than the largest velocity dispersion in our sample ( 291 km s@xmath9 for the central aperture of ngc 5838 ) .
this correction is small for the cat@xmath0 index , which has shown to be fairly constant with velocity dispersion up to 300 km s@xmath9 .
errors in the indices include contributions from radial velocity uncertainties and poisson errors ( see @xcite for more details ) .
the uncertainty from s / n dominates the errors in the indices .
we have corrected our indices for differences between the spectro - photometric system of the data and the model predictions using a set of 10 stars in common with the stellar library on which the models are based .
the following offsets are found ( models - us ) : @xmath10cat@xmath0 = [email protected] , @xmath10cat = 0.205 @xmath110.027 , @xmath10mgi = @[email protected] , @[email protected] .
= 0.07 in lcccccccc index & a & @xmath10a & b & @xmath10b & r@xmath6 & r@xmath13 & @xmath2(r@xmath13 ) & rms + ( 1 ) & ( 2 ) & ( 3 ) & ( 4 ) & ( 5 ) & ( 6 ) & ( 7 ) & ( 8) & ( 9 ) + & & & & 1.2 x 4 & & & & + cat@xmath0 & 8.72 & 0.38 & -0.80 & 0.17 & 0.51 & -0.71 & 7.4e-04 & 0.09 + cat & 8.89 & 0.40 & -0.70 & 0.19 & 0.41 & -0.58 & 8.6e-03 & 0.07 + mgi & 0.00 & 0.08 & 0.11 & 0.04 & 0.13 & 0.20 & 4.2e-01 & 0.03 + & & & & r@xmath5 & & & & + cat@xmath0 & 9.44 & 0.39 & -1.14 & 0.18 & 0.52 & -0.76 & 1.7e-04 & 0.11 + cat & 9.67 & 0.42 & -1.06 & 0.20 & 0.63 & -0.80 & 4.2e-05 & 0.07 + mgi & 0.04 & 0.09 & 0.10 & 0.04 & 0.13 & 0.22 & 3.6e-01 & 0.02 + * note : * parameters of the error - weighted least - squares linear fits to the index - log(@xmath1 ) relations , defined as @xmath14 log(@xmath1 ) at 300 km s@xmath9resolution . columns 2 , 3 , 4 , 5 show the fitted parameters and their errors .
columns 6 , 7 , 8 give the linear correlation coefficients ( r ) , the spearman rank - order correlation coefficients ( r@xmath15 ) and significance levels ( @xmath2(r@xmath13 ) ) .
column 9 lists the rms around the best fit .
the relations of the line - strength indices cat@xmath0 , cat , pat , mgi and velocity dispersion are shown in figure [ fig : indexsigma ] .
we present each relation for the two aperture sizes described in [ sec : obsdata ] ( labeled on the top boxes ) .
early - type spirals ( t @xmath16 ) are depicted with filled symbols , and later types ( t @xmath170 ) with open symbols .
index values for a range of ssp ( single - age , single - metallicity stellar populations ) models at the resolution of the data are shown in the fourth - column panels .
linear least - squares fits are plotted for some of the relations , with best - fit coefficients given in table [ tab : fits ] .
the data values are presented in table [ tab : data ] . a strong anti - correlation exists between the cat and cat@xmath0 indices and log(@xmath1 ) in all apertures .
this is reflected in the low value of the significance levels of the spearman rank - order tests ( table [ tab : fits ] , column 8) , which give the probability of a null correlation , and the rms values about the best fit for the three indices .
the anti - correlation of the cat@xmath0 index with @xmath1 shown here for bulges is similar to that recently found by for elliptical galaxies ( which was , if at all , marginally detected also for ellipticals by ) .
the caii triplet is the first metal line index that is found to anticorrelate with @xmath1 .
the mgi index , in contrast , shows a positive trend with @xmath1 as expected from the behaviour of the mg@xmath18 and mg@xmath3 indices in the optical range ( e.g. @xcite ) .
we have investigated whether different apertures would lead to significant differences in the index@xmath1 relations .
no major difference is found between the gradients of any index at each aperture : the slopes and zero - points agree within the errors ( table [ tab : fits ] ) .
it is interesting to see that none of our bulges give a measurement of the cat or cat@xmath0 indices that lies above the predictions of the models .
this result agrees with the model predictions that the cat and cat@xmath0 indices tend to saturate for metallicities above [ m / h]@xmath19 - 0.4 , which are also supported by the data of and .
given that our sample contains bulges from s0 to sbc , we can study whether the index@xmath1 relations depend on morphological type ( see fig .
[ fig : indexsigma ] ) . due to our limited number of galaxies
, we split the sample in just two groups : one of them containing galaxies of types s0 , s0/a ( t@xmath200 ) and a second group with the sa to sbc galaxies ( t@xmath170 ) .
no differences are found for any of the relations .
we have compared our index@xmath1 relations with those found by measured for an aperture of 2x4 arcsec@xmath6 .
an accurate comparison with is not possible since their aperture is four times smaller than the smallest we can get given the size of some of our bulges .
we perform the comparison with at a similar aperture size and at the same spectral resolution ( 400 km s@xmath9 ) . in figure
[ fig : comparison ] we show the comparison of the cat@xmath0 index .
the results are in good agreement at the 1-@xmath1 level , with the slopes agreeing within the errors ( b[us][email protected] , b[cen02][email protected] ) .
the exclusion of one of our low - velocity dispersion galaxies ( ngc 7457 ) , provides an even better agreement in the slope values ( b[us][email protected] ) .
however the rms of our data around the best fit is much smaller ( rms[cat@xmath0(us)]=0.14 ) than the one measured by ( rms[cat@xmath0(cen02)=0.28 ) . on the other hand ,
our sample shows values for the pat index slightly higher than those of for ellipticals ( @xmath190.03 at 400 km s@xmath9 resolution ) .
previous studies of the same sample by some of us , studying deviations of bulges from the fp of ellipticals and s0s @xcite or color - color diagrams @xcite , conclude that bulge populations are old , but could be younger than ellipticals by up to 2.5 gyr .
the higher pat values , and correspondingly younger ages for bulges could also be due to young stars in the centre of our galaxies .
however , model predictions indicate that young populations ( @xmath41 gyr ) , which increase the value of the patindex , also lower the value of cat@xmath0 , which is opposite to what we find here . given the small significance of the offset it is more likely that flux calibration differences contribute to the observed shift . for the most extreme case in the sample ( ngc 5965,pat=0.91 ) the high pat could be due to star formation induced by the presence of a strong bar in this galaxy @xcite . on the other hand
, we find no correlation of the pat index with @xmath1 , a result supported by for es , and contrary to the strong trend found by .
an explanation for that difference could be in the velocity dispersion correction they have applied to each galaxy in their sample .
a large correction for more massive galaxies could explain the positive correlation of the pat index and the flatness of the cat@xmath0 index as a function of velocity dispersion .
* note : * the log(@xmath1 ) , cat@xmath0 , cat , mgi and pat values for for our sample of bulges in the different apertures , shown in fig .
[ fig : indexsigma ] . uncertainties in the parameters are shown in brackets .
the following offsets must be added to the tabulated values in order to obtain the values shown in fig . 1 : @xmath10cat@xmath0=0.178 , @xmath10cat=0.205 , @[email protected] , @xmath10pat=0.029 .
the complete version of this table is in the electronic edition of the journal .
the printed edition contains only a sample .
the origin and abundance evolution of the atomic element ca has been a matter of debate . due to its nature as an @xmath2 element , an evolution history similar to that of mg is predicted @xcite .
however several observations reveal a more complex picture where ca and mg originate and evolve differently .
the problem has resuscitated again recently in the light of , . for early - type galaxies ,
the ca abundance does not follow mg and there is evidence that it is not enhanced to fe , from , e.g. measurements of the ca4227 line @xcite or the near - ir caii triplet . on the other hand ,
the mg - log(@xmath1 ) relation has been interpreted as a mass - metallicity relation for early - type galaxies @xcite . in the light of the new results obtained by @xcite and model predictions
, such statement does not necessarily hold for the near - ir caii triplet - log(@xmath1 ) relation ( see @xcite and @xcite for extensive reviews ) . the large discrepancy between the observed cat@xmath0 and values for a salpeter - imf model for galaxies with large velocity dispersions
can be interpreted as underabundance of ca w.r.t .
the other metals in more massive galaxies . other possibilities are : 1 ) an imf biased towards low - mass stars , 2 ) composite stellar populations including a low metallicity component or a young and old population of the same metallicity , and 3 ) ca depletion .
we discuss the advantages / disadvantages of each one of these possible explanations in turn .
the problem of the ca underabundance has been previously addressed by several authors ( @xcite , ) .
it is hard to find an evolutionary scenario that can produce non - solar ca / mg values . assuming that both ca and mg are produced by two `` flavours '' of sne in which more massive stars ( 20 - 40 ) produce mg before ( @xmath212.6 myr ) ca , generated in less massive stars ( 12 - 30 )
, one could explain a small discrepancy in abundances between mg and ca @xcite . however , large abundance ratio differences are hard to obtain in this way , and impossible to get for real galaxies , for which the formation timescale is probably much larger than a few myr .
if we would use @xmath2-enhanced isochrones ( @xcite , vaz03 ) , required to get the correct [ mg / fe ] ratios in giant ellipticals , assuming a solar - like element partition for the [ mg / ca ] and [ fe / ca ] , we would predict even larger ca indices , increasing the discrepancy .
next we discuss the apparent ca underabundance at large velocity dispersions in terms of a dwarf - enhanced imf .
the universality of the imf has been a matter of controversy in the last decades ( see @xcite and @xcite for current reviews of observational keys in favour and against this statement ) .
imf time variations among ellipticals have been studied by @xcite in detail .
an advantage of this approach is its ability to explain both , the large mg and the [ mg / fe ] values observed in giant ellipticals @xcite .
the effect of a dwarf - dominated imf is a weakening of the cat@xmath0 index ( vaz03 ; @xcite ) .
this effect is stronger at high metallicities .
as also mentioned in , there are four observational drawbacks against this hypothesis : first : indications of an imf flatter than salpeter in the bulge of our own galaxy @xcite , second : the fact that the predicted feh 9916 feature is significantly stronger than what it is observed @xcite , and third : the discrepancy between visual m / l ratios ( m / l@xmath1710 , @xcite ) and dynamical estimates m / l@xmath216 @xcite .
an additional drawback is that visual - infrared ( @xmath22 ) colors are just at the verge of what it is observed in es , as pointed out by @xcite .
the use of composite stellar populations is analysed by .
their analysis reveals that a combination of populations including a low - metallicity component was not able to explain the low value of cat@xmath0 and cat together with h@xmath23 for the most massive galaxies .
our independent analysis confirms their result .
ca is known to be highly depleted onto dust in the interstellar medium @xcite .
could it be that it is also depleted in stars ?
the evidence is not strong .
such depletion in stars would require that , during star formation , interstellar dust is recycled less than interstellar gas . moreover , depletion would have to be stronger in more dusty galaxies , so one would expect a large scatter in the cat@xmath0 @xmath1 relation due to depletion .
at least , the scatter for bulges of spirals , which are generally more dusty than ellipticals , should be larger .
since this is not seen , we argue that depletion can be excluded as the origin of the low cat@xmath0 values . in summary , many mechanisms can be used to try to explain the apparent underabundance of ca for massive galaxies .
however none of them alone seems to be able to explain all the observations . in the light of the discrepancies found between all the possibilities
, a combination of several effects could be in place .
more work is still needed to provide an answer to this problem .
we thank javier cenarro and javier gorgas for providing us with software and data in electronic form , and for many useful discussions that have helped to improve this letter .
jfb acknowledges the finantial support from a pparc studentship .
the william herschel telescope is operated on the island of la palma by the isaac newton group in the spanish observatorio del roque de los muchachos of the instituto de astrofsica de canarias . | we present measurements of the near - infrared calciumii triplet ( cat , cat@xmath0 ) , paschen ( pat ) and magnesium ( mgi ) indices for a well - studied sample of 19 bulges of early to intermediate spiral galaxies .
we find that both the cat@xmath0 and cat indices decrease with central velocity dispersion @xmath1 with small scatter .
this dependence is similar to that recently found by @xcite for elliptical galaxies , implying an uniform cat@xmath0 @xmath1 relation that applies to galaxies from ellipticals to intermediate - type spirals .
the decrease of cat and cat@xmath0 with @xmath1 contrasts with the well - known increase of another @xmath2-element index , mg@xmath3 , with @xmath1 .
we discuss the role of ca underabundance ( [ ca / fe]@xmath40 ) and imf variations in the onset of the observed relations . |
it is commonly accepted that the agn emission is due to gravitational release of energy by mass falling onto a super compact object .
most of the attention is presently devoted to the geometry of the problem and to physical details in the attempt to explain some spectral characteristics and peculiarity of particular objects , rather than to answer to the fundamental question : + _ where does the accreting mass come from ?
_ in this note , which is part of a more extensive work described elsewhere ( capuzzo dolcetta , 1993 ; capuzzo dolcetta , 1996 ) i try to give an answer to this question , carrying evidence that spherical mass accretion on a compact object in form of dynamically decayed globular clusters in a ( triaxial ) elliptical galaxy can be the source of the power released by agns without invoking ad hoc assumptions .
it seems quite ascertained that the radial distribution of globular clusters is lees peaked than that of bulge stars .
recent wfpc2 hst observations of 14 elliptical galaxies ( forbes et al .
1996 ) confirm the general trend of flattening of the cluster distribution within @xmath0 kpc from the centre .
this difference between the bulge and the globular cluster distributions may be ab initio or due to evolution of the cluster system distribution .
this latter hypothesis is clearly the most `` economical '' and it has been positively checked by capuzzo dolcetta ( 1993 ) . here
i would point the attention to one of the consequences that the evolution of the globular cluster system ( gcs ) has in terms of mass carried to the galactic centre .
the details of the general physical model can be found in capuzzo dolcetta ( 1996 ) ( see also pesce , capuzzo dolcetta and vietri 1992 , and capuzzo
dolcetta 1993 ) .
here i just recall that the gcs evolution in a diskless galaxy is due to the actions of _ dynamical friction _ by field stars and of the _ tidal _ interaction with a compact object in the centre of the galaxy .
the relevant time scales ( @xmath1 , @xmath2 ) depend on the individual cluster initial energy , mass and mass density as well as on the galactic potential and velocity distribution plus ( of course ) on the nucleus mass ( @xmath3 ) .
the ratio of the two time scales @xmath4 shows a competition between the two processes : dynamical friction is effective until a nucleus is accreted in mass enough to destroy the incoming clusters ( @xmath5 is the cluster density averaged over its half mass radius ) .
this suggests the possibility of a self regulated nucleus formation in a galaxy .
various points have to be investigated , among them : \(i ) when @xmath2 and @xmath1 are sufficiently short to be relevant in the life of a galaxy ?
\(ii ) is it possible to build up a compact nucleus in the galactic centre in the form of dynamically decayed globulars ?
\(iii ) what fraction of the mass of frictionally decayed and tidally destroyed clusters is swallowed in the compact nucleus ?
\(iv ) can a quasar like emission be explained by such kind of spherical accretion ?
a definite quantitative answer to question ( ii ) will require an accurate modelization ; the answers to ( i ) , ( iii ) ( as well as to ( iv ) ) have been given in pesce et al .
( 1992 ) , capuzzo dolcetta ( 1993 ) , capuzzo dolcetta ( 1996 ) , and will not be discussed here . in what follows
i limit myself to give support to the positive answer to question ( iv ) .
qualitatively : a black hole of given mass @xmath6 stays at the galactic centre and swallows the surrounding stars which enter a destruction radius @xmath7 where @xmath8 is the schwarzschild s radius and @xmath9 the tidal breakup radius @xmath10 with @xmath11 and @xmath12 mass and radius of the typical star .
the resulting rate of ( spherical ) mass accretion is @xmath13 , which yields a gravitational luminosity @xmath14 ( @xmath15 is the gravitational potential near @xmath16 ) .
they crucially depend on the star density ( @xmath17 ) and velocity dispersion ( @xmath18 ) around the black hole through m _ * = -_*_*<v_*^2>^1/2 + _ * = r_d^2 ( 1 + gm_bhr_d ) + m_bh = -m _ * + l = m_bhc^2 where @xmath19 is the rate of stellar mass swallowed in the black hole and @xmath20 is an efficiency factor , of the order of @xmath21 .
of course the higher @xmath17 the higher @xmath22 while a high @xmath18 favours nucleus accretion increasing the capture time rate but , contemporarily , decreases the swallowing cross section ( 3 ) .
capuzzo dolcetta ( 1996 ) with the help of a model similar to the one now described ( but more detailed ) showed that the nucleus accretion rate can easily increase up to few @xmath23 due to stars which belonged to frictionally decayed globular clusters and to tidally disrupted clusters .
this mass accretion rate is of the order of that needed to sustain a quasar activity ( note that the high velocity bulge stars can not provide for more than @xmath24 ) .
here we report of some results of a model where the gcs is composed by clusters of same mass and the system has an initial density distribution and velocity dispersion equal to those of bulge stars in a typical triaxial galaxy .
figure 1 a ) shows the time evolution of the nucleus luminosity for different choices of the initial black hole mass @xmath25 ( @xmath26 , @xmath27 , @xmath28 , @xmath29 @xmath30 ) .
it is remarkable how , independently of @xmath25 , the highest luminosity reached , @xmath31 , is similar ( @xmath31 varies for a factor of @xmath26 when @xmath25 ranges between @xmath26 to @xmath29 @xmath30 ) .
the nucleus luminosity has a short super
eddington burst , a factor @xmath28 brighter than the luminosity when a relatively slow dimming phase starts ( the noise in the figure is real , due to the graininess of the problem amplified by the non linear coupling of source and sink term in the dynamical equations ) . of course @xmath31
is attained later for smaller values of @xmath25 , for it requires longer to accrete enough mass onto the nucleus . in all the cases the nucleus starts brightening at about @xmath32 gyr . at that age
enough massive globulars have frictionally decayed to the centre and released stars to the black hole .
a flatter increasing slope follows , due to that less massive incoming clusters feed less the nucleus .
the nucleus continues increasing its mass and becomes , rather quickly , an efficient tidal destroyer of clusters capable ( in some cases ) to keep in its potential well a large fraction of the dispersed cluster stars , which are eventually swallowed .
figure 1 b ) shows that the central black hole mass stabilizes around a value ( @xmath33 m@xmath34 in this model ) which is about the same irrespectively its initial value .
panel a ) : time evolution of the nucleus luminosity , for various initial nucleus masses ( as labeled in @xmath30 ) ; panel b ) : time evolution of the nucleus mass , for the same initial values as in panel a ) .
i showed that a possible mechanism to accrete a compact nucleus placed in the centre of a triaxial galaxy is the swallowing of surrounding stars , which belong either to dynamically decayed clusters either to tidally destroyed ones .
the quantitative relevance of this phenomenon , as well as its time scales , depend of course on the orbital structure of the gcs and on its mass and internal density spectrum : a `` cold '' system composed by massive , dilute clusters is the most affected by both dynamical friction and tidal erosion . in the model here
presented , and deeply discussed in capuzzo
dolcetta ( 1996 ) , the peak of mass accretion rate and the final value of the nucleus mass are remarkably independent ( for a fixed typical globular cluster mass ) of the initial black hole mass .
what changes is the age of occurrence of the resulting luminosity peak ( @xmath35 l@xmath34 ) and of the flattening of the mass growth curve .
it is indeed found that the black hole mass grows rapidly until it reaches a value of few @xmath36 m@xmath34 ; after that the mass accretion is much slower .
capuzzo dolcetta , r. : 1993 , 415 616 .
dolcetta , r. : 1996 , in preparation forbes , d.a . ,
franx , m. , illingworth , g.d . ,
carollo , c.m . : 1996 , 467 126 .
pesce , e. , capuzzo
dolcetta , r. , vietri , m. : 1992 , 254 466 . | globular clusters can decay to the center of their mother galaxy carrying a quantity of mass sufficient to sustain the gravitational activity of a small pre existing nucleus and to accrete it in a significant way . the agn luminosity and lifetime
depend on the characteristics of the globular cluster system and it is quite insensitive to the nucleus initial mass .
# 1 # 2 _ mem . soc .
astron .
it . _ * # 1 * , # 2 # 1 # 2 _ the messenger _ * # 1 * , # 2 # 1 # 2 _ astron .
nach . _ * # 1 * , # 2 # 1 # 2 _ astron .
astrophys .
_ * # 1 * , # 2 # 1 # 2 _ astron .
astrophys .
lett . _ * # 1 * , l#2 # 1 # 2 _ astron .
astrophys .
rev . _ * # 1 * , # 2 # 1 # 2 _ astron .
astrophys .
suppl .
ser . _ * # 1 * , # 2 # 1 # 2 _ astron .
j. _ * # 1 * , # 2 # 1 # 2 _ ann .
rev .
astron .
astrophys . _ * # 1 * , # 2 # 1 # 2 _ astrophys .
j. _ * # 1 * , # 2 # 1 # 2 _ astrophys .
j. lett .
_ * # 1 * , l#2 # 1 # 2 _ astrophys .
j. suppl .
_ * # 1 * , # 2 # 1 # 2 _ astrophys .
space sci .
_ * # 1 * , # 2 # 1 # 2 _ adv .
space res .
_ * # 1 * , # 2 # 1 # 2 _ bull .
astron .
inst .
czechosl .
_ * # 1 * , # 2 # 1 # 2 _ j. quant .
spectrosc .
. transfer _ * # 1 * , # 2 # 1 # 2 _ mon . not .
r. astr .
soc . _ * # 1 * , # 2 # 1 # 2 _ mem .
r. astr .
soc . _ * # 1 * , # 2 # 1 # 2 _ phys .
lett .
rev . _ * # 1 * , # 2 # 1 # 2 _ publ .
astron .
. japan _ * # 1 * , # 2 # 1 # 2 _ publ .
astr .
soc .
pacific _ * # 1 * , # 2 # 1 # 2 _ nature _ * # 1 * , # 2 epsf.sty |
while agn are very strong sources of electromagnetic radiation , they are also sources of outflows of matter and magnetic energy . there is a growing body of evidence that outflows are quite common and are an integral part of agn ( e.g. , * ? ? ?
* ; * ? ? ?
moreover , radiation and mass outflows in luminous sources are physically coupled as matter can absorber , scatter and emit radiation and radiation can affect dynamics and ionization of matter .
for example , powerful mass outflows in quasars are very likely winds driven by radiation from accretion disks ( e.g. , reviews by * ? ? ? * ; * ? ? ?
in addition , as we will illustrate here , the same radiation can drive outflows from large scale inflows . in this paper
, we briefly summarize results from numerical simulations of disk winds and large - scale outflows .
we review the properties of these outflows and compare them with those observed in agn .
this will allow us to assess the level of our understanding of agn outflows and their potential role in the agn feedback .
the successes of modelling outflows driven by radiation pressure on spectra lines ( line driving ) from ob stars ( e.g. , * ? ? ?
* ; * ? ? ?
* ) and from accretion disks in cataclysmic variables ( cvs ; e.g. , * ? ? ?
* ) motivate applications of a similar physics to model outflows in agn . in @xcite and @xcite
, we adopted the approach from @xcite to calculate axisymmetric time - dependent hydrodynamical models of line - driven winds from accretion disks in agn .
to apply the line - driven disk wind model developed for cvs to agn , we took into account , in an approximated way , the difference in the spectral energy distribution between agn and cvs .
in particular , we introduced three major modifications to our previous approach : 1 ) calculation of the parameters of the line force based on the wind properties , 2 ) effects of optical depth on the continuum photons , and 3 ) radiative heating and cooling of the gas .
our agn wind calculations follow ( i ) a hot and low density flow with negative radial velocity in the polar region , ( ii ) a dense , warm and fast _ equatorial _ outflow from the disk , and ( iii ) a transitional zone in which the disk outflow is hot and struggles to escape the system .
1 shows an example of a velocity field of a agn disk wind model . to make a direct comparison with observations , we calculated synthetic line profiles based on our model for the c iv @xmath01549 line @xcite .
the synthetic line profiles show a strong dependence on inclination angle : the absorption forms only when an observer looks at the ce through the fast wind ( i.e. , @xmath1 as shown in fig .
3 of @xcite ; see also fig . 2 here ) .
this @xmath2 dependence can explain why only 10% of qso have bals .
( see bottom right corner of each panel for the value of @xmath2 ) .
the solid lines show the profiles due to the absorption only whereas the dashed lines show the profiles due to the absorption and emission ( the line source function includes resonance scattering and thermal emission ) .
note the sensitivity of the lines on the inclination angle .
the zero velocity corresponding to the line center is indicated by the vertical line .
, width=491 ] the model also predicts high column densities for the inclination angles at which strong absorption lines form .
@xcite computed broad band spectra for various inclination angles using the simulation from @xcite .
3 presents some examples and shows that the model is consistent with the observations , i.e. , bal qso are underluminous in x - rays compared to their non - bal qso counterparts ( e.g. , * ? ? ?
* ; * ? ? ?
in the previous section , we showed that the powerful radiation from an agn can drive a strong wind from an accretion disk , the place where the radiation is produced . in turn
, this wind has a significant impact on the radiation .
however , the same radiation can also change the dynamics of the material at large distances from the radiation source . ) from five different lines - of - sight , with different inclinations , through the hydrodynamic simulation of an agn outflow presented in @xcite .
the spectra were calculated with xscort ( v5.18 ) using snapshot 800 of the @xcite simulation to provide self - consistent physical outflow properties .
the input ionizing power - law spectrum is shown in black .
the spectra correspond to inclinations of @xmath3 = 50 , 57 , 62 , 65 , & 67@xmath4 , ( top to bottom ) .
these l.o.s were chosen to highlight the range of spectral shapes that result from the different physical properties throughout the simulated wind .
the figure is from @xcite .
, width=264 ] we have begun studying gas dynamics in agns on subparsec and parsec scales @xcite . in the following , we briefly summarize the main results of this work . in @xcite
, we calculated a series of models for non - rotating flows that are under the influence of super massive bh gravity and radiation from an accretion disk surrounding the bh .
generally , we used the numerical methods developed by @xcite .
our numerical approach allows for the self - consistent determination of whether the flow is gravitationally captured by the bh or driven away by thermal expansion or radiation pressure . for a @xmath5 bh with an accretion luminosity of 0.6 of @xmath6 ( the same parameters as in the disk wind simulations )
, we found that a non - rotating flow settles quickly into a steady state and has two components ( 1 ) an equatorial inflow and ( 2 ) a bipolar inflow / outflow with the outflow leaving the system along the pole .
the first component is a realization of bondi - like accretion flow .
the second component is an example of a non - radial accretion flow becoming an outflow once it is pushed close to the rotational axis of the disk where thermal expansion and radiation pressure can accelerate the flow outward .
the main result of these simplified calculations is that the existence of the above two flow components is robust yet their properties are sensitive to the geometry , sed of the radiation field , and the outer boundaries . in particular , the outflow power and the degree of collimation are higher for the model with radiation dominated by uv / disk emission than for the model with radiation dominated by x - ray / central engine emission .
this sensitivity is related to the fact that thermal expansion drives a weaker and wider outflow , compared to the radiation pressure . .
the infalling gas is weakly rotating ( sub - keplerian ) , and the eddington ratio of the system is 0.6 . the volume rendering representation of the density distributions is shown .
the outflow morphology is bi - conical , but the flow contains relatively cold and dense cloud - like structures which resembles those observed in the nrls of seyfert galaxies . the figure is from @xcite . , width=283 ]
rotation of the inflowing gas changes the geometry of the flow because the centrifugal force prevents gas from reaching the rotational axis ( see fig . 1 in @xcite ) .
this , in turn , reduces the mass outflow rate because less gas is pushed toward the polar region .
we also found that rotation can lead to fragmentation and time variability of the outflow .
as the flow fragments , cold and dense clouds form ( figs . 4 and 12 in @xcite ) .
three - dimensional effects are also important . in @xcite , we considered effects of radiation due to a precessing accretion disk on a spherical cloud of gas around the disk .
on the other hand , in @xcite , we recalculated some models from papers @xcite and @xcite in full 3-d .
our 3-d simulations of a nonrotating gas show small yet noticeable nonaxisymmetric small - scale features inside the outflow .
however , the outflow as a whole and the inflow do not seem to suffer from any large - scale instability . in the rotating case ,
the nonaxisymmetric features are very prominent , especially in the outflow which consists of many cold dense clouds entrained in a smoother hot flow ( e.g. , see figs . 4 and 5 ) .
the 3-d outflow is nonaxisymmetric due to the shear and thermal instabilities .
effects of gas rotations are similar in 2-d and 3-d .
in particular , gas rotation increases the outflow thermal energy flux , but reduces the outflow mass and kinetic energy fluxes .
in addition , rotation leads to time variability and fragmentation of the outflow in the radial and latitudinal directions .
the collimation of the outflow is reduced in the models with gas rotation .
the main difference between the models with rotation in 3-d and 2-d is that the time variability in the mass and energy fluxes is reduced in the 3-d case because of the outflow fragmentation in the azimuthal direction .
@xmath7 plane .
the cold clouds here are defined as the gas with its density higher than @xmath8 and its temperature less than @xmath9 .
the clouds are not spherically distributed , but located near the bi - conic surface ( which appears as an x - shaped pattern here ) defined by the outflowing gas .
note that the length scale are in units of pc .
the figure is from @xcite .
, width=283 ] to be able to compare these new simulations with observations , we are in a process of computing synthetic line profiles , broad band spectra and maps .
however , even without these diagnostics we can check if the models are consistent with the data .
for example , we can compare the kinematics study of ngc 4151 @xcite with the velocity of the cold clouds ( cf .
5 ) projected ( @xmath10 ) toward an observer at the inclination angle @xmath11 , which is the inclination of ngc 4151 .
@xcite used the kinematics model of the outflows with a bi - conic radial velocity law , and found a good fit to their observations when the opening angle of the cone is @xmath12 .
interestingly , we find the opening angle of the outflows in our is also about @xmath13 ( cf . figs . 5 ) .
figure 6 shows @xmath10 of the clouds plotted as a function of the projected vertical distance , which is the distance along the @xmath14-axis in fig . 5 projected onto the plane of the sky for an observer viewing the system with @xmath11 .
the figure shows that the clouds are accelerated up to @xmath15 until the projected distance reaches @xmath16 pc , but the velocity curve starts to flatten beyond this point . towards the outer edges ( near the outer boundaries ) ,
the curve begins to show a sign of deceleration , but not so clearly .
we note that the hot outflowing gas , on the other hand , does show deceleration at the larger radii in our models ( cf .
fig . 9 in * ? ? ?
although the physical size of the long slit observation of ngc 4151 by @xcite is in much lager scale ( @xmath17 times larger ) than that of our model , their radial velocities as a function of the position along the slit ( see their figs . 5 and 6 ) show a similar pattern as in our model ( fig . 6 ) .
the range of @xmath10 in our model is about @xmath18 to @xmath19 while the range of the observed radial velocities in @xcite is about @xmath20 to @xmath21 , which is comparable to ours .
are shown as a function of the projected vertical distance ( the distance along the @xmath14-axis in fig . 5 , but projected on to the plane of the sky for the observer viewing the system with @xmath11 ) .
the negative projected distance indicates the clouds are found in the lower half of the projection plane .
the figure is from @xcite .
, width=302 ]
the models presented here , which numerically simulate the outflows driven by radiation from agn , are in many respects consistent with observations . clearly more work is needed to test the models and improve them .
however , the first few steps toward the development a self - consistent physical model of the agn outflows has been taken .
applications of the model are not limited to agn because other astrophysical objects such as x - ray binaries , in particular micro - quasars have a similar geometry and can be understood within a similar physical framework .
in addition , having a physical model of the agn outflows we can apply it to the so - called agn feedback problem .
some results from the outflow models have been already incorporated to galaxy evolution calculations . in particular ,
@xcite , improved and extended the accretion and feedback physics explored in their previous papers ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) . by using a high - resolution one - dimensional hydrodynamical code
, @xcite studied , the evolution of an isolated elliptical galaxy , where the cooling and heating functions include photoionization and compton effects , and restricting to models which include only radiative or only mechanical feedback ( the latter , in the form of disk winds properties of which were adopted from * ? ? ? * ; * ? ? ? * ) .
these recent calculations confirmed that for eddington ratios above 0.01 , both accretion and radiative outputs are forced by feedback effects to be in burst mode , so that strong intermittencies are expected at early times , while at low redshift the explored models are characterized by smooth , very sub - eddington mass accretion rates punctuated by rare outbursts .
however , the explored models always fail some observational tests . for the high mechanical efficiency of @xmath22
as adopted by some investigators , it was found that most of the gas is ejected from the galaxy , the resulting x - ray luminosity is less than a value typically observed and little super massive black hole growth occurs .
however , models with low enough mechanical efficiency to accommodate satisfactory black hole growth tend to allow too strong cooling flows and leave galaxies at z = 0 with e+a spectra more frequently than is observed .
surprisingly , it was also found that both types of feedback are required .
radiative heating over the inner few kilo parsecs is needed to prevent calamitous cooling flows , and mechanical feedback from active galactic nucleus winds , which affects primarily the inner few hundred parsecs , is needed to moderate the luminosity and growth of the central black hole .
models with combined feedback pass more observational tests ( ciotti et al . , in preparation ) .
thus it emerges from these simulations that to solve the agn feedback problem / explain all aspects of the galaxy evolution , more than one form of feedback is needed .
this work was supported by nasa through grant hst - ar-11276 from the space telescope science institute , which is operated by the association of universities for research in astronomy , inc .
, under nasa contract nas5 - 26555 .
dp acknowledges also support from nsf ( grant ast-0807491 ) .
crenshaw , d.m . ,
kraemer , s.b .
& george i.m .
mass outflow in active galactic nuclei : new perspectives , asp conference proceedings , vol .
255 . edited by d. m. crenshaw , s. b. kraemer , and i. m. george , san francisco : astronomical society of the pacific , 2002 richards , g.t .
& hall , p.b 2004 , agn physics with the sloan digital sky survey , asp conference series , volume 311 , edited by gordon t. richards and patrick b. hall , san francisco : astronomical society of the pacific , 2004 | we present a brief summary of the main results from our multi - dimensional , time - dependent simulations of gas dynamics in agn .
we focus on two types of outflows powered by radiation emitted from the agn : disk winds and winds driven from large - scale inflows .
we show spectra predicted by the simulations and discuss their relevance to observations of broad- and narrow - line regions of the agn .
we finish with a few remarks on whether these outflows can have a significant impact on their environment and host galaxy .
= = = 1=1=0pt = 2=2=0pt |
adjoint functors , since their discovery by daniel kan @xcite in 1958 , have become quite ubiquitous in mathematics , with their universality well - documented already in the wikipedia .
we hope that biadjoint functors , which are pairs of functors @xmath0 such that @xmath1 is both right and left adjoint of @xmath2 , will prove to be of importance as well .
let us begin by reviewing the topological meaning of adjointness and biadjointness .
we will depict a functor @xmath3 by a mark on a horizontal interval , with the half - intervals to the right and left of the mark labelled by @xmath4 and @xmath5 , respectively .
composition @xmath6 of functors @xmath7 is depicted by placing marks for @xmath8 in a row , with the intervals labelled by categories @xmath9 reading from right to left .
the identity natural transformation @xmath10 is depicted by a vertical interval drawn in a rectangular region of the plane , with the two areas labelled by categories @xmath4 and @xmath5 .
a natural transformation @xmath11 , where @xmath12 are functors from @xmath4 to @xmath5 , is depicted by a dot on a vertical line .
a natural transformation @xmath13 can be depicted by merging lines for the identity transformations of @xmath8 into a point and then splitting it into lines for the identities of @xmath14 .
it is useful not to label the identity functor @xmath15 by anything and instead depict it by a horizontal interval labelled @xmath4 .
likewise , the identity natural transformation of this functor is denoted by a region labelled @xmath4 . with these rules
, we can depict @xmath16 by the following picture .
here @xmath5 is the target category for functor @xmath17 and the source category for functor @xmath18 .
similarly , below is a picture for @xmath19 , where @xmath1 is an endofunctor of @xmath4 .
the two possible types of compositions of natural transformations are depicted by either placing diagram in parallel or stacking them vertically ; the next picture is an example of horizontal composition of three compatible natural transformations .
an example of vertical composition of two natural transformations , with some identity transformations thrown in , is given below .
it is important that strands always go `` up '' , that is , strands are not yet allowed to have u - turns .
isotopies of strands are allowed , as long as they do nt create u - turns . on some very informal level , these diagrams are analogous to planar projections of braids ( braids do nt have u - turns either ) .
thus , 2-dimensional planar pictures denote natural transformations , with regions of the picture labelled by categories , strands by functors and nodes by natural transformations .
a planar diagram without boundary points ( a closed diagram ) determines an endomorphism of @xmath15 , where @xmath4 is the label of the outside region , i.e. , an element of the center of category @xmath4 .
this setup , sometimes called string notation for 2-categories , is poincare dual to the more common one where categories are depicted by points , functors by arrows and natural transformations by 2-cells .
the same setup can be used with any 2-category in lieu of the 2-category of natural transformations .
regions of the diagrams will be labelled by objects of the 2-category , edges by 1-morphisms , and nodes by 2-morphisms .
a monoidal category is a 2-category with a single object . in this case
we can avoid labelling regions , denote tensor product of objects of a monoidal category by a sequence of marks on a line and morphisms between tensor products by nodes with source and target arrows ( there is a standard way of dealing with the case when the monoidal category is not strict ; we omit the details ) .
this notation is common in the diagrammatics for the representation categories of simple lie algebras and quantum groups .
we now come to adjointness .
functors @xmath20 and @xmath21 are adjoint if there are isomorphisms @xmath22 functorial in @xmath23 and @xmath24 .
an adjunction is equivalent to having the unit and the counit @xmath25 natural transformations that satisfy the following equations : * natural transformation @xmath26 equals @xmath27 , * natural transformation @xmath28 equals @xmath29 .
let us depict @xmath30 and @xmath31 by a cup and a cap diagram , respectively . to distinguish between functors @xmath1 and @xmath32 we equip the strands with up orientation near @xmath1 and down orientation near @xmath32 .
the equations turn into relations on planar diagrams these relations have a natural interpretation via isotopies of arcs , but give us only two out of four basic isotopy relations on oriented arcs in the plane .
to get complete isotopy invariance , we assume that @xmath1 is also a right adjoint to @xmath32 and the natural transformations describing the second adjointness are fixed @xmath33 they satisfy the relations * natural transformation @xmath34 equals @xmath27 , * natural transformation @xmath35 equals @xmath29 , whose graphical interpretation is that of the two remaining types of arc isotopies : thus , given a biadjoint pair @xmath36 we can make sense out of any diagram built from the four possible @xmath36-cups and caps .
we say that @xmath32 is a biadjoint of @xmath1 and @xmath1 a biadjoint of @xmath32 .
a functor admitting a biadjoint is also called a _
functor , and a biadjoint pair @xmath36 is called a _
frobenius pair_. yet another terminology for a biadjoint pair is _ ambidextrous adjunction _
one of the first interesting examples of biadjoint pairs that appeared in mathematics ( specifically , in the modular representation theory ) were induction and restriction functors for inclusions of finite groups . for a simpler example , take any invertible functor : its inverse is its biadjoint .
one often works with many pairs of biadjoint functors , which requires compatibility , namely given biadjunctions @xmath37 and @xmath38 such that @xmath39 is defined , the biadjunction for @xmath40 should be built in the obvious diagrammatic way via the composition rules .
biadjointness allows us to move dots on strands through cups and caps .
say , we have a transformation @xmath41 and biadjoint pairs @xmath42 .
there exists a unique transformation @xmath43 that satisfies the equality @xmath44 is defined similarly , via the equality from the topological viewpoint , it is convenient to require @xmath45 for all natural transformations @xmath46 between our pairs of biadjoint functors , for then @xmath46 always turns into @xmath47 or @xmath46 no matter what sequence of caps and cups it goes through .
we refer to this property as _ cyclic biadjointness _
( also called _ even - handed structure _ in @xcite ) . with these assumptions ,
there is now a complete isotopy invariance for the string diagrams .
functors for these diagrams are selected from a collection of cyclic biadjoints that satisfy the above compatibility condition for their compositions . any natural transformation between compositions of these functors
can potentially be depicted .
some are glued out of the basic ones , others require introducing new nodes with multiple input and output strands .
once a new node for a natural transformation between composition appears , it may be used as a building block for more complicated planar diagrams . due to cyclicity
, these nodes can be isotoped in the plane . _
remark : _ if the cyclic order of functors around the node has a rotational symmetry , it might not extend to a symmetry of the node ( but in some natural examples it does ) . for instance ,
given a node for @xmath48 where @xmath49 is a biadjoint pair ( of course , @xmath50 is then an endofunctor ) , @xmath51 does not have to be equal to @xmath46 .
the above discussion generalizes , allowing us to depict elements of strict 2-categories with suitable duality properties , mirroring those of cyclic biadjointness .
the planar interpretation of biadjointness has been a folklore for a number of years and appeared in @xcite .
string notation , in the case of 2-categories with one object ( monoidal categories ) , was introduced and made rigorous by joyal and street @xcite .
nowadays , it can be found on youtube , in a series of videos `` string diagrams '' by thecatsters .
planar diagrams of lines labelled by functors and natural transformations can be thought of as suitably decorated 1-dimensional cobordisms embedded in @xmath52 .
there is also a direct relation of biadjoint functors to cobordisms in all dimensions .
let @xmath53 be the 2-category of @xmath54-dimensional smooth cobordisms with boundary and corners .
closed @xmath55-dimensional manifolds @xmath56 are objects of @xmath53 , while an @xmath57-dimensional cobordism @xmath58 is a 1-morphism from object @xmath59 to object @xmath60 .
an @xmath54-dimensional cobordism @xmath61 with corners is a 2-morphism from @xmath62 to @xmath63 . here
@xmath64 are @xmath57-manifolds with boundary , and the boundary of @xmath61 consists of 4 pieces : @xmath64 , and products @xmath65 $ ] , @xmath66.$ ] corners of @xmath61 are four @xmath55-manifolds @xmath67 @xmath68 . for each 1-morphism @xmath58 there is the reverse 1-morphism @xmath69 from @xmath60 to @xmath59 given by flipping @xmath58 .
we claim that @xmath70 constitutes a cyclic biadjoint pair .
multiply @xmath58 by an interval and then fan out the resulting @xmath54-manifold with corners so that it gives two @xmath54-cobordisms between @xmath71 and @xmath72 ( one in each direction ) and two @xmath54-cobordisms between @xmath73 and @xmath74 .
the relations on these four cobordisms are exactly the same as those satisfied by natural transformations @xmath75 of a biadjoint pair ( for more details and pictures see ( * ? ? ?
* section 6.3 ) ) .
in particular , given a 2-functor @xmath1 from @xmath53 into the 2-category of categories , functors , and natural transformations , the functors @xmath76 and @xmath77 are canonically biadjoint . if the target of @xmath1 is the 2-category of additive categories and additive functors , we say that @xmath1 is an @xmath54-dimensional tqft with corners ( or extended tqft ) .
most of the time we require manifolds to be oriented , often additionally decorated by some structure , such as a spin structure , and sometimes place restrictions on topological type of manifolds and cobordisms .
there are interesting mathematically understood examples in dimensions @xmath78 in dimension 2 , they come from commutative frobenius algebras . in dimension 3 , the witten - reshetikhin - turaev tqft and its relatives emerge from chern - simons theory and representation theory of quantum groups , while the rozansky - witten tqft comes from the derived category of coherent sheaves on holomorphic symplectic manifolds .
donaldson - floer , seiberg - witten and heegaard floer theory ( all closely interrelated ) provide famous examples of 4d tqfts .
any extended tqft produces an abundance of biadjoint pairs , namely one pair @xmath79 for each @xmath57-cobordism @xmath58 .
these biadjoint functors go between categories @xmath80 assigned to closed @xmath55-manifolds @xmath56 . when searching for tqfts , it is useful to look for collections of categories that admit many biadjoint functors between each other
. examples of such collections include : * derived categories of coherent sheaves on a calabi - yau variety .
any sheaf on the product @xmath81 induces a pair of convolution functors between @xmath82 and @xmath83 . when @xmath84 , this is a biadjoint pair .
when dimensions do nt match , the functors are almost biadjoint ( biadjoint up to a shift in the derived category ) .
* categories of modules over finite - dimensional symmetric algebras and their derived categories .
the functor of tensoring with a finitely - generated bimodule over symmetric algebras which is left and right projective has a bijadjoint .
* fukaya - floer categories of symplectic manifolds . * derived categories of sheaves on flag varieties . * various categories that appear in categorification of representations of hecke algebras and quantum groups .
for instance , functor @xmath85 that categorifies generator @xmath86 of a simple lie algebra / quantum group is biadjoint ( or almost biadjoint ) to the functor @xmath87 categorifying generator @xmath88 , see @xcite .
some of the earliest examples of categorifications of hecke algebra and @xmath89 representations @xcite came from highest weight categories of representations of @xmath90 , with @xmath86 and @xmath88 being translation functors ( direct summands of tensor products with finite - dimensional modules ) or zuckerman functors .
biadjointness of translation functors was used already in @xcite . in ariki
s categorification @xcite of irreducible @xmath89 and affine @xmath89 representations , functors @xmath85 and @xmath87 are biadjoint as well , being induction and restriction functors between ariki - koike - cherednik cyclotomic quotients of affine hecke algebras .
the issue of biadjointness in the reshetikhin - turaev - witten tqft is not prominent , since the category associated to the circle is semisimple @xmath91-linear and functors associated to surfaces with boundary are @xmath91-linear as well .
such functors are guaranteed to have biadjoints , which can be described in a simple combinatorial way .
in contrast , the category associated to the circle in the rozansky - witten tqft is a version of the derived category of coherent sheaves on a holomorphic symplectic manifold @xmath92 , which is necessarily calabi - yau , so that one of the examples on the above list becomes relevant .
triangulated categories behind extended donaldson - floer and ozsvth - szab theories @xcite are assigned to surfaces and closely related to fukaya - floer categories of the representation variety of the fundamental group of a surface and of the symmetric power of a surface , respectively .
extended 4d tqft that controls ozsvth - szab 3-manifold homology and 4-manifold invariants is being unravelled by lipshitz , ozsvth and thurston @xcite .
we list a few nice features of biadjoint pairs : * a biadjoint functor commutes with both limits and colimits . * if a functor @xmath1 between additive categories has left or right adjoint , it is additive . in particular , in biadjoint
pairs @xmath36 between additive categories both @xmath1 and @xmath32 are additive functors . * in a biadjoint pair @xmath36 of functors between abelian categories both @xmath1 and @xmath32 are additive and exact , take projectives to projectives and injectives to injectives . to an abelian category @xmath4
there is assigned its grothendieck group @xmath93 , an abelian group with generators given by symbols @xmath94 $ ] of objects of @xmath4 and defining relations @xmath94=[m']+[m'']$ ] for each short exact sequence @xmath95 often , it also makes sense to consider the group @xmath96 generated by symbols of projective objects @xmath97 $ ] modulo relations @xmath97=[p']+[p'']$ ] if @xmath98 .
the obvious homomorphism @xmath99 is , in general , neither injective nor surjective .
the homomorphism takes @xmath97 $ ] to @xmath97 $ ] , so this symbol notation might be ambiguous on projectives .
a biadjoint pair @xmath36 induces homomorphisms @xmath100 \ : \
k_0(\mc{a } ) \lra k_0(\mc{b } ) , \hspace{0.2 in } g_0(\mc{a } ) \lra g_0(\mc{b}),\ ] ] @xmath101 \ : \ k_0(\mc{b } ) \lra k_0(\mc{a } ) , \hspace{0.2 in } g_0(\mc{b } ) \lra g_0(\mc{a})\ ] ] that commute with @xmath102 .
these homomorphisms take @xmath94 $ ] to @xmath103 $ ] , respectively @xmath104 $ ] .
in contrast , an exact functor @xmath1 would induce a homomorphism @xmath105 but not necessarily a homomorphism between @xmath106 s , since it might not take projectives to projectives . likewise , a functor taking projectives to projectives would induce a homomorphism on @xmath106 s but not on @xmath107 s .
biadjoint pairs between abelian categories give the best behaving functors from this perspective .
they can be thought of as categorifying pairs of adjoint operators .
namely , assume that @xmath108 are @xmath109-linear , over a field @xmath109 , and hom spaces between objects are finite - dimensional over @xmath109 .
this gives bilinear forms on @xmath110 @xmath111 determined by @xmath112,[q])_{\mc{a } } : = \dim { { \rm hom}}_{\mc{a}}(p , q)\ ] ] for projectives @xmath113 , likewise for @xmath5 .
adjointness isomorphisms @xmath114 descend to relations @xmath115v , w)_{\mc{b } } = ( v , [ g]w)_{\mc{a } } , \ \ ( w,[f]v)_{\mc{b } }
= ( [ g]w , v)_{\mc{a } } , \ \
v\in k_0(\mc{a } ) , w\in k_0(\mc{b}),\ ] ] showing that @xmath116 $ ] and @xmath117 $ ] become adjoint operators on real vector spaces @xmath118 and @xmath119 relative to these two bilinear forms ( in interesting examples the forms are often symmetric and nondegenerate ) .
the iwahori - hecke algebra @xmath120 of the symmetric group @xmath121 has generators @xmath122 , @xmath123 , and defining relations @xmath124 where @xmath125 is a formal parameter . in present - day mathematics
it appears in two seemingly different ways : \i ) as a finite - dimensional quotient of the group algebra of the braid group , providing invariants of braid and links , when the latter are realized as closures of braids .
two braid closures produce the same link if they are related by a finite sequence of markov moves .
the algebraic counterpart of the closure operation is taking trace of an operator .
the ocneanu trace on the hecke algebra behaves well under the markov moves and can be normalized to produce an invariant of links known as the homfly - pt polynomial ( named with the initials of eight people who independently discovered it ) .
more algebraically , the hecke algebra is the commutant of the action of the quantum group @xmath126 on @xmath127 where @xmath128 and @xmath129 .
\ii ) as the endomorphism algebra of the representation of @xmath130 induced from the trivial representation of the subgroup @xmath5 of all invertible upper - triangular matrices . here
@xmath130 is the finite group of invertible @xmath131 matrices with coefficients in the finite field @xmath132 with @xmath125 elements , hence in this interpretation @xmath125 is no longer a formal variable .
case @xmath133 also fits into this framework , corresponding to @xmath134 over the one - element field , which is the symmetric group ( and the iwahori - hecke algebra specializes to the group algebra of the symmetric group when @xmath133 ) .
construction i ) of the hecke algebra is fundamental for low - dimensional topology , construction ii ) and its generalizations is indispensable in representation theory .
the second construction also leads to a categorification of the iwahori - hecke algebra .
the functions on the finite set @xmath135 become sheaves on the flag variety @xmath136 , either in the etale topology over a finite field , or sheaves of @xmath91-vector spaces on the flag variety over the field @xmath91 .
categorification of @xmath120 was constructed by soergel @xcite , who also stated it in a sheaf - free language ( which works for arbitrary weyl groups ) .
we describe it here in the symmetric group case .
it is convenient to introduce @xmath137 and view @xmath120 as a @xmath138$]-algebra .
then @xmath120 is also generated by @xmath139 , @xmath140 , with defining relations @xmath141 soergel s categorification of @xmath120 is built out of bimodules over the polynomial ring @xmath142 $ ] .
the ring is graded , @xmath143 , and all the bimodules are graded as well . grading shift @xmath144 up by one is an automorphism of the category of graded bimodules . for each @xmath140 let @xmath145 be the subring that consists of polynomials invariant under the transposition @xmath146 .
then @xmath147 is a graded @xmath148-bimodule and we let @xmath149 form the category @xmath150 whose objects are tensor products of @xmath151 s and morphisms are homomorphisms of bimodules . by adding finite direct sums , grading shifts , restricting to grading - preserving homomorphisms , and forming the karoubi envelope , one arrives at the category @xmath152 .
we call the objects of this category soergel bimodules
. one of soergel s results is that @xmath152 categorifies the hecke algebra @xmath120 , i.e. the grothendieck ring of @xmath152 is canonically isomorphic to @xmath120 , @xmath153 with @xmath154 $ ] , the symbol of @xmath151 , going to @xmath155 under the isomorphism .
grading shift corresponds to multiplication by @xmath156 : @xmath157= t [ m]$ ] .
multiplication in @xmath120 corresponds to the tensor product of bimodules , @xmath158\cdot [ n ] \ \ : = \ \ [ m\otimes_r n].\ ] ] indecomposables in @xmath152 are in a bijection , up to grading shift , with elements @xmath159 of the symmetric group .
bimodule @xmath160 , where @xmath161 is the trivial permutation , bimodules @xmath162 where @xmath163 , and , inductively , if @xmath164 , @xmath165 is the only indecomposable summand of @xmath166 that is not isomorphic , up to a grading shift , to @xmath167 for some @xmath168 with @xmath169 .
equivalently , @xmath170 is determined by the condition that it appears as a direct summand of @xmath171 , where @xmath172 and @xmath173 is a reduced presentation of @xmath159 , and does not appear as direct summand of any @xmath174 , for sequences @xmath175 of length less than @xmath176 .
defining relations on @xmath155 s become isomorphisms of graded bimodules @xmath177 the last isomorphism comes from decompositions @xmath178 here @xmath179 is the tensor product of two @xmath148 s over the subring of @xmath180-invariants , with the action of @xmath180 by permutations of @xmath181 .
when @xmath182 , the soergel category has 6 indecomposables , up to grading shifts .
they are @xmath183 their images in the grothendieck ring @xmath184 are @xmath185 n. libedinsky @xcite presented @xmath152 in the @xmath186 case ( and , more generally , in the so - called right - angled case , that we do nt discuss ) via generators and relations on morphisms .
we now explain , following @xcite , a generalization of his presentation to an arbitrary @xmath55 .
this approach provides a graphical description of homomorphisms between tensor products of soergel bimodules .
tensoring with a soergel bimodule is an endofunctor in the category of ( graded ) @xmath148-modules and homomorphisms between tensor products corresponds to natural transformations of functors .
start with the simplest soergel bimodule @xmath148 .
the ring @xmath148 is commutative and multiplication by any element of @xmath148 is an endomorphism of the identity functor ( tensoring with @xmath148 ) , thus belongs to the center of the category of @xmath148-modules .
we denote by a box labelled @xmath187 the multiplication by the generator @xmath188 of @xmath148 .
one can think of this box as freely floating in a region .
boxes can float past each other in any direction ( relative height change isotopy corresponds to commutativity of central elements ) .
a collection of floating boxes denotes the product of corresponding generators .
any element of @xmath148 is a @xmath91-linear combinations of products of boxes .
a vertical line labelled @xmath187 will denote the identity endomorphism of the bimodule @xmath151 .
an important feature of @xmath151 is that it is selfadjoint ( in the category of soergel bimodules ) , i.e. left and right adjoint to itself , and this selfadjointness is cyclic .
therefore , we can introduce unoriented cup and cap diagrams labelled by @xmath187 to denote units and counits of biadjunctions ( in pictures below , label @xmath187 is omitted ) .
these diagrams have zero degree .
moreover , @xmath151 is a frobenius object .
namely , there are homomorphisms @xmath189 of degrees @xmath190 respectively , that we depict by @xmath191{dot.eps } } \endxy } } \hspace{0.4 in } { \vcenter{\xy ( 0,0)*{\reflectbox{\includegraphics[scale=.3 , angle=180]{dot.eps } } } \endxy } } \hspace{0.4 in } { \vcenter{\xy ( 0,0)*{\reflectbox{\includegraphics[scale=.3 , angle=180]{merge.eps } } } \endxy}}\hspace{0.4 in } { \vcenter{\xy ( 0,0)*{\includegraphics[scale=.3]{merge.eps } } \endxy}}\ ] ] and that satisfy the axioms of a frobenius algebra object : @xmath192{doublemerge.eps } } \endxy } } = { \vcenter{\xy ( 0,0)*{\reflectbox{\includegraphics[scale=.2]{doublemerge.eps } } } \endxy } } \hspace{0.4 in } { \vcenter{\xy ( 0,0)*{\reflectbox{\includegraphics[scale=.2 , angle=180]{doublemerge.eps } } } \endxy } } = { \vcenter{\xy ( 0,0)*{\includegraphics[scale=.2 , angle=180]{doublemerge.eps } } \endxy}}\ ] ] @xmath192{splitdot.eps } } \endxy } } = { \vcenter{\xy ( 0,0)*{\includegraphics[scale=.5]{line.eps } } \endxy } } = { \vcenter{\xy ( 0,0)*{\reflectbox{\includegraphics[scale=.2]{splitdot.eps } } } \endxy } } \hspace{0.4 in } { \vcenter{\xy ( 0,0)*{\reflectbox{\includegraphics[scale=.2 , angle=180]{splitdot.eps } } } \endxy } } = { \vcenter{\xy ( 0,0)*{\includegraphics[scale=.5]{line.eps } } \endxy } } = { \vcenter{\xy ( 0,0)*{\includegraphics[scale=.2 , angle=180]{splitdot.eps } } \endxy}}\ ] ] @xmath193{cupcap.eps } } \endxy } } \;\ ; = \;\ ; { \vcenter{\xy ( 0,0)*{\includegraphics[scale=.2]{tallline.eps } } \endxy } } \;\ ; = \;\ ; { \vcenter{\xy ( 0,0)*{\reflectbox{\includegraphics[scale=.25]{cupcap.eps } } } \endxy}}\hspace{0.3 in } { \vcenter{\xy ( 0,0)*{\includegraphics[scale=.3]{splitcup.eps } } \endxy } } \ =
\ { \vcenter{\xy ( 0,0)*{\includegraphics[scale=.3]{tallmerge.eps } } \endxy } } \ = \
{ \vcenter{\xy ( 0,0)*{\reflectbox{\includegraphics[scale=.3]{splitcup.eps } } } \endxy}}\ ] ] @xmath191{capdot.eps } } \endxy } } \
= \ { \vcenter{\xy ( 0,0)*{\reflectbox{\includegraphics[scale=.4 , angle=180]{dot.eps } } } \endxy } } \
= \ { \vcenter{\xy ( 0,0)*{\reflectbox{\includegraphics[scale=.3]{capdot.eps } } } \endxy } } \label{twistdot1}\ ] ] @xmath194{cap.eps } } } \endxy } } \ \ = \ \ { \vcenter{\xy ( 0,0)*{\reflectbox{\includegraphics[scale=.3 , angle=180]{dotsplit.eps } } } \endxy } } \hspace{0.2 in } { \vcenter{\xy ( 0,0)*{\includegraphics[scale=.4]{cap.eps } } \endxy } } \ \ = \ \ { \vcenter{\xy ( 0,0)*{\includegraphics[scale=.4]{dotsplit.eps } } \endxy } } \label{twistdot4}\ ] ] we list other relations that involve strands of one color only : @xmath195{multxi.eps } } \endxy}}\ { \vcenter{\xy ( 0,0)*{\includegraphics[scale=.6]{line.eps } } \endxy}}\ + \ { \vcenter{\xy ( 0,0)*{\includegraphics[scale=.4]{multxp.eps } } \endxy}}\ { \vcenter{\xy ( 0,0)*{\includegraphics[scale=.6]{line.eps } } \endxy}}\ & \
= \ & \ { \vcenter{\xy ( 0,0)*{\includegraphics[scale=.6]{line.eps } } \endxy}}\ { \vcenter{\xy ( 0,0)*{\includegraphics[scale=.4]{multxi.eps } } \endxy}}\ + \ { \vcenter{\xy ( 0,0)*{\includegraphics[scale=.6]{line.eps } } \endxy}}\ { \vcenter{\xy ( 0,0)*{\includegraphics[scale=.4]{multxp.eps } } \endxy } } \label{eq - slide1 } \\ \
\ & & \ \ \nonumber \\ { \vcenter{\xy ( 0,0)*{\includegraphics[scale=.4]{multxi.eps } } \endxy}}\ { \vcenter{\xy ( 0,0)*{\includegraphics[scale=.4]{multxp.eps } } \endxy}}\ { \vcenter{\xy ( 0,0)*{\includegraphics[scale=.6]{line.eps } } \endxy}}\ & \ = \ & \ { \vcenter{\xy ( 0,0)*{\includegraphics[scale=.6]{line.eps } } \endxy}}\ { \vcenter{\xy ( 0,0)*{\includegraphics[scale=.4]{multxi.eps } } \endxy}}\ { \vcenter{\xy ( 0,0)*{\includegraphics[scale=.4]{multxp.eps } } \endxy } } \label{eq - slide2 } \\ \
\ & & \ \
\\ { \vcenter{\xy ( 0,0)*{\includegraphics[scale=.4]{multxj.eps } } \endxy}}\ { \vcenter{\xy ( 0,0)*{\includegraphics[scale=.6]{line.eps } } \endxy}}\ & \ = \ & \ { \vcenter{\xy ( 0,0)*{\includegraphics[scale=.6]{line.eps } } \endxy}}\ { \vcenter{\xy ( 0,0)*{\includegraphics[scale=.4]{multxj.eps } } \endxy } } , \hspace{0.2 in } |i - j|>1 .
\label{eq - slide3}\end{aligned}\ ] ] these come directly from the definition of @xmath151 .
furthermore , we impose @xmath196{dotspacedot.eps } } \endxy } } \
= \ { \vcenter{\xy ( 0,0)*{\includegraphics[scale=.4]{multxi.eps } } \endxy}}\ { \vcenter{\xy ( 0,0)*{\includegraphics[scale=.75]{line.eps } } \endxy } } \ - \ { \vcenter{\xy ( 0,0)*{\includegraphics[scale=.75]{line.eps } } \endxy}}\ { \vcenter{\xy ( 0,0)*{\includegraphics[scale=.4]{multxp.eps } } \endxy } } \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ { \vcenter{\xy ( 0,0)*{\includegraphics[scale=.5]{doubledot.eps } } \endxy } } \
= \ { \vcenter{\xy ( 0,0)*{\includegraphics[scale=.4]{multxi.eps } } \endxy } } \ - \
{ \vcenter{\xy ( 0,0)*{\includegraphics[scale=.4]{multxp.eps } } \endxy}}\ ] ] @xmath197{needle2.eps } } \endxy } } \ \ = \ \ 0\ ] ] these relations imply , in particular , @xmath198{twolines.eps } } \endxy } } & \ \ = \ \ { \vcenter{\xy ( 0,0)*{\includegraphics[scale=.4]{twolinescalc2.eps } } \endxy } } \ \ = \ { \vcenter{\xy ( 0,0)*{\includegraphics[scale=.4]{twolinescalc3.eps } } \endxy } } \ - \
{ \vcenter{\xy ( 0,0)*{\includegraphics[scale=.4]{twolinescalc4.eps } } \endxy } } \label{twolines}\\ \
\nonumber \\ & = \ { \vcenter{\xy ( 0,0)*{\includegraphics[scale=.4]{twolinesfinal1.eps } } \endxy } } \ -\ { \vcenter{\xy ( 0,0)*{\includegraphics[scale=.4]{twolinesfinal2.eps } } \endxy } } \nonumber \end{aligned}\ ] ] leading to the decomposition @xmath199 .
we now describe generators and relations for interactions of two adjacent colors @xmath187 and @xmath200 .
below , thin lines represent @xmath151 and thick lines @xmath201 .
the six - valent vertex denotes the composition @xmath202 of projection from the tensor product of 3 bimodules onto its most interesting direct summand and inclusion of this summand into the other triple tensor product .
it turns out that the 6-valent vertex possesses full rotational invariance : @xmath203{ipipiprot1.eps } } \endxy } } \ \ \ = \ \ \ { \vcenter{\xy ( 0,0)*{\includegraphics[scale=.4]{6valent.eps } } \endxy } } \ \ \ = \ \ \ { \vcenter{\xy ( 0,0)*{\reflectbox{\includegraphics[scale=.32]{ipipiprot1.eps } } } \endxy } } \label{eqn - ipipiprot}\ ] ] direct sum decompositions of @xmath204 and @xmath205 are encoded by the relation @xmath191{threelines1.eps } } \endxy}}\ \ = \ \ { \vcenter{\xy ( 0,0)*{\includegraphics[scale=.3]{threelines4.eps } } \endxy } } \ \ - \ \ { \vcenter{\xy ( 0,0)*{\includegraphics[scale=.3]{threelines3.eps } } \endxy } } \label{eqn - threelines}\ ] ] and its relative given by reversing the thickness of lines .
other relations are ( add their reverses as well ) @xmath191{ipipipdot1.eps } } \endxy } } \ \ = \ \ \ { \vcenter{\xy ( 0,0)*{\includegraphics[scale=.3]{ipipipdot2.eps } } \endxy } } \ \ + \ \ { \vcenter{\xy ( 0,0)*{\reflectbox{\includegraphics[scale=.3]{ipipipdot3.eps } } } \endxy } } \label{eqn - ipipipdot}\ ] ] @xmath191{ipipipass.eps } } \endxy } } \ \ = \ \ { \vcenter{\xy ( 0,0)*{\includegraphics[scale=.3 , angle=90]{ipipipass.eps } } \endxy } } \ \ = \ \ { \vcenter{\xy ( 0,0)*{\includegraphics[scale=.3]{ipipipass2.eps } } \endxy } } \ \ = \ \ { \vcenter{\xy ( 0,0)*{\reflectbox{\includegraphics[scale=.3]{ipipipass2.eps } } } \endxy } } \label{eqn - ipipipass}\ ] ] @xmath191{ipipipasswdot1.eps } } \endxy } } \ \ \ \ = \ \ \ \ { \vcenter{\xy ( 0,0)*{\includegraphics[scale=.3]{ipipipasswdot2.eps } } \endxy } } \label{eqn - ipipipasswdot}\ ] ] lastly , @xmath187 and @xmath206-colored lines are allowed to cross when @xmath207 .
these crossings conclude the list of generating 2-morphisms .
the following relations say that the crossings are essentially `` virtual '' , i.e. the lines freely go through each other ( the solid line has color @xmath187 and dashed color @xmath206 ) .
@xmath208{ijijrot1.eps } } \endxy}}\ \ = \ \ { \vcenter{\xy ( 0,0)*{\includegraphics[scale=.4]{4valent2.eps } } \endxy}}\ \ = \ \ { \vcenter{\xy ( 0,0)*{\includegraphics[scale=.4]{ijijrot2.eps } } \endxy } } \label{eqn - ijijrot}\ ] ] @xmath196{farlines.eps } } \endxy}}\ \ = \ \ { \vcenter{\xy ( 0,0)*{\includegraphics[scale=.5]{r2move.eps } } \endxy } } \label{eqn - r2move}\ ] ] @xmath209{ijijdot1.eps } } \endxy } } \ \ = \ \ { \vcenter{\xy ( 0,0)*{\includegraphics[scale=.6]{ijijdot2.eps } } \endxy } } \label{eqn - ijijdot}\ ] ] @xmath208{pullfarthrutrivalent1.eps } } \endxy } } \ \ = \ \ { \vcenter{\xy ( 0,0)*{\includegraphics[scale=.4]{pullfarthrutrivalent2.eps } } \endxy } } \label{eqn - pullfarthrutrivalent}\ ] ] the next relation says that @xmath206-line interacts trivially with trivalent @xmath210-colored 6-vertex ( necessarily @xmath211 or @xmath212 ) .
@xmath208{pullfarthru6valent1.eps } } \endxy } } \ \ = \ \ { \vcenter{\xy ( 0,0)*{\includegraphics[scale=.4]{pullfarthru6valent2.eps } } \endxy } } \label{eqn - pullfarthru6valent}\ ] ] the following relation shows trivial interaction between the crossings of @xmath213-colored lines ( the colors are necessarily far apart @xmath214 )
. @xmath208{r3move.eps } } \endxy } } \ \ = \ \ { \vcenter{\xy ( 0,0)*{\includegraphics[scale=.4]{r3move2.eps } } \endxy } } \label{eqn - r3move}\ ] ] finally , the last and most sophisticated relation is an interation between 6-valent vertices for @xmath215 ( dotted line has color @xmath216 ) .
@xmath208{threecolorass1.eps } } \endxy } } = { \vcenter{\xy ( 0,0)*{\includegraphics[scale=.4]{threecolorass2.eps } } \endxy } } \label{eqn - threecolorass}\ ] ] the main result of @xcite is that the above generators and relations give a presentation of the soergel category @xmath150 .
we can say that @xmath150 is a finitely - presented @xmath91-linear pivotal monoidal category .
this presentation is manifestly planar ( perhaps the term _ planar category _ can be used as a sibstitute for _ pivotal monoidal category _ ) . of course , the self - adjointness of tensoring with @xmath151 was a strong hint that @xmath150 should have a planar description .
below we list some applications of this new viewpoint on the soergel category .
\(a ) planar presentation of @xmath150 leads to a new categorification @xcite of the temperley - lieb algebra @xmath217 .
the temperley - lieb algebra is the quotient of the hecke algebra @xmath120 by the relations @xmath218 .
this algebra is fundamental for the construction of the jones polynomial and admits a categorification via bimodules over certain rings , see @xcite . to categorify the relation @xmath218 observe that after categorification both sides become soergel bimodules @xmath219 and @xmath151 , and that the latter bimodule is a summand of the former .
thus , the equality can be categorified by setting the complementary summand to @xmath220 .
since the 6-valent vertex ( [ eqn - ipipiprot ] ) is a map which goes through this summand , we simply set all 6-valent vertices to 0 and form the quotient monoidal category .
a result of b. elias @xcite says that the grothendieck ring of the quotient category is naturally isomorphic to ( a @xmath221$]-form ) of the temperley - lieb algebra . in an earlier categorification of the temperley - lieb algebra @xcite , a basis of homomorphisms between bimodules categorifying products of @xmath155 s was given by 3-dimensional objects , namely suitable decorated surfaces with boundary and corners embedded in @xmath222 ^ 3 $ ] .
in elias s approach , a basis is given by a certain collection of planar diagrams representing morphisms in the quotient category .
a comparison between the two categorifications gives an example of _ dimensional encoding _ or _ dimensional reduction _
, when 3-dimensional information is flattened onto 2d , see @xcite .
\(b ) r. rouquier @xcite constructed a braid group action on the category of complexes of soergel bimodules up to chain homotopies .
the braid group generators @xmath223 correspond to complexes @xmath224 with the differential being the multiplication map @xmath225 ( counit of @xmath151 in graphical notation ) .
this braid group action extends to an action of the category of braid cobordisms @xcite .
the action can be rethought in the above diagrammatical language @xcite , providing a link between 2-dimensional defining relations in @xmath150 and braid cobordisms ( 4-dimensional objects ) .
\(c ) taking hochschild homology of rouquier complexes in a suitable way leads to homology groups that turn out to depend only on the closure of a braid @xcite .
the resulting homology theory is triply - graded , coincides with the one introduced in @xcite , and can be viewed as a categorification of the ocneanu trace , thus a categorification of the homfly - pt polynomial .
we hope that planar diagrammatics will help to understand this largely mysterious homology theory and its generalization by webster and williamson @xcite .
it is natural to expect a possible relation between diagrammatics for @xmath150 ( and more general planar categories and planar 2-categories ) and 2d ( topological ) field theories .
biadjointness is implicit everywhere in 2d qfts , since the categories that appear there , such as derived categories of coherent sheaves on 3d calabi - yau manifolds , fukaya - floer categories , and categories of matrix factorizations in lg models admit a wealth of biadjoint functors , that come up in a natural way , via convolutions with sheaves on products of a pair of calabi - yau s , via convolutions with lagragians in the product of sympectic manifolds , and via matrix factorizations with potentials @xmath226 .
planar diagrams for morphisms in @xmath150 ( and planar diagrams for 2-morphisms in similar 2-categories ) can perhaps be viewed as world sheets of 2d qfts with defect lines and vertices , with regions of the diagram labelled by different target objects . on a speculative note ,
we suggest a noncommutative geometry language for @xmath152 . recall that we represent soergel bimodule @xmath174 for a sequence @xmath172 by placing dots labelled @xmath227 on a line .
let s imagine that this line with labeled dots is a path , and refer to @xmath174 as a path as well .
more generally , we call any object of @xmath152 a path .
indecomposable objects @xmath170 are _ geodesics_. given two objects @xmath228 of @xmath152 , we think of the graded vector space @xmath229 as categorified quantized area of surfaces stretched between @xmath58 and @xmath61 .
the fact that the grading of @xmath229 is bounded from below is loosely analogous to the same property of energy .
the first example of a categorification of a bialgebra is apparently due to l. geissinger @xcite .
standard inclusions of symmetric groups @xmath230 give rise to induction and restriction functors @xmath231\dmod \times \c[s_m]\dmod \lra \c[s_{n+m}]\dmod , \\ \mathrm{res}_{n , m } & : & \c[s_{n+m}]\dmod \lra \c[s_n \times s_m]\dmod \end{aligned}\ ] ] between categories of ( finite - dimensional ) modules over these group algebras .
summing over all @xmath232 produces functors @xmath233 where @xmath234\dmod , \ \
\ \ \ \ \ \ \ \
\mc{s } \otimes \mc{s } : = { { \mathop{\oplus}\limits_{n , m\ge 0 } } } \c[s_n\times s_m]\dmod\ ] ] is the direct sum of representation categories of symmetric groups , respectively products of symmetric groups .
functors @xmath235 and @xmath236 induce homomorphisms of grothendieck groups @xmath237 \ : \ k_0(\mc{s } ) \otimes k_0(\mc{s } ) \lra k_0(\mc{s } ) , \ \ \ \ \ \ [ \mathrm{res } ] \ : \ k_0(\mc{s } ) \lra k_0(\mc{s } ) \otimes k_0(\mc{s}).\ ] ] here @xmath238\dmod),\ ] ] and the grothendieck group @xmath239\dmod)$ ] is a free abelian group with a natural basis given by symbols of simple @xmath121 modules @xmath240 , parametrized by partitions @xmath241 of @xmath55 ( the group algebra @xmath242 $ ] is semisimple for any finite group @xmath32 , any module is projective , and isomorphism classes of simple modules give rise to a basis of @xmath106 ) .
it is natural to identify @xmath243 with the ring of symmetric functions in countably many variables by taking @xmath244 $ ] to the schur function @xmath245 .
a biring is a bialgebra over @xmath248 .
this result , due to geissinger @xcite , also tells us that the ring of symmetric functions has a natural comultiplication .
the condition that comultiplication @xmath249 $ ] is a homomorphism , @xmath250 , follows from the mackey s decomposition theorem specialized to induction and restriction between products of symmetric groups .
deep generalizations of this construction were developed by a. zelevinsky @xcite , who constructed a similar biring structure on the sum ( over all @xmath55 ) of grothendieck groups of representation categories of @xmath130 , with parabolic induction and restriction in place of the usual induction and restriction .
many other examples of a biring structure on grothendieck groups can be found in bergeron and li @xcite and references therein . in another paper ,
zelevinsky classified irreducible representations of the affine hecke algebra of @xmath121 in terms of multisegments @xcite .
the significance of his result became clear much later , when s. ariki @xcite categorified all integrable irreducible representations of @xmath251 and affine @xmath251 via blocks of representation categories of cyclotomic quotients of affine hecke algebra ( an important earlier milestone was the work of lascoux - leclerc - thibon on categorification of level one representations of affine @xmath251 via the representation categories of hecke algebras of @xmath121 at @xmath252-th root of unity , sum over all @xmath253 ) .
i. grojnowski @xcite , armed with ideas of a. kleshchev from modular representation theory , gave an alternative derivation of ariki s results .
ariki also showed that grothendieck groups of completions of affine hecke algebra at suitable central ideals are canonically isomorphic to weight spaces of @xmath254 and @xmath255 , giving a conceptual categorification - style explanation for zelevinsky s classification of irreducibles and proved that the basis of indecomposable projectives in these grothendieck groups coincides with the @xmath133 specialization of the canonical basis in the quantum group @xmath256 .
in lusztig s construction of the canonical basis @xcite , partially inspired by ringel s work @xcite , canonical basis elements correspond to simple perverse sheaves on the varieties of quiver representations .
in fact , his construction can even be viewed as a categorification of @xmath256 if one passes from the set of simple perverse sheaves to a suitable triangulated category of equivariant sheaves that these simple objects generate .
lusztig @xcite and kashiwara @xcite also provided more elementary approaches to the canonical basis .
we ll now review a very down - to - earth categorification of @xmath256 based on a collection of certain rings , following @xcite , see also @xcite .
grothendieck groups of these rings can be identified with an integral version of @xmath256 , while induction and restriction functors between these rings correspond to multiplication and comultiplication in the quantum group . a direct link between this categorification of @xmath256 and lusztig s perverse sheaves construction was found by varagnolo and vasserot @xcite .
brundan and kleshchev @xcite related these rings to ariki s categorifications of @xmath257 and highest weight representations .
a direct relation to derived categories of coherent sheaves on nakajima quiver varieties @xcite is expected @xcite .
let @xmath258 be an unoriented graph without loop and multiple edges , with set of vertices @xmath259 .
the quantum group @xmath260 is a @xmath261$]-graded @xmath262-algebra with generators @xmath263 of degree @xmath187 and defining relations @xmath264 the integral version @xmath265 is the @xmath221$]-subalgebra of @xmath257 generated by divided powers @xmath266!}$ ] over all @xmath187 and @xmath55 .
both @xmath257 and @xmath265 are twisted bialgebras ( the cartan subalgebra is missing ) , with @xmath267 ( instead of the usual comultiplication @xmath268 in quantum groups ) , and nonstandard algebra structure on @xmath269 : @xmath270 where @xmath271 is the degree , taking values in @xmath261 $ ] , and @xmath272 is the inner product on @xmath261 $ ] with @xmath273 , @xmath274 if @xmath187 and @xmath206 are connected by an edge , and @xmath275 otherwise .
relative to this algebra structure , @xmath276 is a homomorphism @xmath277 ( see @xcite for more details ) .
let @xmath278 , \ \
\nu = \sum_{i \in i } \nu_{i } \cdot i \ ; , \ \
\nu_i \in \n.$ ] for each such @xmath279 define the graded ring @xmath280 , as ring spanned by diagrams of lines in @xmath281 $ ] , with @xmath282 lines colored by label @xmath283 .
lines can intersect , but triple intersections are not allowed , neither are u - turns that create critical points under projection onto the @xmath284-axis .
lines may carry dots ( in the picture below @xmath285 ) .
product is given by concatenation of diagrams ; if the labels of endpoints do nt match the product is zero .
we allow isotopies ( that do not create critical points relative to the @xmath284-axis projection ) and impose the following relations .
make @xmath280 graded by assigning degree @xmath291 to a dot and degree @xmath292 to the @xmath293-crossing . also , choose a base field @xmath109 and consider @xmath280 as a graded @xmath109-algebra .
diagrammatics make it clear that , at the cost of adding simpler diagrams , all dots can be moved above all crossings .
moreover , if two lines intersect twice in a diagram , the diagram can be further simplified .
this argument gives a spanning set in @xmath280 which consists of all monomials in dots times minimal length presentations of symmetric group elements as products of crossings , with arbitrary @xmath279-colorings of the lines .
one can check that this spanning set is a basis , by looking at the action of @xmath280 on a suitable representation .
each sequence @xmath294 of weight @xmath279 gives rise to an idempotent @xmath295 in @xmath296 given by the diagram with @xmath297 vertical lines labelled @xmath298 from left to right .
these idempotents are mutually orthogonal and @xmath299 , where @xmath300 is the set of sequences of weight @xmath279 .
also , each idempotent @xmath295 determines a graded projective module @xmath301 .
this is a crucial proposition .
after passage to the grothendieck group , these isomorphisms become equalities on symbols of projectives : @xmath303 & = & [ p_{ji } ] \hspace{0.2 in } \mbox{if } \hspace{0.1 in } i\cdot j=0 , \\ q [ p_{iji } ] + q^{-1 } [ p_{iji } ] & = & [ p_{iij } ] + [ p_{jii } ] \hspace{0.1 in } \mbox{if } \hspace{0.1 in } i\cdot j=-1 , \end{aligned}\ ] ] showing a perfect match with the defining relations ( [ deq1 ] ) , ( [ deq2 ] ) in @xmath257 after converting @xmath304 $ ] to @xmath305 , @xmath306 $ ] to @xmath307 , etc . the second isomorphism can be refined .
specifically , @xmath308 ( @xmath297 strands identically labelled ) is naturally isomorphic to the nilhecke algebra , generated by multiplication by monomials @xmath309 and divided difference operators @xmath310 .
it is well - known that the nilhecke algebra is isomorphic to the algebra of @xmath311-matrices with coefficients in the ring of symmetric functions in @xmath309 .
consequently , the free module @xmath312 decomposes as a sum of @xmath313 copies of an indecomposable projective module ( the column module ) which we denote by @xmath314 .
the latter has a suitably selected overall grading shift , so that , taking grading into account , @xmath312 is the sum of @xmath315!$ ] copies of @xmath314 . just like with the group algebras of symmetric groups , we can fit rings @xmath280 over all @xmath278 $ ] into a tower of algebras .
there are natural ( nonunital ) inclusions @xmath319 described by placing diagrams representing elements of @xmath280 and @xmath320 in parallel , next to each other .
consider associated induction and restriction functors @xmath321 , @xmath322 .
we would like to look at the maps these functors induce on the grothendieck group .
the induction functor always takes projectives to projectives and induces a homomorphism between grothendieck groups of finitely - generated graded projective modules @xmath323 : k_0(r(\nu))\otimes_{\z[q , q^{-1 } ] } k_0(r(\nu ' ) ) \lra k_0(r(\nu)\otimes_{\bbbk } r(\nu ' ) ) \lra k_0(r(\nu+\nu')).\ ] ] the first arrow comes from tensoring projectives . in our case , the first arrow is an isomorphism , due to absolute irreducibility of simple graded @xmath280-modules , which we prove using methods of kleshchev and grojnowski .
the second arrow comes from the induction functor . for a general inclusion of rings ,
restriction functor does not necessarily take projectives to projectives .
however , @xmath324 ( more accurately , @xmath325 ) is a projective @xmath326-module , so that the restriction does take projectives to projectives and induces a homomorphism @xmath327 : k_0(r(\nu+\nu ' ) ) \lra k_0(r(\nu)\otimes_{\bbbk } r(\nu ' ) ) \cong k_0(r(\nu))\otimes_{\z[q , q^{-1 } ] } k_0(r(\nu')).\ ] ] form the direct sum @xmath328 } } } k_0(r(\nu))\ ] ] and the corresponding sums of functors @xmath329 these induce homomorphisms of @xmath106-groups @xmath330 \ : \
k_0(r)\otimes k_0(r ) \lra k_0(r ) , \ \ \
[ { { \rm res } } ] \ : \
k_0(r ) \lra k_0(r)\otimes k_0(r).\ ] ] @xcite there is a canonical isomorphism of twisted bialgebras over @xmath221 $ ] @xmath331 under which multiplication , respectively comultiplication , in @xmath265 is given by
@xmath332 $ ] , respectively @xmath333 $ ] .
this isomorphism takes the product element @xmath334 to the symbol @xmath335 $ ] of @xmath336 and divided power element @xmath337 to the symbol of @xmath314 , which is an indecomposable direct summand of @xmath338 with a suitably normalized grading .
it is shown by brundan and kleshchev @xcite ( in the @xmath339 and affine @xmath339 case ) and varagnolo and vasserot @xcite ( for general graphs @xmath258 ) that , if @xmath340 , the canonical basis @xcite of @xmath265 goes to the basis of indecomposable projectives under this isomorphism . to get this match ,
when @xmath258 has an odd length cycle , the definition of @xmath280 should be slightly modified , see the above references and @xcite .
rings @xmath280 proved handy in recent constructions @xcite of graded versions of the group algebra of @xmath121 , specht modules , and @xmath125-schur algebras , also see @xcite and references therein for related developments .
moreover , they serve as a building block in webster s categorification of reshetikhin - turaev link and tangle invariants @xcite . at the base
lies the collection of positive integers @xmath341 these are dimensions of weight spaces @xmath342 of @xmath257 , or , more implicitly , coefficients of the kostant partition function .
its categorification is the twisted bialgebra @xmath257 equipped with the lusztig canonical / kashiwara crystal basis .
after the second round of categorification , the weight spaces @xmath342 become grothendieck groups of graded rings @xmath280 , and the canonical basis elements lift to indecomposable projective modules .
equivalently , one can work with the derived category of equivariant constructible sheaves on varieties of quiver representations , with simple perverse sheaves being analogues of indecomposable projective graded @xmath280-modules .
the upper horizontal arrow is a derived equivalence ( for a carefully chosen version of the sheaves category ) exchanging indecomposable projective @xmath280-modules and simple perverse sheaves .
kashiwara crystal graph , together with kashiwara operators , lies midway between the middle and base levels .
vertices of the graph correspond to canonical basis elements , and kashiwara operators remember only `` highest terms '' for the action of @xmath86 s on the canonical basis .
this structure is set - theoretic . despite its position somewhat below the first categorification , it is already incredibly rich .
for instance , key results about simple and projective @xmath280-modules @xcite follow via kleshchev - grojnowski constructions that amount to the representation - theoretical counterpart of the crystal graph structure .
@xmath257 is `` one - half '' of the entire quantum group @xmath343 . in the categorification of @xmath257
the diagrammatics have a braid - like behaviour , in the sense that the lines in diagrams only go up and do not have u - turns .
interestingly , it is possible to enlarge the calculus of rings @xmath280 by allowing u - turns , coloring regions of resulting diagrams by integral weights of @xmath344 and adding more relations , including all isotopies , to categorify the entire quantum group , see lauda @xcite for the @xmath345 case , @xcite for arbitrary @xmath55 , and chuang - rouquier @xcite , rouquier @xcite for related and less rigid axiomatics .
the quantum group first needs to be modified , following beilinson - lusztig - macpherson and lusztig @xcite , by adding idempotents @xmath346 of projection onto integral weights @xmath241 . in this categorification generators @xmath86 and @xmath88 become biadjont functors ( biadjoint up to grading shifts ) .
the grothendieck ring of the resulting 2-category is isomorphic to the blm form of the quantum @xmath90 .
the diagrammatics of biadjoint functors described at the beginning of this paper plays a fundamental role in the definition of the 2-category and computations in it .
recent categorification of the @xmath125-schur algebra by mackaay , stoi , and vaz @xcite links the soergel category and categorified quantum @xmath90 .
the @xmath186 case , due to lauda @xcite , settles a conjecture of igor frenkel , circa 1994 , that there exists a categorification of quantum @xmath345 ( some motivations for the conjecture can be found in @xcite ) . back in the early 90 s igor frenkel , who was the author s advisor in graduate school , envisioned categorification lifting the entire theory of quantum groups , quantum 3-manifold invariants , and conformal field theory . through the work of many people ,
his prophecy is becoming a reality . *
acknowledgments * the author would like to thank hiraku nakajima and the mathematical society of japan for the opportunity to deliver the takagi lectures in june 2009 and for their hospitality .
the author is grateful to aaron lauda , radmila sazdanovi , and joshua sussan for reading the manuscript and providing valuable comments and corrections .
partial support during the writing of present notes came from the nsf , via grant dms-0706924 .
the list of references is grossly inadequate and misses quite a few fundamental contributions - our apologies for the omissions go to many fine practitioners in these areas .
j. bernstein , i. b. frenkel , and m. khovanov , a categorification of the temperley - lieb algebra and schur quotients of u(sl(2 ) ) via projective and zuckerman functors , _ selecta math .
_ * 5 * ( 1999 ) , no .
2 , 199241 , arxiv : math.qa/0002087 . | a diagrammatic presentation of functors and natural transformations and the virtues of biadjointness are discussed .
we then review a graphical description of the category of soergel bimodules and a diagrammatic categorification of positive halves of quantum groups .
these notes are a write - up of takagi lectures given by the author in hokkaido university in june 2009 .
|
that cosmic acceleration is a fact appears indubitable . instead of an exotic new form of dark energy driving the acceleration ,
it may be caused by a modification of gravity .
precise measurements for gravity are only available in the range of scales from a millimeter to that of the solar system we do not have any direct probe of einstein gravity beyond these boundaries . cosmic acceleration
may originate in a breakdown of einstein gravity at distances beyond the range above .
dvali , gabadadze and porrati ( dgp ) @xcite have proposed a braneworld theory in which our universe is a ( 3 + 1)-dimensional brane embedded in an infinite minkowski bulk .
gravity propagates everywhere , but , on the brane , an additional ( 3 + 1)-dimensional gravitational interaction is induced .
this allows for gravitational potentials on the brane of a ( 3 + 1)-dimensional form at small distances to evolve into ( 4 + 1)-dimensional form beyond a crossover scale determined by the unknown energy scale for the bulk gravity .
the cosmological solution of this theory was shown to exhibit accelerated cosmic expansion without the aid of an exotic energy component like dark energy @xcite@xcite .
it has already been shown that the linearized field theory as defined by the dgp model contains ghost degrees of freedom @xcite@xcite@xcite@xcite@xcite , or even may violate causality in certain limits @xcite .
it is known , for instance , that the de sitter background is unstable to classical linear perturbations ; however , it is claimed in @xcite that strong - coupling effects at small radii around matter sources ensure that the theory remains stable .
the point of view of our work is to assume that linear perturbation theory remains valid on the largest scales .
this is motivated by the fact that the late universe is dominated by the gravitational interaction of dark - matter haloes .
the internal structure of the haloes is controlled by the strongly - coupled non - linear theory . on the other hand ,
the radius below which strong - coupling is important for haloes is approximately equivalent to their size and therefore their interactions should be driven by the linear theory analyzed in this work .
we do find that deep into the accelerated era the spacetime becomes unstable on the timescale of the expansion .
however , this is an effect that only becomes important far into the future and is negligible as far as the observational impact today is concerned .
we therefore assume that during the early universe , when the theory does not exhibit instabilities , the analysis for dgp proceeds in exactly the same way as that for gr
. then , deep during the acceleration era , instabilities develop and the theory may or may not be saved by non - linear effects an issue on which we remain agnostic .
this evolutionary history appears to be the only one which is capable of reproducing the universe as we see it .
if strong coupling effects are important straight away and at all scales , the approximation of a homogeneous background cosmology is completely inapplicable and the dgp model would not be able to reproduce observations such as supernova luminosities .
we therefore effectively assume a best - case scenario for dgp : should this analysis fail to predict the observations , the model is excluded .
if it passes the observational tests , a more careful study of the effects of the strong coupling regime during the acceleration era would be required to understand fully the future evolution of the universe in the dgp model . under the above assumptions , the equations of motion for the theory of gravity on the brane , pertinent to the study of cosmology , do not close owing to the interaction with the bulk at first order in perturbation theory .
koyama and maartens @xcite have used a quasi - static approximation , valid well within the horizon , to investigate structure formation at smaller scales .
this solution shows the essential role that the bulk plays in correcting the gravitational potentials , reducing the growth rate .
also , lue _ et al .
_ @xcite have reached a similar conclusion using a different approach , including non - linearities in their calculations . in the following , we present a new scaling ansatz for the master equation , allowing us to solve the equation and calculate the resulting cosmological evolution at all scales for high redshifts , and close to the horizon today . in [ s : eom ] , we review the linearized equations of motion for dgp .
we present our scaling solution in [ s : scaling ] and discuss its cosmological implications in
[ s : cosmo ] .
we study the robustness of the scaling solution in the appendix and discuss these results in
[ s : discussion ] .
in the dgp model , gravity alone propagates in the bulk , and the 5d gravitational theory is complemented by an induced 4d ricci scalar restricted to the brane .
we assume that both the bulk and the brane have zero tensions , i.e. the cosmological constants are zero .
we thus start off with the basic dgp action : @xmath0 \ , , \ ] ] a 3-brane embedded in an empty bulk , with all the standard - model fields localized on the brane at @xmath1 . the constants @xmath2 and @xmath3 define the energy scales of the theories of gravity : one is newton s constant , @xmath4 , the other represents the energy scale of the bulk gravity .
as shown by dvali _
@xcite , the ratio of the two scales defines a cross - over radius beyond which the four - dimensional gravitational theory transitions into a five - dimensional regime @xmath5 this scale is chosen to be of the order of the current hubble length so that the acceleration of the expansion today results from the 4d to 5d transition . for illustrative purposes
, we will take in our calculations @xmath6 , in a matter and radiation universe with @xmath7 and @xmath8 .
when comparing with @xmath9cdm , the concordance set of parameters will be used : @xmath10 , @xmath11 .
the two sets of parameters represent cosmologies which are best fits to supernova luminosity data ( snls @xcite ) and the distance to the last - scattering surface ( wmap @xcite ) , but assuming that the universe is flat .
we discuss the fits to these data in a companion paper , song _ et al .
_ @xcite . for the purpose of cosmological calculations
, we would like to reduce the five - dimensional braneworld to an effective theory on the brane , which could be studied using the usual range of four - dimensional tools . using a 4 + 1 decomposition of the theory @xcite@xcite@xcite
, we can derive the effective on - brane equations of motion a set of modified einstein equations @xmath12 where the greek indices range across all the dimensions , @xmath13 .
@xmath14 is a tensor quadratic in the 4-dimensional einstein and energy - momentum tensors , @xmath15 while @xmath16 is the bulk weyl tensor projected onto the brane using the vector normal to the brane , @xmath17 @xmath18 it can be shown that for branes with maximally - symmetric spatial hypersurfaces , the projected weyl tensor must take the form @xmath19 , with @xmath20 a constant ( see @xcite : it is effectively a constant of integration for the background ) . since the weyl tensor is traceless , it will dilute as rapidly as radiation and will become irrelevant at late times , if it does nt completely dominate the dynamics initially .
we will therefore set @xmath21 , allowing us to find the modified friedman equation for the background evolution of the cosmology on the brane . in the case of the flat brane , which is the only one to be considered here , @xmath22 leading to the result that , for the upper - sign selection , the cosmology tends to a de - sitter phase as the matter density gets diluted away , potentially providing a model for the observed acceleration @xcite .
this is the choice we will make henceforth .
in addition to @xmath23 , there is another scale present in the theory , which thus far we have ignored : the strong - coupling scale given by @xmath24 where @xmath25 is the schwarzschild radius of the point source under consideration . beneath this radius ,
the linear approximation developed above is not valid and the theory returns to 4d gr with small corrections .
it is not understood how two sources superimpose when the linear regime is not valid .
however , the linear density fields are actually constructed out of a spatial average of dark- matter halos , each of which has @xmath26 comparable to its radius .
thus we should be able to use the linear theory to discuss the gravitational dynamics of the spatial averages of the halos , but most likely not their internal structure .
an observer restricted to the brane will perceive the universe as purely four dimensional .
therefore , the most general linear scalar perturbations to the induced flat four - dimensional metric can be parameterized by @xmath27 while he linearized energy - momentum tensor can be written down as @xmath28 here @xmath29 is the density of the cosmological background , @xmath30 is the background equation of state parameter , and @xmath31 is the sound speed for the pressure perturbations .
the full 5d perturbations of the bulk are richer : deffayet in @xcite has shown that their effect on the brane can be reduced to the presence of perturbations to the weyl tensor as an additional source of stress - energy . the precise relationship between the various components of the weyl tensor
is determined by their relationship to a master variable ( see [ s : master ] ) , which in turn encodes all the gauge - invariant perturbations to the full 5d metric throughout the bulk .
we define the linear scalar perturbations to the weyl tensor by @xmath32 \,.\end{aligned}\ ] ] the weyl tensor is traceless , and hence the pressure perturbation will behave like radiation , i.e. @xmath33 .
the linearized poisson and anisotropy equations then are @xmath34 @xmath35 \mu^2\rho a^2 w \pi \notag\\ & \qquad - \frac{1}{2r_{\mathrm{c}}h \left(1+\frac{\dot{h}}{2h^2}\right ) -1 } \frac{\mu^2 \rho}{3}a^2 \pi_e \ , , \label{e : aniso}\end{aligned}\ ] ] with @xmath36 the comoving density contrast and the equivalent definition for @xmath37 : @xmath38 the equations of motion for the energy momentum tensor are given by the conservation law @xmath39 supplemented by equations of state that define the stress fluctuations @xmath40 - { k^2 \over a^2 } q = -3 \dot\phi\ , , \\ & \dot q - 3 c_{\mathrm{s}}^2 q h + { c_{\mathrm{s}}^2}{\delta \over 1+w } -{2\over 3}{w \over 1+w } k^2\pi = -\psi \,.\end{aligned}\ ] ] we have here assumed adiabatic pressure fluctuations in the multicomponent matter system @xmath41 . on the other hand , the weyl tensor is not separately conserved and its equations of motion come from the bianchi identity , @xmath42 , as applied to eq . ,
@xmath43 we can rewrite this as @xmath44 where the source term @xmath45 \,.\end{aligned}\ ] ] thus a non - zero weyl tensor is unavoidably generated by matter perturbations in linear theory @xcite . in order to close the above equations
, we need the analogue of an equation of state to relate the weyl anisotropic stress @xmath46 to the other components of the weyl tensor @xmath47 and @xmath48 . unlike the relationship between the weyl pressure and energy density , this relation requires a consideration of perturbations in the bulk . in @xcite
, mukohyama showed that for maximally - symmetric five - dimensional space - times , the full five - dimensional linear scalar perturbations in the bulk can be described using a master variable @xmath49 .
since the bulk being considered is just minkowski , and the brane is assumed flat , we can parameterize the unperturbed background 5d metric by @xcite @xmath50 where the brane sits at @xmath51 and @xmath52 this parameterization is also valid for branes with non - zero curvature provided that @xmath53 .
the master variable then obeys a hyperbolic equation of motion : @xmath54 we can then express all the gauge - invariant perturbations to the 5d bulk as functions of derivatives of the master variable .
in particular , deffayet @xcite has shown that the components of the weyl tensor evaluated on the brane can be expressed as @xmath55 we will hereafter implicitly assume evaluation at @xmath51 for the master variable in the on - brane equations where no confusion might arise .
we can now rewrite the bianchi identity in terms of the master variable , obtaining , after assuming that the cosmological fluid has no anisotropy stress , @xmath56 where @xmath57 expresses the derivative across the brane @xmath58 we have defined two new functions of the hubble parameter @xmath59 were it not for the @xmath60 derivative across the brane in @xmath57 , this equation would be a simple dynamic equation for @xmath49 , which , given an evolution equation for the source @xmath36 , could be solved as a coupled equation .
the role of the master equation is to define @xmath57 , the relationship between @xmath60 and @xmath49 .
koyama and maartens @xcite adopted a quasi - static approach to solve these equations .
the master equation then implies that the gradient @xmath61 in the bianchi identity the time - derivative and brane - derivative terms are neglected compared to those of order @xmath62 .
this quasi - static approach leads to their solution to which we will refer henceforth as `` qs '' @xmath63 this solution is equivalent to a closure relationship for @xmath46 in terms of @xmath64 through eqs .
( [ e : deltaeomega ] ) and ( [ e : pieomega ] ) . using this , we can define the quasi - static limit ( qs ) of dgp gravity where the poisson and anisotropy equations become @xmath65 with @xmath66 these equations are equivalent to the linear limit of the results obtained by lue _
_ @xcite . in the next two sections
, we will show how the quasi - static solution is dynamically achieved for the perturbations shortly after horizon crossing and discuss large - scale deviations from this solution .
the master equation is a wave equation sourced by the comoving density perturbations on the brane through the bianchi identity . since , in appropriate coordinates , the bulk is just minkowski , the evolution of @xmath49 in the bulk can be seen as a normal propagating wave given a boundary condition from the behavior on the brane . beyond the causal horizon , the bulk should remain unperturbed .
this causal horizon must be invariant in all coordinizations of the bulk ; therefore we can locate it by finding the null geodesic of eq .
, giving us the @xmath67-position of the horizon as @xmath68 before the acceleration epoch , this reduces to @xmath69 for a cosmology with a constant equation of state parameter , i.e. @xmath70 for a radiation - only cosmology and @xmath71 for a matter - only cosmology .
we are making the assumption that the universe has not gone through a period of inflation , which would have moved the horizon much further out . if inflation did take place , it is not unreasonable to expect that any perturbations that existed prior to the inflationary phase will have been pushed far away and the bulk will start in an unperturbed state in the vicinity of the brane at the beginning of radiation domination , resulting in a causal horizon equivalent to that of the cosmology with no inflation .
the constancy of the horizon during the domination of a particular fluid suggests that we can define a new variable , in which the horizon will remain fixed at all times , lying at @xmath72 @xmath73 we can then recast the master equation in @xmath74 and solve it as a boundary value problem with the value of @xmath49 at the horizon , @xmath72 , set to zero . the second boundary condition needed to solve the master equation is the behavior of the master variable on the brane .
during epochs when the source remains scale - free and the bianchi dynamics also do not change , one would expect that the master variable also obeys a scaling ansatz on the brane @xmath75 where @xmath76 and @xmath77 are constants .
likewise , during such epochs we expect the master variable in the bulk to reach a stable solution in the variable @xmath74 , the distance in units of the causal horizon , for a given wavenumber @xmath78 .
we therefore propose a new ansatz for the solution to the master equation : @xmath79 with this assumption , the master equation becomes the ordinary differential equation @xmath80 where the primes denote differentiation with respect to @xmath81the new time coordinate which will be used henceforth and @xmath82 . in deriving the above ,
we have neglected time derivatives of @xmath77 and @xmath83 : the scaling ansatz is not expected to be valid when @xmath77 is not a constant , i.e. during times when the cosmology is undergoing a change from the domination of one fluid to another .
in addition , strictly speaking , @xmath84 is actually a function of both @xmath74 and @xmath78 , even in the scaling limit .
we have therefore also assumed that @xmath78 is a constant , which is valid in the @xmath85 limit . as we explain later , as @xmath78 approaches unity , where its time - derivatives might impact the solution significantly , the character of eq .
changes and terms which do not involve the derivatives of @xmath78 dominate . note
that , provided @xmath86 is constant , one of the denominators in eq . can be re - expressed as @xmath87 thus the equation has a regular singular point @xmath72 , exactly at the causal horizon .
this is not a coordinate singularity ( all the entries of the metric are regular there ) , but is a reflection of the junction between perturbed and unperturbed space - times .
supplying @xmath88 , @xmath77 as a function of the scale factor is enough to solve this ode as a boundary - value problem , requiring that @xmath89 be 1 on the brane and 0 at @xmath90 .
this in turn gives the value of @xmath57 as @xmath91 and closes the evolution equations for the perturbations on the brane .
we will henceforth refer to this solution as the dynamical scaling or just scaling solution and use the acronym `` ds '' . in practice
, one does not know the scaling index @xmath77 a priori and , moreover , it can change during the evolution of a @xmath92 mode as the master variable leaves one scaling regime and enters another .
we therefore solve for @xmath77 iteratively by demanding consistency with the bianchi identity .
to determine the zeroth - order solution for @xmath77 , we substitute the ansatz eq . into eq .
to obtain @xmath93\mu^2\rho\delta \ , , \label{e : peq}\\ j(h ) & = p [ p + { { h}}-3f(h)]+ \frac{1}{k(h)hr_{\mathrm{c } } } + \frac{2hr_{\mathrm{c}}-1}{hr_{\mathrm{c } } } r \ , .
\notag\end{aligned}\ ] ] before the acceleration epoch , when the expansion is dominated by a single fluid , @xmath94 and @xmath95 are constant , while @xmath96 is a simple power law in @xmath97 .
we therefore set @xmath98 , the zeroth order solution for @xmath99 @xmath100}{{\mathrm{d}}\ln a } } } \,.\end{aligned}\ ] ] note that this definition allows transitions between scaling regimes where @xmath86 changes .
we shall see that the solutions for @xmath57 become independent of @xmath77 for @xmath101 and so we use this as the zeroth order solution for all modes , both super- and subhorizon .
finally we need a zeroth order solution for @xmath102 .
modes of interest to large - scale cosmological tests are superhorizon during radiation domination and enter the horizon either during matter domination or the current acceleration epoch . before the acceleration epoch and in the absence of anisotropic stress ,
these modes obey @xcite @xmath103 where @xmath104 . in order to obtain the zeroth - order solution for @xmath105
we assume that the growth prescribed by equation eq . continues until today ( see fig .
[ f : p ] ) . given this ,
we solve the master equation eq . to obtain the off - brane gradient , @xmath57 , and
then dynamically solve the bianchi identity eq .
coupled to the cosmology , for the particular mode .
once the dynamic evolution for the first - order solution @xmath106 is obtained , we can iteratively improve our estimation of @xmath77 by numerically calculating @xmath107 and repeating the above prescription .
we find in the next section that this procedure converges quickly and alters the value of @xmath77 only when @xmath77 is not a constant , as expected .
we display the effects of the iteration on @xmath77 in fig .
[ f : p ] . in the scaling solution
@xmath108 for superhorizon modes , plotted as a function of the scale factor .
the @xmath105 calculation assumes that the density perturbation @xmath36 follows eq . at all times .
the @xmath109 result is the output of an iterative process , where @xmath77 is used to calculate @xmath57 and hence the evolution of @xmath49 which is in turn used to derive a correction to @xmath77 .
[ f : p],height=260 ]
the evolution of the master variable exhibits several distinct phases that are distinguished by the on - brane scaling evolution : that during radiation domination , matter domination and the de - sitter acceleration phase both inside and outside the horizon . the scaling of the bianchi identity eq .
determines the value of @xmath77 at a particular scale factor , while the solution to the master equation eq . determines the off - brane gradient @xmath57 . to better understand the nature of the solution and how it impacts perturbation evolution , we will compare the analytic expectations to the full numerical results in the various phases .
, as defined in eq . , for a selection of modes .
on superhorizon scales , @xmath57 is constant whenever @xmath77 is a constant .
once the mode enters the horizon it rapidly approaches @xmath110 . as the universe enters the de - sitter phase , the modes again leave the horizon and @xmath57 asymptotes to @xmath111 .
[ f : rplot ] ] during radiation domination , @xmath112 , @xmath113 and @xmath114 .
the particular combination in the denominator of @xmath96 causes the first - order contribution of @xmath115 to cancel , leaving us with @xmath116 .
the bianchi identity eq . for superhorizon modes then dictates that @xmath117 since @xmath118 during radiation domination , this gives @xmath119 for superhorizon modes .
inserted back into the master equation under the scaling ansatz eq .
, this value of @xmath77 implies @xmath120 with @xmath57 determined , the equations of motion for the perturbations on the brane are closed .
this analytic expectation also serves as the initial conditions for the numerical scaling solution . in practice , we begin the integration at @xmath121 , when all modes of interest are outside the horizon .
the numerical solution for @xmath77 is shown in fig .
[ f : p ] and for @xmath57 in fig .
[ f : rplot ] .
note that , in the radiation - dominated era , their values stay stable at the analytic prediction for all iterations of the solution .
the large - scale modes of interest remain outside the horizon during the whole radiation - dominated epoch . in the matter - dominated epoch
, the evolution outside the horizon can be obtained by noting , @xmath122 , @xmath123 , @xmath124 , and @xmath112 .
the bianchi identity dictates that @xmath125 given that @xmath126 , @xmath127 .
since the matter dominated solution is of particular interest , we explicitly give the master equation under the scaling ansatz @xmath128 this is solved as a boundary value problem with boundary conditions @xmath129 and @xmath130 . the form of the numerical solution to this equation is shown in fig . [ f : gplot ] .
in the large - scale limit , the gradient reaches @xmath131 in the numerical solution of figs .
[ f : p ] and [ f : rplot ] , these values of @xmath127 and @xmath132 are achieved gradually as the expansion becomes matter dominated .
the iteration of the numerical solution in fact further smooths the transition until a stable form is achieved as would be expected .
note also that @xmath57 is very insensitive to @xmath78 , provided it be less than 1 such that the mode is larger than the horizon . obtained by solving equation during matter domination ( @xmath133 ) , compared to off - brane profiles for the quasi - static ( qs ) solution . for high @xmath78
the profiles are very narrow and effectively independent of the position of the causal horizon : they penetrate very little into the bulk and the behavior of the solution is practically independent of the value of @xmath77 . in this regime ,
the qs solution is practically coincident with the scaling solution . for modes with low @xmath78 ,
the solution is non - zero in the whole interval @xmath134 and therefore it depends strongly on the value of @xmath77 .
qs severely underestimates the gradient of the profile in this regime .
[ f : gplot ] ] the modes of interest cross the horizon either during matter domination or the acceleration epoch . for large values of @xmath78 , the final term of the master equation eq .
dominates over other parts of the coefficients of @xmath89 and @xmath135 . as evidenced in fig .
[ f : gplot ] , in this regime , the solution does not penetrate very far into the bulk .
we can thus expand the master equation around @xmath136 , reducing it to @xmath137 this matches the quasi - static approximation up to first order in @xmath74 .
therefore , for @xmath138 , the relation giving the gradient on the brane is exactly as in the qs approximation @xcite , with no dependence on the value of @xmath77 : @xmath139 the numerical solution for @xmath57 evolved through horizon crossing is shown in fig .
[ f : rplot ] .
it reaches this scaling shortly after horizon crossing . despite this independence of @xmath77 ,
the master variable does achieve a scaling form during matter domination .
the bianchi identity can be reduced to the qs form , eq . , and implies @xmath140 and hence @xmath141 . in general
then , @xmath57 is a function of both @xmath78 and @xmath142 and therefore each mode needs to be followed separately through its evolution both outside and inside the horizon .
however , only outside the horizon does the value of the off - brane gradient actually affect the evolution on the brane , since for high @xmath78 it is subdominant in the bianchi identity eq . . at late times
, the dgp cosmology enters the self - accelerated de - sitter phase . during
this time all modes exit the horizon while the causal horizon in the bulk , @xmath83 , grows rapidly toward infinity .
this allows us to concentrate on the @xmath143 limit . deep into the de - sitter phase of the expansion , @xmath144 , @xmath145
. the master equation can now be rewritten ( in the scaling approximation ) as @xmath146 this equation has an analytic solution .
assuming that the perturbations vanish at the causal horizon , @xmath147 at @xmath148 , and that @xmath149 , we can find the off - brane gradient : @xmath150 this can now be combined with the bianchi identity @xmath151 it can be shown that @xmath36 becomes a constant during the acceleration era . in the limit where @xmath152 and @xmath153 , i.e. at very late times , the solution to the above two equations combined is @xmath154 and @xmath155 or @xmath156 , where we would expect the fastest - growing mode , @xmath157 , to dominate .
we plot the numerical solutions in figs .
[ f : p ] and [ f : rplot ] .
note that iteration of the solution here is crucial but @xmath157 , @xmath154 is approached at late times .
the numerical results only converge to @xmath157 , @xmath154 very slowly .
this can also be understood analytically .
there is a slight correction to this result arising as a result of @xmath83 being finite , but it is already insignificant by @xmath158 .
a more significant correction to the scaling behavior arises as a result of @xmath36 not being zero .
its size can be estimated by assuming that @xmath159 . substituting this into both eq . and
eq . we obtain @xmath160 indeed , as @xmath49 grows , the correction decreases until @xmath157 is achieved .
in particular , the continued growth of @xmath49 implies that the anisotropy at the largest scales will continue growing as @xmath161 deriving an evolution equation for the metric highlights the fact that the metric evolution is not sensitive to the dgp modification until the acceleration epoch . even in the acceleration epoch
the metric obeys a simple equation of motion on superhorizon scales . by appropriately rearranging the linearized einstein equations and the bianchi identity , we can obtain a single differential equation for the gravitational potentials @xmath162\right)(1+w ) = \notag\\ & = \frac{h'}{h}\frac{1}{2hr_{\mathrm{c}}}\left[(w - c_{\mathrm{s}}^2)\delta_e + \left(\frac{1}{3}-w+\frac{h'}{3h}\right)\delta_e \right ] \ , .\end{aligned}\ ] ] in the large - scale limit , @xmath163 , @xmath164 ; if we also assume that @xmath165 or that @xmath166 from the poisson equation , we obtain @xmath167 note that this equation for the metric perturbations is the same as found in general relativity [ e.g. @xcite eq .
( 50 ) ] . therefore
, before the dgp modifications change the expansion rate or generate anisotropic stress , the evolution of the metric is identical to general relativity .
furthermore , as pointed out by bertschinger @xcite in the absence of anisotropic stress , @xmath168 and this equation yields solutions that depend only on the expansion history through @xmath88 even in the acceleration epoch . in dgp gravity
, however , the anisotropy at the largest scales is never negligible and , in fact , grows at late times , as discussed in section [ s : ds ] .
we can rewrite eq . by defining @xmath169 in the de sitter era , when @xmath170 and @xmath171 , and assuming that the two new variables obey a scaling solution with @xmath172 and @xmath173 , eq .
becomes @xmath174 one would expect that @xmath175 would grow with exponent @xmath176 , while @xmath177 would decay away with exponent @xmath178 .
however , because of the need to preserve the bianchi identity with a non - zero @xmath36 , as discussed in section [ s : ds ] , the scaling solution is slightly violated .
we find that @xmath179 while @xmath180 .
this leads to the relation @xmath181 where @xmath182 monotonically . at late times
, our solution tends to a regime where the ratio @xmath183 , with @xmath184 .
it is interesting to note that in this opposite limit to that studied in @xcite where @xmath185 eq .
( [ e : bert ] ) also becomes closed and has solutions that depend only on the expansion history through @xmath88 . for dynamic scaling and quasi - static solutions .
in qs , @xmath49 responds instantaneously to changes in @xmath36 .
fully dynamic solution to the bianchi identity requires time to respond and eventually decays to the qs solution .
initializing the calculation at earlier times changes neither the scale factor at which @xmath49 responds nor its value today .
rapid growth occurs during the time before horizon crossing .
[ f : omegacomp ] ] in fact , eq . is equivalent to the statement that the bardeen curvature , @xmath186 , is conserved during the de sitter era , once the mode leaves the horizon .
this is also true of the comoving density perturbation , @xmath36 , which saturates to a constant during the de sitter era .
however , the comoving and longitudinal hypersurfaces warp with a shift @xmath187 that grows without bound .
it is useful to summarize the differences between the quasi - static ( qs ) and dynamical scaling ( ds ) solutions uncovered in the previous section .
beginning at the initial conditions in the radiation - dominated era , the superhorizon value of the master variable is highly suppressed with @xmath188 ( see fig . [
f : omegacomp ] ) .
as the mode enters the horizon during matter domination , the ds solution for @xmath49 grows rapidly and then executes damped oscillations around the qs solution .
this can be understood analytically since the bianchi identity takes the form of a damped oscillator in @xmath189 that is driven by @xmath36 . during the time when @xmath49 significantly deviates from the qs solution , the weyl corrections to the poisson equation
are suppressed , since @xmath112 , and there is no additional correction to the gravitational potentials over and above that of qs ( see fig .
[ f : phicomp ] ) . for concordance @xmath9cdm , the quasistatic ( qs ) solution , and our new dynamical scaling ( ds ) solution . for scales
@xmath190 mpc@xmath191 the scaling and qs solutions do not differ appreciably : the decay in the potentials is a little faster than @xmath9cdm as a result of slower growth of density contrast . at larger scales ,
the ds solution exhibits significant additional decay owing to the different value of @xmath49 at late times , as exhibited in fig .
[ f : omegacomp ] .
all potentials normalized to 1 at @xmath133 .
[ f : phi ] ] for concordance @xmath9cdm , qs and our ds solution .
the growth function in dgp is suppressed compared to @xmath9cdm , even in the case of a flat cosmology .
there is no significant difference between the scaling ansatz and the quasi - static solution , with ds departing at most by 2% from qs at the largest scales .
all quantities are normalized to 1 at @xmath133 .
[ f : delta ] ] we find that the results of the scaling solution match the quasi - static results for all modes that enter the horizon well within the matter - dominated epoch @xmath192
mpc@xmath191 ( see figs . [
f : omegacomp][f : phicomp ] ) . for larger scales , this is not so : as shown in fig .
[ f : omegacomp ] , @xmath49 now only decays toward the qs solution , rather than oscillating around it .
since at late times the weyl perturbations are no longer suppressed , this now makes a significant contribution to the poisson equation , resulting in additional decay of the potentials .
as shown in fig .
[ f : phicomp ] , there is a 15% deviation from qs in @xmath193 at the scale @xmath194 mpc@xmath191 for the chosen sets of cosmological parameters .
the direction of this effect agrees with estimates made by lue in @xcite and is such that the scaling solution is an even worse fit to cmb anisotropy data than the quasi - static ( see @xcite for a discussion ) . on the other hand , up to the present time ,
the qs solution for comoving density perturbations , @xmath36 , is a very good approximation for the ds solution at all scales .
the additional suppression is of the order of 2% for @xmath194 mpc@xmath191 .
comparison of results of qs and ds solutions : upper panel presents the ratio of @xmath193 in the two approximations . for modes with @xmath192
mpc@xmath191 the two solutions differ by less than 2% .
the difference is much more pronounced for larger scales where @xmath49 has not decayed to the qs value , resulting in additional decays of up to 15% .
lower panel presents the ratio of comoving density perturbations in the two solutions : @xmath36 is affected much less , with approximately a 2% deviation from qs at the largest scale , @xmath194 mpc@xmath191 . ]
we have introduced a new scaling ansatz which allows solutions to linear perturbations in the dgp model on all scales less than the cross - over scale @xmath23 up to the present epoch .
the equations of motion for linear perturbations on the brane require knowledge of the gradient of the so - called master variable into the bulk .
the master variable obeys a master equation in the bulk . to solve the master equation , it is sufficient to have two boundary conditions , one on the brane and the other in the bulk .
our scaling solution begins with an ansatz for the brane boundary condition : that the evolution of the master variable is scale free on the brane .
the second boundary condition is that the master variable vanishes at the causal horizon in the bulk .
with these two boundary conditions , we solve the master equation to determine the gradient . with the gradient known
, we can then replace the scale - free ansatz with the dynamical solution and iterate the solution until convergence .
we find that the quasi - static ( qs ) solution of @xcite is rapidly approached once the perturbation crosses the horizon .
before horizon crossing there are strong deviations from the quasistatic solution . for modes that crossed the horizon only recently during the acceleration epoch , we find that the metric perturbation @xmath193 decays more rapidly that the qs solution .
the qs solution itself has a stronger decay than the @xmath9cdm model .
the extra decay compared with @xmath9cdm is extremely robust to changing the gradient of the master variable into the bulk , the one variable that is required to close the equations of motion on the brane .
we consider the observational consequences of these results in a companion paper @xcite .
we thank sean carroll , chris gordon , nemanja kaloper , arthur lue and roman scoccimarro for useful conversations .
this work is supported by the u.s .
dept . of energy contract de - fg02 - 90er-40560 .
is and wh are additionally supported at the kicp by the nsf grant phy-0114422 and the david and lucile packard foundation .
the kicp is an nsf physics frontier center .
assuming that the scaling ansatz is appropriate , the solution presented in this paper is correct at all scales .
however , the scale - free assumption for the solution in the bulk depends crucially on the existence of only one scale in the problem : the hubble parameter .
it is possible that through some additional physics in the vicinity of @xmath23 this assumption is broken and the evolution of the bulk profile will depend on both @xmath88 and @xmath23 independently .
however , any violation of scaling will enter the equations solely through the off - brane gradient @xmath57 : this is the quantity which we obtained by solving the master equation eq . using the scaling ansatz .
it is important to note that , in our analysis , we have not dropped any terms either arising from the weyl - fluid - driving bianchi identity or the master equation .
we have assumed that the solution in the bulk depends only on @xmath195 but the form of the master equation implies that this scaling assumption should be a good one .
we are only testing the robustness of our solution in order to estimate the effect of any new physics which might be important at scales around @xmath23 and which is not embodied in the master equation already .
one way of testing the robustness of the scaling ansatz is to change the values of @xmath57 at late times and investigate how much of a departure from scaling - ansatz values is necessary to significantly change the behavior of observables .
we concentrate on the change to the evolution of the potential @xmath193 , which drives the isw effect , as we alter the off - brane gradient .
since the qs solution already has a significantly sharper decay than @xmath9cdm , and therefore is a worse fit to the large - angle cmb anisotropy @xcite , and the ds solution decays even more rapidly ( see figure [ f : phi ] ) , we attempt to violate the scaling solution in such a way as to soften this decay .
we present the modification to @xmath57 in fig .
[ f : modr ] : we employ a linear interpolation for @xmath57 between its scaling - ansatz value at @xmath133 and a chosen off - brane gradient value today , @xmath196 .
one should note that this breaking is rather extreme , since the scale under consideration , @xmath197 mpc@xmath191 , is inside the horizon today and it should be well within the quasi - static regime at the present time .
off - brane gradient for @xmath198 mpc@xmath191 for the scaling solution and the scaling - violating scenarios employed in robustness testing .
since any scaling violation is only likely to occur at scale factor close to @xmath199 , we modify the gradient starting at @xmath200 .
a value of the gradient today , @xmath196 , is chosen and the gradient is interpolated linearly between these two scale factors . in this modification
, we disregard the fact that the mode enters the horizon at late times . ]
we have found that choosing negative values for @xmath196 strengthens the decay of @xmath193 , with the scaling solution approximately replicated for @xmath201 .
positive values of @xmath196 reduce the rapidity of the decay at late times : the qs solution is matched for @xmath202 , which is a value much higher than ever achieved by the scaling solution . in order to achieve the low levels of decay exhibited by @xmath9cdm
, @xmath196 needs to be set in the vicinity of 50 .
the effect is different for @xmath36 : here , negative values of @xmath196 also decrease the growth rate even further beneath that of ds ; however , @xmath36 asymptotes to a value approximately 2% beneath that of the qs solution as @xmath196 is sent to infinity . and
@xmath36 for mode @xmath198 mpc@xmath191 for a selection of scaling - violating scenarios .
increasing @xmath196 brings the solution closer to that of qs , and , for very large values , reduces the decay of @xmath193 to that of @xmath9cdm .
the effect of changing @xmath196 on @xmath36 is much smaller , the quantity remains insensitive to the precise details of the scenario . ] and @xmath36 for scaling and scaling - violating scenarios to their values in qs for mode @xmath198 mpc@xmath191 .
dashed line represents the final value of the quantity in qs , dotted in ds , and dot - dashed in @xmath9cdm .
the solution is very sensitive to negative values of @xmath196 , but choosing such scenarios only increases the decay , strengthening the isw effect .
positive values of @xmath196 bring the evolution of the observables closer to that of the qs solution and , for very large values , achieve decays as low as those of @xmath9cdm .
@xmath36 is quite insensitive to the choice of @xmath57 .
all quantities were normalized to the same value at @xmath133 . ] | we develop a scaling ansatz for the master equation in dvali , gabadadze , porrati cosmologies , which allows us to solve the equations of motion for perturbations off the brane during periods when the on - brane evolution is scale - free .
this allows us to understand the behavior of the gravitational potentials outside the horizon at high redshifts and close to the horizon today .
we confirm that the results of koyama and maartens are valid at scales relevant for observations such as galaxy - isw correlation . at larger scales , there is an additional suppression of the potential which reduces the growth rate even further and would strengthen the isw effect . |
in the theory of quantum chaos , the connection between the chaotic behavior of classical systems and specific properties of their quantum mechanical counterparts has received much attention .
the most studied issues in this context are the statistical properties of nuclear and atomic spectra @xcite and of two-(recently three ) dimensional billiards @xcite . in the mid - eighties ,
a conjecture ( which became known as the bgs conjecture @xcite ) was made that the quantal energy level spacing of ( ergodic ) systems whose classical equivalent exhibit chaotic behavior obeys the wigner gaussian - orthogonal - ensemble statistics , known from random matrix theory , whereas for classically integrable systems the poisson distribution applies .
exceptions to this conjecture are known @xcite , but it is generally believed that these form a set of measure zero . another way to make a connection between the chaotic behavior of classical systems and specific properties of their quantum mechanical counterparts is to look at the eigenstates of billiards .
eigenstates of a stadium may exhibit structure that corresponds to the classical orbits @xcite .
these structures , called scars , have been directly related to unstable periodic orbits @xcite .
visual inspection of movies that show the time evolution of the probability distribution of a quantum particle moving in two and three - dimensional ( 2d and 3d ) billiards of different shapes strongly suggest that these sequences of images may contain enough information to distinguish between classically integrable and chaotic systems@xcite . to our knowledge the morphology of the patterns of time ( in)dependent probability distributions has not been studied . in this paper
we propose a method , based on concepts from integral geometry , to perform a morphological image analysis ( mia ) of the probability distribution .
we use mia to characterize the solutions of the time - dependent schrdinger equation for various 2d and 3d billiards .
we demonstrate that indeed , one can define geometrical and topological descriptors that are useful to distinguish between classically integrable and chaotic billiards .
our motivation to analyze time - dependent instead of stationary probability distributions is that the former usually provide information about many eigenstates simultaneously . by following the system over a sufficiently long period of time ,
one collects information on all those eigenstates that have non - negligible overlap with the initial state .
this `` parallelism '' , absent in the stationary - state approach@xcite , reduces the computational work .
mia amounts to the characterization of the geometry and topology of any pattern or series of patterns by means of the so - called minkowski functionals , known from integral geometry @xcite .
these functionals are related to familiar quantities : in two ( three ) dimensions they correspond to the covered area , perimeter and connectivity ( volume , surface area , integral mean curvature and connectivity ) of the pattern . in integral geometry
the calculation of the minkowski functionals is relatively straightforward and requires little computational effort .
mia has proven to be a powerful tool in different fields where the description of patterns is important : statistical physics , to describe the morphology of porous media and complex fluids ; cosmology , to analyze the large scale distribution of matter in the universe ; chemistry , to describe the morphology of patterns in reaction diffusion systems @xcite ; seismology , to describe the spatial complexity of regional seismicity realizations@xcite ; polymer science , to identify and quantify the morphology of mesoscale structures in block copolymers @xcite .
furthermore , in the semi - classical limit the sum of exponentials of eigenfrequencies of a billiard - shaped drum can be written in terms of the minkowski functionals of the billiard@xcite , and the weyl hypothesis states that the number of energy levels up to an energy @xmath0 can be written in a similar manner ( see e.g. @xcite ) . in section 2
we briefly review the theory of minkowski functionals and explain the technique of mia in more detail .
we then apply mia to the motion of a quantum particle in various two and three - dimensional billiards and show that an appropriate quantity can be defined to distinguish classically integrable and chaotic billiards .
finally , we study the transition to chaos for a system whose degree of chaos is parametrized .
at the heart of mia lies the computation of minkowski functionals of an arbitrary ( thresholded ) image . therefore
, in this section we will give some mathematical background on minkowski functionals ( closely following @xcite ) and sketch the outline of an algorithm to compute these functionals .
we first define the minkowski functionals for compact convex sets .
the parallel set @xmath1 of a compact convex set @xmath2 at a distance @xmath3 is the union of all closed balls of radius @xmath3 , the centers of which are points in @xmath2 .
the well - known euclidean measures of @xmath1 , such as the area and volume , can be written as polynomials in @xmath3 .
for example , the parallel volume of a square @xmath4 and a cube @xmath5 with edge lengths @xmath6 are given by @xmath7 in general , the volume @xmath8 in @xmath9 dimensions of the parallel set @xmath1 at distance @xmath3 of @xmath2 is given by the steiner formula @xcite @xmath10 where @xmath11 are the minkowski functionals .
for the first three dimensions it can be shown that@xcite @xmath12 where @xmath13 and @xmath14 denote respectively the length , area , perimeter , volume , surface area , mean breath and euler characteristic or connectivity number of the convex set @xmath2 .
the mean breath @xmath15 is closely related to the mean curvature @xmath16 where @xmath17 and @xmath18 are the principal radii of curvature@xcite .
the euler characteristic @xmath19 is related to the gaussian curvature @xmath20 and is equal to unity for a convex body@xcite .
the mean curvature and the gaussian curvature are two useful measures of the curvature of a surface@xcite .
if the local surface geometry has an elliptic , euclidean or hyperbolic shape respectively then the gaussian curvature is respectively positive , zero or negative@xcite . up to now
, we have limited the discussion to convex sets . in order to deal with the more general sets one encounters in characterizing images
, we consider the convex ring @xmath21 , which is the class of all subsets @xmath22 which can be expressed as finite unions of compact convex sets @xmath23 @xmath24 the euler characteristic @xmath19 is defined as an additive functional on @xmath21 , so that for @xmath25 @xmath26 and where for compact convex sets @xmath2 @xmath27 the euler characteristic describes @xmath22 in a pure topological way , i.e. without reference to any kind of metric .
it can be shown that this definition of @xmath19 permits all minkowski functionals on @xmath21 to be written in terms of @xmath19 and the kinematical density @xcite .
the functional @xmath19 as defined in integral geometry is the same as the euler characteristic defined in algebraic topology @xcite : for @xmath28 , @xmath19 equals the number of connected components minus the number of holes and in three dimensions @xmath19 is given by the number of connected components minus the number of tunnels plus the number of cavities .
for example , for a solid cube @xmath29 , for a hollow cube @xmath30 and for a cube pierced by a tunnel @xmath31 . for multiply connected structures
the euler characteristic is negative . for complex structures it is often difficult to identify the number of connected components , tunnels and cavities .
however , mia directly yields @xmath19 as will be explained below . in order to calculate the morphological properties of a particular image
we first convert the image to a black - and - white picture using a threshold .
pixels ( voxels ) with intensity below the threshold are considered to be part of the background and the others are building up the objects in the 2d ( 3d ) picture . according to integral geometry , the morphological properties of the various objects building up the black - and - white picture can be completely described in terms of minkowski functionals @xcite . in order to calculate the minkowski functionals in an efficient way we consider each pixel ( voxel ) as the union of the disjoint collection of its interior , faces ( for the 3d case only ) , open edges and vertices .
the values of @xmath22 , @xmath32 , @xmath33 , @xmath4 , @xmath15 and @xmath19 for these single open structures can easily be calculated @xcite . by making use of the property of additivity of the minkowski functionals and
the fact that there is no overlap between open bodies on a lattice , we compute the minkowski functionals of the whole image by starting from a complete white image , by calculating the change in the number of interiors , faces ( for the 3d case only ) , open edges and vertices when one pixel ( voxel ) is added to the 2d ( 3d ) image , and this until all pixels ( voxels ) building up the picture are added .
the minkowski functionals can then be computed from @xmath34 where @xmath35 ( @xmath36 ) denote the number of squares ( cubes ) in 2d ( 3d ) , @xmath37 counts the number of faces ( in 3d only ) , @xmath38 and @xmath39 denote the number of edges and vertices respectively . from ( 10 )
and ( 11 ) it follows directly that the minkowski functionals contain information about local 4 ( 8)-point correlation functions in 2d ( 3d ) .
a more detailed description of the algorithm and an example of computer code to compute the minkowski functionals is given in ref.@xcite .
the application of mia to the time evolution of the probability distribution of a quantum particle moving in a billiard is straightforward . we solve the time - dependent schrdinger equation for a particle moving in a billiard by a stable and accurate numerical method@xcite . for practical purposes the results obtained are exact .
a time series of snapshots displaying the probability density can easily be extracted from this data .
a collection of digital videos can be found in @xcite .
each pattern of the time series is converted into a black - and - white picture by applying a threshold .
the result of applying a threshold of 10% to a representative image of the probability density is shown in fig.[fig : fig1 ] .
( 8.5,3.85 ) ( 0.5,0.35 ) ( 4.5,0.35 ) for each black - and - white image in the time series we compute the minkowski functionals using the algorithm described above and analyze the behavior of the minkowski functionals as a function of the control parameters such as threshold , initial conditions of the wave packet , and shape of the billiard .
we use gaussian wave packets as initial states and study a variety of classically integrable and non - integrable billiards ( see figs.[fig : fig2 ] , [ fig : pot3d],[fig : fig4 ] ) .
( 7.6,7.9 ) ( 0,0.25 ) in our numerical work we express lengths in units of a fixed wavelength @xmath40 and rescale energy ( and time , setting @xmath41 ) such that a wave packet with average momentum @xmath42 has an average kinetic energy of one @xcite .
as will become clear from the examples given below , to distinguish between classically integrable and chaotic systems it is expedient to define the time - averaged minkowski functionals @xmath43 and the quantities @xmath44 where @xmath45 ( @xmath46 ) in two ( three ) dimensions .
( 10,7.4 ) ( 0,7.1)classically integrable billiards ( 5.2,7.1)classically non - integrable billiards ( 0,4.4 ) ( 2.7,4.4 ) ( 5.1,4.4 ) ( 7.8,4.4 ) ( 1.1,3.9)k ( 3.5,3.9)l ( 6.3,3.9)k ( 8.7,3.9)l ( 0,0.9 ) ( 2.7,0.9 ) ( 5.1,0.9 ) ( 7.8,0.9 ) ( 1.1,0.4)m ( 3.5,0.4)n ( 6.3,0.4)m ( 8.7,0.4)n ( 3.6,2.0 ) ( 0,0.25 )
5 we present results that illustrate the typical behavior of the minkowski functionals as a function of time , keeping the threshold fixed . the left panel of fig . 5 shows the perimeter @xmath47 as a function of time @xmath48 for the billiards ( a , a ) using a threshold @xmath49 .
billiard a ( solid line ) shows a behavior which is manifestly different from that of billiard a ( dashed line ) : for times @xmath50 the solid line displays large fluctuations , in contrast to the dashed line .
other pairs of billiards ( b , b ) etc .
show very similar , characteristic fluctuations in the perimeter as a function of time ( results not shown ) .
essentially the same behavior is found for the euler characteristic @xmath51 .
an example is given in the right panel of fig.5 for the case of billiards ( b , b ) and @xmath52 .
hence , in conclusion , for integrable billiards @xmath47 and @xmath53 show large fluctuations while for non - integrable billiards these large fluctuations do not occur .
( 17.5,4.5 ) ( 0,0.2 ) ( 8.8,0.2 ) since the euler characteristic @xmath53 is a global measure of the curvatures of the objects in the black - and - white image and provides direct quantitative information about the topology of the probability distribution ( see @xcite , pp .
34 , 112 and 113 ) we will study the behavior of @xmath19 as a function of @xmath48 and @xmath54 in more detail . for each choice of the threshold @xmath54 , the cumulative time average of the euler characteristic @xmath55 approaches a constant value which we denote by @xmath56 .
a representative example of the behavior of @xmath55 is shown in fig.[fig : fig6 ] .
( 8.6,4.5 ) ( 0,0.2 ) in fig.[fig : fig7 ] we plot the results for @xmath56 , for each of the billiards of fig.[fig : fig2 ] . as @xmath54 approaches zero , the thresholded black - and - white picture becomes completely filled with black pixels , hence @xmath57 , for any image .
obviously , for @xmath58 , @xmath56 does not contain any useful information and for reasons of clarity we therefore omit data for @xmath59 in fig .
[ fig : fig7 ] .
( 8.8,9 ) ( 0,0.2 ) .@xmath60 and the corresponding @xmath54 for all billiards shown in fig.[fig : fig2 ] . [ cols="^,^,^",options="header " , ]
an interesting case to confirm that @xmath61 can be used to detect the transition from integrable to chaotic behavior is the lemon - shaped billiard @xcite .
two instances of the lemon - shaped billiard are shown in fig.[fig : fig4 ] .
the boundaries in the @xmath62 plane are defined by @xmath63 , @xmath64 $ ] .
the integrability of the lemon - shaped billiard depends on the shape parameter @xmath65 : for @xmath66 and @xmath67 the billiard is integrable and otherwise it is not .
if @xmath65 slightly deviates from one , the billiard is classically partially chaotic ( periodic orbits and chaotic trajectories coexist ) .
fig.[fig : fig9 ] shows @xmath60 for the lemon - shaped billiard , for the shape parameter @xmath65 varying between 0.5 and 2.0 .
( 8.6,5.7 ) ( 0,0.2 ) the sharp peak centered around @xmath68 shows that @xmath69 is very sensitive for the presence of classically unstable trajectories .
if there are any unstable trajectories in the classical system , the value of @xmath61 for the quantum mechanical equivalent can be classified as chaotic .
thus , the curve in fig .
[ fig : fig9 ] clearly shows the transition from chaotic to integrable behavior and vice versa .
we have described a method , based on concepts of integral geometry , to analyze the time evolution of the probability distribution of a quantum particle moving in two and three - dimensional billiards .
we have demonstrated that the time - averaged euler characteristic can be used to classify billiards as integrable or not . for three - dimensional billiards also the time - averaged mean breadth ( or integral mean curvature ) of the probability density can be employed for this purpose . for lemon - shaped billiards
we have shown that the transition from integrable to chaotic behavior can be determined from the dependence of the time - averaged euler characteristic on the parameter that controls the shape of the billiard .
we believe it may also be of interest to apply the method described in this paper to the time - independent solutions .
we would like to thank the dutch ` stichting nationale computer faciliteiten ' ( ncf ) for their support and a. lande for his critical reading of the manuscript .
99 s. blumberg and g. e. porter , phys rev . * 110 * , 786 ( 1958 ) .
m. v. berry and m. tabor , proc .
london , ser .
a * 356 * , 375 ( 1977 ) .
s. w. mcdonald and a. n. kaufman , phys .
* 42 * , 1189 ( 1979 ) .
v. lopac , i. mrkonji and d. radi , phys .
rev . e.*59 * , 303 ( 1999 ) .
h. primack and u. smilansky , phys .
reports * 327 * , 1 ( 2000 ) .
o. bohigas , m. giannoni and c. schmidt , phys .
lett . * 52 * , 1 ( 1984 ) .
j. zakrzewski , k. dupret and d. delande , phys .
lett . * 74 * , 522 ( 1995 ) .
p. crehan , j. phys .
a : math . gen . * 28 * , 6389 ( 1995 ) .
s. w. mcdonald , lawrence berkeley lab .
report lbl-14837 ( 1983 ) .
r. d. taylor and p. brumer , faraday discuss .
chem . soc . * 75 * , 170 ( 1983 ) .
e. j. heller , phys .
lett . * 53 * , 1515 ( 1984 ) .
e. b. bogomolny , physica d * 31 * , 169 ( 1988 ) .
m. v. berry , proc .
a * 423 * , 219 ( 1989 ) .
m. feingold _
et al . _ ,
* a146 * , 199 ( 1990 ) .
b. li , m. robnik and b. hu , phys .
e * 57 * , 4095 ( 1998 ) .
h. hadwiger , _ vorlesungen ber inhalt , oberflche und isoperimetrie _
( springer - verlag , berlin , 1957 ) .
l. a. santal , _ integral geometry and geometric probability _ ( addison - wesley , reading , massachusetts , 1976 ) .
d. stoyan , w. s. kendall and j. mecke , _ stochastic geometry and its applications _
( akademie verlag , berlin , 1989 ) .
k. r. mecke , int .
b * 12 * , 861 ( 1998 ) .
makarenko , l.m .
karimova , a.g .
terekhov , a.v .
kardashev , physics of the solid earth * 36 * , 305 ( 2000 ) .
k. michielsen , h. de raedt , t. kawakatsu , morphological image analysis , in computer simulation studies in condensed matter physics xiii , d.p .
landau , k.k .
mon , h.d .
schtler , eds .
( in press ) .
k. michielsen , h. de raedt , j.g.e.m .
fraaije , prog .
. suppl . * 138 * , 543 ( 2000 ) .
m. kac , amer .
monthly * 73 * , 1 ( 1966 ) .
h. p. baltes and e. r. hilf , _ spectra of finite systems _ ( bibliographisches institut , mannheim , 1976 ) .
s. hyde , s. andersson , k. larsson , z. blum , t. landh , s. lidin , b.w .
ninham , _ the language of shape ; the role of curvature in condensed matter : physics , chemistry and biology _
( elsevier , amsterdam , 1997 ) .
k. michielsen and h. de raedt , _ morphological image analysis _ , comp .
( in press ) . h. de raedt and k. michielsen , comp . in phys .
* 8 * , 600 ( 1994 ) .
t. papenbrock , phys .
e * 61 * , 4626 ( 2000 ) . | morphological image analysis is applied to the time evolution of the probability distribution of a quantum particle moving in two and three - dimensional billiards .
it is shown that the time - averaged euler characteristic of the probability density provides a well defined quantity to distinguish between classically integrable and non - integrable billiards . in three dimensions the time - averaged mean breadth of the probability density may also be used for this purpose . |
he google books data set is captivating both for its availability and its incredible size .
the first version of the data set , published in 2009 , incorporates over 5 million books @xcite .
these are , in turn , a subset selected for quality of optical character recognition and metadata e.g .
, dates of publication from 15 million digitized books , largely provided by university libraries .
these 5 million books contain over half a trillion words , 361 billion of which are in english . along with separate data sets for american english , british english , and english fiction
; the first version also includes spanish , french , german , russian , chinese , and hebrew data sets .
the second version , published in 2012 @xcite , contains 8 million books with half a trillion words in english alone , and also includes books in italian .
the contents of the sampled books are split into case - sensitive @xmath0-grams which are typically blocks of text separated into @xmath0 = 1 , , 5 pieces by whitespace e.g .
, `` i '' is a 1-gram , and `` i am '' is a 2-gram -gram lengths range from one to five , contain some exceptions to this rule of thumb .
for example , in the 2009 version , `` i am . ''
is tokenized as the 3-gram , `` i am . ''
the period is counted in the corpus as a 1-gram , as are other punctuation marks .
similarly , concatenations are often tokenized as 2-grams , so `` i ve been '' becomes `` i ve been''also a 3-gram . ] . a central if subtle and deceptive feature of the google books corpus , and for others composed in a similar fashion , is that the corpus is a reflection of a library in which only one of each book is available .
ideally , we would be able to apply different popularity filters to the corpus .
for example , we could ask to have @xmath0-gram frequencies adjusted according to book sales in the uk , library usage data in the us , or how often each page in each book is read on amazon s kindle service ( all over defined periods of time ) .
evidently , incorporating popularity in any useful fashion would be an extremely difficult undertaking on the part of google .
we are left with the fact that the google books library has ultimately been furnished by the efforts and choices of authors , editors , and publishing houses , who collectively aim to anticipate or dictate what people will read .
this adds a further distancing from `` true culture '' as the ability to predict cultural success is often rendered fundamentally impossible due to social influence processes @xcite we have one seed for each tree but no view of the real forest that will emerge . we therefore observe that the google books corpus encodes only a small - scale kind of popularity : how often @xmath0-grams appear in a library with all books given ( in principle ) equal importance and tied to their year of publication ( new editions and reprints allow some books to appear more than once ) .
the corpus is thus more akin to a lexicon for a collection of texts , rather than the collection itself .
but problematically , because google books @xmath0-grams do have frequency of usage associated with them based on this small - scale popularity , the data set readily conveys an illusion of large - scale cultural popularity .
an @xmath0-gram which declines in usage frequency over time may in fact become more often read by a particular demographic focused on a specific genre of books . for example , `` frodo '' first appears in the second google books english fiction corpus in the mid 1950s and declines thereafter in popularity with a few resurgent spikes @xcite .
while this limitation to small - scale popularity tempers the kinds of conclusions we can draw , the evolution of @xmath0-grams within the google books corpus their relative abundance , their growth and decay still gives us a valuable lens into how language use and culture has changed over time .
our contribution here will be to show : 1 . a principled approach for exploring word and phrase evolution ; 2 . how the google books corpus is challenged in other , orthogonal respects , particularly by the inclusion of scientific and medical journals ; and 3 .
how future analyses of the google books corpus should be considered . for ease of comparison with related work
, we focus primarily on 1-grams from selected english data sets between the years 1800 and 2000 . in this work
, we will use the terms `` word '' and `` 1-gram '' interchangeably for the sake of convenience .
the total volume of ( non - unique ) english 1-grams grows exponentially between these years , as shown in fig .
[ fig : volume ] , except during major conflicts e.g . , the american civil war and both world wars when the total volume dips substantially .
we also observe a slight increase in volume between the first and second version of the unfiltered english data set .
between the two english fiction data sets , however , the total volume actually decreases considerably , which indicates insufficient filtering was used in producing the first version , and immediately suggests the initial english fiction data set may not be appropriate for any kind of analysis .
the simplest possible analysis involving any google books data set is to track the relative frequencies of a specific set of words or phrases .
examples of such analyses involve words or phrases surrounding individuality @xcite , gender @xcite , urbanization @xcite , and time @xcite , all of which are of profound interest
. however , the strength of all conclusions drawn from these must take into account both the number of words and phrases in question ( anywhere from two @xcite to twenty @xcite or more at a time ) and the sampling methods used to build the google books corpus .
many researchers have carried out broad analyses of the google books corpus , examining properties and dynamics of entire languages .
these include analyses of zipf s and heaps laws as applied to the corpus @xcite , the rates of verb regularization @xcite , rates of word introduction and obsolescence and durations of cultural memory @xcite , as well as an observed decrease in the need for new words in several languages @xcite .
however , these studies also appear to take for granted that the data sets sample in a consistent manner from works spanning the last two centuries .
analysis of the emotional content of books suggests a lag of roughly a decade between exogenous events and their effects in literature @xcite , complicating the use of the google books data sets directly as snapshots of cultural identity .
as we will demonstrate , an assumption of unbiased sampling of books is not reasonable during the last century and especially during recent decades , which is of particular importance to all analyses concerned with recent social change . since parsing in the data sets is case - sensitive , we can give a suggestive illustration of this observation in fig .
[ fig : ffig ] , which displays the relative ( normalized ) frequencies of `` figure '' versus `` figure '' in both versions of the corpus and for both english and english fiction . in both versions of the english data set , the capitalized version , `` figure
, '' surpasses its lowercase counterpart during the 1960s . since the majority of books in the corpus originated in university libraries @xcite , a major effect of scientific texts on the dynamics of the data set is quite plausible
this trend is also apparent albeit delayed in the first version of the english fiction data set , which again suggests insufficient filtering during the compilation process for that version .
because of google books library - like nature , authors are not represented equally or by any measure of popularity in any given data set but are instead roughly by their own prolificacy .
this leaves room for individual authors to have noteworthy effects on the dynamics of the data sets , as we will demonstrate in section [ sec : discussion ] .
lastly , due to copyright laws , the public data sets do not include metadata ( see supporting online material @xcite ) , and the data are truncated to avoid inference of authorship , which severely limits any analysis of censorship @xcite in the corpus . under these conditions
, we will show that much caution must be used when employing these data sets with a possible exception of the second version of english fiction to draw cultural conclusions from the frequencies of words or phrases in the corpus .
we structure the remainder of the paper as follows . in sec . [
sec : methods ] , we describe how to use jensen - shannon divergence to highlight the dynamics over time of both versions of the english and english fiction data sets , paying particular attention to key contributing words . in sec .
[ sec : discussion ] , we display and discuss examples of these highlights , exploring the extent of the scientific literature bias and issues with individual authors ; we also provide a detailed inspection of some example decade decade comparisons .
we offer concluding remarks in section [ sec : conc ] .
we examine the dynamics of the google books corpus by calculating the statistical divergence between the distributions of 1-grams in two given years . a commonly used measure of statistical divergence
is kullback - leibler ( kl ) divergence @xcite , based on which we use a bounded , symmetric measure . given a language with @xmath1 unique words and 1-gram distributions @xmath2 in the first year and @xmath3 in second
, the kl divergence between @xmath2 and @xmath3 can be expressed as @xmath4 where @xmath5 is the probability of observing the @xmath6 1-gram random chosen from the 1-gram distribution for first year , and @xmath7 is the probability of observing the same word in the second year .
the units of kl divergence is bits , and may be interpreted as the average number of bits wasted if a text from the first year is encoded efficiently , but according to the distribution from the latter , incorrect year . to demonstrate this
, we may rewrite the previous equation as @xmath8 where @xmath9 is the shannon entropy @xcite , also the average number of bits required per word in an efficient encoding for the original distribution ; and the remaining term is the average number of bits required per word in an efficient , but mistaken , encoding of a given text .
however , if a single ( say , the @xmath10 ) 1-gram in the language exists in the first year , but not in the second , then @xmath11 , and the divergence diverges .
since this scenario is not extraordinary for the data sets in question , we instead use jensen - shannon divergence ( jsd ) @xcite given by @xmath12 where @xmath13 is a mixed distribution of the two years .
this measure of divergence is bounded between 0 when the distributions are the same and 1 bit in the extreme case when there is no overlap between the 1-grams in the two distributions .
if we begin with a uniform distribution of @xmath1 species and replace @xmath14 of those species with @xmath14 entirely new ones , the jsd between the original and new distribution is @xmath15 , the proportion of species replaced .
the jsd is also symmetric , which is an added convenience .
the jsd may be expressed as @xmath16 from which it is apparent that a similar waste analogy holds as with kl divergence , with the mixed distribution taking the place of the approximation regardless of the year a text was written .
the form for jensen - shannon divergence given in eq .
[ eq : jsd ] can be broken down into contributions from individual words , where the contribution from the @xmath6 word to the divergence between two years is given by @xmath17 some rearrangement gives @xmath18 where @xmath19 , so that contribution from an individual word is proportional to the average probability of the word , and the proportion depends on the ratio between the smaller probability ( without loss of generality ) and the average .
namely , we may reframe the equation above as @xmath20 words with larger average probability yield greater contributions as do those with smaller ratios , @xmath21 , between the smaller and average probability .
so while a common 1-gram such as `` the , '' `` if , '' or a period changing subtly can have a large effect on the divergence , so can an uncommon ( or entirely new ) word given a sufficient shift from one year to the next .
the size of the contribution relative to the average probability is displayed in fig .
[ fig : contribution_curve ] for ratios ranging from 0 to 1 .
@xmath22 is symmetric about @xmath23 ( i.e. , no change ) , so no novel behavior is lost by omitting the case where @xmath24 ( i.e. , when @xmath5 is the larger probability ) .
the maximum possible contribution ( in bits ) is precisely the average probability of the word in question , which occurs if and only if the smaller probability is 0 .
no contribution is made if and only if the probability remains unchanged . between the smaller relative probability of an element and the average
, @xmath25 is the proportion of the average contributed to the jensen - shannon divergence ( see eqs .
[ eq : contribution - messy ] and [ eq : contribution ] ) . in particular ,
if @xmath26 ( no change ) , then the contribution is zero ; if @xmath27 , the contribution is half its probability in the distribution in which it occurs with nonzero probability .
, scaledwidth=35.0% ] we coarse - grain the data at the level of decades e.g . ,
between 1800-to-1809 and 1990-to-1999by averaging the relative normalized frequency of each unique word in a given decade over all years in that decade .
( each year is weighted equally . )
this allows convenient calculation and sorting of contributions to divergence of individual 1-grams between any two time periods .
the dashed lines highlight the divergences to and from the year 1880 , which are featured in fig .
[ fig:1g_divcuts ] .
the off - diagonal elements represent divergences between consecutive years , as in fig .
[ fig:1g_divcuts2 ] .
the color represents the percentage of the maximum divergence observed in the given time range for each data set .
the divergence between a year and itself is zero . for any given year
, the divergence increases with the distance ( number of years ) from the diagonal sharply at first , then gradually .
interesting features of the maps are the presence of two cross - hairs in the first half of the 20th century , which strongly suggests a wartime shift in the language , as well as an asymmetry that suggests a particularly high divergence between the first half century and the last quarter century observed .
, scaledwidth=48.0% ] fig .
[ fig:1g_heatmaps ] shows the jsd between the 1-gram distributions for every pair of years between 1800 and 2000 contributed by 1-grams present above a threshold normalized frequency of @xmath28 for both versions of the english and english fiction data sets ( i.e. , words that appear with normalized frequency at least 1 in @xmath29 ) .
a major qualitative aspect apparent from the heatmaps is a gradual increase in divergence with differences in time the lexicon underlying google steadily evolves though this is strongly curtailed for the second english fiction corpus .
we see the heatmaps are `` pinched '' toward the diagonal in the vicinities of the two world wars . also visible
is an asymmetry that suggests a particularly high divergence between the first half century and the last quarter century observed .
we examine these effects more closely in figs .
[ fig:1g_divcuts ] and [ fig:1g_divcuts2 ] by taking two slices of the heatmaps .
we specifically consider the divergences of each year compared with 1880 ( dashed lines ) , and the divergences between consecutive years ( off - diagonal ) . to verify qualitative consistency
, we also include analogous contribution curves using the more restrictive threshold of @xmath30 . .
contributions are counted for all words appearing above a @xmath28 threshold in a given year ; for the dashed curves , the threshold is @xmath30 .
typical behavior in each case consists of a relatively large jump between one year and the next with a more gradual rise afterward ( in both directions ) .
exceptions include wartime , particularly the two world wars , during which the divergence is greater than usual ; however , after the conclusion of these periods , the cumulative divergence settles back to the previous trend .
initial spikiness in ( d ) is likely due to low volume .
, scaledwidth=48.0% ] . for the solid curves ,
contributions are counted for all words appearing above a @xmath28 threshold in a given year ; for the dashed curves , the threshold is @xmath30 .
divergences between consecutive years typically decline through the mid-19th century , remain relatively steady until the mid-20th century , then continue to decline gradually over time .
, scaledwidth=48.0% ] while the initial divergence between any two consecutive years is noticeable , the divergence increases ( for the most part ) steadily with the time difference .
the cross - hairs from the heatmap resolve into war - time bumps in divergence , which quickly settle in peacetime .
the larger boost to the divergence in recent decades , however , is more persistent suggesting a more fundamental change in the data set , which we will examine in more depth later in this section .
divergences between consecutive years typically decline through the mid-19th century .
divergences then remain relatively steady until the mid-20th century , then continue to decline gradually over time , which may be consistent with previous findings of decreased rates of word introduction and increased rates of word obsolescence in many google books data sets over time @xcite and a slowing down of linguistic evolution over time as the vocabulary of a language expands @xcite .
the initial spikes in divergence in the second version of the fiction data set are likely due to the lower initial volume observed in fig .
[ fig : volume ] .
we present `` word shifts '' @xcite for a few examples of inter - decade divergences in figs .
[ fig : barsalolnew30][fig : barsficnew50 ] , specifically comparing the 1940s to the 1930s and the 1980s to the 1950s for the first unfiltered english data set ( figs .
[ fig : barsalolnew30][fig : barsallnew50 ] ) and both english fiction data sets ( figs .
[ fig : barsficold30][fig : barsficnew50 ] ) .
we provide a full set of such comparisons in the supporting information files s1 , s2 , s3 , and s4 . for each of the four data sets ,
the largest contributions to all divergences generally appear to be from increased relative frequencies of use of words between decades . for the unfiltered data sets , these are in turn heavily influenced by increased mention of years , which is less pronounced for english fiction .
the 1940s literature , unsurprisingly , features more references to hitler and war than the 1930s , along with other world war ii - related military and political terms .
this is seen regardless of the specific data set used and is fairly encouraging .
curiously , regardless of the specific data set , a noticeable contribution is given by an increase in relative use of the words `` lanny '' and `` budd , '' in reference to one character ( lanny budd ) frequently written about by upton sinclair during that decade .
in the fiction data sets , this character dominates the charts . a comparison of the 1930s and 1940s for the second version of the unfiltered english data set ( fig .
[ fig : barsalolnew30 ] ) shows dynamics dominated by references to years .
( the first version is similar . for analogous figures
see the si files s1 , s2 , s3 , and s4 . )
eight of the top ten contributions to the divergence between those decades are due to increased relative frequencies of use of each of years between 1940 and 1949 , their contribution decreasing chronologically , and the other two top ten words are the last two years of the previous decade ( `` 1948 '' and `` 1949 '' appear at ranks 15 and 34 , respectively ) .
the last three years in the 1920s also appear by way of decreased relative frequency of use in the top 60 contributions .
other notable differences include : * the 11th highest contribution is from `` war , '' which increased in relative frequency .
* `` hitler '' and `` nazi '' ( increased relative frequencies ) are ranked 18th and 26th , respectively .
* parentheses ( 13th and 14th ) show increased relative frequencies of use . *
personal pronouns show decreased relative frequencies of use .
* the word `` king '' ( 41st ) also shows a decreased relative frequency , possibly due to the british line of succession . * the top two contributions between the 1950s and the 1980s ( see fig .
[ fig : barsallnew50 ] ) in the english data set are both parentheses , which show dramatically increased relative frequencies of use . *
combined with increased relative frequencies for the colon ( 4th ) , solidus / virgule ( or forward slash ) ( 14th ) , `` computer '' ( 32nd ) , and square brackets ( 58th and 59th ) , this suggests that the primary changes between the 1950s and the 1980s are due specifically to computational sources . *
other technical words showing noticeable increases include `` model '' ( 34th ) , `` data '' ( 35th ) , `` percent '' and the percentage sign ( 37th and 39th ) , `` figure '' ( 40th ) , `` technology '' ( 51st ) , and `` information '' ( 56th ) . *
similarly to the divergence between the 1930s and 1940s , 19 out of the top 30 places are accounted for by increased relative frequencies of use in years between 1968 and 1980 .
* the words `` the '' ( 3rd ) , `` of '' ( 8th ) , and `` which '' ( 16th ) all decrease noticeably in relative frequency and are the highest ranked alphabetical 1-grams .
* unlike the divergence between the 1930s and 1940s , only masculine pronouns show decreases in the top 60 , while `` women '' ( 55th ) increases .
the first version of english fiction shows similar dynamics to the second version of the unfiltered data set between the 1930s and the 1940s ( see fig .
[ fig : barsficold30 ] ) with yearly mentions dominating the ranks .
some exceptions include : * `` lanny '' rising in rank from 49th to 8th .
* parentheses falling from 13th and 14th to 36th and 37th .
`` ml '' ( increased relative frequency of use in the 1940s ) falling from 31st to 55th .
* `` radio '' ( with increased relative frequency ) rising from 51st to 30th .
* `` king '' is no longer in the top 60 contributions . *
`` patient '' enters the top 60 ( ranked 51st ) .
this similarity between the original english fiction data set and the unfiltered data set also appears in the divergence between the 1950s and the 1980s ( see fig . [
fig : barsficold50 ] ) with parentheses and years dominating .
moreover , `` patients '' ranks 13th ( with increased relative frequency of use ) despite not appearing in the top 60 for the unfiltered data set . these observations , combined with increases in `` levels '' ( 47th ) , `` drug '' ( 51st ) , `` response '' ( 55th ) , and `` therapy '' ( 56th )
demonstrate the original fiction data set is strongly influenced by medical journals .
therefore , this data set can not be considered as primarily fiction despite the label .
fortunately , the same is not true for the second version of the english fiction data set .
this is quickly apparent upon inspection of the two greatest contributions to the divergence between the 1930s and the 1940s ( see fig . [
fig : barsficnew30 ] ) .
the first of these is due to a dramatic increase in the relative frequencies of use of quotation marks , which implies increased dialogue .
the second is the name `` lanny '' in reference to the recurring character lanny budd from 11 upton sinclair novels published between 1940 and 1953 .
`` budd '' ranks 11th in the chart ahead of `` hitler '' ( 13th ) . the normalized frequency series for `` lanny '' and `` hitler '' provided in fig .
[ fig : lannyhitlerfig ] demonstrate that lanny received more mention than hitler during this time period .
the chart is littered with the names of fictional characters : * studs lonigan , the 1930s protagonist of a james t. farrel trilogy , secures the 12th spot .
( naturally , he is mentioned fewer times during the 1940s . ) * dinny cherrel from the 1930s _ the forsyte saga _ by john galsworthy secures rank 15 . * wang yuan from the 1930s _ the house of earth _ trilogy by pearl s. buck ranks 17th and 37th . *
detective bill weigand , a recurring character created by richard lockridge in the 1940s , secures rank 33 .
* the eponymous , original asimov robot from the 1940 short story , `` robbie , '' ranks 19th . *
`` mama '' ( ranked 48th ) is none other than the subject of _ mama s bank account _ , published in 1943 by kathryn forbes .
* `` saburov '' ( ranked 22nd ) from _ days and nights _ by konstantin simonov and `` diederich '' ( ranked 45th ) from _ der untertan _ by heinrich mann are subjects of works translated into english in the 1940s .
we note that while marcel proust ( 56th and 33rd ) , who died in 1922 may be present in the 1940s due to letters translated by mina curtiss in 1949 or other references not technically fiction . similarly ,
`` b.m . ''
( 18th ) may refer to the author b. m. bower .
thus , the vast majority of prominent words in the word shift may be traced not only to authors of fiction , but to the content of their work .
moreover , the greatest contributions to divergence appear to correspond to the most prolific authors , particularly upton sinclair . while there are no names of characters in the top divergences between the 1950s and the 1980s , the updated fiction data set ( fig .
[ fig : barsficnew50 ] ) displays far more variety than the original version , including : * decreases in relative frequencies of masculine pronouns e.g .
, `` he '' ( rank 19 ) and `` himself '' ( rank 48)and corresponding increases for feminine pronouns e.g .
, `` her '' ( 3rd ) , `` she '' ( 5th ) , and `` she '' ( 6th ) .
we present times series for `` he '' and `` she '' in fig .
[ fig : heshefig ] . *
an increase in relative frequencies of contractions ( see ranks 9 , 15 , and 21 ) . *
a decrease in `` shall '' ( 16th ) and `` must '' ( 49th ) , and a variety of increased profanity ( particularly ranks 33 and 51 ) . * decreases in `` mr . ''
( 10th ) and `` mrs . ''
( 17th ) . * various shifts in punctuation , particularly fewer semicolons ( 1st ) and more periods ( 2vd ) .
quotation ( 11th ) and question ( 18th ) marks both see increased relative frequencies of use in the 1980s , and the four - period ellipsis ( 20th ) loses ground to the three - period version ( 22td ) . as our jsd analysis has shown above , the unfiltered english data sets feature more general scientific terms and we compare `` percent , '' `` data , '' `` figure , '' and `` model '' in fig .
[ fig : moreseries ] . the original fiction data set also features these , but also places `` patients , '' `` drug , '' `` response , '' and `` therapy '' among the top 60 contributions .
the primary difference between the unfiltered and original fiction data sets in the 1980s ( compared to the 1950s ) appears to consist of the nature of journals sampled .
the unfiltered components predicted and observed for this particular data set seem to be dominated by medical journals .
as well as having more mentions of time and technical terms ( and parentheses ) in the 1980s than in the 1950s , both unfiltered versions and the first fiction data set include both `` et '' and `` al '' with greater relative frequency in the 1980s .
perhaps more importantly , years do not have a large effect on the dynamics in the second english fiction data set .
we see in fig .
[ fig : yearsfig ] that while peaks for years rise in the unfiltered data , they do not in fiction .
the absence of rising peaks in fiction strongly suggests the rise in peak relative frequencies of years in the larger data set is due to a citation bias in the unfiltered data set from high sampling of scientific journals .
this bias casts strong doubt on conclusions that we as a culture forget things more quickly than we once did based on the observation that half - lives for mentions of a given year decline over time @xcite .
the exponential rise in scientific literature is not a new phenomenon , and as de solla price stated in 1963 @xcite ( p. 81 ) when discussing the half - lives for citations of scientific literature , `` in fields embarrassed by an inundation of literature there will be a tendency to bury as much of the past as possible and to cite older papers less often than is their statistical due . ''
it would see that an explanation for declining half - lives in the mentions of years lies in the dynamics of the memory of scientific discoveries rather than that of culture . for the second fiction data set , we observe in fig .
[ fig : moreseries]b , that `` computer '' gains popularity in the fiction data set despite other technical words remaining relatively steady in usage , as we might expect .
this should be encouraging for anyone attempting to analyze colloquial english , despite the prolificacy bias apparent from the authors such as upton sinclair . in the si files
s1 , s2 , s3 , and s4 , we include the top 60 contributions to divergences between each pair of the 20 decades in each of the four data sets analyzed in this paper . in total ,
760 figures are included ( 190 per data set ) for a grand total of 45,600 contributions .
we highlight some of these here . * for divergences to and from the first decade of the 1800s , many of the contributions are due to a reduction of optical character recognition confusion between the letters ` f ' and ` s ' .
for example , in the second unfiltered data set between the 1800s and 1810s , the top two contributions are due to reductions in `` fame '' and `` os , '' respectively .
the word `` same '' ( ranked 11th ) is the first increasing contribution .
decreased relative frequencies of `` os , '' `` sirst , '' `` thofe , '' `` fo , '' `` fay , '' `` cafe , '' `` fays , '' `` fome , '' and `` faid '' ( ranks 3 through 10 , respectively ) and `` lise '' ( 12th ) all suggest digital misreadings of both f and the long ` s ' .
( the 13th contribution is `` napoleon , '' who is mentioned with greater relative frequently in the 1810s . )
* contributions between the 1830s and the 1860s in the second unfiltered data set highlight the american civil war and its aftermath .
`` state '' ( 11th ) , `` general '' ( 19th ) , `` states '' ( 20th ) , `` union '' ( 37th ) , `` confederate '' ( 48th ) , `` government '' ( 52nd ) , `` federal '' ( 56th ) , and `` constitution '' ( 59th ) all show increased relative frequency of use .
religious terms tend to decline during this period
, `` church '' ( 14th ) , `` god '' ( 24th ) , and `` religion '' ( 58th ) . * between the 1940s and 1960s , the second unfiltered dataset shows increases for `` nuclear '' ( 43rd ) , `` vietnam '' ( 47th ) , and `` communist '' ( 50th ) .
the relative frequency of `` war '' ( 25th ) decreases substantially .
meanwhile in fiction , `` lanny '' ( 5th ) declines , while `` television '' ( 38th ) and the hardy boys ( `` hardy '' ranks 51st ) appear with greater relative frequencies . * between the 1960s and 1970s , the second fiction data set is strongly affected by `` garp '' ( _ the world according to garp _ by john irving , 1978 ) at rank 19 , increased relative frequencies of profanity ( ranks 27 , 33 , and 38 ) , and increased mentions of `` nixon '' ( 41st ) and `` spock '' ( 47th , likely due to `` star trek '' novels ) . * between the 1980s and 1990s , the second fiction set shows increased relative frequencies of use of the words `` gay '' ( 15th ) , `` lesbian '' ( 19th ) , `` aids '' ( 24th ) , and `` gender '' ( 27th ) .
female pronouns ( 2nd , 8th , and 9th ) show increased relative frequencies of use in continuance of fig .
[ fig : barsficnew50 ] .
based on our introductory remarks and ensuing detailed analysis , it should now be clear that the contents of the google books corpus do not represent an unbiased sampling of publications . beyond being library - like , the evolution of the corpus throughout the 1900s is increasingly dominated by scientific publications rather than popular works .
we have shown that even the first data set specifically labeled as fiction appears to be saturated with medical literature .
when examining these data sets in the future , it will therefore be necessary to first identify and distinguish the popular and scientific components in order to form a picture of the corpus that is informative about cultural and linguistic evolution .
for instance , one should ask how much of any observed gender shift in language reflects word choice in popular works and how much is due to changes in scientific norms , as well as which might precede the other if they are somewhat in balance .
even if we are able to restrict our focus to popular works by appropriately filtering scientific terms , the library - like nature of the google books corpus will mean the resultant normalized frequencies of words can not be a direct measure of the `` true '' cultural popularity of those words as they are read ( again , frodo ) .
secondarily , not only will there be a delay between changes in the public popularity of words and their appearance in print , normalized frequencies will also be affected by the prolificacy of the authors . in the case of upton sinclair
s lanny budd , a fictional character was vaulted to the upper echelons of words affecting divergence ( even surpassing hitler ) by virtue of appearing as the protagonist in 11 novels between 1940 and 1953 .
google books is at best a limited proxy for social information after the fact .
the google books corpus s beguiling power to immediately quantify a vast range of linguistic trends warrants a very cautious approach to any effort to extract scientifically meaningful results .
our analysis provides a possible framework for improvements to previous and future works which , if performed on english data , ought to focus solely on the second version of the english fiction data set , or otherwise properly account for the biases of the unfiltered corpus . | it is tempting to treat frequency trends from the google books data sets as indicators of the `` true '' popularity of various words and phrases .
doing so allows us to draw quantitatively strong conclusions about the evolution of cultural perception of a given topic , such as time or gender .
however , the google books corpus suffers from a number of limitations which make it an obscure mask of cultural popularity .
a primary issue is that the corpus is in effect a library , containing one of each book .
a single , prolific author is thereby able to noticeably insert new phrases into the google books lexicon , whether the author is widely read or not . with this understood
, the google books corpus remains an important data set to be considered more lexicon - like than text - like .
here , we show that a distinct problematic feature arises from the inclusion of scientific texts , which have become an increasingly substantive portion of the corpus throughout the 1900s .
the result is a surge of phrases typical to academic articles but less common in general , such as references to time in the form of citations .
we highlight these dynamics by examining and comparing major contributions to the statistical divergence of english data sets between decades in the period 18002000 .
we find that only the english fiction data set from the second version of the corpus is not heavily affected by professional texts , in clear contrast to the first version of the fiction data set and both unfiltered english data sets .
our findings emphasize the need to fully characterize the dynamics of the google books corpus before using these data sets to draw broad conclusions about cultural and linguistic evolution . |
weyl semimetals ( wss ) have weyl points , comprised of a linear crossing of valence and conduction bands@xcite near the fermi energy e@xmath4 .
wss can be realized from a lack of time - reversal symmetry ( magnetic ordering , or applied field @xmath5),@xcite or from no inversion symmetry in three - dimensional compounds,@xcite and show peculiar features , such as fermi arcs from surface states connecting two weyl points with different chirality.@xcite experimental evidence has been obtained using x - ray angle - resolved photoemission spectroscopy.@xcite extremely large magnetoresistivity ( xmr ) also occurs in some wss , with a recent analysis by burkov@xcite concluding that xmr should be expected in high mobility semimetals where weyl or dirac points lie near the fermi surface ( fs ) .
xmr was observed also in the semimetal wte@xmath6 , showing unsaturated mr up to 60 t,@xcite attributed to precise electron and hole carrier compensation , as in bi and other stoichiometric semimetals .
recently a `` nodal loop '' weyl semimetal phase with potentially distinctive properties has been proposed on theoretical grounds@xcite in the nonmagnetic non - centrosymmetric isovalent class taas , tap , nbas , and nbp , whose structure is shown in fig .
1 . a previous proposal of nodal loop wss by phillips and aji invoked time - reversal symmetry breaking@xcite rather than the lack of inversion symmetry of this class of pnictides . since these suggestions , xmr has been observed in both taas@xcite and nbp.@xcite nbp , upon which we focus here , shows xmr of 8,500 at 2 k for @xmath7 t , increasing to @xmath8 at @xmath9 k at @xmath5=30 t , to 10@xmath10 at 60 t , and remains unsaturated.@xcite these colossal xmr values reflect the excellent conduction at low temperatures , with resistivity as low as 0.1@xmath11 cm , and a metal to insulator crossover with field at 100 k and below .
a high mobility @xmath12=@xmath13 @xmath14/v - s and low carrier density of @xmath15 @xmath16 at low @xmath17 was inferred from the conventional single band expressions , which however are not quantitative for multiband semimetals . as temperature increases ,
both the hall coefficient @xmath18 changes sign from electron - like to hole - like and the mobility rapidly changes in the 50 - 150 k range , correlating with the metal to insulator crossover .
a multiband model and additional data will be necessary to sort out individual carrier densities and mobilities .
nbp has the smallest spin - orbit coupling ( soc ) of this class , making it the most delicate example of the features of nodal loop wss . from both theoretical and experimental viewpoints ,
nodal loop semimetals comprise a new class of materials with topological behavior . in this paper , we focus on effects of soc on their electronic properties and evaluate the optical and transport coefficients of nbp , using density functional theory based methods that enable a direct theory - experiment comparison . a central feature that emerges
is electronic fine structure driven by the combination of soc and lack of an inversion center , a feature elaborated on by samokhin@xcite from a model viewpoint .
this splitting in a semimetal with small closed fermi surfaces ( fss ) results in a `` russian doll nested '' pair of fss , and new low energy interband transition strength pilfered from the drude strength by non - centrosymmetry .
the fs character is clarified and low energy collective electronic excitations ( plasmons ) arising from the electronic fine structure are predicted .
while plasmons have been studied in conventional wss,@xcite their appearance in nodal loop wss as well as several other properties discussed here are distinctive when the special `` nodal loop semimetal '' character is encountered .
as shown in fig . [ str ] , nbp has a body - centered tetragonal lattice of the space group @xmath19 ( # 109).@xcite this space group has @xmath20 point group , is non - symmorphic , and ( importantly ) lacks inversion symmetry .
the structure can be pictured as columns of face - shared nbp@xmath21 trigonal prisms oriented along the @xmath22-axis , edge - shared along the @xmath23-axis in one layer .
the trigonal columns are @xmath24 rotated in the next layer along the @xmath25-axis , giving overall tetragonal symmetry . in the structure with the experiment lattice parameters@xcite of @xmath26 , @xmath27 , both nb and p ions lie at @xmath28 positions @xmath29 : @xmath30 @xmath31=0.4176 ; the site symmetry is @xmath32 .
our optimized positions using the generalized gradient approximation ( gga ) with the perdew - burke - ernzerhof ( pbe ) exchange - correlation functional@xcite in fplo@xcite coincided with the experimental values .
this structure leads to nearly identical nb - p bond lengths of [email protected] and p - nb - p bond angles around 82@xmath34 .
calculations with the experimental structure were performed using the all - electron full - potential code wien2k,@xcite with selected results confirmed with fplo .
all results are based on the pbe - gga exchange - correlation functional .
one objective is to determine the combined effects of soc and the lack of inversion center , since delicate features around @xmath35 are sensitive to soc and become important .
calculation of optical properties including soc is available in wien2k.@xcite the dielectric function @xmath36 , with only diagonal @xmath37 and @xmath38 components due to tetragonal symmetry , is decomposed into the intra- and inter - band contributions .
the imaginary parts of each contribution are given by@xcite @xmath39 where @xmath40 is the momentum operator , and @xmath41 and @xmath42 are energies of the occupied @xmath43 and unoccupied @xmath44 orbitals , respectively . in the intraband term , which contains the drude divergence for @xmath45 , @xmath46 is the drude plasma frequency and @xmath47 ( chosen to be 10 mev here ) is an inverse scattering lifetime @xmath48 .
the corresponding real parts are obtained by the kramers - kronig relation .
the transport calculations were carried out using semiclassical bloch - boltzmann transport theory , implemented in the _ boltztrap _ code@xcite that is interfaced to wien2k .
no approximations beyond the constant scattering time approximation are made . in all calculations performed here ,
the brillouin zone was sampled with a dense @xmath49-mesh up to @xmath50 to treat the semimetallic fine structure carefully . in wien2k ,
the basis size was determined by @xmath51 , and augmented plane - wave sphere radii of 2.5 bohr ( nb ) and 2.14 bohr ( p ) were used .
we first address the electronic structure obtained with fine @xmath49-point and energy meshes .
figure [ band ] shows the gga and gga+soc band structures near e@xmath4 .
semimetallic character with a band crossing along at least three symmetry directions appears , with valence - conduction overlaps mostly of the order of 150 mev .
this band behavior leads to a pseudogap centered 20 mev below @xmath35 ( fig .
[ band ] ) , with a very small density of states ( dos ) @xmath52=0.046 states / ev per f.u .
while these results are similar to the previous report,@xcite we examine the origins and consequences more completely .
inclusion of soc leads to unconventional consequences . as usual
, soc converts many band crossings along symmetry lines into anticrossings ; here gapping bands by a few tens of mev near e@xmath4 .
however , the lack of inversion symmetry leads to lifting of band degeneracy over entire symmetry planes,@xcite with splittings in the 20 - 200 mev range . in spite of the modest magnitude of soc in @xmath53 nb
, soc will play an important role in determining both the thermal and spectroscopic properties of nbp .
the fss obtained from gga+soc are displayed in fig .
[ fs ] , see the caption for a description .
they are similar to those presented earlier , but they are displayed in an extended zone manner to illustrate their toroidal origin .
have described in detail@xcite the underlying features of the weyl nature of nbp and isovalent partner compounds . without soc , mirror symmetry protected nodal loops lie precisely in the mirror planes @xmath54 and @xmath55=0 .
the nodal loops consist of band degeneracies @xmath56 ( @xmath57 is spin degeneracy ) .
the locus of @xmath58 points at this intersection forms the ring in the @xmath47zn mirror plane , not at constant energy but , for these compounds , in a small energy range including @xmath35 .
the fs intersects the nodal loop , so it must include an even number of doubly degenerate _ fermi points _ where @xmath56 = e@xmath4 .
such nodal loops have also been found in two dimensional srvo@xmath59 nanolayers.@xcite in this non - centrosymmetric structure , soc splits the bands forming the nodal loops , leaving two symmetry - inequivalent pairs of weyl points ( 12 in all ) near but not on the fss , the points being provided explicitly by weng _
et al._@xcite one lies in the @xmath60 plane , the other at a general @xmath58 point lying near the @xmath47nz plane .
the band splitting is highly anisotropic ; each fs ( fig .
[ fs ] ) consists of a flattened torus ( annulus ) pinched off into one `` boomerang '' containing holes and one `` new moon '' containing electrons .
small doping will alter the position of the pinching , hence changing the number of electron and hole carriers . for a stoichiometric compound
the fss contain an equal number of holes and electrons , and from calculated band fillings we obtain 4.2@xmath010@xmath1 electron ( hole ) carriers per f.u .
this value corresponds to high velocity carriers of each sign separated on average by 13 lattice constants in each direction .
the rms fermi velocities are @xmath61=3.7 , @xmath62=1.6 , in 10@xmath63 cm / s , a factor of 2.35 in anisotropy .
note that the velocities are typical of _ metallic _ compounds , not semimetals , due to the dispersive bands crossing e@xmath4 , _ i.e. _ the weyl character ( analogous to graphene ) .
for the drude plasma energy , the semimetallic value of @xmath52 but normal metal velocities lead to @xmath64=1.0 ev , @xmath65=0.4 ev . what can not be seen in fig .
[ fs ] is the doubling of the fss resulting from the interplay of soc with the non - centrosymmetric crystalline symmetry . with the band degeneracy split ,
fermi surfaces become `` russian doll nested '' pairs , one inside the other .
this band - splitting effect has been discussed for pt - based superconductors@xcite and for electronic properties more generally.@xcite the band spitting aspect is effective both in @xmath49-space ( closely nested fermi surfaces ) and in energy , with consequences discussed below .
transport behavior in semimetals is very sensitive to fermiology and to regions of high curvature , but we have found that calculated values based on quasiclassical bloch - boltzmann theory agree very well with observed behavior.@xcite the hall tensor components @xmath66 including soc , displayed in fig .
[ hall](a ) , show a change in sign at @xmath67=125 k , precisely where the experimental r@xmath68 changes sign.@xcite this change of sign is a strong reminder that in a multiband , compensated semimetal , r@xmath68 bears no direct relation to carrier densities.@xcite this statement is particularly true when the fss are strongly non - ellipsoidal and extremely anisotropic .
the experiment - theory agreement reflects not only the precision of the calculations but also the excellent stoichiometry of the samples ; the stoichiometry could not be established precisely but is supported also by the high mobility .
the magnitude of @xmath66 around 250 k is within a factor of 2 of the experimental value .
this is really excellent agreement , because of the compensating contributions from electron and hole surfaces , which have regions of large curvature .
the expression for @xmath18 in bloch - boltzmann theory has the interpretation of giving it as the average of the curvature over the fss .
a strong temperature dependence and a likely change of sign with temperature should be anticipated in nodal loop wss .
the experimental @xmath18 becomes at the lowest temperature two orders of magnitude larger in magnitude than our calculated value , and it is clear from the `` pointed '' fss in fig .
[ fs ] that evaluating r@xmath68 ( which as mentioned is an average of the fs curvature ) at t=0 is a substantial numerical challenge .
the chemical potential @xmath69 , displayed in fig .
[ hall](c ) , decreases with increasing @xmath17 reflecting the strong differences in the electron and hole densities of states .
this variation is different in sign from another xmr compound wte@xmath6,@xcite reflecting different doss of the valence and conduction bands of the two compounds .
the dos at the chemical potential @xmath70 , shown in fig .
[ hall](b ) , is reduced and achieves a minimum around @xmath71 k before turning around .
figure [ seebeck](a ) displays the conductivity tensor components @xmath72 , which incorporates the anisotropy of @xmath46 , which is @xmath735 times larger in - plane .
the @xmath17-dependent seebeck coefficients , shown in fig .
[ seebeck](b ) , have net @xmath74-type sign .
they peak ( in magnitude ) at 100 - 150 k before again becoming small .
the dependence on chemical potential @xmath12 , which can be varied by doping or by gating , is extremely strong near stoichiometry [ see fig .
the near - perfect cancellation of electron and hole contributions ( it is perfect @xmath75 for @xmath12@xmath3 - 15 mev ) at stoichiometry is destroyed , with a maximum of @xmath76 being attained around 0.5% hole doping , or half of that for electron doping .
the sensitivity makes the thermopower an important gauge of the degree of stoichiometry of samples , which otherwise can be very difficult to determine at this low level .
the band splitting induced by soc and the lack of inversion symmetry have qualitative consequences for the optical properties . in _
conventional _ wss where fermi surfaces are simple and model band structures work well , the dielectric response and plasmon signatures of chirality have been well studied.@xcite nodal loop wss bring new processes to the low - energy , small-@xmath77 dielectric response .
they are compensated semimetals thus providing itinerant electron and hole gases .
they are qualitatively different from other semimetals ( including conventional wss ) by having highly non - parabolic and extended fermi surfaces , and of course by having very closely nested fermi surfaces .
new dielectric phenomena must emerge . without either soc or inversion symmetry breaking
, bands are doubly degenerate and contribute accordingly to the intraband drude response .
in nodal loop wss half of this weight is transferred to finite-@xmath77 , finite-@xmath78 response .
the changes will be appreciable only at small energies , and they should involve new response at very small @xmath77 wavevectors that connect the finely nested fermi surfaces , which are on the order of 10@xmath79@xmath80
. the drude response will be relatively unchanged apart from the factor of 2 reduction in weight , but it may be smeared into the new small-@xmath78 , small-@xmath77 structures . the energy loss function , shown in fig . [ optic ] with scattering rate @xmath47=10 mev , displays low energy plasmons when soc is included , at 20 mev for @xmath81-axis polarization and double in - plane plasmon peaks at 40 mev and 60 mev .
the strength of these peaks , and to some degree their positions , are dependent on the presumed scattering rate @xmath47
. the anisotropy we calculate at low energy , based on the strong anisotropy of the fermi surfaces , may be consistent with the angle - dependence of the mr measurements.@xcite given the reported high mobility , the plasmon peaks may be much sharper at low temperature than calculated here .
the peaks can be expected also to have temperature dependence , because the varying @xmath82 induces changes and @xmath47 will be t - dependent as well .
above 200 mev , we do not expect significant t - dependence of the spectral behavior .
far - infrared spectroscopy may reveal unusual behavior as temperature , field , polarization , and carrier density are varied and tuned .
our study has illuminated the origins of the unusual and delicate fermi surfaces of nbp . without consideration of spin - orbit coupling ,
the electronic structure includes nodal loops@xcite lying in the mirror planes that lead to point fermi surfaces as well as small electron and hole pockets . in this non - centrosymmetric structure ,
spin - orbit coupling removes the spin degeneracy , resulting in russian doll nested pairs of fermi surfaces with a pinched - off annulus topology , with electron surfaces converting to hole surfaces across the pinch .
the resulting electronic fine structure accounts well for the change in sign of the hall coefficient at 100 k , which is also the temperature range in which the magnetoresistance begins to grow in size .
these features identify 10 mev as a relevant energy scale .
the unusual fermi surface also accounts for the lack of universal scaling of xmr in nbp and tap ; spin - orbit coupling is much larger in tap and the fermi surfaces are sensitive to this .
finally , the fine structure in nbp results in the appearance of polarization dependent low energy plasmons in the 20 - 60 mev range .
this study substantially broadens the understanding of how ws behavior impacts the physical properties of nbp , which should also apply in similar form to nbas , tap , and taas .
we acknowledge j .-
hong for useful communications on magnetoresistance and a. s. botana for useful discussion on calculations in boltztrap .
this research was supported by national research foundation of korea grant no .
nrf-2013r1a1a2a10008946 ( k.h.a and k.w.l ) , by u.s .
national science foundation grant dmr-1207622 - 0 ( k.w.l . ) and by u.s .
department of energy grant de - fg02 - 04er46111 ( w.e.p . ) .
huang , s .- y .
xu , i. belopolski , c .- c .
lee , g. chang , b.k .
wang , n. alidoust , g. bian , m. neupane , c. zhang , s. jia , a. bansil , h. lin , and m. z. hasan , a weyl fermion semimetal with surface fermi arcs in the transition metal monopnictide taas class .
commun . * 6 * , 7373 ( 2015 ) .
b. q. lv , n. xu , h. m. weng , j. z. ma , p. richard , x. c. huang , l. x. zhao , g. f. chen , c. matt , f. bisti , v. strokov , j. mesot , z. fang , x. dai , t. qian , m. shi , and h. ding , observation of weyl nodes in taas . nat . phys . * 11 * , 724 ( 2015 ) .
m. a. ali , j. xiong , s. flynn , j. tao , q. d. gibson , l. m. schoop , t. liang , n. haldolaarachchige , m. hirschberger , n. p. ong , and r. j. cava , large , non - saturating magnetoresistance in wte@xmath6 .
nature * 514 * , 205 ( 2014 ) . c. zhang , c. guo , h. lu , x. zhang , z. yuan , z. lin , j. wang , and s. jia , large magnetoresistance over an extended temperature regime in monophosphides of tantalum and niobium .
b * 92 * , 041203(r ) ( 2015 ) . c. shekhar , a. k. nayak , y. sun , m. schmidt , m. nicklas , i. leermakers , u. zeitler , w. schnelle , j. grin , c. felser , and b. yan , extremely large magnetoresistance and ultrahigh mobility in the topological weyl semimetal candidate nbp .
phys . * 11 * , 645 ( 2015 ) . | nbp is one member of a new class of nodal loop semimetals characterized by the cooperative effects of spin - orbit coupling ( soc ) and a lack of inversion center . here
transport and spectroscopic properties of nbp are evaluated using density functional theory methods .
soc together with the lack of inversion symmetry splits degeneracies , giving rise to `` russian doll nested '' fermi surfaces containing 4@xmath010@xmath1 electron ( hole ) carriers / f.u . due to the modest soc strength in nb , the fermi surfaces map out the weyl nodal loops .
calculated structure around t@xmath2@xmath3100 k in transport properties reproduces well the observed transport behavior only when soc is included , attesting to the precision of the ( delicate ) calculations and the stoichiometry of the samples .
low energy collective electron - hole excitations ( plasmons ) in the 20 - 60 mev range result from the nodal loop splitting . |
on 3 november 2000 in a seminar at cern the lep higgs working group presented preliminary results of an analysis indicating a possible 2.9@xmath0 observation of a 115 gev higgs boson @xcite . based on this analysis the four lep collaborations requested the continuation of lep to collect more data at @xmath1 gev . however , the arguments presented by the lep collaborations did not convince the lep management and in retrospect , it turned out that the lep accelerator turn - off date of 2 november 2000 ended its eleven years of forefront research .
figure [ fig - masspeak ] , taken from the above mentioned presentation , shows the reconstructed mass distribution of the background and the signal ( on top of the background ) with the data represented by the dots with error bars .
it is shown for three possible selections of the data samples with increasing signal purity .
none of these distributions shows a clear classical 3@xmath0 excess of data over the expected background and some physicists claimed that this evidence was not convincing enough .
however , the statistical arguments presented by the lep higgs working group were not based on these distributions , but rather on a sophisticated , though beautiful statistical analysis of the data .
two years after the event , when the last analysis of the lep data indicated that the significance of a higgs observation in the vicinity of 115 gev went down to less than 2@xmath0 @xcite , it becomes apparent that the lep standard model ( sm ) higgs heritage will in fact be a lower bound on the mass of the higgs boson .
however , the lep higgs working group has taught us powerful and instructive lessons of statistical methods for deriving limits and confidence levels in the presence of mass dependent backgrounds from various channels and experiments .
these lessons will remain with us long after the lower bound becomes outdated . in this note
we are trying to repeat , in a pedagogical way , this lep higgs statistical lesson . to achieve this we developed a toy monte carlo model containing a simulated background and signal similar to lep conditions at @xmath2 gev .
due to mass resolution effects an event with a reconstructed mass @xmath3 can originate from any higgs mass in its vicinity . to appreciate the significance of a higgs candidate with respect to some test mass @xmath4 the event weights are introduced in section 2 .
the weights of all candidate events are summed up in section 3 to give the likelihood of an experiment in order to quantify its signal - like nature .
the probability density function of the likelihood is used in section 4 to estimate the probability of an experiment without a higgs boson to fluctuate and give a more signal - like outcome than the observed one or the probability of an experiment in the presence of a higgs signal to fluctuate and give a more background - like outcome than the observed one .
confidence levels are introduced for that purpose .
later on in that section , signal confidence levels are defined and used to estimate the exclusion sensitivity of an experiment .
the current lep sm higgs search results are also given , allowing the reader to make his own appreciation and judgement .
when does a higgs candidate event become significant ? what can be considered as a leading or gold plated higgs candidate ?
due to detector resolution and missing energy carried by undetected neutrinos which accompany the higgs decay products , the reconstructed mass of a hypothetical higgs candidate @xmath3 , is not necessarily its physical mass , @xmath4 .
one therefore defines a weight to quantify the significance of a higgs candidate with @xmath3 with respect to a hypothetical higgs with @xmath4 .
this weight is given by @xmath5 ( see next section ) where @xmath6 is the expected number of signal events from an hypothetical higgs boson with a test mass @xmath4 in the vicinity of @xmath3 .
this procedure is illustrated in a series of plots ( figure [ fig - spaghetti ] ) . in this example
we assume there is an event candidate with a reconstructed mass , @xmath7 gev .
we then calculate the event weight for hypothetical higgs test masses in the range 100120 gev ( figs [ fig - spaghetti]a - e ) .
the resulting weights are shown in the last figure of this series ( fig .
[ fig - spaghetti]e ) .
when connecting all the weights with a continuous line , one gets what has become to be known as the spaghetti plot " .
also shown is the spaghetti plot for a candidate event with @xmath8 gev .
as expected , the highest weight is achieved when the reconstructed mass and the hypothetical test higgs mass coincide .
the 17 candidates with the highest weight at a test mass , @xmath9 gev are listed in table [ tab : event - list ] and their corresponding spaghetti plots are shown in figure [ fig - spaghetti - lep ] ( taken from @xcite ) .
.properties of the candidates with the highest weight at @xmath10 gev .
table is taken from @xcite .
[ tab : event - list ] [ cols="^,<,^,^,^,^ " , ] where the background confidence level @xmath11 and the signal+background confidence level @xmath12 at @xmath13 gev , for all lep data combined and for various subsets of the data are shown .
note that the delphi @xmath12 probability is 3.3% .
that does not mean that delphi could exclude a 116 gev higgs boson signal hypothesis , but rather that delphi can exclude a 116 gev _
_ s__+_b _ hypothesis .
however , there is a @xmath14 probability for the delphi background to fluctuate and give a signal - like observation .
it was to take this probability into account that one _ apriori defined _ the signal confidence level to be @xmath15 @xcite . this way the signal hypothesis for @xmath16 gev is excluded by definition at only the 73.8% confidence level , @xmath17 .
this procedure for constructing the signal confidence level is obviously conservative since the coverage probability is in general greater than the @xmath18 . of the three plots shown in figure [ fig - pdflep ]
it is clear now that a 110 gev higgs boson , where the @xmath12 probability is too small to be seen , can be excluded by lep .
this is also seen in table [ tab : clb ] where none of the experiments alone or the combined lep can exclude a 116 gev higgs boson . the exclusion power of lep is illustrated in figure [ fig - exclusion ] @xcite where the median expected signal confidence level , @xmath19 , as a function of the higgs test mass together with its 1 and 2@xmath0 bands is shown .
also shown is the observed @xmath20 confidence level .
the intersection of the horizontal line at @xmath21 with the observed and median expected curves give the 95% observed and expected exclusion _
cl_. thus lep excluded , at 95% _ cl _ , a sm higgs boson with a mass below 114.4 gev , while it had the sensitivity to exclude a 115.3 gev higgs boson .
the small excess of data candidate events around 115116 gev is responsible for the slight reduction of the observed higgs lower mass limit with respect to its potential exclusion sensitivity .
lep standard model higgs search results were reviewed with the emphasis on a pedagogical explanation of the statistical procedure based on a toy model constructed specifically for this purpose .
we would like to thank moshe kugler , rick van kooten and alex read for their useful suggestions and illuminations after a critical reading of this note .
one of the authors ( e.g. ) would like to thank the following people : peter zerwas for inviting him to summarize the lep higgs status in susy02 at desy hamburg , and thereby initiating this work ; alex read for teaching him everything he knows about higgs statistics , chris tully for giving him a private lep higgs statistics lesson and inbar mor for the good vibes and support while completing this work .
search for the standard model higgs boson at lep , aleph , delphi , l3 and opal , the lep working group for higgs boson searches , contributed paper for ichep02 ( amsterdam , july 2002 ) , ` http://lephiggs.web.cern.ch/lephiggs/papers/july2002_sm/index.html ` final results of the searches for neutral higgs bosons in @xmath22 collisions at @xmath23 up to 209 gev / heister , a ; et al . - aleph collaboration .
hep - ex/0201014 ; cern - ep-2001 - 095 .
- geneva : cern , 18 dec 2001 . - 19 p. phys .
, b : 526 ( 2002 ) , pp.191 - 205 | we review the results of the standard model higgs boson search at lep .
an emphasis is put on revealing the details behind the statistical procedure developed by the lep higgs working group .
the procedure is explained using a toy model which allows the reader to estimate the significance of the experimental observation which led at the time to a scientific debate on whether lep has observed a 115 gev higgs boson .
24.11.2002 * higgs statistics for pedestrians * eilam gross and amit klier department of particle physics + weizmann institute of science , rehovot 76100 , israel + _ email : [email protected]_ 12 true pt 10 mm 1.cm |
in recent years , observations of type ia supernovae ( sne ia ) [ 1 , 2 ] , cosmic microwave background ( cmb ) fluctuations [ 3 , 4 ] , and large - scale structures ( lss ) [ 5 , 6 ] indicate that the universe is accelerating , therefore some form of dark energy whose fractional energy density is about @xmath3 must exist in the universe to drive this acceleration .
dark energy has been one of the most active fields in modern cosmology since the discovery of accelerated expansion of our universe .
investigation on the nature of dark energy becomes one of the most important tasks for modern physics and modern astrophysics .
up to now , many candidates of dark energy have been proposed to fit various observations which include the simplest one , the einstein s cosmological constant [ 7 ] , or a dynamical scalar field , such as quintessence [ 8 ] , phantom [ 9 ] , k - essence [ 10 ] , tachyon [ 11 ] and so on .
the present data seem to slightly favor an evolving dark energy with the equation - of - state parameter ( eos ) @xmath4 around present epoch and @xmath5 in the near past .
obviously , @xmath0 can not cross @xmath6 for quintessence or phantom alone .
some efforts have been made to build dark energy model whose eos can cross the phantom divide . in a universe filled with quintessence and phantom fields this case
can be realized easily .
this implement of dark energy , called as quintom , has been first proposed in ref .
[ 12 ] , where the quintom model with an exponential potential and the existence , stability of cosmological scaling solutions in the context of spatially homogeneous cosmological models have been investigated .
phase - plane analysis of the spatially flat frw models shows that the phantom - dominated scaling solution is the unique late - time attractor and there exists a transition from @xmath5 to @xmath4 [ 13 ] .
wei and cai [ 14 ] suggested a hessence model , in which a non - canonical complex scalar field plays the role of dark energy .
the cosmological evolution of the hessence dark energy is also investigated ; it is found that the big rip never appears in the hessence model even in the most general case while beyond particular potentials and interaction forms .
the action of dilaton field in the presence of einstein s cosmological constant has been derived in ref . [
the potential is the counterpart of the einstein s cosmological constant in the dilaton gravity theory .
since it can be reduced to the einstein cosmological constant when the dilaton field is set to zero , the dilaton potential is called the cosmological constant term in the dilaton gravity theory .
compared to the ordinary scalar field , the action for phantom scalar field has only a sign difference before the kinetic term .
later , the explicit expression of the phantom potential have been given in ref .
. a model of the universe dominated by the dilaton field with a liouville type potential has been presented in ref . [
17 ] . in this letter
, we investigate the cosmological evolution of a two - field model of dark energy where one is a dilaton field with canonical kinetic energy and the other is a phantom field with a negative kinetic energy term with liouville type potentials .
let us start from a 4-dimensional theory in which gravity is coupled to dilaton and maxwell field with an action : @xmath7 where _ r _ is the scalar curvature , @xmath8 is the usual maxwell contribution , @xmath9 is an arbitrary constant governing the strength of the coupling between the dilaton and the maxwell field , @xmath10 is a potential of dilaton @xmath11 which is given by ref .
[ 15 ] @xmath12,\end{aligned}\ ] ] here @xmath13 is the cosmological constant .
one can verify that the potential reduces to the einstein cosmological constant when @xmath14 or @xmath15 .
compared to the action of the ordinary scalar fields , the phantom field has one negative kinetic term . in order to obtain a real action of the einstein - maxwell field in the presence of the phantom
, we can make substitutions in the action as follows ref .
[ 16 ] @xmath16 where @xmath17 is the imaginary unit .
thus we get the action @xmath18 and the potential for the phantom field @xmath19.\end{aligned}\ ] ] one can also verify that , when @xmath20 or @xmath21 the action reduces to the einstein - maxwell action and when @xmath22 the action reduces to the einstein - phantom action .
we consider the action in a simple model which contains a normal scalar field @xmath11 and a negative - kinetic scalar field @xmath23 , assuming that there is no direct coupling between the phantom field and the normal scalar field with such potentials , @xmath24 where @xmath25 represents the lagrangian density of matter fields . considering a flat universe which is described by the friedmann - robertson - walker metric
, the homogeneous fields @xmath11 and @xmath23 can be described by a fluid with an effective energy density @xmath26 and an effective pressure @xmath27 given by @xmath28 the corresponding equation of state ( eos ) parameter is given by @xmath29 then the equations of motion for the fields and the friedmann equation can be written as @xmath30 where @xmath31 is the density of fluid with a barotropic equation of state @xmath32 with @xmath33 a constant and @xmath34 ( @xmath35 for radiation and @xmath36 for dust mater ) . the equation ( [ as7 ] ) is the friedmann constraint equation .
in this section , we investigate the two - field dilaton model via the conventional phase space analysis . similar as in ref . [ 18 ] , we define the following new dimensionless variables @xmath37 ^ 2},\nonumber \\
x_\psi\equiv \frac{\kappa \dot{\psi } } { \sqrt{3}h } & , & \quad y_\psi \equiv \frac{\kappa \sqrt{v_2(\psi)}}{\sqrt{6}h } , \quad \lambda_\psi\equiv \frac{\sqrt{3}\frac{\partial _ 1(\psi)}{\partial \psi}}{\kappa v_2(\psi)},\quad \gamma_\psi\equiv \frac{v_1(\psi)\frac{\partial^2 v_1(\psi)}{\partial^2 \psi } } { [ \frac{\partial v_2(\psi)}{\partial \psi}]^2},\nonumber \\
z\equiv \frac{\kappa \sqrt{\rho_\gamma}}{\sqrt{6}h},\nonumber\end{aligned}\ ] ] the equations of motion ( [ as6])-([as7 ] ) can be rewritten as the following system of equations : @xmath38 where @xmath39 is the logarithm of the scale factor ( @xmath40 ) , and the fridemann constraint equation ( 11 ) becomes @xmath41 different from the case of a single exponential potential , the parameters @xmath42 and @xmath43 here are variables of @xmath11 and @xmath23 .
strictly speaking , the above system is not an autonomous system .
thus , if we want to discuss the phase plane , we need to find the constraints on the potential , or equivalently the conditions under which the potential may have the property we require in order that we can get some explicit results .
critical points correspond to fixed points where @xmath44 , @xmath45 , @xmath46 , @xmath47 , @xmath48 , @xmath49 , @xmath50 . observing these equations
, one can find that the physically meaningful critical points @xmath51 of the system are : ( note that we will restrict our discussion of the existence and stability of critical points in the expanding universes with @xmath52 ) .
@xmath53 could be fixed by @xmath54 ; ( ii ) .
@xmath55 ; ( iii ) . @xmath56 . in the case of ( i ) , when @xmath54 , then @xmath57 and the two - fields potentials are given by @xmath58 thus we can determine the important parameters of the dilaton field and the phantom field : @xmath59 in this case , the equations ( [ as1])-([as2 ] ) have one two - dimensional hyperbola ( type @xmath60 ) embedded in four - dimensional phase - space corresponding to kinetic - dominated solutions , with eos @xmath61 and fractional energy density @xmath62 ; a fixed point ( type @xmath63 ) corresponding to a dilaton - dominated solution , with @xmath64 and @xmath62 ; a fixed point ( type @xmath65 ) corresponding to a fluid - dilaton - dominated solution , with @xmath66 and @xmath67 ; and a fixed point ( type @xmath68 ) corresponding to a phantom - dominated solution , with @xmath69 and @xmath62 ( listed in table 1 ) .
.the properties of the critical points in a spatially flat frw universe containing a phantom field and a normal scalar field in the case of ( i ) . [
cols="^,^,^,^,^,^,^,^,^",options="header " , ] in order to study the stability of the critical points , using the friedmann constraint equation ( [ as3 ] ) we can reduce eqs.([as1])-([as2 ] ) to four independent equations .
substituting linear perturbations @xmath70 , @xmath71 , @xmath72 and @xmath73 into the four independent equations , we obtain the equation of perturbations to the first - order : @xmath74 the linear perturbations of system ( [ as4])-([as5 ] ) about each fixed point gives four eigenvalues .
the theory of stability requires that the real part of all eigenvalues should be negative .
so we have : type @xmath60 ( the kinetic - dominated solution ) : @xmath75 indicating that this solution is always unstable .
type @xmath63 ( the dilaton - dominated solution ) : @xmath76 indicating that this solution is also unstable .
type @xmath65 ( the fluid - dilaton - dominated solution ) : @xmath77 indicating that this solution is still unstable .
type @xmath68 ( the phantom - dominated solution ) : @xmath78 indicating that this solution is stable . in the case of ( ii ) @xmath79 , the equations ( [ as1])-([as2 ] )
have one fixed point type @xmath80 embedded in six - dimensional phase - space which corresponds to eigenvalues @xmath81 , @xmath82 and @xmath62 .
this indicates that the critical point is a de sitter attractor . in the case of ( iii )
@xmath83 , the equations ( [ as1])-([as2 ] ) have one fixed point type @xmath84 embedded in six - dimensional phase - space which corresponds to eigenvalues @xmath85 , @xmath0 becomes meaningless and @xmath86 .
this indicates that the critical point is not a dynamical attractor .
our numerical studies indicate that the eos parameter @xmath0 changes from @xmath87 to @xmath88 as shown in figure 1 .
we have assumed that there is no direct coupling between the phantom field and the normal scalar field in this paper . without the loss of generality ,
the initial conditions @xmath89 and @xmath90 can be fixed in order to get the eos today @xmath91 [ 19 ] , and the energy density of dark energy today @xmath3 .
the blue line represents the eos of the two - field dilaton model , the black line represents the single dilaton model and the red line represents the single phantom model .
of the two - field dilaton model of dark energy .
the blue line represents the eos of the two - field model , the black line represents the single dilaton model and the red line represents the single phantom model . ]
in summary , we have investigated the possibility of constructing a two - field dark energy model which has the equation of state @xmath0 crossing @xmath6 by using the dilaton and phantom fields . we have made a phase - space analysis of the evolution for a spatially flat frw universe filled with a barotropic fluid and phantom - dilaton fields .
it is shown that there exists the stable late - time attractor solution in the model .
also , we showed that the equation of state @xmath0 can cross @xmath6 naturally .
so the two - field dilaton field is a viable candidate for dark energy .
it is apparent that our model is also plagued with the instability problem at the quantum level which makes its existence doubtful .
in fact , this is a common problem for nearly all phantom models .
however , as argued by carroll et al .
[ 20 ] , these models might be phenomenologically viable if considered as effective field theories valid only up to a certain momentum cutoff . according to their discussions ,
the instability timescale of the phantom quanta can be greater than the age of the universe provided that the cutoff is at or below 100 mev . in this sense ,
the phantom quanta are stable against decay into gravitons and other particles .
therefore , considering astronomical observations favoring the phantom model for dark energy , it remains open if the phantom matter exists and acts as dark energy .
we are grateful to zhang xin - min , wei hao , feng bo for kind help and discussions .
we also thank zhao gong - bo for plotting the fig .
1 . this project was in part supported by the ministry of education of china , directional research project of the chinese academy of sciences under project kjcx2-yw - t03 , and by the national natural science foundation of china under grants 10521001 , 10733010 , and 10725313 , and by 973 program of china under grant 2009cb824800 .
riess a g et al .
1998 _ astron
. j. _ * 116 * 1009 perlmutter s et al .
1999 _ astrophys .
j. _ * 517 * 565 de bernardis p et al .
2000 _ nature _ * 404 * 955 hanany s et al . 2000 , _ astrophys . j. _ * 545 * l5 spergel d n et al .
2003 _ astrophys .
j. s _ * 148 * 175 tegmark m. et al .
2004 _ phys .
rev . _ * d 69 * 103501 krauss l m and turner m s 1995 _ gen .
* 27 * 1137 ; ostriker j p and steinhardt p j 1995 _ nature _ * 377 * 600 ; liddle a r , lyth d h , viana p t and white m 1996 _ mon . not .
* 282 * 281 piazza f and tsujikawa s 2004 _ j. cosmol .
phys . _ * 0407 * 004 ; copeland e j , sami m and tsujikawa s 2006 _ int
* d15 * 1753 - 1936 ( hep - th/0603057 ) ; bludman s a and roos m 2002 _ phys .
rev . _ * d65
* 043503 ; gonzalez - diaz p f 2000 _ phys . lett . _
* b 481 * 353 ; rubano c and scudellaro p 2002 _ gen .
_ * 34 * , 307 ; sen s and seshadri t r 2003 _ int . j. mod . phys . _ * d 12 * , 445 ; steinhardt p j 2003 _ phil . trans .
soc . lond .
_ * a 361 * 2497 ; urena - lopez l a and matos t. 2000 _ phys .
* d 62 * , 081302 ; zhang j f , zhang x and liu h y 2008 _ eur .
j. _ * c 54 * 253 - 258 ; zhang x 2007 _ phys . lett . _ * b 648 * 1 - 7 ; zhang x 2005 _ mod .
lett . _ * a 20 * 2575 ; ma y z and zhang x 2008 _ phys
* b 661 * 239 - 245 boisseau b , esposito - farese g , polarski d and starobinsky a a 2000 _ phys .
lett . _ * 85 * 2236 ; caldwell r r 2002 _ phys .
* b 545 * 23 ; carroll s m , hoffman m and trodden m 2003 _ phys .
_ * d 68 * 023509 ; cline j m , jeon s y and moore g d 2004 _ phys . rev .
_ * d 70 * 043543 ; dabrowskim p , stachowiak t and szydlowski m 2003 _ phys . rev .
_ * d 68 * 103519 ; sun z y and shen y g 2005 _ gen .
grav . _ * 37 * 243 ; gonzalez - diaz p f 2003 _ phys .
rev . _ * d 68 * 021303 ; guo z k et al . 2005 _ phys .
* b 608 * 177 ; hao j g and li x z 2003 _ phys . rev . _ * d 68 * 043501 ; nojiri s and odintsov s d 2003 _ phys .
* b 562 * 147 ; cai r g and wang a z 2005 _ j. cosmol .
phys . _ * 0503 * 002 ; singh p , sami m and dadhich n 2003 _ phys .
_ * d 68 * 023522 armendariz - picon c , mukhanov v and steinhardt p j 2001 _ phys .
* d 63 * 103510 ; chiba t 2002 _ phys .
* d 66 * 063514 ; chimento l p and feinstein a 2004 _ mod .
lett . _ * a 19 * 761 ; malquarti m , copeland e j , liddle a r and trodden m 2003 _ phys .
* d 67 * 123503 ; scherrer r j 2004 _ phys .
lett . _ * 93 * 011301 aguirregabiria j
m and lazkoz r 2004 _ phys .
rev . _ * d 69 * 123502 ; bagla
j s , jassal h k and padmanabhan t 2003 _ phys .
_ * d 67 * 063504 ; choudhury d , ghoshal d , jatkar d p and panda s 2002 _ phys .
* b 544 * 231 ; frolov a v , kofman l and starobinsky a a 2002 _ phys .
_ * b 545 * 8 ; gibbons g w 2003 _ class
grav . _ * 20 * s321 ; gorini v , kamenshchik a y , moschella u and pasquier v 2004 _ phys . rev . _
* d 69 * 123512 ; kim c , kim h b and kim y 2003 _ phys .
lett . _ * b 552 * 111 ; padmanabhan t 2002 _ phys . rev . _
* d 66 * 021301 ; sen a 2005 _ phys .
scripta _ * t 117 * 70 - 75 ( hep - th/0312153 ) ; shiu g and wasserman i 2002 _ phys .
* b 541 * , 6 ; zhang j f , zhang x and liu h y 2007 _ phys .
* b 651 * 84 - 88 wei h and cai r g 2005 _ class .
* 22 * , 3189 ; wei h and cai r g 2005 _ phys .
_ * d 72 * 123507 gao c j and zhang s n 2004 _ phys .
lett . _ * b 605 * 185 ; elvang h , freedman d z and liu h 2007 hep - th/0703201 gao c j and zhang s n 2006 hep - th/0604114 gao c j and zhang s n 2006 astro - ph/0605682 copeland e j , liddle a r and wands d 1998 _ phys .
* d 57 * , 4686 ; hao j g and li x z 2004 _ phys .
* d 70 * 043529 riess a g et al .
2004 _ astron .
j. _ * 607 * 665 carroll s m et al .
2003 _ phys .
_ * d 68 * 023509 | we investigate the cosmological evolution of a two - field model of dark energy where one is a dilaton field with canonical kinetic energy and the other is a phantom field with a negative kinetic energy term .
a phase - plane analysis shows that the phantom - dominated scaling solution is the stable late - time attractor of this type of models .
we find that during the evolution of the universe , the equation of state @xmath0 changes from @xmath1 to @xmath2 , which is consistent with the recent observations . |
selfgravitating system are ubiquous in the universe .
selfgravity governs the formation and evolution of the large scale structure in the universe as well as the dynamics of galaxy formation , stars and interstellar medium @xcite .
it is therefore of great interest to study self - gravitating fluids in thermal equilibrium although most astrophysical and cosmological systems are not in exact thermal equilibrium @xcite-@xcite .
the self - gravitating gas has peculiar properties from the point of view of equilibrium statistical mechanics : the most important configurations are inhomogeneous and the thermodynamic functions * exist * in the * dilute * limit @xcite @xmath2 where @xmath0 is the number of particles and @xmath3 stands for the volume of the box containing them . in such a limit , the internal energy , the free energy and the entropy turns to be extensive .
that is , they take the form of @xmath4 times a function of intensive dimensionless variables . in the case of a spherical box of radius @xmath5
( so that @xmath6 ) there is only one such variable , that is the ratio of the characteristic gravitational energy @xmath7 ( @xmath8 is the particle mass and @xmath9 is newton s constant ) and the kinetic energy of the order of the temperature @xmath10 of a particle in the gas @xcite : @xmath11 for more complex geometries there will be , besides @xmath12 ( in such cases @xmath5 is to be identified with the linear scale of the box ) , other dimensionless shape parameters . in this paper
we restrict the analysis to the simplest case of a spherical box and follow the approach of refs.@xcite where the starting point is the partition function for non - relativistic particles interacting through their gravitational attraction in thermal equilibrium .
the configurational part of this partition function is expressed as a functional integral over a scalar field @xmath13 [ proportional to the the gravitational potential ] with an euclidean action ( that is minus the logarithm of statistical weight ) proportional to @xmath0 .
therefore , the saddle point technique ( or mean field method ) permits to compute the configuration partition function in the large @xmath4 limit . as investigated in refs.@xcite the self - gravitating gas in the canonical ensemble
is in a gaseous phase for @xmath14 where @xmath15 .
( further studies on self - gravitating gases are reported in ref .
@xcite ) . at this point
the isothermal compressibility diverges and the speed of sound inside the sphere becomes imaginary .
this clearly shows that the gas phase collapses at this point in the canonical ensemble as a consequence of the jeans instability .
monte carlo simulations explicitly show that the gas collapses for @xmath16 into a very dense and compact phase@xcite .
it must be recalled that the small fluctuations around the mean field solution in the canonical ensemble are stable for @xmath17@xcite .
namely , fluctuations in the partition function do not feel the divergence of the isothermal compressibility neither the fact that the speed of sound becomes imaginary .
even in the regime where the gaseous phase is locally stable there are always collapsed configurations which have much lower energy than the gas .
in fact , the energy has its absolute minimum in the collapsed phase where all particles fall to one point and this configuration completely dominates the free energy , causing its divergence to minus infinity .
of course , interactions other than gravitational always dominate for short distances .
therefore , the properties the collapsed phase depend on the short distance physics .
this physics is different according to the nature of the ` particles ' considered . they can be from galaxies till molecules or atoms .
still , we can study the regime when collapse happens just introducing a short distance cutoff ( ultraviolet regulator ) to the gravitational interaction .
we present a complete description of the self - gravitating gas , including the collapsed phase , using a newtonian potential regularized at short distances .
we choose the newton potential to be @xmath18 for interparticle distances @xmath19 instead of newton s law @xmath20 .
the cutoff @xmath21 is chosen very small compared with the size of the box .
we give an exact mapping of the @xmath0-body regularized problem into a field theory functional integral .
this functional integral is dominated by its saddle points in the @xmath1 , namely by the mean field equations .
the mean field equations coincide for zero cutoff with the lane - emden equation obtained from hydrostatics and assuming an ideal gas equation of state . in this way
we reproduce the well - known isothermal sphere solutions describing a self - gravitating gas in local thermal equilibrium .
these diluted solutions get small corrections for small cutoff .
we find a host of new solutions ( saddle points ) from our regularized mean field equations .
these new solutions are singular in the limit of zero cutoff and can not be find from the standard ( no cutoff ) lane - emden equation . in these new solutions , which are perfectly regular for non - zero cutoff , a large number of particles concentrate near the origin .
there are two new types of solutions .
the first type are the collapsed saddle point solutions : all the particles are densely concentrated in a region of size of the order of the cutoff .
these solutions are studied analytically and provide the absolute minima of the free energy , which reads , at dominant order for small cutoff @xmath22 : @xmath23 here , @xmath10 is the temperature and @xmath24 is the free energy of an ideal gas .
the first term is just the potential energy of @xmath4 particles clustered in a small sphere of radius @xmath25 , where the regularized gravitational interaction is the same for all @xmath26 particle pairs .
the second term is @xmath10 times the entropy loss in the collapse .
this free energy is large and negative , unbounded from below as @xmath27 . in particular , this free energy is well below the free energy of the dilute solution .
the second type of saddle point solutions interpolate between the gas phase and the collapsed phase , in a sort of pre collapse .
this solution has a finite action in the zero cutoff limit and describes the tunneling transition where a small fraction of particles coalesce into a region of the size of the short distance regulator so that the density near the origin is very large .
the tunneling probability to collapse for a dilute solution turns out to be extremely small for large @xmath0 and small @xmath28 .
this means in practice that dilute solutions which are locally stable can be considered stable . as mentioned above this
happens in the canonical ensemble for @xmath14 where @xmath15 .
the parameter @xmath29 for the collapsed solutions can be arbitrarily large .
that is , the particle density does not need to be dilute @xmath30 [ see eq.([limter ] ) ] but we can have @xmath31 .
we obtain in sec .
[ metab ] an estimate for the lifetime for the dilute phase of the self - gravitating gas : @xmath32 \ ; .\ ] ] one can see that the lifetime becomes infinitely long in the zero cutoff limit as well as when @xmath33 at fixed cutoff [ recall that @xmath34 in the dilute limit of eq .
( [ limter ] ) ] . in summary , we provide through mean field theory a complete statistical description of the self - gravitating gas including the absolute minimum of the free energy .
at short distances , the particle interaction for the self - gravitating gas in physical situations is not gravitational .
its exact nature depends on the problem under consideration ( opacity limit , van der waals forces for molecules etc . ) .
we shall just assume a null short distance pair force ; that is , we consider the hamiltonian @xmath35 where @xmath36 and @xmath22 is the short distance cut - off .
it must be stressed that the results presented in this work using the soft - core cut - off eq.([defva ] ) will be qualitatively the same for other types of cut - off since @xmath37 .
the partition function in the canonical ensemble reads @xmath38 where @xmath39 is the temperature .
computing the integrals over the momenta @xmath40 , @xmath41 = \left(\frac{m}{2\pi \beta}\right)^{3/2 } \ ; , \ ] ] yields @xmath42 where @xmath43\ ] ] is the partition function of the ideal gas .
thus , @xmath44 is the full canonical free energy . we can now replace the integration over the particle coordinates with a functional integration over the configurations of the gravitational field @xmath45 produced by the particles .
we start from the fundamental property of gaussian functional integrals , namely @xmath46 + \int d^3x\ , j({{{\boldsymbol{x}}}})\,\varphi({{{\boldsymbol{x}}}})\right\ } = \exp\left[\frac12\int d^3x \int d^3x'\,j({{{\boldsymbol{x}}}})\ ; { \cal g}({{{\boldsymbol{x}}}},{{{\boldsymbol{x}}}}')\ ; j({{{\boldsymbol{x}}}}')\right ] \ ; , \ ] ] where @xmath47 $ ] is the quadratic action functional @xmath48 = \frac12\int d^3x\int d^3x'\ ; \varphi({{{\boldsymbol{x}}}})\ , { \cal g}^{-1}({{{\boldsymbol{x}}}},{{{\boldsymbol{x}}}}')\ , \varphi({{{\boldsymbol{x}}}}')\ ] ] and the integration measure is assumed to be normalized such that @xmath49\}=1 $ ]
. then we set @xmath50 and we identify the source field @xmath51 as @xmath52 times the microscopic particle density @xmath53 inserting eqs .
( [ potg ] ) and ( [ densu ] ) into eq .
( [ gauso ] ) yields , @xmath54 + \beta \ ; m\sum_{j=1}^n \varphi({{{\boldsymbol{q}}}}_j)\bigg\ } = \exp\bigg[\frac12\sum_{j , k=1}^n \frac{g \ , m^2 \ , \beta}{|{{{\boldsymbol{q}}}}_j-{{{\boldsymbol{q}}}}_k|_a}\bigg ] = e^{\frac{gm^2\beta\,n}{2a } } \!\!\prod_{1\leq j < k\leq n } \exp\left(\frac{g \ ; m^2 \ ; \beta}{|{{{\boldsymbol{q}}}}_j-{{{\boldsymbol{q}}}}_k|_a}\right ) \ ; .\ ] ] this provides a functional integral representation of the product over particle pairs in the integrand of eq .
( [ fpconf ] ) . inserting the identity eq .
( [ idu ] ) in eq .
( [ fpconf ] ) and exchanging the functional integration with the integration over the coordinates @xmath55 we obtain @xmath56\right\ } \ ; \frac1{v^n } \int_{v^n } d^{3n}\!q \ ; \exp\big[\beta \ , m\sum_{j=1}^{n } \varphi({{{{\boldsymbol{q}}}}}_j ) \big ] \ ; .\ ] ] where @xmath48 = -\frac{\beta}{8 \ , \pi \ , g } \int d^3x \ ; \varphi \ , \nabla^2_a\varphi\ ] ] and @xmath57 satisfies @xmath58 since the @xmath59 are dummy variables , we have @xmath60= \left [ \int_{v } d^3q \ ; e^{\beta \ , m \ , \varphi({{{{\boldsymbol{q } } } } } ) } \right]^n\ ] ] and the configuration partition function becomes @xmath61 } \ ; , \ ] ] where @xmath62 = s[\varphi ] + s_{\rm int}[\varphi ] \quad,\qquad s_{\rm int}[\varphi ] \equiv - n\log\big[\frac1{v } \int_v d^3x \ ; e^{-\beta \ , m \ , \varphi({{{\boldsymbol{x}}}})}\big]\ ] ] we want to stress that this field
theoretic functional formulation of the self - gravitating gas is fully equivalent , for any @xmath4 , to the original one in terms of particles .
the functional integral representation of the partition function presented in ref.@xcite for the canonical ensemble is slightly different but equivalent to eqs .
( [ rizc ] ) and ( [ accef ] ) for large @xmath4 .
let us also observe that the nonlocal operator @xmath63 reduces to the standard laplacian operator in the limit @xmath64 .
however , this limit is not so straightforward in the functional integral above , since the interaction term @xmath65 in the action may ( and indeed does ) become unbounded from below in such limit .
we now pass to dimensionless variables by setting @xmath66 and @xmath67 the action for @xmath68 now reads @xmath69 = n \,s[\phi]\quad,\qquad s[\phi ] = -\frac1{8\pi\,\eta}\int d^3r\ ; \phi \ ; \nabla^2_a \phi - \log\left [ \int_{|{{{\boldsymbol{r}}}}|\le 1 } d^3r \ ; e^{\phi}\right ] + \log\frac{4 \ , \pi}3 \ ; , \ ] ] where @xmath70 we recall that the variable @xmath29 is the ratio of the characteristic gravitational energy @xmath71 and the kinetic energy @xmath72 of a particle in the gas . for @xmath73
the ideal gas is recovered .
having factored out of the action the number of particles @xmath0 , it is natural to use the saddle points method to evaluate the functional integral over @xmath74 in the limit @xmath75 at fixed @xmath12 , that is in the dilute limit . strictly speaking
, however , the action per particle @xmath76 $ ] still depends on @xmath0 through @xmath77 .
thus we should better regard the saddle point method as a mean field type approximation yielding the free energy @xmath78 to leading order in @xmath0 , as we shall now show . the stationarity condition @xmath79}{\delta \phi({{{\boldsymbol{r } } } } ) } = 0 \ ; .\ ] ] leads to the regularized self
consistent boltzmann
poisson equation @xmath80 in the limit @xmath81 this becomes the standard self consistent boltzmann
poisson ( or lane
emden ) equation derivable from hydrostatic plus the assumption of a local , ideal gas equation of state @xcite .
we may rewrite eq .
( [ eq;bple ] ) in integral form as @xmath82 where @xmath83 is the normalized particle density associated to @xmath84 .
we may also use the identity involving the standard laplacian @xmath85 , @xmath86 to cast the field equation in the integro differential form @xmath87 notice that the laplacian in the origin is always regular for nonzero cutoff @xmath88 implying , together with the finiteness of @xmath89 , that @xmath90 in the origin .
we can recast the action of the saddle point eq.([accefa ] ) using the regularized equation of motion ( [ eq;bple ] ) as follows , @xmath91 = \frac1{2 \ ; q } \ ; \int_{|{{{\boldsymbol{r}}}}|\le 1 } d^3r \ ; \phi({{{\boldsymbol{r } } } } ) \ ; e^{\phi({{{\boldsymbol{r } } } } ) } - \log\frac{3 \ ; q}{4 \ , \pi } \quad .\ ] ] the free energy for large @xmath4 can be written in terms of the stationary action as@xcite @xmath92 where @xmath93 $ ] , @xmath94 is a solution of eq .
( [ eq;bple ] ) , and @xmath24 is the free energy for the ideal gas . notice that , since @xmath95 vanishes as @xmath96 as @xmath33 _ at fixed _
@xmath21 _ and _ @xmath29 , we can regard @xmath97 as extensive in the particle number , that is linear in @xmath4 , only if it is regular as @xmath98 .
this holds true for stationary point which are regular at the @xmath99 limit .
we show below that there are saddle points which become singular in the @xmath98 limit and consequently the free energy is not proportional to @xmath4 .
similarly , the subleading terms in eq .
( [ fpv ] ) are order one only for saddle points which are regular for @xmath99 .
what is always true is that they are indeed subleading for any given saddle point .
in the case of spherically symmetry the integration over angles in eq .
( [ eq : reg0 ] ) can be performed explicitly , yielding the one dimensional non linear integro differential equation @xmath100 = -\frac{2 \ , \pi \
, \eta}{q \ , a } \int_{|r - a|}^{r+a } dr'\ ; r ' \ , e^{\phi(r ' ) } \ ; , \ ] ] where now @xmath101 eq . [ ecdar ] is to be supplemented with the boundary conditions of smooth joining ( continuity of @xmath102 and @xmath103 at @xmath104 ) with the external monopole solution @xmath105 .
integration over the angles in eq.([ecinta ] ) yields the non linear integral equation @xmath106 e^{\phi(r')}\ ; \theta(1-r')\right\ } \ ; , \end{split}\ ] ] which by itself determines @xmath107 for all values of @xmath108 . in particular eq .
( [ ecinar ] ) implies that any solution is finite in the origin , since @xmath109 \ ; .\ ] ] moreover , eq .
( [ ecdar ] ) or eq .
( [ ecinar ] ) imply @xmath110 so that the density is monotonically decreasing away from the origin .
the simplest way to prove that @xmath111 is to multiply both sides of eq .
( [ ecdar ] ) by @xmath112 and integrate from 0 to an arbitrary value .
this yields @xmath113 which has a negative definite r.h.s . which vanishes when @xmath114 as long as @xmath115 .
the action for spherically symmetric solution becomes from eq.([acci ] ) , @xmath116 = \frac{2 \ , \pi}{q } \ ; \int_0 ^ 1 r^2 \ ; dr \ ; \phi(r ) \ ; e^{\phi(r ) } - \log\frac{3 \ ; q}{4 \ , \pi } \quad .\ ] ]
the saddle point equation ( [ eq;bple ] ) admits solutions which are regular as @xmath118 and reproduce in this limit the solutions of the standard lane emden equation .
we call them dilute solutions .
these are known since longtime in the spherically symmetric case @xcite . to study such solutions we can set the cutoff @xmath119 to zero from the start .
equivalently , one can take the limit @xmath81 in eq .
( [ ecdar ] ) with the assumption that @xmath102 stays finite in the limit for any @xmath112 , including the origin @xmath120 , and that @xmath121 . assuming spherical symmetry we get @xmath122 for @xmath123 ,
while for @xmath124 we have @xmath125 with the boundary conditions @xmath126 clearly there exists one and only one solution for a given @xmath127 , so the classification of all solutions is equivalent to the determination of all the allowed values of @xmath128 .
we set as usual@xcite @xmath129 where @xmath130 must satisfy , upon inserting eq .
( [ eq : phichi ] ) in eq .
( [ eqrad ] ) , @xmath131 and , from the boundary conditions eq .
( [ eq : bc ] ) at @xmath132 , @xmath133 these two relations fix the two parameters @xmath127 and @xmath134 as functions of @xmath12 .
in particular , we can rewrite @xmath102 as @xmath135 the density profile of eq .
( [ ro ] ) reads in terms of @xmath136 , @xmath137 so that the normalization factor can also be written @xmath138 . due to scale
invariant behaviour of the gravitational interaction , eq .
( [ eq : fund ] ) enjoys the following scale covariance property : if @xmath130 is a solution , then also @xmath139 is a solution . using @xmath140 rather than @xmath130
is compensated by the shift @xmath141 .
notice that @xmath107 is indeed defined as a scale transformation of @xmath130 , with the scale parameter @xmath142 ( not uniquely ) fixed by @xmath143 . as a consequence of this scale invariance
all physical quantities must be invariant under the simultaneous replacements @xmath144 and @xmath145 .
let us observe that , if the solutions of eq.([eqrad ] ) or ( [ eq : fund ] ) did not fulfill @xmath121 or @xmath146 , a delta function at the origin would appear in the right hand side and @xmath147 , with @xmath148 , when @xmath149 .
this is the only possible singular behaviour and corresponds to a point particle with negative mass @xmath150 in the origin
. we shall not consider such unphysical solutions , which are ruled out when eq.([eqrad ] ) is considered as the @xmath81 limit of eq .
( [ ecdar ] ) , and stick to @xmath151 .
moreover , scale invariance allows to set also @xmath152 .
this choice completely fixes the scale and we have no more scale invariance left .
one has from the second eq .
( [ eq : bc2 ] ) that @xmath153 for all positive @xmath154 , so that the density profile is monotonically decreasing with the distance . since @xmath155 is monotonically decreasing , the allowed values of @xmath128 [ that is of @xmath156 , see eqs.([ro ] ) and ( [ eq : bc ] ) ] are in one to one correspondence with to the roots of the relation @xmath157 which inverts the second of eqs .
( [ eq : bc2 ] ) .
thus there is only one dilute solution for each value of @xmath142 .
however , @xmath157 is a multiple valued function for a certain range of @xmath12 .
we plot @xmath158 vs. @xmath159 in fig .
[ fig : eta0 ] .
one sees that there is a unique @xmath160 for a given @xmath158 only for @xmath161 , that is @xmath162 . for @xmath163 ( @xmath164 is the absolute maximum , located at @xmath165 , of @xmath166 over @xmath167 ) the relation @xmath157 is indeed multivalued and as @xmath29 approches the value @xmath168 there are increasingly more solutions which accumulate near the purely logarithmic solution as a function of @xmath169 according to eq.([eq : bc2 ] ) for the dilute solutions .
notice the maximum of @xmath29 at @xmath170 , where @xmath171 diverges .
the region beyond the maximum location , at @xmath172 , is locally unstable as discussed in ref.@xcite . @xmath173 and @xmath174 diverge before the maximum , at the point @xmath175 . ]
@xmath176 = 2 - 2\,\log r \ ; .\ ] ] which correspond to the singular density profile @xmath177 . in other words ,
when @xmath12 is exactly 2 there exist an infinite set of solutions corresponding to the increasing sequence of values of @xmath134 which satisfy @xmath178 . in fig .
[ mrho ] we plot in log log scale the density profile @xmath179 [ see eq .
( [ eq : dprof ] ) ] vs. @xmath112 for few terms of this sequence . .
only the smallest one at @xmath180 is locally stable . ]
we recall that , for any given @xmath181 , only the unique dilute solution with @xmath182 is locally stable .
that is , any small fluctuation increases the action . at @xmath183
a zero mode appears in the linear fluctuation spectrum and @xmath184 , the specific heat at constant volume , diverges . for larger values of @xmath142 negative modes
show up , signalling local instability .
actually the locally stable dilute solutions describe a ( locally ) stable dilute phase , in the thermodnamic and mechanical terms , only for @xmath185 . at @xmath186
the isothermal compressibility diverges and the speed of sound inside the sphere becomes imaginary .
the gas phase collapses at this point in the canonical ensemble as a consequence of the jeans instability .
this is confirmed by monte carlo simulations@xcite .
hence , in the case @xmath187 depicted in fig .
[ mrho ] , only the solution with the smallest density contrast , corresponding to the smallest value @xmath188 , is stable .
the unstable solutions that exist for @xmath189 should imply that the dilute phase is metastable for this range of @xmath29 , however , their lifetime is so huge [ see eq.([vida2 ] ) ] that these dilute solutions are in practice stable . in the stable dilute phase the particles
are moderately clustered around the origin with a density that monotonically decreases with @xmath108 .
one sees that the density contrast between the center and the boundary @xmath190 grows with @xmath142 since @xmath191 decreases with @xmath142 [ see eq.([eq : bc2 ] ) ] and fig .
[ mrho ] @xcite .
the action per particle of a dilute solution can be written entirely in terms of the density at the border@xcite @xmath192 - 4\pi\rho(1)= 3-\eta + \log\frac{\lambda^2}{3 \ , \eta}+ \chi(\lambda)- \frac{\lambda^2}{\eta } \ ,
e^{\chi(\lambda ) } \ ; , \ ] ] and is multivalued for @xmath193 .
the solution with the smallest density contrast has the smallest action , in accordance with the linear stability analysis .
we plot @xmath194 as a function of @xmath195 and @xmath12 in fig .
[ fig : action0 ] . vs. @xmath196 ( above ) and vs. @xmath29 ( below ) for the dilute solution .
@xmath194 is multivalued for @xmath197 , with a peculiar self similar structure , as evident from the inset . ]
all physical quantities can be expressed for large @xmath4 in terms of the function @xmath198 we have for the energy , the pressure , the isothermal compressibility @xmath199 , the specific heats ( @xmath171 and @xmath200 ) and the speed of sound at the boundary@xcite @xmath201 - s(\eta ) + { \cal o}\left(\frac1{n}\right ) \quad , \quad \kappa_t = \frac1{f(\eta)+\frac13 \ ; \eta \ ; f'(\eta ) } + { \cal o}\left(\frac1{n}\right ) \ ; , \\
\cr & & c_v = 3 \left [ f(\eta)-\eta \ ; f'(\eta ) -\frac12 \right ] + { \cal o}\left(\frac1{n}\right ) \quad , \quad c_p = c_v + \frac { \left[f(\eta)-\eta \ ; f'(\eta)\right]^2 } { f(\eta)+\frac13 \eta \ ; f'(\eta ) } + { \cal o}\left(\frac1{n}\right ) \ ; , \cr \cr & & \frac{v_s^2}{t } = \frac { \left[f(\eta)-\eta \ ; f'(\eta)\right]^2 } { 3 \left[f(\eta)-\eta \ ; f'(\eta)-\frac12 \right ] } + f(\eta)+\frac13 \ ; \eta \ ; f'(\eta ) + { \cal o}\left(\frac1{n}\right)\ ; . \nonumber\end{aligned}\ ] ] here @xmath202 stands for the entropy of the ideal gas . notice that @xmath199 and @xmath200 , the specific heat at constant pressure , both diverge before @xmath203 , at @xmath204@xcite . moreover , the speed of sound squared , which stays regular at the boundary when @xmath205 , has a simple pole there when evaluated inside the isothermal sphere@xcite .
thus , when @xmath206 the speed of sound is purely imaginary inside the sphere implying that small fluctuations grow exponentially with time .
we have here discussed the dilute solutions in the zero cutoff limit ( @xmath207 ) . including the short distance cutoff only adds small @xmath208 corrections to the dilute solutions .
we present in this section new solutions for which the mean field @xmath107 develops singularities in the limit @xmath27 . in case of spherical symmetry , when eq.([ecdar ] )
is the saddle point equation , the only possible singularity is localized at @xmath209 .
it is convenient then to ` blow up ' the region near @xmath209 by setting [ compare with eq .
( [ eq : phichi ] ) ] @xmath210 where we introduce the variable : @xmath211 we thus obtain for @xmath212 , the following rewriting of the integral equation ( [ ecinar ] ) @xmath213 \\
+ & \frac{x+1}{2x } \big[i_2(m(x ) ) -i_2(|x-1|)\big ] - \frac1{4x } \big[i_3(m(x ) ) - i_3(|x-1|)\big ] - i_2(1 ) + i_1(1)\ ; , \end{split}\ ] ] where , @xmath214 \ ; , \ ] ] and we have used eq . ( [ ecinar0 ] ) to write @xmath89 in terms of @xmath215 as @xmath216 \ ; .\ ] ] notice that all dependence on @xmath142 appears now through the variable @xmath217 .
by construction we have @xmath218 [ see eq .
( [ eq : phip0 ] ) ] , @xmath219 and @xmath220 for all @xmath221 .
notice that @xmath29 does not enter the integral equation above , just as it did not enter the differential equation ( [ eq : fund ] ) for @xmath222 in the setup without short distance regulator .
owing to the normalization condition eq .
( [ qr ] ) , @xmath29 is * computed * once a solution is known , as a function of @xmath223 ( and @xmath95 ) , through @xmath224 \;.\ ] ] > from eq .
( [ ecdar ] ) we may also derive the integro differential form of the equation satisfied by @xmath225 , that is @xmath226 = -\frac12 \big[i_1(x+1 ) -i_1(|x-1|)\big]\;.\ ] ] eq .
( [ eq : varinteq ] ) or ( [ eq : vardiff ] ) can be solved numerically to high accuracy for any given @xmath223 .
we plot the corresponding @xmath12 as a function of @xmath195 in fig .
[ fig : eta1 ] , for better comparison with its behaviour in the dilute case , fig .
[ fig : eta0 ] . * @xmath229 .
that is , @xmath230 . in this regime
we reproduce the curve @xmath231 ( fig . [
fig : eta0 ] ) characteristic of the diluted phase of sec .
iv plus small corrections of order @xmath95 . *
@xmath232 . here
@xmath233 and the damped oscillation pattern around @xmath187 , characteristic of the dilute solutions , gets disrupted : the oscillations regain larger and larger amplitude until @xmath234 . * @xmath235 . here
, a sudden drop to very small @xmath236 takes place [ see fig . [
fig : eta1 ] ] ; then @xmath29 rises again as @xmath237 , for @xmath237 large enough , and keep growing indefinitely ( see fig .
[ fig : eta1 ] and below ) .
this regime corresponds to saddles interpolating between the dilute phase and a collapsed phase ( see below ) .
* @xmath238 .
this regime describes collapsed configurations where all particles are concentrated in a region of size @xmath239 and provide the absolute minimum of the free energy .
notice that new maxima appear in addition to those present for @xmath99 ( compare with fig .
[ fig : eta0 ] ) . those new maxima turn to be degenerate with the old ones up to small corrections of order @xmath95 .
in addition , @xmath29 exhibits oscillations for @xmath232 which are similar to those present at @xmath240 and @xmath241 but reversed and faster
. we shall give below a qualitative explanation for this peculiar behavior .
the new dependence of @xmath12 on @xmath134 described above implies that the number of solutions ( that is the number of distinct @xmath142 for any given @xmath242 ) is finite for any @xmath29 and that , most importantly , there appear new solutions . in particular
, there are two new solutions for @xmath243 in addition to the unique solution regular as @xmath244 and there is one new solution for @xmath245 , the region not accessible to dilute solutions .
these new solutions are singular in the limit @xmath98 , as evident from fig .
[ fig : chifit ] , where we plot @xmath246 vs. @xmath108 for few values of @xmath95 at a fixed value of @xmath12 and @xmath160 placed in the sudden drop of @xmath29 vs. @xmath196 ( the black circles in fig .
[ fig : eta1 ] , scaling as @xmath247 ) . for @xmath248 ( interpolating solutions ) , with comparison between the regularized and unregularized cases .
the dashed black curve is the dilute solution ( that is for @xmath99 ) at the same value of @xmath29 , namely at the location of the magenta circle of fig .
[ fig : eta1 ] .
the other curves correspond to the black circles in fig .
[ fig : eta1 ] . ] in fig .
[ fig : chifit ] we also compare the solutions of the integral equation ( [ ecinar ] ) or ( [ eq : varinteq ] ) to those of the differential equation ( [ eqrad ] ) , with boundary conditions ( [ eq : bc ] ) , in which @xmath127 takes the value obtained by the integral method , upon using eqs .
( [ eq : phixi ] ) and ( [ eq : phi0 ] ) .
we see that the agreement is very good down to @xmath108 of order @xmath95 , when the short distance regularization becomes effective , while the solutions of the differential equation ( [ eqrad ] ) eventually blow to @xmath249 as @xmath250 .
this agreement appears very natural from the integro
differential formulation of eq .
( [ ecdar ] ) , where the short distance regulator @xmath95 can play a significant role only for @xmath251 .
hence , the qualitative behaviour of the singular solutions can be understood already from the differential equation ( [ eq : fund ] ) , provided one drops the initial conditions @xmath254 characteristic of the dilute solutions .
that is , using the relation @xmath255 that follows from eqs .
( [ eq : phichi ] ) and ( [ eq : phixi ] ) , we can solve the integral equation ( [ eq : varinteq ] ) just in the interval @xmath256 , with @xmath257 , to compute @xmath258 at the boundary @xmath259 .
these values for @xmath260 and @xmath261 , both negative by construction , can then be used as initial conditions at @xmath262 to solve the differential equation ( [ eq : fund ] ) in place of the conditions @xmath254 , characteristic of the dilute solutions .
the choice @xmath257 is motivated by a direct numerical analysis but can be explained by the fact that @xmath223 diverges when @xmath118 in the solutions which are singular as @xmath81 .
this implies , on one side , that the r.h.s .
( [ eq : vardiff ] ) is exponentially small in @xmath263 when @xmath264 [ see also eq .
( [ mx ] ) ] and , on the other side , that the initial conditions at @xmath265 affect the function @xmath130 very strongly , causing the rapid drop for @xmath266 observed in fig .
[ fig : chifit ] , as well as the growth followed by the fall to @xmath267 as @xmath268 .
it must be observed , in fact , that @xmath225 has a finite limit as @xmath269 [ see below eq.([scolap ] ) ] , so that to leading order and @xmath270 , using eq.([codin1 ] ) , @xmath271 it follows that the function @xmath130 fulfills approximately the free equation @xmath272 in the rapid drop for @xmath273 as well as in the following plateau ( to the right in fig .
[ fig : chifit ] ) , that is @xmath274 where we used that @xmath275 . this behaviour is valid as long as @xmath276 , that is @xmath277 ) $ ] since @xmath278 and @xmath279 . for larger values of @xmath154
the function @xmath130 tends to the unique large distance fixed point given by the purely logarithmic solution @xmath280 and does this through the characteristic damped harmonic oscillations in @xmath195@xcite responsible for the waving behaviour of @xmath29 in the dilute phase [ see fig .
[ fig : eta0 ] ] .
since the crossover to the logarithmic behaviour takes place for larger and larger values of @xmath154 the smaller is @xmath95 , because @xmath281 grows with @xmath95 as well as the crossover point @xmath282 ) $ ] .
when @xmath130 is almost constant , the characteristic oscillations tend to have the same amplitude they have in the regular solutions .
moreover , this almost constant value of @xmath130 , which is of order @xmath263 , decreases faster in @xmath223 than the end point @xmath283 at which @xmath130 is to be evaluated to give @xmath29 as in the second of eqs .
( [ eq : bc2 ] ) .
this explain why the oscillations of @xmath12 are reversed and faster right before the sudden drop in fig .
[ fig : eta1 ] . in other words , the oscillations in the left part of fig .
[ fig : eta1 ] describe the approach to the limiting solution @xmath280 [ @xmath284 while one is getting * off * this limiting solution in the right part of fig .
[ fig : eta1 ] .
this explains why the oscillations on the right get reversed compared with the oscillations on the left of fig .
[ fig : eta1 ] .
this is the case when @xmath285 as @xmath118 and corresponds to the growing branch on the right of fig .
[ fig : eta1 ] .
notice that we have collapsed solutions for arbitrarily large values of @xmath29 contrary to the dilute solutions which only exist for @xmath286 .
since @xmath225 is negative and monotonically decreasing for @xmath287 , the quantity @xmath288 $ ] is exponentially small except in the interval @xmath289 where @xmath225 would vanish in the limit .
we use now this property to get an approximate but analytic singular solution .
that is , the dominant contribution in the r.h.s . of eq .
( [ eq : varinteq ] ) is obtained when all integrations are restricted to the interval @xmath290 .
one easily realizes that , by consistency , @xmath291 . then one finds @xmath292 and in particular using eqs.([eq : phi0 ] ) and
( [ eq : etamu ] ) , @xmath293 = \frac{\mu^2}{24 } \quad,\qquad \eta = \mu^2\ , a \,i_2\left(\frac1{a}\right ) = \frac{\mu^2\,a}{24 } \quad,\qquad \phi(0 ) = \frac{\eta}{a } \;.\ ] ] finally , the leading form for @xmath102 reads @xmath294\displaystyle \frac{\eta}a + \frac{\eta}{2r}\left(\frac{r}a - \frac12 \right)^3 \left(\frac{r}a - \frac52 \right ) \ ; , & a/2 \le r \le 3a/2 \\[0.3 cm ] \displaystyle \frac{\eta}r \ ; , & r \ge 3a/2 \;. \end{cases}\ ] ] notice that @xmath107 as well as its first and second derivatives are continuous at @xmath295 and @xmath296 . for @xmath297 ,
the effective gravitational field has the free form proper of a locally vanishing mass density .
higher corrections can be obtained by iteration over eq .
( [ eq : varinteq ] ) . that is , using the leading form of @xmath215 that can be read out of in eq .
( [ scolap ] ) to evaluate to next to leading order , as @xmath298 , the integrals @xmath299 in the r.h.s . of eq .
( [ eq : varinteq ] ) .
this procedure can then be repeated to obtain further corrections .
( [ scolap ] ) provides the spherically symmetric , leading order singular solution of the mean field equation ( [ eq;bple ] ) where the particles are densely concentrated around one point in a region of size of the order of the cutoff @xmath21 .
in fact , to leading order in @xmath95 the density reads @xmath300\displaystyle 0 \ ; , & r\ge a/2 \ ; , \end{cases}\ ] ] while to next to leading order we have @xmath301 \;,\ ] ] where @xmath102 has the leading order form of eq .
( [ scolap ] ) .
this corresponds to the situation when the gravitational attraction completely overcomes the kinetic energy and all particles fall towards the origin .
of course , such singular solution only exist mathematically for non - zero values of the cutoff @xmath21 .
let us now compute the action for the solution eq .
( [ scolap ] ) from eq .
( [ accefa ] ) .
the regularized laplacian can be computed from the equation of motion ( [ eq;bple ] ) with the leading order result for @xmath302 , @xmath303 the second term is just @xmath304 so that all together , to leading order we obtain for the free energy @xmath305 the first term is just the potential energy of @xmath4 particles clustered in a small sphere of radius @xmath306 , where the regularized gravitational interaction is the same for all @xmath26 particle pairs .
the second term is @xmath307 times the entropy loss in the collapse .
as one could expect , this free energy is large and negative , unbounded from below as @xmath308 . in particular , this free energy is well below the free energy of the dilute solution eq .
( [ seta ] ) and fig . [ fig : action0 ] for all values @xmath309 .
that is , the singular solution eq .
( [ scolap ] ) provides the * absolute * minimum for the free energy eq .
( [ accefa ] ) for any @xmath310 .
the dilute solution eqs.([eq : fund ] ) and ( [ eq : bc ] ) is only a * local * minimum . in the next subsection
we shall address the issue of the metastability of the dilute phase .
the equation of state has the mean field form @xmath311 in both the dilute and collapsed phase . in the former case the r.h.s .
( [ eq : eos ] ) coincides with the function @xmath312 . in the latter we obtain instead , from eqs .
( [ eq : nnlo ] ) and ( [ scolap ] ) , to next to leading order @xmath313 \;.\ ] ] thus , the pressure at the boundary is exponentially small in the collapsed phase . the complete plot of @xmath314 vs. @xmath29 is given in fig .
[ fig : ffun ] .
one can see how the well known @xcite spiraling towards @xmath315 , characteristic of the dilute solutions , is now limited to the case when @xmath223 is of order @xmath95 .
when @xmath223 reaches values of order @xmath316 the inward spiral comes to a halt and then winds back until @xmath317 . for larger values of @xmath223 ,
that is @xmath285 as in the collapsed phase , @xmath314 drops to values exponentially small in @xmath318 .
analogous plots are obtained for other types of cutoff in refs .
@xcite . vs. @xmath29
when @xmath319 ( blue curve ) .
when @xmath223 is of order @xmath95 the blue curve can not be distinguished from the corresponding curve of the case @xmath240 , here depicted in red ( dilute solutions ) .
when @xmath223 is of order @xmath316 the blue curve ( @xmath320 ) and the red curve separate .
the blue curve then traces the red spiral backwards at a distance of order @xmath321 as long as @xmath322 ( interpolating solutions ) .
finally the blue curve drops to zero when @xmath263 is of order @xmath323 ( collapsed solutions ) . ] the dilute solutions studied in sec .
iv are metastable because the collapsed solution provides the absolute minimum for free energy [ eq.([acsing ] ) ] .
the collapse transition from the dilute phase to the collapsed phase can take place through an interpolating solution in an analogous way to the bubbles in liquid - gas first order phase transitions @xcite .
such interpolating solutions are the saddle points of eq .
( [ eq;bple ] ) , which correspond to the sudden drop of @xmath29 as a function of @xmath195 ( see fig . [
fig : eta1 ] ) located approximately at @xmath325 . in this narrow drop region we have @xmath326 , so the interpolating solutions are allowed only in this interval of @xmath29 .
examples of the interpolating solutions are given in fig .
[ fig : chifit ] . according to the qualitative discussion below eq .
( [ eq : free ] ) , the location of the @xmath29 drop can be obtained approximately as the values of @xmath142 at which the function @xmath191 has the crossover from the free solution @xmath327 to the logarithmic solution @xmath328 .
that is for @xmath329 , since @xmath330 and @xmath331 , see eq .
( [ eq : free ] ) . then the scaling as @xmath81 of @xmath332 , the density in the origin , can be obtained from eq .
( [ eq : phixi ] ) and the numerical analysis as @xmath333 because @xmath334 in this regime . on the other hand , from fig .
[ fig : chifit ] one sees that the value of the density for @xmath335 and in particular the value at the border @xmath156 differ by order @xmath336 w.r.t . the dilute phase .
the equation of state ( [ eq : eos ] ) , which holds for any value of @xmath142 , then implies that the pressure in the dilute and interpolating solution , for the same value of @xmath29 , differ only by order @xmath95 , as evident also from fig .
[ fig : ffun ] . in practice
, one can regard an interpolating configuration as a special spherical fluctuation of the dilute phase in which a fraction of particles of order @xmath337 is almost uniformely removed from everywhere and concentrated in a region of size @xmath95 around the origin .
in fact , this result has an heuristic explanation by a simple energy
entropy argument : if we denote by @xmath338 the variation of the density in a region of size @xmath336 around the origin , we then have @xmath339 since the first term is the gain in potential energy per particle and the second the loss of entropy per particle due to the concentration .
we may neglect the effect due to the rest of the particles if we assume that @xmath340 , the fraction of particle moved around , would vanish as @xmath81 .
minimizing eq .
( [ eq : enent ] ) w.r.t .
@xmath341 yields @xmath342 which is indeed consistent with the requirement that @xmath343 vanishes as @xmath118 and rather accurately approximate our numerical results .
notice that the minimized @xmath344 is to be identified with the difference , between interpolating and dilute solution , of the action per particle , so that @xmath345 this result is in very good agreement with the numerical values obtained by solving eqs.([ecinar ] ) and ( [ accir ] ) . according to the interpretation of the interpolating solutions as saddle points through which the dilute phase may decay to the collapsed one
, the lifetime of the dilute phase is given by @xmath346 where @xmath347 stands for the determinant of small fluctuations around the interpolating solution and @xmath348 for that around the dilute solution .
this ratio of determinants can be estimated as follows : in the interpolating profile there should be only one negative mode necessarily localized in a region of size @xmath95 around the origin , where the density is very large , since @xmath349 plays the role of schroedinger potential in the eigenvalue equation for the small fluctuations .
( [ eq : drho ] ) then implies that @xmath350 provides the scale of the potential felt by this negative mode .
hence the corresponding negative eigenvalue should be of the same order , while the rest of the spectrum should be almost unaffected , since the two density profiles differ only by order @xmath336 for @xmath351 .
therefore @xmath352 we obtain from eqs .
( [ ds])-([dete ] ) to leading order the following lifetime for the dilute phase of the self - gravitating gas : @xmath353 \;.\ ] ] one can see that the lifetime becomes infinitely long in the zero cutoff limit as well as when @xmath75 at fixed cutoff [ recall that @xmath34 in the dilute limit of eq .
( [ limter ] ) ] .
we want to stress that the lifetime of the dilute solutions is extremely long for large @xmath4 and small cutoff @xmath21 .
this is due to the fact that although the collapsed solution has a much lower energy than the dilute solution , in the latter there is a huge entropy barrier against the gathering , in a small domain of size @xmath354 , of a number particles large enough to start the collapse .
a qualitatively similar conclusion is reached in ref .
@xcite .
j. binney , s. tremaine , ` galactic dynamics ' , princeton univ .
press , princeton , nj , 1987 .
j. e. peebles , ` principles of physical cosmology ' , princeton univ . press , princeton , nj , 1993 .
w. c. saslaw , ` gravitational physics of stellar and galactic systems ' , cambridge univ .
press , cambridge 1987 .
r. emden , ` gaskugeln ' , teubner , leipzig und berlin , 1907 .
s. chandrasekhar , ` an introduction to the study of stellar structure ' , chicago univ .
press , 1939 .
r. ebert , z. astrophys .
* 37 * , 217 ( 1955 ) .
w.b.bonnor , mon . not .
. soc . * 116 * , 351 ( 1956 ) . v.
a. antonov , vest .
leningrad univ . 7 , 135 ( 1962 ) .
d. lynden - bell and r wood , mon . not .
. soc . * 138 * , 495 ( 1968 ) .
h. j. de vega , n. snchez , ` statistical mechanics of the self - gravitating gas : thermodynamic limit , phase diagrams and fractal structures ' , astro - ph/0505561 , 9th course of the international school of astrophysics ` daniel chalonge ' , palermo , italy , 7 - 18
september 2002 , nato asi , p. 291 - 324 in the proceedings , n. snchez and yu .
parijskij editors , kluwer , 2002 .
p. h. chavanis , phys .
e 65 , 056123 ( 2002 ) , p. h. chavanis , i. ispolatov , phys .
e 66 , 036109 ( 2002 ) , p. h. chavanis , j. sommeria , mnras 296 , 569 ( 1998 ) .
e. follana , v. laliena , phys .
e 61 , 6270 ( 2000 ) .
langer j s in _ fluctuations , instabilities and phase transitions _ , riste t , ed .
page 19 , plenum n.y .
langer j s in _ solids far from equilibrium _
godreche c , cambridge univ . press ( 1992 ) , page 297 , and _ systems far from equilibrium _ , garrido l , et .
notes in phys .
_ * 132 * springer ( 1975 ) .
gunton j d , san miguel m and sahni p s in _ phase transitions and critical phenomena _
, domb c and lebowitz j j , eds .
vol 8 , academic press , ( 1983 ) ; langer js , _ acta metall . _
* 21 * , 1649 ( 1973 ) .
d. boyanovsky , h. j. de vega , d. j. schwarz , ` phase transitions in the early and the present universe ' , hep - ph/0602002 , to appear in ann .
nucl . part . | the self - gravitating gas in thermal equilibrium is studied using a newtonian potential regularized at short distances .
this short distance cutoff permits us to obtain a complete description of the gas including its collapsed phase .
we give a field theory description of the @xmath0-body regularized self - gravitating gas in the canonical ensemble .
the corresponding functional integral is dominated in the @xmath1 limit by saddle points which provide a mean field description . the well - known dilute solutions ( isothermal spheres )
are recovered .
we find new solutions which are regular in the regularized theory but become singular in the zero cutoff limit .
they describe collapsed configurations where the particles are densely concentrated in a region of the size of the cutoff .
these collapsed solutions provide the absolute minimum of the free energy .
we find further new solutions which interpolate between the collapsed and the dilute configurations and describe tunneling processes where the gas collapses .
the transition probability for such collapse processes turns out to be extremely small for large @xmath0 .
that is , the dilute solutions are in practice stable in the regime where they are locally stable . |
transition metal ( tm ) oxides have been one of the main subjects of materials science for decades .
experimental and theoretical efforts are driven by their rich , complex , and potentially useful behaviors originating from strong correlations between electrons and/or electrons and lattices.@xcite the recent developments in the crystal growth techniques , in particular the ( laser ) molecular - beam epitaxy , have made us recognize the opportunity to further control their behaviors and to generate phenomena that are not realized in the bulk systems.@xcite here , we focus on the magnetic behavior at an interface between perovskite manganite and other tm oxides .
perovskite manganites , especially la@xmath7sr@xmath8mno@xmath9 ( lsmo ) , are particularly important because of their ferromagnetic ( f ) metallic behavior with relatively high curie temperature @xmath10 and large polarization . controlling the magnetic interaction at interfaces involving manganites would cause a technological breakthrough for electronic devices using , for example , a tunneling magnetoresistance ( tmr ) effect@xcite and an exchange bias ( eb ) effect.@xcitethis requires the microscopic information on the orbital states , not only on the spin states as demonstrated for cuprate / manganite interfaces in refs . and .
however , it remains controversial whether the magnetic moment is induced in the cuprate region@xcite or the dead layers appear in the manganite region.@xcite the difficulty dealing with interfaces involving strongly - correlated electron systems comes from the small volume fraction which makes the experimental analysis challenging , and strong - correlation effects which hinder some of theoretical treatments .
therefore , if a goodenough - kanamori - type @xcite transparent description of the interfacial magnetic interaction becomes available , both experiment and theory would greatly benefit . in this paper , we present a general consideration for the magnetic interaction at an interface involving manganites .
we first focus on the interfacial interaction derived by @xmath1 orbitals which have the largest hybridization along the @xmath11 layer - stacking direction .
we see that the sign of the magnetic interaction via the @xmath1 orbitals is naturally fixed based on the molecular orbitals formed at the interface and the generalized hund s rule .
the argument uses localized orbitals , and therefore shows only the qualitative trend .
the molecular orbitals effectively lift the degeneracy between @xmath1 and @xmath12 orbitals by the order of the hopping intensity . in the second part ,
we perform the model hartree - fock calculation and show that the broken degeneracy can lead to the additional change in the magnetic interaction between the interfacial mn layer and its neighboring mn layer .
we also discuss how to generalize the molecular - orbital based argument for more complicated situations including @xmath6 orbitals .
in this section , we consider the magnetic interaction between manganite and other tm oxides focusing on the molecular orbitals formed by @xmath1 orbitals which have the largest overlap at the interface .
the tm region is specified by the number of electrons occupying a @xmath1 orbital . here , @xmath6
electrons are assumed to be electronically inactive and considered as localized spins when finite number of electrons occupy @xmath6 orbitals .
generalization including these electrons will be discussed later . _
@xmath13 system_. let us start from the simplest case , an interface between mn and a @xmath13 system ( fig .
[ fig : d0 ] ) . in this case , the bonding ( b ) orbital is occupied by an electron whose spin is parallel to the localized @xmath6 spin in mn , while the antibonding ( ab ) orbital is unoccupied . when there are other unpaired electrons in the @xmath13 system at @xmath12 and/or @xmath6 orbitals , their spins align parallel to that of the electron in the b orbital due to the hund coupling .
thus , the f coupling is generated between mn and @xmath13 systems .
this is equivalent to the double - exchange ( de ) interaction originally proposed by zener.@xcite when @xmath12 is much lower in energy than @xmath1 and in the interfacial mn layer ( termed @xmath12 order ) , mn and @xmath13 systems are virtually decoupled .
thus , the magnetic coupling is due to the superexchange ( se ) interaction between @xmath6 electrons .
this interaction is either f or antiferromagnetic ( af ) depending on the orbital state and the occupancy of the tm @xmath6 level .
orbitals on mn ( right ) and tm with the @xmath13 configuration ( left ) . ] _
@xmath14 normal_. this simple consideration can be easily generalized to @xmath3 and @xmath4 systems .
first we consider that the @xmath1 and @xmath12 are nearly degenerate in the interfacial mn layer and the unoccupied @xmath12 level in the tm region is much higher than the @xmath1 level ( fig .
[ fig : d12 ] , top figures ) .
we call this configuration `` normal '' ( n ) configuration . in the lowest energy configuration ,
b orbitals and the mn @xmath12 orbital are occupied by electrons . for the @xmath3 system ,
the f interaction is favorable as in the @xmath13 system . on the other hand , for the @xmath4 system ,
the down electron orbital is hybridized with the minority band in the mn region .
thus , the `` down '' b orbital is higher in energy and has larger weight on the tm than the `` up '' b orbital . because of the hund coupling with the `` down '' electron in the b orbital , other unpaired electrons , if they exist in @xmath12 and/or @xmath6 orbitals , tend to be antiparallel to the mn spin .
since @xmath12 orbitals are predominantly occupied in the interfacial mn layer due to the b / ab splitting of @xmath1-based molecular orbitals , further stabilization of @xmath12 orbitals in the mn region , i.e. , @xmath12 order , would not affect the interfacial magnetic coupling discussed here .
but , this could reverse the magnetic coupling between the interfacial mn layer and the second mn layer as discussed in the next section . _
@xmath14 anomalous_. when the @xmath12 level in the tm region becomes lower than the mn @xmath12 level , the charge transfer occurs .
we shall call this configuration `` anomalous '' ( an ) configuration ( fig .
[ fig : d12 ] , middle figures ) .
the electron transferred to the tm @xmath12 orbital has the same spin as the higher energy b orbital due to the hund coupling ( indicated by arrows ) .
therefore , the sign of the magnetic coupling between the mn and @xmath14 systems is unchanged .
note that this argument is applicable when the hopping probability between @xmath12 orbitals in the mn and the tm regions is negligibly small .
the finite hopping probability would make the charge transfer continuous . furthermore , when the hopping probability becomes large , the de interaction is generated .
although the de interaction through @xmath12 bonds may not be realistic , it cooperatively stabilizes the f spin alignment for the @xmath3 case , while it competes with the af tendency for the @xmath4 case .
magnetic interactions discussed so far are insensitive to the electron density in the interfacial mn because the interactions are mainly derived from the virtual electron excitation from the occupied @xmath1 orbital in the tm region to the unoccupied counterpart in the mn region .
next , we consider that the @xmath1 level is much lower than the @xmath12 in the interfacial mn layer due to either the local jahn - teller distortion or compressive strain originating from the substrate ( fig .
[ fig : d12 ] , lower figures denoted by `` jt '' ) .
the magnetic coupling in this case is sensitive to the electron density of the interfacial mn .
orbitals on mn and the @xmath3 system ( left column ) and the @xmath4 system ( right column ) . in the normal ( anomalous ) configurations ,
@xmath1 and @xmath12 orbitals are nearly degenerate in the interfacial mn , and the unoccupied @xmath12 orbital in the neighboring tm is higher in energy ( lower in energy than the occupied mn @xmath12 ) . in the jt case , the @xmath1 level is significantly lower than the @xmath12 level .
black ( light ) lines indicate the level of majority ( minority ) spins .
the up level and down level are exchange split resulting in the level scheme as indicated .
the minority levels are neglected in the upper left two because these are irrelevant . ] _
@xmath3 jt_. when the mn @xmath1 density is close to 1 , the se interaction between the occupied @xmath1 orbitals becomes af . on the other hand ,
when the density is much less than 1 , the f interaction between @xmath13 configuration on mn and @xmath3 becomes dominant . _
@xmath4 jt_. when the mn @xmath1 occupancy is close to 1 , `` up ''
electrons are localized on each sites because both b and ab molecular orbitals are occupied while down electrons can be excited or leaked from the @xmath4 system to the mn minority level , i.e. , down electron density is virtually reduced in the tm region . as a result , unpaired spins , if they exist in @xmath12 and/or @xmath6 orbitals , become parallel to the up spin , i.e. , f coupling .
when the mn @xmath1 density becomes much less than 1 , the up ab orbital becomes less occupied while keeping the occupancy of b orbitals relatively unchanged . eventually , the down density in the tm region becomes larger than the up density , and the magnetic coupling between the mn and @xmath4 regions becomes af .
in the previous section , we discussed the interfacial magnetic coupling controlled by the molecular orbitals .
the separation between b and ab molecular levels can become as large as the order of @xmath15 , the hybridization between @xmath1 orbitals along the @xmath11 direction . since the interfacial mn @xmath1 band is represented by the b ( ab ) @xmath1 orbital for a @xmath16/manganite interface , the @xmath17 degeneracy is effectively lifted in the interface layer .
this degeneracy lifting is expected to affect the magnetic interaction in the mn region . in this section
, we discuss this effect using the microscopic model calculation .
we consider a two - band de model given by @xmath18 here , an electron annihilation operator at site @xmath19 and orbital @xmath20 $ ] is given by @xmath21 , @xmath22 , and the level difference between @xmath23 and @xmath24 is given by @xmath25 .
we consider the large hund coupling limit , in which the spin direction of a conduction electron is always parallel to that of a localized @xmath6 spin on the same site , and omit the spin index . instead , the relative orientation of @xmath6 spins is reflected in the hopping matrix ; @xmath26 is the unitary transformation representing the rotation of the spin direction between sites @xmath19 and @xmath27 . for simplicity ,
we only consider nearest - neighboring ( nn ) hoppings between mn @xmath17 orbitals via oxygen @xmath28 in the middle .
using the slater - koster scheme,@xcite the orbital dependence of @xmath29 is written as @xmath30 , @xmath31 , @xmath32 , and @xmath33 .
the third term represents interorbital coulomb interaction . due to the @xmath17 symmetry ,
@xmath34 is related to the intraorbital coulomb interaction @xmath35 and the interorbital exchange integral @xmath36 as @xmath37 .
the last term represents the af se interaction between nn @xmath6 spins @xmath38 . from the optical measurements ,
the on - site interactions are estimated as @xmath39 ev and @xmath40 ev.@xcite the density - functional theory calculation provides @xmath41 ev.@xcite using the mean - field analysis for the nel temperature @xmath42 of camno@xmath9 , one estimates @xmath43
mev.@xcite a similar value is obtained from the magnon excitations in the a - af phases of 50 % doped pr@xmath7sr@xmath8mno@xmath9 and nd@xmath7sr@xmath8mno@xmath9 supposing that the af interaction is due to the same @xmath44.@xcite thus , in what follows , we take @xmath45 . considering some ambiguity ,
the realistic value for @xmath46 is expected to be @xmath470.05 .
we analyze the model hamiltonian , eq .
( [ eq : hamiltonian ] ) , using the hartree - fock approximation at @xmath48 focusing on the doped region ( carrier density @xmath49 far away from 1 ) . in light of the experimental reports
, we compare the energy of the following eight magnetic orderings : f ordering , planar af ordering in which spins align ferromagnetically in the @xmath50 ( @xmath51 or @xmath52 ) plane [ a ( a ) ] , chain - type af ordering in which spins align ferromagnetically along the @xmath11 ( @xmath53 or @xmath54 ) direction [ c ( c ) ] , zigzag af in which spins form ferromagnetic zigzag chains in the @xmath50 ( @xmath51 or @xmath52 ) plane [ ce ( ce ) ] , and nacl - type af ( g ) . at @xmath55 , in addition to the spin symmetry breaking , orbital symmetry can be broken due to the se mechanism in the present model.@xcite since we are focusing on the metallic regime @xmath56 , we do not consider such a symmetry breaking .
the numerical results for the bulk phase diagram are presented in figs .
[ fig : pd ] ( a)-[fig : pd ] ( c ) . here , all phase boundaries are of first order , and those at small @xmath49 can be replaced by canted af phases or the phase separation between the undoped g - af phase and doped f or af phases . the overall feature is consistent with the previous theoretical reports.@xcite at @xmath57 , a - af and a-af ( c- and c- , ce- and ce- ) are degenerate but the degeneracy is lifted by the finite @xmath25 .
we found the ce phase at @xmath58 at @xmath59 and @xmath60 ( not shown ) as in the previous reports.@xcite at @xmath61 , the ce phase becomes unstable against a- and c - af phases and appears only at the positive @xmath25 with @xmath62 .
@xmath63 reproduces the phase diagram of a high @xmath10 system such as lsmo and pr@xmath7sr@xmath8mno@xmath9 ( refs . and fairly well .
mean - field phase diagrams of doped manganites as a function of electron density @xmath49 and the af interaction @xmath44 for three choices of the level difference between @xmath17 orbitals @xmath25 . at @xmath64
, @xmath1 is lower ( higher ) in energy than @xmath12 . for notations of the magnetic phases ,
see the main text .
( d ) phase diagram for the 20-layer slab with @xmath65 in the surface layers .
@xmath49 in this case corresponds to the mean electron density . for a1 and a2 phases ,
schematic spin alignments are also shown .
a dashed line is the phase boundary between f and a - af phases in the bulk calculation . ]
the main effect of the level separation @xmath25 is changing the stability of planar - type af ( a or a ) with respect to the chain - type af ( c or c ) and f states .
in particular , the a - af phase is stabilized by the positive @xmath25 more strongly than the c - af phase by the negative @xmath25 .
this is because the energy gain by the de mechanism is favorable for the a - af than the c - af .
the result is semiquantitatively consistent with the previous report based on the density - functional theory.@xcite at @xmath65 , the boundary between f and a - af phases is moved down to @xmath66 at @xmath67 .
this behavior suggests that , when the @xmath1 ab level for the @xmath14/manganite interfaces is about @xmath15 higher than the mn @xmath12 level , the magnetic coupling between the interfacial mn and the second mn layers is switched to af while retaining the intraplane f coupling .
we confirmed this behavior by computing the surface phase diagram considering f phase and two af phases : a1(2 ) where the surface layer ( and the second layer ) is antiferromagnetically coupled to its neighbor .
we introduce positive @xmath25 only on the surface layers in the 20-layer slab . as shown in fig .
[ fig : pd ] ( d ) , a large part of f phase is replaced by a1 phase compared with the bulk phase diagram ( b ) .
( precise phase boundary requires detailed information of the surface or interface . )
although the parameter regime is small , it is also possible that surface three layers are af coupled while the other couplings remain f , a2 phase , before the whole system enters a - af when @xmath44 is increased or @xmath49 is decreased . when the @xmath1 orbital is stabilized , the inter - mn - layer coupling remains f but the intraplane f coupling is reduced .
therefore , in - plane canted af structure may result for small @xmath49 .
summarizing , we presented a general consideration on the magnetic interaction between the doped manganite and other transition metal oxides when an interface is formed . using the molecular orbital formed at the interface and the generalized hund s rule , the sign of the magnetic interaction is determined ( sec .
the bonding / antibonding splitting of the molecular orbitals leads to the degeneracy lifting of @xmath17 orbitals on the interface mn layer .
further , the bulk strain lifts the @xmath17 degeneracy .
these effects control the magnetic interaction in the interfacial mn plane and between the interfacial mn plane and its neighbor ( sec .
iii ) . considering these effects
, we summarized the magnetic couplings in @xmath68/manganite interfaces in table [ tab : coupling ] .
although the present argument is rather qualitative , it is physically transparent and can be applied to a variety of systems .
it is also straightforward to generalize the argument to include other orbitals .
therefore , the present argument will also help a more quantitative analysis with detailed information from either the experiment or the first principle theory .
.magnetic interaction at an interface between mn and tm with the @xmath2 configuration .
the interfacial mn is indicated by mn(1 ) , and mn in the second layer by mn(2 ) .
mn(1)-mn(1 ) indicates intraplane interaction while the others interplane interactions .
the tm - mn(1 ) interaction is based on the molecular - orbital picture presented in sec .
ii while the mn - mn interaction is based on the model hartree - fock study presented in sec .
iii . at @xmath12 order , @xmath12 orbital
is stabilized at the ( interfacial ) mn layer . at f
* , mn - mn interaction is weak , and the canted af ordering may result . see the stabilization of the c - af phase by the jt - type distortion @xmath69 in fig .
[ fig : pd ] ( a ) , the stabilization of the c- and a - af phases by reducing the carrier density @xmath49 in fig .
[ fig : pd ] ( b ) , and the stabilization of the a1 phase by the interfacial @xmath12 order in fig .
[ fig : pd ] ( d ) . [ cols="^,^,^,^,^",options="header " , ] it is worth discussing the implication of the present results to the real systems .
an example of the @xmath4 system is high-@xmath70 cuprate .
it has been reported that the magnetic coupling between yba@xmath71cu@xmath9o@xmath72 ( ybco ) and la@xmath7ca@xmath8mno@xmath9 is af,@xcite and @xmath1 and @xmath12 in the interfacial cu have a similar amount of holes.@xcite this corresponds to the an situation .
f coupling due to the de remains in the mn region because of the finite band width of @xmath12 .
an example of the @xmath3 system is bifeo@xmath9 ( bfo ) .
recently , the eb effect was reported at bfo / lsmo interfaces accompanying the `` af '' coupling between bfo and lsmo.@xcite we expect the n situation with @xmath12 ordering at this interface . although the interfacial coupling is f , the af coupling between the interfacial mn and the second mn layers results in the af alignment between bfo and bulk lsmo as observed experimentally and is responsible for the exchange bias effect .
a question one may ask is what causes the `` an situation '' in ybco and the `` n situation '' in bfo ? a qualitative explanation is as follow : in ybco , the unoccupied cu @xmath12 state is right above the mott gap and its position is nearly identical to the occupied band of manganites.@xcite therefore , the charge transfer from mn @xmath17 to cu @xmath12 can easily occur . on the other hand , the high - spin state is realized in bfo , and the unoccupied @xmath12 state with opposite spin with respect to the majority electrons is located far above the gap . in addition , the very close chemical potentials ( i.e. close @xmath1 levels ) of bfo ( ref . ) and lsmo ( ref . ) maximize the b and ab splittings .
this situation is favorable for the @xmath12 ordering in the interfacial mn layer and the resulting af coupling between the first and the second mn layers [ see fig . [
fig : pd ] ( d ) ] .
finally , an example of the @xmath13 system may be non - magnetic srtio@xmath9 , and the coupling with this is expected to affect the magnetic state near the interfacial mn . for small doping @xmath53 of lsmo , the coupling with srtio@xmath9 ( with the smaller lattice constant of srtio@xmath9 ) increases the @xmath1 orbital occupancy suppressing the inplane de effect . for large doping , srtio@xmath9 creates the tensile strain stabilizing @xmath12 , and the out - of - plane f coupling is reduced.@xcite both are expected to cause a more rapid decrease in the ordered moment with increasing temperature than in the bulk region,@xcite resulting in the rapid suppression of the tmr effect in lsmo / srtio@xmath9/lsmo junctions.@xcite the in - plane ( out - of - plane ) spin canting may also be realized in the former ( latter).@xcite for undoped lamno@xmath9 , the out - of - plane ferromagnetic coupling may result because the overlap between the occupied @xmath1 in the first mn layer and the unoccupied @xmath73 or @xmath74 orbitals in the second mn layer is increased , favorable for the f se interaction between mn layers .
orbitals on mn and ti , originally @xmath75 . here , only bonding orbitals are shown .
( a ) full level diagram including both @xmath17 and @xmath6 .
black ( light ) lines indicate the level of majority ( minority ) spins .
the highest occupied molecular orbitals and the magnetic alignment depend sensitively on the detail of the interface as shown in ( b ) and ( c ) .
( b ) [ ( c ) ] the up b orbital of @xmath1 orbital is lower ( higher ) in energy than the down b orbitals of @xmath76 . in ( b ) ,
induced magnetic moment in ti is parallel to mn , i.e. , @xmath13 configuration , while in ( c ) , it depends on the relative occupancy of ti @xmath76 orbitals in the up and down b orbitals . when the occupancy of the ti down @xmath76 orbitals is larger than the up @xmath76 orbitals , net moment induced in ti site becomes antiparallel to the mn moment . ] _ extention to @xmath6 systems_. in @xmath6 systems such as titanates , vanadates , and cromates , coupling between @xmath6 orbitals could become as important as the coupling between @xmath17 orbitals . here
, we discuss how to generalize the molecular - orbital argument presented in sec .
ii to @xmath6 systems . as an example
, we consider an interface between titanate with the @xmath75 configuration and manganites . extending the argument to other systems
is straightforward .
figure [ fig : timn ] ( a ) shows the level diagram of the titanate / manganite interface including both @xmath17 and @xmath6 orbitals . for simplicity , only bonding orbitals are presented .
because the @xmath1 level in titanate is far above the occupied levels in manganite and the unoccupied @xmath6 levels in titanate and manganite ( down electrons for the latter ) are close,@xcite highest occupied molecular orbitals could be either b up @xmath1 or @xmath77 orbital [ fig . [ fig : timn ] ( b ) which is equivalent to fig .
[ fig : d0 ] ] or b down @xmath76 orbitals [ fig .
[ fig : timn ] ( c ) ] .
note that the interfacial hybridization between @xmath78 orbitals on ti and mn is much smaller than those between @xmath79 and between @xmath80 . in the case of fig .
[ fig : timn ] ( b ) , induced moment in the titanate region is tiny but parallel to the moment in manganite region . on the other hand in the case of fig .
[ fig : timn ] ( c ) , the induced moment in the titanate region could be either parallel or antiparallel to the manganite moment .
this depends on the relative weight of the down electron density in the b @xmath76 orbitals with respect to the up electron density in the b @xmath76 orbitals .
the situation in fig .
[ fig : timn ] ( c ) with antiparallel spin arrangement between titanate and manganite could happen when the original electron density in the manganite @xmath17 orbital is large and @xmath1 level both in titanate and manganite is high due , for example , to the in - plane tensile strain .
but in general , the difference between two configurations , figs .
[ fig : timn ] ( b ) and [ fig : timn ] ( c ) with either parallel or antiparallel spin configurations , would be subtle .
therefore , depending on a variety of condition such as the sample preparation , any situation could be realized . when the number of @xmath6 electrons is increased , such as doped titanates , vanadates , and cromates , electrons tend to enter the down b @xmath76 orbitals and , then , the down b @xmath78 , resulting in the antiparallel spin configuration .
however , the antiparallel configuration becomes unstable against the parallel configuration when the electron number in @xmath6 orbitals becomes large and the level separation between @xmath6 and @xmath17 orbitals , i.e. , @xmath81 , becomes relatively small . in this case
, because of the strong on - site coulomb interactions , the energy gain by forming b orbitals becomes small for @xmath6 electrons and comparable to having electrons in both b and ab orbitals with the parallel spin configuration .
the parallel configuration further lowers the energy by forming @xmath1 b orbital and by the hund coupling between the @xmath1 b orbital and @xmath6 electrons on the @xmath6 system , i.e. , the zener s double - exchange ferromagnetism discussed in sec .
when there are more than three @xmath82 electrons with relatively large @xmath81 , a low spin state is realized . in this case , three electrons enter b orbitals formed with @xmath6 minority bands of mn and remaining electrons enter ab orbitals formed with @xmath6 majority band of mn .
thus , the antiparallel configurations persist .
such a situation may be realized in , for example , an interface between manganites and srruo@xmath9 in which ru@xmath83 is in a low spin state with @xmath84 .
this af configuration can be turned to the f configuration when the double - exchange - type interaction becomes dominant due to the formation of @xmath1 b orbital.@xcite so far , we have considered the ideal lattice structure in which orbitals with different symmetry do not hybridize . in reality ,
bond angle formed by two transition metal ions and an oxygen ion in between becomes smaller than 180@xmath85 allowing electrons to hop between orbitals with different symmetry . as a result ,
additional magnetic channels are generated .
a simple argument presented in this paper can be generalized to deal with such a situation .
the author thanks p. yu , r. ramesh , j. santamaria , and c. panagopoulos for stimulating discussions and sharing the experimental data prior to publication , j. kune for discussion , and the kavli institute of theoretical physics , university of california santa barbara , which is supported in part by the national science foundation under grant no . phy05 - 51164 , for hospitality .
this work was supported by the materials sciences and engineering division , office of basic energy sciences , u.s .
department of energy .
m. imada , a. fujimori , and y. tokura , rev .
phys . * 70 * , 1039 ( 1998 ) .
m. izumi , y. ogimoto , y. konishi , t. manako , m. kawasaki , and y. tokura , mater .
b * 84 * , 53 ( 2001 ) .
a. ohtomo , d. a. muller , j. l. grazul , and h. y. hwang , nature ( london ) * 419 * , 378 ( 2002 ) .
s. okamoto and a. j. millis , nature ( london ) * 428 * , 630 ( 2004 ) .
a. ohtomo and h. y. hwang , nature ( london ) * 427 * , 423 ( 2004 ) ; a. brinkman , m. huijben , m. van zalk , j. huijben , u. zeitler , j. c. mann , w. g. van der wiel , g. rijnders , d. h. a. blank , and h. hilgenkamp , nature mater . * 6 * , 493 ( 2007 ) ; n. reyren , s. thiel , a. d. caviglia , l. fitting kourkoutis , g. hammerl , c. richter , c. w. schneider , t. kopp , a .- s .
retschi , d. jaccard , m. gabay , d. a. muller , j .-
m . triscone , and j. mannhart , science * 317 * , 1196 ( 2007 ) .
j. stahn , j. chakhalian , c. niedermayer , j. hoppler , t. gutberlet , j. voigt , f. treubel , h .- u .
habermeier , g. cristiani , b. keimer , and c. bernhard , phys .
b * 71 * , 140509(r ) ( 2005 ) .
j. chakhalian , j. w. freeland , g. srajer , j. strempfer , g. khaliullin , j. c. cezar , t. charlton , r. dalgliesh , c. bernhard , g. cristian , h .- u .
habermeier , and b. keimer , nat . phys .
* 2 * , 244 ( 2006 ) .
a. hoffmann , s. g. e. te velthuis , z. sefrioui , j. santamar ' ia , m. r. fitzsimmons , s. park , and m. varela , phys .
140407 ( r ) ( 2005 ) . j. chakhalian , j. w. freeland , h .- u .
habermeier , g. cristiani , g. khaliullin , m. van veenendaal , and b. keimer , science * 318 * , 1114 ( 2007 ) .
yu , j .- s .
lee , s. okamoto , m. d. rossell , m. huijben , c .- h .
yang , q. he , j .- x .
zhang , s. y. yang , m. j. lee , q. m. ramasse , r. erni , y .- h .
chu , d. a. arena , c .- c .
kao , l. w. martin , and r. ramesh , phys .
lett . * 105 * , 027201 ( 2010 ) .
n. n. kovaleva , a. v. boris , c. bernhard , a. kulakov , a. pimenov , a. m. balbashov , g. khaliullin , and b. keimer , phys .
lett . * 93 * , 147204 ( 2004 ) . c. ederer , c. lin , and a. j. millis , phys .
b * 76 * , 155105 ( 2007 )
. e. o. wollan and w. c. koehler , phys . rev .
* 100 * , 545 ( 1955 ) . h. kawano - furukawa , r. kajimoto , h. yoshizawa , y. tomioka , h. kuwahara , and y. tokura , phys .
b * 67 * , 174422 ( 2003 ) .
r. maezono , s. ishihara , and n. nagaosa , phys .
b * 58 * , 11 583 ( 1998 ) . j. van den brink and d. khomskii , phys
lett . * 82 * , 1016 ( 1999 ) .
z. fang , i. v. solovyev , and k. terakura , phys .
lett . * 84 * , 3169 ( 2000 ) . | a general consideration is presented for the magnetic interaction at an interface between a perovskite manganite and other transition metal oxides .
the latter is specified by the electron number @xmath0 in the @xmath1 level as @xmath2 .
based on the molecular orbitals formed at the interface and the generalized hund s rule , the sign of the magnetic interaction is rather uniquely determined .
the exception is when the @xmath1 orbital is stabilized in the interfacial manganite layer neighboring to a @xmath3 or @xmath4 system . in this case , the magnetic interaction is sensitive to the occupancy of the mn @xmath5 orbital .
it is also shown that the magnetic interaction between the interfacial mn layer and the bulk region can be changed .
manganite - based heterostructures thus show a rich magnetic behavior .
we also present how to generalize the argument including @xmath6 orbitals . |
autoassociative networks are useful models of one of the basic operations of cortical networks@xcite .
` hebbian ' plasticity on recurrent connections , e.g. in the higher level areas of sensory cortex and in the hippocampus , is the crucial ingredient for autoassociation to work , with real neurons @xcite .
neural network models , although very simplified and abstract , allow a comprehensive analysis , indicating whether associative memory retrieval can proceed safely , or whether it must face dynamical hurdles , such as ` spurious ' local minima in a free - energy landscape .
the dynamics of such networks , in the simplest models , is governed by a number @xmath1 of dynamical attractors , each of which corresponds to a distribution of neural activity , i.e. a pattern , which represents a long term memory .
memory is stored by superimposed synaptic weight changes , and the basic operation proceeds by supplying the network with an external signal that acts as a cue , correlated , perhaps only weakly , with a pattern , and which leads through attractor dynamics to the retrieval of the full pattern .
+ how smoothly can such an operation proceed , and how wide are the basins of attraction of the @xmath1 memory states ? clearly , these issues depend critically on whether other attractors exist , that could hinder or obstruct retrieval as a crude example , if the cue is correlated with the image of a mule , the net may be able to retrieve either a horse or a donkey , if no `` mixed '' attractor exist .
if instead the encoding procedure has , unintentionally , created a spurious attractor for the mule itself , the network will likely be stuck in such a mixed memory state . in a slightly more complicated model endowed with some topographic mapping of visual space , a horse cue and a donkey cue
might be presented simultaneously in neighbouring positions .
if they are too close in visual space and spurious attractors exist , this topographic map might retrieve two mules next to each other .
returning to nets without spatial structure and considering for simplicity only symmetric mixtures of patterns embedded with equal strengths , there are obviously @xmath2 2-mixtures , @xmath3 3-mixtures , and so on .
do they correspond to stable attractors , and as such do they influence the network dynamics ? + in addition , connectionist modelers have proposed to describe in terms of spurious states certain psychiatric dysfunctions @xcite .
speech disorders in schizophrenic patients , for instance , might arise from the existence of a large number of spurious states , that obstruct the retrieval of correct patterns @xcite . in their seminal investigation of the hopfield model @xcite , amit , gutfreund and sompolinsky found that while symmetric mixtures of an even number of patterns are unstable , odd mixtures and the spin glass phase can be stable , in a certain region of phase space @xcite . in the hopfield model ,
though , neurons are modelled as binary units , and correspondingly each distribution of activity , in particular each memory pattern , is a binary vector .
either or both of these aspects might be essential in producing the additional minima in the free - energy landscape .
real neurons behave very differently from binary units in many respects , a basic one being that their spiking activity , once filtered with a short time - kernel @xcite , is better approximated by an analog variable .
threshold - linear units reproduce this graded nature of neural response , yet still allow for a simple and complete statistical mechanics analysis of autoassociative network models @xcite . with threshold - linear units , the memory patterns encoded in the synaptic weights can still be taken to be binary vectors , but can also be taken to be drawn from a distribution with several discrete activity values , or from a continuous distribution @xcite .
exponential distributions , in particular , can be argued to be not far from experimentally observed spike count distributions @xcite .
the question of mixture states in analog nets was first addressed in @xcite , arguing that the multiple local minima of the spin glass phase are fewer in number in an associative net of units with more continuous ( sigmoid ) transfer function .
later it was found , considering threshold - linear units , that are both realistic and amenable to analytical treatment @xcite , that the region of stability of the spin glass phase is severely restricted with such units @xcite , again indicative of a general smoothing of the free - energy landscape with analog variables .
although these analyses provide a good starting point , they are not complete in the sense that they did not show what will happen to @xmath0-mixture states with @xmath0 small ( the ones relevant to models of schizophrenia ) , and what is the effect of different coding schemes , that is pattern distributions . here
, we consider instead symmetric @xmath0-mixtures , with @xmath4 , and we consider non - binary memory vectors .
also , from the biological point of view , it is important to study nets with diluted ( incomplete ) connectivity , which are much more realistic descriptions of cortical @xcite and hippocampal networks @xcite , where the probabiliy of a recurrent connection between any two units may be of the order of a few percent . in this manuscript we show that symmetric mixture states give rise to dynamical attractors only in very restricted circumstances , in associative networks of threshold - linear units , both with full and diluted connectivity .
we have analysed the validity of this statement in different coding schemes , and did not find any stable mixture state at all , when memory patterns are not binary .
essentially , we conclude that this type of spurious states are a pathological feature of the simplified binary models considered in the initial studies .
we use a model very similar to that analysed in @xcite .
we consider a fully connected network of @xmath5 units , taken to model excitatory neurons .
the level of activity of unit @xmath6 is a dynamical variable @xmath7 , which corresponds to the short time averaged firing rate of the neuron .
units are connected to each other through symmetric weights .
the specific covariance hebbian learning rule we consider prescribes that the synaptic weight between units @xmath6 and @xmath8 be given as @xmath9 where @xmath10 represents the activity of unit @xmath6 in pattern @xmath11 .
each @xmath10 is taken to be a quenched variable , drawn independently from a distribution @xmath12 , with the constraints @xmath13 , @xmath14 . as in one of the first extensions of the hopfield model @xcite
, we thus allow for the mean activity @xmath15 of the patterns to differ from the value @xmath16 of the original model @xcite .
the model further assumes that the input to unit @xmath6 takes the form @xmath17 where the first term enables the memories encoded in the weights to determine the dynamics ; the second term allows for external signals @xmath18 to cue the retrieval of one or several patterns ; and the third term is unrelated to the memory patterns , but is designed to regulate the activity of the newtork , so that at any moment in time @xmath19 .
the activity of each unit is determined by its input through a threshold - linear function @xmath20 where @xmath21 is a threshold below which the input elicits no output , @xmath22 is a gain parameter , and @xmath23 the heaviside step function .
units are updated , for example , sequentially in random order , possibly subject to fast noise .
the exact details of the updating rule and of the noise are not specified further , here , because they do not affect the steady states of the dynamics , and we take the noise level @xmath24 to be vanishingly small , @xmath25 .
discussions about the biological plausibility of this model for networks of pyramidal cells can be found in @xcite , and will not be repeated here .
subject to the above dynamics , the network evolves towards one of a set of attractor states . in a given attractor
the network may still wander among a variety of configurations , but it reaches a stationary probability distribution of being in any particular configuration .
the average of any quantity over such ` annealed ' probability distribution is denoted by @xmath26 ( whereas @xmath27 denotes the average over the quenched distribution @xmath12 ) .
to analyse such a model one can introduce , as in @xcite the order parameters : @xmath28 where @xmath29 is simply the mean activity of the network , and @xmath30,the subtracted , or specific , overlap of the current state of the network with each of the stored patterns .
two further parameters , @xmath31 } , \end{aligned}\ ] ] can be defined as a function of @xmath32 and @xmath33 , and play a particularly useful role in the analysis in the limit we consider , @xmath25 , when one configuration dominates the annealed average , and @xmath34 .
the characteristic noise scale of the system is @xmath35 @xcite , and we define the storage load @xmath36 . in the limit @xmath37 , the system is thus characterized by the parameters @xmath15 ( mean pattern activity , which also parametrizes the coding sparseness @xcite in the sense that decreasing @xmath15 makes the code sparser ) , @xmath38 ( storage load ) , @xmath22 ( gain ) and @xmath21 ( threshold ) .
we calculate the free energy using the replica trick , for symmetric @xmath0-mixture states ( where @xmath0 overlaps take the same non - zero value , and the rest are zero ) elicited by external signals @xmath39 .
these signals can be purely transient , so that at steady state @xmath40 , but we consider a non - zero steady value for the sake of generality . we look for symmetric states , characterized by non - zero @xmath41 .
the saddle point equations reduce to @xcite : @xmath42 where now the input to each unit can be expresed as : @xmath43 and the free energy reads : @xmath44 if one defines new parameters @xmath45 ( the specific signal - to - noise ratio ) and @xmath46/(t_0\rho)$ ] ( a sort of uniform field - to - noise ratio ) , it is easy to show that the mean field equations can be reduced to : @xmath47 where @xmath48 with @xmath49and @xmath50 . in the equations above , @xmath51 and
the subscript @xmath52 indicates that the @xmath53-average has to be carried out only in the range where @xmath54 . in the following we take @xmath55 .
thus symmetric @xmath0-mixture attractors exist if we can find stable solutions of eqs.[e1 ] , [ e2 ] .
to analyze the stability of the extrema of the free energy , one has to study the hessian matrix @xmath56 around the saddle point . in general , for @xmath0-mixture states , there are three types of eigenvalues : \1 . a non - degenerate eigenvalue , which decides the stability against a uniform increase in the amplitude of the @xmath0 patterns that contribute to the thermodynamic state ( i.e. the condensed patterns ) , while the other overlaps remain zero .
it is ( for @xmath57 ) @xmath58 \2 .
an eigenvalue of degeneracy @xmath59 , associated with any direction which tends to change the relative amplitude of the non - zero overlaps .
it is ( again for @xmath57 ) @xmath60 \3 .
the third eigenvalue , with degeneracy @xmath61 , measures the stability against the appearance of additional overlaps .
in order to proceed further , we restrict the analysis to a number of specific coding schemes , i.e. , to different choices for the distribution @xmath12 . we consider @xmath63 for small values of the load @xmath64 ( and hence of the quenched noise @xmath65 ) , eq.[e2 ] describes an hyperbole , whose center depends on the value of @xmath22 .
eq.[e1 ] instead , for small values of @xmath64 , is a closed curve in the quartant @xmath66 , so that with an appropriate choice of @xmath22 the two curves intersect at two points .
as @xmath64 grows , the region @xmath67 shrinks in size , until at a certain value of @xmath64 , which depends only on @xmath68 and the coding scheme , it reduces to a point and then disappears .
we have investigated not just the existence but also the stability of solutions for symmetric 2- and 3-mixture states .
the solutions behave exactly in the same manner in these two cases : for small values of @xmath15 both intersections discussed above are unstable , in the sense that both @xmath69 and @xmath70 are negative .
this finding is confirmed by computer simulation , in which one of the overlaps tends to grow , reaching the corresponding attractor , whereas the other one ( or the other two in the case of 3-mixtures ) tend to zero .
increasing the value of the sparsity parameter , one finds different results with binary coding and with other types of coding .
let us consider binary coding first .
after a range of @xmath15-values with only one unstable eigenvalues ( @xmath71 or @xmath72 ) , one finds a range where genuinely stable solutions can be found .
thus the retrieval of mixture patterns is possible for binary coding , as can be seen in the simulations shown in fig.[fig1 ] .
the exact stability region in the @xmath73 plane differs for 2-mixtures and 3-mixtures . in both cases ,
it is delimited to the right by the ` critical load ' @xmath74 , i.e. the value at which the island with @xmath67 shrinks to zero , and to the left by the load @xmath38 beyond which no intersection with both @xmath75 and @xmath76 can be found .
fig.[fig2 ] illustrates these stability regions , compared with the critical load for the pure attractor states , as in @xcite . for ternary and exponential coding , the solutions of the saddle point equations
remain unstable even for very high values of the sparsity parameter @xmath15 . again , this was verified by computer simulations .
fig.[fig3 ] illustrates the different situation occurring with ternary and binary coding , by considering a very low load and a sparsity value for which stable solutions for 3-mixtures are easily found in the binary case .
note that the ` critical load ' for 3-mixtures would be considerably higher with ternary patterns ( not shown ) ; the fact is that at each position of the intersection , either @xmath71 or @xmath72 or both turn out to be negative .
this complex behaviour of eigenvalues will be discussed elsewhere in more detail .
we have also extended the analysis to a highly diluted network @xcite . in this case the number of patterns that can be stored scales with the number @xmath77 of connections each unit receives , rather than with the number of units @xmath5 .
one then redefines the load parameter as @xmath78 . the essential difference introduced by the sparse ( i.e. diluted ) connectivity is that noise has less of an opportunity to reverberate along closed loops .
in fact the signal , which during retrieval is simply contributed by the ` condensed ' patterns , propagates coherently and proportionally to @xmath77 , independently of the density of feedback loops in the network .
the fluctuations in the overlaps with the undecondensed patterns , which as @xmath25 represent the sole source of noise , propagate coherently along feedback loops , giving rise to the amplifying factor @xmath79 of the fully connected case . for a given load ( fixed @xmath64 ) ,
diluted connectivity reduces therefore the influence of this ` static ' noise , and performance is better than in the fully connected case with @xmath80 . in particular with the extreme dilution ,
that is , if the condition @xmath81 is satisfied , one can neglect correlations among the @xmath77 inputs to a given unit @xcite , and the mean field equations become @xcite : @xmath82 examining again the stability matrix , we find that the mixture solutions , that were present with binary coding and large values of @xmath15 , still survive . by the token ,
the results for ternary and exponential coding are not affected , in the sense that no stable solutions can be found even in the highly diluted case .
the conclusion is that the existence of stable mixture states in a restricted region of the parameter space should be regarded as almost a pathological feature , resulting from binary coding .
if one considers mixture states as spurious states , to be avoided , then one notes that the introduction of analog variables , a more realistic description of neural activity , goes a long way towards disposing of spurious states , just as it almost eliminated the spin glass phase @xcite .
the remaining region of stability of spurious states is definitely eliminated by non - binary coding schemes , that further contribute to smooth the free - energy landscape .
this result casts doubts upon e.g. models of schizophrenia that are based on the existence of spurious attractors .
these results may well have implications in domains outside computational neuroscience .
the smoothness of the free - energy landscape is a crucial features of many interacting systems used to map optimization problems , such as the travelling salesman@xcite or the graph matching problem @xcite .
optimization generally fails if the dynamics gets stuck into local minima .
our result indicates that undesired local minima may be eliminated by a combination of analog variables and coding schemes , which may in some cases be manipulated while mapping the problem at hand onto a dynamical system .
rolls , a. treves , _ neural networks and brain function _ , ( oxford u.p . , oxford , 1998 ) .
a. treves , e.t .
rolls , network * 2 * 371 ( 91 ) d.j.amit , _ modeling brain function _ , ( cambridge u.p . , cambridge , 1989 ) .
r. e. hoffman , arch . of gen
. psych . * 44 * 178 ( 1987 ) j.j .
hopfield , proc .
usa * 79 * 2554 ( 82 ) d.j .
amit , h. gutfreund , h. sompolinsky , phys . rev . * a 32 * 1007 ( 85 ) , d.j .
amit , h. gutfreund , h. sompolinsky , ann . phys .
( n.y . ) * 173 * 30 ( 87 ) such short time kernel may be thought to correspond to the conversion of presynaptic spikes into postsynaptic potentials .
a. treves , phys .
rev a * 42 * 2418 ( 90 ) w.b .
levy , r.a .
baxter , neural comp .
* 8 * 531 ( 96 ) ; r.j . baddeley
_ et al _ , proc .
( london ) * b 264 * 1775 ( 97 ) ; a. treves _ et al _ , neural comp .
* 11 * 611 ( 99 ) f.r .
et al _ , phys .
lett * 64 * 1986 ( 90 ) a. treves , j. phys .
a : math . gen . * 24 * 2645 ( 91 ) v. braitenberg , a. schtz , _ statistics of the cortex : anatomy and geometry _ ,
( springer , berlin , 1991 ) a. treves , e.t . rolls , hippocampus * 2 * 199 ( 92 ) m.v .
tsodyks , m.v .
feigelman , europhys .
* 6 * 101 ( 88 ) d.j .
amit , m.v .
tsodyks , network * 2 * 259 , 275 ( 91 ) b. derrida , e. gardner , a. zippelius , europhys .
lett . * 4 * 167 ( 87 ) a. treves , j. phys .
a : math . gen .
* 24 * 327 ( 91 ) j.j .
hopfield , d.w . tank , science * 233 * 625 ( 86 ) y. fu , p.w .
anderson , j. phys .
a * 19 * 1605 ( 86 ) | we show that symmetric @xmath0-mixture states , when they exist , are almost never stable in autoassociative networks with threshold - linear units . only with a binary coding scheme we could find a limited region of the parameter space in which either 2-mixtures or 3-mixtures are stable attractors of the dynamics .
2 [ ] |
the majority of young stellar objects ( ysos ) are classified using a scheme based on characteristics of their seds introduced by @xcite ( see also * ? ? ?
* ; * ? ? ?
they proposed three classes : i , ii , and iii ; distinguished from one another by their near - infrared to mid - infrared spectral indices , which were presumed to be indicators of the evolutionary status of the object in question .
class i sources were initially thought to be representative of the earliest stage of pre - main - sequence evolution , _ i.e. _ protostars .
the spectral index , @xmath0 , is defined as @xmath1 for purposes of yso classification , the region of the sed usually used for this calculation is from 2.2 @xmath2 m to 10 @xmath2 m .
the spectral index is the negative of the slope of the sed , when graphed in the frequency domain @xcite .
the most current version of this slope classification scheme is presented in @xcite . for a class i source , the sed tends to be broader than a single - temperature blackbody and @xmath3 , due to the fact that the peak of the sed of such a source is shifted towards the far - infrared because of the very low effective temperature of the yso .
these sources are usually deeply embedded in their nascent dust envelopes . for class ii and class iii ysos , on the other hand , @xmath4 and @xmath5 , respectively .
the sed of a class ii will be much broader than that of a class iii , due to the presence of circumstellar dust with a fairly wide temperature distribution .
@xcite identify a fourth class of `` flat spectrum '' sources , for which @xmath6 , that have near - infrared spectra which are strongly veiled by continuum emission from hot circumstellar dust .
observations of candidate protostars , in particular vla 1623 in @xmath7 ophiuchi a , an extremely cold ( @xmath8 20 k ) object that does not appear at _ iras _ wavelengths , led to the introduction of a new class of young stellar objects ( ysos ) by @xcite .
they contended that class i sources could not be the youngest examples of pre - main - sequence objects ( _ i.e. _ , not true protostars ) , and promptly created a new category for such objects , designated class 0 .
the search for these protostars has gained impetus in the last several years due to the development of highly sensitive submillimeter continuum detectors .
class 0 protostars are the rarest of ysos .
pre - main - sequence stars typically spend @xmath9 years in this phase , on their way towards the onset of hydrogen fusion and stellar birth .
the current set of defining characteristics of a class 0 protostar are as follows @xcite : * detection of a radio continuum source , presence of a molecular outflow , or some other evidence of a central protostellar source ; * centrally peaked but extended submillimeter continuum emission indicating an envelope as opposed to a mere disk ; * a ratio of l@xmath10/l@xmath11 @xmath12 0.5% ( where l@xmath10 is the luminosity radiated by the object at wavelengths longward of 350 @xmath2 m and l@xmath11 is the total luminosity of the object ) .
this last criterion indicates a circumstellar mass greater than the mass of the central protostellar core , as discussed in @xcite .
there is no mention of the spectral index between 2 and 10 @xmath2 m , since these sources have generally not been detected shortward of 10 @xmath2 m using ground - based instruments .
this will change , no doubt , as detector technology evolves .
although reports of newly discovered class 0 protostars have increased dramatically in the last few years , confirmed sources in this category still number well under a hundred .
@xcite . the spectral energy distribution ( sed ) of a class 0 object , with a typical effective temperature of @xmath1340 k , peaks at around 100 ; until fairly recently , essentially the only tool with which one may study the wien side of the seds of protostars has been the _ iras _ database .
the majority of known class 0 sources were detected by _ iras _ at 25 , if not at 12 ; however , due to the nature of the _ iras _ data , the effective beam size at mid - infrared wavelengths is never less than @xmath143040 , even when using the high - resolution data - processing program yoric @xcite .
figure [ fig : hires ] provides an example of coadded 60- _ iras _ data before ( contours ) and after hires processing ( greyscale ) via yoric .
the poor spatial resolution of the _ iras _ data makes associating point sources in the catalog with data from other wavelengths difficult and , in cases of close binary or multiple systems , does not allow the contributions of the individual components to the total flux to be distinguished .
hires point - source modeling , which makes use of a little - known feature of the yoric program , has been used successfully to estimate individual _ iras _ fluxes from sources as close as 20 apart , but this still does not compare with the sub - arcsecond resolution obtainable with millimeter - wave interferometry @xcite .
consequently , the mid - infrared spectral energy distributions of class 0 sources have tended to be poorly constrained .
data from the latest nasa great observatory , the _ spitzer space telescope _ are already contributing significantly to our understanding of the mid - ir characteristics of young stellar objects , but _
spitzer _ lacks the resolution to augment the information on the small - scale structure of their circumstellar environments which may be obtained from submillimeter and millimeter wavelength observations @xcite .
some of the most important questions in star formation studies today center around the multiplicity of these young systems , and how the binary properties evolve with time @xcite .
the majority of nearby young t tauri stars are known to be binary systems , yet only a very few class 0 protobinary systems have been resolved .
it has been postulated that the giant herbig - haro flows driven by many young stellar objects may be initiated by the dynamical decay of unstable multiple systems @xcite . understanding star formation in the context of multiple systems has important implications for understanding the imf and its physical meaning . in order to address the issue of multiplicity and obtain details about the immediate circumstellar environments of class 0 sources , it is necessary to study these objects at as many wavelengths as possible with high angular resolution . a particularly rich cluster of class 0 sources is located in the compact globule l1448 , a molecular cloud which is part of the perseus molecular complex .
l1448 lies at a distance of @xmath14300 pc ; its mass , estimated from ammonia data , is @xmath1450 m@xmath15 @xcite .
three embedded infrared sources were detected by _ iras _ in this cloud ; since then , it has been surveyed at several wavelengths and mapped in a number of molecular lines @xcite . listed from west to east , the _ iras _ point source catalog designations for the three sources are iras 03220 + 3035 , iras 03222 + 3034 , and iras 03225 + 3034 . in the literature ,
these objects are more commonly known as l1448 irs 1 , l1448 irs 2 , and l1448 irs 3 , respectively ( see figure [ fig : hires ] ) . only the western - most of these objects ( irs 1 ) has been detected at near - infrared and shorter wavelengths @xcite ; this source has been identified as a class i yso and hence was not included in our survey @xcite .
the other two _ iras _ point sources have been found to contain multiple class 0 objects .
irs 2 was recently reported by @xcite to be a candidate protobinary system , based on the two distinct molecular outflows which emerge from that position , although the binary components were unresolved at that time .
subsequently @xcite announced that both components of the irs 2 binary had in fact been detected at bima .
irs 3 consists of l1448n : a , b ( a 7 protobinary ) , l1448nw , and l1448c ( note : l1448c is frequently referred to as l1448-mm ) , all of which drive molecular outflows as well @xcite .
the locations of all of these sources , from previously published high - resolution submillimeter or millimeter - wave observations , are indicated by crosses in figure [ fig : hires ] ( * ? ? ?
* ; * ? ? ?
* and references therein ) .
we observed all of the known l1448 class 0 protostars in the mid - infrared , and present herein images of the detected sources plus multi - wavelength photometry and/or upper limits for all five class 0 sources .
low - mass class 0 objects are generally not visible at wavelengths shorter than 10 using current ( ground - based ) technology @xcite . the shortest wavelengths at which these sources are generally thought to be easily detectable are in the @xmath16 @xmath14 20 atmospheric window .
mirlin , jpl s 128x128 pixel si : as mid - infrared camera @xcite , provides access to this wavelength regime .
the observations reported here were carried out using mirlin at nasa s 3-m infrared telescope facility ( irtf ) , on the nights of 2000 feb 1819 ut and 2000 oct 5 ut .
for these observations , the measured pixel scale was @xmath17 , yielding a field of view just over 1 on a side . to remove the thermal background signal from the sky and the telescope ,
the secondary mirror was chopped 28 se nw at about a 5 hz rate , and the telescope nodded the same distance and direction every few hundred chop cycles . the l1448c source and
the l1448n / nw system were both observed at 12.5 , 17.9 , 20.8 , and 24.5 , while l1448 irs 2 was observed using only the latter two filters .
@xmath0 aur , @xmath0 boo , and @xmath0 cmi were observed throughout each night in february , in all four passbands , both to serve as photometric calibrators and to track the weather conditions and airmass dependence . in october
, @xmath0 lyr was used for the same purposes .
the weather was excellent in february but more variable in october .
the data were reduced using our in - house idl routine , mac ( match and combine ) , which performs the background subtraction , identification and shifting of the source peaks , and the coaddition of the registered data .
circular aperture photometry was then performed on both the standards and the l1448 sources using apertures of radius 5 pixels ( @xmath18 ) .
our observations of l1448 irs 3 are presented in figures [ fig : imc ] and [ fig : imn ] ; each of the four panels is labeled with the central wavelength of the filter used , and has an image of the psf at the appropriate wavelength shown in the lower left corner .
figure [ fig : imc ] shows our observations of l1448c which , at these resolutions ( @xmath19 , @xmath20 , @xmath21 , and @xmath22 , at 12.5 , 17.9 , 20.8 , and 24.5 , respectively ) , is indistinguishable from an isolated point source . figure [ fig : imn ] shows our images of the rest of the sources in irs 3 , all of which lie within a single mirlin field of view : the 7 binary , l1448n : a and b , and a third source @xmath1420away , l1448nw @xcite .
although there appears to be a hint of extended emission around l1448n : a in the 20.8 and 24.5 images , the extension is north - west / south - east , in good alignment with the direction of the chop , and therefore is most likely an artifact of the chop pattern . in february 2000
we detected a single source in the l1448n / nw field , at the location of l1448n : a ; this was initially somewhat surprising , since n : b is actually quite a bit brighter than n : a at millimeter wavelengths .
we therefore made additional observations of this field in october 2000 .
using the nearby bright star bs 999 as a position reference , we find that the single source is consistent only with l1448n : a ; for it to be n : b or nw would require telescope positioning errors of @xmath23 and @xmath24 ( 15 and 37 pixels ) , respectively . in figure
[ fig : imn ] , crosses show the positions at which n : b and nw should appear .
these results agree with those presented by @xcite for the l1448 n : a , b system .
figure [ fig : irs2 ] shows the mirlin image of l1448 irs 2 at 24.5 , along with the psf at that wavelength .
irs 2 was observed at 20.8 and 24.5 , but detected ( at 4@xmath25 ) only at the longer wavelength .
table 1 provides the coordinates of each source and presents the mirlin photometry and upper limits .
coordinates for the class 0 sources l1448n : a , n : b and nw were obtained from @xcite ; the positions of l1448c and l1448 irs 2 were taken from @xcite .
errors in the photometry are dominated by the calibration uncertainties ; even the most marginal detection , irs 2 , was 4@xmath25 .
for l1448n : a and l1448c , signal - to - noise ratios varied between 6 and 23 .
figures [ fig : sedc ] [ fig : sedi2 ] present sed plots for the three detected objects ( l1448c , l1448n : a , and l1448 irs 2 ) , as well as for the undetected sources , l1448n : b and l1448nw , constrained in the mid - infrared by the mirlin upper limits ( indicated by arrows ) .
l1448nw , which is shown in figure [ fig : sednw ] for completeness , was not detected and we were unable to improve on previous upper limits in the mid - infrared that were derived from hires point - source modeling @xcite .
previously reported data from the literature are plotted in combination with our mirlin fluxes .
the seds are shown with either single or dual - greybody fits to the data .
those sed figures with dual - greybody fits ( solid lines ) also display single - temperature greybodies as dashed lines , each greybody being a modified blackbody of the form : @xmath26 where @xmath27 is the planck function at frequency @xmath28 and dust temperature t@xmath29 , @xmath30 is the dust optical depth , and @xmath31 is the solid angle subtended by the source .
the sole exception to this convention is figure [ fig : sednab ] , which presents fits for the data from each component of the l1448n protobinary using a dashed line for n : a and a dotted line for n : b , while the solid line represents the sum of those two individual sed fits ; note that this particular pair of fits to the data is _ not _ the only possible solution , and this fact is discussed at length in [ sec : nab ] .
the various dual - greybody model fits to the data , shown by the solid lines in figures [ fig : sedc ] [ fig : sedi2 ] , were produced by simply summing together two greybody curves , each with a different effective dust temperature .
it is clearly not possible to fit all of the data points for any of the sources which were detected by mirlin in the mid - infrared with a single - temperature greybody .
the dual - greybody curves yield more accurate calculations of the bolometric luminosities .
the model fits to the flux data for each source and any necessary assumptions are discussed in detail in [ sec : disc ] .
table 2 lists each source , the number of greybodies ( n@xmath32 ) used to fit the flux data , the associated dust temperatures ( t@xmath33 ) for each greybody , and two physical parameters derived for these sources from the sed fitting : the bolometric luminosity , l@xmath11 , and the infall envelope mass , m@xmath34 , estimated from @xcite : @xmath35 evaluated at @xmath36 mm .
recall that dust emission is generally optically thin at millimeter wavelengths , and is therefore a direct indicator of the _ total _ mass ( gas @xmath37 dust ) contained in molecular cloud cores , given appropriate assumptions about the dust to gas ratio .
the appropriate value of @xmath38 to use depends on the status or `` age '' of the ysos in the core in question , due to the evolution of dust properties @xcite .
we used @xmath39 based on literature values for similar regions @xcite .
in addition , the ratio of the submillimeter luminosity to the bolometric luminosity , l@xmath10/l@xmath11 , where l@xmath10 is calculated using wavelengths longer than 350 , was determined for each source and included in table 2 .
this ratio can be used as a diagnostic tool to determine the evolutionary status of embedded ysos , because it is linked with the ratio of the circumstellar envelope mass to the mass of the central accreting core , and thus decreases with time .
a value of l@xmath10/l@xmath11 @xmath12 0.5% indicates m@xmath34/m@xmath40 @xmath12 1 , _
i.e. _ an envelope mass greater than the protostellar core mass , a distinguishing characteristic of class 0 sources ( see [ sec : intro ] ) .
this class 0 source , which drives a powerful , highly - collimated outflow , is a favorite target of many observers ( see * ? ? ? * and references therein ) .
when first discovered , it was believed to be one of the youngest class 0 sources found , with a kinematic age ( based on the outflow parameters ) of @xmath143500 years @xcite .
this figure was revised upwards to @xmath1432,000 years by @xcite , who found that the outflow was far more extensive than previously thought ( see also * ? ? ?
the sed for l1448c is presented in figure [ fig : sedc ] , with a two - greybody fit to the data shown by the solid line .
the dashed line indicates a single - temperature greybody fit to the data . in order to analyze the seds for l1448n :
a and n : b , and use them to derive the parameters listed in table 2 , it is necessary to make certain assumptions regarding the individual fluxes of these two objects at far - infrared wavelengths .
there are no high - angular - resolution data that can separate the 7 protobinary components at or near the peak of the dust emission ( @xmath14100 ) , and this causes problems when attempting to fit modified blackbodies to the spectral energy distributions of these objects .
due to the excellent correspondence between the 12 and 25 _ iras _ hires point - source model fluxes from @xcite for l1448n ( which represent the sums of the fluxes for both components of the protobinary ) and our mirlin fluxes ( see figure [ fig : sednab ] ) , it seems clear that the mid - infrared fluxes are due chiefly to l1448n : a , with very little , if any , contribution from n : b or nw .
again , these results agree with those presented in @xcite . in the millimeter
, the situation is quite different : n : b produces most of the flux seen at 2.7 mm @xcite .
using hires modeling , @xcite showed it is possible to separate the _ iras _ fluxes produced by l1448nw from the total flux emitted by the l1448n triple system .
the l1448n system produces a total of 89 jy at 100 , while the hires - modeled 100 flux from nw is @xmath1423 jy .
therefore the _ maximum _ possible flux from the protobinary at 100 must be on the order of 66 jy . since some of the emission detected by _
iras _ at that wavelength is presumably due to shock - heated dust caused by the impact of the l1448c outflow on the dense core containing the l1448n protobinary , the true 100 flux from the binary is probably somewhat less than 66 jy @xcite .
the flux ratio is approximately 5:1 in favor of n : b in the millimeter regime @xcite , therefore it seems reasonable to explore a range of possibilities for the 100 fluxes .
we have done so , in the following fashion : ( 1 ) assume that the fluxes at 100 have the same ratio as the 2.7 mm fluxes , so that n : a produces 11 jy and n : b emits the remaining 55 jy ; ( 2 ) assume the inverse scenario ( _ i.e. _ , n : b has a flux of 11 jy , n : a produces the other 55 jy ) .
these two possibilities define the endpoints of the range that we consider in this paper . a third instructive case to consider
is the intermediate scenario which involves dividing the 66 jy into equal portions ; this assumes fluxes of @xmath1433 jy at 100 from each of the protobinary components .
the assumptions inherent in this range of possibilities were used to generate a series of different seds for n : a and n : b ; each of the seds for n : a were then fit with two - greybody models , while those for n : b were done with single - temperature greybodies .
sanity checks were performed for each case by using the various fits to calculate ranges of l@xmath11 for the two sources , and thus to ascertain whether or not the sum of their luminosities approximates the total luminosity expected for the protobinary .
the luminosity of the protobinary was derived by taking the total fluxes of the l1448n system reported in the literature and subtracting the flux data for l1448nw at all wavelengths , then plotting the resultant sed and fitting with the sum of two modified blackbodies ; this yields a maximum expected l@xmath11 of @xmath1414.1 l@xmath15 for the protobinary system .
the derivation of the l1448nw luminosity , using a single greybody fit to the data , is discussed in [ sec : nw ] . from the analysis described above , we find the plausible range of l@xmath11 for l1448n : a is 3.16.6 l@xmath15 , and for n : b is 2.87.4 l@xmath15 . since n
: b dominates in the millimeter , and n : a in the mid - infrared , and the two objects have bolometric luminosities of the same order of magnitude , their seds must cross each other somewhere in the far - infrared , probably in the general vicinity of 100 .
the scenario of approximately equal 100 fluxes thus seems the most illustrative ; it yields bolometric luminosities of @xmath144.8 l@xmath15 for n : a , and 5.5 l@xmath15 for n : b . the sum is indeed less than the maximum expected luminosity .
figure [ fig : sednab ] is a plot of the literature and mirlin flux data for each protobinary component along with a two - greybody fit for l1448n : a ( dashed line ) and a single - temperature greybody fit for n : b ( dotted line ) using the `` equal 100 fluxes '' scenario .
the previously published data points and derived fluxes for the entire l1448n protobinary system are plotted , and the sum of the fits to the individual seds ( i.e. , dashed + dotted ) is shown as a solid line .
the values of the physical parameters derived from the assumption of approximately equal 100 fluxes ( case 3 ) are reported in table 2 .
l1448nw was not detected by mirlin at any wavelength , as may be seen in figure [ fig : imn ] .
this was the expected result , due to the extensive hires point - source modeling work reported in @xcite , which established _ iras _ upper limits for this low - luminosity source of 0.015 and 0.05 jy at 12 and 25 , respectively .
our mirlin upper limits do not improve on these results ( see table 1 ) .
we have incorporated the submillimeter and millimeter - wave flux data from @xcite and @xcite in the plot shown in figure [ fig : sednw ] ; these data are entirely consistent with the fir _ iras _ fluxes derived from hires modeling @xcite .
the results from a single - temperature greybody fit to the data at wavelengths longer than 60 are used to calculate the parameters reported in table 2 , which allow us to place l1448nw on the evolutionary diagram in figure [ fig : evo ] .
l1448 irs 2 was identified as a class 0 source by @xcite , and first proposed as a candidate protobinary in @xcite , based on co observations which mapped two distinct molecular outflows driven by irs 2 .
@xcite claimed detection of both binary components using bima , although the spatial separation was not reported . with the high spatial resolution of mirlin at the irtf ( @xmath412 at all wavelengths ) , we had hoped to be able to detect and resolve both components of the system , but only a single source was found , at the 4@xmath25 level in the 24.5 filter ( see figure [ fig : irs2 ] ) .
this suggests that either the binary components are too close together for mirlin to resolve ( see the first paragraph of [ sec : res ] for mirlin resolutions ) , or that the fluxes from one of the components are below the detection thresholds at all mirlin wavebands .
figure [ fig : sedi2 ] presents the sed for irs 2 , with a two - greybody fit to the data shown by the solid line .
the dashed line indicates a single - temperature greybody fit to the data .
figure [ fig : evo ] is a plot of m@xmath34 vs. l@xmath11 for all of the sources in this paper ( filled squares ) , along with all other class 0 ( open diamonds ) and class i ysos ( filled triangles ) with l@xmath42 from table 1 in @xcite plotted for comparison .
this plot may be used as a diagnostic tool for determining the relative ages of embedded sources @xcite .
the diagram shows infall envelope mass ( along the vertical axis ) vs. bolometric luminosity ( horizontal axis ) .
l@xmath11 is directly related to the mass of the central protostellar core at very early stages of evolution , assuming all luminosity is generated by gravitational infall , according to : @xmath43 where a protostellar radius of @xmath143 r@xmath15 is assumed @xcite .
our results depicted in figure [ fig : evo ] for l1448c and l1448nw are consistent ( within errors ) with those presented in @xcite for those two sources .
the new results in the diagram presented here are ( 1 ) l1448 irs 2 is plotted on the diagram to allow comparison with the other sources from this cloud , and ( 2 ) we have plotted the l1448n protobinary components _ separately _ on this graph , using the assumptions discussed in [ sec : nab ] .
all of the target objects discussed in this paper lie within the class 0 region of the evolutionary diagram in figure [ fig : evo ] , although l1448n : a appears to be close to the transition zone between class 0 and class i , indicating n : a may be the most evolved source in this cloud after the class i source l1448 irs 1 . by the same criteria ,
l1448n : b is very likely the youngest source that we attempted to image in this cloud .
these classifications are reflected in the values of the parameter l@xmath10/l@xmath11 , listed in table 2 , which range from 1.77.3% : the value of 1.7% corresponds to l1448n : a , and 7.3% to n : b ( for the case 3 scenario of equal 100 fluxes ) .
we note that recently @xcite have argued that l1448n : a is probably a class i object . however , our smallest value of l@xmath10/l@xmath11 for n : a , derived for case 2 ( assuming 100 flux for n : a @xmath44 jy , for n : b @xmath45 jy ) , was 1.5% , which lies well above the lower - limit value of 0.5% established for class 0 objects by @xcite .
we point out two additional factors which bolster the argument for the class 0 status of n : a : ( 1 ) its molecular outflow is extremely well - collimated , similar to outflows associated with other class 0 sources @xcite .
( 2 ) we find that it is not detected in the near - infrared by 2mass , although the class i source l1448 irs 1 is strongly detected in all 2mass bands ( jhk@xmath46 ) .
although fluxes are presented in all bands for a `` source '' detected by 2mass within a few arcseconds of l1448n : a , the j - band flux is presented as an upper limit , and the h - band and k - band fluxes suffer photometric confusion .
the 2mass quicklook images reveal a bow - shaped knot north of l1448n : a in all three bands , and an additional jet - like feature south of l1448n : a in the k - band image .
inspection of these features reveals that they correspond to h@xmath47 outflow features imaged by @xcite . the jet - like feature in the k - band image of l1448n : a is part of the blueshifted outflow lobe associated with l1448c , which is located @xmath48 to the southeast of the l1448n sources .
this outflow was also identified in hires - processed _ iras _ images and
high - velocity co emission @xcite the bow - shaped knot that appears just north of l1448n : a in the j , h , & k@xmath46 2mass images is very prominent in the h@xmath47 image .
there is no trace of a near - infrared point source counterpart to l1448n : a in the 2mass data .
additional near - infrared data for this source became available during a late revision of this paper : a flamingos survey of l1448 down to @xmath49 17 mag @xcite .
again , no near - ir counterpart to l1448n : a was found during this survey . however , these same investigators also used _
chandra _ to observe l1448 ; they detected a weak x - ray source in the general vicinity of n : a .
no x - ray emission was detected from any of the other class 0 sources under discussion in this paper . @xcite
discuss the fact that , given the non - detection in the near - infrared bands , l1448n : a must have an extremely steep near - to - mid - infrared spectral index ( @xmath50 ) , and that nearly all known class i protostars have @xmath51 ( the single exception being the peculiar young stellar object , wl 22 , with @xmath52 ) .
we therefore contend that all currently available data for l1448n : a indicate that it is a bonafide ( albeit borderline ) class 0 source ; the weak x - ray detection does not rule this out , and all other data support this conclusion . using ammonia data , @xcite report that the high - velocity l1448c outflow has had a significant affect on the northern core , creating a deep `` dent '' in its southern edge , directly south of the protobinary .
@xcite suggest that the impact of the outflow from l1448c on the dense core fragmented it and actually induced the formation of the l1448n / nw system . at first glance
, this scenario of outflow - induced fragmentation appears to be at odds with our findings above that n : a is the most evolved class 0 source in this cloud , while n : b is the youngest and most deeply embedded . of course , if these two sources do not form a gravitationally - bound protobinary system , it is unnecessary for them to be coeval .
but for the purposes of this discussion , we will assume with @xcite , @xcite , @xcite , and @xcite , that n : a and n : b are the bound components of a protobinary .
is it possible for presumably coeval sources , which appear to be physically similar , to travel different evolutionary paths ?
one recently - proposed explanation for such a dichotemy in evolutionary status of protobinary components is found in the disintegration of unstable triple or higher - order multiple systems ; such events may cause disk truncation for some of the components , subsequent episodic outflow activity linked to giant herbig - haro flows , and accelerated evolution of one or more components due to the rapid dissipation of circumstellar envelopes @xcite .
it is true that most of the young sources in l1448 have been identified as driving sources of parsec - scale herbig - haro flows @xcite .
however , at @xmath142000 au separation , with a rotational period on the order of 60,000 years @xcite , the l1448n : a , b system does not seem to fit into this particular theoretical paradigm , since a remnant binary from such a former higher - multiplicity system is predicted to have a much smaller separation .
the l1448n : a , b components appear to be too young to have had time to interact with each other or with l1448nw in any significant way . furthermore , according to the @xcite theory , binary components that are remnants of unstable multiples having undergone ejection events might be expected to have more pronounced differences between their apparent evolutionary stages ; _
e.g. _ , one component may transition abruptly to class ii status , while the other remains in class 0 or class i. all of the l1448 sources under discussion in this paper still fall within the same evolutionary category , although n : a appears to be close to transitioning to class i. we propose an alternative explanation for the ( admittedly minor ) difference between the protobinary components : this could be nothing more than a chance side effect of the powerful l1448c outflow . lacking a good understanding of the three - dimensional relative geometry of the sources in this cloud , we are unable to verify our hypothesis , but one possible scenario is that the outflow from l1448c has partially stripped the envelope surrounding n : a , while leaving n : b deeply embedded .
this would allow n : a and n : b to be coeval sources that formed due to the impact of the l1448c outflow , while explaining the different _ apparent _ ages of these objects .
we have conducted a mid - infrared imaging survey of known class 0 protostars in l1448 with mirlin at the irtf , using the 12.5 , 17.9 , 20.8 , and 24.5 filters .
of the five confirmed class 0 sources in this cluster , we imaged and obtained photometry for three , including a single source that was marginally detected at 24.5 at the location of the protobinary in l1448 irs 2 .
we report upper limits for two undetected objects .
seds are presented for all five sources , and are used to derive various physical properties of these sources , allowing us to place them on an evolutionary status diagram .
the components of the protobinary system l1448n are plotted separately ; we find that the source known as l1448n : a lies significantly closer to the class i portion of the diagram than does its more deeply embedded companion , n : b .
one possible explanation for this difference in apparent `` age '' may be found in the `` stripping '' of the envelope surrounding n : a by the l1448c outflow , which is known to be colliding with the northern core .
we thank the irtf support staff for assistance with the data acquisition .
jo acknowledges financial support by nasa grants to the wide - field infrared explorer project and the space infrared telescope facility project at the jet propulsion laboratory , california institute of technology .
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 / california institute of technology , funded by the national aeronautics and space administration and the national science foundation .
this research has made use 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 .
portions of this work were carried out at the jet propulsion laboratory , california institute of technology , under contract with the national aeronautics and space administration .
development of mirlin was supported by the jpl director s discretionary fund and its continued operation is funded by an sr+t award from nasa s office of space science . for a discussion ) .
some of the parameters of the single- and dual - greybody fits are presented in table 2 .
the solid line represents the sum of the fits to the individual flux data points ( dashed + dotted ) .
the filled triangles represent _ iras _ photometry from hires point - source modeling @xcite ; the open triangles with error bars and the upper limits are from mirlin observations ( this work ) .
filled squares indicate jcmt data from @xcite , the plus represents iram data from @xcite .
the `` x ' 's indicate fluxes from @xcite .
the open diamonds are from @xcite , while the filled diamond is the sum of those fluxes .
[ fig : sednab],width=566 ] vs. l@xmath11 : * we plot the positions of the l1448 protostars ( filled squares ) along with a number of other well - known class 0 ( open diamonds ) and class i sources ( filled triangles ) from @xcite and @xcite in the circumstellar mass vs. bolometric luminosity diagram .
since bolometric luminosity is an indirect measure of central core mass , this is essentially a plot of envelope mass vs. central core mass .
the `` error bars '' on l1448n : a and n : b represent the full range of likely values for each object ( all three cases , with the filled square being case 3 , see [ sec : nab ] ) .
generally , sources to the upper left of the diagram are in the class 0 region , while sources to the lower right are class i objects.[fig : evo],width=566 ] lrclrclccccc l1448 irs 2 & 03 & 25 & 22.4 & @xmath3730 & 45 & 12 & & & @xmath410.46 & [email protected] + l1448nw & 03 & 25 & 35.653 & @xmath3730 & 45 & 34.20 & @xmath410.15 & @xmath410.30 & @xmath410.27 & @xmath411.85 + l1448n : b & 03 & 25 & 36.339 & @xmath3730 & 45 & 14.94 & @xmath410.15 & @xmath410.30 & @xmath410.27 & @xmath411.85 + l1448n : a & 03 & 25 & 36.532 & @xmath3730 & 45 & 21.35 & [email protected] & [email protected] & [email protected] & [email protected] + l1448c & 03 & 25 & 38.8 & @xmath3730 & 44 & 05.0 & [email protected] & [email protected] & [email protected] & [email protected] + [ tab : phot ] lccccccc l1448 irs 2 & 2 & 20,50 & 5.6 & 0.73 & 3.3 + l1448nw & 1 & 20 & 3.5 & 0.68 & 5.0 + l1448n : b & 1 & 17.5 & 5.5 & 1.75 & 7.3 + l1448n : a & 2 & 23,90 & 4.8 & 0.29 & 1.7 + l1448c & 2 & 19,85 & 10.0 & 1.10 & 3.7 + [ tab : phys ] | we present ground - based mid - infrared observations of class 0 protostars in ldn 1448 . of the five known protostars in this cloud , we detected two , l1448n : a and l1448c , at 12.5 , 17.9 , 20.8 , and 24.5 , and a third , l1448 irs 2 , at 24.5 .
we present high - resolution images of the detected sources , and photometry or upper limits for all five class 0 sources in this cloud . with these data ,
we are able to augment existing spectral energy distributions ( seds ) for all five objects and place them on an evolutionary status diagram . |
financial market modelling lies in the field of interests of both theoreticians and practitioners . among those who develop these models are also econophysicisits . a view on a financial market as a complex system of investors similar to complex physical systems appeared to be very fruitful , since it allowed to reproduce many characteristic features of such a market @xcite .
a special group constitute approaches based on the ising model @xcite or its generalization to a three - state model @xcite .
they identify agents with spin variables which can take specific values depending on agents decisions . in much the same way as ising spins ,
agents interact with each other , which leads to a herding behavior and , as a consequence , bubbles or crashes .
an important target for such models is a reproduction of real market stylized facts such as fat - tailed returns distribution , clustered volatility , or long range correlations of absolute returns @xcite . in this paper
we proposed a simple model of financial markets , based on the granovetter threshold model of collective behavior , which is in a good agreement with the above facts .
an inspiration for our work was the bornholdt model @xcite , corresponding to the ising spin model with an additional _ minority term _ , where agents act under the influence of their neighbors ( ising part ) and a global magnetization .
the local field for i - th spin in the bornholdt model is defined as : @xmath0 with a global constant @xmath1 .
the model interprets the price of an asset in terms of the magnetization , which enables the authors to reproduce some stylized facts of financial markets .
here we developed a generalization of the ising spin model that also uses the absolute value of the magnetization as a factor controlling the dynamics .
let us consider a model of n interacting market agents where each agent takes one of three actions : `` buy , '' `` sell , ''
or `` stay inactive . ''
such an agent can be represented by a three - state spin variable @xmath2 taking values @xmath3 when the agent is buying , @xmath4 when the agent is selling , and @xmath5 in the remaining case .
the agents interact with each other according to an interaction matrix @xmath6 . in our model
we assumed the interaction matrix corresponding to a 2-dimensional squared lattice in which each agent interacts only with its four nearest neighbors with an equal strength @xmath7 .
the interaction strength can be _
ferromagnetic _ ( @xmath8 ) when investors try to act like their neighbors or _ antiferromagnetic _
( @xmath9 ) when they try to play against their neighbors .
we chose the ferromagnetic case to introduce the herding behavior . at each time step
@xmath10 the agent @xmath11 takes its value according to the following formula , @xmath12,\ ] ] where @xmath13 is a threshold signum function , @xmath14 the function @xmath15 has random values from the gaussian distribution with @xmath5 mean and variance equal to @xmath16 .
the term @xmath17 simulates an individual erratic opinion of the i - th investor and the parameter @xmath18 defines strength of individual opinions .
we define the magnetization of the network , @xmath19 which together with a constant @xmath20 forms a threshold parameter @xmath21 of the sign function .
let us notice that for @xmath22 our model is identical with the 2-valued ising spin model .
we simplified the procedure of price calculation presented in @xcite by omitting the influence of _ chartists _ and _ fundamentalists _ in the population of investors and redefining the price of an asset as : @xmath23 where @xmath24 is , in agreement with the efficient market hypothesis @xcite , a geometric brownian walk corresponding to fundamental price changes .
let us put @xmath25 constant for more simplicity .
we got therefore a logarithmic rate of return : @xmath26 according to ( 2 ) , each agent is under the influence of three factors .
the first is an imitation of their neighbors , associated with the matrix @xmath6 which is responsible for the herding behavior .
the second factor is an individual opinion of the agent provided by the term @xmath17 .
so far , it is the standard ising model .
yet , there is one more factor , @xmath27 , which plays the role of a threshold parameter .
only those agents that are able to exceed the threshold are allowed to trade .
the value of the threshold depends on the absolute magnetization . according to ( 5 ) , the magnetization measures a deviation from the fundamental value .
so , when it is large , the agents are afraid of trading , unless they have a strong support from the neighbors or from their private opinions .
using a square @xmath28 lattice of agents with periodic boundary conditions we computed the history of the magnetization , and thus the evolution of the stock price .
the agents were allowed to interact only with their nearest neighbors .
the simulation was started with a random configuration of spins . at each time
step a randomly chosen spin was updated according to the evolution equation ( 2 ) and this was repeated @xmath29 times .
the first @xmath30 time steps were ignored as a period of system thermalization .
we observed that for proper parameters price returns show volatility clustering ( fig . 1 ) and fat tails ( fig .
2 ) . in time
@xmath10 for parameters @xmath31 , @xmath32 and different @xmath20 . ] [ cols="^,^ " , ] for real prices , fat tails of returns distribution ( @xmath33 ) are getting slimmer with rising time lag @xmath34 .
finally , the distribution became normal for sufficiently large @xmath34 . in figure 3
we presented the distribution @xmath35 for different @xmath34 .
the tails are slimming down with rising @xmath34 .
the distribution has gaussian tails although its shape is slightly different .
volatility clustering can be quantitatively shown using the autocorrelation function of a time series @xmath36 : @xmath37 where @xmath38 is the variance of @xmath36 , and @xmath39 means the average over @xmath10 .
we computed the autocorrelation function of returns @xmath40 ( fig .
4 ) and the absolute value of returns @xmath41 ( fig . 5 ) .
the autocorrelation of returns decays very fast , which is consistent with observations of real markets .
it has also the same shape as the distribution presented in [ 10 ] .
the autocorrelation function of the absolute returns is a slowly decaying exponential similar to the results of [ 13 ] and [ 10 ] .
, obtained for parameters @xmath31 , @xmath32 , @xmath42 . ] , obtained for parameters @xmath31 , @xmath32 , @xmath42 in log - log scale , and in semi - log scale ( inset ) . ]
we proposed a simple model of financial markets based on the granovetter model .
we introduced a threshold controlled by a magnetization of the system .
this makes an agent take an action only if its confidence is strong enough to overwhelm the threshold .
we defined a logarithmic rate of return as a change of the magnetization .
the model with such a defined price reproduces main stylized facts of financial markets meaning a fat - tailed distribution of returns , volatility clustering , very fast decaying autocorrelation of returns and much slower decay of autocorrelation of absolute returns .
it has been observed for real markets that the distribution of returns @xmath35 becomes gaussian for large @xmath34 .
the distribution of returns generated by our model approaches the gaussian distribution for rising @xmath34 , however its shape differs from the normal distribution .
the autocorrelation function of the absolute returns is a slowly decaying exponential function , while the empirical study of the markets reveals a power - law behavior .
both these issues can be considered as weaker points of our model .
we would like to dedicate this paper to prof .
marcel ausloos and to prof .
dietrich stauffer on the occasion of their 65th birthdays and to thank them for their inspiring contributions to econophysics and sociophysics research .
+ the work has been supported by the project stochdyn by european science foundation and by polish ministry of science and higher education ( grant esf/275/2006 )
. 1 r. cont , j .-
bouchaud , macroecon .
dynamics 4 , 170 ( 2000 ) , t. lux , m. marchesi , nature 397 , 498 ( 1999 ) , d. chowdhury , d. stauffer , eur .
j. b 8 , 447 ( 1999 ) , i. giardina , j .- p .
bouchaud , phys . a 324 , 6 ( 2003 ) , s. bornholdt , int . j. mod .
c 12 , 667 ( 2001 ) , t. kaizoji , phys . a 287 , 493 ( 2000 ) , a. krawiecki , j. a. hoyst , d. helbing , phys .
89 , 158701 ( 2002 ) , w .- x .
zhou , d. sornette , eur .
j. b 55 , 175 ( 2007 ) , g. iori , int . j. mod
c 10 , 1149 ( 1999 ) , t. takaishi , int . j. mod
c 16 , 1311 ( 2005 ) , a - h .
sato , phys .
a 382 , 258 ( 2007 ) , r.n .
mantegna , h.e .
stanley , _ an introduction to econophysics _ , cup , cambridge , uk 2000 , t. kaizoji , s. bornholdt , y. fujiwara , phys .
a 316 , 441 ( 2002 ) , e. f. fama , j. finance 25 , 383 ( 1970 ) . | we proposed a model of interacting market agents based on the generalized ising spin model .
the agents can take three actions : `` buy , '' `` sell , '' or `` stay inactive . ''
we defined a price evolution in terms of the system magnetization .
the model reproduces main stylized facts of real markets such as : fat - tailed distribution of returns and volatility clustering .
_ faculty of physics , center of excellence for complex systems research , warsaw university of technology , koszykowa 75 , + pl-00 - 662 warsaw , poland _ * pacs . * 89.65.gh |
when one considers a universe following the standard einstein cosmology , that is , when the dynamics of the universe is described by the general relativity equations , one deduces that there exists a mysterious singularity at the beginning of time , so called big bang .
this singularity may be considered as a deficiency of einstein cosmology at high energies @xcite .
actually , one can assert that the big bang implies the breakdown of general relativity at scales with high energies , whereas we know from the observational evidences , such as the existence of the cosmic microwave background , that the big bang model works at scales lower than planck energy scale . at these scales ,
the universe was full of a hot photon - baryon combination almost described by a radiation fluid .
this hot combination was cooling down as the universe was experiencing the expansion , and due to the different scaling behaviors of pure radiation and pure matter , the energy density of non - relativistic matter started to dominate over the energy density of radiation and led to the formation of structures .
loop quantum gravity ( lqg ) is one of the potential candidates for the study of quantum gravity @xcite .
homogenous and isotropic space time , reduces loop quantum gravity ( lqg ) to loop quantum cosmology ( lqc ) @xcite . according to this theory , due to the effective quantum gravitational effects
, the big bounce happens to remove the big bang singularity , after which a super - inflation phase occurs , and then the universe enters a normal inflation regime @xcite , @xcite .
the loop quantum cosmology effects manifest themselves in a new effective form of modified friedmann equation .
these effective equations have been derived in the literature for flrw loop quantum cosmologies considering different matter sectors at early universe @xcite .
an extended scenario of the matter bounce cosmology , considering a single scalar field with an approximately exponential potential , was proposed as an alternative to the slow roll inflation , in the context of ( lqc ) , in which the universe has experienced a quasi - matter contracting phase with a variable equation of state parameter @xcite .
this matter bounce scenario , in the teleparallel version of ( lqc ) , was shown to be in good agreement with the new bicep2 data @xcite .
a gauss - bonnet extension of ( lqc ) , by introducing holonomy corrections in modified @xmath0 theories of gravity was also developed where the authors have provided a perturbative expansion in the critical density as well as a parameter characteristic of ( lqg ) , and obtained leading order corrections to the classical @xmath0 theories of gravity .
they also presented a reconstruction method which makes possible to find the ( lqc ) corrected @xmath0 theory capable of realizing various cosmological scenarios @xcite . in another work , in order to avoid singularities of @xmath1 model , holonomy corrections to this model were introduced in einstein frame and a detailed analytical and numerical study was performed when holonomy corrections are taken into account in both jordan and einstein frames .
they obtained , in jordan frame , a dynamics which is different qualitatively , from the one of the original model , at early times . according to this dynamics
, the universe is not singular , neither at early times in the contracting phase , nor at bouncing to enter in the new expanding inflationary phase .
this dynamics may lead to better predictions for the inflationary phase in comparison to the current observations @xcite .
motivated by the above mentioned success of ( lqc ) in describing the very early stage of inflationary universe , we are encouraged to study the evolution of universe after the inflationary era and investigate the possible establishment of a specific kind of dark matter ( see below ) and study its contribution in the subsequent evolution of the universe .
explicitly , we assume that after inflation the universe includes ordinary dark matter ( bosonic particles ) , dark energy , radiation and baryonic matter , and that the ordinary dark matter component in the sufficiently cooled universe has gone through a phase transition provided by the bose - einstein condensation , which might have happened during the early stages of cosmological evolution of the universe with a low temperature comparable to the critical temperature for bose - einstein condensation @xmath2 where @xmath3 is the particle mass , @xmath4 is the particle density and @xmath5 is boltzmann s constant @xcite-@xcite .
particularly , in the bose - einstein condensation ( bec ) model , the dark matter can be described as a non - relativistic newtonian gravitational condensate , whose pressure and energy density are related by a barotropic equation of state .
based on the above arguments , we assume the possibility that the bose - einstein condensation might have happened during the early stages of cosmological evolution of the universe with a temperature comparable to the critical temperature for bose - einstein condensation @xmath2 where @xmath3 is the particle mass , @xmath4 is the particle density and @xmath5 is boltzmann s constant @xcite-@xcite .
the modification of cosmological evolution of the early universe , in the presence of condensate dark matter in standard cosmology , has already been studied in @xcite . in this paper
, we generalize this model from standard cosmology to lqc .
the organization of the present paper is as follows . in section
@xmath6 we outline the modification of friedmann equations in lqc model and the basic properties of the normal and bec dark matter in addition to the presentation of the relevant physical quantities and the equation of state . in section
@xmath7 we give a brief explanation about the cosmological dynamics of the bose - einstein condensate .
next , we introduce the numerical values of the cosmological parameters at the condensation point , and then obtain the important @xmath8 parameter as a volume fraction of matter in the bose - einstein condensed phase which explains the evolution of the dark matter energy density during the transition process , and extract the order of bec time interval in lqc model . in the following , we study the post condensation phase in lqc model describing the evolution of the scale factor parameter under the effect of the existence of the bec dark matter in lqc .
the paper ends with a conclusion .
in this section , we intend to review the bose - einstein condensation and explore the dynamics of the condensation and some of the relevant cosmological implications .
we consider the line element of the flat friedmann - robertson walker metric @xmath9 where @xmath10 is the scale factor presenting the cosmological expansion .
we describe the matter by the perfect fluid energy momentum tensor , @xmath11 where ( @xmath12 , @xmath13 ) stands for ( @xmath14 , @xmath15 ) , ( @xmath16 , @xmath17 ) , and ( @xmath18 , @xmath19 ) for baryonic matter , radiation and dark matter , respectively .
note that we do not consider the interaction between these energy terms , in other words the energy of these components are individually satisfying the conservation equation .
+ + taking into account the well - known loop quantum gravity constraints and coupling the matter to classical phase space of flrw universe , we obtain @xcite @xmath20 @xmath21 where @xmath22 is the matter hamiltonian , @xmath23@xcite represents the barbero - immirzi parameter and also @xmath24 is the length gap .
meanwhile , @xmath25 is the conjugate momentum and @xmath26 is proportional to the physical volume of a cubical cell , with unit comoving volume @xmath27 .
it is noticeable that dot means derivative with respect to the time . by rewriting the equations ( [ eq4 ] ) and ( [ eq5 ] )
, we have @xmath28 @xmath29 where @xmath30 is the hubble parameter and the matter density is limited as @xmath31 where @xmath32 .
combination of ( [ eq6 ] ) and ( [ eq7 ] ) reduces to the modified flrw equation in loop quantum cosmology model @xcite @xmath33 as we mentioned , the effective equations have been derived for flrw loop quantum cosmologies considering different matter sectors at early universe
so , we may use the modified flrw equation in loop quantum cosmology model , namely eq.([eq8 ] ) , and assume the matter density @xmath12 as the combination of different matter components @xmath14 , @xmath16 , @xmath18 , and @xmath34 indicating the baryonic matter , radiation , dark matter and cosmological constant respectively , so that @xmath35 the second modified flrw equation is also given by @xmath36 and also for the energy density conservation equation we have @xmath37 now , regarding the cosmological evolution of radiation and baryonic matter we consider the relations @xmath38 and @xmath39 , respectively ; meanwhile @xmath40 and @xmath41 are the energy densities corresponding to their values at @xmath42 , and also for the dark matter we work with a general form of density as @xmath43 . it should be noted that @xmath44 is an arbitrary function which depends on the special dark matter model . as we know
we have the critical density as @xmath45 , where @xmath46 is the value of the hubble parameter at @xmath42 .
in addition , we can write dimensionless parameters known as density parameters as @xmath47 , with @xmath48 . using these relations helps us to write the following new form of modified flrw equation @xmath49 where @xmath50 is the dark energy density parameter .
thus , we can write the constraint @xmath51 now , we suppose that at early universe the bosonic particles , with mass @xmath52 and a very high temperature @xmath53 , were in equilibrium state with other particles in a hot relativistic plasma . after the expansion of universe and rapid decrease of the temperature it was decoupled from this equilibrium state at a high chemical potential @xmath54 or low decoupling temperature @xmath55 . at this decoupling temperature , we assume that the bosonic particles are still in kinetic equilibrium among themselves in an isotropic gas with low temperature . consequently ,
for the spatial number density we have @xmath56 where @xmath57 is the helicity states number , @xmath58 is planck s constant , and @xmath59 is defined as follows @xmath60 - 1\right)},\end{aligned}\ ] ] where @xmath61 is the energy @xmath62 and @xmath13 is the momentum of the particle . due to the red - shift in the momentum of particles , we have @xmath63 where @xmath64 and @xmath65 denote for the momentum of particles and the scale factor of universe , respectively , at the temperature @xmath66 denoting for the decoupling temperature from the rest of plasma . besides , we have the scaling evolution relation for the number density given by @xmath67 @xcite . for the extreme - relativistic case @xmath68 ( @xmath69 and @xmath70 ) , the distribution function
is derived as @xmath71 - 1\right)^{-1}$ ] , hence for the nonrelativistic case @xmath72 ( @xmath73 ) the distribution function is given by @xmath74 - 1\right)^{-1}$ ] , where the evolution equations are as @xmath75 and @xmath76 @xcite . for a frozen distribution of dark matter ,
the associated energy density @xmath77 and kinetic energy - momentum tensor @xmath78 are as follows , respectively @xmath79 @xmath80 the mentioned pressure is described by @xmath81 where we have used @xmath82 @xcite . the density of dark matter @xmath18 for the nonrelativistic case with @xmath83 and @xmath84 is given by @xmath85 whereas the dark matter pressure is given by @xmath86 which leads to @xmath87 where we have used @xmath88 , @xmath89 being the one - dimensional velocity dispersion . by means of the dark matter conservation equation @xmath90 we can find the following general solution @xmath91 where @xmath92 is the density of dark matter at the present value of scale factor @xmath42 . in the standard model , for the description of the dynamics of our universe with a normal dark matter we work with the following equation@xmath93.\ ] ] however , in lqc model we will have the following modification @xmath94 since we aim to extract numerical results for dynamical bec quantities , we have to mention the values of several parameters and quantities such as the hubble constant @xmath95 @xmath96 @xmath97 , the hubble time @xmath98 @xmath99 , the critical density @xmath100 @xmath101 , and the present day dark matter densities @xmath102 , @xmath103 , @xmath104 , @xmath105 respectively @xcite .
it is worth to remind that the following dark matter case is non - relativistic so the global cosmological evolution of the universe will not be affected by the variation of the numerical values of @xmath106 .
as is explained in the introduction , in the very low temperature all dilute bose gas particles are condensed to the same quantum ground state which leads to the bec formation . when the particle s wavelengths overlap , they will experience correlation with each other and the mean inter - particles distance @xmath107 becomes definitely smaller than the thermal wavelength @xmath108 .
this is possible when the temperature is almost @xmath109 @xmath110 , where @xmath3 is the particle mass in the condensate , @xmath5 is boltzmann s constant @xcite and @xmath12 is the density . under the condition that the universe is sufficiently large and
temperature is low enough , there will be a coherent state in progress due to the overlap of the particles wavelengths .
our main assumption is that the dark matter halos may be constructed by a strongly coupled dilute bec at absolute zero . as a result ,
all the particles are in the condensate .
therefore in this situation , just binary collisions at low energy are considerable .
these collisions are specified by just one parameter @xmath111 , _ s - wave scattering length _ , which is thoroughly independent of the two - body potential .
consequently , the interaction potential can be replaced by an effective interaction @xmath112 , where @xmath113 @xcite .
the gross - pitaevskii ( gp ) equation , describing the ground state properties of the dark matter , for the dark matter halos can be obtained from the gp energy functional as follows @xmath114=\int\left[\frac{\hbar^{2}}{2m_{\chi}}|\nabla\psi(\vec{r})|^{2}+\frac{u_{0}}{2}|\psi(\vec{r})|^{4}\right]d\vec{r}- \frac{1}{2}gm^{2}_{\chi}\int\int\frac{|\psi(\vec{r})|^{2}|\psi(\vec{r}^{'})|^{2}}{|\vec{r}-\vec{r}'|}d\vec{r}d\vec{r}^{'},\end{aligned}\ ] ] where @xmath115 @xcite and obviously we have applied @xmath116 as the wave function of the condensate . in the equation ( [ eq22 ] ) , the first and second terms are the quantum pressure and the interaction energy , respectively , and also the third term is the gravitational potential energy . it should be noted that the condensate dark matter mass density is as follows @xmath117 and that the wave function satisfies the normalization condition @xmath118 , where @xmath119 is the total number of particles .
eventually , by imposing the variational procedure @xmath120-\mu~\delta\int|\psi(\vec{r})|^{2}d\vec{r}=0 $ ] , the gp equation is extracted @xmath121 while @xmath122 plays the role of chemical potential .
in addition the gravitational potential @xmath26 follows the well - known poison equation @xmath123 . having considered time dependency , the generalized gp equation , showing a gravitationally trapped rotating bec ,
is found as @xmath124\psi(\vec{r},t).\end{aligned}\ ] ] by means of the well - known madelung representation of the wave function , we can reach to our purpose more easily than using the last generalized gp equation . in order to go trough this , we should consider the following representation of @xmath125 @xmath126,\end{aligned}\ ] ] where @xmath127 represents the classical action . inserting the last expression of @xmath128 into equation ( [ eq25 ] ) leads to the following differential equations @xmath129
@xmath130=- \nabla p_{\chi}\left(\frac{\rho_{\chi}}{m_{\chi}}\right ) - \rho_{\chi}\nabla\left(\frac{v}{m_{\chi}}\right)-\nabla v_{q},\end{aligned}\ ] ] where @xmath131 is the velocity of the quantum fluid and @xmath132 is the quantum potential , given by @xmath133 by defining @xmath134 , the condensate effective pressure is obtained @xmath135 the quantum pressure term makes a great contribution just close to the boundary of the condensate provided that the number of particles in the gravitationally bounded bec becomes large enough . as a result , the quantum stress term in the condensate equation of motion can be ignored , which is so called thomas - fermi approximation .
when the number of particles in the condensate goes to infinity , the mentioned thomas - fermi approximation becomes accurate .
it is also related to the classical limit of the theory . according to its definition , the velocity field is irrotational following the condition @xmath136 . in accordance with equations ( [ eq10 ] ) and ( [ eq30 ] )
we can write @xmath137 which has the following general solution with the integration constant @xmath138 @xmath139 having imposed the condition @xmath140 for @xmath42 , we can find the density of the condensate as @xmath141 where @xmath142
for an interacting bose system case , where all dark matter particles are supposed to be in a single - particle state , the transition procedure from the normal to the condensed phase , and the order of the phase transition has been extensively studied in the recent literature @xcite-@xcite . in principle , bec represents a spontaneous breaking of @xmath143 gauge symmetry , with the condensate fraction showing the order parameter , leading to the second order phase transition @xcite .
nevertheless , it has been shown that bec is a first - order phase transition@xcite , @xcite . from another point of view , in thermodynamics
the first - order phase transitions result in a genuine mathematical singularity .
the question that can finite systems in nature explicitly show such a behavior ? " is a mysterious controversial question in physics @xcite .
nonetheless , a significant study about the thermodynamic instability in an ideal bose gas , confined in a cubic box , has shown that a system including a finite number of particles can exhibit an interrupted phase transition , describing a real mathematical singularity , with constant pressure @xcite .
of course the mentioned result was obtained without considering a continuous approximation or a thermodynamic limit .
therefore , according to above discussion , we will assume that the intrinsic dynamics of bec is described in the framework of a first - order phase transition . as we know
, in general , the chemical potential @xmath122 in a physical system is a function which depends on the temperature and particle density @xmath144 ( @xmath4 is the total particle number and @xmath26 is the volume ) . by considering the property of extensivity of the helmholtz free energy @xmath145
, we are allowed to write not only @xmath146 but also @xmath147 , where @xmath148 and @xmath149 . using the helmholtz free energy
we can obtain @xmath150 and @xmath151 having the same information . imposing the thermodynamical requirements ,
the chemical potential and the pressure are supposed to be single - valued which means that for arbitrary values of @xmath4 , @xmath152 and @xmath53 , there must be just one choice for @xmath122 or @xmath13 , respectively @xcite .
hence , the pressure at the transition moment during the bec process must be continuous .
imposing this thermodynamic condition on equations ( [ eq18 ] ) and ( [ eq30 ] ) , helps us to fix the transition density @xmath153 as follows @xmath154 apparently , in order to have the numerical value of the transition density , we should know the scattering length , the dark matter particle mass and the particle velocity dispersion .
supposing typically the mass of dark matter particle , the mean velocity square , and the scattering length to be of the order of @xmath155 @xmath156 , @xmath157 @xmath158 and @xmath159 @xmath160 , respectively , we can write @xmath161 therefore , the critical temperature at the bec moment is obtained by @xmath162 where @xmath163 is the riemann zeta function , or equivalently @xmath164 the critical pressure of the dark matter fluid at the critical point is given by @xmath165 using the equation ( [ eq35 ] ) we can calculate the critical value of scale factor as @xmath166 consequently , we can find the critical red - shift as follows @xmath167 by inserting the above mentioned numerical values of the parameters in the two last relations we have @xmath168 and @xmath169 the obtained critical values of density , temperature , pressure , scale factor and redshift are important in the study of subsequent ( lqc ) evolution of the the universe during the bose - einstein condensation toward the end time of phase transition when the universe enters in the complete bose - einstein condensed dark matter phase .
we will use these critical values in the study of this evolution in the following subsection . using all these equations
, we consider the time evolution of the volume fraction of the condensed matter dark energy density during the bec process , which is describing the dynamics of this process in the cosmic history in lqc model , as @xmath197 here , we have used @xmath198 as a dimensionless time variable @xmath199 and @xmath200 . considering the equation ( [ eq48 ] ) and @xmath201
, we deduce the approximation @xmath202 in the following calculations .
another approximation comes from neglecting the radiation energy density contribution , so the equation ( [ eq53 ] ) can be integrated to extract @xmath8 @xmath203 where we have defined @xmath204 and @xmath205 in order to obtain the necessary time interval for the entire converting process of the total normal dark matter into the bose - einstein condensed phase , we put @xmath206 in our last calculation @xmath207.\end{aligned}\ ] ] by inserting the value of dark energy density parameter @xmath50 as 0.68 and also considering the standard values @xmath208 @xmath209 , @xmath210 @xmath211 , @xmath212 @xmath213 , and @xmath214 , we can obtain the following expression for the conversion time of the dark matter to the bose - einstein condensed phase @xmath215\times t_{h}\rm s.\end{aligned}\ ] ] for instance , for @xmath216 and @xmath217 , we will have @xmath218 . since it is thoroughly useful to depict the time evolution of @xmath8 , we have represented ( [ eq54 ] ) for different values of @xmath219 in fig.1 .
after the complete conversion to the bec phase , the post - condensation phase begins , which starts at @xmath220 and @xmath221 .
similar to the era , during the condensation phase , in which the cosmological dynamics is affected by the bec ( see ( [ eq50 ] ) ) , the bose - einstein condensed dark matter , in the post condensation phase , may change the cosmological dynamics of the universe in lqc model .
moreover , the bose - einstein condensed dark matter may have considerable impact on the subsequent structure formation in the universe .
therefore , the post condensation phase and the corresponding cosmological values in lqc model are important and deserve more scrutiny . according to ( [ eq33 ] ) ,
we can obtain the cosmological density of the dark matter as follows @xmath222 where , the numerical value of @xmath223 is obtained from ( [ eq34 ] ) @xmath224 clearly , the density ( [ eq59 ] ) should be positive which leads to the condition @xmath225 . simplifying the equation ( [ eq21 ] ) and defining @xmath226 gives us the time evolution of the scale factor in the bec phase
as follows @xmath227 by integration of this relation , we obtain exactly an expression with the following form @xmath228 where @xmath229 is an integration constant , determined from the constraint @xmath230 at @xmath220 as @xmath231 for the universe presented here , filled with radiation , dark energy , baryonic matter and bose - einstein condensed dark matter , the equation describing time evolution of the scale factor is given by @xmath232 which can be integrated by considering the initial conditions @xmath233 . to study the time evolution of the scale factor in such a universe which contains bec dark matter
, it will be useful to depict it for different values of the @xmath223 in fig.@xmath234 .
it is remarkable that the existence of the condensed dark matter changes the cosmological dynamics of the universe in the post condensation phase , and the magnitude of this change decreases with the increase of the bose - einstein condensate parameter @xmath223 .
this behavior is expected because as @xmath223 increases , the corresponding gravity contributes stronger to the dynamics and decreases the expansion rate . on the other hand ,
it is seen from fig.2 that as @xmath223 increases , the bose - einstein condensation phase ( where @xmath235 ) occurs at earlier times .
this is also justified because as @xmath223 increases , the @xmath236 increases as well and so the scale factor @xmath237 corresponding to this increased density tends toward the scale factors having higher density , which naturally occurs at earlier times .
when the bose - einstein condensation happens , dark matter contains two phases , the normal and the condensed phase , respectively . under a thermodynamic condition , we expect to have the continuity of the pressure profile of these two phases right at the beginning of the condensation process .
this leads to uniquely determination of dark matter density at the condensation time as well as other thermodynamical quantities , like temperature and pressure .
the explicit numerical values of the condensation quantities is related to mean square velocity of the normal dark matter @xmath106 and the mass @xmath52 of the dark matter particle and the scattering length @xmath111 . as a result of the uncertainty of the values of @xmath106 , @xmath52 and @xmath111 ,
apparently it is difficult to predict the precise and exact cosmological moment of the bose - einstein condensation and consequently the corresponding cosmological quantities . nevertheless , by considering some `` standard '' numerical values , it will be possible to obtain a total qualitative picture of the transition .
consequently , we estimate the values of the dark matter mass and the mean velocity of the non - relativistic dark matter particles , of the order of @xmath238@xcite , @xcite and @xmath239 , respectively .
according to the general analysis of the condensation process , the first phase of condensation belongs to the standard @xmath34 cold dark matter ( @xmath34cdm ) model . in this paper
, we have generalized this analysis and have gone through the loop quantum cosmology context . with the standard numerical values of the dark matter quantities ,
the condensation was begun at a redshift around @xmath240 . here , we have studied the time evolution of the fraction factor @xmath8 of the condensed dark matter through the loop quantum cosmology . in the next step ,
we have extracted the modified phase transition time interval in lqc , which is approximately of the order of hubble time and shows the viability of our results .
we have studied the evolution during the bose - einstein condensation phase and found that this condensation process modifies the expansion rate of the universe .
also , we have studied the post - condensation phase in lqc model , describing the time evolution of universe right after the end of phase transition , which leads to ( [ eq64 ] ) .
the cosmic time dependence of the global formation of bec dark matter ( at @xmath184 ) is an advantage , compared to the ordinary dark matter , which makes bec dark matter to be preferred than the ordinary dark matter .
this is an important feature which makes bec dark matter to be similar to the global dark matter candidates like neutrinos , wimps , axions and lsp .
the ordinary dark matter has not such capability and can not form global sources of dark matter at a definite cosmic time , and that is why we have favoured bec dark matter in the present study .
we have plotted the time behavior of equations ( [ eq53 ] ) and ( [ eq64 ] ) in figures 1 and 2 for different values of @xmath219 and @xmath223 , showing the definite effect of the phase transition and bose - einstein condensed dark matter density in the modified cosmological history in the loop quantum cosmology .
a major point regarding the present study is that assuming two phases of dark matter , one as the bosonic particles before bec , and the other one as bec dark matter , `` is there any experimental indication for the occurrence of bec , the existence of these two phases , and the preference of one to the other one ? '' or `` why it is necessary to distinguish ordinary dark matter from bec dark matter ? '' . of course
, at present there is no direct evidence for the occurrence of bec , the existence of these two phases , and the preference of one to the other one .
however , in order to answer properly these questions , we may try to answer the alternative question that `` is there any indication for distinguishing between the dynamics of the universe having bosonic particles without the occurrence of bec , and the one with bec dark matter '' . in this regard , we may point out that the time evolution of the scale factor with bec is different from that of without bec for many possible reasons such as i ) the equation of state parameter of bec dark matter is different from that of bosonic particles without bec , due to the first order phase transition , ii ) in the absence of bec , the bosonic particles may interact or participate in some possible transformations into other matter components , like visible matter or dark energy , each resulting in different dynamical history of the universe ; however with bec the condensed bosonic particles can not decay into other forms of matter and they remain as the bec dark matter permanently , with a definite dynamical history distinguishable from variety of dynamics corresponding to the regular bosonic particles having different kinds of interaction with other matter components .
this work has been supported financially by research institute for astronomy and astrophysics of maragha ( riaam ) under research project no.1/4165 - 96 .
k. bamba , j. de haro and s.d .
odintsov , jcap 1302 , 008 ( 2013 ) .
a. ashtekar and j. lewandowski , j. math .
36 , 2170 ( 1995 ) .
a. ashtekar and j. lewandowski , class .
21 , r53 ( 2004 ) . c. rovelli , liv .
rel . 1 , 1 ( 1998 ) .
a. ashtekar et al , adv .
phys . 7 , 233 ( 2003 ) .
p. wu and s. n. zhang , jcap 0806 , 007 ( 2008 ) ; x. fu , h. yu and p. wu , phys .
d 78 , 063001 ( 2008 ) ; s. chen , b. wang and j. jing , phys . rev .
d 78 , 123503 ( 2008 ) ; r. lamon and a. woehr , phys . rev .
d 81 , 024026 ( 2010 ) ; s. li and y. ma , eur . phys .
j. c 68 , 227 ( 2010 ) ; m. jamil , d. momeni and m. a. rashid , eur .
j. c 71 , 1711 ( 2011 ) ; m. jamil and u. debnath , astrophys . space .
sci 333 , 3 ( 2011 ) .
j. de haro , yi - fu .
cai , gen .
grav 47 , 8 ( 2015 ) . j. de haro , j. amors , jcap 08 , 025 ( 2014 ) .
j. de haro , a. n. makarenko , a. n. myagky , s. d. odintsov , v. k. oikonomou , phys . rev .
d 92 , 124026 ( 2015 ) .
j. j. amors , j. de haro , s. d. odintsov , phys .
d 89 , 104010 ( 2014 ) .
f. dalfovo , s. giorgini , l. p. pitaevskii , and s , stringar ; rev .
71 , 463 ( 1999 ) .
m. h. anderson , j. r. ensher , m.r .
matthews , c.e .
wieman , and e.a .
cornell , science 269 , 198 ( 1995 ) ; c. c. bradley , c. a. sackett , j. j. tollet , and r. g.hulet , phys . rev .
lett . 75 , 1687 ( 1995 ) ; k. b. davis , m. o. mewes , m. r. andrews , n. j. van prutten , d. s. durfee , d. m. kurn , and w. ketterle , phys . rev . lett . 75 , 3969 ( 1995 ) . t. fukuyama and m. morikawa , prog .
115 , 1047 ( 2006 ) .
t. fukuyama m. morikawa , and t. tatekawa , j. cosmol .
astropart . phys .
06 , 033 ( 2008 ) . t. fukuyama and m. morikawa , phys , rev .
d 80 , 063520 ( 2009 ) .
e. a. cornell and c. e. wieman , rev .
phys . 74,875 ( 2002 ) ; w. ketterle , rev .
74 , 1131 ( 2002 ) ; l. pitaevskii and s. stringari , _ bose - einstein condensation _
( clarendon press , oxford , 2003 ) ; r. a. duine and h. t. c. stoof , phys .
396 , 115 ( 2014 ) ; q. chen , j. stajic , s. tan , and k. levin , phys .
rep.412 , 1 ( 2005 ) . c. j. pethick and h. smith , _ bose - einstein condensation in dilute gases _
( cambridge university press , cambridge,2008 ) ; m. yu .
khlopov , b. a. malomed and ya .
b. zeldovich , mon . not .
soc , 215 , 575 ( 1985 ) ; z. g. berezhiani , m. yu .
khlopov , z. phys .
c , 49 , 73 ( 1991 ) ; z. g. berezhiani , a. s. sakharov and m. yu .
khlopov , yadernaya fizika , 55 , 1918 ( 1992 ) [ english translation : sov .
phys , 55 , 1063 ( 1992 ) ] ; a. s. sakharov , d. d. sokoloff and m. yu .
khlopov , yadernaya fizika , 59 , 1050 ( 1996 ) [ english translation : phys . atom .
nucl , 59 , 1005 ( 1996 ) ] ; m. yu .
khlopov , a. s. sakharov and d. d. sokoloff , nucl .
suppl . ) 72 , 105 ( 1999 ) ; i. g. dymnikova , m. yu .
khlopov , mod .
lett . a 15 , 2305 ( 2000 ) ; i. g. dymnikova , m. yu .
khlopov , eur .
j. c , 20 , 139 ( 2001 ) . j. f. navarro , c. s. frenk , and s. d. m. white , astrophys .
j. 490 , 493 ( 1997 ) .
a. burkert , astrophys .
447 , l 25 ( 1995 ) .
w. hu , r. barkana , and a. gruzinov , phys .
85 , 1158 ( 2000 ) ; j. goodman , new astron .
5 , 103 ( 2000 ) ; p. j. e. peebles , astrophys .
l 127 ( 2000 ) ; v. sahni and l. wang , phys .
d 62 , 103517 ( 2000 ) .
t. harko , mon . not .
r. astron .
413 , 3095 ( 2011 ) .
chavanis , [ arxiv : gr - qc/1103.2698 ] .
t. harko , phys . rev .
d 83 , 123515 ( 2011 ) .
a. ashtekar , and d. sloan , gen .
43 , 3619 ( 2011 ) ; m. bojowald , h. h. hernndez , m. kagan , p. singh and a. skirzewski , phys.rev.d74 : 123512 ( 2006 ) ; m. bojowald , living rev . rel . 8 : 11 ( 2005 ) ; m. bojowald , j. phys .
24 77 ( 2005 ) ; m. bojowald and h. a. morales - tecotl , lect .
notes phys .
646 : 421 ( 2004 ) ; m. bojowald and a. skirzewski , rev .
18 ( 2006 ) 713
. k. a. meissner , class .
quant . grav .
21 , 5245 ( 2004 ) ; a. ghosh and p. mitra , phys . rev .
d 71 , 027502 ( 2005 ) .
a. ashtekar , t. pawlowski , and p. singh , phys .
96 , 141301 ( 2006 ) ; a. ashtekar , t. pawlowski , p. singh , phys . rev .
d 74 , 084003 ( 2006 ) ; a. ashtekar , t. pawlowski , p. singh , and k. vandersloot , phys . rev .
d 75 , 024035 ( 2007 ) ; k. vandersloot , phys .
d 75 , 023523 ( 2007 ) ; a. corichi , t. vukasinac , and j. a. zapata . c. j. hogan and j. j. dalcanton , phys . rev .
d 62 , 063511 ( 2000 ) ; j. madsen , phys . rev .
d 64 , 027301 ( 2001 ) .
s. l. shapiro and s. a. teukolsky , _ black holes , white dwarfs , and neutron stars : the physics of compact objects _ ( wiley , new york , 1983 ) . g. hinshaw et al , astrophys .
180 , 225 ( 2009 ) .
j. o. anderson , rev .
mod . phys .
76 , 599 ( 2004);e . h. lieb , r. seiringer , and j. yngvason , phys .
94 , 080401 ( 2005 ) .
l. olivares - quiroz , and v. romero - rochin , j. phys .
b 43 , 205302 ( 2010 ) .
park , and s .- w .
kim , phys .
a 81 , 063636 ( 2010 ) .
m. kastner , rev .
mod . phys .
80 , 167 ( 2008 ) .
d. boyanovsky , h. j de vega , and n. g. sanchez , phys .
d 77 , 043518 ( 2008 ) . | we consider the flrw universe in a loop quantum cosmological model filled with the radiation , baryonic matter ( with negligible pressure ) , dark energy and dark matter .
the dark matter sector is supposed to be of bose - einstein condensate type .
the bose - einstein condensation process in a cosmological context by supposing it as an approximate first order phase transition , has been already studied in the literature . here , we study the evolution of the physical quantities related to the early universe description such as the energy density , temperature and scale factor of the universe , before , during and after the condensation process .
we also consider in detail the evolution era of the universe in a mixed normal - condensate dark matter phase . the behavior and time evolution of the condensate dark matter fraction is also analyzed . |
effective models are at the very center of theoretical physics since they allow to focus on the essential physics of a problem without being distracted by unnecessary complexity .
hence it is very important to dispose of _ systematic _ means to derive effective models . here we will present the mathematical structure of a certain kind of effective models , namely effective models where the elementary excitations above the ground state can be viewed as particles above a complex vacuum .
this type of view is very common in low - temperature physics .
many experiments can be understood on the basis of this picture . in this paper
, we will elucidate the global structure of the hamiltonian and of the observables if the model is transformed to a model which _ conserves _ the number of particles .
such a mapping is often possible and renders the subsequent calculation of physical quantities much easier .
the determination of the effective hamiltonian is facilitated by the decomposition into @xmath0-particle irreducible parts .
we set up such a classification at zero temperature for _ strong - coupling _ situations , i.e.no weak - coupling limit is needed and no non - interacting fermions or bosons are required .
generically , we deal with hard - core bosons .
the necessity for the decomposition into @xmath0-particle irreducible parts has arisen in perturbative calculations of the effective hamiltonians because only the @xmath0-particle irreducible interactions are independent of the system size .
the second main point of this article is the perturbative computation of effective hamiltonians and observables .
such computations are a standard technique for ground state energies ( 0-particle terms ) and dispersion relations ( 1-particle terms ) , see ref . and references therein .
but the possibility to compute multi - particle contributions has only recently been realized @xcite and continues to be exploited intensively .
the key ingredient is to define a similarity transformation on the operator level ( see below ) . a promising alternative route , which we can only sketch in this article , consists in the non - perturbative , renormalizing realization of the transformation of the initial model to the effective model which conserved the number of particles . examples of this approach are realized in fermionic models @xcite .
we consider models which are defined on a lattice @xmath1 . at each site of the lattice
the system can be in a number @xmath2 of states spanning the local hilbert space .
let us assume that @xmath2 is finite .
the dynamics of the system is governed by a hamiltonian @xmath3 acting in the tensor - product space of the local hilbert spaces . for simplicity
we do not consider antisymmetric , fermionic situations although this is also possible .
so we are focusing on physical systems which can be described in terms of hard - core bosons .
the hamiltonian @xmath3 is assumed to be of finite range .
this means that it is composed of _ local _ operators @xmath4 acting on a _ finite _ number of sites in the vicinity of the site @xmath5 .
@xmath6 we further assume , that @xmath3 can be split as @xmath7 so that the spectrum of @xmath8 is simple ( see below ) and that the system does not undergo a phase transition from @xmath9 to the range of values we are finally interested in . these requirements do not necessarily imply that @xmath10 has to be small .
but it is helpful if this is the case .
the ground state of @xmath8 and its lowest lying eigen - states shall be known .
the latter will be viewed as elementary excitations from which the whole spectrum can be built .
we assume that we can view the elementary excitations above the ground state as ( quasi-)particles above the vacuum . for simplicity , we will drop the prefix ` quasi- ' ; it is understood that ` particle ' is a synonym for elementary excitation .
we assume that the physical picture sketched for @xmath11 is linked _
continuously _ to the range @xmath12 where @xmath13 is the critical value at which a phase transition occurs .
at the critical value @xmath13 the picture breaks down and can not be used beyond @xmath14 .
generically , a mode of @xmath15 will become soft at @xmath13 .
furthermore , the particles for @xmath9 shall be local in the sense that we can assign a site to each of them .
let @xmath16 be the operator that counts the number of particles .
as a concrete example , the reader may think of an antiferromagnetic heisenberg model made up from strongly coupled ( coupling @xmath17 ) pairs of spins ( ` dimers ' ) which are weakly coupled ( coupling @xmath18 ) among themselves , e.g. . at @xmath9 ,
the ground state is the product state with singlets on all dimers ; the elementary excitations are local triplets
. the number of these local triplets , i.e.the number of dimers which are not in the singlet state , shall be given by the operator @xmath16 .
a considerable simplification of the problem can be achieved by mapping the initial problem @xmath15 to an effective hamiltonian @xmath19 in which the number of elementary excitations does not change .
that is the number of particles should be a conserved quantity .
then the computation of many physical quantities is significantly simplified . in this article
, we advocate to use a continuous unitary transformation ( cut ) @xcite in order to achieve a systematically controlled mapping of the kind described above which leads to @xmath20=0\ , \ ] ] i.e. @xmath21 conserves the number of particles .
such an approach has three major advantages : 1 .
conceptual clarity + using a unitary transformation guarantees that no information of the orginial model is lost . in particular , it is clear that the _ same _ transformation @xcite can be applied to obtain the effective observables @xmath22 from the original observables @xmath23 .
technical simplicity + to implement the unitary transformation in a continuous fashion only the computation of commutators is required since the mapping is split into infinitesimal steps leading to a differential equation @xcite @xmath24\ ] ] where @xmath25 $ ] is an auxiliary parameter parametrizing the continuous transformation with starting point @xmath26 and end point @xmath27 .
good controllability + by an appropriate choice of the infinitesimal generator @xmath28 of the transformation it can be designed such that is preserves block - band diagonality @xcite . moreover , it is renormalizing in the sense that matrix elements between energetically very different states are transformed more rapidly than those between energetically adjacent states @xcite .
we like to point out , however , that the general structure of operators does not depend on the details of the method by which the effective particle - conserving model is obtained . also other methods than cuts are conceivable . in the present paper
we will focus on perturbative realizations of the cuts .
this approach @xcite was the first which realized the computation of bound states in higher orders @xcite .
the concept of a similarity transformation is indispensible for a conceptually clear computation of multi - particle effects @xcite . in sect .
[ lct ] we analyse global structural aspects of effective operators . the basic prerequisite will be eq .
( [ h_q_com ] ) .
furthermore , we show that the linked cluster property holds .
therefore the effective operators which hold in the thermodynamic limit can be computed in finite systems .
[ h_trafo ] is a preparatory section in which the perturbative cut for hamilton operators of certain kind is constructed .
low - dimensional spin models on lattices are among the models which can be treated in this way .
[ o_trafo ] contains a detailed description of how the perturbative cut can be extended to transform general observables .
series expansions in @xmath10 for the effective observables are obtained which allow to compute the experimentally relevant spectral functions .
so the extension from hamiltonians to observables is an important one .
the article is concluded in sect . v.
in this section , we _ assume _ that we are able to construct a mapping such that @xmath21 fulfills eq .
( [ h_q_com ] ) .
the eigen - states of the particle number operator @xmath16 serve as a basis for the hilbert space of the system .
if the mapping is realized perturbatively , the matrix elements of @xmath21 and @xmath22 are polynomials in @xmath10 .
we will show that @xmath21 can be written as @xmath29 where @xmath30 is an @xmath0-particle irreducible operator , i.e. @xmath30 measures @xmath0-particle energies .
moreover , each thermodynamic matrix element of any of the components @xmath30 can be obtained on finite clusters for a given order in @xmath10 if the original hamiltonian is of finite range .
the components @xmath30 can be defined recursively in ascending order in @xmath0 .
( [ h_hauptzerl ] ) comprises already a route to determine the properties of @xmath21 in a sequence of approximate treatments .
the very first step is to know the ground state energy which defines @xmath31 .
the second level is to describe the dynamics of a single particle ( elementary excitations ) correctly which is possible by knowing @xmath32 .
the third level is reached if @xmath33 is included which contains the information on the interaction of two particles .
true three - particle interactions are contained in @xmath34 and so on . from the generic experience in condensed matter theory ,
the three- and more particle terms can very often be neglected .
so the first three terms in eq .
( [ h_hauptzerl ] ) provide the systematically controlled starting point of a broad class of problems .
let us clarify some notation .
we define the following eigen - states of the particle number operator @xmath16 @xmath35 i.e. @xmath36 , @xmath37 and @xmath38 and so on .
these states span the global hilbert space @xmath39 of the physical system under study . dealing with ( hard - core ) bosons @xmath40 and @xmath41 are identical states .
this indistinguishability causes a certain ambiguity .
this ambiguity can be remedied for instance by assuming that coefficients depending on several indices @xmath42 are even under permutation of any pair of these indices @xcite . for simplicity
, the ground state @xmath43 is assumed to be unique .
let @xmath44 be an arbitrary operator acting on @xmath45 and conserving the number of particles @xmath46=0 $ ] .
by @xmath47 we denote the restricted operator acting on @xmath48 spanned by all states with exactly @xmath0 particles .
now we define the operators @xmath30 [ class ] @xmath49 where @xmath50 is the identity operator .
note that these operators are defined on the full hilbert space @xmath39 .
the operators @xmath51 are _ local _ operators that annihilate ( create ) particles at site @xmath52 .
they are bosonic operators .
their definition can be tailored to include a hard - core repulsion between the particles to account for the common situation that at maximum one of the particles may be present at given site @xmath52 . if the particles have additional internal quantum numbers , i.e. if there can be different particles at each site , the indices @xmath52 and @xmath53 are substituted by multi - indices @xmath54 and @xmath55 . as an example
let us consider that there are three kinds of particles per sites , but that at maximum one of these particles can occupy a given site .
then each site corresponds to a four - level system ; the particles are hard - core bosons .
such a situation arises in antiferromagnetic dimerized spin systems where each dimer represents a four - level system .
the ground state is the unique singlet while the three particles are given by the three - fold degenerate triplet states . in this case
we have the multi - indices @xmath56 , where @xmath52 denotes the site and @xmath57 takes for instance the three values of the @xmath58 component @xmath59 . in the local basis @xmath60 , where @xmath61 denotes the singlet , the local creation operators @xmath62 are the @xmath63-matrices [ e_def ] @xmath64 it is understood that the action at all other sites but @xmath52 is the identity so that the operators in ( [ e_def ] ) are defined on the whole hilbert space .
the annihilation operators @xmath65 are given by the hermitian conjugate matrices .
all possible commutators can easily be computed within the matrix representation .
finite matrix elements in the lower right @xmath66 block can be viewed as combined annihilation & creation processes : the matrix @xmath67 with all elements zero except the one at @xmath68 corresponds to the process @xmath69 .
a finite matrix element in the upper left @xmath70 block , i.e. the singlet - singlet channel , can be expressed in normal - ordered fashion as @xmath71 . in this way
the operators ( [ e_def ] ) and their hermitian conjugate define a complete algebra which in turn enables us to classify contributions of the hamiltonian according to the number of particles affected as done in eqs .
( [ h_hauptzerl ] ) and ( [ class ] ) .
the decomposition ( [ h_hauptzerl ] ) is physically very intuitive .
yet the next important question is whether and how the operators @xmath30 are unambiguously defined .
this issue is addressed by noting that @xmath72 vanishes for @xmath73 .
this follows directly from the normal - ordering of the creation and annihilation operators in eq .
( [ class ] )
. then we can proceed iteratively by requiring that @xmath21 applied to @xmath0 particles corresponds to @xmath74 ( @xmath0 arbitrary but fixed ) . solving for @xmath30 yields the recursions [ rekurs ] @xmath75 assuming that @xmath21 is calculated beforehand one starts by evaluating @xmath76 by means of the first definition .
the result entirely defines @xmath31 .
the restriction @xmath77 is then used in the second equation to extract the @xmath78 of @xmath32 and so on .
generally , @xmath30 is defined on the full many - particle hilbert space , not only for @xmath0 particles .
but it is sufficient to know the action of @xmath30 on the subspace of @xmath0 particles to determine all its matrix elements in ( [ class ] ) .
it is the essential merit of the notation in second quantization ( [ class ] ) that it provides the natural generalization of the action of a part of the hamiltonian on a _ finite _ number of particles to an _ arbitrary _ number of particles . since eq .
( [ rekurs - fin ] ) holds for any number of particles and since @xmath72 vanishes for @xmath73 we obtain eq .
( [ h_hauptzerl ] ) , neglecting the precise definition of convergence which is beyond the scope of the present paper . in conventional many - body language ,
@xmath30 stands for the @xmath0-particle irreducible interaction .
the subtractions in eq .
( [ rekurs ] ) ensure that @xmath30 contains no reducible contributions , i.e. contributions which really act only on a lower number of particles .
it should be emphasized that the formalism above does not require that a simple _ free _ fermionic or bosonic limit exists .
it is possible to start from any type of elementary particles counted by some operator @xmath16 .
moreover , the formalism presented in this section does not depend on how @xmath21 is obtained .
it does not matter whether a perturbative , a renormalizing procedure or a rigorously exact method was used to obtain @xmath21 .
here we focus on formal aspects of a perturbative approach generalizing results obtained previously for 0-particle properties
@xcite and for 1-particle properties @xcite .
the feature that the hamiltonian is of finite range on the lattice is exploited .
then the eqs .
( [ rekurs ] ) can be evaluated on finite subsystems ( clusters , see below ) .
still , the thermodynamically relevant matrix elements of the operators @xmath30 are obtained as we show in the following paragraphs . to proceed further definitions
are needed .
a _ cluster _
@xmath79 of the thermodynamic system is a _
finite _ subset of sites of the system and their linking bonds . by @xmath80
we denote an operator which acts only on the hilbert space @xmath81 of @xmath79 .
if @xmath82 denotes the sites of the total system which are not included in @xmath79 , the restricted operator @xmath80 is lifted naturally to an operator @xmath44 in the total hilbert space @xmath83 by @xmath84 note that it is not possible to define a restricted operator @xmath80 from an arbitrary operator @xmath44 acting on @xmath39 since @xmath44 will not have the product structure ( [ prodstruc ] ) in general . two clusters @xmath85 and @xmath86 are said to form a _ disconnected cluster _ @xmath87
iff they do not have any site in common @xmath88 _ and _ there is no bond linking sites from @xmath85 with sites from @xmath86 .
otherwise the clusters @xmath85 and @xmath86 are said to constitute together a _ linked cluster _
given a disconnected cluster @xmath87 an operator @xmath80 is called _ cluster additive _
iff it can be decomposed as @xmath89 with these definitions we show that @xmath21 and @xmath30 are cluster additive .
but @xmath90 is not !
this will turn out to be another important reason to introduce the @xmath30 .
the cluster additivity of @xmath91 is obvious since @xmath85 and @xmath86 are assumed to be disconnected .
so they can be viewed as physically independent systems .
hence @xmath92 similarly , we deduce from ( [ rekurs ] ) the operators @xmath93 and @xmath94 which act on @xmath95 and @xmath96 , respectively .
then it is straightforward to verify that the operators @xmath97 fulfil the recursion ( [ rekurs ] ) for the operators defined for the cluster @xmath79 .
hence the operators @xmath21 and @xmath30 are indeed cluster additive .
it is instructive to see that @xmath90 is _ not _ cluster additive , contrary to what one might have thought .
let us consider the tentative identity @xmath98 this equation can not be true since on the left hand side the number of particles is fixed to @xmath0 while on the right hand side the number of particles to which the identities @xmath99 and @xmath100 are applied is not fixed .
so no cluster additivity is given for the @xmath90 .
the fact that cluster additivity holds only for particular quantities was noted previously for @xmath101 @xcite . for @xmath102 ,
the subtraction procedure was first applied in the calculations in ref .
( though not given in detail ) .
in refs .
the subtractions necessary to obtain the irreducible 2-particle interaction were given in more detail .
the general formalism presented in this article shows on the _ operator _ level why such subtractions are necessary and where they come from .
thereby , it is possible to extend the treatment to the general @xmath0-particle irreducible interaction .
the notation in terms of second quantization ( [ class ] ) renders the cluster additivity almost trivial .
this is so since the creation and annihilation operators are defined locally for a certain site .
it is understood that the other sites are not affected .
hence the same symbol @xmath103 can be used independent of the cluster in which the site @xmath52 is embedded . in particular
, one identifies automatically @xmath104 with @xmath105 if @xmath106 and with @xmath107 if @xmath108 .
hence cluster additivity is reduced to trivial statements of the kind that @xmath109 implies @xmath110 in this sense , the notation in second quantization is the most natural way to think of cluster additivity . following gelfand and co - workers
@xcite we conclude that the cluster additive quantities possess a cluster expansion .
hence all the irreducible matrix elements @xmath78 possess a cluster expansion and can be computed on finite clusters . since @xmath21 conserves the number of particles , i.e. eq . ( [ h_q_com ] ) , its action is to shift existing particles .
let us denote the relevant matrix elements for a linked cluster @xmath85 by [ a_def ] @xmath111 where the indices @xmath112 may be multi - indices from now on . put differently , @xmath113 is the matrix element of @xmath114 , the @xmath115 are the matrix elements of @xmath116 , the @xmath117 those of @xmath118 and so on .
the number @xmath113 is the ground state energy of cluster @xmath85 .
the recursive definitions ( [ rekurs ] ) imply [ t_def ] @xmath119 where @xmath120 comprises six terms resulting from @xmath31 , @xmath121 comprises 18 terms resulting from @xmath32 and @xmath122 comprises 36 terms resulting from @xmath33 .
the explicit formulae are given in appendix [ 3part ] .
the recipe in deriving the above equations is straightforward .
for a given @xmath0-particle process @xmath123
( @xmath124 ) one has to subtract all possible processes which move less than @xmath0 particles . since the @xmath125-particle processes with @xmath73
have been computed before the procedure is recursive .
note that all coefficients must be computed for the _ same _ cluster .
the cluster additivity or , equivalently , the existence of a cluster expansion can be exploited to compute the irreducible matrix elements on finite clusters given that the hamiltonian is of finite range .
there are two strategies to do so .
the first strategy is to choose a cluster large enough to perform the intended computation without finite - size effects .
this strategy works particularly well if the dimensionality of the problem is low .
let us assume for simplicity that the hamiltonian links only nearest - neighbour sites .
aiming at a given matrix element , for instance @xmath126 , which shall be computed in a given order @xmath127 , the large enough cluster @xmath128 contains all possible subcluster @xmath129 with two properties : ( i ) they have @xmath127 or less bonds , ( ii ) they link the concerned sites @xmath130 among themselves @xcite .
clearly , @xmath128 depends on the order @xmath127 .
but it depends also on the sites @xmath130 under study so that the notation @xmath131 is appropriate .
note , that the order of the sites does not matter .
if some sites are omitted the constraints for the subclusters @xmath129 are diminished since less sites must be linked .
this implies in particular @xmath132 . hence there can be a cluster @xmath85 which contains @xmath131 but does _
not _ contain @xmath133 so that the hopping matrix element @xmath134 is _ not _ the thermodynamic one , but the interaction @xmath126 is without finite - size correction .
so intermediate steps in the calculations ( [ t_def ] ) can display finite - size effects although the final result does not . in refs .
@xcite we followed this strategy . the second strategy is to compute for a given order @xmath127 the _ net _ contributions of all clusters @xmath79 with @xmath135 bonds which link the sites under study .
the advantage of this approach is that only smaller clusters need to be treated ( @xmath136 bonds ) .
the price to pay is an overhead in determining the _ net _ contribution .
this requires to deduct from the total contribution of @xmath79 the contributions of all subcluster of @xmath79 with less bonds which link the points under study .
this must be done in order to avoid double counting .
more details on this strategy can be found in ref . . for hamiltonians with relatively simple topology ,
the second strategy is more powerful . for more complicated hamiltonians , however , the task to implement the overhead without flaw can quickly become impracticable while the first strategy can still be used , at least up to a certain order of the perturbation .
an effective hamiltonian conserving the number of particles is useful to determine characteristic energies of the considered systems .
but it is not sufficient to determine physical quantities which require more knowledge than the eigen - energies of the system . in particular , we aim at determining dynamic correlations such as @xmath137 . then the mapping of the original hamiltonian @xmath3 to the effective hamiltonian @xmath21 must be extended to a mapping of the original observable @xmath138 to the effective observables @xmath22 . here
we will assume that this has been achieved by an appropriate unitary transformation , for instance in a continuous fashion as described in the introduction .
the structure of the observables can be described best by using the notation of second quantization .
thereby it can be denoted clearly how many particles are involved .
the most important difference compared to the hamiltonian is that there is no particle conservation .
generically an observable creates and annihilates excitations , i.e. particles .
hence we define the operators @xmath139 the local operators @xmath140 have been described after eq .
( [ class ] ) .
again they shall appear normal - ordered , i.e. all creation operators are sorted to the left of the annihilation operators .
the first index @xmath2 indicates how many particles are created ( @xmath141 ) or annihilated ( @xmath142 ) by application of @xmath143 .
the second index @xmath144 denotes how many particles have to be present before the operator @xmath143 becomes active .
the result of @xmath143 acting on a state with less than @xmath0 particles is zero . in analogy to eq .
( [ h_hauptzerl ] ) the effective observables can be decomposed into partial observables like @xmath145 the additional feature in comparison to eq .
( [ h_hauptzerl ] ) is the sum over @xmath2 .
[ o_zersplittert ] sketches the structure of the terms appearing in the partial observables
@xmath143 .[o_zersplittert ] list of terms appearing in the partial observables @xmath143 which form together the effective observable @xmath22 according to eq .
( [ o_hauptzerl ] ) . no prefactors or indices
are given for clarity . [ cols="^,^,^,^,^",options="header " , ] let us assume that we computed @xmath22 by some technique , for instance by a cut .
then the partial observables can be determined recursively by [ o_rekurs ] @xmath146 here @xmath147 denotes the restriction of an operator to act on the @xmath0-particle subspace @xmath148 ( domain ) and to yield states in the @xmath149+@xmath150-particle subspace @xmath151 ( co - domain ) .
the recursion is set - up in analogy to ( [ rekurs ] ) .
it is again used that an operator @xmath143 effectively vanishes if it is applied to _ less _ than @xmath0 particles . barring possible problems to define convergence , the validity of the recursion ( [ o_rekurs ] ) for all
@xmath2 and @xmath0 implies the decomposition ( [ o_hauptzerl ] ) . as for the hamiltonian
the partial observables @xmath143 can be viewed as the @xmath0-particle irreducible part of the particular observable .
the notation in second quantization elegantly resolves the question how the observables act on clusters as was explained in the section [ clustadditiv ] .
hence the definition ( [ o_dn ] ) ensures cluster additivity and there exist cluster expansions for the partial observables .
so they can be computed on finite clusters .
if dynamical correlations at zero temperature @xmath152 shall be described , the observables are applied to the ground state @xmath43 which is the particle vacuum @xcite .
then only the partial observables @xmath153 with @xmath154 matter . according to ( [ o_rekursa ] )
no corrections are necessary , i.e.the structure of the relevant part of the effective observable is given by @xmath155 this structure has been used so far in a number of investigations of spectral weights @xcite and spectral densities @xcite .
it turned out that it is indeed sufficient to consider a restricted number of particles @xcite .
but the question how many particles are required to describe a certain physical quantity sufficiently well depends on the considered model , the chosen basis ( what do we call a particle ? ) and the quantity under study . at finite temperatures a certain number of particles
will already be present in the system due to thermal fluctuations .
then the action of the partial observables @xmath143 with @xmath156 will come into play as well .
this constitutes an interesting route to extend the applicability of effective models , which were derived in the first place at zero temperature , to finite temperatures .
the recursive equations for matrix elements which can be derived from ( [ o_rekurs ] ) are very similar to those obtained for the hamiltonian ( [ t_def ] ) .
we illustrate this for the matrix elements of @xmath157 .
let the bare matrix elements on a cluster @xmath85 be [ v_def ] @xmath158 from ( [ o_rekurs ] ) we obtain the irreducible elements as [ w_def ] @xmath159 as for the irreducible interactions the strategy is straightforward .
one has to subtract from the reducible @xmath0-particle matrix elements @xmath160 the contributions which come from the @xmath125-particle irreducible matrix elements @xmath161 with @xmath73 . with this strategy
also other irreducible matrix elements can be determined in a straightforward manner .
so far our considerations were general in the sense that it did not matter how we achieved the mapping .
next we focus on the actual _ perturbative _ evaluation of the matrix elements on finite clusters . for simplicity , we assume as before that the perturbative part of the hamiltonian links only nearest - neighbour sites .
let us consider for instance @xmath162 .
we assume that the observable @xmath138 is also local , i.e. acts on a certain site only , or is a sum of such terms . if the observable is a sum of local terms then the transformation of each term separately and subsequent summation yields the result .
so without loss of generality we consider @xmath138 to affect only site @xmath163 .
then we have to compute the matrix elements for clusters linking the _ four _ sites @xmath164 .
if @xmath138 itself is a product of operators affecting several sites @xmath165 then the observable @xmath138 itself links these sites @xmath165 .
apart from this difference compared to the matrix elements of the effective hamiltonian , we may copy the remaining steps from there : there are again the two strategies .
either the calculation in order @xmath127 is performed on a cluster @xmath128 large enough so that all subclusters of @xmath127 bonds linking the relevant sites @xmath164 are comprised in @xmath128 @xcite . or one has to add the _ net _ contributions of all different clusters with @xmath127 or less bonds which link the relevant sites @xmath164 @xcite . in either way the results for spectral densities can be obtained
so far no particular property of the transformation providing the effective operators @xmath21 and @xmath22 was assumed .
the only prerequisites were the existence of a counting operator @xmath16 , which counts the number of elementary excitations , i.e. particles , and the conservation of this number of particles by @xmath21 : @xmath166=0 $ ] .
here we specify a particular transformation leading to @xmath166=0 $ ] .
this section is a very brief summary of ref . which is necessary to fix the ideas and the notation for the subsequent section dealing with the transformation yielding the effective observables . for simplicity
we restrict the considered systems in the following way : the problem can be formulated as perturbation problem as in eq .
( [ h_pert ] ) with the properties a. the unperturbed part @xmath8 has an equidistant spectrum bounded from below .
the difference between two successive levels is the energy of a particle , i.e. @xmath167 .
b. there is a number @xmath168 such that the perturbing part @xmath169 can be split as @xmath170 where @xmath171 increments ( or decrements , if @xmath172 ) the number of particles by @xmath0 : @xmath173=nt_n$ ] . condition ( a ) allows to introduce the particularly simple and intuitive choice @xmath167 .
note that the restrictions of ( a ) are not too serious in practice since very often the deviations from an equidistant spectrum can be put into the perturbation @xmath169 . conditions ( a ) and ( b ) together imply that the starting hamiltonian @xmath3 has a block - band - diagonal structure as depicted in fig .
[ h - matrix ] .
the perturbation @xmath169 connects states of different particle numbers only if the difference is a finite number @xmath174 . note that very many problems in physics display this property , for a discussion of interacting fermions see ref . .
so far , most applications consider @xmath175 @xcite and @xmath176 @xcite , but calculations for higher @xmath177 are also possible @xcite .
we solve the flow equation ( [ flow ] ) for the hamiltonian ( [ h_pert ] ) obeying the conditions ( a ) and ( b ) perturbatively , that means up to a certain order in the expansion parameter @xmath10 .
the _ ansatz _ used is @xmath178 with unknown real functions @xmath179 for which the flow equation ( [ flow ] ) yields non - linear recursive differential equations @xcite .
the notation comprises [ sym_def ] @xmath180 the second sum in ansatz ( [ h_ansatz ] ) runs over all indices @xmath181 of length @xmath182 . thereby , @xmath183 includes all possible virtual excitation processes @xmath184 in a given order @xmath185 multiplied by the weight @xmath186 .
the optimum choice for the infinitesimal generator @xmath28 of the unitary transformation reads @xmath187 in the eigen - basis @xmath188 of @xmath16 , i.e. @xmath189 , the matrix elements of the generator @xmath28 read @xmath190 with the convention @xmath191 .
this choice keeps the the flowing hamiltonian block - band diagonal also at intermediate values of @xmath192 @xcite .
for @xmath193 the generator ( [ gen ] ) eliminates all parts of @xmath183 changing the number of particles so that @xmath166=0 $ ] with @xmath194 . for the functions
@xmath179 a set of coupled differential equations is determined by inserting eqs .
( [ h_ansatz ] ) and ( [ eta_bfree ] ) in the flow equation ( [ flow ] ) and comparing coefficients . the differential equations are recursive @xcite .
the functions @xmath195 of order @xmath196 , i.e. @xmath186 with @xmath197 , are determined by the functions @xmath195 of order @xmath127 .
the initial conditions are @xmath198 for @xmath199 and @xmath200 for @xmath201 .
the functions are sums of monomials with structure @xmath202 , where @xmath203 are integers .
this allows to implement a computer - aided iterative algorithm for the computation of the functions @xmath195 @xcite .
the following symmetry relations hold [ f_sym ] @xmath204 relation ( [ sym1 ] ) reflects the hermitecity of the hamiltonian .
the block - band diagonality for all @xmath192 implies @xmath205 in the limit @xmath193 the coefficients @xmath206 are obtained .
they are available in paper form @xcite and electronically @xcite .
the effective hamiltonian is given by the general form @xmath207 where @xmath208 reflects the conservation of the number of particles .
the action of @xmath21 can be viewed as a weighted sum of particle - number conserving virtual excitation processes each of which is encoded in a monomial @xmath209 .
we want to emphasize that the effective hamiltonian @xmath210 with known coefficients @xmath211 can be used straightforwardly in all perturbative problems that meet conditions ( a ) and ( b ) . block - band diagonal hamilton matrix for @xmath175 in the eigen - basis @xmath188 of the operator @xmath16 which counts the number of particles .
the unperturbed hamiltonian @xmath11 and the effective hamiltonian @xmath21 have matrix elements in the dark areas only : @xmath166=0 $ ] . for non - degenerate ground state @xmath212 is a @xmath213 matrix .
the dimension of @xmath214 grows roughly like @xmath215 with system size @xmath216 .
the perturbation @xmath169 can lead to overlap matrices indicated as light boxes .
the empty boxes contain vanishing matrix elements only.,width=188 ]
to calculate physical quantities which do not only depend on the eigen - energies the relevant observables must also be known .
the conceptual simplicity of unitary transformations implies that the observables must be subject to the _ same _ unitary transformation as the hamiltonian . in this section
we describe how the perturbative cut method can be extended to serve this purpose .
consider the observable @xmath138 .
it is mapped according to the flow equation @xmath217\ , \ ] ] where the _ same _ generator @xmath218 , given in eq .
( [ eta_bfree ] ) , as in eq .
( [ flow ] ) is to be used to generate the transformation . in analogy to eq .
( [ h_ansatz ] ) we employ the ansatz @xmath219 where the @xmath220 are real - valued functions for which the flow equation ( [ q_flow ] ) yields recursive differential equations .
the operator products @xmath221 are given by @xmath222 where we use the notation of the eqs .
( [ sym_def ] ) .
the integer @xmath52 denotes the position in @xmath223 at which the operator @xmath138 is inserted in the sequence of the @xmath224 .
the starting condition is @xmath225 and the final result is found at @xmath27 : @xmath226 . inserting the ansatz ( [ o_mi_def ] ) for @xmath227 and the generator @xmath218 from ( [ eta_bfree ] ) into the flow equation ( [ q_flow ] ) yields @xmath228\ .\end{aligned}\ ] ] the functions @xmath186 are known from the calculations described in the previous section [ h_trafo ] pertaining to the transformation of the hamiltonian .
the sums denoted by expressions of the type @xmath229 run over all multi - indices @xmath230 of length @xmath127 . comparing coefficients in eq .
( [ g_dgl_1 ] ) yields a set of recursive differential equations for the functions @xmath231 . to ease the comparison of coefficients we split a specific @xmath230 with @xmath127 fixed in two parts as defined by @xmath52 @xmath232 with @xmath233 and @xmath234 such that the splitting reflects the structure of @xmath235 in ( [ o_mi_def ] ) .
then the exlicit recursions can be denoted by @xmath236 the recursive nature of these equations becomes apparent by observing that the summations @xmath237 and @xmath238 are performed over all non - trivial breakups of @xmath239 and @xmath240 .
for instance , the restriction @xmath241 with @xmath242 means , that one has to sum over the breakups @xmath243 this implies that the @xmath244 appearing on the right side of eq .
( [ g_dgl_2 ] ) are of order @xmath245 or less .
once they are known the function on the left hand side of order @xmath127 can be computed . by iteration , all functions can be determined .
the initial conditions follow from @xmath246 and read [ start_cond ] @xmath247 by iteration of ( [ g_dgl_2 ] ) , all functions can be determined .
we briefly discuss two examples to illustrate how the eqs .
( [ g_dgl_2 ] ) work .
let us assume @xmath176 .
all zero order functions @xmath248 are equal to @xmath249 . since there is no breakup of @xmath250 , as would be required by the sums on the right hand side of eqs .
( [ g_dgl_2 ] ) , the right hand sides vanish identically , whence @xmath251 for all values of @xmath192 .
the first order function @xmath252 is given by @xmath253f(\ell;(1))\cdot g(\ell;();1)\\ & = & e^{-\ell}\cdot 1\ , \label{example_1}\end{aligned}\ ] ] where @xmath254 is taken from eq .
( 15 ) in ref . .
with the initial condition @xmath255 from ( [ start_cond ] ) the differential equation ( [ example_1 ] ) yields @xmath256 as a second example we consider a second order function where we can use the above result [ example_2 ] @xmath257f(\ell;(-2,1 ) ) g(\ell;(),1)\\ & & \hspace*{5mm}+{\rm sgn}\left[m((-2))\right]f(\ell;(-2 ) ) g(\ell;(1),2)\\ & & = -\left(e^{-3\ell}-e^{-\ell}\right)\cdot 1 - e^{-2\ell}\cdot \left(1-e^{-\ell}\right)\\ & & = e^{-\ell}-e^{-2\ell}\label{ex - fin - res } \ .\end{aligned}\ ] ] again the functions @xmath195 are taken from eq .
( 15 ) in ref . .
integrating the result ( [ ex - fin - res ] ) using the initial condition ( [ start_cond ] ) leads to @xmath258 this kind of calculation carries forward to higher orders . the functions @xmath259 like the functions @xmath195 are sums of simple monomials @xmath202 , where @xmath203 are integers
. thus the integrations are always straightforward @xmath260\end{aligned}\ ] ] and can easily be implemented in a computer - algebraic programme .
the remaining implementation follows very much the same line as described previously for the functions @xmath195 @xcite . in analogy to eqs .
( [ f_sym ] ) for @xmath195 two symmetry relations hold for @xmath259 . with @xmath261
they read [ g_sym ] @xmath262 as can be shown by induction . the first symmetry ( [ osym1 ] ) holds if @xmath138 is hermitian .
unfortunately , there is no equivalence to eq .
( [ f_block ] ) so that a possible initial block - band structure in @xmath263 is generically lost in the course of the transformation , i.e. for @xmath264 . in the limit @xmath265 the coefficients @xmath266 are obtained as rational numbers .
so we retrieve finally @xmath267 similar to eq .
( [ h_eff ] ) .
we will make the coefficients @xmath268 available electronically @xcite .
note that @xmath22 is _ not _ a particle - conserving quantity as is obvious from the fact that the sum over @xmath269 is not restricted to @xmath208 . in order to see the net effect of @xmath270 on the number of particles
explicitly it is helpful to split the bare operator accordingly @xmath271 , where @xmath272 increments ( or decrements , if @xmath172 ) the number of particles by @xmath0 : @xmath273=nt'_n$ ] .
the difference between the bare initial observable @xmath138 and the representation ( [ o_eff_1 ] ) must be viewed as vertex correction which comes into play since the bare initial excitations are not the true eigen - excitations of the interacting system .
we like to stress that the formalism presented introduces the notions of @xmath0-particle irreducibility , vertex correction and so on _ without _ starting from the limit of _ non - interacting _ conventional particles such as bosons or fermions .
in this article we have presented an approach to calculate energies and observables for quantum multi - particle systems defined on lattices .
the article has two main parts . in the first part ( sect .
ii ) , we assumed the existence of a mapping of the original problem to an effective one in which the number of elementary excitations , the so - called ( quasi-)particles , is conserved .
the general structure of the effective hamiltonians and the observables is analysed .
we found that a classification of the various contributions in terms of the number of particles concerned is most advantageous . to this end
we introduced a notation in second quantization which does not , however , require non - interacting fermions or bosons .
generically , hard - core bosons are involved .
we found the formulation in second quantization particularly intuitive .
it provides in a natural way the irreducible quantities on the _ operator level _ which display cluster additivity .
we like to emphasize that the definition of irreducible operators is not a trivial task if a strong - coupling situation is considered as was done in the present paper .
no limit of non - interacting bosons or fermions is assumed .
since the definition of irreducible operators is completely general it allows to compute the @xmath0-particle contribution for arbitrary @xmath0 .
for instance the formulae for the @xmath274-particle interactions are given for the first time in the literature .
the _ irreducible _ interactions and vertex corrections possess a cluster expansion so that they can be computed on finite clusters provided that the hamiltonian is of finite range .
this property is the basis for the real - space treatment of many spin systems . in the second part ( sects .
iii and iv ) , we described an actual mapping which provides effective operators .
the mapping is based on continuous unitary transformations . in this paper
we constructed the mapping perturbatively ( see ` outlook ' ) . in sect .
iii the treatment of the hamiltonian is given .
the computation of the effective hamiltonian requires the solution of a set of recursive non - linear differential equations . for the perturbative set - up under study
these equations can be solved in full generality , i.e.no particular details of the model must be known . in sect .
iv we have given the calculational steps to compute effective observables .
again recursive differential equations have to be solved .
but they are linear since the transformation of the hamiltonian is known . for the perturbative set - up under study also
the equations for the observables can be solved in full generality , i.e.no particular details of the model must be known .
the above approach has been used to compute spectral functions , i.e. dynamical correlations , in a number of models @xcite .
these results may serve as examples for the utility of the approach presented .
we like to point out two important consequences of the formulation of the effective operators in second quantization .
both implications are based on the observation that the irreducible operators are defined on the whole hilbert space , i.e. not only for a small number of particles .
the matrix elements of the @xmath0-particle irreducible operators can be computed considering only @xmath0-particles .
but the resulting operators hold for arbitrary number of particles .
[ [ consequence-1 ] ] consequence 1 : + + + + + + + + + + + + + + the effective hamiltonian is valid at finite temperatures .
hence it is possible to extend the results obtained in the first place at _ zero _ temperature to _ finite _ temperatures .
the technical difficulty arising is to treat the interactions properly , in particular the hard - core constraint .
but the description in terms of effective particles helps to tackle this situation .
let us recall that at zero temperature no excitation , i.e. no particle , is present . at low temperatures only a _ small _ density of particles will be in the system .
so it is well justified to use a ladder approximation .
this approximation is also suited to deal with the hard - core constraint ( brckner approach ) @xcite .
note that the problems linked to the existence of anomalous green functions @xcite do not occur if the particle - conserving , effective hamiltonian is used .
thus the brckner approach for the _ effective _ hamiltonian after a suitable mapping @xcite is well justified and represents a very promising route to treat finite temperatures .
[ [ consequence-2 ] ] consequence 2 : + + + + + + + + + + + + + + so far the mapping to an effective model has been constructed perturbatively .
that means that all operators , the hamiltonian @xmath3 , the generator @xmath28 and the observables @xmath138 , are given in a series of some small parameter @xmath10 . in actual applications
these series are suitably extrapolated . but
a certain caveat persists if the starting point is a local hamiltonian .
then a calculation up to a certain order describes processes of a certain _ finite _ range only .
this restriction can be partly overcome by extrapolating in momentum space , e.g.for dispersion relations @xmath275 .
but it is difficult to extrapolate the matrix elements of the @xmath276-particle irreducible interaction because it is not diagonal in all momenta .
this problem can be overcome by performing the continuous unitary transformation directly on the level of the @xmath0-particle irreducible operators .
an ansatz for the effective hamiltonian is chosen comprising for instance all possible irreducible @xmath0-particle terms and similar terms creating and annihilating particles .
this ansatz is inserted in the flow equation ( [ flow ] ) .
comparison of the coefficients @xmath277 and @xmath278 in front of the terms @xmath279 yields coupled non - linear differential equations .
these differential equations represent renormalization equations for the problem under study .
we call this type of transformation a self - similar one since the _ kind _ of terms retained stays the same . again
, the formulation in second quantization allows a significant generalization .
we like to stress again that the approach presented does not require a weak - coupling limit . for illustration , however , the reader is referred to the weak - coupling examples in refs . .
[ [ concluding - remark ] ] concluding remark + + + + + + + + + + + + + + + + + in this paper we discussed the general structure of effective operators and a perturbative unitary transformation to derive them .
no concrete application is presented since such applications can be found elsewhere .
the two continuative points above show along which lines the general structure can be exploited to extend the applicability beyond zero temperature results and beyond finite range processes .
work along these lines is partly under way , but deserves definitely further attention .
the authors are grateful for many inspiring discussions with m. grninger , h. monien and e. mller - hartmann .
one of us ( gsu ) acknowledges very helpful discussions with c. pinettes on the problems of anomalous green functions in the brckner approach and its applicability at finite temperatures .
here we complete the formulae for the irreducible 3-particle interaction which was given in eq . ( [ t_def ] ) .
the corrections @xmath120 , @xmath121 and @xmath122 result from @xmath31 , @xmath32 and @xmath33 , respectively , as given in ( [ rekurs ] ) .
they read @xmath280\nonumber\\ & & \\
\!&= & t^a_{j_1;i_1}\tilde\delta_{j_2j_3;i_2i_3 } + t^a_{j_1;i_2}\tilde\delta_{j_2j_3;i_1i_3 } + t^a_{j_1;i_3}\tilde\delta_{j_2j_3;i_1i_2 } + \nonumber\\ & & t^a_{j_2;i_1}\tilde\delta_{j_1j_3;i_2i_3 } + t^a_{j_2;i_2}\tilde\delta_{j_1j_3;i_1i_3}+ t^a_{j_2;i_3}\tilde\delta_{j_1j_3;i_1i_2}+\nonumber\\ & & t^a_{j_3;i_1}\tilde\delta_{j_1j_2;i_2i_3 } + t^a_{j_3;i_2}\tilde\delta_{j_1j_2;i_1i_3 } + t^a_{j_3;i_3}\tilde\delta_{j_1j_2;i_1i_2 } \\
\!&= & \delta_{j_1i_1}\tilde t^a_{j_2j_3;i_2i_3 } + \delta_{j_1i_2}\tilde t^a_{j_2j_3;i_1i_3 } + \delta_{j_1i_3}\tilde t^a_{j_2j_3;i_1i_2 } + \nonumber\\ & & \delta_{j_2i_1}\tilde t^a_{j_1j_3;i_2i_3 } + \delta_{j_2i_2}\tilde t^a_{j_1j_3;i_1i_3}+ \delta_{j_2i_3}\tilde t^a_{j_1j_3;i_1i_2}+\nonumber\\ & & \delta_{j_3i_1}\tilde t^a_{j_1j_2;i_2i_3 } + \delta_{j_3i_2}\tilde t^a_{j_1j_2;i_1i_3 } + \delta_{j_3i_3}\tilde t^a_{j_1j_2;i_1i_2}\end{aligned}\ ] ] while the actual formulae are lengthy the underlying principle is straightforward ( see main text ) .
note that in concrete realizations it is often advantageous to denote only one representative of the states which do not change on interchange of particles ( @xmath282 ) .
furthermore , certain problems allow to exploit higher particular symmetries like spin rotation symmetry
. then additional permutation symmetries among the various quantum numbers constituting the multi - index can be exploited leading to the appearance of exchange - parity factors .
depending on the details of the interaction on the bonds it may be sufficient to consider smaller clusters than mentioned in the main text , for instance a pure nearest - neighbour spin exchange reduces the range of virtual excursions .
frustration is another mechanism which reduces the range of the effective processes , see e.g. the shastry - sutherland model @xcite . | for many - particle systems defined on lattices we investigate the global structure of effective hamiltonians and observables obtained by means of a suitable basis transformation .
we study transformations which lead to effective hamiltonians conserving the number of excitations .
the _ same _ transformation must be used to obtain effective observables .
the analysis of the structure shows that effective operators give rise to a simple and intuitive perspective on the initial problem .
the systematic calculation of @xmath0-particle irreducible quantities becomes possible constituting a significant progress .
details how to implement the approach perturbatively for a large class of systems are presented . |
following the successful launch and early operations of the _ galaxy evolution explorer _ satellite in 2003 , we decided to undertake an initial spectroscopic assay of sources detected by _
galex _ using the deimos multiobject spectrograph @xcite on the keck ii telescope .
although _ galex _ data have been matched to relatively deep optical surveys @xcite our study is the first to extend the modeling technique of @xcite from @xmath14 to @xmath15 .
the power of the _ galex _ medium imaging survey to select interesting objects is noteworthy .
we recall that the mis observations are 1500 sec in duration , imaging fuv and nuv simultaneously ( @xcite & @xcite ) .
the _ galex _
satellite has investigated the ultraviolet universe to @xmath16 with the goal of determining among other things , the star formation rate ( sfr ) of galaxies in the local universe , @xmath17 @xcite and the evolution of the global star formation density out to @xmath16 @xcite .
the ultimate goal of such work along with other surveys and observations of galaxies at similar and higher redshifts is to constrain the baryonic physics of galaxy formation and evolution .
questions still remain as to how the sfr depends on different factors such as environment , mass , morphology etc . @xcite .
previous work deriving sfrs with a uv - selected sample was carried out by @xcite who derive sfrs both from uv luminosities at 2000 and h@xmath18 luminosities from optical spectra for a sample of galaxies at redshifts of @xmath19 detected in the uv by the balloon - borne telescope foca @xcite .
estimates of local star formation rates before this were only possible through either optical emission line luminosities of the recombination lines of hydrogen , mostly h@xmath18 , and [ oii]@xmath203727 or the far infrared luminosity from 10 - 100@xmath21 m @xcite .
in this paper we present the observations and results from the pilot program for a sample of objects detected by _
galex _ and sdss with spectra obtained from the deimos spectrograph at the keck ii telescope .
the analysis of this sample is enhanced greatly by the techniques of salim et al.(2005 ) that have been expanded to include the derivation of galaxy physical parameters such as stellar mass , from _ galex _ + sdss photometry and redshifts .
the main goal of this paper is to characterize the galaxies detected by _
galex _ that are fainter than sdss spectroscopic galaxy sample , in terms of mass , sfr , and uv - optical color .
we are predominantly ( about two - thirds of the sample ) seeing the faint blue galaxy population @xcite and are also sensitive to qsos . while _ galex _
observations have been undertaken in many deep fields with much higher median redshifts @xcite we have the advantage in this sample of obtaining spectroscopy with keck / deimos , which gives wavelength coverage sufficient to derive some detailed physical parameters from the spectra . in
2 we describe the sample and the photometric and spectroscopic data .
redshift and emission line flux measurements from the spectroscopy are presented in 3 .
the distributions of uv color , uv - optical color , and the sdss _ r _
magnitude for the sample are described in 4 .
the derived galaxy parameters from sed fits are described in 5 .
we give a discussion and summary of our findings in 6 and 7 . throughout this paper
we assume _
km s@xmath10 mpc@xmath10 , @xmath23=0.3 , and @xmath24=0.7 .
_ galex _ is a nasa small explorer mission aimed to survey the uv emission from galactic and extragalactic sources from 700 km circular orbit @xcite .
_ galex _ images the sky simultaneously in two bands , the far - uv ( fuv 1344 - 1786 ) and the near - uv ( nuv 1771 - 2831 ) .
each _ galex _
circular field is 1.25 deg . in diameter .
we use fuv and nuv magnitudes and magnitude errors derived in elliptical apertures .
the photometry is taken from the _ galex _ internal release 1.1 ( ir1.1 ) .
the _ galex _
mis photometry for tile number 10273 has a limiting magnitude , m@xmath25(ab)@xmath26 .
we use optical photometry for our objects obtained from sdss data release 4 ( dr4 ) @xcite .
the sdss photometric data are taken with the 2.5 m telescope at apache point observatory .
imaging is obtained in _
bands @xcite .
the imaging data are photometrically @xcite and astrometrically @xcite calibrated . an overview of the sdss data pipelines and products
can be found in @xcite .
our sample is derived from objects with detections in the _ galex _ medium imaging survey ( mis ) tile 10273 , centered at a right ascension / declination of 17@[email protected]@xmath29 @xmath30 57@[email protected]@xmath33 .
the _ galex _ field being much larger than the deimos slit mask , we selected objects from the _ galex _ field to populate two deimos slit masks .
the two deimos fields were initially chosen to maximize the number of mis objects with optical counterparts in the sdss classified as galaxies with @xmath34 photometry from an early version of the _ galex _ data reduction pipeline .
the number of galaxies matching this criteria was much smaller than the number of available slits , and so to take full advantage of the deimos field of view , the slits were populated by any galaxy detected by both _
galex _ and sdss .
in addition , the remainder of available slits in both fields was populated with blue stellar - like objects ( qsos and white dwarfs ) , main sequence outlier stars ( @xmath35 or @xmath36 ) , and bright alignment stars .
the first deimos field contains 26 objects classified as galaxies in sdss , and 10 blue stellar objects and main sequence outlier stars .
the second deimos field contains 19 objects classified as galaxies in sdss , and 9 blue stellar objects and main sequence outlier stars .
the spectroscopic data were obtained at the keck ii telescope with the deimos multi - object spectrometer on october 1 , 2003 @xcite .
spectra were obtained using two slit masks , for a total 64 spectra .
the 830 grooves mm@xmath10 grating was used with a slit width of @xmath37 giving a resolution of @xmath38@xmath39fwhm . the signal - to - noise ratio obtained over the continuum of each spectrum varied between sources from 1 to @xmath40 , depending on the brightness of the source .
the integration time for both slit masks was 30 minutes .
flux calibration of the spectra was not performed as no flux standards were observed .
the spectra cover a wavelength range of @xmath41 .
figure 1 shows a panel of several one dimensional spectra for star - forming galaxies in the sample at a range of redshifts , plotted in the rest wavelength , and figure 2 shows the spectra for objects with measured @xmath3 .
in all , we measured 50 redshifts , 45 for sources classified as galaxies in sdss and 5 for sources classified as stars in sdss . of those five , three are qso s at @xmath3 and the other two are star - forming galaxies .
the remaining 14 of the sources classified as stars by the sdss have stellar spectra . the redshifts for objects with detected emission lines
were measured by eye . for objects with @xmath42 ,
multiple emission lines were detected including either h@xmath18 and/or h@xmath43 and [ oii]@xmath203727 , [ oiii]@xmath204959 , [ oiii]@xmath205007 , [ nii]@xmath206584,[sii]@xmath206717 , and [ sii]@xmath206731 .
for all objects where the doublet [ oii]@xmath203727 was detected , h@xmath18 was redshifted beyond the wavelength range of the detector .
for one source , at z=1.028 , we detected two features that we assigned as [ oii]@xmath203727 and mgii@xmath202800 emission .
for the other three sources with @xmath3 we find only 1 feature that we identify as mgii@xmath202800 , based on the width of the line and the absence of other features in the spectrum .
figure 3 shows the redshift distribution of the sample .
the mean redshift of the sample is @xmath44 .
two of the sources in the deimos sample have corresponding sdss spectra .
the redshifts obtained for these two objects agree with the sdss redshifts .
one of the objects with sdss spectra is the qso at @xmath45 and the other is a galaxy at @xmath46 .
both the deimos spectra and the sdss spectra for the latter object have significant detections of only h@xmath18 and [ nii]6584 .
we measure spectral emission line fluxes and equivalent widths for [ oii]@xmath203727 , h@xmath43 , [ oiii]@xmath204959 , [ oiii]@xmath205007 , h@xmath18 , [ nii]@xmath206584 , [ sii]@xmath206717 , and [ sii]@xmath206731 .
fluxes , equivalent widths and errors for the emission lines were measured in idl using the mpfit function .
the continuum of each spectrum was first fit by a polynomial , then the emission lines were simultaneously fit with gaussians .
table 1 lists the objects , their equivalent widths and flux ratios of [ oiii]@xmath205007/h@xmath43 and [ nii]@xmath206584/h@xmath18 for objects with a 3@xmath47 measurement of at least one of the above lines .
we did not perform a reddening correction to the fluxes , since the standard procedure requires either detections of both h@xmath18 and h@xmath43 or a measurement of the radio continuum @xcite .
the effect of this on the flux ratios of [ oiii]@xmath205007/h@xmath43 and [ nii]@xmath206584/h@xmath18 is thought to be negligible due to the small wavelength separation of the emission lines . in the deimos sample we only have detections of both h@xmath18 and h@xmath43 for 18 objects , all at @xmath48 , and chose for consistency not to perform corrections on any of the spectra .
we note that the foreground galactic extinction , e(b - v ) , calculated from the dust maps of @xcite is @xmath49 magnitudes for the field .
fourteen objects have measured fluxes of h@xmath43 , [ oiii]@xmath205007 , h@xmath18 , and [ nii]@xmath206584 .
figure 4 shows the emission line diagnostic of @xcite used to distinguish star - forming galaxies and type ii agn .
the solid curve is the boundary between star - forming galaxies and agn determined through modeling of starburst galaxy spectra by @xcite .
a star forming galaxy with only @xmath40% of the optical emission line flux due to an agn would lie above the star - forming / agn boundary @xcite .
the dashed curve is the more stringent demarcation used by @xcite to distinguish between agn and star - forming galaxies in sdss .
the shaded histogram shows the distribution of emission line ratios for a sample of 51,000 sdss / galex objects ( see 4 ) .
the curve of @xcite distinguishes between the two different sources of emission line flux in this large sample with the agn occupying the parameter space to the right of the distribution of star - forming galaxies . in the deimos sample
only three galaxies of the fourteen have emission line flux ratios above this curve , implying that some portion of the flux is due to an agn , though it is still within the errors that the emission line flux for two of these objects results entirely from star - forming regions inside the host galaxies .
objects can also be agn if @xmath50 [ oiii]/h@xmath51 or @xmath50 [ nii]/h@xmath52 .
another 13 objects also have only detections of h@xmath43 and [ oiii]@xmath205007 . of the thirteen ,
one has a flux ratio of @xmath50 [ oiii]/h@xmath51 , indicative of an agn .
@xcite find a similar ratio of galaxies classified as star - forming / agn from examination of emission lines in their sample of uv selected galaxies .
the deimos sample is @xmath53 magnitudes deeper in _
r _ than the sdss spectroscopic sample .
figure 5 shows a histogram of the sdss _
r _ magnitudes for the sample plotted with the sdss _ r _
magnitude for _ galex _ mis sources in ir1.1 with matches in the sdss dr2 spectroscopic sample .
the ir1.1/dr2 sample consists of @xmath54 galaxies at redshifts @xmath55 with derived star - formation histories ( sfhs ) by sed fitting ( see 5 ) .
the ir1.1/dr2 histogram is normalized to the size of the deimos galaxy sample . the ir1.1/dr2 sample is mostly bounded in _
r _ by the sdss spectroscopic survey limits , @xmath56 @xcite . using the redshifts determined for the deimos sample we calculate the absolute _ r _ magnitude , m@xmath57 , of galaxies in the sample , k - corrected to the mean redshift of the ir1.1/dr2 sample , @xmath58 .
k - corrections for all bandpasses were calculated using the publicly available code of @xcite version 4_1_4 . in figure 6
we construct a color magnitude diagram ( cmd ) of @xmath59 versus @xmath60 .
we overplot our sample onto the ir1.1/dr2 sample , plotted as a shaded contour plot . in the figure
the diamond symbols correspond to deimos objects with spectroscopically measured redshifts at @xmath61 , and the crosses correspond to objects with spectroscopically measured redshifts at @xmath62 .
the contours enclose 40% and 80% of the objects in the ir1.1/dr2 sample .
the cmd of the ir1.1/dr2 sample clearly shows the bimodality of galaxies seen by _
galex _ in the nearby universe , @xmath61 , with distribution peaks at blue @xmath59 colors of @xmath63 and at red @xmath5 colors of @xmath64 @xcite .
we henceforth call these the blue and red sequences .
the galaxies in the blue sequence are generally late - type in morphology and have spectra with emission line - ratios indicative of star - formation @xcite , while the galaxies in the red sequence typically have absorption line spectra and lower star - formation rates than blue sequence galaxies . the majority of the objects in the deimos sample lie in the region of the diagram occupied by blue star - forming galaxies in the ir1.1/dr2 sample , mostly along the blue edge of the distribution , with only one having a @xmath5 color greater than 4 . the spectrum for this object , at @xmath65 ,
is shown in figure 1 .
the spectrum shows absorption features : nad , and mgib , but also shows weak emission features of h@xmath18 , [ nii]6584 , and [ sii]6717,6731 . in the bottom - left panel of figure 7
we again plot the color magnitude diagram for the sample , highlighting the objects with @xmath66 detections ( filled circles ) .
the other 3 panels of this figure show the uv color distribution for the 24 galaxies in the deimos sample with @xmath66 detections .
plotted for reference are galaxies from the ir1.1/dr2 sample with @xmath66 detections shown as the shaded countour plots or shaded histogram in the respective panels . besides the 3 galaxies in the deimos sample with @xmath67 the uv color for this sample shows a similar distribution with the ir1.1/dr2 sample . in the fuv cmd
the objects with m@xmath68 show a range of uv color not seen in the local blue sequence galaxies of similar optical luminosity .
both figures 6 and 7 show that this sample probes the type of galaxies that one obtains when looking at objects detected by _
several magnitudes deeper than the sdss spectroscopic limit .
we derive the following galaxy parameters according to the approach of @xcite : the v - band dust attenuations , a@xmath69 in magnitudes , stellar metallicity z , the current star formation rate , sfr , averaged over the past 100 myr in m@xmath9 yr@xmath10 , the present - day stellar mass , m@xmath70 , of the galaxy in m@xmath9 , the fraction of stellar mass formed in bursts over the last 100 myr , f@xmath71 , and the @xcite b parameter , defined as the ratio of the current sfr to the past time - averaged sfr ( averaged over the estimated age , not hubble time ) .
the galaxy parameters are derived from model libraries of galaxies at redshifts between 0.1 and 1.6 at redshift increments of 0.1 for the deimos sample , and at redshifts of 0.05,0.1,0.15,0.2 , and 0.25 for the ir1.1/dr2 sample .
each library consists of up to @xmath410@xmath72 models .
each model is parameterized according to galaxy age , optical depth , star formation history ( sfh ) , and metallicity .
the sfh of each model is parameterized according to @xcite , with an underlying , continuous , exponentially declining sfr upon which bursts of star formation , random in time and amplitude , are superimposed .
dust attenuation in each model is parameterized using the prescription of @xcite using an effective @xmath73-band optical depth @xmath74@xmath69 and absorption curve , @xmath75 resulting from both giant molecular clouds and the diffuse ism , with the fraction @xmath21 of @xmath74@xmath69 contributed only by the diffuse ism .
the @xmath73-band optical depth from giant molecular clouds is taken to only affect stars younger than 10 myr .
a description of the prior distributions of the model parameters is discussed in @xcite .
spectral energy distributions ( seds ) are created for each galaxy in the library using the population synthesis code of @xcite .
the model seds are convolved with the _ galex _ and sdss filter response curves .
statistical estimates of physical galaxy parameters are derived by comparing the observed 7 band galex@xmath30sdss fluxes of each galaxy to all the convolved model seds in the nearest redshift library .
probability density functions ( pdfs ) for each physical parameter are created by assigning weights to the parameters .
the @xmath76 goodness of fit of the model determines the weight ( @xmath77 $ ] ) that is assigned to the parameters of that model . the median ( or most typical parameter value ) of the pdf
is taken as the estimate of the galaxy parameter .
we perform sed fits for galaxies not identified as qsos . in table 2 we list the galaxies , their redshifts , @xmath78 non k - corrected colors , and their derived parameters .
the parameters derived from the sed fits are not used unless the reduced @xmath76 fit of at least one model is below 10 .
only two of the 46 galaxies do not meet this criterion .
figures 8 and 9 show the derived parameters for the deimos sample of galaxies overplotted on the ir1.1/dr2 sample plotted as shaded contour plots .
the contours are labeled and encompass 52% , 84% , and 97% of the data .
the diamonds as before are objects with @xmath61 , and the crosses are objects with @xmath62 .
the objects span the range of derived parameters of the ir1.1/dr2 sample for z , and a@xmath69 .
the derived metallicities , @xmath79 $ ] , for the sample are not very well constrained . the typical error , @xmath80 of the difference between the 97.5 percentile and the 2.5 percentile of the pdf , of the derived metallicities are @xmath8143% z@xmath9 . the derived m@xmath70 range from 10@xmath7 to 10@xmath8 m@xmath9 with an average error of @xmath82dex , but the majority of the m@xmath70 are at or below 10@xmath83m@xmath9 , about an order of magnitude below the mean of the ir1.1/dr2 sample . the derived sfrs lie between @xmath84 and @xmath85 m@xmath86 , with five galaxies having derived sfrs greater than 10 m@xmath86 .
the average error on the derived sfrs is typically @xmath87dex . comparing the sfrs between galaxies with similar @xmath5 color or m@xmath88 , the galaxies at @xmath62 have sfrs about an order of magnitude greater than the galaxies at @xmath61 .
this is mostly a selection effect since the more distant galaxies are also more luminous on average .
a comparison of these sed derived sfrs with h@xmath18 derived sfrs can not be performed since the spectra are not flux calibrated and many lack h@xmath18 .
the range of sfrs we find for this sample is similar to the range of sfrs for local uv galaxies found by @xcite derived both from h@xmath18 luminosity and from the uv luminosity , but is an order of magnitude larger than the sfrs found by @xcite in their sample of goods - north treasury keck redshift survey galaxies at @xmath89 when compared to galaxies in the deimos sample with similar redshifts .
we find that the galaxies are largely starbursts with @xmath90 .
this is shown in figure 10 .
the distribution of @xmath91 for objects in the deimos sample with sed fits are plotted with the distribution for the ir1.1/dr2 sample normalized to the number of objects in the deimos sample with sed fits .
the deimos sample shows a highly peaked value of @xmath91 compared to the ir1.1/dr2 sample which has two small peaks near @xmath92 and @xmath93 , showing that most of these galaxies currently have less star - formation now than in the past .
the value @xmath94 that describes most of the galaxies in the deimos sample indicates that while these galaxies have less star formation than in the past , they are currently or have recently gone through a burst of star - formation .
two of the galaxies have @xmath95 indicating that these galaxies are going through major starbursts in their histories . from a statistical standpoint ,
the sed fits reveal with a 95% reliability that at least three galaxies have not had a burst of starformation in the last 100 myr .
half of the galaxies in the sample could have formed as much as @xmath4 10% of their stellar mass in a burst within the last 100 myr , 10 could have formed up to @xmath96 and five of the less massive systems could have had up to @xmath97 of their stellar mass form in bursts within the last 100myr .
in the plot of @xmath98 vs @xmath6 our sample lies along a `` blue sequence '' of star forming galaxies and they have among the bluest colors for star forming galaxies in the local universe found by _ galex _ and sdss .
while all the galaxies show blue @xmath99 color this is not a homogenous sample .
fifteen of the galaxies show disk structure in the sdss images .
the remaining two - thirds of the sample fit into the heterogeneous class of faint blue galaxies at intermediate redshifts previously studied by @xcite and @xcite in the hubble deep field and adjoining fields , and by @xcite and @xcite in the canada - france redshift survey .
the faint blue subset of the deimos sample has similar luminosities , sfrs and optical colors to the compact galaxies found in the hubble deep field by @xcite , though none are `` compact '' ( optical half - light radii @xmath100.5 . )
this is a selection effect in our sample .
galaxies with half - light radii this small are classified as stars in sdss and would not have been selected for spectroscopy in our sample . in figure 11
we plot , as filled diamonds , the rest frame absolute b magnitude versus the b magnitude surface brightness and sfr versus velocity dispersion .
the absolute b magnitudes are calculated from the observed bandpasses using the k - correction code of @xcite , and the surface brightness are calculated with the half - light sdss r petrosian radii .
the velocity dispersions are derived from the measured linewidths of [ oiii]5007 for objects with @xmath101 detections of the emission line . for comparison ,
we also plot , as stars and squares , the faint blue galaxy samples of @xcite and @xcite making the necessary corrections to our adopted cosmology . while our sample of galaxies have luminosities , sfrs , and velocity dispersions comparable with the previous samples , the size selection effect separates our sample from the previous samples in surface brightness .
@xcite and @xcite propose that the subset of the faint blue population of galaxies with compact geometry ( half light radii @xmath100.5 ) and narrow emission lines ( @xmath102 ) will fade to become dwarf spheroidals by z=0 , while @xcite claims that the most luminous of these are too massive to become dwarf spheroidals , and instead will become the bulges of spiral galaxies .
the most massive intermediate redshift galaxies ( @xmath103 ) in our sample could likely follow this latter evolution path .
indeed , we see that the galaxies in our sample that show extended structure at low redshifts ( @xmath48 ) are among the the most massive of the sample .
the conclusions of @xcite and @xcite rest partially on the assumption that galactic winds from the last starburst event will remove the remaining gas from these systems and halt star formation causing them to fade several magnitudes by z=0 . through modeling the uv and optical broadband colors
we find that about half the galaxies in our present sample could have formed at most @xmath104 of their stellar mass in bursts within the past 100 myr , and only fifteen ( all having @xmath105 ) could have formed over 20% of their stellar mass within the last 100 myr .
if the last starburst event removes all remaining gas from these systems , as @xcite propose , then why did the previous starforming events in these galaxies that produced the majority of the stars in these systems not halt star formation ? our derived starformation histories of these galaxies
argues that the current star formation events will not halt starfromation in every galaxy .
but what percentage of these will continue to experience further bursts of star formation is unknown .
however , even in the low mass systems of the present sample , residual star formation or another burst of star formation is ultimately unlikely to change the evolutionary outcome proposed by @xcite and @xcite for the most compact low mass galaxies in our sample .
we suggest that the more extended sources of our sample are more likely to become present - day dwarf irregulars rather than dwarf spheroidals .
we have presented deimos spectra for objects detected by _
galex _ in the mis survey with imaged counterparts in sdss ; a total exposure time of 30 min per slitmask was used .
galex has proven to be a sensitive instrument for wide field galaxy surveys .
we have shown that the _ galex _ medim imaging survey followed up with a 30 m integration with keck / deimos yields redshifts and line measurements for star forming galaxies to z@xmath4 0.7 , and has yielded 4 qso s 3 of which are previously un cataloged .
the matched sample reaches approximately 3 magnitudes fainter in _ r _ than the sdss spectroscopic survey limits .
the sample is not a homogenous sample , but is indicative of the types of galaxies forming stars out to z@xmath106 .
we have derived physical parameters for these galaxies from the seds and compared this sample to a sample of 50,000 sdss galaxies with galex detections .
we find that roughly one - third of the galaxies are starforming late type disk galaxies , four are qsos at @xmath3 , and the remaining galaxies are faint blue low mass starbursts .
approximately 3 out of 14 star formimg galaxies show emission line ratios indicative of an agn .
a similar fraction was found by @xcite in their uv selected sample .
the masses of the galaxies are typically lower than what is found locally in the sdss spectroscopic sample . the range of @xmath107 for the deimos sample spans from @xmath108 to @xmath109 , whereas the median of the sdss spectroscopic sample is @xmath110 .
+ + 4 . the sfrs of the galaxies at @xmath62
are roughly an order of magnitude greater than the sfrs for the galaxies at @xmath61 .
besides three of the most massive , and reddest in nuv -r color , the remaining galaxies show evidence of a starburst in the last 100 myr , with @xmath111 .
fifteen of the galaxies in the lower mass range of the sample could have formed more than 20% of their stellar mass in bursts of star formation within the last 100 myr .
our sample has similar velocity dispersions , sfrs , and b luminosities to previous samples of faint blue galaxies , though the galaxies in our present study are 2 mag fainter in surface brightess .
_ galex _ is a nasa small explorer , launched in april 2003 .
we gratefully acknowledge nasa s support for construction , operation , and science analysis for the galex mission , developed in cooperation with the centre national detudes spatiales of france and the korean ministry of science and technology .
the authors wish to recognize and acknowledge the very significant cultural role and reverence that the summit of mauna kea has always had within the indigenous hawaiian community .
we are most fortunate to have the opportunity to conduct observations from this mountain .
the analysis pipeline used to reduce the deimos data was developed at uc berkeley with support from nsf grant ast-0071048 .
abazajian , k. et al .
2004 , , 128 , 502 baldwin , j. a. , phillips , m. m. , & terlevich , r. 1981 , pasp , 93 , 5 bertin , e. , & arnouts , s. 1996 , a&as , 117 , 393 blanton , m. et al .
2003 , , 125 , 2276 brinchmann , j. , charlot , s. , white , s. d. m. , tremonti , c. , kauffmann , g. , heckman , t. , & brinkmann , j. 2004 , mnras , 351 , 1151 bruzual , g. & charlot , s. 2003 , mnras , 344 , 1000 charlot , s. & longhetti 2000 , mnras , 323 887 charlot , s. & fall , s. m. 2000 , apj 539 , 718 davis m. et al . in discoveries and research prospects from 6- to 10-meter - class telescopes
ii . edited by guhathakurta , puragra
. proceedings of the spie , volume 4834 , pp .
161 - 172 ( 2003 ) ellis , r. , s. 1007 , araa , 35 , 389 faber , s. m. , et al .
2003 , , 4841 , 1657 fukugita , m. , ichikawa , t. , gunn , j. e. , doi , m. , shimasaku , k. , & schneider , d. p. 1996
, , 111 , 1748 guzm@xmath112n , r. , koo , d. c. , faber , s. m. , illingworth , g. d. , takamiya , m. , kron , r. g. , & bershady , m. a. 1996 , apj , 460 , 5 guzm@xmath112n et al .
1997 , apj , 489 , 559 guzm@xmath112n , r. , jangren , a. , koo , d. c. , bershady , m. a. , & simard , l. 1998 apj 495 , 13 hammer , f. , gruel , n. , thuan , t. x. , flores , h. , & infante , l. 2000 , apj , 550 , 570 hogg , d. w. , finkbeiner , d. p. , schlegel , d. j. , and gunn , j. e. 2001 , , 122 , 2129 kauffmann , f. et al .
2003 , mnras , 341,33 kennicutt , r. c. , ara&a , 36 , 189 kennicutt r. c. , bresolin , f. & garnett , d. r. 2003 , apj , 591 , 801 kewley , l. j. , dopita , m. a. , sutherland , r. s. , heisler , c. a. , & trevena , j. 2001 , apj , 556,121 kobulnicky , h. a. & kewley , l. 2004 , apj , 617 , 240 koo , d. c. , guzm@xmath112n , r. , faber , s. m. , illingworth , g. d. , bershady , m. a. , kron , r. g. , & takamiya , m. 1995 , apj , 440 , 49 martin , c. d. et al .
2005 , apj , 619 , 1 martin , c. d. et al .
2005 , apj , 619 , 59 milliard , b. , donas , j. , laget , m. , armand , c. , vuillemin , a. 1992 a&a , 257 , 24 morrissey , p. et al .
2005 , apj 619 , 7 mallen - ornelas , g. , lilly , s. j. , crampton , d. , & schade , d. 1999 , apj , 518 , 83 osterbrock , d. e. 1989 , astrophysics of gaseous nebulae and active edmunds , m. g. , blackwell , d. e. , chun , m. s. , & smith , g. 1979 , mnras , 189 , 95 phillips , a. c. et al .
1997 , apj , 489 , 543 pier , j. r. , munn , j. a. , hindsley , r. b. , hennessy , g. s. , kent , s. m. , lupton , r. h. , and ivezic , z. 2003 , 125 , 1559 salim , s. et al .
2005 , apj , 619 , 39 salim , s. et al 2007 , arxiv:0704.3611 scalo , j. m. 1986 . _ fundam .
_ 11:1 - 278 schiminovich , d. et al .
2005 , apj , 619,47 d.j .
schlegel , d.p .
finkbeiner , & m. davis , apj , 500 , 525 ( 20 june 1998 ) smith , j. a. et al .
2002 , , 123 , 2121 stoughton , c. et al .
2002 , , 123 , 485 strauss , m. a. et al .
2002 , aj , 124 , 181 sullivan , m. , treyer , m. a. , ellis , s. r. , bridges , t. j. , milliard , b. , donas , j. 2000 , mnras , 312 , 442 tremonti , c. a. et al .
2004 , apj , 613 , 898 treyer , m. a. et al .
2005 , apj , 619 , 15 wyder , t. k et al . 2005 , apj , 619 , 15 wyder , t. k. 2006 , in preparation 587725578037494738 & 0.474 & 7.23 @xmath81 5.29 & 2.11 @xmath81 0.51 & 2.54 @xmath81 0.29 & * * * @xmath81 * * * & * * * @xmath81 * * * & 2.83 @xmath81 0.72 & * * * @xmath81 * * * + 587725578037494286 & 0.134 & * * * @xmath81 * * * & 2.05 @xmath81 0.20 & 2.57 @xmath81 0.22 & 4.51 @xmath81 0.05 & 3.46 @xmath81 0.13 & 1.37 @xmath81 0.18 & 0.25 @xmath81 0.01 + 587725591459201372 & 0.615 & 4.63 @xmath81 1.24 & 3.02 @xmath81 0.50 & 3.74 @xmath81 0.33 & * * * @xmath81 * * * & * * * @xmath81 * * * & 1.46 @xmath81 0.28 & * * * @xmath81 * * * + 587725578037495014 & 0.213 & * * * @xmath81 * * * & 6.16 @xmath81 4.88 & 4.83 @xmath81 4.78 & 58.32 @xmath81 30.70 & 3.91 @xmath81 16.30 & 1.12 @xmath81 1.38 & 0.98 @xmath81 0.60 + 587725578037494283 & 0.084 & * * * @xmath81 * * * & 0.82 @xmath81 0.25 & 1.17 @xmath81 0.25 & 2.36 @xmath81 0.05 & 1.68 @xmath81 0.10 & 1.84 @xmath81 0.69 & 0.27 @xmath81 0.02 + 587725591459201726 & 0.439 & 5.21 @xmath81 0.35 & 3.17 @xmath81 0.18 & 3.34 @xmath81 0.06 & * * * @xmath81 * * * & * * * @xmath81 * * * & 4.95 @xmath81 0.30 & * * * @xmath81 * * * + 587725591459201125@xmath113 & 0.066 & * * * @xmath81 * * * & 7.74 @xmath81 0.19 & 2.74 @xmath81 0.06 & 3.57 @xmath81 0.01 & 3.13 @xmath81 0.02 & 1.40 @xmath81 0.05 & 0.29 @xmath81 0.01 + 587725591459201562 & 0.068 & * * * @xmath81 * * * & 0.81 @xmath81 0.82 & 1.51 @xmath81 0.75 & 2.66 @xmath81 0.19 & 1.17 @xmath81 0.51 & 2.93 @xmath81 3.32 & 0.14 @xmath81 0.06 + 587725591459201436 & 0.066 & * * * @xmath81 * * * & * * * @xmath81 * * * & 2.28 @xmath81 0.10 & 2.25 @xmath81 0.03 & 1.65 @xmath81 0.12 & * * * @xmath81 * * * & 0.14 @xmath81 0.010 + 587725591459201670 & 0.651 & 6.22 @xmath81 0.58 & 3.66 @xmath81 0.77 & 5.06 @xmath81 0.21 & * * * @xmath81 * * * & * * * @xmath81 * * * & 3.98 @xmath81 0.84 & * * * @xmath81 * * * + 587725591459201656 & 0.157 & * * * @xmath81 * * * & 2.04 @xmath81 0.51 & 2.10 @xmath81 0.11 & 2.53 @xmath81 0.12 & 1.60 @xmath81 1.65 & 5.81 @xmath81 1.47 & 0.02 @xmath81 0.02 + 587725591459201666 & 0.078 & * * * @xmath81 * * * & 1.03 @xmath81 1.80 & 1.08 @xmath81 0.62 & 2.10 @xmath81 0.11 & 1.50 @xmath81 0.51 & 2.36 @xmath81 4.33 & 0.18 @xmath81 0.06 + 587725591459201614 & 0.188 & * * * @xmath81 * * * & 2.62 @xmath81 0.40 & 2.69 @xmath81 0.15 & 3.38 @xmath81 0.10 & 2.02 @xmath81 0.83 & 3.01 @xmath81 0.49 & 0.08 @xmath81 0.03 + 587725591459136404 & 0.220 & * * * @xmath81 * * * & 0.74 @xmath81 0.59 & 3.43 @xmath81 0.57 & 4.95 @xmath81 0.23 & 2.77 @xmath81 0.67 & 7.21 @xmath81 5.93 & 0.18 @xmath81 0.04 + 587725591459136351 & 0.319 & 1.34 @xmath81 1.16 & 2.38 @xmath81 0.72 & 0.76 @xmath81 0.45 & * * * @xmath81 * * * & * * * @xmath81 * * * & 0.34 @xmath81 0.23 & * * * @xmath81 * * * + 587725591459136031 & 0.192 & * * * @xmath81 * * * & 0.25 @xmath81 0.22 & 0.59 @xmath81 0.34 & 2.98 @xmath81 0.12 & 2.79 @xmath81 0.18 & 2.42 @xmath81 2.55 & 0.63 @xmath81 0.05 + 587725591459136024@xmath113 & 0.245 & * * * @xmath81 * * * & 2.00 @xmath81 0.44 & 1.81 @xmath81 0.47 & 3.87 @xmath81 0.12 & 2.54 @xmath81 0.32 & 0.85 @xmath81 0.29 & 0.23 @xmath81 0.03 + 587725591459136330 & 0.565 & 8.01 @xmath81 0.95 & 3.13 @xmath81 0.32 & 3.43 @xmath81 0.11 & * * * @xmath81 * * * & * * * @xmath81 * * * & 18.69 @xmath81 1.78 & * * * @xmath81 * * * + 587725591459135913 & 0.293 & * * * @xmath81 * * * & 2.32 @xmath81 0.61 & 2.23 @xmath81 0.47 & 2.76 @xmath81 0.09 & 2.18 @xmath81 0.26 & 1.29 @xmath81 0.44 & 0.19 @xmath81 0.02 + 587725591459136277 & 0.290 & * * * @xmath81 * * * & * * * @xmath81 * * * & 1.81 @xmath81 1.07 & 0.82 @xmath81 0.44 & 2.80 @xmath81 0.41 & * * * @xmath81 * * * & 3.97 @xmath81 2.21 + 587725591459136339 & 0.188 & * * * @xmath81 * * * & 2.97 @xmath81 0.35 & 2.69 @xmath81 0.35 & 5.10 @xmath81 0.07 & 5.15 @xmath81 0.17 & 0.95 @xmath81 0.17 & 0.36 @xmath81 0.01 + 587725578037166749 & 0.320 & * * * @xmath81 * * * & 2.30 @xmath81 0.35 & 2.45 @xmath81 0.86 & * * * @xmath81 * * * & * * * @xmath81 * * * & 0.59 @xmath81 0.22 & * * * @xmath81 * * * + 587725578037166468 & 0.330 & * * * @xmath81 * * * & 1.49 @xmath81 0.20 & 0.94 @xmath81 0.17 & * * * @xmath81 * * * & * * * @xmath81 * * * & 0.62 @xmath81 0.14 & * * * @xmath81 * * * + 587725578037166557 & 0.168 & * * * @xmath81 * * * & 2.47 @xmath81 0.11 & 3.06 @xmath81 0.07 & 4.11 @xmath81 0.03 & 3.06 @xmath81 0.09 & 2.39 @xmath81 0.12 & 0.20 @xmath81 0.01 + 587725578037166705 & 0.192 & * * * @xmath81 * * * & 1.54 @xmath81 0.35 & 2.71 @xmath81 0.29 & 2.50 @xmath81 0.11 & 1.24 @xmath81 0.50 & 2.94 @xmath81 0.73 & 0.06 @xmath81 0.03 + 587725578037166672 & 0.189 & * * * @xmath81 * * * & 3.85 @xmath81 0.15 & 2.92 @xmath81 0.20 & 5.48 @xmath81 0.03 & 5.96 @xmath81 0.07 & 0.59 @xmath81 0.05 & 0.34 @xmath81 0.01 + 587725578037166295 & 0.560 & 7.13 @xmath81 0.52 & 5.94 @xmath81 0.38 & 5.37 @xmath81 0.29 & * * * @xmath81 * * * & * * * @xmath81 * * * & 1.29 @xmath81 0.11 & * * * @xmath81 * * * + 587725578037166628 & 0.652 & 0.95 @xmath81 0.65 & 3.54 @xmath81 0.43 & 0.86 @xmath81 0.45 & * * * @xmath81 * * * & * * * @xmath81 * * * & 0.17 @xmath81 0.09 & * * * @xmath81 * * * + 587725578037166629 & 0.321 & 2.15 @xmath81 4.43 & 2.93 @xmath81 0.17 & 3.21 @xmath81 0.07 & * * * @xmath81 * * * & * * * @xmath81 * * * & 3.02 @xmath81 0.19 & * * * @xmath81 * * * + 587725578037100839 & 0.561 & 5.25 @xmath81 0.34 & 2.84 @xmath81 0.15 & 4.40 @xmath81 0.25 & * * * @xmath81 * * * & * * * @xmath81 * * * & 1.70 @xmath81 0.13 & * * * @xmath81 * * * + 587725578037101267 & 0.293 & * * * @xmath81 * * * & 7.99 @xmath81 8.21 & 6.65 @xmath81 0.68 & 2.60 @xmath81 1.68 & 0.17 @xmath81 0.51 & * * * @xmath81 * * * & 0.10 @xmath81 0.30 + 587725578037101221 & 0.518 & 5.79 @xmath81 0.95 & 3.31 @xmath81 0.35 & * * * @xmath81 * * * & * * * @xmath81 * * * & * * * @xmath81 * * * & * * * @xmath81 * * * & * * * @xmath81 * * * + 587725578037100922 & 0.168 & * * * @xmath81 * * * & 1.24 @xmath81 0.44 & 1.06 @xmath81 0.45 & 3.54 @xmath81 0.11 & 2.82 @xmath81 0.20 & 0.85 @xmath81 0.47 & 0.48 @xmath81 0.04 + 587725578037101176 & 0.281 & * * * @xmath81 * * * & 2.28 @xmath81 0.29 & 3.52 @xmath81 0.27 & 2.49 @xmath81 0.08 & * * * @xmath81 * * * & 2.07 @xmath81 0.31 & * * * @xmath81 * * * + 587725591458873892 & 0.281 & * * * @xmath81 * * * & 2.49 @xmath81 0.21 & 3.14 @xmath81 0.26 & * * * @xmath81 * * * & * * * @xmath81 * * * & 1.21 @xmath81 0.14 & * * * @xmath81 * * * + 587725578037100775@xmath114 & 0.176 & * * * @xmath81 * * * & * * * @xmath81 * * * & * * * @xmath81 * * * & 0.86 @xmath81 0.11 & 1.07 @xmath81 0.08 & * * * @xmath81 * * * & 1.30 @xmath81 0.20 + 587725591458808530 & 0.427 & 2.53 @xmath81 0.96 & 3.58 @xmath81 0.27 & 3.92 @xmath81 0.38 & * * * @xmath81 * * * & * * * @xmath81 * * * & 0.94 @xmath81 0.11 & * * * @xmath81 * * * + 587725591458808528 & 0.598 & 5.42 @xmath81 0.31 & 3.57 @xmath81 0.20 & * * * @xmath81 * * * & * * * @xmath81 * * * & * * * @xmath81 * * * & * * * @xmath81 * * * & * * * @xmath81 * * * + 587725578037101077 & 0.387 & 6.51 @xmath81 1.41 & 3.96 @xmath81 0.33 & 3.12 @xmath81 0.13 & * * * @xmath81 * * * & * * * @xmath81 * * * & 2.40 @xmath81 0.22 & * * * @xmath81 * * * + 587725578037101473 & 0.496 & 4.81 @xmath81 1.31 & 2.44 @xmath81 0.52 & 2.95 @xmath81 0.13 & * * * @xmath81 * * * & * * * @xmath81 * * * & 4.70 @xmath81 1.01 & * * * @xmath81 * * * + 587725578037166675 & 0.498 & 7.79 @xmath81 0.92 & 3.15 @xmath81 0.23 & 3.32 @xmath81 0.14 & * * * @xmath81 * * * & * * * @xmath81 * * * & 1.39 @xmath81 0.11 & * * * @xmath81 * * * lccllcclll 587725578037166705 & 0.192 & 17@xmath27 37@xmath28 30.681@xmath29 & 57@xmath31 20@xmath32 8.70@xmath33 & 22.53 & 23.11 & 1.51 & 8.69 & -0.42 & -0.09 + 587725578037166675 & 0.498 & 17@xmath27 37@xmath28 34.409@xmath29 & 57@xmath31 21@xmath32 13.71@xmath33 & * * * & 22.62 & 0.97 & 9.50 & 0.50 & -0.07 + 587725578037166672 & 0.189 & 17@xmath27 37@xmath28 39.053@xmath29 & 57@xmath31 21@xmath32 30.10@xmath33 & * * * & 23.04 & 3.36 & 10.05 & 0.15 & -0.55 + 587725578037166749 & 0.320 & 17@xmath27 37@xmath28 41.177@xmath29 & 57@xmath31 18@xmath32 46.29@xmath33 & 22.87 & 22.12 & 1.88 & 9.91 & 0.65 & -0.14 + 587725578037100839 & 0.561 & 17@xmath27 37@xmath28 43.542@xmath29 & 57@xmath31 24@xmath32 5.51@xmath33 & * * * & 22.19 & 1.19 & 10.40 & 0.85 & -0.32 + 587725578037166468 & 0.330 & 17@xmath27 37@xmath28 45.923@xmath29 & 57@xmath31 19@xmath32 17.19@xmath33 & 22.38 & 21.50 & 2.15 & 10.85 & 0.48 & -0.84 + 587725578037166557 & 0.168 & 17@xmath27 37@xmath28 58.250@xmath29 & 57@xmath31 20@xmath32 16.80@xmath33 & 20.61 & 20.12 & 1.18 & 9.47 & 0.53 & -0.08 + 587725578037166295 & 0.560 & 17@xmath27 37@xmath28 59.495@xmath29 & 57@xmath31 23@xmath32 21.80@xmath33 & * * * & 22.16 & 0.89 & 9.70 & 0.79 & -0.07 + 587725578037166628 & 0.652 & 17@xmath27 38@xmath28 2.622@xmath29 & 57@xmath31 23@xmath32 26.99@xmath33 & * * * & 22.73 & 1.51 & 10.51 & 1.11 & -0.24 + 587725578037166629 & 0.321 & 17@xmath27 38@xmath28 7.756@xmath29 & 57@xmath31 23@xmath32 34.90@xmath33 & * * * & 22.32 & 1.04 & 9.09 & 0.28 & -0.05 + 587725578037101267 & 0.293 & 17@xmath27 38@xmath28 22.192@xmath29 & 57@xmath31 25@xmath32 2.82@xmath33 & * * * & 23.36 & 1.66 & 9.16 & -0.06 & -0.18 + 587725578037100922 & 0.168 & 17@xmath27 38@xmath28 23.628@xmath29 & 57@xmath31 27@xmath32 56.42@xmath33 & 21.48 & 21.97 & 2.67 & 9.97 & -0.24 & -0.76 + 587725578037101473 & 0.496 & 17@xmath27 38@xmath28 27.876@xmath29 & 57@xmath31 26@xmath32 33.90@xmath33 & 22.91 & 22.28 & 1.22 & 9.76 & 0.68 & -0.10 + 587725578037101077 & 0.387 & 17@xmath27 38@xmath28 36.526@xmath29 & 57@xmath31 30@xmath32 38.59@xmath33 & * * * & 22.32 & 0.52 & 9.11 & 0.51 & -0.01 + 587725578037101221 & 0.518 & 17@xmath27 38@xmath28 37.317@xmath29 & 57@xmath31 26@xmath32 53.92@xmath33 & * * * & 22.50 & 1.09 & 10.07 & 0.62 & -0.26 + 587725578037100775@xmath114 & 0.176 & 17@xmath27 38@xmath28 37.646@xmath29 & 57@xmath31 29@xmath32 13.99@xmath33 & 21.03 & 20.34 & 3.08 & 11.08 & 0.69 & -0.88 + 587725578037101176 & 0.281 & 17@xmath27 38@xmath28 41.873@xmath29 & 57@xmath31 28@xmath32 9.30@xmath33 & * * * & 22.33 & 0.99 & 9.24 & -0.09 & -0.19 + 587725591458873892 & 0.281 & 17@xmath27 38@xmath28 48.259@xmath29 & 57@xmath31 29@xmath32 10.40@xmath33 & 22.69 & 22.23 & 1.80 & 9.76 & 0.43 & -0.13 + 587725591458808530 & 0.427 & 17@xmath27 38@xmath28 50.596@xmath29 & 57@xmath31 29@xmath32 54.31@xmath33 & * * * & 22.88 & 2.89 & 10.50 & 2.12 & 0.00 + 587725591458808528 & 0.598 & 17@xmath27 38@xmath28 54.873@xmath29 & 57@xmath31 30@xmath32 5.80@xmath33 & * * * & 22.58 & 1.83 & 10.37 & 1.33 & -0.09 + 587725578037494286 & 0.134 & 17@xmath27 39@xmath28 45.205@xmath29 & 56@xmath31 40@xmath32 2.20@xmath33 & 20.87 & 20.47 & 1.17 & 9.66 & -0.15 & -0.28 + 587725578037494409 & 0.355 & 17@xmath27 39@xmath28 51.035@xmath29 & 56@xmath31 39@xmath32 13.11@xmath33 & * * * & 23.23 & 1.78 & 9.14 & 0.36 & -0.04 + 587725578037495014 & 0.213 & 17@xmath27 39@xmath28 51.343@xmath29 & 56@xmath31 39@xmath32 58.89@xmath33 & * * * & 22.73 & 0.95 & 8.18 & -0.66 & -0.02 + 587725578037494283 & 0.084 & 17@xmath27 39@xmath28 53.130@xmath29 & 56@xmath31 40@xmath32 18.91@xmath33 & 22.24 & 22.00 & 2.34 & 9.17 & -0.57 & -0.35 + 587725578037494050 & 0.077 & 17@xmath27 39@xmath28 56.243@xmath29 & 56@xmath31 38@xmath32 17.20@xmath33 & * * * & 22.11 & 5.46 & 10.76 & -0.80 & -1.91 + 587725578037494494 & 0.356 & 17@xmath27 39@xmath28 56.287@xmath29 & 56@xmath31 37@xmath32 21.00@xmath33 & * * * & 22.20 & 1.96 & 9.99 & 1.14 & -0.03 + 587725578037494738 & 0.474 & 17@xmath27 39@xmath28 56.704@xmath29 & 56@xmath31 37@xmath32 53.80@xmath33 & * * * & 22.72 & 0.86 & 9.67 & 0.38 & -0.16 + 587725591459201436 & 0.066 & 17@xmath27 40@xmath28 1.663@xmath29 & 56@xmath31 44@xmath32 3.12@xmath33 & 20.89 & 20.74 & 2.20 & 9.21 & -0.41 & -0.35 + 587725591459201125@xmath113 & 0.066 & 17@xmath27 40@xmath28 3.853@xmath29 & 56@xmath31 41@xmath32 59.82@xmath33 & 20.37 & 19.99 & 1.96 & * * * & * * * & * * * + 587725591459201656 & 0.157 & 17@xmath27 40@xmath28 6.724@xmath29 & 56@xmath31 43@xmath32 49.51@xmath33 & 22.64 & 22.12 & -0.52 & 8.68 & -0.97 & -0.24 + 587725591459201627 & 0.079 & 17@xmath27 40@xmath28 8.196@xmath29 & 56@xmath31 44@xmath32 57.80@xmath33 & 22.18 & 21.68 & 1.51 & 8.48 & -0.61 & -0.09 + 587725591459201372 & 0.615 & 17@xmath27 40@xmath28 8.328@xmath29 & 56@xmath31 39@xmath32 39.82@xmath33 & * * * & 22.68 & 1.80 & 10.53 & 0.81 & -0.51 + 587725591459201929 & 0.436 & 17@xmath27 40@xmath28 15.117@xmath29 & 56@xmath31 46@xmath32 4.70@xmath33 & 22.36 & 22.16 & 0.24 & 9.27 & 0.22 & -0.07 + 587725591459201726 & 0.439 & 17@xmath27 40@xmath28 17.307@xmath29 & 56@xmath31 41@xmath32 31.09@xmath33 & * * * & 22.42 & 0.94 & 9.98 & 0.37 & -0.22 + 587725591459136339 & 0.188 & 17@xmath27 40@xmath28 17.468@xmath29 & 56@xmath31 48@xmath32 5.50@xmath33 & 21.23 & 20.57 & 0.76 & 9.98 & -0.06 & -0.47 + 587725591459201670 & 0.651 & 17@xmath27 40@xmath28 19.175@xmath29 & 56@xmath31 43@xmath32 39.18@xmath33 & * * * & 22.46 & 0.14 & 8.93 & 0.61 & 0.29 + 587725591459201562 & 0.068 & 17@xmath27 40@xmath28 20.969@xmath29 & 56@xmath31 42@xmath32 43.30@xmath33 & * * * & 22.77 & 2.90 & 8.77 & -1.06 & -0.53 + 587725591459201614 & 0.188 & 17@xmath27 40@xmath28 23.225@xmath29 & 56@xmath31 45@xmath32 42.52@xmath33 & 22.37 & 21.50 & 0.59 & 8.72 & -0.06 & -0.04 + 587725591459136404 & 0.220 & 17@xmath27 40@xmath28 23.708@xmath29 & 56@xmath31 46@xmath32 34.10@xmath33 & 22.53 & 21.76 & 0.79 & 9.59 & -0.47 & -0.55 + 587725591459135913 & 0.293 & 17@xmath27 40@xmath28 29.685@xmath29 & 56@xmath31 50@xmath32 27.71@xmath33 & 21.64 & 21.93 & 1.18 & 9.55 & 0.12 & -0.21 + 587725591459136024@xmath113 & 0.245 & 17@xmath27 40@xmath28 32.498@xmath29 & 56@xmath31 49@xmath32 12.51@xmath33 & 22.73 & 20.99 & -0.08 & * * * & * * * & * * * + 587725591459201666 & 0.078 & 17@xmath27 40@xmath28 33.193@xmath29 & 56@xmath31 44@xmath32 5.32@xmath33 & * * * & 20.82 & 1.06 & 9.00 & -0.90 & -0.42 + 587725591459136277 & 0.290 & 17@xmath27 40@xmath28 35.691@xmath29 & 56@xmath31 50@xmath32 29.29@xmath33 & 22.66 & 22.23 & 3.27 & 10.99 & 0.49 & -0.98 + 587725591459136330 & 0.565 & 17@xmath27 40@xmath28 46.362@xmath29 & 56@xmath31 49@xmath32 13.40@xmath33 & * * * & 22.18 & 1.33 & 10.03 & 1.06 & -0.08 + 587725591459136031 & 0.192 & 17@xmath27 40@xmath28 56.177@xmath29 & 56@xmath31 49@xmath32 15.38@xmath33 & 20.94 & 20.09 & 2.45 & 10.92 & 0.97 & -0.47 + 587725591459136351 & 0.319 & 17@xmath27 40@xmath28 56.594@xmath29 & 56@xmath31 48@xmath32 52.20@xmath33 & * * * & 21.35 & 2.15 & 10.90 & 0.75 & -0.60 + + + 587725578037100834 & 1.515 & 17@xmath27 38@xmath28 9.360@xmath29 & 57@xmath31 25@xmath32 21.36@xmath33 & * * * & 23.40 & 2.63 & * * * & * * * & * * * + 587725591459201473 & 1.387 & 17@xmath27 40@xmath28 9.836@xmath29 & 56@xmath31 40@xmath32 7.68@xmath33 & * * * & 22.67 & 1.14 & * * * & * * * & * * * + 587725591459136108 & 1.632 & 17@xmath27 40@xmath28 43.682@xmath29 & 56@xmath31 48@xmath32 45.36@xmath33 & * * * & 23.05 & 1.67 & * * * & * * * & * * * + 587725591459135981@xmath114 & 1.028 & 17@xmath27 40@xmath28 49.197@xmath29 & 56@xmath31 47@xmath32 23.65@xmath33 & 22.62 & 21.20 & 1.94 & * * * & * * * & * * * | we report results from a pilot program to obtain spectroscopy for objects detected in the _ galaxy evolution explorer _ ( _ galex _ ) medium imaging survey ( mis ) .
our study examines the properties of galaxies detected by _
galex _
fainter than the sloan digital sky survey ( sdss ) spectroscopic survey .
this is the first study to extend the techinques of @xcite to estimate stellar masses , star formation rates ( sfr ) and the b ( star formation history ) parameter for star - forming galaxies out to @xmath0 .
we obtain redshifts for 50 _ galex _ mis sources reaching @xmath1 ( ab mag ) , having counterparts in the sdss data release 4 ( dr4 ) . of our sample ,
43 are starforming galaxies with @xmath2 , 3 have emission line ratios indicative of agn with @xmath2 , and 4 objects with @xmath3 are qsos , 3 of which are not previously cataloged .
we compare our sample to a much larger sample of @xmath450,000 matched galex / sdss galaxies with sdss spectroscopy ; while our survey is shallow , the optical counterparts to our sources reach @xmath43 magnitudes fainter in sdss _
r _ than the sdss spectroscopic sample .
we use emission line diagnostics for the galaxies to determine that the sample contains mostly star - forming galaxies .
the galaxies in the sample populate the blue sequence in the @xmath5 vs @xmath6 color - magnitude diagram .
the derived stellar masses of the galaxies range from 10@xmath7 to 10@xmath8 m@xmath9 and derived sfrs are between 10@xmath10 and 10@xmath11 m@xmath9@xmath12 .
our sample has sfrs , luminosities , and velocity dispersions that are similar to the samples of faint compact blue galaxies studied previously in the same redshift range by @xcite , @xcite , & @xcite .
however , our sample is @xmath13 mag fainter in surface brightness than the compact blue galaxies .
we find that the star - formation histories for a majority of the galaxies are consistent with a recent starburst within the last 100 myr . |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.